The Value of 3 Tesla Magnetic................

The Value of 3 Tesla Magnetic ResonanceImaging for the Detection and Aggressiveness

Assessment of Prostate Cancer

- From Theory to Practice -

THOMAS HAMBROCK

SIEMENS

The Value of 3 Tesla Magnetic Resonance Imaging for the Detection and Aggressivenes


From Theory to Practice

THOMAS HAMBROCK

1

The studies presented in this thesis were carried out at the Department of Radiology, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands, This project was supported by the Queen Wilhelmina Fund from the Dutch Cancer Society.

Nov 2012

Copyright © Thomas Hambrock

Publisher:

ISBN:

2

.

978-90-9027256-6

Nov 2012

Copyright © Thomas Hambrock

Publisher: Drukkerij Efficiënt Nijmegen

ISBN: 978-90-902756-6

The Value of 3 Tesla Magnetic Resonance Imaging for the Detection and Aggressivenes


From Theory to Practice

PROEFSCHRIFT

TER VERKRIJGING VAN DE GRAAD VAN DOCTOR AAN DE RADBOUD UNIVERSITEIT NIJMEGEN OP GEZAG VAN DE RECTOR MAGNIFICUS, PROF.

MR. S.C.J.J. KORTMANN, VOLGENS BESLUIT VAN HET COLLEGE VAN DECANEN IN HET OPENBAAR TE VERDEDIGEN OP DINSDAG 4 DECEMBER

2012 OM 13:30 UUR PRECIES

DOOR

THOMAS HAMBROCK

GEBOREN OP 1 SEPTEMBER 1978 TE PRETORIA, ZUID-AFRIKA

3

PROMOTOR : Prof. dr. J.O. Barentsz

COPROMOTOREN : Dr. ir. H.J. Huisman

Dr. ir. T.W.J. Scheenen

Dr. C.A. Hulsbergen-van de Kaa

MANUSCRIPTCOMMISIE : Prof. dr. J. van Krieken

Dr. E. van Lin

Prof. dr. A. Villers (University of Lille)

4

PROMOTOR : Prof. dr. J.O. Barentsz

COPROMOTOREN : Dr. ir. H.J. Huisman

Dr. ir. T.W.J. Scheenen

Dr. C.A. Hulsbergen-van de Kaa

MANUSCRIPTCOMMISIE : Prof. dr. J. van Krieken

Dr. E. van Lin

Prof. dr. G. Villeirs (University of Gent)

Dedicated to my grandfather . . . . . .

HERMANN AUGUST HAMBROCK * 8. MAY 1907 -

Who died at the young age of 63 years due to metastatic prostate cancer.

5

Sir Bertrand Russels (1872-1970)

6

Table of Contents

PART ONE - INTRODUCTION AND BACKGROUND

Chapter 1 . . . . 11 28

Introduction.

Chapter 2 . . . . 29 59

Background to functional MR imaging.

PART TWO - DETECTION OF PRIMARY AND RECURRENT PROSTATE CANCER -

Chapter 3 . . . . 61 82

32-Channel 3T MR guided biopsies of prostate tumor suspicious regions identified on multi-

modality 3T MR imaging : Technique and feasibility. Invest Radiol 2009 [HAMBROCK T,

FÜTTERER JJ, HUISMAN HJ et al.]{Impact factor 4.7}

Chapter 4 . . . . 83 100

MRI guided prostate biopsies in men with repetitive negative biopsies and elevated PSA.

J Urol 2010 [HAMBROCK T, SOMFORD DM, HOEKS C et al.]{Impact factor 3.9}

Chapter 5 . . . . 101 114

MR guided prostate biopsies of DCE-MR imaging suspicious tumor regions for the diagnosis

of prostate cancer following radiotherapy. Invest Radiol 2010 [YAKAR D; HAMBROCK T,

HUISMAN HJ et al.] {Impact factor 4.7}

Chapter 6 . . . . 115 140

Multiparametric MR imaging for detection and localization of low vs. high-grade transition

zone prostate cancer. Radiology Accepted [HOEKS C, HAMBROCK T, YAKAR D et al.]

{Impact factor 6.1}

7

Table of Contents

PART THREE - ASSESSMENT OF PROSTATE CANCER AGGRESSIVENESS -

Chapter 7 . . . . 142 -165

The Relation of Apparent Diffusion Coefficient and prostate cancer Gleason grade in

Peripheral Zone. Radiology 2011 [HAMBROCK T, SOMFORD DM, HUISMAN HJ et al.]

{Impact factor 6.1}

Chapter 8 . . . . 166 182

Initial experience with identifying high-grade prostate cancer using diffusion-weighted MR

imaging in patie -guided biopsy. Invest

Radiol 2012 [HAMBROCK T, SOMFORD DM, OORT I et al.]{Impact factor 4.7}

Chapter 9 . . . . 183 199

In vivo assessment of prostate cancer aggressiveness using three-dimensional proton MR

spectroscopy at 3T with the combined endorectal coil and pelvic phased array coil. Eur

Urol 2011 [KOBUS T, HAMBROCK T, HULSBERGEN C et al.] {Impact factor 8.8}

Chapter 10 . . . . 200 219

Prospective Assessment of Prostate Cancer Aggressiveness using 3 Tesla diffusion weighted

MR imaging guided biopsies versus a systematic 10-Core transrectal ultrasound prostate

biopsy cohort Eur Urol 2012 [HAMBROCK T, HOEKS C, HULSBERGEN C et al.] {Impact

factor 8.8}

PART 4 CALIBRATION/COMPUTER ASSISTED DIAGNOSIS OF PROSTATE CANCER -

Chapter 11 . . . . 221 236

The effect of inter-patient normal peripheral zone apparent diffusion coefficient variation

on the prediction of prostate cancer aggressiveness. (Radiology Accepted) [LITJENS G,

HAMBROCK T, BARENTSZ JO et al.]{Impact factor 6.1}

Chapter 12 . . . . 238 259

Computer-aided diagnosis of prostate cancer using multiparametric 3T MR imaging: Effect

on observer performance. Radiology Accepted [HAMBROCK T, VOS P, BARENTSZ JO et

al.] {Impact factor 6.1}

8

Table of Contents

PART THREE - ASSESSMENT OF PROSTATE CANCER AGGRESSIVENESS -

Chapter 7 . . . . 142 -164

The Relation of Apparent Diffusion Coefficient and prostate cancer Gleason grade in

Peripheral Zone. Radiology 2011 [HAMBROCK T, SOMFORD DM, HUISMAN HJ et al.]

{Impact factor 6.1}

Chapter 8 . . . . 165 182

Initial experience with identifying high-grade prostate cancer using diffusion-weighted MR

imaging in patie -guided biopsy. Invest

Radiol 2012 [HAMBROCK T, SOMFORD DM, OORT I et al.]{Impact factor 4.7}

Chapter 9 . . . . 183 199

In vivo assessment of prostate cancer aggressiveness using three-dimensional proton MR

spectroscopy at 3T with the combined endorectal coil and pelvic phased array coil. Eur

Urol 2011 [KOBUS T, HAMBROCK T, HULSBERGEN C et al.] {Impact factor 8.8}

Chapter 10 . . . . 200 220

Prospective Assessment of Prostate Cancer Aggressiveness using 3 Tesla diffusion weighted

MR imaging guided biopsies versus a systematic 10-Core transrectal ultrasound prostate

biopsy cohort Eur Urol 2012 [HAMBROCK T, HOEKS C, HULSBERGEN C et al.] {Impact

factor 8.8}

PART 4 CALIBRATION/COMPUTER ASSISTED DIAGNOSIS OF PROSTATE CANCER -

Chapter 11 . . . . 221 236

The effect of inter-patient normal peripheral zone apparent diffusion coefficient variation

on the prediction of prostate cancer aggressiveness. (Radiology Accepted) [LITJENS G,

HAMBROCK T, BARENTSZ JO et al.]{Impact factor 6.1}

Chapter 12 . . . . 238 259

Computer-aided diagnosis of prostate cancer using multiparametric 3T MR imaging: Effect

on observer performance. Radiology Accepted [HAMBROCK T, VOS P, BARENTSZ JO et

al.] {Impact factor 6.1}

Table of Contents

PART 5 - DISCUSSION, CONCLUSIONS AND FUTURE PERSPECTIVES -

Chapter 13 . . . . 261 292

Discussion, conclusions and future perspectives

Chapter 14 . . . . 293 308

English summary - Nederlandse samenvatting

PART SIX POSTLUDE

A. List of Publications . . . . 310 314

B. List of Presentations Scientific Paper Presentations . . . . 315 316

C. List of Presentations Presentations on Invitation . . . . 317 317

D. List of Awards . . . . 318 318

E. Curriculum Vitae . . . . 319 319

F. Dankwoord . . . . 320 322

9

PART SIX POSTLUDE

A. List of Publications . . . . 311 315

B. List of Presentations Scientific Paper Presentations . . . . 316 317

C. List of Presentations Presentations on Invitation . . . . 318 318

D. List of Awards . . . . 319 319

E. Curriculum Vitae . . . . 320 320

F. Dankwoord . . . . 321 323

PART ONE

INTRODUCTION AND BACKGROUND

— CHAPTER 1 —

Introduction

CHAPTER 1CHAPTER 1

Introduction 1

GENERAL INTRODUCTION

March 2011, the Netherlands:

[UROLOGIST] : “Mr. v. R, your PSA levels have now continuously been rising over the last 8

years from 3 ng/ml to 50 ng/ml. You have severe allergies to multiple antibiotics. So I think

prostate biopsies are no option.”

[PATIENT MR. v. R]: “How can we exclude prostate cancer then?”

[UROLOGIST]: “Apart from biopsies, which we cannot perform in your case, there is no

alternative.”

[PATIENT MR. v. R]: “I have heard that MRI can be used for prostate cancer detection.”

[UROLOGIST]: “Nonsense, you cannot see prostate cancer on MRI!”

This is a true conversation which has taken place in a first world country at the beginning of the

year 2011. The patient mentioned above, was the second last patient (prior to writing this

introduction) the author of this book had scanned using multi-parametric MRI for the evaluation

of prostate cancer. After patient persistence, a biopsy was performed under special antibiotic

coverage. The final diagnosis on biopsy: prostate cancer in all 10 biopsy cores left and right,

Gleason Score 4+3=7. Following MR imaging, extracapsular extension with neurovascular

bundle infiltration and seminal vesicle invasion was diagnosed. In addition, the presence of

metastatic lymph nodes was also established. Radiological stage T3B N1 M0 disease. Treatment

with intent to cure: unlikely.

The author can only stand in awe and disbelief when such hesitancy, ignorance and lack of

knowledge in the year 2011 is still present in a first world country. A change has to be brought

about. Not a mere change, but more importantly “A PARADIGM SHIFT”.

There is probably no single word in human history which has caused so much fear, suffering,

and inner turmoil both on the side of the patient as well as the side of relatives and friends as the

word: “CANCER”. Celsus (28 BC - 50 AC), a Roman doctor, translated the Greek word "carcinos"

12

Introduction 1

Introduction 1

into the word "cancer", a Latin word meaning crab or crayfish as a symbol of being eaten and

torn apart by a crab (cancer). The fear this word arose has elicited the greatest battle fought on

earth, the battle of the inner mind. The author subjectively is of the opinion that in the 21st

century, cancer has become more of a mental burden to humans than a physical one, without

downplaying the severe physical suffering of millions of patients who die to this conglomerate of

diseases or suffer severe morbidity thereof.

The cover page of Section One – Introduction and Background, reveals the scene from Leonardo

da Vinci’s (1452-1519) lost painting, “The Battle of Anghiari” (1505), believed to be still hidden

beneath later frescoes in the Hall of Five Hundred in the Palazzo Vecchio, Florence, Italy. The

current picture is a painting of the original made by the famous Flemish painter Peter Paul

Rubens (1577-1640) and this copy can still be appreciated in the Louvre, Paris, France. To the

author, this painting is the perfect reflection on the current state of the diagnosis and

management of prostate cancer. It is a state of war, blood, tears, swords and horses and sounds

of thunder. Truly a state of chaos!

It is therefore the humble vision of the author that this current thesis may provide a stepping

stone to bring about A PARADIGM SHIFT in the Prostate World of Warcraft. This thesis is not

THE paradigm shift, but merely the beginning of a stone that has become dislodged, one that in

combination with many new scientific insights will bring about this future shift. It is inherent to

human nature to resist an alteration of one’s chosen path, especially if one had trod that path for

so many years. As the medical community gains more scientific evidence, and as the plight of the

patient is increasingly recognized, a more peaceful path can be taken along the long journey of

prostate cancer. There will be a different way of thinking and a different way of “doing”. The

author agrees with the Joker from the movie Batman, who said: “I like to rattle cages”. If cages

are rattled and people are stirred by the content of this thesis and awakenings happen, then it

has fulfilled its purpose already. May this hold true for radiologists, oncologists, urologists,

radiotherapists and patients alike!

It is exceptional in modern days for “new” discoveries, inventions or advances in science to be

labeled to the sole genius of a single persons work. For all scientific facts that we unravel and

discover in current year and age, the author humbly and whole heartedly is obliged to agree with

Bernard of Chatres († 1124) who said:

13

Introduction 1

Introduction 1

"Nos esse quasi nanos, gigantium humeris insidentes, ut possimus plura eis et remotiora videre, non

utique proprii visus acumine, aut eminentia corporis, sed quia in altum subvenimur et extollimur

magnitudine gigantea"

“We are like dwarfs on the shoulders of giants, so that we can see more than they, and things at a

great distance, not by virtue of any sharpness of sight on our part, or any physical distinction, but

because we are carried high and

Cedalion standing on the shoulders of the giant Orion, by Nicolas Poussin, 1658

In the current chain of inventions and groundbreaking discoveries, as well as the small

improvements which have lead to some of the MR imaging advances in prostate cancer as

outlined in this thesis, the author can merely acknowledge that we and our work are dwarfs on

the shoulders of giants. Many giants have been before us lifting us high to see what we currently

see. It is important in this thesis to mention a few of the hundreds of giants: Conrad Röntgen

(1845-1923) the discoverer of Röntgen rays and therefore the father of the most exciting field in

medicine: Radiology; Rudolf Virchow (1821-1902) whose pioneering work in the field of

14

Introduction 1

Introduction 1

histopathology has introduced the “gold standard” of all our work – the father of Histopathology.

Furthermore the ingenious work of Max Planck (1858-1947), regarded as the father of

Quantumphysics which underlies the crucial fundamentals of magnetic resonance imaging and

with which Walther Gerlach (1889-1979) later discovered the spin quantification in a magnetic

field, thereby serving as the beginning point of MRI. The ground breaking and fundamental

work on prostate cancer pathology and assessment, later referred to as the Gleason Scoring

system cannot be omitted by mentioning the giant: Donald F. Gleason (1920-2008).

In memory of these and many other “Giants” who paved the way of science…..

Conrad Röntgen Max Planck Rudolf Virchow

Walther Gerlach Donald Gleason

15

Introduction 1

Introduction 1

INTRODUCTION TO PROSTATE CANCER

Why is the prostate so important? This little organ which serves a crucial role in the

reproductive capabilities of the human species. However, for other reasons it has much greater

importance. The greatest importance is the effect that this organ has on the mind of the man

(agreeing that the external male sexual organs probably have a larger impact). It is rumored

that the sexual wellbeing of a woman is directly related to the wellbeing of the male prostate.

Most articles dealing with prostate cancer (PCa) all start with mentioning the prevalence and

mortality and the great burden prostate cancer has on society. Especially the notion that 1 in 6

men will develop prostate cancer is enough the let the war trumpets sound for the Battle of

Anghiari. A second disturbing phrase is so often mentioned in combination with PCa: “Surely sir,

you are more likely to die with prostate cancer than from it.” While this is true for a large

proportion of patients, this undoubtedly adds to the turmoil, chaos and bewilderment mostly on

the side of the patient. It is almost unfathomable that such a minute organ as the prostate has

become galactic in size (regarding the amount of literature on it and for some patients, this

organ, even in the benign state, can become truly gigantic). It is therefore purposeless and futile

to give a full introduction on this topic. To vaguely unravel the chaos of the Battle of Anghiari,

the author wishes to mention a few epidemiological, statistical and pathological points,

especially to introduce the unacquainted reader with some background to assist in reading this

thesis. More details especially on the anatomy and pathology of prostate cancers will become

evident in later chapters.

Indeed, PCa has become the most widely diagnosed cancer in males with 29% of all cancer

diagnoses being prostate cancer. Equally, in females, breast cancer has become the mostly

diagnosed malignancy, representing 29% of all cancers being diagnosed. For both, the absolute

mortality figures are only surpassed by lung cancer. These figures are surely shocking as regard

to the absolute numbers, being slightly over 240 000 diagnosed new cases for prostate and

220 000 breast cancers in the U.S.A. (1) and around 10 000 prostate cancers diagnosed in the

Netherlands in 2010. According to the American cancer statistics of 2012, around 12% of men

diagnosed with PCa succumb from their prostate cancer compared to 17% for females with

breast cancer. Therefore, a considerable difference exists in being diagnosed with PCa and

eventually dying from it. It is however important to consider the fact that patients currently

16

Introduction 1

Introduction 1

dying from PCa are more often elderly men who have “missed” the earlier screening and

effective therapy now offered to younger patients. Therefore the likelihood of dying from

prostate cancer when one is currently (in the year 2012) diagnosed with the disease, is expected

to be much lower. About 3.5% of all male deaths are from PCa, making the lifetime risk that a

man will succumb of this disease about 1 in 28. This compares to the lifetime risk of dying in a

car accident, about 1 in 4 000 or in an airline accident about 1 in 100 000 (2).

Figure 1. Cancer Statistics 2012 in U.SA. from Siegel et al.(1)

Prostate cancer has been known as a disease of elderly men. It is therefore not surprising that

2/3 of all PCa deaths occur in men > 75 years. Although ~ 1 in 7 (14%) of men die from PCa,

only 1 in 20 (5%) of these deaths are “premature” (the author acknowledges that this might be

arbitrarily in the modern age), occurring in men younger < 75 years. Diagnosis is rare before

age 50, but after this age incidence increases exponentially and the rate of increase is faster than

seen in other malignancies.

17

Introduction 1

Introduction 1

Figure 2. Worldwide incidence and mortality of prostate cancer. (3;4)

Yet, many publications advocated that at least 50% of currently diagnosed prostate cancer are

indolent or insignificant, comprising small (< 0.5 cc) tumours with only well differentiated

components (Gleason grade 3 or less) that apparently won’t lead to death or morbidity. From a

statistical point of view, this is true. There is a great advocacy that men diagnosed with such

cancers should be left alone and sent home. The author wishes to make a bold and provocative

statement that this is not true. Every cancer begins with one cell, then two then four …. until it

has reached great size and metastasizes. Additionally, it appears that most humans are not born

with cancers or with an increased likelihood of genetic mutation. These occur most often de

novo during their life time. Even if there are hereditary components, a two-hit sequence is often

needed for eventual manifestation of disease. From postmortem examination, the true

prevalence of prostate cancer in all ages is at the least to say, catastrophic. The author

contemplates that he himself (at the age of 33 years) has a likelihood of around 25% of

harboring PCa as these sentences are written. Yet this “toothless lion” sometimes grows teeth

and sometimes not. Sometimes it sleeps in its den and sometimes it comes out to feed. When

and why, we don’t know. At least, not yet!

18

Introduction 1

Introduction 1

Figure 3. Prevalence of PCa in Autopsy of white American males. Delongchamps et al.(3)

There is no current evidence (and this is also very difficult to prove) to show that aggressive

prostate cancers (that eventually lead to morbidity and mortality), start their life as aggressive

(meaning Gleason grade 4/5) small tumours. The vast majority of aggressive tumours have well

differentiated components as well. Would an aggressive proliferating cell suddenly turn benign?

The contrary is rather the case. Therefore, a substantial number of tumours begin as well

differentiated “good little cancers” – the wolf in sheepskin. For reasons unknown, some of these

tumours (probably undergoing additional mutations under carcinogenic or other influences)

undergo further dedifferentiation into tumours that cause eventual clinical problems.

It is the burden of the scientific community to unravel which tumours have the potential to cause

problems in the future from those who don’t. Probably this is not possible at all, as visiting the

“Oracle of Delphi” is not an option anymore and future additional mutations cannot be predicted.

They happen when they happen. Therefore the author undoubtedly is of the opinion that these

“good little cancers” should not be diagnosed at all (in their sheepskin phase) and only the

tumours with potential asocial behavior identified early and treated, BUT that all patients should

be offered a reliable method to follow them through life to identify when good tumours show

19

Introduction 1

Introduction 1

aggressive dedifferentiation and further treatment is necessary. This thesis presents some

important concepts that will lift the veil of what is possible, both now, and in the future.

Nearly every aspect of PCa generates controversies for both doctors and patients. While dietary

fat of animal origin has repeatedly been associated with the risk of developing and dying of

prostate cancer, there is no clear evidence yet that dietary alterations or supplementation of

micronutrients can prevent or modify the course of this disease. Similarly, diagnosis and staging

are also controversial. While the introduction of the serum Prostate Specific Antigen (PSA) has

dramatically changed the rate of PCa diagnosis and altered the stage at diagnosis, it has often

been criticized for its lack of specificity resulting in over performing random prostate biopsies

and leading to over detection of innocuous cancers. Despite this, its role in the current and

future diagnosis of prostate cancer remains.

Unfortunately many (especially older, for some reason, mostly German) physicians rely and

swear on their own fingers capabilities to “feel” and therefore diagnose prostate cancers with a

high certainty. Needless to say, only a small proportion of the prostate gland can be felt by

digital rectal examination (DRE). DRE is unfortunately too often used to stage prostate cancer

and to make decisions regarding preservation (albeit not) of the neurovascular bundles during

radical prostatectomy. The overall sensitivity for the digital rectal examination is only 37% (5).

The author is of the opinion that the digital rectal examination should rather be placed in the

spiritual, more theological realm. One definitely needs a divine finger to rely on (Fig. 4), when

dealing with prostate cancer. As Prof. Barentsz always says, “Our fingers have no eyes but we have

MRIs!” The fallen “Finger of God” in South-West Africa (fallen 1988) (Fig. 5) serves as an

understatement of the current role of DRE in prostate cancer diagnosis and management. Of

course, the exception is the case, where tumours are only initially identified using DRE. It

however still has some role to play in the subjective experience, both by patient and physician,

that a “thorough” examination was performed.

20

Introduction 1

Introduction 1

Figure 4. The Divine finger (needed to be good in cancer detection and staging) by

Michelangelo, Cistine chapel, 1511. Figure 5. in Namibia, fallen down

in 1988.

PCa is often (not always) detected by multiple, “systematic” but actually random, transrectal

needle biopsies of the prostate, rather than targeted biopsy of a palpable nodule or a lesion

visible by imaging. There is no other solid organ in which “blinded” biopsies are performed to

make a diagnosis. Both patient and clinicians alike would vehemently resist the possibility that

for example breast cancer should be diagnosed by performing 20 odd blind biopsies of the

breast on each side, hoping to have struck the tumour nodule hiding in the abundance of fat and

glandular tissue. This however was and is still the case with prostate cancer. Different biopsy

schemes have been advocated to “strike gold” more often.

Figure 6. Prostate biopsy schemes and cancer detection rates as advocated by Presti (6)

21

Introduction 1

Introduction 1

Once detected, the size, location and extent of the lesion, as well as its grade, are difficult to

determine with precision. It is commonly known that biopsy results underestimate the extent

and grade of the cancer. Unfortunately many clinicians still lack confidence in imaging, despite

substantial progress being made in the field of MRI. Clinicians’ recommendation for treatment,

therefore, arises from a profound sense of uncertainty about the precise nature of the cancer

they are treating. After diagnosis, treatment decisions are hampered by further difficulties in

accurately staging the disease. In this atmosphere it is no wonder that treatment decisions are

so difficult for patient and their physicians. For the patient who chooses active rather than

deferred therapy, which treatment is best: radical prostatectomy, external beam irradiation,

brachytherapy or some combination? Not only do the treatments differ in timing of onset and

degree of side effects, but the likelihood of cancer control. The number of complications and

side effects depend as much on the specific technique employed as well as the expertise of the

treating physician on the method of therapy chosen.

22

Introduction 1

Introduction 1

AIM OF THESIS

Figure 7. PCa

Undoubtedly a substantial number of factors play a role in the chaotic nature of the Battle of

Anghiari. The author started his thesis with the confident hope of trying to unravel this battle

with one overriding consideration:

, identify the clinical problem, then identify why there is a problem, then find a solution to

solve this problem.

The author was fortunate to have been able to build on the important foundations of MR imaging

of the prostate, laid down by the valuable research done in Nijmegen by his PhD predecessors.

Only with this foundation, is the continuous work and developments highlighted in this thesis

possible.

Problems faced by Clinicians:

1. Patients with an elevated/elevating PSA value but persistent negative TRUS

biopsies, are of considerable concern. Does he have cancer or not? Should further

TRUS biopsies be performed or not?

2. If MRI is accurate in identifying a tumour location, what effective method is there

to reliably obtain histological proof of this location?

23

Introduction 1

Introduction 1

3. After radiation therapy for PCa, diagnosing local recurrence vs. metastatic disease

when the PSA starts rising again, is challenging.

4. Pretreatment identification of prostate cancer aggressiveness is crucial for

management and prognostication. The current methods to determine this are

inaccurate. What in vivo methods are available to reliably predict PCa nature?

5. When a patient is diagnosed with PCa Gleason Score 3+3=6 on biopsy, is there a

method available to reliably aid in differentiating those patients where biopsies

represent an undergrading (and therefore need more radical therapy) from those

where it is a correct prediction (and therefore may be managed more

conservatively)?

6. Transrectal ultrasound guided biopsies only reflect the true aggressiveness i.e.

Gleason grade in about 60% of patients. Are there any methods to improve the

tumour aggressiveness representativeness in biopsy samples on which further

management decisions can be based?

Problems faced by Radiologists:

7. No prostate looks alike. In particular the transition zone is a radiologically chaotic

region. What multi-parametric MR imaging features and techniques are available

8. What is normal? In one sue looks different from

tissue in a different

quantitative measurements and our assessment of what is malignant?

9. It is often mentioned that prostate multi-parametric MR imaging should be left to

the experts. The prostate is too complex, too many imaging modalities are needed

and tumours are very heterogeneous. Is there any help for the non-expert?

THE AIM of this thesis therefore is to target these specific problems faced by doctors and

patients and develop and validate methods in order to provide solutions for them.

24

Introduction 1

Introduction 1

OUTLINE OF THESIS

In contrast to PhD theses in other non-medical fields, the fragmented nature of writing a thesis

in a medical science is based on the fact that a number of chapters should be based on peer

reviewed articles. Therefore each chapter has a similar repetitive composition being outlined in

an introduction, materials and methods, results, discussion and conclusions. Many chapters

start of with a similar introduction and to the unacquainted, this might appear cumbersome and

excessively repetitive. This is unfortunately a drawback of medical PhD theses. Yet, each

chapter can be read as a small thesis in itself, with a sufficient overview to provide the reader

with insight into the clinical question addressed in that chapter. Most chapters therefore will be

read and understood separately instead of seeing the complete picture. Many pictures, diagrams

and images are provided throughout the thesis. Radiologists are undoubtedly visually

stimulated creatures who are notoriously easily bored by great amounts of text.

This thesis is spread in 6 principal parts. PART ONE includes Chapter 1 and 2 which provide an

Introduction and Background to this thesis. PART TWO deals with the detection of primary

and recurrent prostate cancer and includes Chapter 3-6. PART THREE consists of the

assessment of PCa aggressiveness and includes Chapter 7-10. PART FOUR deals with

computerized calibration of “normal” peripheral zone tissue for increased accuracy in

assessment of aggressiveness and this is presented in Chapter 11. Chapter 12 deals with

computer aided diagnosis. PART FIVE is the finale, dealing with a discussion, conclusion and

future perspective, being outlined in Chapter 13. An English and Dutch summary is given in

Chapter 14. PART SIX is the usual postlude including a list of publications, presentations and

prizes, words of gratitude and ending with a short curriculum vitae.

25

Introduction 1

Introduction 1

PART ONE INTRODUCTION and BACKGROUND

Chapter 1 provides the non acquainted reader with a broad introduction in the interesting field

of prostate cancer demographics, diagnostics and sets the basis for the problem-solution

orientated approach as is presented in this thesis.

Chapter 2 provides an introduction and basis of understanding for the advanced functional MR

imaging modalities that are tested and validated in this thesis. These modalities consist of

Dynamic Contrast Enhanced MRI (DCE-MRI) and Diffusion Weighted Imaging (DWI) with the

derived Apparent Diffusion Coefficient (ADC) maps.

PART TWO DETECTION of PRIMARY and RECURRENT PROSTATE CANCER

Chapter 3 describes the feasibility and method of using an MR compatible transrectal biopsy

device within a 3 Tesla MRI scanner, to obtain biopsies of tumour suspicious regions on multi-

parametric MR imaging.

Chapter 4 goes further and determines the value of MR guided biopsies on the yield of prostate

cancer in men with elevated PSA > 4 ng/ml and more than two prior negative TRUS guided

biopsy sessions. Furthermore, it also determines the location of tumours not found by

conventional biopsy techniques and the significance of the detected tumours.

Chapter 5 determines if DCE-MRI can be a useful technique to detect local recurrence of PCa

following external beam radiotherapy. Furthermore it evaluates if the MR guided biopsy

procedure is a useful technique for providing definite histological proof thereof.

Chapter 6 deals with transition zone cancers and evaluates the role of the individual anatomical

and functional MR imaging modalities to detect and localize low- vs. high-grade tumours.

PART THREE ASSESMENT OF PROSTATE CANCER AGGRESIVENESS

Chapter 7 evaluates the relationship between ADC values of tumour in the peripheral zone and

aggressiveness of prostate cancer and establishes a basis for a further study which prospectively

determines the Gleason grades prior to treatment.

Chapter 8 reports on the initial experience with identifying PCa undergrading using DWI

derived ADC values in patients wi TRUS-guided biopsy.

26

Introduction 1

Introduction 1

Chapter 9 determines the potential value of 1H-MRSI in the assessment of prostate cancer

aggressiveness.

Chapter 10 evaluates the value of using DWI combined with MR guided biopsies to

prospectively improve the pretreatment prediction of true prostate cancer aggressiveness.

These findings are then compared to the conventional 10-core TRUS biopsy scheme.

PART FOUR COMPUTER AIDED DIAGNOSIS OF PROSTATE CANCER

Chapter 11 builds on the findings reported in Chapter 7. The substantial variation in normal

peripheral zone ADC values is used for calibration. Re-assessment is done for this this mixed

model incorporating both normal peripheral zone as well as tumour ADC values for

improvement in tumour aggressiveness differentiation.

Chapter 12 shows the development of a computer aided diagnosis (CAD) technique using both

quantitative pharmacokinetic parameters derived from DCE-MRI combined with quantitative

ADC values from DWI to differentiate tumour from benign tissue (but with tumour suspicious

characteristics) with a high diagnostic accuracy. This CAD system is then tested on multiple

less-experienced and experienced readers in evaluation of prostate MRI and determines if the

less-experience reader can be aided to improve his/her assessment of prostate tumours.

PART FIVE - DISCUSSION, CONCLUSION, FUTURE PERSPECTIVES

Chapter 12 provides a detailed discussion, conclusion and considers the future perspectives.

Chapter 13 an English and Dutch Summary

PART SIX THE POSTLUDE

List of Publications

List of Presentations both Scientific and Invited

Awards

27

Introduction 1

Introduction 1

REFERENCES

1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J.Clin. 2012 Jan;62(1):10-29.

2. Scardino PT, Kelman J.K. Dr. Peter Scardino's Prostate Book. New York, Avery Press: 2005.

3. Delongchamps NB, Singh A, Haas GP. The role of prevalence in the diagnosis of prostate cancer. Cancer Control 2006 Jul;13(3):158-68.

4. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J.Clin. 2005 Mar;55(2):74-108.

5. Schroder FH, van der MP, Beemsterboer P, Kruger AB, Hoedemaeker R, Rietbergen J, Kranse R. Evaluation of the digital rectal examination as a screening test for prostate cancer. Rotterdam section of the European Randomized Study of Screening for Prostate Cancer. J.Natl.Cancer Inst. 1998 Dec 2;90(23):1817-23.

6. Presti JC, Jr., O'Dowd GJ, Miller MC, Mattu R, Veltri RW. Extended peripheral zone biopsy schemes increase cancer detection rates and minimize variance in prostate specific antigen and age related cancer rates: results of a community multi-practice study. J.Urol. 2003 Jan;169(1):125-9.

28

Introduction 1

CHAPTER 2

Background to Functional MR Imaging of the Prostate

T. Hambrock; C. Hoeks, R. Somford et al.

CHAPTER 2CHAPTER 2

Background to Functional MRI of the Prostate 2

The content of this chapter is principally based on three publications:

Dynamic contrast enhanced MR imaging in the diagnosis and management of prostate

cancer; Categorical Course in Diagnostic Radiology: Genitourinary Radiology 2006.

Hambrock T, Padhani A, Tofts P et al.

Diffusion and perfusion MR imaging of the prostate; Magnetic Resonance Imaging Clinics

of North America 2008. Somford R, Fütterer J, Hambrock T.

Prostate Cancer: Multiparametric MR imaging for Detection, Localization and Staging;

Radiology 2011. Hoeks C, Barentsz J, Hambrock T et al.

INTRODUCTION

The development of clinical utilization of MRI, culminating from multiple important

breakthroughs in quantum mechanics, is according to the author one of the most ingenious

developments of the 20th century. The basic principals underlying magnetic resonance imaging

are extremely complex (and extremely interesting) and it is definitely beyond the scope of this

thesis to provide a thorough introduction to physical principals underlying it. However, the

principals underlying more recent developments in functional MR imaging modalities incl.

Diffusion Weighted Imaging (DWI) and Dynamic Contrast Enhanced Imaging (DCE) will be

explained in more detail as these were the most important techniques evaluated in this thesis.

Additionally, a brief overview is given of the pathophysiological processes which underlie

imaging of the prostate. For a more detailed understanding of the physics underlying MRI, the

reader is referred to: MRI in Practice by Catherine Westbrook (1).

BASIC PRINCIPALS OF MAGNETIC RESONANCE IMAGING

Certain atoms are characterized by their tendency to align their axis of magnetic moment to an

external magnetic field. This happens because of their inherent angular moment or spin, as they

contain positively charged protons, that is, they possess electrical charge. The laws of

electromagnetic induction (as described originally by Faraday) refer to three individual forces

30



motion, magnetism and charge. The law of Faraday states that if two of these forces are present,

then the third is automatically induced. Atoms that have the properties of aligning along a

magnetic field are amenable to MR imaging. The most important one is hydrogen (1H), partially

because of its profound abundance in living matter, but also because of its large gyromagnetic

ratio. Certain isotopes of particular nuclei including carbon (13C), phosphorus (31P) and fluorine

(19F) are also amenable to MR imaging.

Figure 1. Random positioning of hydrogen atoms in the absence of a magnetic field (left)

and alignment against/with the main magnetic field (depending on their energy) (right).

When hydrogen atoms are aligned within the external magnetic field, a precession occurs

around the field with a specific frequency (for a 3T magnet this is 127 MHz). If an external

radiofrequency pulse is applied at this exact frequency, low energy hydrogen atoms aligned with

the magnetic field (y-axis) are flipped over causing a change in the net magnetization moment.

The net magnetization is now in the transverse plane (x-axis). After the RF pulse, the net

magnetization slowly returns to the y-axis. Recovery of longitudinal magnetization (y-axis) is

caused by nuclei giving up their energy to the surrounding environment. The rate of recovery is

an exponential process with a recovery time constant called the T1 time. This is the time it takes

63% of the longitudinal magnetization to recover. T2 decay of transverse magnetization is

caused by nuclei exchanging energy with neighboring nuclei. The rate of loss of coherent

transverse magnetization is also an exponential process, with the T2 relaxation time of a tissue,

the time it takes for 63% of the transverse magnetization to be lost. Using a receiver coil placed

31



over the patient this change in magnetization after the RF pulse can be measured. The T1 and

T2 time for different tissues and molecules differs

between tissues on MR imaging.

Figure 2. Schematic presentation of the sequence of events in an MRI scanner. After the

RF pulse on the hydrogen atoms, placed in a magnetic field, spin magnetization is flipped

into the transverse direction.

A clinical 1.5T MRI scanner T1-w image of the prostate T2-w image of the prostate

The T1-weighted (T1-w) and T2-weighted (T2-w) images are considered anatomical images and

the functional MR imaging modalities considered hereafter are DCE-MRI and DWI-MRI with the

exception of MR spectroscopic imaging, not included in this chapter.

32



A. DYNAMIC CONTRAST ENHANCED MRI

Pathophysiological basis - Angiogenesis and the prostate

For a prostate tumour, one critical factor that affects development, growth, invasiveness, and

progression into the metastatic form is the ability of the tumour to generate new blood vessels.

Angiogenesis, which we define as the sprouting of new capillaries from existing blood vessels,

and vasculogenesis, the de novo generation of new blood vessels, are the two primary methods of

vascular expansion by which nutrient supply to tumour tissue is adjusted to match physiologic

needs. Angiogenesis is an essential component of several normal physiologic processes,

including menstrual cycle changes in the ovaries and uterus, organ regeneration, wound healing,

and the spontaneous growth of collateral vessels in response to ischemia (2). Pathologic

angiogenesis is an integral part of a number of disease states, including rheumatoid disease, age-

related macular degeneration, proliferative retinopathy, and psoriasis, as well as being critical

for the growth and metastasis of malignant tumours (3).

A number of different mechanisms are involved, including vessel sprouting and bridge

formation. These processes depend on the migration and proliferation of endothelial cells.

Circulating endothelial progenitor cells derived from bone marrow are also recruited to sites of

active angiogenesis by tumour-derived growth factors such as vascular endothelial growth

factor (VEGF) (4). Tumour growth larger than 1 2 mm in diameter in solid tissues cannot occur

without vascular support (5). Tumour neovascularization often lags behind tumour growth,

leaving areas of low oxygen tension (hypoxia). The decrease in oxygen tension stimulates

further angiogenesis through various signaling pathways by the production of numerous

transcriptional factors, the most important being hypoxia-inducible factors (HIFs), especially

HIF-1 and HIF-2 (6). In the presence of hypoxia, HIF-1 binds to HIF-1 at the HIF response

elements (HREs); this is made possible because HIF-1 does not undergo hydroxylation and

subsequent degradation. Many of the genes activated by the HIF-HRE complex are beneficial to

tumour survival, including those involved in angiogenesis (VEGF), glucose metabolism (glucose

transporter 1), proliferation (insulin-like growth factor 2), and pH regulation (carbonic

anhydrase 9) (7) (Fig 1). Mediation of the physiologic and pathologic stimulation that causes a

33



change in cellular phenotype is enacted by a variety of pro-angiogenic factors, which include the

following: (a) VEGF, the most prominent of the angiogenic stimulators; (b) thymidine

phosphorylase, also known as platelet-derived endothelial growth factor; (c) matrix

metalloproteinases (MMPs), a multifarious family of proteolytic enzymes involved in the

breakdown of extracellular matrix; (d) carbonic anhydrase 9, an enzyme that catalyzes the rapid

conversion of carbon dioxide and water into carbonic acid, protons, and bicarbonate ions; and

(e) cyclooxygenase-2, a key enzyme in the prostaglandin biosynthesis pathway that converts

arachidonic acid to prostaglandin (7).

Figure 1. Cascade of gene activations with HIF, after hypoxic stimulation. Cascade results

in eventual angiogenesis to overcome the hypoxia. CA-9 = carbonic anhydrase 9, COX-2 =

cyclooxygenase-2, GLUT-1 = glucose transporter 1.

The importance of angiogenesis in prostate cancer is well established. Angiogenesis is an

integral part of benign prostatic hyperplasia, is associated with prostatic intraepithelial

neoplasia, and is a key factor in the growth and metastasis of prostate cancer (Fig 2). The results

of some studies have demonstrated a direct correlation of angiogenesis with Gleason score,

tumour stage, progression, metastasis, and survival (8,9). Angiogenesis is not directly associated

34



with serum PSA levels, which might reflect the ability of PSA to convert plasminogen to

angiostatin-like fragments, possibly contributing to the slow growth of prostate tumours.

Expression of angiogenic cytokines in prostate cancer might be induced as a response to hypoxic

stress or by hormonal stimulation but can also result from activation of oncogenes. The

angiogenic process in prostate cancer is highly dependent on VEGF. VEGF is produced in

abundance by the prostatic secretory epithelium of normal, hyperplastic, and tumour-containing

prostate glands. With respect to the vasculature, it is clear that VEGF is required for vascular

homeostasis in benign prostatic hyperplasia, and the overproduction of VEGF maintains a high

fraction of immature vessels (those without investing pericytes and/or smooth muscle cells) in

prostate cancers (10,11). In the prostate, production of VEGF requires continual stimulation by

androgens, and at androgen withdrawal, VEGF expression is down-regulated, and tumours

undergo vascular regression before tumour cell death (12). VEGF has a positive association with

microvessel density, tumour stage and grade, and disease-specific survival in patients with

prostate cancer (13). As noted earlier in this section, HIF-1 is a key mechanism for VEGF

regulation, and it is known that HIF-1 is up-regulated in the majority of prostate tumour tissues

and that its expression is induced in prostate cancer in situ (14).

Figure 2. Growth and metastasis of tumour with hypoxia-induced angiogenesis, mediated

by VEGF.

35



Role of DCE-MRI in visualizing angiogenesis in the prostate

A number of distinguishing features are characteristic of malignant vasculature, many of which

are amenable to study with dynamic contrast agent enhanced MR imaging methods. These

features include (a) spatial heterogeneity and chaotic structure: little hierarchy of vascular

structures is observed, with abrupt changes in diameter and blind-ending vessels, particularly

within the centers of tumours; and few structurally complete arteries or veins are found with

sinusoidal capillary plexuses prevailing; the remodeling of the vasculature seen in inflammation

or wound healing is largely missing; (b) poorly formed fragile vessels with high permeability to

macromolecules because of the presence of large endothelial cell gaps or fenestrae, incomplete

basement membrane, and relative lack of pericytes or smooth muscle association with

endothelial cells; (c) arteriovenous shunting, high vascular tortuosity, and vasodilatation; (d)

intermittent or unstable blood flow (with acutely collapsing vessels and areas of spontaneous

hemorrhage; and (e) extreme heterogeneity of vascular density, with areas of low vascular

density mixed with regions of high angiogenic activity. These features are distinct from the

organized regular structure and normal blood flow seen in mature vessels. Angiogenic vessels

are also leaky, a feature that aids extracellular matrix signaling and metabolism, as well as

contributing to tumour cell invasion and metastasis (10). These tumour-induced vascular and

structural abnormalities result in functional impairments that are important to dynamic

contrast-enhanced MR imaging observations, including the following:

1. The interstitial pressure is increased because of an increased vascular permeability and

poor lymphatic drainage. As a result, the interstitial space is enlarged (by as much as five

times), allowing low-molecular weight contrast agents to accumulate. The higher interstitial

pressure also leads to compression of vessels and thus increased vascular resistance and

regional areas of acute perfusion-related hypoxia.

2. The transcapillary permeability increases, allowing a more rapid exchange of low-

molecular-weight contrast agents. As a consequence, contrast agents are more easily able to

access the interstitial extracellular space and flow out when plasma levels drop. This can be

observed as an increase in MR signal intensity followed by a subsequent decrease.

36



3. The total vascular cross-sectional area may increase and can be combined with arterio-

venous shunts. This gives rise to increased blood flow overall. The global increase in flow in

cancers causes the bolus of contrast agent to arrive just a little earlier than it does in

surrounding normal tissue. In the prostate, differences in arrival time between normal and

abnormal tissue are short (differences of only 1 second have been observed). It is important

to remember that all of these functional changes do not necessarily occur homogeneously

throughout the tumour but most often are heterogeneously distributed, and they need not

coincide spatially. Thus, areas of increased interstitial volume may occur separately (at

different locations) from areas of increased permeability.

Fast T1-weighted dynamic contrast-enhanced MR imaging for monitoring the uptake of an

intravascular contrast agent has proved itself to be a powerful technique for studying the

characteristics of the microvasculature of prostate tumours and normal prostatic tissues. The

essence of fast prostate T1- weighted dynamic contrast-enhanced MR imaging lies in the

differences in microvascular characteristics that have been observed between normal and

malignant prostatic tissues. Differences in the enhancement pattern observed in the prostate are

due to three physiologic processes in the microvasculature:

a) perfusion, or blood flow; the higher the perfusion, the quicker the contrast agent will be

available for diffusion into the extravascular extracellular space;

b) capillary permeability; the higher the permeability and the greater the microvessel

surface area, the faster the transfer of contrast agent to the extravascular extracellular

space and the greater the rate of T1-weighted enhancement; and

c) cellular density; the higher the cellular density outside the vasculature, the less free

interstitial fluid is available for relaxivity changes induced by the gadolinium-based

contrast agent (i.e., reduced extravascular extracellular space).

37



Permeability, or leakiness, of capillaries refers to the ability of molecules to pass through

interendothelial fenestrae and junctions into the interstitial compartment. Note that most

normal tissues are leaky to micromolecules (the exception being the brain because of the blood-

brain barrier), but macromolecular permeability is specific for tumours; that is, high

permeability of the vasculature is a characteristic of pathologic blood vessels in inflamed tissues

and tumours. It is because both benign and malignant tissues are leaky to low-molecular-weight

contrast agents that simple pre- and post-contrast images are usually ineffective in detecting the

intraprostatic location of prostate cancer; only minimal differences between benign tissue and

prostate cancer are seen unlike the case in the brain, which, as stated previously, has an

intrinsic low vascular permeability. Imaging performed for a few minutes after administration of

contrast agent has been described as a way to detect breast lesions (19); however, in the

prostate, nearly all tissues tend to enhance similarly on these images (20). It is for these reasons

that dynamic sequences acquired at high temporal resolution by exploiting differences in

perfusion are currently the only way of differentiating prostatic tissues.

Principals underlying DCE-MRI

Dynamic contrast-enhanced MR imaging with the routinely available low-molecular-weight

gadolinium chelates enables noninvasive imaging of tissue functional vascular features. The

three essential aspects of dynamic contrast-enhanced MR imaging include:

a) Fast dynamic imaging, referring to the temporal (time) component in imaging; complete

coverage of the anatomic area with a fast T1-weighted sequence is required before and

after the bolus injection of a lowmolecular-weight contrast agent;

b) Contrast agent administration, that is, intravenous administration of a low-molecular-

weight, usually gadolinium-based contrast agent; increases in signal intensity are seen

on the dynamically acquired T1-weighted MR images; and

38



c) Quantification of signal intensity changes, that is, semiquantitative and quantitative

estimation of signal intensity changes to determine the kinetic parameters of the

contrast agent.

Depending on the technique used, data can be obtained reflecting the tissue perfusion (blood

flow, blood volume, and mean transit time), the microvessel permeability surface area product,

and the extracellular leakage space. Insights into these physiologic processes can be obtained by

the evaluation of kinetic enhancement curves or by the application of complex compartmental

modeling techniques. In addition to the signal intensity increases observed with T1-weighted

MR sequences, it is possible to observe the effects of the contrast agent while still confined to the

early vascular phase. While in the vascular space, concentrated contrast agent produces focal

magnetic field inhomogeneities that result in a decrease in the signal intensity of the

surrounding tissues (T2* effect). Thus, MR sequences can be designed to be:

a) Sensitive to the vascular phase of contrast agent delivery (the so-called T2*-weighted or

susceptibility-based methods), which reflect tissue perfusion and blood volume; or

b) Sensitive to the presence of contrast agent in the extravascular space (also termed T1-

weighted or relaxivity-based methods) and reflecting the perfused microvessel area and

permeability, as well as the extravascular extracellular leakage space.

The choice of the dynamic contrast-enhanced MR imaging sequences and parameters to be used

will depend on the required anatomic coverage, the acquisition times, the susceptibility to

artifacts resulting from magnetic field variations, and the need for quantification (10).

Analysis of the tissue signal intensity or the uptake of gadolinium-based contrast agent can be

done semiquantitatively (eg, with the onset time, the maximum enhancement, or the time to

peak) or with more complicated but quantitative pharmacokinetic modeling approaches. The

latter methods quantify enhancement with parameters like the transfer constant (Ktrans), the

volume of interstitial extravascular extracellular space (Ve), and the rate constant (kep) (15). The

quantification of kinetic parameters has the advantages of being biologically meaningful, helping

39



to establish objective criteria for classifying tissues, and being able to be used to objectively

assess the response to therapy (9,16).

A relationship exists between the uptake rate for gadolinium-based contrast agent and the

surface area of perfused microvessels. Histopathologic examination can only show microvessel

density and does not provide information with regard to the functionality (perfusion) of the

microvessels. It is important to note that implanted tumour xenograft data show that there is a

discrepancy between perfused and visible microvessels at histologic examination. The perfusion

of microvessels shows a variation from 20% to 85% at any given time (17,18). Dynamic MR

imaging can therefore provide additional information on tumour neovascularity as well as the

perfused fraction of vessels. Two aspects of dynamic MR imaging are of extra importance:

contrast agents and microvasculature of the prostate. These two will be discussed in the

following sections.

Contrast Agents

A number of different groups of contrast agents could be used for assessment of the angiogenic

status in tumours. These groups include (a) low-molecularweight agents (<1000 Da) that rapidly

diffuse into the extracellular fluid space (extracellular fluid agents), such as gadoterate

meglumine and gadopentetate dimeglumine; (b) intermediate-molecular- weight agents; (c)

high-molecular-weight agents (>30 000 Da) designed for prolonged intravascular retention

(macromolecular contrast agents or blood pool agents), such as gadofosveset trisodium, which

itself is of low molecular weight but binds rapidly to plasma albumin and so effectively behaves

like a macromolecular contrast agent; however, there is a small fraction that remains unbound,

particularly in the first 1 minute after contrast agent administration; and (d) agents intended to

accumulate at sites of concentrated angiogenesis nanoparticulate gadoliniumcontaining

liposomes. In the United States, for dynamic contrast-enhanced MR imaging, only low-

molecular-weight gadolinium chelate contrast agents are currently approved. They shorten the

longitudinal (spin lattice) T1 relaxation of protons, resulting in increased signal intensity on T1-

weighted MR images. The increase in signal intensity is dependent on the native T1 relaxation of

tissue, the dose of the contrast agent, the imaging sequence and parameters used, and the gain

and scaling factors of the MR imaging equipment. These agents are unable to cross cell

40



membranes and thus will stay in the intravascular extracellular space (blood plasma) or the

extravascular extracellular space (interstitial fluid space). Note that although gadolinium

chelates affect protons in their immediate vicinity, proton diffusion occurs sufficiently quickly

for their sphere of influence to extend to the intracellular compartment. Thus, although

gadolinium chelates cannot enter intact cells, they can and do affect the proton relaxation in

cells. In Europe, a wider range of agents has recently become available: superparamagnetic iron

oxide agents like ferumoxides, which might be used in dynamic susceptibility-weighted MR

imaging, as well as the first clinically available blood pool contrast agent, gadofosveset

trisodium. The exact role of these contrast agents in oncologic imaging still needs to be defined.

Figure 3 shows a work-flow diagram for the technique of DCE-MR imaging.

Figure 3. Work-flow diagram to show the technique of DCE-MR imaging, from intravenous

(IV) administration of contrast agent to the generation of colored images.

41



Contrast Agent Administration

For optimal qualitative and quantitative estimation of dynamic contrast agent related changes

in prostatic tissue, controlled administration of a bolus of contrast agent into a peripheral vein is

required. Manual administration can result in distorted enhancement characteristics. To

minimize this problem, automatic power injectors should be used with fixed administration

rates (usually 2.5 mL/sec although a higher rate of. 4-5 ml/sec seems more advantageous). After

injection of the bolus of gadolinium-based contrast agent, a normal saline flush is also needed to

clear the line and to chase the injected bolus of contrast agent into the central circulation.

Dynamic Sequences

T1-weighted sequences. The T1-weighted signal intensity increase in tissue (Fig. 4) is

dependent on the baseline T1 value. In general, quantification is improved by estimating

changes in the T1 relaxation rate at each time point during the dynamic acquisition. T1-weighted

sequences, usually of gradient-echo or saturation-recovery/inversion-recovery snapshot types,

are used for data acquisition. High spatial resolution to cover the whole prostate can only be

achieved by compromising the temporal resolution and vice versa.

42



Figure 4. Graphs of signal intensity versus time for a T1-weighted dynamic contrast-

enhanced MR imaging acquisition. Different semiquantitative parameters are calculated

from the graph after curve-fitting algorithms have been applied.

As a result, two types of schemes having different temporal resolutions are used when

performing dynamic contrast-enhanced MR imaging of the prostate: a) slow sequences

(temporal resolution, approximately 30 seconds) with high spatial resolution; these in general

have high sensitivity and low specificity; and b) Fast sequences (imaging techniques with

temporal resolution of 1 4 seconds) with lower spatial resolution; these have low sensitivity

and high specificity.

The optimal temporal resolution and spatial resolution still need to be established to achieve the

highest sensitivity and specificity, and this will depend on the clinical question. To date, most

researchers have used strategies of high temporal resolution, but it seems that cancer might be

accurately depicted, at least in the peripheral zone, by using slower sequences (21). The great

advantage of higher temporal resolution, compared with low temporal resolution, is the ability

to accurately quantify enhancement parameters and gain valuable pharmacokinetic information.

Although most studies emphasize high temporal resolution at the expense of spatial resolution,

lower spatial resolution may not depict critical features needed for optimal staging (e.g., minimal

capsular penetration). Despite this, studies with high temporal resolution have shown that

43



dynamic contrast-enhanced MR imaging can improve the staging capabilities of less-experienced

radiologists (22).

Figure 5. Graphs of signal intensity versus time showing the difference between fast (left)

and slow (right) acquisition methods.

Data Processing

T1-weighted sequence data. From the raw data acquired with the T1-weighted sequence, a

pixel-by pixel analysis of signal intensity changes is made. Signal enhancement seen on T1-

weighted dynamic contrast-enhanced MR images can be assessed in two ways:

a) Semiquantitative analysis of signal intensity changes and

b) Quantitative analysis of contrast agent concentration (change in relaxivity) by using

pharmacokinetic modeling techniques.

Semiquantitative parameters describe signal intensity changes by using a number of descriptors.

These parameters include curve shape, onset time (t 0 = time from injection or appearance in an

44



artery to the arrival of contrast agent in the tissue of interest), gradient of the slope of

enhancement curves, maximum signal intensity, area under the signal intensity curve at a fixed

time point (usually 60 90 seconds after onset time), and washout gradient (late washout). These

parameters have the advantage of being relatively straightforward for calculation, but they are

limited by the fact that they are not biologically meaningful, may not accurately reflect contrast

agent concentration in tissues, and can be influenced by the imaging equipment s settings

(including gain and scaling factors). These factors limit the usefulness of semiquantitative

parameters and make between-patient and between-system comparisons difficult.

Quantitative techniques use pharmacokinetic modeling, which is usually applied to changes in

the contrast agent concentrations in tissue. Signal intensity changes observed during dynamic

acquisition are used to estimate contrast agent concentration in vivo (23). Concentration-time

curves are then mathematically fitted by using one of a number of recognized pharmacokinetic

models, and quantitative kinetic parameters are derived. Examples of modeling parameters

include the volume transfer constant of the contrast agent (Ktrans [formally called the

permeability surface area product per unit volume of tissue], measured in units per minute), the

interstitial fluid space as a percentage of unit volume of tissue (Ve), and the rate constant (kep,

measured in units per minute). These standard parameters are related mathematically (24): kep

= Ktrans/Ve (1). Quantitative parameters are more complicated to derive than those derived

semiquantitatively, which deters their use. However, commercially available software is

beginning to appear, and if contrast agent concentration can be measured accurately and if the

type, volume, and method of administration of contrast agent are consistent, then it is possible

to directly compare pharmacokinetic parameters acquired serially in a given patient and in

different patients imaged at the same or different imaging sites (25). Uncertainties exist with

regard to the reliability of kinetic parameter estimates derived from the application of contrast

agent kinetic models to T1-weighted dynamic contrast-enhanced MR imaging data. These

uncertainties derive from assumptions implicit in kinetic models and those assumptions made

for the measurement of the contrast agent concentration in tissue. The vascular input function

used in the calculations also affects the reliability of the data obtained; robust methods for

measuring arterial input function for routine dynamic contrast-enhanced MR imaging studies

are currently emerging but are still not widely available (26 28).

45



Figure 6. Body compartments accessed by low-molecular-weight contrast agents. IV =

intravenous.

Quantitative Dynamic Contrast-enhanced Parameters

1. Extravascular Extracellular Space Volume (Ve)

The volume of extravascular extracellular space (Ve) is defined as:

where [Cgd]plateau_prostate is the prostate gadolinium concentration at plateau of peak

enhancement (i.e., the signal amplitude at which the exponential curve levels off), and

[Cgd]plateau_ref_tissue is the gadolinium concentration at plateau of peak enhancement in the

reference tissue used for calibration purposes. Ve refers to the space into which gadolinium

can leak from a capillary and has the benefit of specifically excluding the vascular space.

There may be regions (such as fibrous tissue) that are in the extravascular extracellular space

and yet are inaccessible to gadolinium-based contrast agents. Alternatives terms would

therefore be leakage space or distribution space. This is a theoretical parameter, though; and

in practice, in leaky tumours the contribution of plasma contrast agent and interstitial

46



contrast agent cannot be discriminated. Thus, Ve in practice measures the total extravascular

extracellular volume and therefore, 1 - Ve represents the cellular fraction.

2. Rate Constant (kep)

The rate constant (kep) is defined as:

where ttpprostate is the time to peak enhancement in the prostate, and ttpref_tissue is the time to

peak enhancement in the reference tissue. The rate constant kep is formally the diffusion rate

constant between the extravascular extracellular space and blood plasma. Both the volume

transfer constant and the rate constant have the same units (units per minute). kep Is always

greater than the transfer constant Ktrans. For a range of typical extravascular extracellular

space fractional volumes seen in tumours (Ve = 20% 50%), kep is two to five times higher

than Ktrans. The kep is the exponential decay constant for tissue concentration that would result

if the arterial concentration could be (a) instantaneously raised from zero to a constant value

or (b) dropped to zero. The kep is also the mean residence time for contrast agent in the

extravascular extracellular space after a bolus arterial input (24).

3. Volume Transfer Constant (Ktrans)

The volume transfer constant Ktrans is defined as follows:

47



Ktrans has several physiologic interpretations, depending on the balance between capillary

permeability and blood flow in the tissue of interest. In high-permeability situations, where

diffusion through the interendothelial fenestrae is limited by flow, Ktrans is equal to the blood

plasma flow per unit volume of tissue. In the other limiting case of low permeability, where

contrast agent diffusion is limited by permeability, Ktrans is equal to the permeability surface

area product of the capillary vessel walls, per unit volume of tissue (24).

Limitations of the DCE-MERI technique

It should be evident that dynamic contrast-enhanced MR imaging combined with high-spatial

resolution T2-weighted imaging and Diffusion weighted imaging will remain the mainstay of

prostate cancer MR imaging for the foreseeable future. However, the limitations of dynamic

contrast-enhanced MR imaging should be borne in mind. The transition zone, often replaced by

benign prostatic hyperplasia, can be highly vascularized and show rapid and high levels of

enhancement. As noted previously, discriminating normal transition zone and benign prostatic

hyperplasia from tumours within the same region is often challenging. Pathologic but

nonmalignant lesions within the prostate can often also mimic tumour on dynamic contrast-

enhanced MR images. The most common of these lesions are high-grade prostatic intraepithelial

neoplasia and prostatitis; the underlying reasons for the overlap with tumour lies in the fact that

these lesions also incite angiogenic responses in tissues. Administration of a contrast agent is an

invasive procedure with additional costs and potential side effects. For quantitative dynamic

contrast-enhanced MR imaging to be widely applied in clinical practice, it is necessary to

develop standardized robust analytic approaches for the measurement of enhancement. This

includes the need for commercial equipment manufacturers to provide robust methods for

rapidly measuring time-varying change in T1 relaxation rates, incorporation of arterial input

function into kinetic modeling processes (or other reliable methods that substitute for arterial

input function measurement), and robust analytic software that allows input from the different

MR imagers (17). Finally, interpretation requires a certain level of experience because no

quantitative parameter is able to be used to reliably separate tumour from benign tissues. In

conclusion, dynamic contrast-enhanced MR imaging has established itself as a valuable imaging

tool with a wide variety of applications for patients with prostate cancer.

48



Figure 7. Dynamic contrast-enhanced MR imaging in localization of prostate cancer. (a)

Histologic determination of tumour areas. (b) T2-weighted MR image. Arrows indicate

tumour. (c e) Semiquantitative parameters. (c) Wash-in rate. (d) Late washout. (e)

Relative enhancement. Quantitative parameters: (f-h). (f) Rate constant (kep). (g) Leakage

space (Ve). (h) Volume transfer constant (Ktrans). (i) Graph of signal intensity versus time

shows the difference between enhancement characteristics of tumour and normal

peripheral zone (PZ).

49



B. DIFFUSION WEIGHTED IMAGING (DWI)

Pathophysiological basis and the role of DWI in depicting prostatic tissue

Water molecules exhibit random motion in tissue, related to temperature (Brownian effect)(29).

The intra- and extracellular movement of molecules in tissue is largely restricted by membranes

forming barriers to diffusion., The more barriers water molecules meet in a certain time interval,

the smaller the mean movement (diffusion) distance (32). The degree of restriction to water

diffusion in biological tissue is inversely correlated to tissue cellularity and the integrity of cell

membranes. Free motion of water molecules is more restricted in tissues with a high cellular

density. DWI can quantify this water motion in an indirect manner (30,31). The DWI pulse

sequence labels hydrogen nuclei in space, of which most is water molecules at any moment, and

determines the length of the path that water molecules travel over a short period of time

(labeling time in the order of 50 ms). DWI estimates the mean distance traveled by all hydrogen

nuclei in every voxel of imaged tissue. The greater this mean distance the higher the apparent

mobility of the water molecules in the tissue. In the clinical setting, diffusion-weighted

sequences are sensitized to detect diffusion distances ranging from 1 to 20 m predominantly

measuring microcapillary water movement (5% of total volume of voxel), intracellular and

extracellular space diffusion. From the DWI images quantitative values can be calculated, called

the apparent diffusion coefficients (ADC) with high values indicating free water movement and

low ADC values indicating restrictions to free movement.

DWI was initially used for the early detection of cerebral ischemia (36). The evolution of DWI

characteristics in cerebral ischemia over time has classically been attributed to the extracellular

to intracellular distribution of hydrogen nuclei caused by different types of edema (37). It has

been postulated that extracellular water molecules have a much higher range of mobility,

because they are not bound within membranes or by other cellular structures (38,39). When

this is translated to prostate tissue, which is predominantly glandular tissue, the predominant

contribution of the extracellular component is from tubular structures and their fluid content,

whereas the intracellular component is determined by the epithelial and stromal cells.

50



A prerequisite for the correct interpretation of diffusion and ADC images relies on good

knowledge of the diffusion characteristics of the different anatomic zones of the prostate and of

benign prostatic conditions compared with prostate cancer (40). The normal prostatic gland is

rich in tubular structures. This allows for abundant self-diffusion of water molecules within

these structure and provides high ADC values. In most cases, the peripheral zone can be easily

discriminated from the transition zone on DWI, because it displays relatively higher ADC values

(41-43). The exact background of this phenomenon remains unclear, because the exact ratio of

extracellular to intracellular components for the different anatomic zones of the prostate has not

yet been described. Moreover, exchange of water over membranes can obscure a fully

compartmentalized interpretation of the diffusion characteristics. The transition zone by

microscopic observation consists of more compact smooth muscle and sparser glandular

elements than the peripheral zone, leading to a lower extracellular to intracellular fluid ratio

(44). Furthermore, an age-related increase of T2 signal intensity of the peripheral zone

compared with the transition zone has also been demonstrated (45) and an age-related increase

in ADC values in both transition zone and peripheral zone has been seen (46), which are most

likely caused by atrophy in the prostate leading to reduced cell volume and enlarged glandular

ducts. Benign prostatic hyperplasia (BPH) gives rise to nodular adenomas in the transition zone

and with time these compress the central zone to form a pseudocapsule, consequently occupying

the complete transition zone. The peripheral zone is usually not affected by BPH and retains its

own histologic characteristics. BPH is defined by hyperplasia of all cells that constitute the

transition zone, with glandular, muscular, and fibrous compartments involved in various

degrees within a patient and between patients. This nodular hyperplasia gives rise to

inhomogeneous diffusion patterns and because tubular structures often remain in place, the

increased cellular density of hyperplasia, which is far less predominant than in prostate

carcinoma, might explain the observed reduction in ADC levels of the transition zone on DWI.

However since BPH has inhomogeneous diffusion characteristics, not only reduction in ADC but

also increases in ADCs have also been described (46).

Prostatitis almost uniquely originates in the peripheral zone. With respect to MR imaging,

chronic prostatitis is of far more importance than the acute prostatitis counterpart. Chronic

prostatitis is asymptomatic in many cases and symptoms may mimic BPH. Furthermore, both

are often associated with elevated prostate-specific antigen levels, raising the suspicion of

51



prostate cancer. Histologically, chronic prostatitis is characterized by extracellular edema

surrounding the involved prostatic cells with concomitant aggregation of lymphocytes, plasma

cells, macrophages, and neutrophils in the prostatic stroma. This abundance in cells as compared

with normal prostatic tissue may lead to an ADC decrease because of increased cellular density

and therefore restriction free water motion

on the DWI characteristics of chronic prostatitis.

Prostate carcinoma is histologically characterized by a higher cellular density than normal

prostate tissue, with replacement of the normal glandular tissue. This leads to a decrease in ADC

values, compared with normal prostate gland (Fig. 1) (40,47). Concomitantly with destruction of

tubular structures in prostate carcinoma, fractional anisotropy is also reduced (48,49).

Interestingly, whereas well-differentiated prostate carcinomas display some tubular formation,

with worsening differentiation the tubular structures become less predominant, and the cellular

component of the cancer increases.

Figure 8. Schematic presentation of the increased diffusivity of molecules in low-cellular

tissue (left) and severe restriction in highly cellular tissue (right).

52



Figure 9. The left shows a histological slide of the glandular composition of normal

peripheral zone tissue with high diffusivity. The histological images on the right

represent a highly cellular Gleason grade 5 prostate adenocarcinoma.

Principals underlying DWI

The sensitivity of the DWI sequence to water motion is introduced by diffusion sensitizing

gradients in which a certain combination of amplitude, duration, and spacing in time is

expressed in a b value. The relationship between signal intensity and b value encompasses a

continuous spectrum ranging from fast signal decay due to flowing water molecules in

microcappillaries, intermediate signal decay due to freely diffusing water down to slow signal

decay due to many restrictions. The number of b-values and the signal-to-noise of the

measurements define whether it is feasible to fit anything more than a mono-exponential decay

(50) curve to this data. The slope of the decay curve at low b-values quantifies a fast apparent

diffusion coefficient (ADCfast) reflecting distances traveled by protons in microcapillaries. At

higher b-values, the slope of the signal decay curve quantifies the slow component of the

apparent diffusion coefficient (ADCslow), reflecting distances traveled by protons in the 1 10 m

range in the extracellular space. In theory, very high b-values can be used to interrogate the

short diffusion distances traveled by intracellular protons as they encounter intracellular

membranes, but in practice the low signal-to-noise ratio (SNR) from these very high b-value

images makes the data unreliable (51).

53



Figure 10. (a) Graph illustrating the signal intensity versus b-values at diffusion-weighted

imaging (DWI) of a model of two compartments: tissue with normal versus restricted

diffusion. (b) Graph illustrates the logarithm of signal intensity versus b values at

diffusion-weighted imaging of normal peripheral zone (PZ) prostate tissue versus

prostate tumour. The signal of water molecules decays exponentially with increasing b-

values for different tissue types. The decay in signal is reduced in tissues with restricted

diffusion (e.g., tumour). The ADC represents the slope (gradient) of the plotted lines

(logarithmic conversion of the exponential decay). The greater the number of b- values

used in the analysis, the more accurate the ADC calculation.

In a large volume of pure water, self-diffusion is equal in all directions, hence isotropic, and not

restricted by any barrier. In organized tissue water mobility can have a preferred direction: if

water molecules experience less restrictions in one dimension their mobility can appear

anisotropic. The fractional anisotropy is determined along the axis of the tubular structures of

normal prostate tissue, and can potentially be used for tissue characterization.. Because

diffusion in tissue is limited by cellular structures, to establish a reliable estimate of this mean

distance traveled by hydrogen nuclei, DWI is acquired in at least three different orthogonal

directions for each b-value (32,33). In linearly aligned tissue this anisotropy is more pronounced

because there is one direction that contributes most to the DWI. Diffusion tensor imaging is a

specific technique that quantifies the level of anisotropy in tissue, expressed in a fractional

anisotropy value. This is low in imaged tissue without substantial anisotropy and is higher in

54



imaged tissue in which the larger part of diffusion takes place in one direction (33,34). Diffusion

tensor imaging can be used in addition to DWI to determine the structural organization of tissue

along which diffusion takes place.

Images from DWI typically have both T2-weighted and diffusion-weighted characteristics. The

intensity of the signal on the diffusion-weighted image represents a combination of signal from

the inherent T2 relaxation of the tissue as well as the dephasing caused by water motion in the

presence of the diffusion gradients. At low b-values there is greater contribution from the T2-

weighted signal, and at higher b-values, contrast is determined more by the relative diffusion of

molecules. When a high b-value diffusion weighted image has high signal intensities, this can be

due to two phenomena: a) the brightness (signal intensity) is due to an inherent long T2

relaxation of the tissue, referred to as the - or; b) the tissue of interest

portrays clear restriction in the Brownian proton movement due to increased cellularity,

macromolecules etc. Therefore ADC maps should also be obtained in every instance to

differentiate these two effects. The current guidelines advocated at least three different b-values

to obtain accurate ADC estimates using a low b-value, between 50-100 s/mm2, and two higher b-

values around 400-500 and 800-1000 s/mm2. Because vascular microperfusion can contaminate

the signal attenuation in DWI acquisition, it is essential not to include b-values < 50 s/mm2,

except when more complex bi-exponential fitting of the signal data is performed. To minimize

the influence of bulk motion as a distorting factor and minimizing T2 shine-through, typically a

TE as short as possible is chosen in addition to performing parallel imaging acquisition

techniques.

Limitations of Diffusion-Weighted MR Imaging

One of the main drawbacks of DWI of the prostate is its suboptimal spatial resolution, even with

currently widely available 3-T MR imaging systems, combining pelvic phased array surface coil

in combination with an endorectal coil for signal reception. The authors

improved spatial resolution with the use of DWI at 3 T improves zonal and tumour delineation

and allows improved ability to compare ADC mapping with whole-mount sectioned

prostatectomy specimens for research purposes. It has been shown that use of an endorectal coil

significantly improves imaging quality in T2-weighted imaging. Rectal gas in the absence of an

endorectal coil may lead to susceptibility artifacts (46). The endorectal coil enables better

55



staging performance and improves sensitivity for the localization of prostate carcinoma with

conventional MR imaging (52, 53). In the experience, the use of an endorectal coil in

conjunction with surface coils and parallel imaging improves image quality of DWI. This may

result in improved overall performance of DWI in the localization, characterization, and

delineation of prostate carcinoma. A further drawback of DWI is that it is very susceptible to

motion artifact resulting in distorted inaccurate ADC calculation. To some degree this can be

overcome by using a combination of surface and endorectal coil, which facilitate shortened

imaging time and allow using a lower echo time (TE).

Figure 11. A 68-year-old man with prostate cancer of the right peripheral zone. (A) Axial

T2-weighted MR image shows a low signal intensity area in the right peripheral zone.

Color parametric maps were calculated (B) and demonstrated increased washout in the

right peripheral zone, (C and D) increased Ktrans and kep. (E) ADC map at the same level

as in image A shows reduced ADC compared with the normal peripheral zone. (F)

Histopathology confirmed these findings and showed a tumour with Gleason Score of

4+3=7.

56



REFERENCES

1. Westbrook C. MRI in Practice. 4ed. John Wiley and Sons; 2011.

2. Conway EM, Collen D, Carmeliet P. Molecular mechanisms of blood vessel growth. Cardiovasc Res 2001;49(3):507 521.

3. Folkman J. Angiogenesis in cancer, vascular, rheumatoid and other disease. Nat Med 1995;1(1):27 31.

4. Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature 2000;407:249 257.

5. Folkman J. New perspectives in clinical oncology from angiogenesis research. Eur J Cancer 1996;32A(14):25342539.

6. Harris AL. Hypoxia: a key regulatory factor in tumour growth. Nat Rev Cancer 2002;2:38 47. 7. Charlesworth PJ, Harris AL. Mechanisms of disease: angiogenesis in urologic malignancies. Nat Clin Pract Urol

2006;3(3):157 169.

8. Borre M, Offersen BV, Nerstrom B, Overgaard J. Microvessel density predicts survival in prostate cancer patients subjected to watchful waiting. Br J Cancer 1998;78:940 944.

9. Bono AV, Celato N, Cova V, Salvadore M, Chinetti S, Novario R. Microvessel density in prostate carcinoma. Prostate Cancer Prostatic Dis 2002;5(2):123 127.

10. Padhani AR, Harvey CJ, Cosgrove DO. Angiogenesis imaging in the management of prostate cancer. Nat Clin Pract Urol 2005;2(12):596 607.

11. Jackson MW, Bentel JM, Tilley WD. Vascular endothelial growth factor (VEGF) expression in prostate cancer and benign prostatic hyperplasia. J Urol 1997;157:2323 2328.

12. Jain RK, Safabakhsh N, Sckell A, et al. Endothelial cell death, angiogenesis, and microvascular function after castration in an androgen-dependent tumour: role of vascular endothelial growth factor. Proc Natl Acad Sci U S A 1998;95(18):10820 10825.

13. Borre M, Nerstrom B, Overgaard J. Association between immunohistochemical expression of vascular endothelial growthfactor (VEGF), VEGF-expressing neuroendocrine-differentiated tumour cells, and outcome in prostate cancer patients subjected to watchful waiting. Clin Cancer Res 2000;6:1882 1890.

14. Du Z, Fujiyama C, Chen Y, Masaki Z. Expression of hypoxiainducible factor 1alpha in human normal, benign, and malignant prostate tissue. Chin Med J (Engl) 2003;116(2):1936 1939.

15. Leach MO, Brindle KM, Evelhoch JL, et al. The assessment of antiangiogenic and antivascular therapies in early-stage clinical trials using magnetic resonance imaging: issues and recommendations. Br J Cancer 2005;92:15991610.

16. van der Sanden BP, Rozijn TH, Rijken PF, et al. Noninvasive assessment of the functional neovasculature in 9L-glioma growing in rat brain by dynamic 1H magnetic resonance imaging of gadolinium uptake. J Cereb Blood Flow Metab 2000;20(5):861 870.

17. Collins DJ, Padhani AR. Dynamic magnetic resonance imaging of tumour perfusion. IEEE Eng Med Biol Mag 2004;23(5):65 83.

18. Endrich B, Vaupel P. The role of microcirculation in the treatment of malignant tumours: facts and fiction. Blood Perfusion Microenviron Hum Tumours 1998;3:19 39.

19. Boetes C, Barentsz JO, Mus RD, et al. MR characterization of suspicious breast lesions with a gadolinium-enhanced TurboFLASH subtraction technique. Radiology 1994;193(3):777 781.

20. Huisman HJ, Engelbrecht MR, Barentsz JO. Accurate estimation of pharmacokinetic contrast-enhanced dynamic MRI parameters of the prostate. J Magn Reson Imaging 2001;13(4):607 614.

21. Rouviere O, Valette O, Grivolat S, et al. Recurrent prostate cancer after external beam radiotherapy: value of contrast-enhanced dynamic MRI in localizing intraprostatic tumour correlation with biopsy findings. Urology 2004;63(5):922 927.

22. Fütterer JJ, Engelbrecht MR, Huisman HJ, et al. Staging prostate cancer with dynamic contrast-enhanced endorectal MR imaging prior to radical prostatectomy: experienced versusess experienced readers. Radiology 2005;237(3):541 549.

23. Parker GJ, Suckling J, Tanner SF, et al. Probing tumour microvascularity by measurement, analysis and display of contrast agent uptake kinetics. J Magn Reson Imaging 1997;7:564 574.

57



24. Tofts PS, Brix G, Buckley DL, et al. Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J Magn Reson Imaging 1999;10(3):223 232.

25. Padhani AR, Leach MO. Antivascular cancer treatments: functional assessments by dynamic contrast-enhanced magnetic resonance imaging. Abdom Imaging 2005;30:324 341.

26. Rijpkema M, Kaanders JH, Joosten FB, van der Kogel AJ,Heerschap A. Method for quantitative mapping of dynamic MRI contrast agent uptake in human tumors. J Magn Reson Imaging 2001;14:457 463.

27. Port RE, Knopp MV, Brix G. Dynamic contrast-enhanced MRI using Gd-DTPA: interindividual variability of the arterial input function and consequences for the assessment of kinetics in tumors. Magn Reson Med 2001;45:1030 1038.

28. Hara N, Okuizumi M, Koike H, Kawaguchi M, Bilim V. Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is a useful modality for the precise detection and staging of early prostate cancer. Prostate 2005;62(2):140 147.

29. Crank J. The mathematics of diffusion. New York: Oxford University Press; 1956.

30. Stejskal EO, Tanner JE. Spin diffusion measurements: spin echoes in the presence of a time-dependentfieldgradient. J Chem Phys 1965;42:288 92.

31. Basser PJ, Mattiello J, LeBihan D. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magn Reson 1994;103:247 54.

32. Bammer R, Skare S, Newbould R, et al. Foundations of advanced magnetic resonance imaging. NeuroRx 2005;2:167 95.

33. Basser PJ. Inferring microstructural features and the physiological state of tissues from diffusionweighted images. NMR Biomed 1995;8:333 44.

34. Westin CF, Maier SE, Mamata H, et al. Processing and visualization for diffusion tensor MRI. Med Image Anal 2002;6:93 108.

35. Bammer R. Basic principles of diffusion-weighted imaging. Eur J Radiol 2003;45:169 84.

36. Schaefer PW, Copen WA, Lev MH, et al. Diffusionweighted imaging in acute stroke. Magn Reson Imaging Clin N Am 2006;14:141 68.

37. Moseley ME, Butts K, Yenari MA, et al. Clinical aspects of DWI. NMR Biomed 1995;8:387 96.

38. Moseley ME, Kucharczyk J, Mintorovitch J, et al. Diffusion- weighted MR imaging of acute stroke: correlation with T2-weighted imaging and magnetic susceptibility-enhanced MR imaging in cats. AJNR Am J Neuroradiol 1990;11:423 9.

39. Lansberg MG, Norbash AM, Marks MP, et al. Advantages of adding diffusion-weighted magnetic resonance imaging to conventional magnetic resonance imaging for evaluating acute stroke. Arch Neurol 2000;57:1311 6.

40. Anderson AW, Xie J, Pizzonia J, et al. Effects of cell volume fraction changes on apparent diffusion in human cells. Magn Reson Imaging 2000;18:689 95.

41. Kumar V, Jagannathan NR, Kumar R, et al. Apparent diffusion coefficient of the prostate in men prior to biopsy: determination of a cut-off value to predict malignancy of the peripheral zone. NMR Biomed 2007;20:505 11.

42. Kim CK, Park BK, Lee HM, et al. Value of diffusionweighted imaging for the prediction of prostate cancer location at 3T using a phased-array coil: preliminary results. Invest Radiol 2007;42:842 7.

43. Tamada T, Sone T, Toshimutsu S, et al. Age-related and zonal anatomical changes of apparent diffusion coefficient values in normal human prostatic tissues. J Magn Reson Imaging 2008;27:552 6.

44. Hricak H, Dooms GC, McNeal JE, et al. MR imaging of the prostate gland: normal anatomy. AJR Am J Roentgenol 1987;148:51 8.

45. Allen KS, Kressel HY, Arger PH, et al. Age-related changes of the prostate: evaluation by MR imaging. AJR Am J Roentgenol 1989;152:77 81.

46. Ren J, Huan Y, Wang H, et al. Diffusion-weighted imaging in normal prostate and differential diagnosis of prostate diseases. Abdom Imaging 2008.

47. Song SK, Qu Z, Garabedian EM, et al. Improved magnetic resonance imaging detection of prostate cancer in a transgenic mouse model. Cancer Res 2002;62:1555 8.

48. Manenti G, Carlani M, Mancino S, et al. Diffusion tensor magnetic resonance imaging of prostate cancer. Invest Radiol 2007;42:412 9.

49. Gibbs P, Pickles MD, Turnbull LW. Diffusion imaging of the prostate at 3.0 Tesla. Invest Radiol 2006;41: 185 8.

58



50. Riches SF, Hawtin K, Charles-Edwards EM, de Souza NM. Diffuison-weighted imaging of the prostate and rectal wall : comparison of biexponential and mono-exponential modelled diffusion and associated perfusion coefficients. NMR Biomed 2009; 318-325

51. Kim CK, Park BK, Kim B. High-b-value diffusion weighted imaging at 3 T to detect prostate cance: comparison between b values of 1000 and 2000 s/mm2. AJR 2010; 194:172 (33-37).

52. Futterer JJ, Engelbrecht MR, Jager GJ, et al. Prostate cancer: comparison of local staging accuracy of pelvic phased-array coil alone versus integrated endorectal-pelvic phased-array coils: local staging accuracy of prostate cancer using endorectal coil MR imaging. Eur Radiol 2007;17:1055 65.

53. Futterer JJ, Engelbrecht MR, Huisman HJ, et al. Staging prostate cancer with dynamic contrastenhanced endorectal MR imaging prior to radical prostatectomy: experienced versus less experienced readers. Radiology 2005;237:541 9.

59


PART TWO

DETECTION OF PRIMARY AND

RECURRENT PROSTATE CANCER

CHAPTER 3

32-Channel Coil 3T MR Guided Biopsies of Prostate Tumour Suspicious Regions on MRI

- Technique and Feasibility -

T. Hambrock; J. Fütterer; Henkjan Huisman et al.

CHAPTER 3CHAPTER 3

MR Guided Biopsies of the Prostate Technique and Feasibility - 3

Thirty-two-Channel Coil 3 Tesla Magnetic Resonance Imaging Guided Biopsies of Prostate Tumour Suspicious Regions Identified on

Multiparameteric 3T MRI: Technique and Feasibility

Hambrock T, Fütterer JJ, Huisman HJ, Hulsbergen-van de Kaa CA, van Basten JP, van Oort I, Witjes JA, Barentsz JO

First Prize Award Society of Uroradiology and European Society of Uroradiology,

Bonita Springs, Apr 2007

Advances in Knowledge

Tumour suspicious regions identified on multiparametric 3T MR imaging can effectively

be translated to T2-weighted images during an MR biopsy session.

MR guided biopsies of tumour suspicious regions, with an MR compatible biopsy device

using a 32 channel coil, 3T MRI, is a feasible technique, which can be performed in a

clinically acceptable time and needs only a low number of biopsy cores.

Implications for Patient Care

This study shows the great potential for using an MR guided biopsy device to improve

tumour detection in patients with previous negative biopsies and tumour suspicious PSA

levels.

Summary Statement

MR guided biopsies of tumour suspicious regions identified on multiparametric 3T MR imaging, using an endorectal MR compatible

biopsy device, is a feasible technique to establish a diagnosis of tumour in patients with repeat negative biopsies but persistent

suspicion for tumour based on PSA values.

62


ABSTRACT

Objectives: To test the technique and feasibility of translating tumour suspicious region maps

in the prostate, obtained by multi-modality, anatomical and functional 3T MRI data to 32

channel coil, T2-weighted, 3T MR images, for directing MR guided biopsies. Furthermore to

evaluate the practicability of MR guided biopsy on a 3T MR scanner using a 32-channel coil and a

MR compatible biopsy device.

Materials and Methods: 21 Patients with a high PSA (> 4.0 ng/ml) and at least two prior

negative transrectal ultrasound guided biopsies of the prostate, underwent an endorectal coil 3T

MRI, which included T2-weighted, Diffusion Weighted and Dynamic Contrast Enhanced MR

imaging. From these multi-modality images, tumour suspicious regions (TSR) were determined.

The 3D localization of these TSRs within the prostatic gland were translated to the T2-weighted

MR images of a subsequent 32 channel coil 3T MRI. These were then biopsied under 3T MR

guidance.

Results: In all patients, TSRs could be identified and accurately translated to subsequent 3T MR

images and biopsied under MR guidance. Median MR biopsy procedure time was 35 min. Of the

21 patients, 8 (38%) were diagnosed with prostate cancer, 6 (29%) had evidence of prostatitis, 6

(29%) had combined inflammatory and atrophic changes while only 1 (5%) patient had no

identifiable pathology.

Conclusions: Multi modality, 3T MRI determined TSRs, could effectively be translated to T2-

weighted images, to be used for MR biopsies. 3T MR guided biopsy based on these translated

TSRs was feasible, performed in a clinical useful time and resulted in a high number of positive

results.

63


INTRODUCTION

Like many cancers, prostate cancer is treated most effectively when detected early(1). Two of

the most important tests for the early diagnosis of prostate cancer are the digital rectal

examination (DRE) and the prostate specific antigen (PSA) blood test(2). Currently, prostate

cancer is most often confirmed following biopsy of the prostate(3) and the histopathological

examination of tissue obtained from these biopsies remains the reference standard for diagnosis

of prostate cancer.

Standard sextant transrectal ultrasound (TRUS) guided biopsy of the prostate up till few years

ago was the most common method used to detect prostate cancer in patients following an

abnormal DRE or high serum PSA levels(4). Recently, it was shown that extended schemes

incorporating laterally directed biopsies, significantly increase the detection rate(5;6). Biopsy

techniques consisting of 12 cores including laterally directed cores, seem a current compromise

between maximizing the cancer detection rate and minimizing adverse events(6).

Tumour detection rates at initial biopsy sessions vary according to the extent and site of

biopsies. Schemes involving sextant and octant cores, reported cancer detection rates of 22-

29%(7;8), while ten to twelve core schemes reported initial detection rates of between 33-

36%(9;10). If patients with initial negative biopsies were subjected to subsequent sextant or

octant biopsies the tumour detection rates were between 10-19%(7;11;12) while using

extended 10-12 core schemes, tumour detection rates at second biopsy, were reported to range

between 17-35%(9;10;12;13).Unfortunately, the literature is still very heterogeneous on the

biopsy schemes performed and no clear consensus is reached yet by urologists(5;8;14-19).

Because PSA is a non-specific marker for prostate cancer, urologists are often faced with the

dilemma of managing a patient with a high index of suspicion for prostate cancer after an initial

set of negative prostate biopsies. Hence, the possibility remains that these patients may still

have tumour, as prostate cancer is often multifocal and heterogeneous in nature and the volume

of prostatic tissue sampled relatively small(7). Patient anxiety about the possibility of cancer is

particularly high when there is a high index of tumour suspicion(20). Therefore, more accurate

methods need to be found to detect or rule out significant disease.

MR imaging of the prostate has established itself as a very useful modality to accurately localize

prostate cancer within the gland(21;22). On a T2-weighted (T2-w) MR image, the characteristic

pattern of prostate cancer is a low-signal-intensity lesion. But despite high-resolution imaging

(e.g., when using an endorectal coil for MR imaging at 3T), the diagnostic localization accuracy of

64


T2-weighted imaging, remains quite low(23;24). Therefore, functional MR imaging modalities

have been used to increase this accuracy. Dynamic contrast enhanced MR imaging (DCE-MRI),

diffusion weighted imaging (DWI) and proton spectroscopic MR imaging (H-MRS) have all been

established as reliable techniques for this purpose: localization accuracies of DCE-MRI being

between 80 90%(21;25), DWI-MRI between 82-89%(26-28), and H-MRS around 85

%(21;29;30). Because the sensitivity of gray-scale ultrasound to localize prostate cancer is quite

low (38% to 44%)(31), MRI, with its higher tumour localization ability, can potentially be used

as a modality for directing biopsies of tumour lesions. Despite the fact that most current data on

MR of the prostate is for 1.5T imaging, recent publications on functional imaging at 3T, show a

tendency of increased accuracy(26;32;33).

Recent publications on the use of MR spectroscopic imaging to guide TRUS biopsies have shown

an improved detection yield(34;35). Because translation of MR images to ultrasound images is

technically challenging, prostate tissue sampling techniques under direct MR guidance have

been developed. Initial experience of using such MR guided devices at 1.5T, to biopsy tumour

suspicious regions on T2-w imaging, have shown promising results(36;37).

If MR guided biopsies are considered and functional imaging modalities (DCE-MRI, DWI or H-

MRS) added to anatomical images (to increase tumour localization accuracies), a robust and

accurate technique is needed to exchange information from the localization MRI to the biopsy

MR images. This accurate translation of data is crucial to improve tumour detection yield,

especially if this is to be done with a low number of cores and performed in a clinically

acceptable time.

The principal aim of our study was to test the technique and feasibility of translating tumour

suspicious region maps obtained by multi-modality, anatomical and functional, 3T MRI data to

32 channel coil, T2-weighted, 3T MR images for directing MR guided biopsies. Furthermore, we

evaluated the practicability of MR guided biopsy on a 3T MR scanner using a 32-channel coil and

a MR compatible biopsy device.

65


MATERIALS AND METHODS

Patients

Between Aug 2006 and Mar 2007, 21 consecutive patients were referred for tumour localization,

from the departments of urology at the Radboud University Nijmegen Medical Centre (RUMCN)

and the Canisius Wilhelmina Hospital (CWZ) in Nijmegen, Netherlands, for MR imaging of the

prostate. In 20 of these patients there was a suspicion for prostate cancer based on an elevated

PSA of > 4.0 ng/ml and/or abnormal DRE. One patient had a PSA of 1ng/ml with an abnormal

DRE and previous high-grade prostate intra-epithelial neoplasia on biopsy. All patients had

received at least two prior negative TRUS guided biopsy sessions of the prostate. In all patients,

prior biopsies had been at least 6 weeks before referral. Eight patients have received more than

two prior TRUS biopsies because of continuous concern for the presence of prostate cancer,

based on excessive high PSA > 10ng/ml or continuous rising PSA.

In 8 of 21 patients, at least one extended biopsy session was performed, which included a 9-core

sampling technique (6 lateral peripheral zone, 2 transition zone, one apically directed) while the

remaining 13 had at least one 10-core biopsy (8 peripheral zone cores and 2 transition zone

cores). Median age was 62 years (range 54 to 71) and the median PSA value was 15 ng/ml

(range 1 to 123). This study was approved by the Institutional Review Board and the

requirement for signed informed consent was waivered.

Localization MR Imaging

To identify possible tumour location(s), MR imaging was performed on these patients using a 3T

MR scanner (Siemens Trio Tim, Erlangen, Germany) with the use of an endorectal coil (Medrad,

Pittsburgh, U.S.A). The endorectal coil was inserted and filled with a 40 ml perfluorocarbon

preparation (FOMBLIN, Solvay-Solexis, Milan, Italy). Peristalsis was suppressed with an

intramuscular administration of 20 mg butylscopolaminebromide (BUSCOPAN, Boehringer-

Ingelheim, Ingelheim, Germany) and 1 mg of glucagon (GLUCAGEN, Nordisk, Gentofte,

Denmark). Following this, all patients were examined using a 3T MRI.

The imaging protocol, after fast evaluation of correct endorectal coil position with fast gradient

echo imaging, included the following sequences: first, T2-weighted turbo spin echo sequences

were performed with an in-plane resolution of 0.4 x 0.4 mm (TR 3250 ms/TE 116 m; flip angle

120; 15-19 slices; 3 mm slice thickness; echo train length 15; 180 x 180 mm field of view and

448 x 448 matrix) in axial, coronal and sagittal planes, covering the prostate and seminal

vesicles. Second, a single-shot-echo-planar imaging sequence with diffusion module and fat

66


suppression pulses was implemented. Water diffusion in 3 directions was measured using b-

values of 0, 50, 500 and 800 s/mm2 and a TR of 2500ms, TE of 91 ms, slices 15-19, 3mm slice

thickness and an in-plane resolution of 1.5 x 1.5 mm. ADC-maps were automatically calculated

by the scanner software. Thirdly, 3D T1-weighted spoiled gradient-echo images (TR/TE 34/1.6

ms, 14° flip angle, 10 transverse partitions on a 3D slab, 4-mm section thickness, 192-mm field of

view, 128 x 128 matrix, GRAPPA parallel imaging, factor 2) were acquired during an intravenous

bolus injection of a paramagnetic gadolinium chelate 0.1 mmol of gadopentetate dimeglumine

(DOTAREM, Guerbet, Paris, France) per kilogram of body weight which was administered with

a power injector (Spectris; Medrad) at 2.5 ml/sec and followed by a 15-ml saline flush. With this

sequence, a 3D volume with 10 partitions was acquired every 2.5 seconds during 210 seconds,

with the same positioning angle and center as the transverse T2-weighted sequence, covering

the entire prostate. Before contrast material injection, the same transverse 3D T1-weighted

gradient echo sequence (with the exception of TR/TE of 800/1.6 and an 8° flip angle) was used

to obtain proton-density images, with identical positioning to allow calculation of the relative

gadolinium chelate concentration curves.

Localization MR Data Analysis

The prostate images were viewed on an in-house developed analytical software workstation

(38), which calculated the pharmacokinetic DCE-MRI parameters and projected these

parameters as color overlay maps over the T2-w images. Additionally, ADC maps calculated

from DWI were also projected as color overlays. DWI images as such were not used as part of

the evaluation. If patient related movement caused misregistration of the different modalities,

these were corrected using a manual co-registration tool, built into the software.

Images of all patients were read in consensus by two readers with one (T.H) and four years (J.F)

experience in prostate MR imaging. The high-resolution, axial T2-w images were used as basis

for evaluation of the prostate and all other functional imaging modalities were interpreted in

relation to these. On T2-w imaging, the generally known tumour criteria were used to detect

TSRs. These included (a) low signal intensity areas in the peripheral zone (PZ), (b) within the

transition zone, a homogeneous low T2 signal intensity area with ill-defined margins or a

lenticular shape(39) and (c) within the central zone, areas of homogenous low signal intensity

with an ill-defined margin. After identification of tumour suspicious regions on T2-w images,

the ADC maps and multi-parametric pharmacokinetic (DCE-MRI derived) color maps - Ktrans, Ve,

Kep and WashOut, were analyzed in a color overlay mode on the T2-w images. The generally

known features of tumour on DCE-MR imaging(21;40) (high Ve, Ktrans, Kep and negative

67


WashOut) as well as areas of restriction on ADC maps (especially in the PZ and transition zone),

were used to increase the specificity of the T2-w identified TSRs.

Additionally, after the functional data from DWI and DCE images were evaluated in relation to

the TSR findings on the T2-w images, the DWI and DCE images were viewed separately and in

combination to determine additional TSRs not evident on T2-w images. Eventually, the

information from all the imaging modalities were combined and used to determine the (up to

three) most suspicious TSRs within the prostate.

Figure 1 shows an example of how the prostate was divided into different axial and sagittal

regions for 3D spatial position estimation of the TSR. This was done as follows: on the sagittal

T2-w images, the prostate was divided into 5 slabs, equating to: apex, apex-mid, mid, mid-basis

and basis levels. These slabs were equal in thickness and parallel to the axial T2-w images. The

axial T2-w slice(s) containing the TSR were then related to the corresponding sagittal slab. On

the axial T2-w images, distinction was made between peripheral zone, central zone and

transition zone and the relationship of the TSRs to these was noted. Furthermore, each axial

zone was divided into the following sub-zone regions. Peripheral zone for each left and right

half of the prostate, using the urethra as dividing point : 1) anterior horn 2) dorso-lateral region

3) dorsal region. The central zone was divided in four quadrants while the transition zone was

divided into left and right regions only. The apical and basal slabs, where the PZ and CG often do

not clearly co-exist, were divided into simple quadrants. The TSR in relation to the sub-zone

region was then noted. Therefore, for each TSR position, the slab, zone and sub-zone, were

recorded. This position was used as basis for re-identification of a TSR on the T2-w images

during the biopsy MR session.

68


Figure. 1. Anatomical T2-weighted images in the axial (a,c,e) and sagittal plane (g). The

sagittal images (h) were divided into 5 levels: base (B), mid-base (MB), mid (M), apex-mid

(AM) and apex (A). On the axial images the apical (b) and basal slabs (f) were divided into

quadrants, while the axial images corresponding to the apex-mid, mid and mid-basis

slabs (d) were subdivided into anatomical regions corresponding to the peripheral zone

(blue), central zone (red) and transition zone (yellow). The transition zone was divided

into left/right halves, the central zone into a quadrant while the left/right halves of the

peripheral zone, each were subdivided into the anterior horns (1), dorso-lateral region

(2) and dorsal region (3).

69


MR Guided biopsy

On average, two weeks (range 1 week 4 weeks) after the initial tumour localization MRI,

patients received a 32-channel coil (Invivo, Schwerin, Germany), 3T MR guided biopsy (Trio Tim,

Siemens, Germany). For antibiotic prophylaxis, all patients took oral ciprofloxacin 500 mg

(CIPROXIN, Bayer, Leverkusen, Germany) the evening before, on the morning of the biopsy, as

well as 6 hours post biopsy. Prostate biopsies were performed with the patient in the prone

position, with the 32 channel coil elements positioned beneath as well as on the back of the

patient.

A gadolinium filled needle guider was inserted rectally and attached to the arm of a MR

compatible biopsy device (Invivo, Schwerin, Germany). Figure 2 shows the MR guided biopsy

device attached to the patient, lying prone on the scanner table. The used method and

adjustments made to the device during scanning, were previously described (36;37). In

summary, the arm onto which the needle guider was attached, enabled the needle guider to be

rotated, moved forward and backward, and adjusted in height. The insertion angle can be

adjusted by rotating the needle guide about a point inside the rectum. The needle guide was

then directed to the defined TSR within the prostate. After correct alignment, the needle guide

was fixed in position for obtaining tissue samples with an 18-gauge, fully automatic, core-needle,

double-shot biopsy gun (Invivo, Schwerin, Germany) with needle length of 150 mm and tissue

core sampling length of 17 mm.

T2-w turbo spin-echo images in the axial and sagittal direction (TR 3500 ms/TE 116 ms, flip

angle 180, slice thickness 3 mm, in-plane resolution 0.7 x 0.7 mm, number of slice 15) were

obtained for anatomical visualization of the prostate. The axial T2-w slices had a similar

angulation relative to the dorsal surface of the prostate as during the localization MR session.

The MR images of the previous localization MRI were projected on a monitor, positioned next to

the MR console.

Re-identification of the TSRs on the new T2-w images was done firstly by using the relative 3D

position, which incorporated the 5 slabs, the zones and different sub-zonal regions. For this

purpose, on the sagittal T2-w images, the prostate was divided into the five slabs from apex to

base. The axial T2-w slices corresponding to the TSR slab were then identified. These were

visually divided into the different zones and sub-zone regions (see localization for details).

When the desired sub-zone region was re-identified, the positioning was a fine-tuned. This was

done by noting heterogeneous features within the gland, e.g. features relating to nodular and

stromal appearances, position of urethra and other prominent characteristics within the zones.

70


These were established on the T2-w images of the initial MRI, which were projected on the

computer screen next to the console and then compared to the features visible on the biopsy MR

images. If a low-signal intensity lesion was evident on the original T2-w images and re-

identified on the current T2-w images within the desired zone-region, a certain TSR re-

identification was evident. Otherwise one or two slices were scrolled up or down (axial T2-w

slices) for identification of the T2-w evident lesion.

After re-identification of the desired TSR, adjustments were made to the biopsy device to aim

the gadolinium-filled needle guider exactly towards this area. In-between adjustments, fast T2-

w TRUE-FISP images (TR 4.48ms/TE 2.24/Flip angle 70º, FOV 228 x 228 mm, Matrix 228 x 228,

Slice thickness 3mm) were made in the axial and sagittal direction (imaging time 11 seconds for

each direction) to visualize correct guider position and used to plan further adjustments.

Biopsies were obtained and to verify correct needle position within the TSR, fast T2-w TRUE-

FISP images were again obtained with the needle left in situ. This was done for each TSR that

was biopsied. One to three biopsies were taken per TSR, depending on the certainty of correct

needle position within a TSR as well as the size of the TSR. A maximum of three different TSRs

were biopsied per patient. All biopsies were performed by one radiologist (TH).

Samples were subsequently processed by a routine fixation in formaldehyde, embedded in

paraffin, stained with hematoxylin-eosin, before being evaluated by a histopathologist for the

presence of tumour or other benign pathologies.

Figure 2. Patient in prone position, with a gadolinium filled needle guider inserted

endorectally and attached to a 3D manipulating, MR compatible biopsy device (Invivo,

Schwerin, Germany)

71


RESULTS

The tumour localization MRI had an average scanning duration of 22 min. In all 21 patients

referred for a tumour localization MRI, one or more TSRs could be identified. On average two

TSRs were identified per patient (range 1 to 3). Using the described translation technique, all

TSR regions could be confidently re-identified during the biopsy MR session. Patients tolerated

the MR guided biopsies well and apart from one transient transurethral hemorrhage

immediately following the procedure, only minor pain after tissue sampling was reported as a

side effect by some patients. By imaging with the needle left in situ, identification of needle

position confirmed correct sampling of all the TSRs.

In total, 84 prostate cores were obtained from 40 different TSRs in 21 patients. The average

number of biopsies per patient was 4 (range 1 to 7). Of the 40 different TSRs sampled, 23%

(9/40) contained tumour, while 77% (26/40) were normal, containing benign pathological

changes in 65% (26/40) and no changes in 13% (5/40). Histopathological analysis of the

prostate samples revealed adenocarcinoma in 31% of cores (26/84) and in 38% of patients (8

out of 21). Of all core samples, 27% revealed prostatitis (23/84) while 21% (18/84) showed

combined atrophic and inflammatory changes, 1 core (1%) showed atypia and 1 (1 %) core

revealed necrosis. No identifiable pathology was found in 18% (15/84) of cores and in only one

patient.

Of the 8 patients identified with adenocarcinoma, one patient had a tumour with Gleason score

5, four had a tumour with Gleason score 6, one with two TSRs positive for tumour, each with a

Gleason score 7 and two patients with a Gleason score of 8. Of the 9 TSRs positive for tumour,

67% (6 out of 9) were in the ventral aspect of the transition zone, 22% (2 out of 9) in the

peripheral zone and 11% (1 out of 9) in the central zone.

The median duration of MR imaging guided biopsies was 35 min (range 21-75 min). A learning

curve was evident in manipulating the biopsy device and performing biopsies, quickly and

effectively. This was also reflected in the median imaging time of 41 min for the first 10 patients,

which subsequently decreased to 32 min for the following 11 patients. Table 1 reveals a

summary of the patient and biopsy findings. The imaging features of the TSRs, positive for

tumour, are summarized in Table 2. Figure 3 shows the multi-modality MR images of a patient,

which include: T2-w, DWI and DCE-MRI, used for identifying the TSRs. In this patient, the

tumour was situated in the right ventral aspect of the prostate, on the border between the

transition and central zone. Figure 4 shows this TSR, which was subsequently re-identified on

the T2-w images during the MR guided biopsy session.

72


Patient Age PSA Nr.previous TRUS biopsies

Nr. TSRs

Nr. Biopsies

Diagnosis Tumour Gleason Score

1 69 20 3 2 3 Tumour 6 (3+3)

2 63 1 2 1 1 Prostatitis

3 65 17 4 2 3 Tumour 6 (3+3)

4 66 20 2 2 3 Prostatitis/

Necrosis

-


Atrophy

-


Atrophy

-

7 60 4 2 1 3 Prostatitis -



10 54 12 2 1 1 Prostatitis -

11 57 8 2 2 6 N.A.D -


Atrophy

-

13 58 123 4 2 5 Tumour 6 (3+3)

14 63 14 2 3 6 Tumour 7 (4+3)


Atrophy

-


17 70 34 2 1 4 Tumour 8 (4+4)

18 61 16 4 3 3 Prostatitis

19 62 5 3 1 6 Tumour 5 (3+2)

20 71 17 3 1 3 Tumour 6 (3+3)

21 70 15 2 1 4 Tumour 8 (5+3)

Table 1. Patient and biopsy characteristics of the 21 patients biopsied.

73


TSR T2-w DWI DCE-MRI

1 + + +

2 + + +

3 0 + +

4 0 + +

5 + + -

6 + + -

7 + - +

8 - - +

9 - + +

Table 2. MR imaging features of TSRs positive for tumour on biopsy. A “+” indicates that a

lesion is visible while “-“ indicates no lesion visibility. Features indicated with “0” denote

non-specific heterogeneous low-signal intensity lesions on T2-w, within the central gland.

DISCUSSION

Our principal aim was to test the technique and feasibility of translating multi-modality,

anatomical and functional, 3T MRI data of tumour suspicious regions, to subsequent 32 channel

coil, T2-w, 3T MR images. We have shown this to be feasible and can be performed without

much difficulty. We further demonstrated that such a technique could be used to direct and

perform MR guided biopsies in a clinically acceptable manner on a 3T MR scanner using a 32-

channel coil and an MR compatible biopsy device.

To achieve accurate tumour localization, we performed an endorectal coil, 3T MRI, using

validated, multi-modality MR sequences with high diagnostic accuracies for tumour localization.

Additionally, because reading of these images was performed using a high sensitivity approach,

TSRs could be identified in all patients. To use multi-modality 3T MR imaging for directing 3T

MR guided biopsies is unique and contrasts to prior studies in which only T2-w imaging at low-

field systems was utilized(36;37;41).

We established that using high-resolution T2-w images as basis for image interpretation of

functi -

tuning, resulted in an accurate 3D translation of TSRs between the MR sessions. In all 21

74


patients, we were confident that every TSR could be adequately re-identified during the biopsy

MR session. Apart from the translation technique, this can also be related to using high-

resolution images, obtained with a 3T 32-channel coil for biopsy guidance.

We found that 32-channel coil 3T MR imaging guided biopsy, with the patient in the prone

position, is well tolerated and feasible to perform within a clinically acceptable time. The

practicability of using the current biopsy technique and equipment is in agreement with two

prior 1.5T studies (36;37) which describe a similar setup. In comparison to these prior studies,

our imaging time was remarkably reduced with an average duration of 35 min, compared to 55

min and 2 hours respectively. It should however be mentioned, that the latter reported imaging

times were also particularly high because of an initial learning curve in performing MR biopsies

and additionally a higher number of cores were obtained. We found that using the fast (11 s) T2-

w TRUE-FISP sequences at 3T, resulted in adequate visibility of anatomical details to orientate

the needle guider effectively. Faster imaging and higher anatomical detail is probably the

biggest advantage of 3T MR biopsy over 1.5T, especially with the use of a 32 channel surface coil

and parallel imaging.

Despite the fact that re-identification of the translated TSRs remains somewhat subjective; the

high tumour detection rate directed towards these regions (31% of samples and 38% of

patients) and the high prevalence of benign pathological diagnoses (67% - 27/40 of TSRs), is a

good indirect validation of satisfactory re-identification. The preliminary results also show a

higher tumour detection percentage (38%) compared to sextant and octant TRUS guided

biopsies after two prior negative biopsies, reported in the literature (8-14%)(7;11;12). It has to

be emphasized that the detection of prostate cancer in subsequent biopsy sessions is strongly

dependent on the biopsy scheme used during the initial and subsequent sessions(42) as well as

patient related factors like PSA value, prostate volume and the population prevalence of prostate

cancer(43;44). In an attempt to increase the detection rate during rebiopsy, saturation biopsy

techniques with greatly increased the number of samples (> 20 cores), have been advocated by

some researchers. These saturation biopsy strategies were shown to increase detection rates

(25-41%)44,45,46, but at the expense of pain and complications, as well as the unduly high cost of

processing the large amount of pathological material. The preliminary results of our current

research may not show an increased detection performance to such saturation schemes but at a

similar performance, our technique will be a more appealing alternative. Furthermore,

saturation biopsies are not yet advocated by the European Guidelines for Prostate Cancer and

only performed by few urologists in our country.

75


Figure 3. A case example with images from the endorectal coil, multi-modality 3T MRI.

The axial T2-w image (a) shows a low signal intensity area in the transitions zone, right

(arrow). The ADC map (b) from the DWI shows the same area with restriction in diffusion

ability while the WashOut pharmacokinetic map (c), calculated from the DCE-MRI data,

shows an enhanced removal rate of gadolinium. Images (d) and (e) show the TSR

identified by the readers. The subsequent “crude” estimation of the 3D location within

the gland shows the TSR to be predominantly within the right aspect of the transition

zone (f) and in the mid-base (MB) slab (g).

76


Figure 4. The TSR (yellow interrupted circle) identified from the multi-modality,

endorectal coil, 3T MR images, is translated to the 32 channel coil T2-w images using the

3D localization method. The mid-base (MB) level containing the TSR is first identified (a).

On the axial T2-w images (b) within this level, the right transition zone area with TSR is

found. The gadolinium filled needle guider is directed on sagittal (c) and axial (d) images

towards this re-identified region. Correct needle positioning and tissue sampling from

the TSR is verified by imaging with the needle left insitu (e).

77


All our patients have received at least one prior TRUS session with tissue sampling of the

transition zone. Despite this, 67 % of our tumour containing TSRs were situated in the

transition zone. Although our results are too preliminary to make definite conclusions, it does

seem as if the lack of extended transition zone sampling in rebiopsy session could account for

the large number of tumours missed. It is known that 25% of prostate cancers arise in the

transition zone(45), yet most studies on extended biopsy techniques, conclude that laterally

directed cores, apical sampling or sampling of the anterior horns, increases the detection yield,

while additional sampling of the transition zone, does not. (6;46-48)

The transrectal sampling approach has the advantage of being the least invasive and requiring

no anesthesia, compared to other described MR biopsy approaches, of which transperineal(49)

and transgluteal(41;50) are the best known techniques. However, infection risk is increased

when an endorectal approach to tissue sampling is used(51). Djavan et al.(51) has shown that

complications due to prostatic biopsies are not trivial. Of all their patients that underwent

biopsies, 11% developed infectious complications, 63% had bleeding related problems while up

to 8% had significant pain or discomfort. We have demonstrated that by limiting the biopsy

cores to MR determined, tumour suspicious regions only, a lower number of cores can be

obtained, with an average of 4 in our study. This has the advantageous potential to reduce

bleeding risk, infection risk as well as patient discomfort and pain.

Other authors have shown that using T2-w imaging alone for tumour localization can be

valuable in guiding biopsies performed in a closed-bore MRI scanner. It is known that T2-w

imaging on its own has a quite poor sensitivity for localizing prostate cancer, reported in the

literature to range between 58% and 65%(32;52). Only one study has looked at a possible

advantage of adding DCE-MRI imaging to T2-w imaging for guidance of MR biopsies.

Unfortunately, a too small patients group was evaluated to assess the possible benefit(53).

Because of the currently available data on tumour localization accuracies by MRI, we decided to

maximize the possibility of tumour detection likelihood, by utilizing high-spatial resolution

imaging with an endorectal coil at 3T and obtaining DCE and DWI imaging in addition to T2-w

imaging. This approach however, has the drawback that performing a localization MRI and

biopsy MRI in the same imaging session is deemed unpractical. Changing the endorectal coil,

awaiting software post-processing of the DCE data and the time consuming evaluation of images,

were all considered practical reasons to postpone the MR guided biopsy to another session.

Additionally because recent evidence by Heijmink et al. (23) has shown that the localization

accuracy of endorectal coil, T2-w 3T MR imaging is significantly improved compared to using a

78


surface coil for T2-w, 3T imaging (Az 0.68 vs 0.62), we decided to use the endorectal coil for

localization.

The preliminary data on the 9 TSRs positive for tumour (Table 2) indicated that for 2 TSRs,

functional imaging modalities directed the biopsies towards tumour whereas T2-w images

showed no lesion visible. Additionally, for another 2 TSRs, the heterogeneous nature of the

central gland with diffuse low signal intensities on T2-w imaging was found unhelpful for clear

localization of the TSRs, something that was only possible with the addition of functional data.

One of the limitations of this study is the small number of patients used. However, this study

was designed as a pilot to assess the technique of translating multi-modality MR data and the

feasibility of MR imaging guided biopsies at 3 Tesla. Additionally, an arbitrarily defined

maximum of 3 TSRs was chosen, as increasing this number further would have resulted in

prolonged, clinically unacceptable imaging time and discomfort to the patient. Interpreting

anatomical and functional MR images of the prostate requires a degree of experience, which

might make routine clinical application of this technique somewhat difficult. However, recently

Vos et al.(54) showed that by using computer aided diagnostic (CAD) software support for DCE-

MRI, a method can be found to aid radiologists to improve the accuracy of tumour localization.

CONCLUSIONS

Our conclusions are that TSR maps identified on multi-modality, anatomical and functional, 3T

MR imaging can effectively be translated to T2-weighted images during a MR biopsy session by

using simple visual criteria. Furthermore, MR guided biopsies of these TSRs, with the MR

compatible biopsy device using a 32 channel coil, 3T MRI, is a feasible technique, which can be

performed in a clinically acceptable time and needs only a low number of biopsy cores. This

shows great potential for improving tumour detection in patients with previous negative

biopsies and tumour suspicious PSA levels. Assessing the true overall tumour detection

capabilities, the benefit of MR guided biopsies over extended biopsy techniques in rebiopsy

sessions and the need for using an endorectal coil, is part of an ongoing study.

79


REFERENCES

1. McLeod DG. Success and failure of single-modality treatment for early prostate cancer. Rev.Urol. 2004;6 Suppl 2:S13-S19.

2. Bigliocchi M, Marini M, Nofroni I, Perugia G, Shahabadi H, Ciccariello M. Prostate cancer detection rate of transrectal ultrasonography, digital rectal examination, and prostate-specific antigen: results of a five-year study of 6- versus 12-core transperineal prostate biopsy. Minerva Urol.Nefrol. 2007 Dec;59(4):395-406.

3. Hodge KK, McNeal JE, Terris MK, Stamey TA. Random systematic versus directed ultrasound guided transrectal core biopsies of the prostate. J.Urol. 1989 Jul;142(1):71-4.

4. Raja J, Ramachandran N, Munneke G, Patel U. Current status of transrectal ultrasound-guided prostate biopsy in the diagnosis of prostate cancer. Clin.Radiol. 2006 Feb;61(2):142-53.

5. Djavan B, Margreiter M. Biopsy standards for detection of prostate cancer. World J.Urol. 2007 Mar;25(1):11-7.

6. Eichler K, Hempel S, Wilby J, Myers L, Bachmann LM, Kleijnen J. Diagnostic value of systematic biopsy methods in the investigation of prostate cancer: a systematic review. J.Urol. 2006 May;175(5):1605-12.

7. Djavan B, Ravery V, Zlotta A, Dobronski P, Dobrovits M, Fakhari M, Seitz C, Susani M, Borkowski A, Boccon-Gibod L, et al. Prospective evaluation of prostate cancer detected on biopsies 1, 2, 3 and 4: when should we stop? J.Urol. 2001 Nov;166(5):1679-83.

8. Gosselaar C, Roobol MJ, Roemeling S, Wolters T, van Leenders GJ, Schroder FH. The value of an additional hypoechoic lesion-directed biopsy core for detecting prostate cancer. BJU.Int. 2008 Mar;101(6):685-90.

9. Mian BM, Naya Y, Okihara K, Vakar-Lopez F, Troncoso P, Babaian RJ. Predictors of cancer in repeat extended multisite prostate biopsy in men with previous negative extended multisite biopsy. Urology 2002 Nov;60(5):836-40.

10. Philip J, Hanchanale V, Foster CS, Javle P. Importance of peripheral biopsies in maximising the detection of early prostate cancer in repeat 12-core biopsy protocols. BJU.Int. 2006 Sep;98(3):559-62.

11. Keetch DW, Catalona WJ, Smith DS. Serial prostatic biopsies in men with persistently elevated serum prostate specific antigen values. J.Urol. 1994 Jun;151(6):1571-4.

12. Roehl KA, Antenor JA, Catalona WJ. Serial biopsy results in prostate cancer screening study. J.Urol. 2002 Jun;167(6):2435-9.

13. Hong YM, Lai FC, Chon CH, McNeal JE, Presti JC, Jr. Impact of prior biopsy scheme on pathologic features of cancers detected on repeat biopsies. Urol.Oncol. 2004 Jan;22(1):7-10.

14. Djavan B, Zlotta AR, Ekane S, Remzi M, Kramer G, Roumeguere T, Etemad M, Wolfram R, Schulman CC, Marberger M. Is one set of sextant biopsies enough to rule out prostate Cancer? Influence of transition and total prostate volumes on prostate cancer yield. Eur.Urol. 2000 Aug;38(2):218-24.

15. Djavan B, Remzi M, Marberger M. When to biopsy and when to stop biopsying. Urol.Clin.North Am. 2003 May;30(2):253-62, viii.

16. Eskew LA, Bare RL, McCullough DL. Systematic 5 region prostate biopsy is superior to sextant method for diagnosing carcinoma of the prostate. J.Urol. 1997 Jan;157(1):199-202.

17. Presti JC. Prostate biopsy: current status and limitations. Rev.Urol. 2007;9(3):93-8. 18. Presti JC, Jr., Chang JJ, Bhargava V, Shinohara K. The optimal systematic prostate biopsy scheme should include

8 rather than 6 biopsies: results of a prospective clinical trial. J.Urol. 2000 Jan;163(1):163-6.

19. Roobol MJ, Grenabo A, Schroder FH, Hugosson J. Interval cancers in prostate cancer screening: comparing 2- and 4-year screening intervals in the European Randomized Study of Screening for Prostate Cancer, Gothenburg and Rotterdam. J.Natl.Cancer Inst. 2007 Sep 5;99(17):1296-303.

20. Carlsson S, Aus G, Wessman C, Hugosson J. Anxiety associated with prostate cancer screening with special reference to men with a positive screening test (elevated PSA) - Results from a prospective, population-based, randomised study. Eur.J.Cancer 2007 Jul 20.

21. Futterer JJ, Heijmink SW, Scheenen TW, Veltman J, Huisman HJ, Vos P, Hulsbergen-Van de Kaa CA, Witjes JA, Krabbe PF, Heerschap A, et al. Prostate cancer localization with dynamic contrast-enhanced MR imaging and proton MR spectroscopic imaging. Radiology 2006 Nov;241(2):449-58.

22. Tanimoto A, Nakashima J, Kohno H, Shinmoto H, Kuribayashi S. Prostate cancer screening: the clinical value of diffusion-weighted imaging and dynamic MR imaging in combination with T2-weighted imaging. J.Magn Reson.Imaging 2007 Jan;25(1):146-52.

23. Heijmink SW, Futterer JJ, Hambrock T, Takahashi S, Scheenen TW, Huisman HJ, Hulsbergen-Van de Kaa CA, Knipscheer BC, Kiemeney LA, Witjes JA, et al. Prostate cancer: body-array versus endorectal coil MR imaging at 3 T--comparison of image quality, localization, and staging performance. Radiology 2007 Jul;244(1):184-95.

80


24. Jager GJ, Ruijter ET, van de Kaa CA, de la Rosette JJ, Oosterhof GO, Thornbury JR, Barentsz JO. Local staging of prostate cancer with endorectal MR imaging: correlation with histopathology. AJR Am.J.Roentgenol. 1996 Apr;166(4):845-52.

25. Ito H, Kamoi K, Yokoyama K, Yamada K, Nishimura T. Visualization of prostate cancer using dynamic contrast-enhanced MRI: comparison with transrectal power Doppler ultrasound. Br.J.Radiol. 2003 Sep;76(909):617-24.

26. Kim CK, Park BK, Lee HM, Kwon GY. Value of diffusion-weighted imaging for the prediction of prostate cancer location at 3T using a phased-array coil: preliminary results. Invest Radiol. 2007 Dec;42(12):842-7.

27. Haider MA, van der Kwast TH, Tanguay J, Evans AJ, Hashmi AT, Lockwood G, Trachtenberg J. Combined T2-weighted and diffusion-weighted MRI for localization of prostate cancer. AJR Am.J.Roentgenol. 2007 Aug;189(2):323-8.

28. Miao H, Fukatsu H, Ishigaki T. Prostate cancer detection with 3-T MRI: Comparison of diffusion-weighted and T2-weighted imaging. Eur.J.Radiol. 2007 Feb;61(2):297-302.

29. Futterer JJ, Scheenen TW, Heijmink SW, Huisman HJ, Hulsbergen-Van de Kaa CA, Witjes JA, Heerschap A, Barentsz JO. Standardized threshold approach using three-dimensional proton magnetic resonance spectroscopic imaging in prostate cancer localization of the entire prostate. Invest Radiol. 2007 Feb;42(2):116-22.

30. Jung JA, Coakley FV, Vigneron DB, Swanson MG, Qayyum A, Weinberg V, Jones KD, Carroll PR, Kurhanewicz J. Prostate depiction at endorectal MR spectroscopic imaging: investigation of a standardized evaluation system. Radiology 2004 Dec;233(3):701-8.

31. Halpern EJ, Strup SE. Using gray-scale and color and power Doppler sonography to detect prostatic cancer. AJR Am.J.Roentgenol. 2000 Mar;174(3):623-7.

32. Kim CK, Park BK, Kim B. Localization of prostate cancer using 3T MRI: comparison of T2-weighted and dynamic contrast-enhanced imaging. J.Comput.Assist.Tomogr. 2006 Jan;30(1):7-11.

33. Manenti G, Carlani M, Mancino S, Colangelo V, Di RM, Squillaci E, Simonetti G. Diffusion tensor magnetic resonance imaging of prostate cancer. Invest Radiol. 2007 Jun;42(6):412-9.

34. Prando A, Kurhanewicz J, Borges AP, Oliveira EM, Jr., Figueiredo E. Prostatic biopsy directed with endorectal MR spectroscopic imaging findings in patients with elevated prostate specific antigen levels and prior negative biopsy findings: early experience. Radiology 2005 Sep;236(3):903-10.

35. Yuen JS, Thng CH, Tan PH, Khin LW, Phee SJ, Xiao D, Lau WK, Ng WS, Cheng CW. Endorectal magnetic resonance imaging and spectroscopy for the detection of tumour foci in men with prior negative transrectal ultrasound prostate biopsy. J.Urol. 2004 Apr;171(4):1482-6.

36. Anastasiadis AG, Lichy MP, Nagele U, Kuczyk MA, Merseburger AS, Hennenlotter J, Corvin S, Sievert KD, Claussen CD, Stenzl A, et al. MRI-guided biopsy of the prostate increases diagnostic performance in men with elevated or increasing PSA levels after previous negative TRUS biopsies. Eur.Urol. 2006 Oct;50(4):738-48.

37. Beyersdorff D, Winkel A, Hamm B, Lenk S, Loening SA, Taupitz M. MR imaging-guided prostate biopsy with a closed MR unit at 1.5 T: initial results. Radiology 2005 Feb;234(2):576-81.

38. Futterer JJ, Engelbrecht MR, Huisman HJ, Jager GJ, Hulsbergen-Van de Kaa CA, Witjes JA, Barentsz JO. Staging prostate cancer with dynamic contrast-enhanced endorectal MR imaging prior to radical prostatectomy: experienced versus less experienced readers. Radiology 2005 Nov;237(2):541-9.

39. Akin O, Sala E, Moskowitz CS, Kuroiwa K, Ishill NM, Pucar D, Scardino PT, Hricak H. Transition zone prostate cancers: features, detection, localization, and staging at endorectal MR imaging. Radiology 2006 Jun;239(3):784-92.

40. Engelbrecht MR, Huisman HJ, Laheij RJ, Jager GJ, van Leenders GJ, Hulsbergen-Van de Kaa CA, de la Rosette JJ, Blickman JG, Barentsz JO. Discrimination of prostate cancer from normal peripheral zone and central gland tissue by using dynamic contrast-enhanced MR imaging. Radiology 2003 Oct;229(1):248-54.

41. Zangos S, Eichler K, Engelmann K, Ahmed M, Dettmer S, Herzog C, Pegios W, Wetter A, Lehnert T, Mack MG, et al. MR-guided transgluteal biopsies with an open low-field system in patients with clinically suspected prostate cancer: technique and preliminary results. Eur.Radiol. 2005 Jan;15(1):174-82.

42. Hong YM, Lai FC, Chon CH, McNeal JE, Presti JC, Jr. Impact of prior biopsy scheme on pathologic features of cancers detected on repeat biopsies. Urol.Oncol. 2004 Jan;22(1):7-10.

43. Dong F, Jones JS, Stephenson AJ, Magi-Galluzzi C, Reuther AM, Klein EA. Prostate cancer volume at biopsy predicts clinically significant upgrading. J.Urol. 2008 Mar;179(3):896-900.

44. Presti JC, Jr., Chang JJ, Bhargava V, Shinohara K. The optimal systematic prostate biopsy scheme should include 8 rather than 6 biopsies: results of a prospective clinical trial. J.Urol. 2000 Jan;163(1):163-6.

45. Sakai I, Harada K, Hara I, Eto H, Miyake H. A comparison of the biological features between prostate cancers arising in the transition and peripheral zones. BJU.Int. 2005 Sep;96(4):528-32.

81


46. Damiano R, Autorino R, Perdona S, De SM, Oliva A, Esposito C, Cantiello F, Di LG, Sacco R, D'Armiento M. Are extended biopsies really necessary to improve prostate cancer detection? Prostate Cancer Prostatic.Dis. 2003;6(3):250-5.

47. Inahara M, Suzuki H, Kojima S, Komiya A, Fukasawa S, Imamoto T, Naya Y, Ichikawa T. Improved prostate cancer detection using systematic 14-core biopsy for large prostate glands with normal digital rectal examination findings. Urology 2006 Oct;68(4):815-9.

48. Siu W, Dunn RL, Shah RB, Wei JT. Use of extended pattern technique for initial prostate biopsy. J.Urol. 2005 Aug;174(2):505-9.

49. D'Amico AV, Tempany CM, Cormack R, Hata N, Jinzaki M, Tuncali K, Weinstein M, Richie JP. Transperineal magnetic resonance image guided prostate biopsy. J.Urol. 2000 Aug;164(2):385-7.

50. Zangos S, Herzog C, Eichler K, Hammerstingl R, Lukoschek A, Guthmann S, Gutmann B, Schoepf UJ, Costello P, Vogl TJ. MR-compatible assistance system for punction in a high-field system: device and feasibility of transgluteal biopsies of the prostate gland. Eur.Radiol. 2007 Apr;17(4):1118-24.

51. Djavan B, Waldert M, Zlotta A, Dobronski P, Seitz C, Remzi M, Borkowski A, Schulman C, Marberger M. Safety and morbidity of first and repeat transrectal ultrasound guided prostate needle biopsies: results of a prospective European prostate cancer detection study. J.Urol. 2001 Sep;166(3):856-60.


53. Singh AK, Krieger A, Lattouf JB, Guion P, Grubb RL, III, Albert PS, Metzger G, Ullman K, Smith S, Fichtinger G, et al. Patient selection determines the prostate cancer yield of dynamic contrast-enhanced magnetic resonance imaging-guided transrectal biopsies in a closed 3-Tesla scanner. BJU.Int. 2008 Jan;101(2):181-5.

54. Vos P, Hambrock T, Hulsbergen-Van de Kaa CA, Futterer JJ, Barentsz J, Huisman HJ. Computerized analysis of prostate lesions in the peripheral zone using dynamic contrast enhanced MRI. Med.Phys. 2008;Accepted, awaiting publication.

82

— CHAPTER 4 —

Magnetic Resonance Imaging Guided Prostate Biopsies in Men

with Repetitive Negative Biopsies and Elevated PSA

T. Hambrock; D. Somford; C. Hoeks et al.

CHAPTER 4CHAPTER 4

Magnetic Resonance Imaging Guided Prostate Biopsies in Men with Repetitive Negative Biopsies and Elevated PSA

Journal of Urology, 2010 Feb; 183(2):520-7

Hambrock T, Somford DM, Hoeks C, Bouwense SA, Huisman HJ, Yakar D,

van Oort IM, Witjes JA, Fütterer JJ, Barentsz JO

Cum Laude Award Society of Computed Body Tomography and Magnetic Resonance, 2009


3T Multiparametric MRI is a highly effective method for the detection and localization of

clinically significant prostate cancer.

MR guided biopsies towards tumour suspicious regions on MRI is a very useful method

for accurately validating correct sampling of prostatic tissue.

Tumour detected in patients with repeat negative biopsies are mostly located in areas

not explicitly sampled by routine schemes.


MRI should be considered essential in any workup protocol of patients who are

suspected of harboring malignancy but who have successive negative biopsies.

Because of the low numbers of cores needed, MR guided biopsies is an appealing

alternative to procedures such as saturation biopsies.

Summary Statement

3T Multiparametric MR imaging in combination with MR guided biopsies, represent a very effective method for diagnosing clinically significant prostate cancer in men with repeat negative biopsies and

abnormal PSA values.

84

MRI Guided Prostate Biopsies in Men with Repetitive Negative Biopsies and Elevated PSA

4

ABSTRACT

Background: Undetected cancer in repeat transrectal ultrasound guided prostate biopsies

(TRUS-GB) in patients with elevated Prostatic specific antigen (PSA) > 4 ng/mL is a considerable

concern. We investigated the tumour detection rate of biopsies from tumour suspicious regions

on multi-modality 3T Magnetic Resonance Imaging (m-m MRI) and subsequent MR-guided

biopsy (MR-GB) in sixty eight men with repetitive negative TRUS-GB and compared this to a

matched TRUS-GB population. Furthermore we aimed to determine the clinical significance of

detected tumours.

Methods: Seventy one consecutive patients with PSA >4 ng/ml and -GB

sessions received a m-m MRI. In 68 patients this was followed by MR-GB directed towards

tumour suspicious regions. A matched, multi-session TRUS-GB population from our institutional

database was used for comparison. Clinical significance of tumours detected was established

using accepted criteria (PSA, gleason grade, stage, tumour volume).

Results: The tumour detection rate (DR) with m-m MRI MR-GB was 59% (40/68) using a

median of 4 cores. The tumour DRs were significantly (p <0.01) higher than TRUS-GB in all

patient subgroups except those with PSA >20 ng/ml, prostate volumes >65 cc and PSA density

>0 tumours,

93% (37/40) were considered highly likely to harbor clinically significant disease.

Conclusions: Multi-modality MRI is an effective technique for localizing prostate cancer and

MR-guided biopsy of tumour suspicious regions is an accurate method of detecting clinically

significant prostate cancer in men with repetitive negative biopsies and elevated PSA.

85


4

INTRODUCTION

In 2008, prostate cancer was the most frequently diagnosed cancer in men, accounting for 25%

of all cancers. The most widely used tests for the screening of prostate cancer are digital rectal

examination and the prostate specific antigen (PSA) blood-serum test. An elevated PSA is not

cancer-specific as numerous benign prostatic conditions can elevated PSA levels. Urologists are

increasingly faced with the dilemma of how best to manage a patient with elevated PSA in which

repeat Transrectal Ultrasound Guided Biopsies (TRUS-GB) reveal no cancer.

Systematic TRUS-GB of the prostate is the standard procedure for histological sampling of the

prostate. Prostate cancer is often multifocal and the volume sampled by systematic TRUS-GB

relatively small. The value of MRI in accurately localizing prostate cancer is well established(1).

The accuracy of cancer localization on anatomical T2-w imaging remains low(2), therefore

dynamic contrast enhanced MR imaging (DCE-MRI) and diffusion weighted imaging (DWI) have

been implemented and shown to improve the accuracy of prostate cancer localization. A multi-

modality approach using a combination of these techniques appears the optimal approach(3).

Imaging-guided biopsies have been advocated to improve tumour detection. However, grey-

scale TRUS, the most commonly used technique for guidance of biopsies, has a low sensitivity

for localizing prostate cancer(4). A combined approach using systematic and additional lesion

directed biopsies of contrast-enhanced suspicious areas, appears more useful(5;6).

The principal aim of our study was to determine the tumour detection yield of m-m MRI

followed by directed MR-GB in a large patient group with clinical suspicion of cancer but with

repetitive negative systematic TRUS-GBs. Furthermore we aimed to determine whether the

detected tumours were clinically significant.


Patients

Between August 2006 and March 2008, 71 consecutive patients with PSA >4 ng/ml and

negative TRUS-GB sessions (of which the last session included at least an extended scheme of 8-,

9- or 10-cores, including transition zone sampling) were referred from the departments of

Urology at the Radboud University Nijmegen Medical Centre (RUNMC) and the Canisius

86


4

Wilhelmina Hospital (CWZ) in Nijmegen, the Netherlands, for clinically routine m-m MRI of the

prostate. The Institutional Review Board approved this study.

The RUNMC histopathology database was searched for consecutive patients (January 2000 - Dec

2006) with two or more TRUS-GB sessions. Only patients who underwent at least one

systematic biopsy protocol of 8- to 10-cores, including transition zone (TZ) biopsies, were

included. At each TRUS-GB session, the principal diagnosis, age, PSA and prostate volumes were

noted. To remove the bias effect of differences in PSA and prostate volumes, tumour detection

rates for patients were compared by subgroup analysis of PSA, prostate volume and PSA density.

Localization MR Imaging

To identify possible tumour location(s), MR imaging was performed using a 3T MR scanner

(Siemens Trio Tim, Erlangen, Germany) with endorectal coil (Medrad, Pittsburgh, U.S.A) in 28

patients and the pelvic phased array coils only, in 40 patients. Axial, sagittal and coronal T2-w

images, axial DWI and DCE-MR images (using15 ml gadopentetate dimeglumine (DOTAREM,

Guerbet, Paris, France) were obtained.

Localization MR Data Analysis

Prostate images were viewed on an in-house developed analytical software workstation, which

projected the calculated DCE-MRI parameters(7) and apparent diffusion coefficient (ADC) maps

as color overlays over the T2-w images. All patient images were read in consensus by two

readers with two (T.H) and five (J.F) years experience in prostate MR imaging. MR-images were

used to determine up to three tumour suspicious regions (TSRs) for biopsy, using features

described in literature(8;9).

MR Guided biopsy

On average two weeks (range 1– 6) after the initial MRI to localize possible tumours and

identify TSR for MR-GB planning, patients received a 3T MR-GB of the prostate. Antibiotic

prophylaxis using oral ciprofloxacin 500 mg (CIPROXIN, Bayer, Leverkusen, Germany) was

given. A previously described(10) translation technique using an MR compatible biopsy device

(Invivo, Schwerin, Germany) was used for obtaining 18-gauge biopsy cores (MR-compatible

biopsy gun, Invivo, Germany) of re-identified TSRs. In brief: a needle guider attached to the arm

of the biopsy device was inserted rectally. It was subsequently adjusted to aim towards the TSR

in the prostate. Biopsies were then obtained through the needle guider. No random and only

TSRs directed biopsies were obtained. All biopsies were performed by one radiologist (TH).

87


4

Samples were subsequently processed by routine histopathological fixation and staining and

evaluated by a histopathologist.

Statistical analysis

Chi-square tests were performed to calculate for significant difference between MR-GB and

TRUS-GB subgroups. The Mann-Whitney U was performed for comparing mean age, mean PSA,

mean prostate volumes and mean PSAD between groups. A significant difference was considered

when p <0.05. Statistical analyses were performed with SPSS software (SPSS, version 16.0.01,

Chicago, U.S.A).

Evaluation of clinical significance of tumours

The clinical significance of tumours detected was determined by using currently accepted

criteria(11-14). In patients where prostatectomy was performed after a positive biopsy, a

Gleason grade 4 or 5 component, stage pT3 or tumour volume >0.5 cc were considered to

represent clinical significant disease (CSD). In patients diagnosed with cancer in whom no

prostatectomy was performed, cancer was considered significant if PSA values at biopsy were

>10 ng/ml and PSA densities were >0.15 ng/ml/cc, or a Gleason grade 4 or 5 was found on

biopsy.

RESULTS

In 70/71 patients, TSRs could be identified on MRI. One patient refused biopsies, while in

another patient no biopsies were performed because of a high bleeding risk and no certain

evidence of tumour on MRI, leaving 68 patients who received MR-GB.

The 68 patients had a mean age of 63 years (range 48-74), a median PSA of 13 ng/ml (range 4-

243) and a median of 3 (range 2-7) previous negative TRUS-GB sessions. MR-GBs of the prostate

had a median procedure duration of 30 min (range 14-75).

The tumour DR with MR-GB was 59% (40/68). In total 260 prostate cores -only directed and

no random cores- were obtained from 114 different TSRs (TSR tumour DR 40%, 46/114). The

median number of biopsies per patient was 4 (range 2 to 7). A summary of patient and

pathological findings is given in Table 1.

88


4

All patients Tumour No Tumour Patient Characteristics - Number of patients 68 40 28 - Mean Age in years (range) 63 (48-76) 65 (48-76) 62 (54-70) - Median PSA ng/ml (range) 13 (4-243) 13 (5-243) 12 (4-58) - Positive family history for prostate cancer 4 (6%) 2 (50%) 2 (50%) - Suspicious DRE 2 (3%) 1 (50%) 1 (50%) - Median number of previous TRUS-GB 3 (2-8) 3 (2-7) 3 (2-8) Imaging and Biopsy Characteristics -TSRs per patient (range) 1 (1-3) 1 (1-3) 2 (2-3) - Median biopsy cores obtained (range) 4 (2-7) 3 (2-6) 4 (2-7) - Median MR biopsy time – min. (range) 30 (14-75) 29 (14-75) 33 (15-55) - Principal histological diagnosis ~ Tumour ~ Chronic Prostatitis ~ Prostatitis + Reactive Atypia ~ Acute Prostatitis ~ Atypia suspect for malignancy ~ Atypical adenomatous hyperplasia ~ Fibromuscular Nodule ~ Necrosis ~ No abnormality detected

40

15 2 1 2 2 2 1 3

- Patient highest tumour GS distribution (40 Patients) 5 6 7 8 9

3 (7%) 19 (48%) 10 (25%) 5 (12%) 3 (8%)

- Clinically significance of disease (40 Patients with tumour) Criteria (Determined in RP) (Determined in MR-GB + no RP) ~ Tumour ~ Stage pT3 on RP ~ PSA > 10 + PSA density > 0.15 ~ Insignificant disease cannot be ruled out

18 (10) (9) 8 1 2 5

10

3

- Advanced disease ~ M+/N+ ~ Stage T3 in RP

20% (8/40) 3 5

Table 1. Summary findings of patient, radiological and pathological features. (RP =

radical prostatectomy, M = Metastasis, N = Nodal Metastasis)

89


4

Radical prostatectomy was performed in 20 of the 40 patients with tumour. Gleason score (GS)

(30%). In 10/20 a tumour volume > 0.5cc (with GS 6) was found. All of the prostatectomy

patients therefore harbored CSD. In the remaining 20 patients, external beam radiotherapy

(11/20), brachytherapy (3/20), hormonal ablation (3/20) or active surveillance (3/20) was

performed. In these patients, biopsy GS, prostatic volumes and PSA values were used to assess

the presence of CSD with 9/20 patients having a bi

PSAD > 0.15ng/ml/cc. One patient had skeletal metastasis. Therefore CSD was considered to be

present in 85% (17/20) of these patients and in 93% (37/40) of all tumour patients. Aggressive

cancer was evident in at least 48% (19/40) of all tumour The

tumour characteristics are summarized in Table 2.

2nd TRUS-GB 3rd TRUS-GB MR-GB Statistical significance

p value Number of patients ~ Tumour patients ~ Non-tumour patients

248 55 (22%) 193

65 10 (15%) 55

68 40 (59%) 28

A ** B **

Mean Age in years (range) ~ All patients ~ Tumour patients ~ Non-tumour patients

64 (74-80) 66 (52-78) 64 (52-78)

65 (48-76) 64 (57-75) 65 (48-76)

63 (48-76) 65 (48-76) 62 (54-70)

A B A B

Median PSA ng/ml (range) ~ All patients ~ Tumour patients ~ Non-tumour patients

-63) 9 (2-60) -63)

10 (2-36) 13 (6-29) 10 (2-36)

13 (4-243) 13 (5-243) 12 (4-58)

A ** A ** A ** B

Atypia in previous biopsies ~ All patients ~ Tumour patients ~ Non-tumour patients

34 (14%) 10 (29%) 24 (71%)

14 (22%) 4 (29%) 10 (71%)

14 (21%) 6 (43%) 8 (57%)

A B A B

HGPIN in previous biopsies ~ All patients ~ Tumour patients ~ Non-tumour patients

12 (5%) 4 (33%) 8 (66%)

7 (11%) 2 (29%) 5 (71%)

3 (4%) 2 (67%) 1 (33%)

A B A B

90


4

2nd TRUS-GB 3rd TRUS-GB MR-GB Signif. p-val. PSA Subgroups ~ PSA < 4 ng/ml

Total patients Tumour patients

~ PSA 4 – 10 ng/ml Total patients Tumour patients

– Total patients Tumour patients

~ – Total patients Tumour patients

~ Total patients Tumour patients

22 4 (18%) 142 29 (20%) 45 8 (18%) 16 4 (25%) 23 10 (43%)

4 0 (0%) 28 4 (11%) 20 1 (5%) 6 2 (33%) 7 3 (43%)

0 - 21 10 (48%) 17 11 (65%) 14 11 (79%) 16 8 (50%)

A ** B ** A** B ** A ** B ** A B

Prostate Volumes ~ Median prostate volume

Total

Tumour patients

~ < 30 cc Total patients Tumour patients

– 50 cc Total patients Tumour patients

– 65 cc Total patients Tumour patients


# 58 (15-201) 42 (15-160) 31 9 (29%) 67 21 (31%) 50 10 (20%) 96 14 (14%)

## 61 (17-212 30 (17-94) 7 5 (71%) 14 1 (7%) 15 1 (7%) 27 2 (7%)

48 (12-152) 42 (12-83) 14 14 (100%) 21 15 (71%) 14 7 (50%) 19 5 (26%)

A B ** A B A ** B ** A ** B ** A ** B ** A B

PSA Density ~ Median PSA density (range)

Total

Tumour patients





# 0 - - 138 19 (14%) 68 17 (25%) 28 11 (39%) 10 7 (70%)

## - - 32 2 (6%) 20 2 (20%) 6 1 (17%) 5 4 (80%)

- - 12 2 (17%) 23 13 (57%) 17 12 (71%) 16 13 (81%)

< 0.01A ** 0.01B ** 0.02A ** 0.69B

A B A **

B ** A ** B ** A B

91


4

Table 2. Comparison between TRUS-GB and MR-GB populations - tumour detection rates

a different PSA levels, prostate volumes and PSADs. (Statistical significance is calculated

for MR-GB vs. each TRUS-GB population seperately; A denotes MR-GB vs. 2nd session

TRUS-GB and B MR-GB vs. 3rd session TRUS-GB; N.A Not applicable; ** Statistically

Significant; # - in 4 patients no prostate volume obtained; ## - in 2 patients no prostate

volume obtained)

The principal tumour location was the most ventral aspect of the transition zone (TZ) in 57%

(26/46), followed by the paramedian region of the peripheral zone (PZ) in 20% (9/46) and

anterior horns of the PZ in 11% (5/46) (Figure 1). Figure 5 shows the MR images of a patient in

whom tumour was detected with MR-GB.

Figure 1. Schematic presentation of the prevalence of the 46 different tumour positive

TSRs (for 9 biopsies, two adjacent regions were both positive giving 55 tumour maps)

within the prostate as detected with MR-GB. The prostate was divided into 5 cranial-

caudal segments equating to a) apex b) apex-mid c) mid d) mid-base and e) basal level.

R=Right; L=Left; VT=Ventral Transition zone; MT=Middle Transition zone; DT=Dorsal

Transition zone; AH=Anterior Horn of peripheral zone; DL=Dorsolateral peripheral zone;

D=Dorsal peripheral zone.

92


4

From our reference database, 248 patients were identified with at least 2 TRUS-GB sessions and

65 patients with 3 sessions. No MR imaging was performed prior to biopsy in these subjects and

biopsies were performed on a systematic basis only. The overall tumour DR at the second and

third biopsy sessions were 22% (55/248) and 15% (10/65) respectively. The comparison of

tumour DR for TRUS-GB and MR-GB subgroups, stratified according to PSA, prostate volume and

PSAD are given in Table 2. MR-GB achieved significantly higher tumour DRs for all PSA

ts with a

PSA >20 ng/l (Figure 2), prostate volumes >65 cc (Figure 3 l/cc and

), where superior results were evident but not significant (p > 0.05).

One self-limiting transurethral hemorrhage and one uncomplicated urinary tract infection were

the only procedure related complications.

Figure 2. Tumour detection rates (%) in different PSA subgroups, at 2nd, 3rd TRUS-GB and

MR-GB session.

93


4

Figure 3. Tumour detection rates (%) in different prostate volume subgroups, at 2nd, 3rd

TRUS-GB session and MR-GB session.

Figure 4. Tumour detection rates (%) in different prostate PSA density subgroups, at 2nd,

3rd TRUS-GB session and MR-GB session (PSA density values in ng/ml/cc).

94


4

DISCUSSION

By using state-of-the-art multi-modality 3T MRI for tumour localization, we have shown that a

definite diagnosis of prostate cancer could be made in 59% (40/68) of patients in whom

repetitive prostate biopsies remained negative, but continuous concern regarding the presence

of cancer was evident. Ninety-three percent (37/40) of patients diagnosed with prostate cancer

were considered to harbor CSD. It is therefore justifiable to deduce, that MRI of the prostate

accurately portrays the locations of tumours, and thus offers urologists a method to improve

their biopsy outcomes. As MR-GB is limited by the restricted general availability, other methods

of MRI targeted biopsy techniques, such as MR-TRUS fusion(15) during TRUS biopsies, could be

considered. Nevertheless, MR-GB is probably the most accurate technique, because translation

of the TSR to another imaging modality is not required.

For comparison, we selected a TRUS-GB population from our institution which was clinically

matched for: age, prevalence of atypia in previous biopsies, PSA, prostate volume and PSAD. We

also determined the tumour DR for each subgroup. Most literature reporting on extended

schemes of 8-12 cores detect cancer in around 10-17%(16-18) of patients at second biopsy. The

overall tumour detection rate of 22% at the second and 15% at third TRUS-GB session in our

institution is therefore in agreement with reported data.

As different PSA values can predict the likelihood of finding tumour and constitutes a bias for

comparison, patients were substratified according to different PSA levels. Our study shows that

the MR-GB DR is superior (p< 01) to the repeat TRUS-GB sessions in all PSA subgroups except

in the very high PSA of >20 ng/ml group, where a similar DR was achieved (50% vs. 43%).

Prostatic volume constitutes another important factor that plays a role in the tumour detection

rate achieved by different biopsy protocols. In previous series it appears that 8-cores are

appropriate in patients with prostates <30 cc, whereas 10-12 cores are needed in 30-50 cc sized

prostates and >12-cores in large prostates of >50 cc(19). For all prostate volume groups, MR-GB

significantly outperformed TRUS-GB in tumour detection , except in excessively large

prostates of >65 cc, where similar rates were achieved.

In patients with tumour, a 15 ng/ml/cc is considered a good prognostic feature, with

low rates of progression(13). MR-GB did not achieve significant detection improvements over

TRUS-GB in this subset of patients with probable insignificant disease. On the contrary, in the

50 ng/ml/cc, tumours are likely to have a larger volume and therefore more

easily diagnosed with TRUS-GB.

95


4

Figure 5. Endorectal coil 3T MR images of a 64 yr old male with 4 previous negative

TRUS-GB (incl. 2 x 8-, 10-, 12-core) and a PSA of 18 ng/ml. T2-weighted axial (a) and

coronal (b) images show a low signal intensity lesions in the ventral portion of the apex.

This area also shows restriction on the ADC map (c) as well as a high Ktrans on DCE-MRI

(d). Axial (e) T2-w TRUE-FISP images during the MR-GB session, showing the needle

guider directed towards the TSR. Histopathological analysis of the biopsy cores, revealed

an adenocarcinoma with GS 4+3=7. Subsequent pelvic lymph node dissection revealed

metastatic disease.

96


4

In case of negative TRUS-GBs, radical measures involving saturation biopsies of 24-64 cores

have been advocated, with reported detection rates of between 18-34%(20;21) at second

session. No widespread application and acceptance of this technique by urologists exist with

conflicting results having been published(22;23). Moreover, saturation biopsies do appear to

offer a method of increasing tumour detection in high-risk patients however the additional use

of analgesia/anaesthesia, the higher incidence of side-effects and the high cost of processing the

large amount of pathological material are the biggest drawbacks of these techniques. As our

study has shown that MR-GB has a high tumour detection yield, requiring a very low number of

cores (median 4), this method could offer a very appealing alternative to the patient, urologist

and pathologist alike.

Whether MR-GB of the prostate detects a substantial proportion of potentially insignificant

tumours, is a very legitimate question. It stands to reason that a higher sensitivity for detecting

tumours implies a higher chance of finding tumours that do not need treatment, so called

clinically insignificant cancers. For prostate cancer the discussion on overtreatment of these

tumours remains a controversial issue(24). The concept of ‘insignificant’ prostate cancer, based

on tumour size and favourable pathological characteristics, was proposed in the 1990’s(25), and

the clinical criteria for predicting such tumours were 10 ng/ml, GS

pT2 and tumour 5 cc(11;12;26).

According to prostatectomy series, the predominant location of tumours, is the peripheral zone

in almost 70% of cases(27). Therefore, current systematic biopsy schemes extensively sample

the peripheral zone and thus the dorsal region of the prostate. In contrast, 68% (31/46) of

tumours in our series were very anteriorly located, 57% ( 26/46) in the ventral TZ and 11%

(5/46) in the anterior horns of the PZ. This may be the explanation within our group of patients

for having previous negative biopsies.

The current literature on MR guided biopsies of the prostate(28-30) is sparse and the few

reports currently available included small numbers of patients, had excessively long imaging and

biopsy times, and reported only on the use of conventional T2-weighted MR imaging to

determine TSRs for MR-GB after one previous negative TRUS biopsy.

Limitations of the current study relate principally to the fact that a direct comparison of our

results to other literature could not be made. This was because of differences in PSA values,

prostatic volumes, the number of previous biopsies as well as the biopsy schemes used during

the initial sessions. A prospective randomized trial would have appeared superior. However to

determine the potential benefit, we compared our study patient cohort with our own

97


4

institutional database, selecting similar patients but with TRUS guided multiple biopsies and

sub-grouped according to PSA, prostate volume and PSAD. Since our institution is a referral

hospital, the MR-GB patients had very heterogeneous previous biopsy protocols, with the

highest number of cores per biopsies session, ranging from 8-, 9-, 10-, 12- , 18 core and even

saturation biopsies. Patients that were selected for inclusion based on PSA of >10 ng/ml, can

represent a selection bias in relation to determining clinical significance of the detected

tumours.

CONCLUSIONS

In conclusion this study most importantly indicates that MRI is a highly effective method for the

detection and localization of clinically significant prostate cancer. As we have shown that guided

biopsies towards TSRs on MRI, detect clinically significant tumour in a substantial portion of

patients, MRI should be considered essential in any workup protocol of patients who are

suspected of harboring malignancy but who have successive negative biopsies. Secondly, this

study also concludes that MR-GB directed towards TSRs on m-m MRI is a very useful method of

accurately validating correct sampling of suspicious prostatic tissue. Because of the low

numbers of cores needed, MR-GB appears an appealing alternative to procedures such as

saturation biopsies. Finally we have shown that tumours detected were mostly located in areas

not explicitly sampled by routine schemes. Future studies for tumour DR using MR-TRUS fusion

during TRUS-GB incl. saturation targeting of suspicious areas, transperineal sampling of the

anterior prostate or changing locations for sampling under TRUS-GB in repeat sessions patients,

are needed and ideally should be compared to an MRI directed MR-GB technique.

98


4

REFERENCES

1. Umbehr M, Bachmann LM, Held U, Kessler TM, Sulser T, Weishaupt D, Kurhanewicz J, Steurer J. Combined Magnetic Resonance Imaging and Magnetic Resonance Spectroscopy Imaging in the Diagnosis of Prostate Cancer: A Systematic Review and Meta-analysis. Eur.Urol. 2008 Oct 18.



4. Halpern EJ, Strup SE. Using gray-scale and color and power Doppler sonography to detect prostatic cancer. AJR Am.J.Roentgenol. 2000 Mar;174(3):623-7.

5. Mitterberger M, Pinggera GM, Horninger W, Bartsch G, Strasser H, Schafer G, Brunner A, Halpern EJ, Gradl J, Pallwein L, et al. Comparison of contrast enhanced color Doppler targeted biopsy to conventional systematic biopsy: impact on Gleason score. J.Urol. 2007 Aug;178(2):464-8.

6. Pelzer A, Bektic J, Berger AP, Pallwein L, Halpern EJ, Horninger W, Bartsch G, Frauscher F. Prostate cancer detection in men with prostate specific antigen 4 to 10 ng/ml using a combined approach of contrast enhanced color Doppler targeted and systematic biopsy. J.Urol. 2005 Jun;173(6):1926-9.

7. Huisman HJ, Engelbrecht MR, Barentsz JO. Accurate estimation of pharmacokinetic contrast-enhanced dynamic MRI parameters of the prostate. J.Magn Reson.Imaging 2001 Apr;13(4):607-14.



10. Hambrock T, Futterer JJ, Huisman HJ, Hulsbergen-vandeKaa C, van Basten JP, van O, I, Witjes JA, Barentsz JO. Thirty-two-channel coil 3T magnetic resonance-guided biopsies of prostate tumour suspicious regions identified on multimodality 3T magnetic resonance imaging: technique and feasibility. Invest Radiol. 2008 Oct;43(10):686-94.

11. Augustin H, Hammerer PG, Graefen M, Erbersdobler A, Blonski J, Palisaar J, Daghofer F, Huland H. Insignificant prostate cancer in radical prostatectomy specimen: time trends and preoperative prediction. Eur.Urol. 2003 May;43(5):455-60.

12. Bastian PJ, Mangold LA, Epstein JI, Partin AW. Characteristics of insignificant clinical T1c prostate tumours. A contemporary analysis. Cancer 2004 Nov 1;101(9):2001-5.

13. Nakanishi H, Wang X, Ochiai A, Trpkov K, Yilmaz A, Donnelly JB, Davis JW, Troncoso P, Babaian RJ. A nomogram for predicting low-volume/low-grade prostate cancer: a tool in selecting patients for active surveillance. Cancer 2007 Dec 1;110(11):2441-7.

14. Ochiai A, Troncoso P, Babaian RJ. The relationship between serum prostate specific antigen level and tumour volume persists in the current era. J.Urol. 2007 Mar;177(3):903-6.

15. Xu S, Kruecker J, Turkbey B, Glossop N, Singh AK, Choyke P, Pinto P, Wood BJ. Real-time MRI-TRUS fusion for guidance of targeted prostate biopsies. Comput.Aided Surg. 2008 Sep;13(5):255-64.

16. Mian BM, Naya Y, Okihara K, Vakar-Lopez F, Troncoso P, Babaian RJ. Predictors of cancer in repeat extended multisite prostate biopsy in men with previous negative extended multisite biopsy. Urology 2002 Nov;60(5):836-40.

17. Philip J, Hanchanale V, Foster CS, Javle P. Importance of peripheral biopsies in maximising the detection of early prostate cancer in repeat 12-core biopsy protocols. BJU.Int. 2006 Sep;98(3):559-62.

18. Roehl KA, Antenor JA, Catalona WJ. Serial biopsy results in prostate cancer screening study. J.Urol. 2002 Jun;167(6):2435-9.

19. Eskicorapci SY, Guliyev F, Akdogan B, Dogan HS, Ergen A, Ozen H. Individualization of the biopsy protocol according to the prostate gland volume for prostate cancer detection. J.Urol. 2005 May;173(5):1536-40.

20. Campos-Fernandes JL, Bastien L, Nicolaiew N, Robert G, Terry S, Vacherot F, Salomon L, Allory Y, Vordos D, Hoznek A, et al. Prostate Cancer Detection Rate in Patients with Repeated Extended 21-Sample Needle Biopsy. Eur.Urol. 2008 Jun 23.

21. Pepe P, Aragona F. Saturation prostate needle biopsy and prostate cancer detection at initial and repeat evaluation. Urology 2007 Dec;70(6):1131-5.

99


4

22. Stav K, Leibovici D, Sandbank J, Lindner A, Zisman A. Saturation prostate biopsy in high risk patients after

multiple previous negative biopsies. Urology 2008 Mar;71(3):399-403.

23. Walz J, Graefen M, Chun FK, Erbersdobler A, Haese A, Steuber T, Schlomm T, Huland H, Karakiewicz PI. High incidence of prostate cancer detected by saturation biopsy after previous negative biopsy series. Eur.Urol. 2006 Sep;50(3):498-505.

24. Taylor JA, III, Gancarczyk KJ, Fant GV, McLeod DG. Increasing the number of core samples taken at prostate needle biopsy enhances the detection of clinically significant prostate cancer. Urology 2002 Nov;60(5):841-5.

25. Dugan JA, Bostwick DG, Myers RP, Qian J, Bergstralh EJ, Oesterling JE. The definition and preoperative prediction of clinically insignificant prostate cancer. JAMA 1996 Jan 24;275(4):288-94.

26. Miyake H, Sakai I, Harada K, Hara I, Eto H. Prediction of potentially insignificant prostate cancer in men undergoing radical prostatectomy for clinically organ-confined disease. Int.J.Urol. 2005 Mar;12(3):270-4.

27. Sakai I, Harada K, Hara I, Eto H, Miyake H. A comparison of the biological features between prostate cancers arising in the transition and peripheral zones. BJU.Int. 2005 Sep;96(4):528-32.


29. Beyersdorff D, Winkel A, Hamm B, Lenk S, Loening SA, Taupitz M. MR imaging-guided prostate biopsy with a closed MR unit at 1.5 T: initial results. Radiology 2005 Feb;234(2):576-81.

30. Engelhard K, Hollenbach HP, Kiefer B, Winkel A, Goeb K, Engehausen D. Prostate biopsy in the supine position in a standard 1.5-T scanner under real time MR-imaging control using a MR-compatible endorectal biopsy device. Eur.Radiol. 2006 Jun;16(6):1237-43.

100


4

— CHAPTER 5 —

3T DCE-MRI Guided Biopsy in Localizing Prostate Cancer Recurrence after Radiation

Therapy

D. Yakar; T. Hambrock; H. Huisman et al.

CHAPTER 5CHAPTER 5

Feasibility of 3 Tesla Dynamic Contrast Enhanced Magnetic Resonance Guided Biopsy in Localizing Local Prostate Cancer Recurrence after

Radiation Therapy

Investigative Radiology 2010 Mar; 45(3):121-5

Yakar D, Hambrock T, Huisman H, Hulsbergen-vandeKaa CA, van Lin E, Vergunst H, Hoeks CM, van Oort IM, Witjes JA, Barentsz JO, Fütterer JJ

Young Investigators Award European Society for Urogential Radiology, München,

2008 (Derya Yakar)


Dynamic contrast enhanced MR imaging is a very accurate method in detecting prostate

cancer recurrence following external beam radiotherapy.

It is feasible to perform MR guided biopsies of patients with abnormal MRI after

radiotherapy to make a definite diagnosis of local recurrence.


In patients with elevated PSA following external radiotherapy, DCE-MRI can play an

important role in diagnosing local recurrence and therefore guide therapy for systemic

vs. local therapy.

Summary Statement

3T DCE-MRI followed by MR guided biopsies is a feasible and useful technique for diagnosing local recurrence of prostate cancer following

external beam radiotherapy.

102

3T DCE-MRI Guided Biopsy in Localizing Prostate Cancer Recurrence after Radiation Therapy

5

ABSTRACT

Objectives: The objective of this study was to assess the feasibility of the combination of

magnetic resonance (MR)-guided biopsy (MRGB) and diagnostic 3T MR imaging in the

localization of local recurrence of prostate cancer (PCa) after external beam radiation therapy

(EBRT).

Materials and Methods: Twenty-four consecutive men with biochemical failure suspected of

local recurrence after initial EBRT were enrolled prospectively in this study. All patients

underwent a diagnostic 3T MR examination of the prostate. T2-weighted and dynamic contrast-

enhanced MR images (DCE-MRI) were acquired. Two radiologists evaluated the MR images in

consensus for tumour suspicious regions (TSRs) for local recurrence. Subsequently, these TSRs

were biopsied under MR-guidance and histopathologically evaluated for the presence of

recurrent PCa. Descriptive statistical analysis was applied.

Results: Tissue sampling was successful in all patients and all TSRs. The positive predictive

value on a per patient basis was 75% (15/20) and on a per TSR basis 68% (26/38). The median

number of biopsies taken per patient was 3, and the duration of an MRGB session was 31

minutes. No intervention-related complications occurred.

Conclusions: The combination of MRGB and diagnostic MR imaging of the prostate was a

feasible technique to localize PCa recurrence after EBRT using a low number of cores in a

clinically acceptable time.

103


5

INTRODUCTION

Approximately 30% of the patients diagnosed with prostate cancer (PCa) are treated with

external beam radiation therapy (EBRT) as initial definitive treatment (1). Of these patients,

20% to 60% develop biochemical failure (2). Biochemical failure is considered to represent

cancer recurrence (3) and correlates well with clinical progression as evidenced by local clinical

control rates after 5 years in patients with and without biochemical failure of 86% and 99%,

respectively (4).

The diagnosis of recurrent PCa entails a prostate-specific antigen (PSA) test, digital rectal

examination (DRE), transrectal ultrasound (TRUS)-guided prostate biopsy, and a bone scan.

However, each of these diagnostic tools has definite shortcomings (5). The PSA nadir, PSA half-

life, time interval from treatment to PSA nadir, and PSA doubling time are all associated with

both clinical and biochemical failure (6). Yet, there is no absolute cut-off value to accurately

discriminate on an individual basis, between local recurrence and distant metastases (7).

Diagnosing local recurrence after EBRT by DRE is challenging because of radiation-induced

fibrosis and shrinkage of the prostate, whereas the sensitivity and specificity of TRUS after

radiotherapy were found to be 49% and 57%, respectively (8). Magnetic resonance (MR)

imaging of the prostate is a well established modality for accurately localizing PCa. Evaluation of

the radiated prostate gland is restricted because of treatment-induced changes, resulting in

homogeneously low signal on T2-w images (9,10). Hence, MR imaging of the prostate after EBRT

is much more challenging.

Functional imaging techniques, like dynamic contrast enhanced MR imaging (DCE-MRI)(11,12),

diffusion-weighted imaging, and proton MR spectroscopy, can improve the accuracy of localizing

recurrence after EBRT (13–15). This diagnostic information can be used for directing biopsies to

tumour suspicious regions (TSRs). Thus far, TRUS is the most widely used imaging modality, and

systematic sextant TRUS-guided biopsy is considered as the gold standard for histologic

assessment of a local recurrence, with a positive predictive value (PPV) of only 27% (8).

However, this may be improved by DCE-MRI with a PPV of 46% to 78% (11,12). To our

knowledge, there are no studies available on 3T MR-guided biopsy (MRGB) in the localization of

local recurrence of PCa after EBRT.

The purpose of our study was to assess the feasibility of the combination of MRGB and

diagnostic 3T MR imaging in the localization of local recurrence of PCa after EBRT.

104


5


Patients

This study was approved by the ethics review board of our institution, and informed consent

was obtained. From October 2006 to March 2009, 24 consecutive patients (median age, 70

years; range, 60–83) with biochemical failure (using the ASTRO definition of 3 consecutive rises

in PSA after reaching PSA nadir) after initial EBRT were enrolled in this prospective study.

Patients were referred for MR imaging of the prostate followed by MRGB. Exclusion criteria

were contraindications to MR imaging (eg, cardiac pacemakers, intracranial clips). The pre- and

postradiotherapy characteristics are given in Table 1.

Median age (yr) 70 (60-83)

Preradiotherapy stage

T1

T2

T3

3

4

13

Median preradiotherapy Gleason Score 7 (5-9) 2

Median preradiotherapy PSA (ng/ml) 15.6 (6.1-96.0) 1

Median radiation therapy dose (Gy) 67.5 (66.0-78) 3

No. patients who received hormonal therapy 15

Median PSA nadir (ng/ml) 0.26 (0.0-6.3)

Median PSA (ng/ml) prior to MRI 4.4 (1.1-13.4)

Median time (wk) between MRI and MR biopsy 4.1 (1.3-10.0)

Table 1. The pre- and postradiotherapy characteristics of the patients

MR Imaging Protocol

MR imaging of the prostate was performed using a 3T MR scanner (Siemens Trio Tim, Erlangen,

Germany) with the use of a phased array coil. The ERC was inserted and filled with either a 40

ml perfluorocarbon or water preparation (FOMBLIN, Solvay-Solexis, Milan, Italy). Peristalsis was

suppressed with an intravenous injection of 20 mg butylscopolamine bromide (BUSCOPAN,

Boehringer-Ingelheim, Ingelheim, Germany), intramuscular injection of 20 mg

butylscopolaminebromide and 1 mg of glucagon (GLUCAGEN, Nordisk, Gentofte, Denmark).

105


5

The imaging protocol included the following sequences: T2-weighted turbo spin echo sequences

were acquired (TR 4260 milliseconds/ TE 99 milliseconds; flip angle, 120; 3 mm slice thickness;

echo train length, 15; 180x90 mm field of view, and 448x448 matrix; voxel size, 0.4x0.4x 3 mm)

in axial, coronal, and sagittal planes. A 3-dimensional (3D) T1-weighted gradient echo sequence

(TR 4.90 milliseconds/TE 2.45 milliseconds; flip angle, 10; 176 slices per 3D slab; 0.9 mm slice

thickness; 288 x 288 field of view and 320 x 320 matrix; voxel size 0.9 mm x 0.9 mm x 0.9 mm)

was used to assess lymph node and skeletal status. Finally, an axial 3D T1-weighted gradient

echo sequence (TR 800 milliseconds/TE 1.47 milliseconds; flip angle, 14; 3 mm slice thickness;

field of view 230 x 230 and 128 x 128 matrix; voxel size 1.8 mm x 1.8 mm x 3 mm) was used to

obtain proton-density images, with the same positioning angle and center as the axial T2-

weighted sequence (to allow calculation of the relative gadolinium chelate concentration

curves), followed by 3D T1-weighted spoiled gradient-echo images (TR 38 milliseconds/TE 1.35

milliseconds; flip angle, 14; 10 transverse partitions on a 3D slab; 3 mm section thickness; 230 x

230 mm field of view; 128 x 128 matrix; voxel size 1.8 x 1.8 x 3 mm; GRAPPA parallel imaging

factor 2; 2.5s temporal resolution; and 2 minutes 30 seconds acquisition time) acquired during

an intravenous bolus injection of a paramagnetic gadolinium chelate— 0.1 mmol of

gadopentetate dimeglumine (DOTAREM, Guerbet, Paris, France) per kilogram of body weight.

This was administered with a power injector (Spectris; Medrad) at 2.5 ml/s and followed by a

20-mL saline flush.

MR Data Analysis

The prostate images of all patients were read in consensus by 2 radiologists with respectively 2

and 5 years of experience in prostate MR imaging. Functional dynamic imaging parameters were

estimated from a fitted general bi-exponential signal intensity model for each MR signal

enhancement–time curve, as described previously (16,17). The pharmacokinetic parameters

(Ktrans, Ve, Kep, and WashOut) were computed using the standard 2-compartment model (18)

and the arterial input function was estimated using the reference tissue method (19) and

automated per-patient calibration (20). Finally, these parameters were projected as color

overlay maps over the T2-weighted images.

This patented procedure (21) for calculating pharmacokinetic parameters is being used in

multiple centers. For the first 2 patients, lesions were identified as suspicious purely whether a

focally enhancing region with DCE-MRI was seen irrespective of the degree of enhancement. To

determine whether regions enhancing in the prostate were representing tumour, we applied a

lesion directed biopsy approach. To establish a “cut-off” in the color coding of the

106


5

pharmacokinetic maps, we used the “normal” regions of these 2 patients as a cut-off for future

color coding. Above this “normal” threshold, radiologists then defined suspicious regions as

focally enhancing spots shown on the color maps.

The criterion for TSRs on DCE-MRI in the peripheral zone, the transition zone, and the seminal

vesicles was a cutoff value of 3.5 (/s) for Ktrans and -0.225 (AU) for washout. The criterion for

TSRs on T2-weighted MR images was a low signal-intensity region within the prostate. Per

patient, the T2-w images were evaluated for TSRs individually and in color overlay (DCE-MRI).

To locate the TSRs, the prostate was divided into 22 different axial and sagittal segments, to

make a 3D spatial position estimation of the identified TSRs which was used during the second

MRGB-session. A similar translation technique was described before by Hambrock et al. (22).

MR-Guided Biopsy

After the initial tumour localization MR examination (median time between biopsy and initial

localization MRI was 4.1 weeks; range, 1.3–10.0), patients underwent an MRGB using an MR-

compatible biopsy device (In vivo, Schwerin, Germany) at 3T. All patients received oral

ciprofloxacin 500 mg (CIPROXIN, Bayer, Leverkusen, Germany) the evening before, in the

morning of the biopsy, and 6 hours after biopsy.

Relocation of the TSRs (by using the 3D spatial position estimation) determined during the first

MR imaging localization was done by obtaining T2-weighted anatomic images in the axial

direction. Prostate biopsies were performed with the patient in prone position, and a needle

guider inserted rectally, which was attached to the arm of the biopsy device. The needle guider

was pointed toward the TSR before obtaining the biopsy specimen (22). All biopsies were

supervised by one experienced radiologist (4 years of experience) with MRGB of the prostate.

Histopathology

Samples were subsequently processed by a routine fixation in 10% buffered formalin, embedded

in paraffin, stained with hematoxylin-eosin, before being evaluated by an experienced

genitourinary pathologist (18 years experience of PCa histopathology) for the presence of

tumour. The biopsies were classified as negative if there was no evidence of carcinoma or

residual indeterminate carcinoma with severe treatment effect, defined as isolated tumour cells

or poorly formed glands with abundant clear or vacuolated cytoplasm. All positive biopsies were

assigned a Gleason score.

107


5

Statistical Analysis

Descriptive statistical analysis was applied. The positive predictive value of TSRs seen on MR

images was calculated. The patient characteristics were calculated by using SPSS 16.0

RESULTS

Three patients with contraindications to an ERC (e.g., anorectal surgery, inflammatory bowel

disease) were scanned with a pelvic phased-array coil only. Metastatic disease was evident on

MR imaging in 4 of 24 patients. One patient had multiple low signal intensity areas in the left

iliac and sacral bone and multiple enlarged lymph nodes (diameter greater than 10 mm) next to

the right internal iliac artery. Two patients had multiple enlarged lymph nodes next to the right

internal iliac artery. One patient had multiple areas with focal low signal intensities in the body

of L4, the right acetabulum, and in the right anterior superior iliac spine. These patients received

hormonal therapy and were excluded in the further analysis.

In the remaining 20 patients, a total of 38 TSRs were identified on combined T2-weighted and

DCE-MRI and subsequently biopsied with MRGB. All TSRs that were identified on T2-weighted

imaging were also identified on DCE-MRI. With DCE-MRI 8 TSRs, in 5 patients, were identified

that were not identified on T2-weighted imaging. One patient had a hyperintense region

compared with uninvolved prostate tissue as TSR on T2-weighted MR imaging and increased

permeability on DCE-MRI (Fig. 1), which turned out to be recurrent PCa on histologic

examination.

Median MRGB time was 31 minutes (range, 18–47) per patient and a median of 3 biopsy cores

per patient (range, 2–5) was obtained. Median number of biopsies per TSR was 2 (range, 1–4).

The MRGBs were tolerated well and no procedure-related complications occurred. Tissue

sampling was successful in all patients and TSRs.

Histologically proven local recurrence was evident in 15 patients (Fig. 2). These patients were

either treated with salvage cryosurgery (5 patients), salvage prostatectomy (1 patient), wait and

see (1 patient), hormonal therapy (2 patients), were lost to follow-up (4 patients), or died (2

patients). Of the 38 different TSRs identified on MR imaging, 26 contained histologically proven

recurrence (68%), 8 revealed radiotherapy induced atypia in preexisting glands (21%), 1

contained residual indeterminate PCa with severe radiation changes (3%), and the remaining 3

contained fibrosis (8%).

108


5

Figure 1. MR images obtained from a 70 year old man with a TSR seen in the right ventral

part in the midprostate, with a PSA of 0.4 ng/ml. The T2-weighted images showed a

hyperintense TSR in the right ventral part in the midprostate (arrow) (A), also on DCE-

MRI there was a TSR (high Ktrans) visible (arow) (B). During a second session, an MRGB

was performed, the needle guide was pointed toward the TSR in axial (TRUE-FISP image)

(C) plane, and subsequently biopsied. Histopathology revealed a Gleason 9 prostate

cancer recurrence.

The PPV of MRGB for detecting local recurrence on per patient basis and per TSR basis was 75%

(15/20) and 68% (26/38), respectively. There was no significant difference of the PPV when

peripheral, transition zone, and seminal vesicles were considered separate and also the use of a

pelvic phased-array coil only had no influence on these results. Gleason score 10 was present in

109


5

2/26 (8%), Gleason score 9 in 8/26 (31%), Gleason score 8 in 3/26 (12%), Gleason score 7 in

9/26 (35%), and Gleason score 6 in 4/26 (15%) TSRs.

The site of recurrence within the prostate was present in the apical region in 6 of 26 TSRs, in the

apex-mid in 5, in the mid in 9, in the midbase in 3, in the base in 1, and in the seminal vesicle in 2

TSRs. The local recurrence was identified in the peripheral zone in 19 of 26 (73%) TSRs and in

the transition zone in 7 of 26 (27%) TSRs.

Of the 5 patients, 3 with negative histology received hormonal therapy, 1 underwent salvage

cryosurgery, and 1 underwent a follow-up MR examination. The patient with the follow-up MR

had unchanged MR findings in combination with a declining PSA (PSA during the first MR

examination was 6.4 ng/ml and during the follow-up MR imaging was 3.7 ng/ml).

DISCUSSION

Results of our study show that local recurrence after EBRT could be localized with the

combination of MRGB and diagnostic MR imaging in a substantial proportion of patients (PPV of

8% and 75% on a per TSR and a per patient basis, respectively). With a median intervention

time of 31 minutes, and no procedure-related complications, MRGB can be considered a feasible

method in localizing local PCa recurrence following EBRT.

MRGB was not able to assess the effect of false negatives, i.e., areas of prostate cancer which did

not enhance, because of the relative time-consuming character of this procedure. However,

previous studies that have used 6 core TRUS-guided biopsy as the standard of reference in

localizing radiation therapy recurrence, including from regions that did not enhance on DCE-

MRI, have shown that the negative predictive value of this technique is between 78% and 95%

(11,12). Undoubtedly, 6 core TRUS-guided biopsies have many limitations when used as the gold

standard (e.g., high false negatives and underestimating of true Gleason score), which probably

means that these negative predictive values of the above mentioned studies are somewhat

overrated.

Correlating DCE-MRI with radical prostatectomy samples would be the best option and should

be the next step in correlating recurrent disease seen on DCE-MRI. Nonetheless, when selecting

patients with prostate cancer recurrence the mostly followed treatment strategy is still some

form of whole-gland salvage therapy. Hence, false negatives are less of a problem. Merely

detecting a recurrence, rather than mapping the recurrent tumour, will suffice for this particular

110


5

group of patients. Thus, only biopsying TSRs seen on DCE-MRI should be considered as an

advantage rather than a disadvantage of MRGB biopsies.

MRGB is capable of detecting a recurrence with only 3 biopsies per patient omitting TRUS biopsy

core schemes (with a PPV of 27%) that use between 6 and 12 cores per patient. Consequently, it

may lead to higher patient satisfaction. The main advantage of TRUS-guided biopsy over 3T DCE-

MRGB is that the expertise and the technique itself are more generally available in routine

clinical practice. Probably, performing MR-TRUS fusion for targeted biopsies by combining the

high spatial resolution of MR imaging with the wide-spread availability of TRUS is the most

optimal strategy as evidenced by promising results in 2 different studies (23,24).

However, MRGB has the advantage of being directed toward TSRs using the same imaging

modality used for localization. This way spatial misregistration between MR imaging and the

biopsy core can be reduced to a minimum, which probably makes it a more precise method.

Unfortunately, randomized controlled trials comparing MRGB, TRUS-guided biopsy, and MR-

TRUS fusion are not available yet. Future studies should focus on inclusion of a larger number of

patients and to improve the reproducibility of quantitative pharmacokinetic parameters (25) as

well as to limit the subjectiveness of reader evaluation, an automated per-patient arterial input

function estimation, which was shown in a recent publication to be superior to fixed input

models (20) is more preferred.

The widely known and accepted criterion for prostate cancer on T2-weighted MR imaging, a

region of low signal intensity, was used by us for localizing recurrence of PCa. It is interesting to

note that we had a specific case with a hyperintense region compared with surrounding prostate

tissue as a TSR, which was confirmed on histologic evaluation as PCa recurrence. This indicates

that after EBRT a recurrence can also have a hyperintense character, compared with

surrounding prostate tissue on T2-weighted MR imaging. Furthermore, future studies may

include other functional MR imaging techniques such as diffusion-weighted imaging26 and MR

spectroscopy in guiding MRGB after EBRT. This may lead to a higher detection rate with a

minimum number of biopsy cores.

111


5

Figure 2. MR images obtained from a 68-year-old man with a TSR seen in the right

peripheral zone at the apex-mid, with a PSA of 2.1 ng/mL. The T2-weighted images

showed a diffuse low signal intensity of the entire prostate and no TSR was detectable (A);

however, on DCE-MRI, there was a TSR (high Ktrans) visible (arrow) (B). During a second

session, an MRGB was performed, by pointing the needle guider toward the TSR in axial

(TRUE-FISP image) (C), plane and subsequently biopsied. Histopathology revealed a

Gleason 7 prostate cancer recurrence.

Limitations of our study are related to the relatively small number of patients included and the

incomplete data concerning patient characteristics, which was due to patient referral from

outside our university hospital. In our study, we did not have any patient with a negative DCE-

MRI on local prostate level. In other words, each patient we imaged had at least one TSR in the

prostate detected with DCE-MRI. This can be interpreted as a selection bias. However, this is not

112


5

surprising because of the inclusion criteria being 3 consecutive rises of PSA after reaching PSA

nadir.

Only 3 patients were examined without the use of an ERC. Because of the small number of this

group, no further conclusions can be drawn from this result. Because our current study is

prospective and no other studies exist on defining true pharmacokinetic cut-off values for the

irradiated prostate we had to base our cut-off values on the first 2 patients we imaged and

biopsied. Defining a cut-off for tumour is impossible with only 2 patients. These 2 were only

used to define a cut-off for “normal.” Above this “normal” threshold, radiologists then defined

suspicious regions as focally enhancing spots shown on the color maps. To define true

pharmacokinetic cut-off values for tumour versus benign enhancing spots versus normal in the

irradiated prostate is subjective of a future publication and can only be determined

retrospectively on a larger group of patients.

Salvage therapies can offer a possibility of cure in selected patients. This underlines the

importance to select the patients who would benefit from it carefully (27). Unfortunately, the

conventional methods for diagnostic workup of PCa recurrence (e.g., DRE, PSA, TRUS, bone scan)

all have their limitations. Therefore, there is a need for more accurate and versatile diagnostic

tools like MR imaging, which has the advantage that both local and distant prostatic disease can

be evaluated at the same time. Moreover, the addition of other functional MR imaging techniques

such as DWI can even possibly play a role in the assessment of tumour aggressiveness (28).

CONCLUSION

In conclusion, the combination of MRGB and diagnostic MR imaging of the prostate was a

feasible technique to localize PCa recurrence after EBRT using a low number of cores in a

clinically acceptable time.

113


5

REFERENCES

1. Stanford JL, Stephenson RA, Coyle LM, et al. Prostate Cancer Trends 1973–1995, SEER Program. Bethesda, MD: National Cancer Institute; 1999.

2. D’Amico AV, Whittington R, Malkowicz SB, et al. Biochemical outcome after radical prostatectomy or external beam radiation therapy for patients with clinically localized prostate carcinoma in the prostate specific antigen era. Cancer. 2002;95:281–286.

3. Aus G, Abbou CC, Bolla M, et al. EAU guidelines on prostate cancer. Eur Urol. 2005;48:546 –551. 4. Horwitz EM, Vicini FA, Ziaja EL, et al; American Society of Therapeutic Radiology and Oncology. The correlation

between the ASTRO Consensus Panel definition of biochemical failure and clinical outcome for patients with prostate cancer treated with external beam irradiation. Int J Radiat Oncol Biol Phys. 1998;41:267–272.

5. Nudell DM, Wefer AE, et al. Imaging for recurrent prostate cancer. Radiol Clin North Am. 2000;38:213–229. 6. Kestin LL, Vicini FA, Ziaja EL, et al. Defining biochemical cure for prostate carcinoma patients treated with

external beam radiation therapy. Cancer. 1999;86:1557–1566. 7. Pound CR, Brawer MK, Partin AW. Evaluation and treatment of men with biochemical prostate-specific antigen

recurrence following definitive therapy for clinically localized prostate cancer. Rev Urol. 2001;3:72– 84. 8. Crook J, Robertson S, Collin G, et al. Clinical relevance of trans-rectal ultrasound, biopsy, and serum prostate-

specific antigen following external beam radiotherapy for carcinoma of the prostate. Int J Radiat Oncol Biol Phys. 1993;27:31–37.

9. Chan TW, Kressel HY. Prostate and seminal vesicles after irradiation: MR appearance. J Magn Reson Imaging. 1991;1:503–511.

10. Coakley FV, Hricak H, Wefer AE, et al. Brachytherapy for prostate cancer: endorectal MR imaging of local treatment-related changes. Radiology. 2001; 219:817– 821.

11. Rouviere O, Valette O, Grivolat S, et al. Recurrent prostate cancer after external beam radiotherapy: value of contrast-enhanced dynamic MRI in localizing intraprostatic tumour—correlation with biopsy findings. Urology. 2004;63:922–927.

12. Haider MA, Chung P, Sweet J, et al. Dynamic contrast-enhanced magnetic resonance imaging for localization of recurrent prostate cancer after external beam radiotherapy. Int J Radiat Oncol Biol Phys. 2008;70:425– 430.

13. Coakley FV, Teh HS, Qayyum A, et al. Endorectal MR imaging and MR spectroscopic imaging for locally recurrent prostate cancer after external beam radiation therapy: preliminary experience. Radiology. 2004;233:441–448.

14. Menard C, Smith IC, Somorjai RL, et al. Magnetic resonance spectroscopy of the malignant prostate gland after radiotherapy: a histopathologic study of diagnostic validity. Int J Radiat Oncol Biol Phys. 2001;50:317–323.

15. Kim CK, Park BK, Lee HM. Prediction of locally recurrent prostate cancer after radiation therapy: incremental value of 3T diffusion-weighted MRI. J Magn Reson Imaging. 2009;29:391–397.

16. Futterer JJ, Heijmink SW, Scheenen TW, et al. Prostate cancer localization with dynamic contrast-enhanced MR imaging and proton MR spectroscopic imaging. Radiology. 2006;241:449–458.

17. Huisman HJ, Engelbrecht MR, Barentsz JO. Accurate estimation of pharmacokinetic contrast-enhanced dynamic MRI parameters of the prostate. J Magn Reson Imaging. 2001;13:607– 614.

18. Tofts PS, Brix G, Buckley DL, et al. Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J Magn Reson Imaging. 1999;10:223

19. Kovar DA, Lewis M, Karczmar GS. A new method for imaging perfusion and contrast extraction fraction: input functions derived from reference tissues. J Magn Reson Imaging. 1998;8:1126 –1134.

20. Vos PC, Hambrock T, Barentsz JO, et al. Automated calibration for computerized analysis of prostate lesions using pharmacokinetic magnetic resonance images. In: Yang GZ, Hawkes DJ, Rueckert D, et al, eds. Medical Image Computing and Computer-Assisted Intervention MICCAI 2009. Part II, Volume 5761. Berlin, Germany: SpringerLink; 2009:836–843.

21. Huisman HJ, Karssemeijer N. Processing and displaying dynamic contrastenhanced MRI 2008 22. Hambrock T, Futterer JJ, Huisman HJ, et al. Thirty-two-channel coil 3T magnetic resonance-guided biopsies of

prostate tumour suspicious regions identified on multimodality 3T magnetic resonance imaging: technique and feasibility. Invest Radiol. 2008;43:686–694.

23. Kaplan I, Oldenburg NE, Meskell P, et al. Real time MRI-ultrasound image guided stereotactic prostate biopsy. Magn Reson Imaging. 2002;20:295–299.

24. Singh AK, Kruecker J, Xu S, et al. Initial clinical experience with real-time transrectal ultrasonography-magnetic resonance imaging fusion-guided prostate biopsy. BJU Int. 2008;101:841– 845.

25. Lowry M, Zelhof B, Liney GP, et al. Analysis of prostate DCE-MRI: comparison of fast exchange limit and fast exchange regimen pharmacokinetic models in the discrimination of malignant from normal tissue. Invest Radiol. 2009;44:577–584.

26. Kim CK, Park BK, Park W, et al. Prostate MR imaging at 3T using a phased-arrayed coil in predicting locally recurrent prostate cancer after radiation therapy: preliminary experience. Abdom Imaging. In press.

27. Stephenson AJ, Scardino PT, Bianco FJ, et al. Salvage therapy for locally recurrent prostate cancer after external beam radiotherapy. Curr Treat Options Oncol. 2004;5:357–365.

28. Gibbs P, Liney GP, Pickles MD, et al. Correlation of ADC and T2 measurements with cell density in prostate cancer at 3.0 Tesla. Invest Radiol. 2009;44:572–576.

114


5

CHAPTER 6

Multiparametric MR Imaging for Detection and Localization of

Transition Zone prostate cancer

C. Hoeks, T. Hambrock; D. Yakar et al.

CHAPTER 6CHAPTER 6

Multiparametric MR imaging for detection and localization of transition zone prostate cancer

Radiology (Accepted)

Hoeks C, Hambrock T, Yakar D, Hulsbergen-van de Kaa CA, Wijtes JA,

Fütterer JJ, Barents JO

Advances in knowledge:

3T Multiparametric MR imaging, consisting of T2-weighted imaging (T2-w) + Diffusion

Weighted imaging + Dynamic Contrast-Enhanced MR imaging is of added value for the

detection of Gleason Grade (GG) 2-3 transition zone (TZ) cancers.

Detection rates of Gleason Grade 4-5 TZ cancers are significantly higher than detection

rates of Gleason Grade 2-3 TZ cancers on multiparametric MR imaging.

The overall localization accuracy for both GG 2-3 and GG 4-5 TZ cancers with

multiparametric MRI is superior compared to T2-w imaging alone .

Implications for patient care:

In patients with an increased PSA and one or more negative TRUS biopsy sessions, MR

imaging can detect the vast majority of aggressive tumours and almost the half of all low-

grade TZ tumours.

MR imaging can have a valuable role in identifying and selecting patients undergoing

active surveillance or targeting focal therapy.

Summary Statement:

Multiparametric 3T MR imaging is an accurate technique to detect and localize clinically significant, aggressive transition zone cancer.

116

Multi-parametric MR Imaging for Detection and Localization of Transition Zone Prostate Cancer

6

ABSTRACT

Purpose: To retrospectively determine GG 2-3 and GG 4-5 transition zone (TZ) cancer detection

and localization accuracy of multiparametric MR imaging (MP-MRI) using radical prostatectomy

(RP) specimens as a gold standard.

Materials and Methods: The need for informed consent was waived by the IRB. Inclusion

criteria were TZ cancer >0.5 cm 3 upon RP and prior performed 3T endorectal multiparametric

MR imaging (T2-weighted (T2-w), diffusion weighted and dynamic contrast enhanced MRI

(DCE-MRI)). From 96 RP, 20 patients with TZ cancers were included and twenty-two patients

without TZ but with PZ cancers were randomly selected as control-group. Four radiologists

randomly scored patients for T2-w, and MP-MRI protocols: T2-w+ADC, T2-w+DCE-MRI and T2-

w+ADC+DCE-MRI consecutively with an interval of TZ cancer suspicion was rated (5-

point scale) in 6 regions of interest (ROI). A score of 4-5 was defined as a positive detection

result on a patient level. Localization was analyzed using ROI-ROC with generalized estimation

equations.

Results: MP-MRI (T2-w+ADC+/DCE-MRI, accuracy 60-61 %) improved detection of GG 2-3 TZ

cancers compared to T2-w alone (44%, p=0.01-0.02). Significantly more GG 4-5 (91%) versus

GG 2-3 (47%) TZ cancers were detected (p<0.05). Localization on ROI level was slightly

improved by MP-MRI for GG 4-5 TZ cancers only (AUC: 0.94 versus T2-w 0.91, p=0.02).

Conclusion: Only for GG 2-3 TZ cancers MP-MRI improves detection accuracy in comparison to

T2-w. Detection rates of GG 4-5 cancers are significantly higher than those of GG 2-3 cancers.

MP-MRI is only of slightly added value to T2-w for TZ cancer localization.

117


6

INTRODUCTION

Prostate cancer accounts for more than a quarter of male cancer incidence and had mortality

rates over 15% in 2008 (1). Twenty-five to 30% of these cancers are transition zone (TZ)

cancers (2;3). No uniform histopathologic definition of TZ cancers exists, although a cancer

volume of 50-70% or higher within the TZ, is commonly used as a histopathologic definition for

probable TZ origin (4). TZ prostate cancers have shown to have a relatively lower Gleason score

(5), local stage (6) and biochemical recurrence rate (7;8) in comparison to peripheral zone (PZ)

cancers. However, zonal location of a high Gleason Grade (GG) prostate cancer has shown not to

influence biochemical relapse-free survival (9). Furthermore, in prostatectomy series, a GG 4 or

5 with extra capsular extension and positive resection margins was present in 9% of all TZ

cancers (10). Therefore, improvement of TZ cancer detection remains an important goal in

prostate cancer diagnostics. Especially detection of TZ cancers with a GG 4-5 is important, as a

higher GG is an independent factor for poorer prognosis (11).

In current prostate cancer diagnostics, especially very anteriorly situated TZ cancers are often

subject to biopsy error due to dorsal to medial sample reach in random transrectal ultrasound

guided biopsies (TRUSGB) (12). Moreover, in patients with an elevated prostate specific antigen

(PSA) and 2 or more previous negative TRUSGB sessions, 57% of MR guided biopsy detected

cancers are situated within the ventral TZ (13).

On T2-weighted MR imaging (T2-w), TZ prostate cancers are difficult to differentiate from

benign prostatic hyperplasia, as the latter has heterogeneous signal intensity (SI)

similar to those of prostate cancer. In spite of this, T2-w has been advocated as an accurate

technique for TZ cancer detection (14;15). Several features on T2-w, like a homogeneously low

SI (sensitivity 76-78%, specificity 78-87%), ill-defined margins (sensitivity 76-78%, specificity

78-89%) and lenticular shape (sensitivity 48-56%, specificity 85-98%) may accurately predict

TZ cancer (16-18).

Functional MR imaging on 1.5 T has not shown to be of added value to T2-w for TZ cancer

detection (19). Diffusion weighted MR imaging (DWI) has shown to increase TZ cancer detection

accuracy when high b-values (>1000 s/mm2) are used (20-23). In dynamic contrast enhanced

MR imaging (DCE-MRI) quantitative parameters like Ktrans are able to differentiate TZ cancer

from glandular BPH, however not from stromal BPH (24). To our knowledge, 3T endorectal

118


6

multiparametric MR imaging including DWI and DCE-MRI has not earlier been evaluated for TZ

cancer detection and localization accuracy. Application of a higher field strength may have the

advantage of a higher SNR in the ventral prostate.

Therefore, the purpose of this study is to retrospectively determine GG 2-3 and GG 4-5 transition

zone (TZ) cancer detection and localization accuracy of multiparametric MR imaging (MP-MRI)

using radical prostatectomy (RP) specimens as a gold standard.


Patients

The need for informed consent was waived by the Institutional Review Board. Inclusion criteria

were patients who had TZ cancer with a cancer volume > 0.5 cm3 upon prostatectomy and a pre-

prostatectomy endorectal 3T multiparametric MR imaging examination including T2-w, DWI

and DCE-MRI. Patients who had prior radiotherapy or transurethral resection of the prostate

were excluded.

Twenty-three TZ cancer patients were retrospectively selected from 98 consecutive

prostatectomies performed within our referral center between January 2007 and May 2009.

Subsequently, 25 patients with PZ cancer without co-existent TZ cancer were randomly selected

to serve as a control group, of which 22 patients were finally included. Three PZ cancer patients

were excluded due to earlier transurethral resection of the prostate (n=2) or due to a post-

radiotherapy status (n=1). A flow diagram for patient selection is shown in Figure 1.

MR imaging acquisition protocol

MR images were obtained on a 3T MR system (Trio Tim, Siemens, Erlangen, Germany) with the

use of a pelvic phased-array coil and an endorectal coil (Medrad, Pittsburgh, USA) filled with 40

ml of perfluorocarbon (Fomblin, Solvay-Solexis, Milan, Italy). Multiparametric MR imaging

parameters are presented in Table 1.

119


6

Axial T2-w images of the prostate and seminal vesicles were followed by axial DWI with the

same slice positions using diffusion modules and fat suppression pulses. Proton diffusion was

measured in three directions and apparent diffusion coefficient (ADC) maps were calculated

automatically. Prior to the DCE series an axial 3D proton density weighted gradient echo was

obtained. This was followed by an axial 3D T1 weighted spoiled gradient-echo sequences,

acquired during intravenous administration of 0.1 mmol of gadopentetate dimeglumine

(DOTAREM, Guerbet, Paris, France) per kilogram of bodyweight at a rate of 2.5 ml/s followed by

a 20 ml saline flush. Data post-processing of DCE-MR imaging was performed on in-house

software. Quantitative pharmacokinetic analysis was based on the Toft's model (25), using an

automatic per-patient calibration for estimation of the arterial input function (26).

MR image interpretation

Four radiologists (two with 2 years, one with 4 years and one with 9 years of experience with

prostate MR imaging) independently scored all cases in random order on in-house developed

software. Radiologists were blinded for the zonal location of the prostate cancer. T2-w, T2-

w+ADC, T2-w+DCE-MRI and T2-w+ADC+DCE-MRI were prospectively scored in 4 separate

consecutive sessions with an interval of at least 2 weeks. T1-weighted MR images were available

for evaluation of post-biopsy hematoma to avoid false-positive results. The TZ was divided into 6

regions of interest (ROI). In the coronal plane the TZ was divided in 3 parts: the level of the

verumontanum and the level inferior and superior to this plane. The sagittal plane through the

verumontanum was used to divide the TZ into a right and a left half.

A five-point scale rating was used for every ROI and for every reading session: 1) definitely no,

2) probably no, 3) possible, 4) probable and 5) definite presence of TZ cancer. For T2-w, existing

TZ cancer multiparametric MR imaging features were used to evaluate the presence of TZ

cancer. These included: homogeneously low SI, irregular boundaries around a low SI, lenticular

shape, interruption of the pseudocapsule and invasion of the anterior fibromuscular stroma

(AFS) or the ventrolateral TZ boundary (27-29). For DWI, a homogeneously low ADC value in

comparison to the surrounding TZ was used (30).

120


6

Figure 1. Flow diagram for patient selection. MR= magnetic resonance, TZ= transition

zone, PZ= peripheral zone, DWI= diffusion weighted MR imaging, DCE-MRI = dynamic

contrast enhanced MR imaging, TURP= transurethral resection of the prostate and

GT= ground truth or reference standard, 3T= 3 tesla.

Radical Prostatectomies

2007-2009 n= 98

Endorectal 3T MR imaging

including T2w, DWI and DCE

n=59

Exclusion: No combined DWI

and/or DCE n=28

No endorectal coil n= 11

TZ cancers n= 20

Random selection control group:

PZ cancers without co-existent

n= 25

TZ cancers positive GT

n= 20 Control group: PZ cancers without co-existent TZ

n= 22

Exclusion control group: n=2 TURP n= 1 post-radiotherapy

121


6

ADC maps were shown using window levels on which the highest ADC values within the bladder

ranged from 1800-2843 x 10-6 mm2/s. The darkest ADC values within the endorectal coil were 0

mm2/s and ADC values of the darkest regions within the TZ ranged from 251-1200 x 10-6 mm2/s.

For DCE-MRI (31), a pharmacokinetic signal intensity time-curve, in which washout occurred

immediately after early enhancement, was considered suspicious for TZ cancer. Both the lack of

washout and the presence of a plateau phase after early enhancement were considered less

suspicious for TZ cancer presence. The following features were applied for TZ cancer suspicion:

homogeneous, asymmetric and AFS enhancement (32-34), washout and the lack of a type 1 and

2 curve on Ktrans parametric maps (35).

Histopathologic analysis

Prostatectomy specimens served as a reference standard for multiparametric MR imaging

results and were processed into 4 mm step section slices (36). An experienced urogenital

pathologist, who was blinded for multiparametric MR imaging results, determined location,

stage and GG components for every individual cancer (according to 2005 ISUP criteria (37)).

Cancer volume was calculated while assuming elliptical tumour shape (38). Presence of a cancer

(39).

Correlation of MR data to the gold standard

Two radiologists (one with 2 years and one with 4 years of experience in prostate

multiparametric MR imaging) correlated MR data and prostatectomy in consensus using

landmarks like the verumontanum and calcifications. Correlation of prostate multiparametric

MR imaging to histopathology is known to be difficult (40).

Statistical Analyses

For all analyses significance levels of 2 sided p-value < 0.05 were used. An independent t-test

was used to test for differences in characteristics between patient groups. A Z-score was used to

test for differences in proportions.

TZ cancers consisting of a GG 4 and/or 5, were considered high GG TZ cancers, while TZ cancers

consisting of a GG 2 and/or 3 only were considered low GG TZ cancers.

122


6

Localization of TZ cancer was defined as finding a cancer in a certain ROI area of the prostate.

Localization accuracy for all TZ cancers, for GG 2-3- and for GG 4-5 TZ cancers was analyzed by

comparing areas under the receiver operating characteristic (ROC) curve (AUC) of different MR

imaging protocols using a generalized estimation equation (GEE) to account for correlation of

the same patient. The GENMOD procedure (version 9, SAS Institute, Cary,

NC) was used to calculate Pearson X2 tests for differences in ROI-ROC values between

multiparametric MR imaging protocols.

(41).

Protocol Sequence TR (ms)

TE (ms)

Slice thickness

FOV Matrix size

Voxel size

b-values (s/mm2)

Temp. resol.(s)

T2-w

TSE axial

coronal

sagittal

4280

3599

4290

99

98

98

3

3

3

179 179

192 192

192 192

448 448

384 384

384 384

0.4 0.4

0.5 0.5

0.5 0.5

DWI SSEPI 2600 90 3 204 204 136 136 1.5 1.5 0/50/500/800

DCE-MRI GE 800 1.51 3 192 192 128 128 1.5 1.5

DCE-MRI Spoiled GE 36 1.4 3 192 192 128 128 1.5 1.5 2.5

Table 1. MR imaging parameters. T2-w = T2-weighted MR imaging, DWI= diffusion

weighted MR imaging, DCE-MRI= dynamic contrast enhanced MR imaging, TSE= turbo

spin echo, SSEPI= single-shot echoplanar imaging, GE= gradient echo, TR= repetition time

and TE=echo time.

121


6

123

MRI All TZ cancers vs. no TZ cancers GG 4-5 TZ vs. no TZ cancers GG 2-3 TZ vs. no TZ cancers

Sens. Spec. Accur. Sens. Spec. Accur. Sens. Spec. Accur.

T2-w 60%

(48/80)

[49-70]

74%

(65/88)

[64-82]

67%

(113/168)

[60-74]

86%

(38/44)

[73-94]

74%

(65/88)

[64-82]

78%

(103/132)

[70-84]

28%

(10/36)

[14-41]

51%

(45/88)

[41-61]

44%

(54/124)

[35-52]

T2-w + ADC

68%

(54/80)

[57-77]

68%

(60/88)

[58-77]

68%

(114/168)

[60-74]

86%

(38/44)

[73-94]

68%

(60/88)

[58-77]

74%

(98/132)

[66-81]

44%

(16/36)

[30-60]

68%

(60/88)

[58-77]

61%

(76/124)

[52-69]

T2-w + DCE-MRI

63%

(50/80)

[52-72]

69%

(61/88)

[59-78]

66%

(111/168)

[59-73]

89%

(39/44)

[76-96]

72%

(63/88)

[61-80]

77%

(102/132)

[69-84]

33%

(12/36)

[20-50]

72%

(63/88)

[61-80]

60%

(75/124)

[52-69]

T2-w + ADC +

DCE-MRI

59%

(47/80)

[48-69]

65%

(57/88)

[54-74]

62%

(104/168)

[54-69]

91%

(40/44)

[78-97]

81%

(71/88)

[71-88]

84%

(111/132)

[77-89]

47%

(17/36)

[32-63]

65%

(57/88)

[54-74]

60%

(74/124)

[51-68]

Table 3. Diagnostic accuracy for detection of all TZ cancers, of GG 4-5 and of GG 2-3 TZ

cancers for all readers for the different multiparametric MR imaging protocols.

Percentages are placed as numbers and proportions are placed in between brackets. 95%

confidence intervals are placed in square brackets. T2-w = T2-weighted MR imaging,

ADC= apparent diffusion coefficient, DCE-MRI= dynamic contrast enhanced MR imaging,

TZ= transition zone, MR= magnetic resonance, GG= Gleason Grade.

RESULTS

For 19 patients with TZ cancer only one tumour was present and for 1 patient 2 TZ cancers, both

with a GG 3+2 were present in the TZ upon prostatectomy specimens. Eleven out of 20 (55%) TZ

cancers had a GG 4-5 and nine out of 20 (45%) TZ cancers had a GG 2-3. Mean cancer volume of

GG 4-5 TZ cancers (6.6 cm3) did not differ significantly from mean cancer volume of GG 2-3 TZ

cancers (6.8 cm3, p=0.96). This was also the case for PZ cancers (GG 4-5 PZ cancers mean volume

of 0.95 cm3 versus GG 2-3 PZ cancers mean volume of 0,73 cm3, p=0.51). Mean PSA levels,

histopathological stage pT2A and cancer volume differed significantly between TZ and PZ cancer

124


6

patients (respectively p=0.01, p=0.02 and p<0.001). Differences between patients with TZ

cancers and patients with PZ cancers are presented in Table 2.

For detection of all TZ cancers, GG 4-5 TZ cancers and GG 2-3 TZ cancers on a patient level,

diagnostic accuracies are presented in Table 3. The accuracy did not differ significantly between

MR imaging protocols (T2-w, T2-w+ADC, T2-w+DCE-MRI and T2-w+ADC+DCE-MRI) for

detection of both all TZ cancers and for GG 4-5 TZ cancers. However, for detection of GG 2-3 TZ

cancers accuracies of T2-w+ADC, T2-w+DCE-MRI and T2-w+ADC+DCE-MRI were significantly

higher in comparison to T2-w: T2-w+ADC versus T2-w p=0.01, T2-w+DCE-MRI versus T2-w

p=0.01 and T2-w+ADC+DCE-MRI versus T2-w,p=0.02.

For both T2-w and multiparametric MR imaging, detection rates of GG 4-5 TZ cancers were

significantly higher in comparison to detection rates of GG 2-3 TZ cancers: T2-w (28% of GG 2-3

cancers versus 86% of GG4-5 cancers, p=0.001), T2-w+ADC (44% versus 86%, p=0.001), T2-

w+DCE-MRI (33% versus 89%, p=0.001) and T2-w+ADC+DCE-MRI (47% versus 91%, p=0.001).

These results are presented in Table 4.

For localization of GG 4-5 TZ cancers on an ROI level, AUC of T2-w+ADC+DCE-MRI (0.94) was

higher in comparison to AUC of T2-w alone (0.91, p=0.02) and in comparison to AUC of T2-

w+DCE-MRI (0.91, p=0.01). For localization of all TZ cancers and GG 2-3 TZ cancers on an ROI

level, there were no significant differences between different multiparametric MR imaging

protocols. Results of these ROI-ROC analyses are presented in Table 5 and Figure 2.

Interobserver agreement for TZ cancer detection was moderate to substantial. Mean weighted

kappa values for all protocols (T2-w, T2-w+ADC, T2-w+DCE-MRI and T2-w+ADC+DCE-MRI)

ranged from 0.56-0.71.

125


6

Characteristic Patients with

TZ cancer

(n = 20)

Patients with PZ cancer only

(n = 22)

Independent t-test patients with

vs. patients without TZ

cancer

Patients with GG4-5 TZ

cancer

(n=11)

Patients with GG2-3 TZ

cancer

(n=9)

Age (yr) (range) 67 (55-73) 64.5 (53-71) 0.98 67 (55-73) 67 (55-70)

PSA (ng/ml) (range) 10.3 (1.9-44) 6.3 (3.2-14.8) 0.01* 17.6 (5.3-44.0) 8.7(1.9-11)

Tumour volume (cm3) 4.9 (0.5-22) 0.5 (0.01-2.36) <0.001* 7.4 (0.5-15.7) 3.5 (0.5-22)

MR I to Surgery interval (wk) (range)

7(1-21) 7 (1-21) 0.78 5 (1-21) 8(1-22)

Histopathologic Stage

pT2A 0 7 p=0.02* 0 1

pT2C 9 9 p=0.96 3 6

pT3A 7 6 p=0.83 5 1

pT3B 2 0 p=0.43 2 0

pT4 2 0 p=0.43 1 1

Gleason score in prostatectomy

2+2 0 1 p=0.96 0 0

2+3 3 0 p=0,20 0 3

2+4 1 0 p=0.96 1 0

3+2 3 1 p=0.54 0 3

3+3 3 9 p=0.13 0 3

3+4 2 7 p=0.18 2 0

4+2 1 0 p=0.96 1 0

4+3 5 4 p=0.87 5 0

4+5 2 0 p=0.43 2 0

Table 2. Patient and prostatectomy characteristics. TZ= transition zone, PZ= peripheral

zone, PSA= prostate specific antigen. *= under threshold for significance (2 tailed p-value

<0.05), N.A. = not available, n= number, GG= Gleason Grade. Median values are reported.

126


6

MR imaging protocol Detection rate

GG 2-3 TZ cancers Detection rate

GG 4-5 TZ cancers Z score, p value

T2-w

28% (10/36)

86% (38/44)

5.09, p<0.001*

T2-w+ADC

44% (16/36)

86% (38/44)

3.74, p<0.001*

T2-w+DCE

33% (12/36)

89% (39/44)

4.89, p<0.001*

T2-w+ADC+DCE

47% (17/36)

91% (40/44)

4.05, p<0.001*

Table 4. Comparison of detection rates of GG 2-3 versus GG 4-5 TZ cancers. Percentages

are placed as numbers and proportions are placed in between brackets. TZ= transition

zone, * = under threshold for significance (2 tailed p-value <0.05), GG= Gleason Grade,

T2-w= T2-weighted MR imaging, ADC= apparent diffusion coefficient, DCE-MRI= dynamic

contrast enhanced MR imaging, MR= magnetic resonance.

Localization all TZ cancers X2 test and AUC

Localization GG 4-5 TZ cancers versus PZ cancers

Localization GG 2-3 TZ cancers versus PZ cancers

X2(3)=6.68, p=0.08 X2(3)=8.55, p=0.04 X2(3)=4.44, p=0.22

T2-w 0.78 (0.71-0.84)

T2-w 0.91

(0.86-0.96)

T2-w 0.56

(0.43-0.68)

T2-w+ADC 0.82 (0.76-0.89)

T2-w+ADC 0.92

(0.86-0.97)

T2-w+ADC 0.66

(0.53-0.79)

T2-w+DCE-MRI 0.80 (0.73-0.86)

T2-w+DCE-MRI 0.91

(0.87-0.96)

T2-w+DCE-MRI 0.60

(0.47-0.73)

T2-w+ADC+

DCE-MRI 0.83 (0.77-0.89)

T2-w+ADC+DCE-MRI 0.94

(0.91-0.98)

T2-w+ADC+DCE-MRI

0.64

(0.53-0.75)

Significant differences Significant differences Significant differences

N.S. T2-w < T-2w+ADC+DCE

T2-w+DCE-MRI <

T2-w+ADC+DCE-MRI

p=0.02

p=0.01

N.S.

Table 5. Results of ROI-ROC analyses for localization of all TZ cancers, of GG 4-5 and of GG

2-3 TZ cancers. ROI-ROC values were compared using a generalized estimation equation

(GEE). Pearson X2 tests were calculated to test for differences in ROI-ROC values between

MR imaging protocols. 95% confidence intervals are placed between brackets. T2-w= T2-

weighted MR imaging, ADC= apparent diffusion coefficient maps, DCE-MRI= dynamic

contrast enhanced MRI, TZ= transition zone, PZ= peripheral zone, GG= Gleason Grade.

127


6

Figure 2 Receiver operating characteristic curves of the different multiparametric MR

imaging protocols for localization of TZ cancers (a), of GG 4-5 TZ cancers (b)( AUC values

of T2-w+ADC+DCE-MRI (0.94) were significantly higher in comparison to AUC of T2-w

only (0.91, p=0.02) and to AUC of T2-w+DCE-MRI (0.91, p=0.01)) and of GG 2-3 TZ cancers

(c) respectively. Diagonal black line= reference line of a 0.50 area under the curve, TZ=

transition zone, T2-w= T2-weighted MR imaging, ADC= apparent diffusion coefficient

maps, DCE-MRI= dynamic contrast enhanced MR imaging, GG= Gleason Grade. A legend on

the right side indicates which colors belong to which MR imaging protocol, AUC= area

under the receiver operating characteristic curve, *= under threshold for significance (2

tailed p-value <0.05).

128


6

DISCUSSION

To our knowledge, this is the first study, which compared more than one endorectal MR imaging

technique at 3T for TZ prostate cancer detection and localization. Our results indicate that

multiparametric MR imaging (T2-w+ADC and/or DCE-MRI) significantly improves detection

accuracy in comparison to T2-w for GG 2-3 TZ cancers, as opposed to GG 4-5 TZ cancers. For

detection of GG 4-5 TZ cancers, T2-w may be sufficient as multiparametric MR imaging did not

significantly improve detection accuracy in comparison to T2-w. All analyses within our study

reflect significantly increased detection rates of GG 4-5 TZ cancers versus GG 2-3 TZ cancers for

both T2-w only and for multiparametric MR imaging. Cancer volume may have influenced these

measurements to some extent. Whereas mean cancer volumes did not differ significantly

between GG 2-3 and GG 4-5 TZ cancers, the latter ones were, nonetheless, often larger in volume.

A few studies on 1.5T have evaluated detection and localization of TZ cancer with

multiparametric MR imaging. These studies, however, did not always use a histopathological

definition for TZ cancer, used different b-values and applied either pelvic phased-array or

endorectal coils (42-45). For TZ cancer detection, our accuracy for T2-w+ADC of 62-68% for all

TZ cancers is lower compared to other studies (46-49). Possible reasons for better DWI results

of Yoshizako et al. (accuracies of 62-81% respectively) are inclusions of larger lesions (10-28

mm versus our median of 3.56 cc), and the use of higher b- 2), which have

improved TZ cancer detection accuracy for T2-w+DWI(50;51). Our results are in agreement

with studies, which used b-values of <1000 mm/s2 (52;53), and did not show added value of

DWI in general TZ cancer detection. Furthermore the lack of use of a standard ADC window-

level (54) and quantitative threshold values for ADC reader evaluation (55) may further explain

our lower detection accuracy for T2-w+ADC in comparison to results of Haider et al. (accuracy of

81%).

Considering T2-w+DCE-MRI, our results are comparable to those of Yoshizako et al. at 1.5T

(sensitivity 69%, specificity 68% and accuracy 69%)(56). However, our results differ from those

of Delongchamps et al.(57), this study had a sensitivity of 47% and a specificity of 77% for TZ

cancer detection. Their lower sensitivity in comparison to ours (63%) may be explained by their

population existing predominantly out of GG 2-3 TZ cancers and by using a lower field strength.

Performed studies of the TZ often used semi-quantitative analysis instead of our quantitative

pharmacokinetic analysis, which makes comparison difficult (58;59). Our detection results for

multiparametric MR imaging are in contrast with those of Delongchamps et al., who found no

added value of multiparametric MR imaging techniques compared to T2-w for TZ cancer

129


6

detection in a sub-group of predominantly low GG TZ cancers (60). The lower field strength and

the lack of a histopathologic definition for TZ cancers may explain these conflicting results for

detection of low GG TZ cancers. Within the same study, also no influence of Gleason score on

multiparametric MR imaging TZ cancer detection accuracy was found. This may be due to the

low number of GG 4-5 TZ cancers in their population.

For TZ cancer localization on ROI level, multiparametric MR imaging adds little value to T2-w.

Only for GG 4-5 TZ cancers, multiparametric MR imaging (T2-w+ADC+DCE-MRI AUC 0.94)

slightly improves TZ cancer localization accuracy in comparison to T2-w (AUC 0.91, p=0.02) and

in comparison to T2-w+DCE-MRI (AUC 0.91, p=0.01). Our TZ cancer localization accuracy for T2-

w+ADC is lower (AUC 0.82) in comparison to Delongchamps et al. (AUC 0.88), while our T2-

w+DCE-MRI (AUC 0.80) results are in agreement with their T2-w+DCE-MRI results (AUC

0.70)(61). The mentioned lack of use of a standard ADC window-level (62) and quantitative

thresholds may explain our lower T2-w+ADC performance. Our T2-w+ADC+DCE-MRI

localization results (AUC 0.83) are somewhat higher in comparison to the same study (AUC

0.75). As mentioned, this difference may be explained by our use of a higher field strength.

This study has limitations. As mentioned, we did not use high b-values, standard ADC settings

and quantitative ADC reading in DWI. However, no quantitative ADC thresholds have been

defined for ADC maps created out of DWI with b values< 1000 s/mm2. Secondly, in order to

acquire the most optimal gold standard, we only selected prostatectomy patients. This may have

introduced selection bias in our results. Our population is rather small and consists for 54% out

of TZ cancers. This is not an optimal representation of the general population where prevalence

of TZ cancers is 25%(63). Thirdly, significant differences between TZ cancer and PZ cancer

(healthy TZ) patient groups were present for PSA level, cancer volume and clinical stage. This

could not be adjusted for with stratification as patient numbers in subpopulations were too low

for valid analyses. A relatively higher PSA level, cancer volume and clinical stage in TZ cancer

patients versus PZ cancer patients, may have biased our results. Fourthly, we only included TZ

cancers with a volume >0.5 cm3. The median volume of undetected TZ cancers in a 1.5T study

with a higher in plane resolution than our study (0.31 0.31 mm), was 0.44 cm3 (range 0.26-

0.58)(64). By including TZ cancers with a volume> 0.5 cm3, all TZ cancers should be detectable

on 3T MR imaging.

130


6

Figure 3. Sixty-five year old male patient, with a PSA of 44 ng/mL, 2 previous TRUS guided

prostate biopsy sessions and a clinical stage T2 and a Gleason score of 4+5=9 prostate

cancer found in a third session in all five cores on the right side, who underwent

endorectal MR imaging for local prostate cancer localization and staging. The cancer

suspicious region is visible on T2WI as well as on different multi-parametric MR imaging

protocols. (a) Axial endorectal T2WI turbo spin echo at mid-prostate level shows a low

zone (white demarcation). This finding is a feature of transition zone cancer on T2WI. (b)

Axial ADC map at the same level as image a. A low ADC value (mean ADC 758 x 10-6

mm2/s) is present in the ventral transition zone (white demarcation). A low ADC value

within the transition zone may be suspect for transition zone cancer. (c) Axial T2-

weighted turbo spin echo at the same level as (a), with a superimposed Ktrans parametric

map. Homogeneous enhancement of the transition zone is present. (d) Axial whole mount

section histopathology at the corresponding level (a-c) demonstrates a pT4N1R1 Gleason

score 4+5=9 prostate TZ cancer (black demarcation).

131


6

Figure 4. A 59-year old man with a PSA of 10.3 ng/ml, a clinical stage T2 and a Gleason

score 3+3 prostate cancer in 2 out of 5 cores on the left side and in 1 out of 5 cores on the

right side upon random TRUS guided biopsy. Endorectal MR imaging was performed upon

a clinical localization and staging indication. The cancer suspicious region is more clearly

defined on multiparametric MR imaging techniques in comparison to T2-w.

a) Axial T2-weighted images at the apex-mid level. A homogeneous low signal intensity

te arrows).

b) Axial ADC map at the same level as a. In the right transition zone an irregular

shaped low ADC value (median 422x10-6 mm2/s), suspect for TZ cancer is present

(white and black demarcation). In the left TZ a low ADC value is also present (white

line demarcation).

132


6

c) Axial T2-w image at the same level as (a) with a superimposed wash-out parametric

map The right TZ shows an asymmetric increased contrast washout, which is

suspect for transition zone cancer. A red-colored ring shape pattern can be

appreciated. Some washout is also present in the left transition zone.

d) The relative Gadolinium-concentration-

lesion.

e) Axial whole mount section histopathology at the corresponding level (a-c) shows a

multifocal prostate cancer, with a dominant lesion in the right ventral transition

zone Gleason score 3+4=7 (black demarcation). The lesion in the left transition

zone (L) had a Gleason score of 3+2=5.

133


6

Figure 5. A 67-year old male patient with a PSA of 18 ng/ml, a clinical stage T3 Gleason score of

7 prostate cancer underwent endorectal MR imaging for pre-surgical staging.

a) Axial T2-weighted image at mid-prostate level. Next to a well defined inhomogeneous

nodular pattern of the dorsal transition zone, a low homogeneous signal intensity with

poorly defined edges is seen in the right ventral transition zone. A lenticular shape and

interruption of the pseudo-capsule are present (white demarcation). These findings are

suspect for transition zone cancer.

b) Axial ADC map at the same level as (a). A low ADC value (median 511x10-6 mm2/s) is

present in the right ventral transition zone (white demarcation).

134


6

c) Axial endorectal T2-weighted turbo spin echo at the same level as (a) with a

superimposed Ktrans parametric map. A suspect sign for transition zone cancer is the

increased enhancement of the anterior fibromuscular stroma (white demarcation).

d) The relative Gadolinium-contrast-versus-

fast rise, fast time-to-peak and wash-out, which is suspect for prostate cancer.

e) Axial whole mount section histopathology at the corresponding level (a-c) shows a pT3b

Gleason score 4+5=9 transition zone cancer in the right transition zone with extension to

the left (black demarcation).

135


6

Future studies, in which (high b-value) endorectal multiparametric MR imaging at 3T is

evaluated for detection of clearly defined TZ cancers in large populations, are necessary. Our

results have the following clinical implications; (a) in patients with increased PSA levels and one

or more negative TRUS biopsy sessions, addition of either DWI ADC maps or DCE-MRI to T2-w

is required to establish a diagnosis in case of low Gleason Grade (2-3) TZ cancer presence.

Timely TZ cancer diagnosis prevents unnecessary PSA measurements and TRUSGB sessions and

may also prevent progression of lower GG cancers and positive anterior resection margins. And

(b) in patients with proven TZ cancer who are considered for focal therapy, application of 3T T2-

w may be sufficient for TZ cancer localization, as 3T endorectal multiparametric MR imaging

(T2-w+ADC+DCE-MRI) only slightly improves T2-w localization results.

CONCLUSION

In conclusion, multiparametric MR imaging (T2-w+ADC and/or DCE-MRI), improves detection

accuracy in comparison to T2-w alone for GG 2-3 TZ cancers only, as opposed to GG 4-5 cancers.

GG 4-5 TZ cancers have significantly higher detection rates in comparison to their GG 2-3

counterparts, both on T2-w and on multiparametric MR imaging. For TZ cancer localization at

3T, multiparametric MR imaging is of little added value to T2-w alone.

136


6

REFERENCES

1. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Murray T, Thun MJ. Cancer statistics, 2008. CA Cancer J.Clin. 2008 Mar;58(2):71-96.

2. Augustin H, Erbersdobler A, Graefen M, Fernandez S, Palisaar J, Huland H, Hammerer P. Biochemical recurrence following radical prostatectomy: a comparison between prostate cancers located in different anatomical zones. Prostate 2003 Apr 1;55(1):48-54.

3. McNeal JE, Redwine EA, Freiha FS, Stamey TA. Zonal distribution of prostatic adenocarcinoma. Correlation with histologic pattern and direction of spread. Am.J.Surg.Pathol. 1988 Dec;12(12):897-906.


5. Augustin H, Erbersdobler A, Hammerer PG, Graefen M, Huland H. Prostate cancers in the transition zone: Part 2; clinical aspects. BJU Int 2004 Dec;94(9):1226-9.

6. Greene DR, Wheeler TM, Egawa S, Weaver RP, Scardino PT. Relationship Between Clinical Stage and Histological Zone of Origin in Early Prostate-Cancer - Morphometric Analysis. British Journal of Urology 1991;68(5):499-509.

7. Erbersdobler A, Huhle S, Palisaar J, Graefen M, Hammerer P, Noldus J, Huland H. Pathological and clinical characteristics of large prostate cancers predominantly located in the transition zone. Prostate Cancer Prostatic.Dis. 2002;5(4):279-84.

8. King CR, Ferrari M, Brooks JD. Prognostic significance of prostate cancer originating from the transition zone. Urol Oncol. 2009 Nov;27(6):592-7.

9. King CR, Ferrari M, Brooks JD. Prognostic significance of prostate cancer originating from the transition zone. Urol Oncol. 2009 Nov;27(6):592-7.

10. Shannon BA, McNeal JE, Cohen RJ. Transition zone carcinoma of the prostate gland: a common indolent tumour type that occasionally manifests aggressive behaviour. Pathology 2003 Dec;35(6):467-71.

11. Gleason DF, Mellinger GT. Prediction of prognosis for prostatic adenocarcinoma by combined histological grading and clinical staging. J.Urol 1974 Jan;111(1):58-64.

12. Pelzer AE, Bektic J, Berger AP, Halpern EJ, Koppelstatter F, Klauser A, Rehder P, Horninger W, Bartsch G, Frauscher F. Are transition zone biopsies still necessary to improve prostate cancer detection? Results from the tyrol screening project. Eur Urol 2005 Dec;48(6):916-21.

13. Hambrock T, Somford DM, Hoeks C, Bouwense SA, Huisman H, Yakar D, van O, I, Witjes JA, Futterer JJ, Barentsz JO. Magnetic resonance imaging guided prostate biopsy in men with repeat negative biopsies and increased prostate specific antigen. J.Urol 2010 Feb;183(2):520-7.


15. Li H, Sugimura K, Kaji Y, Kitamura Y, Fujii M, Hara I, Tachibana M. Conventional MRI capabilities in the diagnosis of prostate cancer in the transition zone. AJR Am.J.Roentgenol. 2006 Mar;186(3):729-42.

16. Hricak H, Choyke PL, Eberhardt SC, Leibel SA, Scardino PT. Imaging prostate cancer: a multidisciplinary perspective. Radiology 2007 Apr;243(1):28-53.



19. Delongchamps NB, Rouanne M, Flam T, Beuvon F, Liberatore M, Zerbib M, Cornud F. Multiparametric magnetic resonance imaging for the detection and localization of prostate cancer: combination of T2-weighted, dynamic contrast-enhanced and diffusion-weighted imaging. BJU Int 2010 Nov 2.



22. Yoshizako T, Wada A, Hayashi T, Uchida K, Sumura M, Uchida N, Kitagaki H, Igawa M. Usefulness of diffusion-weighted imaging and dynamic contrast-enhanced magnetic resonance imaging in the diagnosis of prostate transition-zone cancer. Acta Radiol. 2008 Dec;49(10):1207-13.

137


6

23. Katahira K, Takahara T, Kwee TC, Oda S, Suzuki Y, Morishita S, Kitani K, Hamada Y, Kitaoka M, Yamashita Y.

Ultra-high-b-value diffusion-weighted MR imaging for the detection of prostate cancer: evaluation in 201 cases with histopathological correlation. Eur Radiol. 2011 Jan;21(1):188-96.

24. Oto A, Kayhan A, Jiang Y, Tretiakova M, Yang C, Antic T, Dahi F, Shalhav AL, Karczmar G, Stadler WM. Prostate Cancer: Differentiation of Central Gland Cancer from Benign Prostatic Hyperplasia by Using Diffusion-weighted and Dynamic Contrast-enhanced MR Imaging. Radiology 2010 Sep 15.

25. Tofts PS, Brix G, Buckley DL, Evelhoch JL, Henderson E, Knopp MV, Larsson HB, Lee TY, Mayr NA, Parker GJ, et al. Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J.Magn Reson.Imaging 1999 Sep;10(3):223-32.

26. Vos PC, Hambrock T, Barenstz JO, Huisman HJ. Automated calibration for computerized analysis of prostate lesions using pharmacokinetic magnetic resonance images. Med Image Comput.Comput.Assist.Interv. 2009;12(Pt 2):836-43.

27. Hricak H, Choyke PL, Eberhardt SC, Leibel SA, Scardino PT. Imaging prostate cancer: a multidisciplinary perspective. Radiology 2007 Apr;243(1):28-53.



30. Kim CK, Park BK, Han JJ, Kang TW, Lee HM. Diffusion-weighted imaging of the prostate at 3 T for differentiation of malignant and benign tissue in transition and peripheral zones: preliminary results. J.Comput.Assist.Tomogr. 2007 May;31(3):449-54.

31. Ren J, Huan Y, Wang H, Chang YJ, Zhao HT, Ge YL, Liu Y, Yang Y. Dynamic contrast-enhanced MRI of benign prostatic hyperplasia and prostatic carcinoma: correlation with angiogenesis. Clin.Radiol. 2008 Feb;63(2):153-9.


33. Lemaitre L, Puech P, Poncelet E, Bouye S, Leroy X, Biserte J, Villers A. Dynamic contrast-enhanced MRI of anterior prostate cancer: morphometric assessment and correlation with radical prostatectomy findings. Eur Radiol. 2008 Aug 29.

34. Engelbrecht MR, Huisman HJ, Laheij RJ, Jager GJ, van Leenders GJ, Hulsbergen-Van De Kaa CA, de la Rosette JJ, Blickman JG, Barentsz JO. Discrimination of prostate cancer from normal peripheral zone and central gland tissue by using dynamic contrast-enhanced MR imaging. Radiology 2003 Oct;229(1):248-54.

35. Ren J, Huan Y, Wang H, Chang YJ, Zhao HT, Ge YL, Liu Y, Yang Y. Dynamic contrast-enhanced MRI of benign prostatic hyperplasia and prostatic carcinoma: correlation with angiogenesis. Clin.Radiol. 2008 Feb;63(2):153-9.

36. van Niekerk CG, van der Laak JA, Borger ME, Huisman HJ, Witjes JA, Barentsz JO, Hulsbergen-Van De Kaa CA. Computerized whole slide quantification shows increased microvascular density in pT2 prostate cancer as compared to normal prostate tissue. Prostate 2009 Jan 1;69(1):62-9.

37. Epstein JI, Allsbrook WC, Jr., Amin MB, Egevad LL. The 2005 International Society of Urological Pathology (ISUP) Consensus Conference on Gleason Grading of Prostatic Carcinoma. Am.J.Surg.Pathol. 2005 Sep;29(9):1228-42.

38. van Niekerk CG, van der Laak JA, Borger ME, Huisman HJ, Witjes JA, Barentsz JO, Hulsbergen-Van De Kaa CA. Computerized whole slide quantification shows increased microvascular density in pT2 prostate cancer as compared to normal prostate tissue. Prostate 2009 Jan 1;69(1):62-9.

39. Bouye S, Potiron E, Puech P, Leroy X, Lemaitre L, Villers A. Transition zone and anterior stromal prostate cancers: zone of origin and intraprostatic patterns of spread at histopathology. Prostate 2009 Jan 1;69(1):105-13.


41. Kundel HL, Polansky M. Measurement of observer agreement. Radiology 2003 Aug;228(2):303-8.



138


6

44. Delongchamps NB, Rouanne M, Flam T, Beuvon F, Liberatore M, Zerbib M, Cornud F. Multiparametric magnetic

resonance imaging for the detection and localization of prostate cancer: combination of T2-weighted, dynamic contrast-enhanced and diffusion-weighted imaging. BJU Int 2010 Nov 2.






50. Katahira K, Takahara T, Kwee TC, Oda S, Suzuki Y, Morishita S, Kitani K, Hamada Y, Kitaoka M, Yamashita Y. Ultra-high-b-value diffusion-weighted MR imaging for the detection of prostate cancer: evaluation in 201 cases with histopathological correlation. Eur Radiol. 2011 Jan;21(1):188-96.













139


6

63. Reissigl A, Pointner J, Strasser H, Ennemoser O, Klocker H, Bartsch G. Frequency and clinical significance of

transition zone cancer in prostate cancer screening. Prostate 1997 Feb 1;30(2):130-5.

64. Villers A, Puech P, Mouton D, Leroy X, Ballereau C, Lemaitre L. Dynamic contrast enhanced, pelvic phased array magnetic resonance imaging of localized prostate cancer for predicting tumour volume: correlation with radical prostatectomy findings. J.Urol 2006 Dec;176(6 Pt 1):2432-7.

140


6

PART THREE

ASSESSMENT OF PROSTATE CANCER

AGGRESSIVENESS

CHAPTER 7

Relation of Apparent Diffusion Coefficient Values at 3T MRI with Prostate Cancer Gleason Grade in

the Peripheral Zone

T. Hambrock; D. Somford; H. Huisman et al.

CHAPTER 7CHAPTER 7

Relation of ADC Values at 3T MRI with Prostate Cancer Gleason Grade in the Peripheral Zone

7

Relation of Apparent Diffusion Coefficient Values at 3 Tesla Magnetic Resonance Imaging with Prostate Cancer Gleason Grade in the

Peripheral Zone

Hambrock T, Somford D, Huisman H, van Oort I, Witjes J, Hulsbergen-van de Kaa C, Scheenen T, Barentsz JO

First Prize Award International Cancer Imaging Society, Bath, UK, Oct 2008


Apparent diffusion coefficient (ADC) values of prostate cancer in the peripheral zone

inversely relate to prostate cancer Gleason grades, with low-, intermediate- and high-

grade tumours showing significant differences in ADC values (p<0.001)

Using the median ADC values of the most aggressive tumour regions a high

discriminatory accuracy is achieved for discerning low-grade from combined

intermediate- and high-grade cancers (AUC=0.90).


Non-invasive prediction of Gleason grades may improve patient management by more

accurate risk-stratification, follow-up in patients undergoing active surveillance or

targeting biopsies towards the most aggressive components.

Summary Statement

ADC values determined from DW-MR imaging at 3T represents a useful biomarker for prostate cancer aggressiveness in the peripheral

zone.

142


7


7

ABSTRACT

Purpose: To retrospectively determine the relation of apparent diffusion coefficients (ADC)

from 3 Tesla Diffusion weighted MRI with prostate cancer Gleason grades in the peripheral zone.

Materials and methods: IRB approval was waived. 51 Patients underwent MR imaging prior to

prostatectomy including DWI-MRI using b-values 0, 50, 500 and 800s/mm2. In prostatectomy

specimens, separate slice-by-slice determinations of Gleason grades groups (GGG) based on

primary, secondary and tertiary Gleason grades was done. Additionally, qualitative grade (QG)

groups (low-; intermediate- or high-grade) of tumours were made. ADC maps were aligned to

step-sections and regions-of-interest annotated for each tumour slice. Median ADC (mADC) of

tumours was related to QG groups with a linear mixed model regression analysis. The accuracy

of mADC of the most aggressive tumour component, to differentiate low- from combined

intermediate- and high-grade tumours was summarized using the area (AUC) under the receiver

operating characteristics curve (ROC).

Results: In 51 prostatectomy specimens, 62 different tumours and 251 step-section tumour

lesions were identified. Tumour mADC values showed a negative association with GGG and were

significantly different between the three QG groups (p<0.001). Overall, with an increase of one

QG group, the mADC decreases with 0.18x10-3 mm2/s (±0.02). Low-, intermediate- and high-

grade tumours had a mADC of.1.30±0.30 x10-3 mm2/s, 1.07±0.30 x10-3 mm2/s and 0.94±0.30

x10-3 mm2/s respectively. ROC analysis showed a discriminatory performance of AUC=0.90 in

discerning low-grade from combined intermediate- and high-grade lesions.

Conclusion: 3T ADC values inversely relate to prostate cancer Gleason grades in the peripheral

zone. A high discriminatory performance is achieved in differentiating between low-,

intermediate- and high-grade cancer.

143


7


7

INTRODUCTION

Gleason grade of prostate cancer is an important determinant of biological activity and

aggressiveness. A vast body of literature has established Gleason score as one of the paramount

pathologic factors in predicting disease outcome in prostate cancer. In fact, the grading scheme

has now become so vital that it is often used as an integral piece of information in both

management and treatment stratification of patients with prostate cancer before and after

definitive therapy(1-5). Pre-treatment knowledge of final Gleason grade would be an important

advance, but currently, such information remains elusive.

Biopsy determination of Gleason grade often does not provide an accurate reflection of final

Gleason Grade, i.e., whole-organ pathology(6-8). Partin tables and risk stratification(9;10)

schemes that incorporate information from biopsy Gleason grades into decision making are

therefore rendered less accurate and less reliable. A definite need for a more accurate and non-

invasive method therefore exists to improve the accuracy of determination of true pretreatment

Gleason grades.

Diffusion Weighted MR imaging (DW-MRI) is a functional imaging technique that quantifies

random Brownian motion properties of water molecules (diffusion) in tissue. The degree of

restriction to water diffusion in biologic tissue is inversely correlated to tissue cellularity and the

integrity of cell membranes(11). Diffusion of molecules also occurs across tissues, especially

from areas of restricted diffusion to areas with free diffusion. The net displacement of molecules

is called the apparent diffusion coefficient (ADC). On MR imaging, the ADC can be calculated by

acquiring two or more images with a different magnetic field gradient duration and

amplitude (b-values). The contrast in the ADC map depends on the spatially distributed

diffusion coefficient of the acquired tissues and does not contain T1 and T2* values (12).

The role of DW-MRI in tumour localization within the prostate has been extensively reported

before(13-16). However, its use in stratifying low and high grade prostate cancer has not

received much attention and is limited to biopsy-determined Gleason grades (17;18).

144


7 Relation of ADC Values at 3T MRI with Prostate Cancer Gleason Grade in the Peripheral Zone

7


7

The purpose of our study was to determine the relationship between ADC values from 3T DW-

MRI and peripheral zone (PZ) prostate cancer Gleason grades determine from step-section

specimens after prostatectomy.


Patients

Between August 2006 and January 2009, 70 consecutive patients with biopsy proven prostate

cancer, scheduled for radical prostatectomy, were referred from the departments of urology at

the Radboud University Nijmegen Medical Centre and the Canisius Wilhelmina Hospital in

Nijmegen, Netherlands, for a clinically routine preoperative MRI of the prostate. The need for

informed consent was waived by the Institutional Review Board.

MR Imaging Protocol

MR imaging was performed using a 3T MR system (Siemens Trio Tim, Erlangen, Germany) with

the use of combined endorectal coil (ERC)(Medrad, Pittsburgh, U.S.A) and pelvic phased array

(PPA) coils. The ERC was filled with a 40-mL Perfluorocarbon preparation (Fomblin;Solvay-

Solexis, Milan, Italy). Peristalsis was suppressed with an intramuscular administration of 20-mg

Butylscopolaminebromide (Buscopan; Boehringer-Ingelheim, Ingelheim, Germany) and 1 mg of

glucagon (Glucagen; Nordisk, Gentofte, Denmark).

The imaging protocol, after evaluation of correct endorectal coil position with fast gradient echo

imaging, included the following sequences: first, T2-weighted turbo spin echo sequences were

performed with an in-plane resolution of 0.4 x 0.4 mm (TR 3250 ms/TE 116 m; flip angle 120;

15-19 slices; 3 mm slice thickness; echo train length 15; 180 x 180 mm field of view and 448 x

145


7


7

448 matrix) in axial, coronal and sagittal planes, covering the prostate and seminal vesicles.

Second, a single-shot-echo-planar imaging sequence with diffusion module and fat suppression

pulses was implemented. Water diffusion in 3 directions was measured using b-values of 0, 50,

500 and 800 s/mm2 and a TR of 2500ms, TE of 81 ms, parallel imaging factor 3, 15-19 slices,

3mm slice thickness and an in-plane resolution of 1.5 x 1.5 mm. ADC-maps were automatically

calculated by the scanner software using all 4 b-values.

Reconstructed whole-mount step-section preparation

Following radical prostatectomy, prostate specimens were uniformly processed and entirely

submitted for histological investigation. Immediately after surgical resection, specimens were

fixed in 10% neutral-buffered formalin, using fine needle formalin injections and fixation

overnight. Subsequently, the entire surface was marked with ink using three different colours,

after which the entire prostate specimen was cut into serial transverse 4 mm thick slices,

perpendicular to the dorsal-rectal surface and all slices were macroscopically photographed

with a CCD-camera. The apex and base were sagittally sectioned to assess the caudal and cranial

surgical margins. Seminal vesicles were amputated at their junction with the prostate and

sectioned parallel to their junction and embedded in total. The remaining slices were subdivided

into halves or quadrants to fit routine cassettes. After histological staining all specimens were

evaluated by one expert urological pathologist (C.H, 17 years experience). Tumours were

outlined on the microscopic slides and subsequently mapped on the macroscopic photographs

to allow reconstruction of tumour extent and multifocality. Each individual tumour was graded

according to the 2005 ISUP Modified Gleason Grading System (19). Tumours were staged

according to the 2002 TNM classification.

Annotations of MR images

Retrospectively, after radical prostatectomy, annotations of MR images were performed in

consensus by one radiologist (T.H) and one urologist (D.S). To achieve good objective spatial co-

alignment accuracy, a number of strategies were applied. First, both ADC maps and

146


7


7

prostatectomy step-sections were obtained perpendicular to the dorsal surface of the prostate.

Secondly, a similar slice thickness was chosen. Thirdly, objective mapping of MR slices to

prostatectomy step-sections, was performed by aligning the apex and base on MR and step-

sections in the cranial-caudal direction (Figure1A). Starting from the apex, each consecutive

ADC map was matched to the consecutive pathology step-section. Finally, a per-slice subdivision

was made. On each slice, the PZ and transition zone (TZ) were identified and using the urethra

as reference, the tumour maps were translated to a schematic subdivision of the peripheral zone

(Figure 1B) incl. anterior horns (AH), dorso-lateral region (DL) and dorsal segment (D) in both

left and right halves. The PZ, TZ and urethra are well visible on ADC maps, thus allowing the

schematic subdivision applicable to ADC maps as well. The urethra again served to identify the

AH, DL and D segments. This schematic mapping allowed objective translation of tumour

containing regions from prostatectomy to ADC maps with a high degree of certainty.

A Region-of-Interest (ROI) was annotated and drawn to match the size and extent of the tumour

(ROITumour) determined from histology, as closely as possible (Fig. 1). ROIs were also placed in

the contra lateral segment of the peripheral zone in mirror position (ROINorm_Mirror) to the tumour

and of similar size as the ROITumour. Normal regions were annotated purely to provide a visual

reference of heterogeneity within the peripheral zone compared to tumour values. Each

separate step-section - ADC slice match, was annotated as a different tumour ROI and normal

ROI. Only tumours originating in the peripheral zone were annotated. Annotation of the

following were omitted if applicable: 1) ROITumour of a particular slice if the corresponding

pathology step-section revealed a tumour < 5x5 mm. This was due to the limit in spatial

resolution of the DWI images. 2) ROINorm_Mirror of a particular slices if the tumour was extending

beyond the midline (no mirror possible) or a second tumour was present in the mirror position

of the first tumour,

147


7


7

Figure 1A. Matching prostatectomy step-sections with ADC maps according to a

systematic method. The prostate was cut into step-sections perpendicular to the dorsal

surface of the prostate (left). According to the number of step-sections obtained, the MR

images were divided into the same number of slices (centre). For each step-section, the

corresponding ADC map was identified.

148


7


7

B. The step-sections were subsequently schematically subdivided into three different

peripheral zone regions: AH (Anterior Horns), DL (Dorso-lateral) region, D (dorsal)

region. The position of the urethra was identified. The peripheral zone on the

corresponding ADC was similarly divided. The relative position of the tumour (T) was

therefore translatable and annotated on the ADC maps, to match size and distribution of

the tumour.

149


7


7

Histological tumour grading assessment

Following annotations of ROIs on ADC maps, one radiologist (T.H) together with one

genitourinary pathologist (C.H) re-evaluated all step-sections containing tumour. For each

tumour present in the peripheral zone of the prostate, separate Gleason grades were identified

and quantified in percentages of the tumour section volume. For each step-section, the primary,

secondary and tertiary tumour grade component were noted, referred hereafter as the Gleason

Grade Group. An additional qualitative grading per step-section was also made: a) low-grade

lesion consisting of grade 2 or 3 components only; b) intermediate-grade lesions consisting of

grade 4 as secondary or tertiary component (without any grade 5) c) high-grade lesions

consisting of grade 4 as primary and/or grade 5 as primary, secondary or tertiary component.

Each ROITumour on ADC maps was subsequently correlated to the matching Gleason grade group

and qualitative grade assessments made per step-section slice. An in-house developed MR

analytical software workstation was used to draw ROIs and summarize the median and standard

deviation (SD) of ADC values (in x10-3 mm2/s) calculated for each ROI.


To determine the relation between tumour median ADC (mADC) values and ordinal Gleason

grade groups, a linear mixed effect regression model with random tumour effect was used. This

mixed-model regression analysis incorporates the dependency of repeated measurements

within the same tumour. In an additional mixed-model analysis, the differences in mADC

between the three QG groups were estimated.

Apart from establishing a relation between ADC and Gleason score, the diagnostic accuracy of

using ADC in differentiating low-grade from combined intermediate- and high-grade tumours is

of clinical importance. To this end, for every tumour, the histopathology slice with the highest

Gleason grade was matched to the corresponding ADC slice, thereby identifying the mADC value

matching to the most aggressive part of the tumour. If identical highest Gleason grade

compositions were evident for different slices within the same tumour, the slice showing the

150



7


7

lowest mADC was used. The diagnostic accuracy (AUC) of mADC values in discriminating low-

vs. combined intermediate- and high-grade grade groups was quantified with the area under the

receiver operating characteristic curve (ROC). A significant difference was considered when p

<0.05. Statistical analyses were performed with SPSS software (SPSS, version 16.0.01, Chicago,

U.S.A).

RESULTS

Of the 70 consecutive patients, 56 had clinically significant peripheral zone tumours (>0.5cc). In

the remaining 14 patients, 11 patients had transition zone tumours only and in 3 patients,

peripheral zone tumours were < 0.5 cc in volume (with Gleason grade 2/3 components only). In

addition, 5 of the 56 patients were excluded due to: severe motion artifacts (3); widespread

intraprostatic hemorrhage (1) or severe ghosting artifacts on the MRI images (1). In the 51

prostatectomy specimens from these patients, histological analysis revealed a total of 62

different PZ tumours and 251 tumour lesions on different step-sections of the specimens. In

none of the patients tumours were identified with a volume < 0.5 cc and containing a Gleason

grade 4/5 component. In total 14 different Gleason grade groups were identified according to

primary, secondary and tertiary features present. The patient and tumour characteristics are

summarized in Table 1 and the ADC demographics summarized in Table 2.

The tumour mADC values showed an association with the 14 Gleason grade groups (Fig. 2). The

linear mixed model analysis showed an inverse relationship (slope -0.18x10-3 mm2/s, SE ± 0.04,

p<0.001) between the mADC and the three QG groups (Fig. 3). Additional mixed model analysis

revealed that the mADC difference between low- and intermediate grade tumours was 0.22x10-3

mm2/s (SE±0.03)(p<0.001). The difference between intermediate- and high-grade tumours was

0.14x10-3 mm2/s (SE±0.03) (p<0.001) and between low- and high-grade was 0.36x10-3 mm2/s

(SE±0.04) (p<0.001). Low-, intermediate- and high-grade tumours had a mADC of 1.30x10-3

mm2/s (SE±0.30), 1.07x10-3 mm2/s (SE± 0.30) and 0.94x10-3 mm2/s (SE± 0.30), respectively.

Overall, mADC values for mirror normal peripheral zone were 1.60x10-3 mm2/s (SE±0.25).

151



7


7

Number of patients 51

Clinical Demographics

- Median PSA ng/ml (range)

- Median Age yrs (range)

6.8 (1.7-42)

64 (49-69)

Pathology Demographics

Stage

- T2a

- T2c

- T3a

- T3b

- T4

5

23

18

4

1

Number of different PZ tumours

- Gleason 3+2

- Gleason 3+3

- Gleason 2+4

- Gleason 3+4

- Gleason 3+4+5

- Gleason 4+3

- Gleason 4+3+5

- Gleason 4+4

- Gleason 4+5

3

18

1

13

4

13

5

2

3

Table1. Patient clinical data, pathological stage , and Gleason Grade

152



7


7

Number Median ADC (±SE)

(x10-3s/mm2)

Peripheral Zone lesions

- Gleason 2+3

- Gleason 3+2

- Gleason 2+3+4

- Gleason 3+2+4

- Gleason 3+3

- Gleason 3+3+4

- Gleason 2+4

- Gleason 3+4

- Gleason 3+4+5

- Gleason 4+3

- Gleason 4+3+5

- Gleason 4+4

- Gleason 4+4+5

- Gleason 4+5

251

4

11

3

3

74

3

1

46

8

54

7

17

2

19

1.02 (±0.29)

1.40 (±0.18)

1.16 (±0.14)

0.95 (±0.04)

1.20 (±0.05)

1.36 (±0.26)

1.29 (±0.02)

1.25 (±0)

0.97 (±0.22)

0.99 (±0.11)

0.92 (±0.17)

0.79 (±0.15)

0.68 (±0.13)

0.74 (±0.02)

0.79 (±0.10)

Peripheral Zone lesions

a) Low-grade

b) Intermediate-grade

c) High-grade

94

50

107

1.30 (±0.30)

1.07 (±0.30)

0.94 (±0.30)

Normal mirror PZ 1861

1.60 (±0.25)

Table 2. Gleason grades and ADC Values of 251 Step Sections . 1 In 65 matches no normal

mirror PZ could be annotated due to presence of contra lateral tumour or tumour

extending beyond midline)

153


7


7

Using ROC analysis of the most aggressive part of the tumour only, mADC was able to

discriminate between low-grade and combined intermediate and high-grade tumours with an

AUC=0.90 (CI 0.81-0.98) (Fig. 4). Furthermore, it was noted that in 94% of tumours (58 of 62),

the ADC slice with lowest mADC for tumour was in exact concordance with the most aggressive

composition slice in pathology. Figure 5 shows the visibility of different grade tumours on ADC

maps.

Figure 2. Association between median tumour ADC vs. Gleason grade groups.

1554


7

154


7

Figure 3. Relation between median ADC vs. Qualitative grade groups and normal mirror

PZ using tumour slice with lowest mADC value. Linear mixed effect regression model

slope estimate -0.18x10-3mm2/s.

155


7


7

Figure 4. ROC curve of the discriminating performance of mADC to differentiate between

low-grade vs. intermediate- and high-grade lesions using tumour slice with highest

Gleason grade composition.

DISCUSSION

This study has shown that Gleason Grade, and by inference aggressivity, is related to ADC-MRI

values. Using a linear mixed model approach, we determined that the mADC significantly

decreased, on average 0.18x10-3 mm2/s per QG group interval. Further analysis showed a larger

difference in mADC between low and intermediate grade tumours compared to the difference

between mADC of intermediate and high-grade tumours. Using the most aggressive component

within the tumour as end-point, mADC revealed an AUC=0.90 in separating low-grade tumours

from combined intermediate- and high-grade ones.

The Gleason grade sub-grouping allows a better comparison and assessment of the effect of

microscopic glandular differentiation, growth features and structure of different prostatic

156


7


7

carcinoma sub-grades, on the free diffusivity of water. Correlation of ADC to qualitative grade

groups potentially allows a more practical utilization of the information in routine clinical

decision making, risk stratification and patient tailored treatment options. Furthermore, the

subdivision into low-, intermediate- and high-grade, can allow meaningful cut-off points to be

defined and used to help differentiate patient groups with different prognoses and therefore

different management needs.

DW-MRI is increasingly being incorporated into oncologic imaging and information obtained

from this technique is appealing as an imaging biomarker(20). The low ADC values found in

most tumours has been attributed to increased cellular density but diffusion can also be

influenced by fibrosis, glandular and stromal organization and shape(21). Within the prostate,

the predominant contribution of DW-MRI signal is from the extracellular component (from

tubular structures and their fluid content), with a lesser contribution from the extracellular

stromal space and the intracellular components (epithelial and stromal cells). Because of the

abundant self-diffusion of water molecules within the predominant tubular components within

the peripheral zone, their contents provides a high signal on ADC(22).

A rationale for the relation of prostate cancer aggressiveness with ADC can be suggested from

the current understanding of the structural and organizational features of epithelial, glandular

and extraductal components existing in different grades of cancer(23-25). With increasing

Gleason grade, the change in tissue organization to a more solid and compact architecture (with

higher cellular density) ought to be reflected in restrictions in the distances of free water motion

within the tissue. Well-differentiated prostate carcinomas display tubular formation with a

concomitant higher contribution of unrestricted water motion to ADC values. Lower grade

tumours are also known to have a remarkable heterogeneity in glandular size and ability to

grow between pre-existing ducts. In contrast, poorly-differentiated adenocarcinomas show

more expansile masses of small, tightly packed cell groups with small to absent lumina. Gleason

Grade 2 tumours are defined histologically by tightly packed, well-differentiated glandular

components, while grade 3 tumours show wider spaced tubuli with heterogeneity in ductal size

and density, imposing less restrictions on extra-glandular free water diffusivity motion. This

basis also seems to be reflected in the slightly lower ADC for tumours with grade 2 component

compared to tumours that are purely grade 3. Of the different Gleason grade groups, pure grade

157


7


7

3 (3+3) tumours show the largest variation in mADC possibly reflecting the heterogeneity of

sparse vs. dense growing lesions, akin to these. A large space for diffusion both between ducts

and within ductal lumina in well-differentiated compared to poorly differentiated tumours is the

most likely explanation for the observed differences in ADC values in low-grade compared to

high-grade tumours.

Despite the fact that true Gleason score is not represented in transrectal ultrasound guided

biopsy cores in 30-50% of patients, biopsy determined GS remains one of the most important

factors in decision-making. An accurate non-invasive method that improves prediction of

prostate cancer GS may allow a significant improvement in patient management by better

treatment selection, performing more targeted and therefore Gleason representative biopsies,

better risk-stratification and follow-up in patients on active surveillance protocols as well as

planning intensity modulated radiotherapy to the dominant aggressive component.

For correlation analysis, each step-section containing tumour was matched to an ADC map as a

separate tumour lesion. The reasoning behind this approach was that tumours display

remarkable intratumoural heterogeneity in their Gleason grade patterns and ability to grow in-

between existing normal ducts and stromal tissue. This is evident for example, when for the

same tumour, different sections reveal pure grade 4 components for the one, a mixed of grade 4

and 3 for another and a pure grade 3 for a final. As the ADC maps and pathology section are

matched to a high degree of certainty and for each an individual Gleason grade and ADC

quantification made, a better matching on the assessment of the effect of Gleason grade on water

diffusivity is obtained. On DWI, the slice with tumour showing the lowest mADC values will in

clinical practice most often be used prospectively to predict aggressivity, guide therapy or direct

targeted biopsies. Because this study was set up as a validation study,, data selection for ROC

analysis was done by choosing the tumour slice with the highest Gleason grade composition (i.e.

tumour slice with the highest proportions of Gleason grade 4 or 5 components). In 94% of

tumours, this was the exact same slice that showed the lowest mADC for the tumour, therefore

indicating that in a prospective setting, using the tumour slice with lowest mADC as starting

point,might be useful.

The clinical relevance of this imaging biomarker has noticeable potential. On a solitary basis,

mADC may contribute in risk-stratifying patients. With a good discriminatory performance

158


7


7

between low-, intermediate- and high-grade tumours, incorporating this information into

decision-making, will depend on the clinical question and off-course the particular sensitivity

and specificity that is desired. Differentiating men who can be managed expectantly (active

surveillance) from those requiring active management is therefore a potential application of

DWI-MRI (26). Patient with high-grade cancer including Gleason grade 4 as primary, or Gleason

grade 5 as primary, secondary or tertiary pattern, represent a group with a particularly

detrimental prognosis(27-29). Non-invasively identifying these patients before surgery could

be of importance to avoid unnecessary surgery, consider early adjuvant therapy or warrant

additional diagnostic test for metastasis assessment. A prospective advantage of identifying the

most abnormal part of the tumour based on mADC is that this can facilitate targeted biopsies to

obtain cores from the regions with the worst Gleason scores providing a better basis for further

patient management. Furthermore, when focal therapy (i.e. IMRT) is used, the most aggressive

component could receive the highest dose and therefore improve outcome. With currently

available prognostic factors such as preoperative PSA, stage, biopsy Gleason grading, such a

selection cannot be made with sufficient accuracy on an individual level(30). Our findings

suggest a potential role that DWI-MR imaging can play as a non-invasive adjuvant in

characterizing prostatic carcinomas. To which degree our findings will in practice affect

individual patient management, should be assessed by future prospective studies.

Though multi-parametric MR imaging has firmly defined its role in accurately staging and

localizing prostate cancer(31-34), limited data is currently available on the value in improving

the prediction of prostate cancer aggressiveness. A correlation of H-MRS determined

choline+creatine/citrate ratios at 1.5T to prostatectomy GS has been reported(35;36), however,

the overlapping groups appear to be too large to determine meaningful cut-off points. Further

observations have confirmed that T2-w signal intensity correlates to GS(37), as poorly

differentiated tumours are more readily detected on T2-w imaging compared to well

differentiated ones(38). A correlation between ADC values and prostate cancer cellularity,

proliferation activity and density of growth has recently been demonstrated in two studies

(39;40). Additionally, a correlation between ADC and biopsy determined GS has been reported

by Tamada et al.(41). These authors have shown a significant correlation ( =0.497, 0.0001)

of the biopsy Gleason grade findings with ADC. Furthermore, the same visual trend in

159


7


7

association between Gleason grades and ADC as shown in our Gleason grade group vs. mADC

was also shown.

Our study had a number of limitations. We did not include transition zone (TZ) tumours in this

study. TZ tumours are known to have different genetic mutations, biological behavioral features

and prognoses(42-44). Therefore the conclusions drawn from this study cannot be applied to

TZ tumours which are known to have different ADC values compared to the PZ(45). Another

potential limitation is the reliability of the method of matching axial MR images to step-section

maps from histology we used(31;36;46). We believe that using a number of strategies to

improve the spatial mapping of MR images and step-sections, allowed us to obtain good

matching with a high degree of certainty. Following slice-by-slice matching of step-sections to

ADC maps, we annotated ROIs based on a schematic translation of the ground truth based on

zonal subdivision and urethral land marking.

Having demonstrated the relationship between ADC and tumour aggressivity, in the future it is

necessary to prospectively validate the ability of DW-MR imaging to improve risk stratification

on an individual patient basis, in addition to clinical parameters through a prospective multi-

reader study. Evaluating the impact of such stratification on patient management is also

necessary. Reproducibility of absolute ADC values between vendors and field-strengths as well

as correlating ADC with TZ cancers, should have future priority as well.

CONCLUSIONS

Our conclusions are that quantitative DW-MR imaging may be a well-suited non-invasive

biomarker for prostate cancer aggressiveness. Median tumour ADC values inversely relate to

Gleason grade groups and qualitative grade groups. A high discriminatory accuracy of

AUC=0.90 suggests that ADC will prove to be a useful biomarker that can help improve

identification of patients with particular tumour aggressiveness risk.

160


7


7

Figure 5. Histological step-section and corresponding ADC maps for three patients with

tumours of different aggressivity. Window levels were kept the same for all patients.

Patient with a low-grade tumour (T) - Gleason grade 3+3 (1) and tumour mADC of 1.24

(x10-3mm2/s). Intermediate-grade tumour Gleason grade 3+4 (2) in patient where

mADC of tumour was 0.99 (x10-3mm2/s). Patient (3) with high-grade tumour Gleason

4+5 with tumour mADC of 0.66 (x10-3mm2/s). Tumour region on ADC indicated with red

dashed ROI.

161


7


7

REFERENCES

1. Blute ML, Bergstralh EJ, Iocca A, Scherer B, Zincke H. Use of Gleason score, prostate specific antigen, seminal vesicle and margin status to predict biochemical failure after radical prostatectomy. J.Urol. 2001 Jan;165(1):119-25.

2. Brachman DG, Thomas T, Hilbe J, Beyer DC. Failure-free survival following brachytherapy alone or external beam irradiation alone for T1-2 prostate tumours in 2222 patients: results from a single practice. Int.J.Radiat.Oncol.Biol.Phys. 2000 Aug 1;48(1):111-7.

3. Egevad L, Granfors T, Karlberg L, Bergh A, Stattin P. Prognostic value of the Gleason score in prostate cancer. BJU.Int. 2002 Apr;89(6):538-42.

4. Egevad L, Granfors T, Karlberg L, Bergh A, Stattin P. Percent Gleason grade 4/5 as prognostic factor in prostate cancer diagnosed at transurethral resection. J.Urol. 2002 Aug;168(2):509-13.

5. Narain V, Bianco FJ, Jr., Grignon DJ, Sakr WA, Pontes JE, Wood DP, Jr. How accurately does prostate biopsy Gleason score predict pathologic findings and disease free survival? Prostate 2001 Nov 1;49(3):185-90.

6. Divrik RT, Eroglu A, Sahin A, Zorlu F, Ozen H. Increasing the number of biopsies increases the concordance of Gleason scores of needle biopsies and prostatectomy specimens. Urol.Oncol. 2007 Sep;25(5):376-82.

7. Kvale R, Moller B, Wahlqvist R, Fossa SD, Berner A, Busch C, Kyrdalen AE, Svindland A, Viset T, Halvorsen OJ. Concordance between Gleason scores of needle biopsies and radical prostatectomy specimens: a population-based study. BJU.Int. 2008 Dec 22.

8. Rajinikanth A, Manoharan M, Soloway CT, Civantos FJ, Soloway MS. Trends in Gleason score: concordance between biopsy and prostatectomy over 15 years. Urology 2008 Jul;72(1):177-82.

9. Dong F, Kattan MW, Steyerberg EW, Jones JS, Stephenson AJ, Schroder FH, Klein EA. Validation of pretreatment nomograms for predicting indolent prostate cancer: efficacy in contemporary urological practice. J.Urol. 2008 Jul;180(1):150-4.

10. Makarov DV, Trock BJ, Humphreys EB, Mangold LA, Walsh PC, Epstein JI, Partin AW. Updated nomogram to predict pathologic stage of prostate cancer given prostate-specific antigen level, clinical stage, and biopsy Gleason score (Partin tables) based on cases from 2000 to 2005. Urology 2007 Jun;69(6):1095-101.

11. Uhl M, Altehoefer C, Kontny U, Il'yasov K, Buchert M, Langer M. MRI-diffusion imaging of neuroblastomas: first results and correlation to histology. Eur.Radiol. 2002 Sep;12(9):2335-8.

12. Pagani E, Bizzi A, Di SF, De SN, Filippi M. Basic concepts of advanced MRI techniques. Neurol.Sci. 2008 Oct;29 Suppl 3:290-5.


14. Kim CK, Park BK, Lee HM, Kwon GY. Value of diffusion-weighted imaging for the prediction of prostate cancer location at 3T using a phased-array coil: preliminary results. Invest Radiol. 2007 Dec;42(12):842-7.

15. Miao H, Fukatsu H, Ishigaki T. Prostate cancer detection with 3-T MRI: Comparison of diffusion-weighted and T2-weighted imaging. Eur.J.Radiol. 2007 Feb;61(2):297-302.

16. Somford DM, Futterer JJ, Hambrock T, Barentsz JO. Diffusion and perfusion MR imaging of the prostate. Magn Reson.Imaging Clin.N.Am. 2008 Nov;16(4):685-95, ix.

17. deSouza NM, Riches SF, Vanas NJ, Morgan VA, Ashley SA, Fisher C, Payne GS, Parker C. Diffusion-weighted magnetic resonance imaging: a potential non-invasive marker of tumour aggressiveness in localized prostate cancer. Clin.Radiol. 2008 Jul;63(7):774-82.

18. Van AN, Charles-Edwards E, Jackson A, Jhavar S, Reinsberg S, Desouza N, Dearnaley D, Bailey M, Thompson A, Christmas T, et al. Correlation of diffusion-weighted MRI with whole mount radical prostatectomy specimens. Br.J.Radiol. 2008 Jun;81(966):456-62.


162


7


7

20. Padhani AR, Liu G, Mu-Koh D, Chenevert TL, Thoeny HC, Takahara T, Dzik-Jurasz A, Ross BD, Van CM, Collins D, et al. Diffusion-weighted magnetic resonance imaging as a cancer biomarker: consensus and recommendations. Neoplasia. 2009 Feb;11(2):102-25.

21. Chenevert TL, Sundgren PC, Ross BD. Diffusion imaging: insight to cell status and cytoarchitecture. Neuroimaging Clin.N.Am. 2006 Nov;16(4):619-ix.

22. Tamada T, Sone T, Jo Y, Yamamoto A, Yamashita T, Egashira N, Imai S, Fukunaga M. Prostate cancer: relationships between postbiopsy hemorrhage and tumour detectability at MR diagnosis. Radiology 2008 Aug;248(2):531-9.

23. Histological grading of prostatic carcinoma. In Pathology of the Prostate. Churchill Livingstone; 1990. p. 83-93.

24. Amin MB, Crignon DJ, Humphrey PA, Srigley JR. Gleason Grading of Prostate Cancer: A contemporary apprroach. Lippincott Williams & Wilkins; 2004.

25. Gleason DF. Histologic grading of prostate cancer: a perspective. Hum.Pathol. 1992 Mar;23(3):273-9.

26. Bastian PJ, Carter BH, Bjartell A, Seitz M, Stanislaus P, Montorsi F, Stief CG, Schroder F. Insignificant Prostate Cancer and Active Surveillance: From Definition to Clinical Implications. Eur.Urol. 2009 Mar 6.

27. Chan TY, Partin AW, Walsh PC, Epstein JI. Prognostic significance of Gleason score 3+4 versus Gleason score 4+3 tumour at radical prostatectomy. Urology 2000 Nov 1;56(5):823-7.

28. Hattab EM, Koch MO, Eble JN, Lin H, Cheng L. Tertiary Gleason pattern 5 is a powerful predictor of biochemical relapse in patients with Gleason score 7 prostatic adenocarcinoma. J.Urol. 2006 May;175(5):1695-9.

29. Whittemore DE, Hick EJ, Carter MR, Moul JW, Miranda-Sousa AJ, Sexton WJ. Significance of tertiary Gleason pattern 5 in Gleason score 7 radical prostatectomy specimens. J.Urol. 2008 Feb;179(2):516-22.

30. Epstein JI, Partin AW, Sauvageot J, Walsh PC. Prediction of progression following radical prostatectomy. A multivariate analysis of 721 men with long-term follow-up. Am.J.Surg.Pathol. 1996 Mar;20(3):286-92.



33. Reinsberg SA, Payne GS, Riches SF, Ashley S, Brewster JM, Morgan VA, deSouza NM. Combined use of diffusion-weighted MRI and 1H MR spectroscopy to increase accuracy in prostate cancer detection. AJR Am.J.Roentgenol. 2007 Jan;188(1):91-8.


35. Shukla-Dave A, Hricak H, Kattan MW, Pucar D, Kuroiwa K, Chen HN, Spector J, Koutcher JA, Zakian KL, Scardino PT. The utility of magnetic resonance imaging and spectroscopy for predicting insignificant prostate cancer: an initial analysis. BJU.Int. 2007 Jan 11.

36. Zakian KL, Sircar K, Hricak H, Chen HN, Shukla-Dave A, Eberhardt S, Muruganandham M, Ebora L, Kattan MW, Reuter VE, et al. Correlation of proton MR spectroscopic imaging with gleason score based on step-section pathologic analysis after radical prostatectomy. Radiology 2005 Mar;234(3):804-14.

37. Wang L, Mazaheri Y, Zhang J, Ishill NM, Kuroiwa K, Hricak H. Assessment of biologic aggressiveness of prostate cancer: correlation of MR signal intensity with Gleason grade after radical prostatectomy. Radiology 2008 Jan;246(1):168-76.

38. Ikonen S, Karkkainen P, Kivisaari L, Salo JO, Taari K, Vehmas T, Tervahartiala P, Rannikko S. Magnetic resonance imaging of prostatic cancer: does detection vary between high and low gleason score tumours? Prostate 2000 Apr 1;43(1):43-8.

39. Wang XZ, Wang B, Gao ZQ, Liu JG, Liu ZQ, Niu QL, Sun ZK, Yuan YX. Diffusion-weighted imaging of prostate cancer: correlation between apparent diffusion coefficient values and tumour proliferation. J.Magn Reson.Imaging 2009 Jun;29(6):1360-6.

40. Langer DL, van der Kwast TH, Evans AJ, Sun L, Yaffe MJ, Trachtenberg J, Haider MA. Intermixed normal tissue within prostate cancer: effect on MR imaging measurements of apparent diffusion coefficient and T2--sparse versus dense cancers. Radiology 2008 Dec;249(3):900-8.

163


7


7

41. Tamada T, Sone T, Jo Y, Toshimitsu S, Yamashita T, Yamamoto A, Tanimoto D, Ito K. Apparent diffusion coefficient values in peripheral and transition zones of the prostate: comparison between normal and malignant prostatic tissues and correlation with histologic grade. J.Magn Reson.Imaging 2008 Sep;28(3):720-6.


43. Guo CC, Zuo G, Cao D, Troncoso P, Czerniak BA. Prostate cancer of transition zone origin lacks TMPRSS2-ERG gene fusion. Mod.Pathol. 2009 Apr 24.

44. Noguchi M, Stamey TA, Neal JE, Yemoto CE. An analysis of 148 consecutive transition zone cancers: clinical and histological characteristics. J.Urol. 2000 Jun;163(6):1751-5.

45. Pickles MD, Gibbs P, Sreenivas M, Turnbull LW. Diffusion-weighted imaging of normal and malignant prostate tissue at 3.0T. J.Magn Reson.Imaging 2006 Feb;23(2):130-4.


164


7

— CHAPTER 8 —

Initial Experience with Identifying High-Grade Prostate Cancer using

Diffusion-Weighted MRI in Patients

Schematic TRUS-guided Biopsy.

T. Hambrock; D. Somford; van Oort et al.

CHAPTER 8CHAPTER 8

Initial Experience with Identifying High-grade Prostate Cancer using Diffusion Weighted Magnetic Resonance Imaging in Patients with a

Gleason Score Biopsy. A radical Prostatectomy Correlated Series.

Hambrock T, Somford D, van Oort I, Witjes J, Hulsbergen-van de Kaa C, van Basten J, Fütterer J, Barentsz J


Transrectal ultrasound guided biopsies of the prostate substantially undergrade prostate

cancer tumours.

Quantitative 3T Diffusion Weighted Imaging is able to accurately differentiate patients

e cancer who represent undergrading of true

Gleason score from those subjects where it is a correct assessment of true Gleason score

at radical prostatectomy

A high diagnostic accuracy using ADC values is achieved in separating patients with low

and high-grade prostate cancer.


Accurate pretreatment assessment of true prostate aggressiveness is of paramount

importance in selecting optimal treatment methods.

3T DWI imaging can be a valuable method in assessing pretreatment aggressiveness and

should be part of any patient diagnosed with prostate cancer.

Summary Statement

3T DWI MR imaging is a very valuable technique for accurately identifying patients in whom biopsies represent an underestimation

of prostate cancer aggressiveness.

166

Identifying High-Grade Prostate Cancer using DW-MRI in Patients with a Biopsy

8

ABSTRACT

Introduction: Diffusion-weighted magnetic resonance (MR) Imaging (DWI) might be able to

fulfill the need to accurately identify high-grade prostate carcinoma, in patients initially selected

for active surveillance in the PSA screening era based upon transrectal ultrasound (TRUS)-

guided biopsy Gleason score. We aimed to retrospectively determine whether DWI is able to

and/or 5 components in their radical prostatectomy (RP) specimen.

Materials and methods: Whole-mount RP specimens were used to identify regions of interest

(ROI) corresponding with tumour on the DWI-derived Apparent Diffusion Coefficient (ADC)

maps in 23 patients with a Glea ADC values were correlated with RP

Gleason grades. Statistical analysis was performed using the area under the ROC-curve (AUC) of

median ADC to separate patients with Gleason 4 and/or 5 components vs. those without. Mann-

Whitney U-testing was performed to detect differences in mean ADC values for tumours with

Results: Median ADC values had an AUC of 0.88 for identifying patients subject to TRUS-guided

biopsy undergrading using RP Gleason score as a reference. In patients harboring a Gleason 4

and/or 5 component the median ADC was 0.86 x10-3 mm2/s (±0.21), whereas patients harboring

no Gleason 4 and/or 5 component displayed a median ADC of 1.16x10-3 mm2/s (±0.19) for the

single tumour slice with lowest median ADC (p<0.002).

Conclusions: 3T DWI is accurate in predicting the presence of high-grade tumour in patients

wi

treatment selection.

167


8

INTRODUCTION

PSA testing for prostate carcinoma has led to earlier detection of prostate carcinoma in most

men, with a tendency to downstaging for the entire population(1;2). Recent publications from

the ERSPC trial showed a significant prostate cancer mortality reduction from PSA screening,

however at the cost of 1410 men screened and, more importantly, 48 additional cases of

prostate cancer treated to prevent one prostate cancer death(3).

The dilemma of the clinical insignificant tumour has been addressed with increasing frequency

and would even become more important with the implementation of PSA screening(4;5). In a

recent European Association of Urology position statement on PSA screening for prostate

cancer, the authors underline the paramount importance of the development of reliable

monitoring and prognostic markers and/or imaging modalities to prevent overtreatment before

widespread implementation of population-based PSA screening (6). Clinical staging and accurate

grade assessment of prostate carcinoma has become of utmost importance in decision making

regarding the need for active treatment at any time point following the diagnosis of prostate

cancer in individual cases.

Accurate identification of insignificant and/or low-risk prostate carcinoma remains the

cornerstone of selection of patients for active surveillance, but is currently severely hampered

by absence of reliable pre-treatment predictors(7;8). PSA levels in patients with histological

proven prostate carcinoma do grossly correlate with the risk of extraprostatic extension (EPE),

seminal vesicle invasion and positive surgical margins, but correlate poorly with

differentiation(9). Other prostate cancer markers, such as PCA3 or hK2, have not been able to

identify low-risk prostate cancers with sufficient accuracy for clinical decision making(10;11). In

current practice, transrectal ultrasound (TRUS)-guided schematic prostate biopsies are the

predominant method to obtain a histological diagnosis of prostate carcinoma and to determine

pathological characteristics of the tumour. Subsequently, biopsy-determined combined Gleason

score remains a cornerstone of pre-treatment risk stratification for localized prostate carcinoma.

However, when correlated with radical prostatectomy (RP) specimens, Gleason grading

obtained by TRUS-guided biopsy has been shown to underestimate tumour Gleason score in up

to 40% of cases(12;13), a phenomenon further referred to as Gleason undergrading in this

paper.

RP series including patients considered eligible for active surveillance according to

contemporary inclusion criteria showed that up to 27% of patients had a Gleason score of at

least 7 upon RP(14;15). An active surveillance series by Duffield et al. outlined that most

168


8

patients progressing on such a protocol did so 1 to 2 years after diagnosis, suggesting significant

undergrading upon initial TRUS-guided biopsy(16). Only 52% of patients consequently

need

for more accurate grading and staging of prostate carcinoma precluding inclusion in active

surveillance protocols.

Diffusion-weighted magnetic resonance (MR) Imaging (DWI) is a functional MR technique that

quantifies the freedom of movement of protons, predominantly a part of water molecules in

tissue. In prostate carcinoma the diffusion of water will be limited due to increased cellular

density of tumour compared with normal glandular prostate tissue, leading to lower apparent

diffusion coefficient (ADC) levels in prostate carcinoma when compared with benign prostate

tissue(17). Previous reports on the value of DWI in the detection and localization of prostate

cancer have been published(18-20). Furthermore the correlation of ADC to tumour Gleason

score and tumour volume has recently been shown(21-23). Therefore, another merit of DWI

might be in correctly identifying those patients that would have been selected for active

surveillance protocols based on their TRUS-guided bio

Gleason 4 and/or 5 components not sampled by random biopsies. In this series we aimed to

establish the potential value of DWI to identify patients subject to pre-operative Gleason

undergrading by TRUS-guided biopsy, using RP Gleason score as a gold standard, thus enabling

more accurate pre-treatment risk stratification and treatment decision making.


Study population

Inclusion criteria were histologically proven prostate cancer with a Gleason s

upon TRUS-guided biopsy in patients consequently scheduled for radical prostatectomy (RP).

Patients were referred for MRI from two hospitals, following the histological diagnosis of

prostate cancer by 8-10 core schematic TRUS-guided biopsies. In these patients, endorectal coil

multi-parametric MR imaging at 3 Tesla (3T) preceding RP was performed. Patients in whom the

diagnosis of prostate cancer was established using MR-guided biopsy were excluded. Patient

characteristics for the complete cohort were registered.

169


8

Imaging parameters


Germany) combined with an endorectal coil (Medrad, Pittsburgh, USA) in combination with a

pelvic phased array coil. The routine MR imaging protocol consisted of anatomical T2-weighted

turbo spin echo sequences in the axial, sagittal and coronal direction, covering the prostate and

seminal vesicles. Axial images were obtained perpendicular to the dorsal surface of the prostate

to facilitate comparison with whole-mount sectioned RP specimens. DWI was performed using a

fat saturated single-shot-echo-planar imaging sequence with 3-scan trace imaging with b-values

of 0, 50, 500, and 800 s/mm2. DWI images were obtained in 4:32 min. The scanner software

automatically calculated ADC maps using all b-values. Further imaging parameters are shown in

table 1.

Sequence

name Sequence

Type

TR (ms) TE (ms) Voxel size Slice thickness

b-values FOV (mm)

Matrix

T2-w axial TSE 3250 ms 108 ms 0.4x0.4 mm 3 mm - 192 448

T2-w coronal TSE 4100 ms 108 ms 0.5x0.5 mm 3 mm - 192 384

T2-w sagital TSE 3760 ms 108 ms 0.5x0.5 mm 3 mm - 192 384

DWI axial SS-EPI 2700 ms 81 ms 1.5x1.5 mm 3 mm 0, 50, 500, 800

180 120

Table 1. Imaging parameters at 3T. TSE = Turbo Spin Echo; SS-EPI = Single Shot Echo

Planar Imaging; TR = Repitition time; TE = Echo time; ms = milliseconds; b-values in

s/mm2; FOV = Field of view.

Specimen handling

Following RP, prostate specimens were fixed overnight in 10% neutral buffered formaldehyde

and routinely processed according to protocol(24). In brief, after inking of the surface, the

prostate specimen was cut into serial transverse 3-4 mm thick slices, perpendicular to the

dorsal-rectal surface and all slices were macroscopically photographed. The apex and base were

sagittally sectioned to assess the caudal and cranial surgical margins. Seminal vesicles were

amputated at their junction with the prostate and sectioned parallel to their junction and

embedded in total. The remaining slices were subdivided into halves or quadrants to fit routine

cassettes. After histological staining, all specimens were evaluated by one expert urological

170


8

pathologist (CH). Tumours were outlined on the microscopic slides and subsequently mapped

on the macroscopic photographs to allow reconstruction of tumour extent and multifocality. For

every RP specimen and for each separate tumour, the presence of each primary, secondary and

tertiary Gleason grade pattern as well as a combined Gleason score was recorded. For every

tumour slice a separate Gleason grade assessment was made. The presence of EPE was reported

for all cases.

Data retrieval

Retrospectively, after radical prostatectomy, annotations of MR images were performed in

consensus by one urologist (DS) and one radiologist (TH). To achieve good objective spatial co-

alignment accuracy, a number of strategies were applied. First, both ADC maps and

prostatectomy step-sections were obtained perpendicular to the dorsal surface of the prostate.

Secondly, a similar slice thickness was chosen. Thirdly, objective mapping of MR slices to RP

step-sections, was performed by aligning the apex and base on MR and step-sections in the

cranial-caudal direction. Starting from the apex, each consecutive ADC map was matched to the

consecutive pathology step-section. Finally, a per-slice subdivision was made. On each slice, the

peripheral zone (PZ) and transition zone (TZ) were identified and using the urethra as

reference, the tumour maps were translated to a schematic subdivision of the PZ (Figure 1)

including anterior horns, dorsolateral region and dorsal segment in both left and right halves.

The PZ, TZ and urethra are well visible on ADC maps, thus allowing the schematic subdivision

applicable to ADC maps as well. This schematic mapping allowed objective translation of

tumour containing regions from RP to ADC maps with a high degree of certainty. A region of

interest (ROI) was annotated and drawn to match the size and extent of the tumour determined from histology, as closely as possible. Separate ROI’s were placed on every ADC slice containing tumour. Therefore, for each tumour, multiple ROI’s were determined, depending on the tumour

extent on different step-sections. Tumour foci with a inplane area less than 5x5 mm were

excluded from analysis due to the limit in spatial resolution obtained with DWI (inplane voxel

sizes 1.5 x 1.5 mm).

171


8

Figure 1. Schematic translation of tumour regions from macroscopic step-section

histopathology maps to ADC maps. Left image shows prostatectomy step-section with

tumour in the left-peripheral zone. The middle image shows the subdivision of the

prostate in the peripheral zone (PZ – red dashed lines) and transition zone (TZ – blue

dashed line). The urethra (U) is identified in the centre. The peripheral zone is

furthermore subdivided using the urethra as reference landmark into left and right

anterior horns (AH), dorsolateral regions (DL) and dorsal (D) region. The schematic

subdivision is anatomically translated to the corresponding ADC maps (image on right)

with tumour translated and annotated in yellow dashed lines. For each tumour slice, the

pathologist made a separate mention of the presence and proportions of the primary,

secondary and tertiary Gleason grade components. Prim.= Primary; Second. = Secondary;

Tert. = Tertiary. From each of the designated ROI’s, median ADC values were calculated on a per-slice basis.

Following median ADC estimation, for the purpose of this study, the index tumour was identified

as the tumour within the whole RP specimen revealing the highest Gleason Score and having a

tumours revealed the same Gleason score, the tumour with the

largest volume was subsequently identified as the index tumour. For this tumour only, the slice

with the lowest median ADC was used for further analysis. As prior studies have identified ADC

values to correlate with Gleason score, the assumption was made that the region within the

tumour with the lowest ADC should also reflect the region with largest proportions of highest

Gleason grade components. To validate this, for the index tumour, the pathology slice with

highest Gleason grade components were matched to the corresponding ADC slices.

Subsequently this ADC slice was evaluated if it also represented the ADC slice where the lowest

median ADC value was identified. In addition, the index tumour volume and zonal location was

also noted

172


8


-guided biopsy were stratified according to

their final radical prostatectomy revealing the presence or absence of a Gleason 4 and/or 5

component. This resulted in two groups. The first, where the TRUS-guided biopsy Gleason score

and the second, where TRUS-guided

for the tumour slice with lowest ADC values was identified and matched to these two groups.

Area-under the receiver operating characteristic (AUC) curves were determined for the median

for median PSA values predicting correct Gleason grading versus Gleason undergrading was

calculated. The Mann-Witney U-test was performed to determine whether there was a

significant difference in mean PSA, mean ADC (of the single slice with the lowest median tumour

ADC) and mean index tumour volume for these groups. Level of significance was set at P<0.05.

Statistical analysis was performed using SPSS software (SPSS, version 16.0.01, Chicago, Illinois,

USA.)

RESULTS

Twenty-three patients with a TRUS- -

parametric 3T MR imaging before RP. The mean age was 61 years (range 42 - 69) with a mean

PSA of 8.0 ng/ml (range 1.7 - 37.5). In 23 prostatectomies, 56 different tumours were found.

The prevalence of PZ tumours was 68% (38/56) and for TZ tumours this was 32% (18/56).

In all patients, one index prostate cancer exceeding a volume of 0.2cc in the RP specimen was

identified. The median index tumour volume was 4.09 cc (range 0.31 – 28cc). Almost all index

tumour were located in the PZ (96%; 22/23). In one patient tumour substantially involved both

the PZ and TZ, therefore primary zone of origin was not determinable. A summary of the patient

and pathology characteristics is provided in Table 2.

Eleven of the 23 (48%) included patients had a primary or secondary Gleason 4 and/or 5

component in their final RP specimen, leaving 12 cases that were correctly identified as low-

grade prostate cancer by pre-operative TRUS-guided biopsy. Patients subject to Gleason

undergrading had a median PSA of 6.10 (range 1.7 - 37.5), while patients with a Gleason score of

RP had a median PSA of 6.08 (range 2.2 - 9.8; p=0.11). Furthermore, the median

index tumour volume in patients with Gleason undergrading was 6.62 cc (range 0.31 – 28 cc)

173


8

nge 0.36 –

10.01; p=0.006). None of the patients correctly identified by TRUS-guided biopsy as Gleason

undergraded by TRUS-guided biopsy displayed EPE upon RP.

The diagnostic accuracy of median ADC in correctly discriminating patients subject to pre-

operative Gleason undergrading was an AUC of 0.88 (95% CI: 0.64-1.00) (figure 2). In patients

harboring a primary or secondary Gleason 4 and/or 5 component the median ADC was 0.86 x

10-3 mm2/s (SD±0.21), whereas patients harboring no Gleason 4 and/or 5 component displayed

a significantly higher median ADC of 1.16x10-3 mm2/s (SD±0.19; p<0.002) for the tumour slice

with the lowest ADC (figure 3).

Retrospective analysis revealed that the index tumour slice with the lowest median ADC in all

patients also corresponded with the highest Gleason grade component of the tumour upon RP.

The diagnostic accuracy of mean PSA values in discriminating patients into these two groups

revealed an AUC of 0.58 (95% CI: 0.32 – 0.83).

No Undergrading Undergrading p-values

Number 12 11 N.A.

Median PSA value ng/ml (range) 6.08 (2.2 - 9.8) 6.10 (1.7 - 37.5) 0.11

Median Index Tumour Volume cc (range)

2.59 (0.36 – 10.01) 6.62 (0.31 – 28) 0.006 *

Stage T3 disease 0% (0/12) 82% (9/11) N.A.

Median ADC values x10-3 mm2/s (±SD) 1.16 (±0.19) 0.86 (±0.21) 0.002 *

Table 2. Patient, pathology and ADC characteristics

174


8

Figure 2. ROC curve for differentiation of low-grade (no Gleason 4 and/or 5 component)

and high-grade prostate carcinoma upon RP using median ADC in a TRUS-biopsy Gleason

Figure 3. Box-plot of median ADC of low-grade (no primary or secondary Gleason 4

and/or 5 component) and high-grade prostate carcinoma upon RP in a TRUS-biopsy

175


8

DISCUSSION

In this study we have primarily shown that median quantitative ADC values obtained from 3T

prostate cancer upon TRUS-guided biopsy represents undergrading of true Gleason score from

those subjects where it is a correct assessment of true Gleason score at radical prostatectomy.

Because it is known that biopsy Gleason score is a poor predictor of true Gleason score identified

in RP, a definite need exists to improve identification of undergraded patients as this has

important implications in treatment selection and prognostication. From our results, it seems

that DWI has a strong potential to fill this current gap in pretreatment aggressiveness

determination for prostate cancer.

DWI has been established to reliably localize areas of prostate cancer within a 3T multi-

parametric MR imaging approach. Reported ADC values for prostate cancer (1.13 – 1.38x10-3

mm2/s) and normal prostate tissue (1.58 – 1.95x10-3 mm2/s) differ widely (20;25;26), which to

some degree can be explained by different sequences using varied b-values, and thus obtaining

different levels of diffusion-weighting. Also, population-based differences in Gleason score

prevalences can also account for differences in ADC values for prostate cancer in different series.

Reliable pre-treatment grading of prostate cancer remains a major issue, especially with the

emergence of active surveillance programs for low-risk prostate cancer and the growing interest

for focal ablative therapies. A RP correlated series by Haider et al. showed a very promising role

of DWI in combination with T2-weighted imaging in the detection of significant prostate cancer,

tumour diameter > 4mm. They reported a sensitivity of 81%

and a specificity of 84% for T2-weighted MRI and DWI combined(27).

ADC-values have been shown to correlate with cellular density in human cancers. The

proliferative activity of low-grade prostate cancer is relatively lower than that of higher-grade

cancers. Low-grade tumours reveal low tumour cellularity, intermixed with various amounts of

normal prostatic stroma and glands as well as showing larger extracellular and glandular

luminal spaces compared with higher-grade tumours. The latter are characterized by higher

cellularity density and loss of glandular duct formation (28;29). As a consequence, the space for

free water movement both intraluminally and extracellularly, reduces, the higher the grade of

the tumour becomes. Based on the results from Wang et al.(30) it is evident that the ADC of

prostate cancer decreased with an increase in tumour cellularity and proliferation rate. This

association has also been shown by Zelhof et al(31).

176


8

Apart from inherent tumour cellularity, the degree of tumour intermixing with normal prostatic

tissue, also infers variation in ADC values between tumours. Langer et al.(32) identified sparse

vs. dens growing prostate tumours and determined a correlation with ADC. They identified that

all sparse growing tumours Therefore,

it appears that inherent tumour cellularity (which is related to the Gleason grade) as well as

intermixing pattern, are important factors that influence the diffusion characteristics of prostate

cancer on ADC.

The ability of ADC to predict biopsy Gleason score has been established in several series(33-35),

but this approach is methodologically hampered by the well-known phenomenon of Gleason

undergrading of true combined Gleason score by pre-operative biopsies. Two earlier reports on

the correlation of ADC and radical prostatectomy Gleason score have recently been published

showing a high diagnostic accuracy of ADC in predicting high-grade prostate cancer(36;37). To

our knowledge we are the first to report on the use of DWI in identifying patients subject to

Gleason undergrading upon TRUS-guided biopsy.

For selection of patients for active surveillance protocols or focal therapy, a reliable technique

with a high sensitivity for any Gleason 4 and/or 5 component could be a parameter of

paramount importance to increase reliability and safety of such protocols. In this retrospective

series we were able to identify cases subject to pre-operative biopsy Gleason undergrading with

great accuracy using median ADC of the most aggressive part of the tumour. It is also known that

higher PSA values as such are associated with increased odds of undergrading by TRUS biopsies.

Isariyawongse et al.(38) have shown that patients with PSA values between 10-20 ng/ml had

odds ratios of 2.11 compared to patients with PSA < 10 ng/ml for representing undergrading of

true GS in prostatectomy. Despite this, PSA values alone are insufficient for accurate

stratification in this regard. This was reaffirmed by the relative poor AUC of 0.58 achieved using

PSA values as classifier.

A major limitation of our series might be the establishment of pre-operative Gleason score based

upon 8-10 core TRUS-guided biopsies. However, while more extensive TRUS-guided biopsy

schemes have been shown to improve detection rates and decrease the rate of Gleason

undergrading, this issue still remains substantial(39). We therefore are of the opinion that the

effect of more TRUS-guided biopsies taken might not have altered the outcome of our series

significantly; further investigation in a series with extended biopsy schemes is however

warranted.

177


8

A fur

TRUS-guided biopsy for our analysis. However, as the clinical approach is based on a biopsy

-guided biopsy as representative of the true tumour features,

we opted to only include these patients for this series. This however resulted in a fairly low

number of patients in each subgroup. A further drawback is that although TZ tumours

represented 32% of all tumours found in our patients, the index tumour in 96% of patients (22

of 23) was a peripheral zone tumour. Because the normal TZ and PZ are known to have different

ADC values, our results may therefore be biased towards revealing the discriminatory

performance of ADC more in the light of PZ tumour undergrading. The larger cohort with more

patients harboring TZ index tumour undergrading is needed to view the value of ADC in a larger

perspective.

CONCLUSIONS

DWI has been established as a diagnostic modality in oncology now for over a decade. Its main

merits lie in the detection of solid tumour within surrounding normal tissue and more recently

research has focused upon the ability of DWI to characterize aggressiveness of tumours. We

confirmed this potential of DWI to characterize prostate cancer aggressiveness. DWI should be

an integral part of any multi-parametric MRI approach for prostate cancer, whether localizing or

characterizing the tumour is the aim. Its main contribution to the diagnostic arena for prostate

cancer might lie in its ability to identify high-grade components in prostate cancer precluding

adequate pre-treatment risk stratification and aiding in therapeutic decision making.

Prospective research will need to focus upon the performance of DWI in candidates for active

surveillance to predict and evaluate progression to curative therapy.

178


8

Figure 4. Patient with biopsy Gleason 3+3=6 and PSA 5.2 ng/ml. Histopathological step

sections 1a - 3a reveal the extent of prostate carcinoma in the peripheral zone (light-blue

area). For every seperate slice, a Gleason grade composition expressed in grade and

percentage of tumour region is given. Every histopathology slice is matched to the

corresponding ADC map (1b - 3b) and the tumour containing region translated for

placement of a ROI placed over the tumour. For each ADC slice, a seperate median ADC

(mADC) value was calculated. The mADC values are expressed in x10-3 mm2/s. The

region of tumour with the largest proportion of higher Gleason grades (3a) corresponds

also to the slice with the lowest mADC tumour values (3b). The window level for ADC

maps are defined to range from 0.5 - 1.5 x 10-3 mm2/s. A small additional insignificant

transition zone tumour (green region) is also shown in 1a. The red line indicates the area

of extraprostatic extension. The final diagnosis on RP was Gleason 3+4=7, stage pT3a.

179


8

REFERENCES

1. Postma R, Schroder FH, van Leenders GJ, Hoedemaeker RF, Vis AN, Roobol MJ, van der Kwast TH. Cancer detection and cancer characteristics in the European Randomized Study of Screening for Prostate Cancer (ERSPC)--Section Rotterdam. A comparison of two rounds of screening. Eur.Urol. 2007 Jul;52(1):89-97.

2. Cremers RG, Karim-Kos HE, Houterman S, Verhoeven RH, Schroder FH, van der Kwast TH, Kil PJ, Coebergh JW, Kiemeney LA. Prostate cancer: trends in incidence, survival and mortality in the Netherlands, 1989-2006. Eur.J.Cancer 2010 Jul;46(11):2077-87.

3. Schroder FH, Hugosson J, Roobol MJ, Tammela TL, Ciatto S, Nelen V, Kwiatkowski M, Lujan M, Lilja H, Zappa M, et al. Screening and prostate-cancer mortality in a randomized European study. N.Engl.J.Med. 2009 Mar 26;360(13):1320-8.


5. Jeldres C, Suardi N, Walz J, Hutterer GC, Ahyai S, Lattouf JB, Haese A, Graefen M, Erbersdobler A, Heinzer H, et al. Validation of the contemporary epstein criteria for insignificant prostate cancer in European men. Eur.Urol. 2008 Dec;54(6):1306-13.

6. Abrahamsson PA, Artibani W, Chapple CR, Wirth M. European Association of Urology Position Statement on Screening for Prostate Cancer. Eur.Urol. 2009 May 19.

7. Anast JW, Andriole GL, Bismar TA, Yan Y, Humphrey PA. Relating biopsy and clinical variables to radical prostatectomy findings: can insignificant and advanced prostate cancer be predicted in a screening population? Urology 2004 Sep;64(3):544-50.

8. Bastian PJ, Carter BH, Bjartell A, Seitz M, Stanislaus P, Montorsi F, Stief CG, Schroder F. Insignificant Prostate Cancer and Active Surveillance: From Definition to Clinical Implications. Eur.Urol. 2009 Mar 6.

9. Freedland SJ, Hotaling JM, Fitzsimons NJ, Presti JC, Jr., Kane CJ, Terris MK, Aronson WJ, Amling CL. PSA in the new millennium: a powerful predictor of prostate cancer prognosis and radical prostatectomy outcomes--results from the SEARCH database. Eur.Urol. 2008 Apr;53(4):758-64.

10. Raaijmakers R, de Vries SH, Blijenberg BG, Wildhagen MF, Postma R, Bangma CH, Darte C, Schroder FH. hK2 and free PSA, a prognostic combination in predicting minimal prostate cancer in screen-detected men within the PSA range 4-10 ng/ml. Eur.Urol. 2007 Nov;52(5):1358-64.

11. Hessels D, van Gils MP, van HO, Jannink SA, Witjes JA, Verhaegh GW, Schalken JA. Predictive value of PCA3 in urinary sediments in determining clinico-pathological characteristics of prostate cancer. Prostate 2010 Jan 1;70(1):10-6.

12. Fine SW, Epstein JI. A contemporary study correlating prostate needle biopsy and radical prostatectomy Gleason score. J.Urol. 2008 Apr;179(4):1335-8.

13. Narain V, Bianco FJ, Jr., Grignon DJ, Sakr WA, Pontes JE, Wood DP, Jr. How accurately does prostate biopsy Gleason score predict pathologic findings and disease free survival? Prostate 2001 Nov 1;49(3):185-90.

14. Griffin CR, Yu X, Loeb S, Desireddi VN, Han M, Graif T, Catalona WJ. Pathological features after radical prostatectomy in potential candidates for active monitoring. J.Urol. 2007 Sep;178(3 Pt 1):860-3.

15. Louie-Johnsun M, Neill M, Treurnicht K, Jarmulowicz M, Eden C. Final outcomes of patients with low-risk prostate cancer suitable for active surveillance but treated surgically. BJU.Int. 2009 Nov;104(10):1501-4.

16. Duffield AS, Lee TK, Miyamoto H, Carter HB, Epstein JI. Radical prostatectomy findings in patients in whom active surveillance of prostate cancer fails. J.Urol. 2009 Nov;182(5):2274-8.

17. Somford DM, Futterer JJ, Hambrock T, Barentsz JO. Diffusion and perfusion MR imaging of the prostate. Magn Reson.Imaging Clin.N.Am. 2008 Nov;16(4):685-95, ix.


19. Hosseinzadeh K, Schwarz SD. Endorectal diffusion-weighted imaging in prostate cancer to differentiate malignant and benign peripheral zone tissue. J.Magn Reson.Imaging 2004 Oct;20(4):654-61.

20. Mazaheri Y, Shukla-Dave A, Hricak H, Fine SW, Zhang J, Inurrigarro G, Moskowitz CS, Ishill NM, Reuter VE, Touijer K, et al. Prostate cancer: identification with combined diffusion-weighted MR imaging and 3D 1H MR spectroscopic imaging--correlation with pathologic findings. Radiology 2008 Feb;246(2):480-8.

21. Mazaheri Y, Hricak H, Fine SW, Akin O, Shukla-Dave A, Ishill NM, Moskowitz CS, Grater JE, Reuter VE, Zakian KL, et al. Prostate tumour volume measurement with combined T2-weighted imaging and diffusion-weighted MR: correlation with pathologic tumour volume. Radiology 2009 Aug;252(2):449-57.

180


8

22. Verma S, Rajesh A, Morales H, Lemen L, Bills G, Delworth M, Gaitonde K, Ying J, Samartunga R, Lamba M.

Assessment of aggressiveness of prostate cancer: correlation of apparent diffusion coefficient with histologic grade after radical prostatectomy. AJR Am.J.Roentgenol. 2011 Feb;196(2):374-81.

23. Hambrock T, Somford DM, Huisman HJ, van O, I, Witjes JA, Hulsbergen-van de Kaa CA, Scheenen T, Barentsz JO. Relationship between Apparent Diffusion Coefficients at 3.0-T MR Imaging and Gleason Grade in Peripheral Zone Prostate Cancer. Radiology 2011 May;259(2):453-61.

24. van O, I, Bruins HM, Kiemeney LA, Knipscheer BC, Witjes JA, Hulsbergen-van de Kaa CA. The length of positive surgical margins correlates with biochemical recurrence after radical prostatectomy. Histopathology 2010 Mar;56(4):464-71.

25. Pickles MD, Gibbs P, Sreenivas M, Turnbull LW. Diffusion-weighted imaging of normal and malignant prostate tissue at 3.0T. J.Magn Reson.Imaging 2006 Feb;23(2):130-4.

26. Sato C, Naganawa S, Nakamura T, Kumada H, Miura S, Takizawa O, Ishigaki T. Differentiation of noncancerous tissue and cancer lesions by apparent diffusion coefficient values in transition and peripheral zones of the prostate. J.Magn Reson.Imaging 2005 Mar;21(3):258-62.


28. Amin MB, Crignon DJ, Humphrey PA, Srigley JR. Gleason Grading of Prostate Cancer: A contemporary apprroach. Lippincott Williams & Wilkins; 2004.

29. Gleason DF. Histologic grading of prostate cancer: a perspective. Hum.Pathol. 1992 Mar;23(3):273-9.

30. Wang XZ, Wang B, Gao ZQ, Liu JG, Liu ZQ, Niu QL, Sun ZK, Yuan YX. Diffusion-weighted imaging of prostate cancer: correlation between apparent diffusion coefficient values and tumour proliferation. J.Magn Reson.Imaging 2009 Jun;29(6):1360-6.

31. Zelhof B, Pickles M, Liney G, Gibbs P, Rodrigues G, Kraus S, Turnbull L. Correlation of diffusion-weighted magnetic resonance data with cellularity in prostate cancer. BJU.Int. 2009 Apr;103(7):883-8.

32. Langer DL, van der Kwast TH, Evans AJ, Sun L, Yaffe MJ, Trachtenberg J, Haider MA. Intermixed normal tissue within prostate cancer: effect on MR imaging measurements of apparent diffusion coefficient and T2--sparse versus dense cancers. Radiology 2008 Dec;249(3):900-8.



35. Woodfield CA, Tung GA, Grand DJ, Pezzullo JA, Machan JT, Renzulli JF. Diffusion-weighted MRI of peripheral zone prostate cancer: comparison of tumour apparent diffusion coefficient with Gleason score and percentage of tumour on core biopsy. AJR Am.J.Roentgenol. 2010 Apr;194(4):W316-W322.

36. Hambrock T, Somford DM, Huisman HJ, van O, I, Witjes JA, Hulsbergen-van de Kaa CA, Scheenen T, Barentsz JO. Relationship between Apparent Diffusion Coefficients at 3.0-T MR Imaging and Gleason Grade in Peripheral Zone Prostate Cancer. Radiology 2011 May;259(2):453-61.

37. Verma S, Rajesh A, Morales H, Lemen L, Bills G, Delworth M, Gaitonde K, Ying J, Samartunga R, Lamba M. Assessment of aggressiveness of prostate cancer: correlation of apparent diffusion coefficient with histologic grade after radical prostatectomy. AJR Am.J.Roentgenol. 2011 Feb;196(2):374-81.

38. Isariyawongse BK, Sun L, Banez LL, Robertson C, Polascik TJ, Maloney K, Donatucci C, Albala D, Mouraviev V, Madden JF, et al. Significant discrepancies between diagnostic and pathologic Gleason sums in prostate cancer: the predictive role of age and prostate-specific antigen. Urology 2008 Oct;72(4):882-6.

39. San F, I, Dewolf WC, Rosen S, Upton M, Olumi AF. Extended prostate needle biopsy improves concordance of Gleason grading between prostate needle biopsy and radical prostatectomy. J.Urol. 2003 Jan;169(1):136-40.

181


8

CHAPTER 9

In vivo assessment of prostate cancer aggressiveness using MR

Spectroscopic Imaging at 3T

T. Kobus; T. Hambrock; C. Hulsbergen van de Kaa et al.

Leonardo da Vinci “Anatomical sketches”

CHAPTER 9CHAPTER 9

assessment of prostate cancer aggressiveness using MR

Spectroscopic Imaging at 3T with an endorectal coil

Kobus T, Hambrock T, Hulsbergen-van de Kaa C, Wright A, Barentsz J,

Heerschap A, Scheenen T


3T MR spectroscopic imaging is an in vivo technique to identify changes in prostate cancer

metabolism related to tumour aggressiveness.

The maximum Choline+Creatine/Citrate ratio, maximum Choline/Crreatine ratio, and

malignancy rating of a standardized threshold approach can all separate low from higher

grade tumours with considerable accuracy.


In vivo pretreatment knowledge of tumour aggressiveness and biochemical characteristics

can offer important information for patient treatment selection and assessment of tumour

response to therapy.

Summary Statement

3T 1H-MRSI offers potential for non-invasive assessment of prostate cancer aggressiveness.

184

In Vivo Assessment of Prostate Cancer Aggressiveness using MR Spectroscopic Imaging at 3T

9

ABSTRACT

Background: One of the most important clinical challenges in prostate cancer management is an

assessment of cancer aggressiveness.

Objective: To validate the performance of MR spectroscopic imaging (MRSI) of the prostate at

3T to assess tumour aggressiveness, based on the choline plus creatine to citrate ratio

(Cho+Cr/Cit) and choline to creatine ratio (Cho/Cr), using the Gleason score of the radical

prostatectomy (RP) specimen as the gold standard.

Design, Setting, and Participants: A total of 43 biopsy-proven prostate cancer patients with 53

clinically relevant tumour foci were retrospectively included in this study.

Measurements: Patients underwent a MRI and MRSI exam followed by RP. From MRSI, all

spectroscopy voxels containing tumour were selected by a radiologist guided by the

prostatectomy histopathology map only. For each tumour, two spectroscopists determined the

maximum Cho+Cr/Cit, Cho/Cr and malignancy rating by a standardized threshold approach,

incorporating both metabolic ratios. The maximum Cho+Cr/Cit, Cho/Cr, and malignancy ratings

showed a relation to tumour aggressiveness and so were used to differentiate between tumour

aggressiveness classes.

Results and limitations: The maximum Cho+Cr/Cit ratio, maximum Cho/Cr ratio, and

malignancy rating of a standardized threshold approach separated low from higher grade

tumours with areas under the receiver operating characteristic curves of 0.70, 0.74 and 0.78,

respectively.

Conclusions: MR spectroscopic imaging offers possibilities for an non-invasive

assessment of prostate cancer aggressiveness. The combination of the different metabolite ratios

was used with promising results for discrimination between different aggressiveness classes.

185


9

INTRODUCTION

Prostate cancer is the most frequently diagnosed non-cutaneous cancer in European men and

accounted for 9.3% of cancer-related deaths in men in 2008(1). A large European study(2)

showed that screening for prostate cancer by means of prostate-specific antigen (PSA) levels will

result in a mortality reduction of 20% at the cost of overdiagnosis of many indolent cancers. If

unnecessary side-effects due to overtreatment of indolent tumours are to be prevented, while at

the same time all aggressive tumours are to be treated, accurate discrimination between

indolent and life-threatening cancers is essential.

Generally, transrectal ultrasound-guided biopsies are performed to confirm the presence of

prostate cancer and to determine the Gleason score of the tumour. However, the multi-focal

nature and heterogeneity of these tumours cause sampling errors and may lead to

underestimation of their aggressiveness. Several studies demonstrate discrepancies between the

Gleason score identified in biopsies and the subsequent radical prostatectomy (RP)

specimens(3, 4). For optimal diagnosis the most aggressive tumour focus should be identified.

Proton magnetic resonance spectroscopic imaging (1H-MRSI) provides spatial mapping of the

tissue levels of the metabolites citrate, choline and creatine in the whole prostate gland(5, 6).

Prostate cancer tissue is characterized by lower citrate levels and/or higher choline levels

compared to normal tissue(7), resulting in the ratio of choline and creatine to citrate

(Cho+Cr/Cit) as a marker for prostate cancer(7, 8). MRI and 1H-MRSI have detected high grade

tumours in patients with elevated PSA with high sensitivity(9). Other studies found a

correlation(10) and a trend(11) between the Gleason score and the Cho+Cr/Cit ratio at 1.5T

with the use of an endorectal coil. No relation with aggressiveness was found at 1.5T without the

use of an endorectal coil(12). At 3T also no relation was found using just body-array coils(6).

The use of an endorectal coil increases the signal to noise and might provide enough sensitivity

at 3T to classify tumour aggressiveness.

Incorporation of the choline/creatine (Cho/Cr) ratio is interesting since choline supposedly

increases in malignant tissue due to altered phospholipid metabolism(13) and increased choline

to creatine ratios have been used in standardized scoring systems for 1H-MRSI of the

prostate(14, 15). High-resolution Magic-Angle-Spinning NMR of prostate biopsies

186


9

showed a significant correlation of the Gleason score to the Cho/Cr ratio, among other

ratios(16). These studies suggest that the Cho/Cr ratio has additional potential to classify

tumour aggressiveness .

The purpose of this study is to validate the performance of 1H-MRSI of the prostate at 3T with an

endorectal coil to assess tumour aggressiveness based on the Cho+Cr/Cit and Cho/Cr ratio using

the Gleason score from histopathology of the RP specimen as the gold standard.


Patients

This study was approved by the institutional ethics review board and the need for informed

consent was waived for the retrospective study. 108 consecutive patients with biopsy-proven

prostate cancer had a 3T MR exam between October 2006 and February 2009 as well as a RP

between October 2006 and April 2009. Of these patients, 72 had a 1H-MRSI exam and were

retrospectively selected for this study. Patients were excluded if they were not examined with an

endorectal coil (n=4); had prior neoadjuvant therapy (n=9); had no tumour foci with a volume of

at least 0.5cc according to the histopathological analysis (n=14); or had no reliable

histopathology (n=2).

MR data acquisition

All MR-exams were performed on a 3T MR system (Magnetom Trio, Siemens, Erlangen,

Germany). An endorectal surface coil (Medrad, Pittsburgh, PA) was combined with body-array

coils for signal reception. In all patients an intramuscular injection of 1mg glucagon (Glucagen,

Nordisk, Gentofte, Denmark) and an injection of 20 mg of Butylscopolaminebromide (Buscopan,

Boehringer-Ingelheim, Ingelheim, Germany) were used to suppress peristalsis.

Fast gradient-echo-sequences were used to check the position of the coils. High spatial

resolution T2-weighted images were made in three directions. A prostate specific MRSI

sequence was used(17), including suppression of water and fat signals and an adapted sampling

scheme(5, 18). The actual volume of the spherical voxels was 0.37 or 0.64cm3(5).

187


9

Histopathological analysis

All RP specimens were uniformly processed and completely embedded according to a clinical

protocol(19). After formalin fixation and inking of the surface, the prostate specimen was

serially sectioned at 4 mm intervals, perpendicular to the dorsal-rectal surface, and all slices

were macroscopically photographed. One urological pathologist (C.A.H.K.) evaluated all RP

specimens and outlined for each slice the location of the tumour(s) on the photographs. Each

tumour focus was graded according to the 2005 ISUP Modified Gleason Grading System(20) and

each patient was staged following the 2002 TNM classification(21).

The primary, secondary and tertiary Gleason grades were used for a qualitative grading of the

tumour aggressiveness. Tumours were classified as low grade if consisting only of grades 2

and/or 3; intermediate grade with a secondary or tertiary grade of 4, but no 5 component; or

high grade with 4 as primary and/or 5 as primary/secondary or tertiary grade (Table 1).

Voxel selection

One radiologist (T.H.), blinded to the spectra, used the results of the histopathological analysis to

assign all spectroscopic voxels containing tumour tissue on high resolution T2-weighted images

with the voxel matrix of the spectroscopic exam projected over these images (Fig. 1c). Only

clinically significant tumours with a minimal size of 0.5cc(22, 23) on the histopathologic analysis

were included. Figure 1 shows an example of a T2-weighted image and corresponding

histopathology.

The signals of interest were fitted with a prototype software package (Metabolite report,

Siemens Medical Solutions, Germany) and the Cho+Cr/Cit and Cho/Cr ratios were calculated

automatically. Two spectroscopists independently (T.K. and T.W.J.S.) inspected the quality of the

automated fit for sufficient signal to noise, the absence of lipid signals and the absence of

baseline distortions. Only voxels that were approved by both spectroscopists were included in

the remainder of the study.

For every tumour, the voxel with the highest Cho+Cr/Cit ratio and the voxel with the highest

Cho/Cr ratio were determined. The highest Cho+Cr/Cit and Cho/Cr values of all data from each

188


9

aggressiveness class were checked for normality and compared with each other using a Kruskal-

Wallis test. The Pearson correlation coefficient was used to determine the correlation between

Number of patients 43

Mean PSA level ** 8.33 ng/ml (range 2.08 - 40.96)

Mean age 61 years (range 42-70)

Clinical stage T2 T3 T4

19 23 1

Gleason score

Peripheral zone:

3+2

3+3

3+3+4

3+4

3+4+5

4+3

4+3+5

4+4

4+5

Central gland:

2+3

3+2

3+2+4

2+4+5

4+3

4+3+5

Both zones:

4+3

4+4

Number of tumours:

2

10

1

10

3

9

2

1

2

2

3

1

1

1

3

1

1

low: 17

intermediate: 12

high: 24

Table 1. Patient and tumour characteristics

* The tumours are classified according to their aggressiveness as explained in the

materials and method section.

** PSA = Prostate specific antigen

189


9

the two ratios. To test the correlation between the ratios and the different aggressiveness

classes the Kendall-tau rank-correlation coefficient was determined.

To enable incorporation of both metabolite ratios in discrimination of aggressiveness classes the

standardized threshold approach(14, 15) was used. The conventional standardized threshold

approach uses an initial malignancy rating (ranging from 1, definitely benign, to 5, definitely

malignant) based upon the mean and standard deviation of the Cho+Cr/Cit ratio of non-cancer

tissue (distinction is made between peripheral zone (PZ) and central gland (CG)) with an

adjustment to the initial rating based upon the Cho/Cr ratio (Table 2.)(15). This malignancy

rating was determined for all accepted voxels in each tumour using the mean Cho+Cr/Cit ratios

and standard deviations of non-cancer tissue described earlier(6). In order to optimize the

standardized scoring system for aggressiveness assessment with 3T data, the Cho/Cr cut off

level, which adjusts the initial malignancy rating, was varied between 1 and 4 with 0.1 intervals.

For all 31 of these Cho/Cr adaptation levels the highest malignancy rating per tumour was

calculated and used to obtain ROC curves to determine the accuracy.

Five-point standardized scoring system

Score and Score definition PZ Cho+Cr/Cit ratio

CG Cho+Cr/Cit ratio

Cho/Cr ratio adjustment

adjust 3 and 2 into 4.

If Cho/Cr ratio < 2, then: adjust 5 into 4 and 4 into 3.

1: Definitely benign tissue

2: Probably benign tissue

3: Possibly malignant tissue

4: Probably malignant tissue

5: Definitely malignant tissue

>0.70 >0.90

PZ = peripheral zone, CG = central gland, Cho+Cr/Cit and CC/C = choline plus creatine to citrate, Cho/Cr = choline to creatine

Table 2. Conventional Standardized Threshold Approach definitions

190


9


The AUCs of the ROC curves were determined to study the performance of discrimination of low

grade from higher grade tumours and high grade from lower grade tumours using the maximum

Cho+Cr/Cit, maximum Cho/Cr and the 5 point scale of the standardized threshold approach. All

AUCs were compared for statistical differences(24).

Statistical analyses were performed with GraphPad Prism (GraphPad Software Inc, La Jolla, USA)

and Matlab (The Mathworks Inc, Natick, USA). For all statistical tests a P-value of 0.05 was used

to show significance.

RESULTS

Of the 108 patients, 43 passed the inclusion criteria. A summary of patient and tumour

characteristics is given in Table 1. The mean time between the MR exam and RP was 6 weeks

(range 0-21 weeks). The MR-exam (example in Fig. 1) was performed on average 46 days after

transrectal ultrasound-guided biopsy (range 19-107 days). The total number of clinically

significant tumours was 53 in 43 patients. 40 tumours were located in the PZ, 11 in the CG and 2

tumours covered both zones (assigned as PZ tumour for analysis). The average tumour size on

histopathology was 6.3 cm3 (range 0.52-33.5 cm3). The patient Gleason scores of the tumours

that were excluded because the volume was smaller than 0.5cc were 2+3 (n=2), 3+3 (n=9), 3+4

(n=2) and no malignancy detected (n=1).

In total, 1892 tumour voxels were selected by the radiologist. Of these voxels, 77% (1463/1892)

passed the quality inspection of both spectroscopists. Three patients, and therefore 5 tumours,

had to be excluded from the analysis, since none of the spectra of the selected voxels were

usable.

A small, but significant correlation was found between the maximum Cho+Cr/Cit ratio and the

aggressiveness classes (p=0.02, r=0.27)(Fig 2). The comparison of the medians of the three

aggressiveness classes revealed a significant difference between the low and high grade

tumours. The maximum Cho/Cr ratio also correlated significantly with the aggressiveness

191


9

classes (p<0.01, r=0.31)(Fig 3). The median Cho/Cr of the low grade tumours was significantly

different from the high grade tumours. Table 3 shows the AUCs representing the performance of

the aggressiveness assessments.

The correlation between the maximum Cho+Cr/Cit ratio and the maximum Cho/Cr ratio was

0.51 (p<0.001). For each tumour the maximum Cho+Cr/Cit was plotted against the maximum

Cho/Cr value (Fig 4). A Cho/Cr adaptation level of 2.3 for the standardized threshold approach

gave the highest AUCs when discriminating high from low and intermediate grade tumours

(AUC=0.73) and low from the combined high and intermediate grade tumours (AUC=0.78). The

malignancy ratings using an adaptation level of 2.3 are shown in Figure 5 and the median

malignancy rating of the high (median=5) and low (median=3) grade tumours was significantly

different. In all discriminations between aggressiveness classes (Table 3), the AUCs of the

standardized threshold method with adaptation level of 2.3 were higher than the AUCs for the

maximum Cho+Cr/Cit or Cho/Cr ratio alone, but these differences were not significant.

Low vs. high and intermediate grade tumours

High vs. low and intermediate grade tumours

Maximum Cho+Cr/Cit 0.70 0.69

Maximum Cho/Cr 0.74 0.71

Standardized threshold approach*

0.78 0.73

Table 3. The areas under the ROC curves for the discrimination between different

aggressiveness classes using the maximum Cho+Cr/Cit and Cho/Cr ratios and the

standardized threshold approach.

* For the standardized threshold approach an optimized Cho/Cr rating adaptation level of

2.3 was used.

192


9

Figure 2. The distribution of the maximum Cho+Cr/Cit ratios per tumour aggressiveness

class. The horizontal bars indicate the median. One high grade tumour had a maximum

Cho+Cr/Cit ratio of 58 which has been plotted at a Cho+Cr/Cit ratio of 3 for displaying

purposes.

Figure 3. Distribution of the maximum Cho/Cr ratio per aggressiveness class. The

horizontal bars indicate the median.

193


9

Figure 4. The maximum Cho+Cr/Cit ratio for each tumour focus plotted against the

maximum Cho/Cr ratio of that focus. The tumours are divided by aggressiveness class.

The Cho+Cr/Cit ratio of one high grade tumour was cut off at 3 for displaying purposes

.

Figure 5. The highest malignancy ratings according to the standardized threshold

approach for each tumour in the aggressiveness classes. A Cho/Cr adaptation level of 2.3

was used.

194


9

DISCUSSION

With an increasing number of reports on overdiagnosis(2, 25), an individualized prostate cancer

aggressiveness assessment is essential. Currently the pre-treatment aggressiveness is

determined with biopsies, but often the Gleason score is underestimated(3, 4). MR(S)I could be a

prognostic non-invasive approach to differentiate between patients as suitable candidates for

active surveillance and patients that need immediate treatment. Moreover, it would enable

targeted therapy, like intensity-modulated radiotherapy or high-intensity focused ultrasound, to

be applied to the most aggressive part of the tumour. In a detection setting, it could guide a

biopsy to be taken from the most aggressive part of a tumour. Validation is the first step in the

development of such a prognostic MRSI approach. We assessed the tumour aggressiveness

differentiation by the AUC of the ROC curves and this gave similar results when using either the

Cho+Cr/Cit (0.70) or the Cho/Cr (0.74) ratio. The performance of combining both ratios was

better (0.78), though this was not statistically significant, due to a relative small sample size. The

AUCs reflect considerable overlap between the ratios in the different aggressiveness classes,

similar to what has been reported in other studies to the relationship between the Cho+Cr/Cit

ratio and Gleason score(10, 11).

We investigated whether a combination of both ratios would increase the performance of

separating low from higher grade tumours. The original standardized threshold approach was

developed to improve tumour localization at 1.5T with a Cho/Cr level of 2 to adjust the initial

malignancy rating(14, 15). As the timing of the MRSI pulse sequence at 3T is different(17) from

the timing at 1.5T, we optimized the Cho/Cr adaptation level of the standardized threshold

approach for tumour aggressiveness classification at 3T, retaining the original concept of

defining the thresholds based on the mean and standard deviations of non-cancer PZ and CG

tissues. When all tumours with a malignancy rating of 4 or 5 would be classified as high grade,

only 10% of the high grade tumours would be misclassified as a lower grade. 52% of the

intermediate and low grade tumours would be misclassified as high grade, which is better than

previous results (68%) using MRI and 1H-MRSI in patients with an elevated PSA level(9).

As in current clinical practice MR examinations of the prostate are often multiparametric, the

standardized threshold approach could even be extended to include more predictive variables

for the assessment of aggressiveness. Features like the apparent diffusion coefficient or

195


9

pharmacokinetic parameters of dynamic-contrast-enhanced MRI could be included to improve

the performance(26).

The tumours were divided in three clinically useful classes of aggressiveness: low, intermediate

and high. The nature of the Gleason score does not lend itself to a linear relationship with an

aggressiveness marker. Using three aggressiveness classes, the tertiary Gleason scores could be

taken into account and Gleason 7 tumours could be divided in primary grade 3 and primary

grade 4 tumours. The primary grade 4 tumours are more aggressive and have an increased risk

for progression(27, 28). Tumours with a tertiary Gleason score of 5 have a higher risk of

extraprostatic extension(29) and PSA recurrence(28, 30) than tumours with the same score

without a tertiary pattern. For these reasons, three classes of aggressiveness were used, which

still enabled us to answer clinically relevant questions on aggressiveness.

Our study had several limitations. This was a retrospective single-institution study with a

limited number of patients; therefore, our results may not be extrapolated to the general patient

population. Therefore, the data can be considered as very promising but preliminary, and our

conclusions need confirmation by a prospective multicentre trial, which is currently underway.

We did not correct for possible correlation between multiple tumours in a single patient (6 cases

of two clinically significant tumours within one prostate). Although we distinguished between

tumours in the PZ and those in the CG, we did not do a separate aggressiveness analysis for the

two regions due to the limited number of CG tumours(n=11). We excluded 14 patients from our

initial patient cohort because of tumour size on histopathology of less than 0.5cc. Next to the fact

that these small tumours are clinically insignificant (no high grade cancer present), the actual

volume of the voxels of our MRSI examinations was too large to represent tumour foci of these

sizes without too much surrounding non-cancer tissue within the voxel of interest. Metabolite

ratios of these voxels would represent a mixture between cancer and non-cancer tissue.

CONCLUSION

In conclusion, this study showed that 1H-MRSI offers potential for non-invasive

assessment of prostate cancer aggressiveness, which has important implications. We have

modified the existing standardized threshold approach to assess tumour aggressiveness at 3T.

196


9

The initial results of combining the Cho+Cr/Cit and Cho/Cr ratio were promising for the

discrimination between different aggressiveness classes.

Figure 1. MRI and MRSI of a 65-year old patient with prostate cancer (PSA 5.3 ng/ml,

Gleason Score 4+5) (A) T2 weighted image of the prostate with tumour in the left

peripheral zone. Histopathology (B) is used as the gold standard. The spectroscopy grid

is displayed on top of the T2 weighted image in (C). In the right and left peripheral zone a

non-cancer voxel and cancer voxel, respectively, are indicated. (D) shows a spectrum of

non cancer tissue, while in (F) the spectrum of a cancer voxel is shown with deviating

signal intensities for the different metabolites. The calculated choline and creatine to

citrate ratio (Cho+Cr/Cit) and the choline to creatine ratio (Cho/Cr) are indicated in the

spectra of both tissues. The metabolite map in (E) shows the Cho+Cr/Cit distribution over

the prostate.

197


9

REFERENCES

1. Ferlay J, Parkin DM, Steliarova-Foucher E. Estimates of cancer incidence and mortality in Europe in 2008. Eur J Cancer 2010; 46:765-781.

2. Schroder FH, Hugosson J, Roobol MJ, et al. Screening and prostate-cancer mortality in a randomized European study. N Engl J Med 2009; 360:1320-1328.

3. Tomioka S, Nakatsu H, Suzuki N, Murakami S, Matsuzaki O, Shimazaki J. Comparison of Gleason grade and score between preoperative biopsy and prostatectomy specimens in prostate cancer. Int J Urol 2006; 13:555-559.

4. Rajinikanth A, Manoharan M, Soloway CT, Civantos FJ, Soloway MS. Trends in Gleason Score: Concordance Between Biopsy and Prostatectomy over 15 Years. Urology 2008; 72:177-182.

5. Scheenen TW, Klomp DW, Roll SA, Futterer JJ, Barentsz JO, Heerschap A. Fast acquisition-weighted three-dimensional proton MR spectroscopic imaging of the human prostate. Magn Reson Med 2004; 52:80-88.

6. Scheenen TW, Heijmink SW, Roell SA, et al. Three-dimensional proton MR spectroscopy of human prostate at 3 T without endorectal coil: feasibility. Radiology 2007; 245:507-516.

7. Kurhanewicz J, Vigneron DB, Hricak H, Narayan P, Carroll P, Nelson SJ. Three-dimensional H-1 MR spectroscopic imaging of the in situ human prostate with high (0.24-0.7-cm3) spatial resolution. Radiology 1996; 198:795-805.

8. Heerschap A, Jager GJ, Van Der Graaf M, et al. In vivo proton MR spectroscopy reveals altered metabolite content in malignant prostate tissue. Anticancer Res 1997; 17:1455-1460.

9. Villeirs GM, De Meerleer GO, De Visschere PJ, Fonteyne VH, Verbaeys AC, Oosterlinck W. Combined magnetic resonance imaging and spectroscopy in the assessment of high grade prostate carcinoma in patients with elevated PSA: A single-institution experience of 356 patients. Eur J Radiol 2009; Epub ahead of print.

10. Kurhanewicz J, Vigneron DB, Males RG, Swanson MG, Yu KK, Hricak H. The prostate: MR imaging and spectroscopy: Present and future. Radiol Clin North Am 2000; 38:115-138.

11. Zakian KL, Sircar K, Hricak H, et al. Correlation of proton MR spectroscopic imaging with gleason score based on step-section pathologic analysis after radical prostatectomy. Radiology 2005; 234:804-814.

12. Jambor I, Borra R, Kemppainen J, et al. Functional imaging of localized prostate cancer aggressiveness using 11C-acetate PET/CT and 1H-MR spectroscopy. J Nucl Med 2010; 51:1676-1683.

13. Ackerstaff E, Pflug BR, Nelson JB, Bhujwalla ZM. Detection of increased choline compounds with proton nuclear magnetic resonance spectroscopy subsequent to malignant transformation of human prostatic epithelial cells. Cancer Res 2001; 61:3599-3603.

14. Jung JA, Coakley FV, Vigneron DB, et al. Prostate depiction at endorectal MR spectroscopic imaging: investigation of a standardized evaluation system. Radiology 2004; 233:701-708.

15. Fütterer JJ, Scheenen TWJ, Heijmink SWTPJ, et al. Standardized threshold approach using three-dimensional proton magnetic resonance spectroscopic imaging in prostate cancer localization of the entire prostate. Invest Radiol 2007; 42:116-122.

16. Van Asten JJA, Cuijpers V, Hulsbergen-Van De Kaa C, et al. High resolution magic angle spinning NMR spectroscopy for metabolic assessment of cancer presence and Gleason score in human prostate needle biopsies. MAGMA 2008; 21:435-442.

17. Scheenen TWJ, Gambarota G, Weiland E, et al. Optimal timing for in vivo 1H-MR spectroscopic imaging of the human prostate at 3T. Magn Reson Med 2005; 53:1268-1274.

18. Pohmann R, von Kienlin M. Accurate phosphorus metabolite images of the human heart by 3D acquisition-weighted CSI. Magn Reson Med 2001; 45:817-826.

19. Van Oort IM, Bruins HM, Kiemeney LALM, Knipscheer BC, Witjes JA, Hulsbergen-Van De Kaa CA. The length of positive surgical margins correlates with biochemical recurrence after radical prostatectomy. Histopathology 2010; 56:464-471.

20. Epstein JI, Allsbrook Jr WC, Amin MB, et al. The 2005 International Society of Urological Pathology (ISUP) consensus conference on Gleason grading of prostatic carcinoma. Am J Surg Pathol 2005; 29:1228-1242.

21. Greene FL, Page DL, Flemming ID, et al. AJCC Cancer Staging Manual. 6th edition New York: Springer-Verlag, 2002.

198


9

22. Augustin H, Hammerer PG, Graefen M, et al. Insignificant prostate cancer in radical prostatectomy specimen: time trends and preoperative prediction. Eur Urol 2003; 43:455-460.

23. Stamey TA, Freiha FS, McNeal JE, Redwine EA, Whittemore AS, Schmid HP. Localized prostate cancer. Relationship of tumor volume to clinical significance for treatment of prostate cancer. Cancer 1993; 71:933-938.

24. Metz CE, Herman BA, Roe CA. Statistical comparison of two ROC-curve estimates obtained from partially-paired datasets. Med Decis Making 1998; 18:110-121.

25. Andriole GL, Crawford ED, Grubb RL, 3rd, et al. Mortality results from a randomized prostate-cancer screening trial. N Engl J Med 2009; 360:1310-1319.

26. Seitz M, Shukla-Dave A, Bjartell A, et al. Functional Magnetic Resonance Imaging in Prostate Cancer. Eur Urol 2009; 55:801-814.

27. Chan TY, Partin AW, Walsh PC, Epstein JI. Prognostic significance of Gleason score 3+4 versus Gleason score 4+3 tumor at radical prostatectomy. Urology 2000; 56:823-827.

28. Hattab EM, Koch MO, Eble JN, Lin H, Cheng L. Tertiary Gleason Pattern 5 is a Powerful Predictor of Biochemical Relapse in Patients With Gleason Score 7 Prostatic Adenocarcinoma. J Urol 2006; 175:1695-1699.

29. Mosse CA, Magi-Galluzzi C, Tsuzuki T, Epstein JI. The Prognostic Significance of Tertiary Gleason Pattern 5 in Radical Prostatectomy Specimens. Am J Surg Pathol 2004; 28:394-398.

30. Pan CC, Potter SR, Partin AW, Epstein JI. The prognostic significance of tertiary Gleason patterns of higher grade in radical prostatectomy specimens: A proposal to modify the Gleason grading system. Am J Surg Pathol 2000; 24:563-569.

199


9

Prospective Assessment of Prostate Cancer Aggressiveness

using 3T DWI-MRI Guided Biopsies vs. a 10-Core TRUS Biopsy Cohort

T. Hambrock; C. Hoeks, C.Hulsbergen-van de Kaa et al.

CHAPTER 10CHAPTER 10

Prospective Assessment of Prostate Cancer Aggressiveness using 3T Diffusion Weighted Magnetic Resonance Imaging Guided Biopsies versus a 10-Core Transrectal Ultrasound Prostate Biopsy Cohort

Hambrock T, Hoeks C, Scheenen T, Hulsbergen-van de Kaa C, Bouwense S, Schröder F, Fütterer J, Huisman J, Barentsz J

Lauterbur Award Society of Body CT and MR, San Diego, Mar 2010

Advances in knowledge: 10-Core Transrectal ultrasound (TRUS) guided systematic prostate biopsy determined

highest Gleason grades reveal a poor concordance with true highest Gleason grades

(HGG) in prostatectomy specimens.

TRUS biopsy show a substantial undergrading for tumours with a HGG of 5.

3T Diffusion weighted MR imaging is an accurate and valuable technique for identifying

the most aggressive components within a prostate tumour.

MR guided biopsies targeted towards the most abnormal regions on DWI has a vastly

superior accuracy for determining the true HGG compared to TRUS.

Implications for patient care: Biopsies targeted towards the most abnormal regions on 3T DWI MR imaging represent

a substantially improved method for assessment of true tumour aggressiveness and can

therefore represent an indispensable tool in the diagnosis and management of patients

with prostate cancer.

Summary Statement

MR guided biopsies targeted towards the most abnormal regions on DWI MRI represent a substantially improved prospective method for

assessment of true tumour aggressiveness

202

Prospective Assessment of PCa Aggressiveness using 3T DWI-MRI Guided Biopsies vs. 10-core TRUS Biopsy Cohort

10

ABSTRACT

Background: Accurate pretreatment assessment of prostate cancer aggressiveness is becoming

more important in decision making. Gleason grade is one of the most important predictors of

prostate cancer aggressiveness. Transrectal ultrasound guided biopsies (TRUS-GB) show

substantial undergrading of Gleason grade found in prostatectomy specimens. Diffusion

weighted MR-imaging (DWI) has been shown to be a valuable biomarker of tumor

aggressiveness.

Objective: To improve pretreatment assessment of prostate cancer aggressiveness, this study

prospectively evaluated the value of MRI guided prostate biopsies (MR-GB) of abnormalities

determined on DWI apparent diffusion coefficient (ADC) maps. Results were compared to those

of a clinical cohort using 10-core TRUS-GB. Prostatectomy findings were the gold standard.

Measurements: A multi-parametric 3T MRI incl. DWI was performed to identify tumour

suspicious regions in patients (n=34) with a negative TRUS-GB. Subsequently, the regions with

highest restriction on ADC maps within the suspicions regions, were used to direct MR-GB. A 10-

core TRUS-GB was used in a matched cohort of 64 men. Following prostatectomy, the highest

Gleason grades (HGG) in biopsies and prostatectomy (RP) specimens were identified. Biopsy and

prostatectomy Gleason grade performances were evaluated using Chi-square analysis.

Results and Limitations : No significant differences were observed for the proportions of

patients on RP having a HGG=3 (35% vs. 28%; p=0.50), HGG=4 (32% vs. 41%; p=0.51) and

HGG=5 (32% vs. 31%; p=0.61) for the MR-GB and TRUS-GB cohort respectively. MR-GB showed

an exact performance with RP for overall HGG in 88% (30/34) while for the TRUS-GB this was

55% (35/64; p=0.001). In the MR-GB cohort, an exact performance with a HGG=3 was 100%

(12/12), for HGG=4, 91% (10/11) and for HGG=5, 73% (8/11). The corresponding performance

rates for TRUS-GB were 94% (17/18; p=0.41), 46% (12/26;p=0.02) and 30% (6/20; p=0.01)

respectively.

Conclusions: This study prospectively shows the ability of DWI directed MR-GBs to improve

pretreatment risk-stratification by obtaining biopsies, which are representative for true

prostatectomy Gleason grade.

203


10

INTRODUCTION

The Gleason grading system is one of the most important features describing the pathological

characteristics of prostate cancer. Of all clinically determinable parameters, the Gleason score

(GS) has been proven to be one of the most important in measuring aggressiveness, disease

outcome and risk of mortality from prostatic cancer(1).

Currently, transrectal ultrasound guided prostate biopsy (TRUS-GB) is the most accepted

method for establishing a definite diagnosis of prostate cancer in patients with a clinical

suspicion based on prostate specific antigen (PSA) values or digital rectal examination (DRE).

The most widely used biopsy schemes include sampling by 10-12 cores with emphasis on the

lateral peripheral zone and transition zone (2;3). The tumor containing tissues cores obtained

at TRUS-GB are scored according to the Gleason grading scheme in order to determine prostate

cancer aggressiveness and prognosis .

Prostate cancer can be multifocal in location and also heterogeneous in composition, often

presenting with well, moderately- and poorly-differentiated components in the same tumor.

TRUS-GB determined GS has been shown (4-6) to be substantially discordant (under-grading in

34-38%) with the GS determined in analysis of radical prostatectomy (RP) specimens. Because

risk-stratification will affect individualized treatment decisions and prognosis, an accurate

pretreatment prediction of GS, which is the major component of the currently used risk

nomograms, remains essential.

Multi-parametric MR imaging, including T2-weighted imaging (T2-w MRI), diffusion-weighted

imaging (DWI) and dynamic contrast enhanced MR imaging (DCE-MRI) have all been shown

(especially in combination) to accurately localize prostate cancer (7;8). Due to improved

localization, suspicious regions on multi-parametric MRI have also been biopsy targeted under

direct MR guidance and shown to substantially increase the tumour detection rates (9;10).

Especially DWI has recently been shown to provide information about tumor aggressiveness

(11;12).

The aim of this study, therefore, was to prospectively determine whether 3T DWI guided

prostate tissue sampling could improve the pretreatment assessment of prostate cancer

aggressiveness. These results were compared to a standard clinical cohort of patients, where

204


10

tissue sampling was done using a systematic 10-core sampling approach for the same purpose.

In both cohorts the performance of Gleason grades in biopsy and prostatectomy specimens,

serving as a gold standard, was determined.


Patients

Between Aug 2006 and Apr 2009, 123 consecutive patients underwent a RP at the Radboud

University Nijmegen Medical Centre, Nijmegen, Netherlands and were retrospectively included if

a diagnosis of prostate cancer was made either with a 10-core systematic TRUS-GB scheme or

MR-GB. Patients with hormonal or radiotherapy prior to prostatectomy were excluded.

TRUS-GB sampling

Extended systematic 10-core TRUS-GB's (incl. 6 lateral and 4 transition zone biopsies) were

obtained using a Pro Focus B and K ultrasound device (Medical, Herlen, Denmark) with an

Endfire probe transducer (8667 convex array, 8.0 MHz, B and K Medical, Herlen, Denmark) and

an 18 Gauge needle with 17 mm sampling length. The indication for performing biopsies was

based on routine clinical parameters i.e. an elevated PSA above 4 ng/ml and/or abnormal DRE,

requiring further assessment. TRUS-GB represented the first biopsy session in these patients.

MR imaging

Multi-parametric MR imaging at 3T (Trio Tim, Siemens, Erlangen, Germany) which included

DWI, T2-w MRI, and DCE-MRI was performed in patients with at least one prior negative 10-core

TRUS-GB session, however with ongoing clinical suspicion for prostate cancer. This suspicion

was defined by a rising or persistently elevated PSA. MR imaging parameters are presented in

Table 1. Apparent diffusion coefficient (ADC) maps were automatically calculated from the DWI

by the scanner software. Two radiologists determined up to 3 tumour suspicious regions (TSR)

per patient in consensus using the combined information of the features suspicious for

malignancy on the different modalities of the multi-parametric MRI. The patient PSA values were

available to radiologists for all evaluated patients. Each of the three imaging modalities was

205


10

scored on a tumour probability scale of 1 to 5 with a maximum cumulative score of 15. Per

modality, the scale is defined in as such that 1 indicates: definitely no tumour; 2: probably no

tumour; 3: possibly tumour; 4: probably tumour and 5: definitely tumour. A biopsy indication

8/15. Subsequently, for each TSR

that was biopsied, the region on the ADC for that particular TSR, revealing the darkest spot, was

used as target for the MR-GB.

Sequence

Type Slice

thickness Number of slices

In-plane resolution

TR TE Averages b-values

T2-w Axial

TSE 4 mm 15-19 0.6 x 0.6 mm 3540 ms 104 ms 2 -

T2-w Coronal

TSE 4 mm 15-19 0.6 x 0.6 mm 3350 ms 105 ms 2 -

T2-w Sagital

TSE 4 mm 15-19 0.6 x 0.6 mm 3810 ms 105 ms 2 -

DWI SE-EPI 4 mm 15-19 2.0 x 2.0 mm 2800 ms 81 ms 10 0, 50, 500, 800 mm2/s

T1-w DCE

GRE FLASH

4 mm 14 1.8 x 1.8 mm 37 ms 1.47 ms 1 -

Table 1. MR imaging sequence parameters. T2-w: T2-weighted; T1-w: T1-weighted; DWI:

Diffusion weighted imaging; DCE: Dynamic contrast enhanced imaging; TSE: Turbo Spin

Echo; SE-EPI: Spin Echo – Echo Planar Imaging; TR: Repetition Time; TE: Echo Time;

GRAPPA: Parallel imaging factor; GRE: Gradient echo imaging; FLASH: Fast low angle shot

imaging.

206


10

MRI guided biopsy (MR-GB)

On average 4 weeks (range 2 – 6 weeks) following the tumour detection multi-parametric MRI

study, MR-GB (using MR compatible 18 gauge needles with sampling length of 17 mm) of the

previously determined TSRs was performed using a commercially available transrectal MR-

biopsy device (Invivo, Schwerin, Germany) under direct 3T MR guidance. The translation of

initial MR imaging findings to the subsequent MR-GB has been described in detail before (13).

The lowest signal areas on the ADC maps within the TSR were used to target the biopsy cores.

Histopathological analysis of biopsy specimens

Immediately after biopsy, tissue cores were fixed in 10% neutral-buffered formalin. After

histological staining with hematoxylin and eosin (H&E), tissue section of 5 m were prepared

and thereafter evaluated by one urogenital pathologist (C.A.H.K) with 17 years experience in

prostate pathology. For all included patients, clinical features incl. PSA were available to the

histopathologist. For cores containing cancer, a Gleason score using the International Society of

Urogential Pathology (ISUP) criteria of 2005 was assessed. The primary, secondary and tertiary

Gleason grades were determined and the highest Gleason grade (HGG) was identified within the

biopsy cores.

Reconstructed whole-mount step-section preparation


submitted for histopathological investigation. After inking of the surface, the prostate specimens

were cut into 4-mm thick slices, perpendicular to the dorsal-rectal surface in a plane parallel to

the transverse T2-weighted imaging plane, and macroscopically photographed with a CCD-

camera. All slices were completely processed and evaluated on 5 μm hematoxylin-eosins stained

sections. The presence and extent of cancer were outlined on the glass slide cover by the same

expert urological pathologist who reviewed all prior biopsies. Subsequently tumours were

mapped on the macroscopic photographs to allow reconstruction of tumour extent and

multifocality. Each individual tumour was graded according to the 2005 ISUP Modified Gleason

Grading System (14). As with the assessment of biopsies, the primary to tertiary Gleason grades

and the HGG identified within the prostate was noted.

207


10


Cross tabulation analysis of the biopsy and RP findings was done. For both MR-GB and TRUS-GB

cohorts, performance rates (%) with RP were determined for the HGG. Secondly, for a RP HGG of

5, undergrading was further defined as if the Bx HGG was 3. Thirdly, performance

rates between Bx and the RP HGG groups were determined separately for respectively patients

with a PSA 10 ng/ml and those with a PSA >10ng/ml. Chi-square analyses with Fisher’s exact tests were performed to evaluate the significance of differences between MR-GB and TRUS-GB

performance rates. The t-test was performed to determine for differences in mean PSA, prostate

volume and dominant tumour volume. A significant difference was considered present when p

<0.05. Statistical analyses were performed with SPSS software (SPSS, version 16.0.01, Chicago,

U.S.A).

RESULTS

Ninety-eight patients fulfilled the inclusion criteria. In 34/98 patients a tumour diagnosis was

made using MR-GB (median of 3 cores, range 1-5; median number of biopsies per TSR: 2, range 1

to 3) and in 64/98 patients a diagnosis using 10-core TRUS-GB. The median procedure duration

for MR-GB was 29 min (range 15-75 min). The median duration between MR-GB and RP was 6

weeks (range 3- 11 weeks) and between TRUS-GB and RP, 5 weeks (range 2-9 weeks). The

patient demographic and clinical parameters are summarized in Table 2.

No significant differences between the MR-GB and TRUS-GB cohorts were observed for

percentage pT3 tumours (35% vs. 38%; p=0.83), mean dominant aggressive tumour volume

(4.85 cc vs. 4.52 cc; p=0.69) or mean prostate volume (41 cc vs. 36 cc; p=0.61). Furthermore, no

significant differences were observed for the overall proportions of patients on RP having a

HGG=3 (35% vs. 28%; p=0.50), HGG=4 (32% vs. 41%; p=0.51) and HGG=5 (32% vs. 31%;

p=1.00) for the MR-GB and TRUS-GB cohort respectively. In our two cohorts, the RP presence of

HGG 4 was associated with extra-capsular extension in 39-46% and the presence of a HGG of 5 in

64-70%.

208


10

MR-GB 10-core Significance (p-value)

Number of Patients 34 64 N.A.

Age 66 (51-74) 66 (41-74) 0.22

Number of biopsies (range)

3 (1-5) 10 N.A.

Stage

- pT2

- pT3

22/34 (65%)

12/34 (35%)

40/64 (62%)

24/64 (38%)

0.83

Prostate Volume

- Median [cc] (range)

PSA

- Median [ng/ml] (range)

D.A. Tumour Volume

- Median [cc] (range)

41 (12-79)

12 (3 – 40)

4.85 (0.1-33)

36 cc (17-126)

8 (2– 47)

4.52 (0.1-33.5)

0.61

0.02 *

0.69

Prevalence of Tumours in RP

HGG category

- HGG 3

- HGG 4

- HGG 5

35% (12 of 34)

0% (0/12)

32% (11 of 34)

45% (5/11)

32% (11 of 34)

64% (7/11)

28% (18 of 64)

0% (0/18)

41% (26 of 64)

38% (10/26)

31% (20 of 64)

80% (16/20)

0.50

N.A.

0.51

0.73

1.00

0.41

Tabel 2. Patient and pathology characteristics. (D.A. = Dominant aggressive; *=denotes

significance; N.A.=not applicable)

209


10

A summary of the Bx and RP findings are presented in Tables 3 and 4. When patients were

categorized based on the HGG on Bx and RP, the overall performance rate for MR-GB was 88%

(30/34) vs. 55% (35/64) for TRUS-GB patients. In the MR-GB cohort, an exact performance with

a RP HGG=3 was 100% (12/12), for HGG=4 this was 91% (10/11) and for HGG=5, 73% (8/11).

The corresponding performance rates for TRUS-GB were 94% (17/18; p=0.41), 46%

(12/26;p=0.01) and 30% (6/20; p=0.02) respectively. For biopsies determined as low-grade

(HGG=3), the positive predictive value (PPV) for MR-GB to represent true low-grade tumour was

92% (12/13) while the PPV for TRUS-GB was 45% (17/38; p=0.001). Overall, undergrading of

tumors with a RP HGG 4 or 5 was 46% (25/46) for the TRUS-GB cohort and 5% (1/22) for the

MR-GB cohort.

Prostatectomy

HGG 3 HGG 4

HGG 5

HGG 3 17 14 8 44% (17/39) TRUS-GB HGG 4 1 12 6 63% (12/19) HGG 5 0 0 6 100% (6/6)

94% (17/18) 46% (12/26)

73%(8/11)

55% (35/64)

Prostatectomy

HGG 3

HGG 4

HGG 5

HGG 3 12 1 0 92% (12/13) MR-GB HGG 4 0 10 3 77% (10/13) HGG 5 0 0 8 100% (8/8)

100%(12/12) 91% (10/11)

73% 8/11)

88% (30/34)

Table 3. Crosstabs for cohorts based on HGG grouping

210


10

Performance rates

MR-GB

Performance rates

10-core TRUS-GB

Significance (p-value)

- Overall Bx concord. with RP HGG

- Bx concord. with RP HGG 3



- Bx concord. with RP HGG 4/5

- PPV for Bx and RP HGG=3

88% (30/34)

100% (1212)

91% (10/11)

73% (8/11)

95% (21/22)

92% (12/13)

55% (35 of 64)

94% (17 of 18)

46% (12 of 26)

30% (6 of 20)

54% (25 of 46)

45% (17/38)

0 .001*

0.41

0.01 *

0.02 *

0.001 *

0.003 *

PSA

-

~ Overall HGG performance

- > 10 ng/ml

~ Overall HGG performance

12/34 (35%)

100% (12/12)

22/34 (65%)

82% (18/22)

44/64 (69%)

59% (26/44)

20/64 (31%)

45% (9/20)

0.01 *

0.01 *

0.01 *

0.01 *

Table 4. Performance analysis between Bx and RP cohorts. (PPV=Positive predictive

value; *=denotes significance; N.A.= Not applicable)

With MR-GB there was no over-grading while this was the case in one patient (false HGG=4

instead of 3) in the TRUS-GB cohort. The under-grading for a RP HGG=5 was 27% (3/11) for MR-

GB compared to 70% (14/20) for TRUS-GB. Furthermore, TRUS-GB showed a substantial under-

grading in 57% (8/14) of the under-graded HGG=5 cases, whereas no substantial under-grading

occurred with MR-GB.

As the MR-GB (median PSA of 12 ng/ml) and TRUS-GB groups (median PSA of 8 ng/ml) showed

a significant difference in pre-biopsy PSA levels (p=0.02), a sub-group analysis was performed

showed a 0% (0/12) undergrading

for MR-GB in patients 10 ng/ml. In contrast, in this PSA sub-group, with TRUS-GB,

under-grading occurred in 41% (18/44; p=0.01). For patients with PSA levels >10 ng/ml, MR-GB

revealed an under-grading of 18% (4/22) and TRUS-GB revealed an under-grading of 55%

(11/20; p=0.01).

211


10

Figure 1. Performance rates according to HGG categorization. 10-Core TRUS-GB vs. MR-

GB

DISCUSSION

In this prospective study, 3T Diffusion Weighted MR imaging (DWI) guided prostate tissue

sampling improved the pretreatment assessment of prostate cancer aggressiveness. The Gleason

grades as determined with DWI showed a high performance rate of 88% with prostatectomy.

This is in sharp contrast with a clinical routine 10-core TRUS biopsy protocol, which showed a

performance rate of only 55%, which is in agreement with rates reported in literature (15-17).

In this study the most abnormal regions on ADC, following a multi-parametric localization

approach of the tumour, were used to target the biopsies. To our knowledge, this is the first

prospective report on the use of DWI in obtaining prostate cancer tissue samples, which are

more representative for true prostatectomy specimen Gleason grade. These results confirm

prior retrospective studies on the ability of MR imaging to visualize tumour aggressivity and

provide a method to improve pretreatment prediction of true Gleason grades(18;19).

212


10

The importance of establishing a correct pretreatment assessment of prostate cancer

aggressiveness has generally been accepted. In recent years, a shift from radical therapy to a

more patient tailored focal therapy has been advocated (20). One of the cornerstones of

pretreatment risk stratification in order to tailor treatment on a patient level is the correct

prediction and assessment of the true Gleason grades within the tumour. Patients without

Gleason grade 4/5 components, are potential candidates for less invasive treatment options,

such as active surveillance or local therapy incl. brachytherapy or high-intensity focused

ultrasound (21). Patients with the presence of high-grade components are in definite need for

additional evaluation for the presence of extra capsular extension or metastasis to lymph nodes

or bone. In addition, high-grade PCa managed with non-curative intent, substantially reduces life

expectancy (22). An EORTC trial showed that high-risk stratified patients have a definite benefit

from adjuvant hormone therapy. Therefore, correctly stratifying patients into low/high risk is of

utmost importance (23).

Numerous studies have addressed the correlation between Gleason scores in needle biopsy and

corresponding radical prostatectomy specimens. These show, that increasing the number of

biopsies, increases the performance. For earlier studies, using sextant biopsies, under-grading

was reported in 44-60% of cases (24;25) while more recent studies with extended biopsy

schemes reported lower values of 32-38% (4;5;16;24). Most studies have shown that over-

grading (8-10%)(15;17) by biopsy cores is of less importance than under-grading. When

comparing the overall performance rates between studies, the most important factor that needs

careful consideration and interpretation is the prevalence of low-grade tumours. Using an

extended scheme with a median of 12-cores, San Fransisco et al. (26) showed an exact GS

performance rate of 76%. However, the prevalence of low-grade tumours in their RP was 72%.

This artificially increases the overall performance rates. When only evaluating high-grade

tumours (HGG of 4/5), an under-grading of 32% was still evident in their series. Data from a

large cohort from John Hopkins Hospital (27) revealed an overall GS agreement of 76%. Also

here, the prevalence of low-grade tumours in RP was high at 67%. When only the high-grade

tumours on RP were chosen, an under-grading of 42% was noted.

Our 10-core TRUS-GB revealed a 46% under-grading of tumours indentified as HGG=4/5 on RP.

This is in agreement with these two prior studies. Yet, for MR-GB, only 5% under-grading of

213


10

high-grade tumours was seen. In addition, TRUS-GB revealed a substantial under-grading in

57% of cases of the HGG 5 tumor, i.e. showing a HGG of 3. In all cases of HGG 5 under-grading,

biopsies performed by MR targeting, revealed a HGG of 4, thus showing a more acceptable

underestimation. The prevalence of HGG 3, 4 and 5 groups in our two cohorts did not show

statistically significant differences. We therefore feel that our results with MR-GB show a

substantial improvement of performance rates compared to current practice and literature. In

addition, with MR-GB, only a median of 3 biopsy cores per patient were taken, instead of 10 with

TRUS-GB.

A number of clinical important factors exist which may be associated with prostate biopsy

undergrading. Isariyawongse et al.(28) have shown that both age and PSA values are important

in this respect. Biopsies in patients with PSA values 10-20 ng/ml and PSA > 20 ng/ml had odds

ratios of 2.11 and 3.64 respectively compared to PSA < 10 ng/ml for representing undergrading

of true Gleason scores in prostatectomy. Our overall baseline PSA values for the two cohorts did

indeed show a significant difference, however, to the detriment of the MR-GB where slightly

higher PSA values were found. Usually a PSA cut-off value of 10 ng/ml is used as an integral part

in decision-making regarding further diagnostic tests or type of treatment, i.e. opting for active

surveillance(29). We therefore, performed

and those >10 ng/ml. For both subgroups, MR-

compared to TRUS-GB. Evidently, the PSA value did not influence the performance rates of

biopsies with RP findings in our study. Stackhouse et al.(30) evaluated additional factors that

may predict undergrading in biopsies. Of relevance to our study would also be their identified

factors: patient age and prostate weight (and thus prostate volume). Increasing age has been

shown to have increasing odds ratios for undergrading. In our cohort both groups had the same

median ages of 66 years (p=0.22). No significant differences in prostate volumes (p=0.61) or

dominant tumour volume (p=0.69) were seen in our cohorts. Furthermore, no difference in the

prevalence of stage pT3 disease was noticed (p=0.83) between the two cohorts.

214


10

Figure 2. Patient with PSA of 11 ng/ml. TRUS-GB revealed a Gleason 3+4. a) T-2w image

shows a large tumor region in the entire dorsal peripheral zone (arrows). On b) the ADC

maps, restriction is clearly visible for the same lesion. On DCE imaging, the Ktrans map (c)

shows irregular enhancement of the tumor. However d) within the restricted regions,

two regions with higher restriction are visible (yellow asterisks). On the corresponding

pathology step section, the tumor is delineated in light blue, corresponding to the findings

on MR. Regions with focal Gleason grade 5 are delineated with a dotted line correspond

exactly to the ADC “hot spots” findings. Final pathology showed Gleason 3+4+5, pT3

tumour.

In addition, we have evaluated two further factors that in our opinion may also represent biases

in cohorts possibly having an influence on degree of undergrading: dominant aggressive tumour

volume and tumour stage on at radical prostatectomy. In a paper by Resnick et al.(31), biopsy

and prostatectomy features of patients at first, second and third TRUS-GB session were

215


10

Figure 3. Patient with PSA 12ng/ml. Four times prior negative TRUS-GB. a) T2-w image

with focal lesion visible in right peripheral zone. b) On DCE-Ktrans map, diffuse

enhancement of the peripheral zone is seen. c) ADC derived from DWI shows focal small

lesion with clear restriction (yellow asterisk). d) True-FISP images during biopsy with

the needle guider directed towards the most suspicious region, prior to taking a MR-GB.

MR-GB revealed a Gleason 4 component. e) Prostatectomy step section showed a pT2c

tumor in the right peripheral zone (light blue=Gleason 3 and red=Gleason 4 component).

The volume of the Gleason 3 component is underestimated by the MRI, however volume

of the focal “hot-spot” on the ADC images exactly match with final pathology: Gleason 4.

evaluated. In their large cohort of 2411 patients, with each increase in the number of biopsy

sessions, the undergrading of Gleason score

second to 58% at third biopsy session, despite the increasing overall prevalence of Gleason score

6 tumors with every subsequent session. These findings would actually suggest an increased

likelihood of undergrading for our repeat biopsies. On the contrary however, despite

representing a re-biopsy session, our MR-GB still outperformed a first session 10-core TRUS-GB.

We therefore are of the opinion, that despite these minor differences between our cohorts, no

216


10

important clinical or pathological factor could be determined that might bias our MR-GB cohort

to a more favorable group regarding the likelihood of undergrading.

Diffusion weighted MR imaging is rapidly gaining importance as a valuable non-invasive

biomarker for determining tumour response to therapy in a large variety of tumours (32). In

addition, DWI is also increasingly being used to non-invasively determine tumour

aggressiveness. It’s role for assessment of aggressiveness and cellularity in breast tumours (33),

soft tissue sarcomas (34), renal tumours (35) and hepatocellular tumours (36) has been

reported before. For prostate cancer, recent data has shown that ADC values derived from the

DWI images, have a high discriminatory performance in separating low- vs. combined

intermediate- and high-grade cancers(18).

A practical utilization of our findings on a larger scale warrants some consideration. We have

chosen to obtain biopsies of the abnormal region under direct MR guidance. Despite being more

exact in targeting biopsies, its use is limited by widespread availability and practicality for a

large number of patients. We therefore envision that DWI in future should facilitate MR based

targeted biopsies under TRUS guidance once fusion software has become widely available.

A number of limitations exist. A randomized trial between MR imaging biopsies vs. TRUS biopsy

or performing both TRUS and MR guided biopsies in the same patient would represent the ideal

scenario. Our approach was, however, to determine the performance in a routine clinical setup

as performed in our hospital. A second limitation represents the relative low number of patients.

Nonetheless, differences were statistically significant, even with this low number of patients.

Furthermore, because of the equal predominant proportion of high-grade tumours in both our

cohorts, valid conclusions on the amount of TRUS-GB and MR-GB undergrading can be made,

based even on this relatively low number of patients. Although a multi-parameteric approach is

proven to be the most useful for evaluation of prostate cancer on MR imaging, it still requires a

high level of expertise and is known to suffer from observer variability(37). Our results

therefore represent findings of an expert centre which utilizes in-house developed analytical

software and numerous years of experience. This might therefore be an overoptimistic

prediction of performance attainable in smaller, non-expert institutions. A final limitation, as

thoroughly discussed previously, is the potential differences of the two cohorts. MRI and

subsequent MR guided biopsies were only performed when a first systematic TRUS-GB was

217


10

negative despite clinical suspicion for prostate cancer. However, there were no differences

between the groups, except for PSA value, where a separate sub-group analysis was made.

CONCLUSIONS

Our final conclusions are that biopsies targeted towards the most abnormal regions on 3T DWI

MR imaging represent a substantially improved method for assessment of true tumour

aggressiveness and can therefore represent an indispensable tool in the diagnosis and

management of patients with prostate cancer. This will probably also hold true for other

malignancies. Its utilization is therefore strongly advocated.

218


10

REFERENCES

1. Egevad L, Granfors T, Karlberg L, Bergh A, Stattin P. Prognostic value of the Gleason score in prostate cancer. BJU.Int. 2002 Apr;89(6):538-42.

2. Presti JC, Jr., O'Dowd GJ, Miller MC, Mattu R, Veltri RW. Extended peripheral zone biopsy schemes increase cancer detection rates and minimize variance in prostate specific antigen and age related cancer rates: results of a community multi-practice study. J.Urol. 2003 Jan;169(1):125-9.

3. Siu W, Dunn RL, Shah RB, Wei JT. Use of extended pattern technique for initial prostate biopsy. J.Urol. 2005 Aug;174(2):505-9.

4. Divrik RT, Eroglu A, Sahin A, Zorlu F, Ozen H. Increasing the number of biopsies increases the concordance of Gleason scores of needle biopsies and prostatectomy specimens. Urol.Oncol. 2007 Sep;25(5):376-82.

5. Kvale R, Moller B, Wahlqvist R, Fossa SD, Berner A, Busch C, Kyrdalen AE, Svindland A, Viset T, Halvorsen OJ. Concordance between Gleason scores of needle biopsies and radical prostatectomy specimens: a population-based study. BJU.Int. 2008 Dec 22.

6. Lattouf JB, Saad F. Gleason score on biopsy: is it reliable for predicting the final grade on pathology? BJU.Int. 2002 Nov;90(7):694-8.

7. Delongchamps NB, Rouanne M, Flam T, Beuvon F, Liberatore M, Zerbib M, Cornud F. Multiparametric magnetic resonance imaging for the detection and localization of prostate cancer: combination of T2-weighted, dynamic contrast-enhanced and diffusion-weighted imaging. BJU.Int. 2010 Nov 2.

8. Kitajima K, Kaji Y, Fukabori Y, Yoshida K, Suganuma N, Sugimura K. Prostate cancer detection with 3 T MRI: comparison of diffusion-weighted imaging and dynamic contrast-enhanced MRI in combination with T2-weighted imaging. J.Magn Reson.Imaging 2010 Mar;31(3):625-31.


10. Hambrock T, Somford DM, Hoeks C, Bouwense SA, Huisman H, Yakar D, van O, I, Witjes JA, Futterer JJ, Barentsz JO. Magnetic resonance imaging guided prostate biopsy in men with repeat negative biopsies and increased prostate specific antigen. J.Urol. 2010 Feb;183(2):520-7.



13. Hambrock T, Futterer JJ, Huisman HJ, Hulsbergen-vandeKaa C, van Basten JP, van O, I, Witjes JA, Barentsz JO. Thirty-two-channel coil 3T magnetic resonance-guided biopsies of prostate tumor suspicious regions identified on multimodality 3T magnetic resonance imaging: technique and feasibility. Invest Radiol. 2008 Oct;43(10):686-94.


15. Boorjian SA, Karnes RJ, Crispen PL, Rangel LJ, Bergstralh EJ, Sebo TJ, Blute ML. The impact of discordance between biopsy and pathological Gleason scores on survival after radical prostatectomy. J.Urol. 2009 Jan;181(1):95-104.

16. Coogan CL, Latchamsetty KC, Greenfield J, Corman JM, Lynch B, Porter CR. Increasing the number of biopsy cores improves the concordance of biopsy Gleason score to prostatectomy Gleason score. BJU.Int. 2005 Aug;96(3):324-7.

17. Serkin FB, Soderdahl DW, Cullen J, Chen Y, Hernandez J. Patient risk stratification using Gleason score concordance and upgrading among men with prostate biopsy Gleason score 6 or 7. Urol.Oncol. 2008 Dec 29.

18. Hambrock T, Somford DM, Huisman HJ, van O, I, Witjes JA, Hulsbergen-Van de Kaa CA, Scheenen T, Barentsz JO. Relationship between apparent diffusion coefficients at 3.0-T MR imaging and gleason grade in peripheral zone prostate cancer. Radiology 2011 May;259(2):453-61.

219


10


10


20. Ahmed HU, Pendse D, Illing R, Allen C, van der Meulen JH, Emberton M. Will focal therapy become a standard of care for men with localized prostate cancer? Nat.Clin.Pract.Oncol. 2007 Nov;4(11):632-42.

21. de la RJ, Ahmed H, Barentsz J, Johansen TB, Brausi M, Emberton M, Frauscher F, Greene D, Harisinghani M, Haustermans K, et al. Focal therapy in prostate cancer-report from a consensus panel. J.Endourol. 2010 May;24(5):775-80.

22. Tewari A, Divine G, Chang P, Shemtov MM, Milowsky M, Nanus D, Menon M. Long-term survival in men with high grade prostate cancer: a comparison between conservative treatment, radiation therapy and radical prostatectomy--a propensity scoring approach. J.Urol. 2007 Mar;177(3):911-5.

23. Bolla M, Van TG, Warde P, Dubois JB, Mirimanoff RO, Storme G, Bernier J, Kuten A, Sternberg C, Billiet I, et al. External irradiation with or without long-term androgen suppression for prostate cancer with high metastatic risk: 10-year results of an EORTC randomised study. Lancet Oncol. 2010 Nov;11(11):1066-73.

24. Paulson DF. Impact of radical prostatectomy in the management of clinically localized disease. J.Urol. 1994 Nov;152(5 Pt 2):1826-30.

25. Steinberg DM, Sauvageot J, Piantadosi S, Epstein JI. Correlation of prostate needle biopsy and radical prostatectomy Gleason grade in academic and community settings. Am.J.Surg.Pathol. 1997 May;21(5):566-76.

26. San F, I, Dewolf WC, Rosen S, Upton M, Olumi AF. Extended prostate needle biopsy improves concordance of Gleason grading between prostate needle biopsy and radical prostatectomy. J.Urol. 2003 Jan;169(1):136-40.

27. Fine SW, Epstein JI. A contemporary study correlating prostate needle biopsy and radical prostatectomy Gleason score. J.Urol. 2008 Apr;179(4):1335-8.

28. Isariyawongse BK, Sun L, Banez LL, Robertson C, Polascik TJ, Maloney K, Donatucci C, Albala D, Mouraviev V, Madden JF, et al. Significant discrepancies between diagnostic and pathologic Gleason sums in prostate cancer: the predictive role of age and prostate-specific antigen. Urology 2008 Oct;72(4):882-6.

29. Dall'Era MA, Konety BR, Cowan JE, Shinohara K, Stauf F, Cooperberg MR, Meng MV, Kane CJ, Perez N, Master VA, et al. Active surveillance for the management of prostate cancer in a contemporary cohort. Cancer 2008 Jun 15;112(12):2664-70.

30. Stackhouse DA, Sun L, Schroeck FR, Jayachandran J, Caire AA, Acholo CO, Robertson CN, Albala DM, Polascik TJ, Donatucci CF, et al. Factors predicting prostatic biopsy Gleason sum under grading. J.Urol. 2009 Jul;182(1):118-22.

31. Resnick MJ, Lee DJ, Magerfleisch L, Vanarsdalen KN, Tomaszewski JE, Wein AJ, Malkowicz SB, Guzzo TJ. Repeat prostate biopsy and the incremental risk of clinically insignificant prostate cancer. Urology 2011 Mar;77(3):548-52.

32. Harry VN, Semple SI, Parkin DE, Gilbert FJ. Use of new imaging techniques to predict tumour response to therapy. Lancet Oncol. 2010 Jan;11(1):92-102.

33. Costantini M, Belli P, Rinaldi P, Bufi E, Giardina G, Franceschini G, Petrone G, Bonomo L. Diffusion-weighted imaging in breast cancer: relationship between apparent diffusion coefficient and tumour aggressiveness. Clin.Radiol. 2010 Dec;65(12):1005-12.

34. Schnapauff D, Zeile M, Niederhagen MB, Fleige B, Tunn PU, Hamm B, Dudeck O. Diffusion-weighted echo-planar magnetic resonance imaging for the assessment of tumor cellularity in patients with soft-tissue sarcomas. J.Magn Reson.Imaging 2009 Jun;29(6):1355-9.

35. Squillaci E, Manenti G, Cova M, Di RM, Miano R, Palmieri G, Simonetti G. Correlation of diffusion-weighted MR imaging with cellularity of renal tumours. Anticancer Res. 2004 Nov;24(6):4175-9.

36. Muhi A, Ichikawa T, Motosugi U, Sano K, Matsuda M, Kitamura T, Nakazawa T, Araki T. High-b-value diffusion-weighted MR imaging of hepatocellular lesions: estimation of grade of malignancy of hepatocellular carcinoma. J.Magn Reson.Imaging 2009 Nov;30(5):1005-11.

37. Lim HK, Kim JK, Kim KA, Cho KS. Prostate cancer: apparent diffusion coefficient map with T2-weighted images for detection--a multireader study. Radiology 2009 Jan;250(1):145-51.

220


10

PART FOUR

COMPUTER ASSISTED DIAGNOSIS OF PROSTATE

CANCER

CHAPTER 11

The Effect of Inter-patient Normal Peripheral Zone Apparent Diffusion

Coefficient Variation on the Prediction of Prostate Cancer

Aggressiveness

Geert Litjens; Thomas Hambrock; Jelle Barentsz et al.


The effect of inter-patient normal peripheral zone Apparent Diffusion Coefficient variation on the Prediction of Prostate Cancer Aggressiveness


Litjens G, Hambrock T, Barentsz J, Huisman J

Advances in knowledge: A large inter-patient variability exists for normal peripheral zone apparent diffusion

coefficient values (1.2 2.0 x10-3 mm2/s) derived from diffusion weighted MR imaging at 3

Tesla. This inter-patient variability is significant (p=0.006).

Correcting for inter-patient variability results in a significant increase in diagnostic accuracy

for separating low-grade and high-grade cancer, from 0.91 to 0.96 for the area under the

ROC curve (p=0.04).

A clinically useful nomogram is created to aid the radiologists in improving his assessment

of prostate cancer aggressiveness.


Correcting for patient related variation in tissue composition allows an improved method

for noninvasively predicting prostate Gleason grade, which is a very important marker used

for prognostication and management.

Summary Statement

Peripheral Zone ADC values show a significant inter-patient variation,

which has a significant effect on the prediction of prostate cancer

aggressiveness. Correcting this effect results in a significant increase in

diagnostic accuracy.

222

Effect of Inter-patient Normal Peripheral Zone ADC Variation on Prediction of Prostate Cancer Aggressiveness

11

ABSTRACT

Purpose: To determine the inter-patient variability of prostate peripheral zone (PZ) apparent

diffusion coefficient values (ADC) at 3T and the effect this has on the assessment of prostate cancer

aggressiveness.

Materials and Methods: The requirement to obtain institutional review board approval was

waived. Intra- and inter-patient variation of PZ ADC values was determined by repeated

measurements of normal PZ ADC values in a retrospective cohort of 10 consecutive patients with

high PSA level, negative transrectal ultrasound biopsy and three separate MR imaging sessions at

3T. In these patients no signs of cancer were found in all three imaging sessions. The effect of the

intra, and inter-patient variation on assessment of prostate cancer aggressiveness was examined in

a second retrospective cohort of 51 patients with PZ prostate cancer who underwent an MRI, prior

to prostatectomy. Whole-mount step-section pathology served as reference standard for placement

of regions of interest (ROIs) on PZ tumours and normal peripheral zone. Repeated-measures

ANOVA were performed to determine the significance of the inter-patient variations in PZ ADC

values. Linear logistic regression was used to assess whether incorporating normal PZ ADC values

improves the prediction of cancer aggressiveness. The effect on the diagnostic performance was

assessed using receiver-operating characteristic (ROC) analysis.

Results: The repeated-measures ANOVA revealed that inter-patient variability (1.2 2.0 x10-3

mm2/s) was significantly larger than measurement variability (average measurement standard

deviation 0.068 ± 0.027 x10-3 mm2/s) (p = 0.006). Analysis of standalone tumour ADC values

showed an AUC of 0.91 for discriminating low- vs. high-grade tumours. Incorporating normal PZ

ADC using linear logistic regression, significantly improved the AUC to 0.96 (p = 0.04).

Conclusion: Peripheral zone ADC values show a significant inter-patient variation, which has a

substantial effect on the prediction of prostate cancer aggressiveness. Accounting for this effect

results in a significant increase in diagnostic accuracy.

223


11

ABSTRACT

Purpose: To determine the inter-patient variability of prostate peripheral zone (PZ) apparent

diffusion coefficient values (ADC) at 3T and the effect this has on the assessment of prostate cancer

aggressiveness.

Materials and Methods: The requirement to obtain institutional review board approval was

waived. Intra- and inter-patient variation of PZ ADC values was determined by repeated

measurements of normal PZ ADC values in a retrospective cohort of 10 consecutive patients with

high PSA level, negative transrectal ultrasound biopsy and three separate MR imaging sessions at

3T. In these patients no signs of cancer were found in all three imaging sessions. The effect of the

intra, and inter-patient variation on assessment of prostate cancer aggressiveness was examined in

a second retrospective cohort of 51 patients with PZ prostate cancer who underwent an MRI, prior

to prostatectomy. Whole-mount step-section pathology served as reference standard for placement

of regions of interest (ROIs) on PZ tumours and normal peripheral zone. Repeated-measures

ANOVA were performed to determine the significance of the inter-patient variations in PZ ADC

values. Linear logistic regression was used to assess whether incorporating normal PZ ADC values

improves the prediction of cancer aggressiveness. The effect on the diagnostic performance was

assessed using receiver-operating characteristic (ROC) analysis.

Results: The repeated-measures ANOVA revealed that inter-patient variability (1.2 2.0 x10-3

mm2/s) was significantly larger than measurement variability (average measurement standard

deviation 0.068 ± 0.027 x10-3 mm2/s) (p = 0.006). Analysis of standalone tumour ADC values

showed an AUC of 0.91 for discriminating low- vs. high-grade tumours. Incorporating normal PZ

ADC using linear logistic regression, significantly improved the AUC to 0.96 (p = 0.04).

Conclusion: Peripheral zone ADC values show a significant inter-patient variation, which has a

substantial effect on the prediction of prostate cancer aggressiveness. Accounting for this effect

results in a significant increase in diagnostic accuracy.

INTRODUCTION

Only 15% of men diagnosed with prostate cancer show a disease specific mortality. The mortality in

the US in 2010 was 30000, with 220000 new prostate cancer cases diagnosed (1). Thus in order to

tailor treatment from more radical therapy to active surveillance protocols, accurate cancer

aggressiveness risk stratification is very important. The most useful estimator of cancer

aggressiveness is the Gleason score (GS), a histopathological scoring system used on biopsy and

prostatectomy specimens. It has become such an integral part in prostate cancer evaluation, that

patient management is largely influenced by the assessment thereof (2,3,4).

Recently, the apparent diffusion coefficient (ADC) values determined from diffusion-weighted

magnetic resonance imaging (DWI-MRI) showed to be inversely correlated to GS (5,6,7). As a result,

ADC has been proposed as a useful non-invasive biomarker for prostate cancer aggressiveness (5,7).

However, the discriminative power of ADC depends in part on the variability of the ADC

measurement. This variability is machine i.e. vendor, settings, noise - and patient dependent, the

latter caused by natural tissue heterogeneity. Based on the large inter-patient distribution of normal

PZ ADC values (1.2 2.2 x10-3 mm2/s) observed on a single MR scanner, we hypothesize that a

substantial histo-physiological heterogeneity between patients must exist (inter-patient variation)

(8,7).

Inter-patient ADC variation could affect the discriminative power of ADC both for prostate cancer

localization as well as for the determination of prostate cancer aggressiveness. Since normal

prostate PZ tissue fluctuates significantly in ADC value, the ADC values of an aggressive tumour may

show similar fluctuations. If normal PZ and tumour ADC are correlated, considering both

simultaneously, may lead to better estimates of aggressiveness.

The purpose of this study was to determine the inter-patient variability of prostate peripheral zone

(PZ) apparent diffusion coefficient values (ADC) at 3T and the effect this has on the assessment of

prostate cancer aggressiveness.

224


11


Patients

Imaging data of two retrospective patient cohorts was used in our experiments. The requirement to

obtain institutional review board approval was waived for both cohorts. To determine the

significance of the inter-patient variance relative to the measurement variability, we included 10

consecutive patients (February 2008 to June 2011, interval between scans: 6 12 months) who had

repeated measurements of normal PZ ADC values at 3 three separate MR imaging sessions at 3T.

The indication for the studies was continuously high PSA level and at least one negative transrectal

ultrasound biopsy. Patients were followed up if PSA level remained high. In these patients no cancer

was found in all three imaging sessions by an expert radiologist (J.O.B., with 18 years of experience).

If a suspicious lesion was indicated by the radiologist subsequent MR-guided biopsy found no traces

of tumour.

In addition, to determine the effect of the inter-patient variation of ADC on the prediction of

prostate cancer aggressiveness, a second cohort was included. Between August 2006 and January

2009, 51 consecutive patients with biopsy proven PZ prostate cancer (in total 61 tumours),

scheduled for radical prostatectomy, were referred from the departments of urology at the Radboud

University Nijmegen Medical Centre and the Canisius Wilhelmina Ziekenhuis in Nijmegen for MRI of

the prostate (7).

MR Imaging Protocol


Germany). The first cohort of 10 patients was scanned only with the pelvic phased array coils.

The second cohort was scanned with the use of a combined endorectal coil (ERC) (Medrad,

Pittsburgh, U.S.A) and pelvic phased array coils. The ERC was filled with a 40-mL Perfluorocarbon

preparation (Fomblin, Solvay-Solexis, Milan, Italy)

225


11

In both cohorts peristalsis was suppressed with an intramuscular administration of 20-mg

Butylscopolaminebromide (Buscopan, Boehringer-Ingelheim, Ingelheim, Germany) and 1 mg of

glucagon (Glucagen, Nordisk, Gentofte, Denmark).

The MR imaging protocol included: anatomical T2-weighted turbo spin echo sequences in axial,

sagittal and coronal planes covering the entire prostate and seminal vesicles. Axial diffusion

weighted imaging was performed using a single-shot-echo-planar imaging sequence with diffusion

modules and fat suppression pulses implemented. Water diffusion was measured in 3-scan trace

mode using b-values of 0, 50, 500, and 800 s/mm2. ADC-maps were automatically calculated by the

scanner software using all b-values. . Complete pulse sequence details can be found in table 1 for the

first cohort containing 10 patients with repeated measurements and table 2 for the second cohort.

Sequence Type

Slice thickness

Number of slices

In-plane resolution

TR TE Averages GRAPPA b-values

T2-w Axial

TSE 3.5-4 mm 13-19 0.6 mm 3540 ms 104 ms 2 -3 - -

T2-w Coronal

TSE 3.5-4 mm 15-19 0.6 0.8 mm 3350 ms 105 ms 2 3 - -

T2-w Sagital

TSE 3.5-4 mm 15-19 0.6-0.8 mm 3810 ms 105 ms 2 3 - -

DWI SE-EPI 3.5-4 mm 15-20 1.6-2.0 mm 2300 ms 61 ms 6 - 10 2 0, 50, 500, 800 mm2/s

T1-w DCE GRE

(FLASH 3D)

3.5-4 mm 14 1.8 mm 37 ms 1.47 ms 1 - -

Table 1: Pulse sequence details for the first patient cohort with repeated measurements. In-

plane resolution is the same in both directions.

226


11

Sequence Type

Slice thickness

Number of slices

In-plane resolution


T2-w Axial

TSE 4 mm 15-19 0.4 mm 3540 ms 104 ms 2 - -

T2-w Coronal

TSE 4 mm 15-19 0.5 mm 3350 ms 105 ms 2 - -

T2-w Sagital

TSE 4 mm 15-19 0.5 mm 3810 ms 105 ms 2 - -

DWI SE-EPI 4 mm 15-19 2.0 mm 2800 ms 81 ms 10 2 0, 50, 500, 800 mm2/s

T1-w DCE GRE

(FLASH 3D)

4 mm 14 1.8 mm 37 ms 1.47 ms 1 - -

Table 2: Pulse sequence details for the second patient cohort. In-plane resolution is the same in both

directions.

Reconstructed Whole-Mount Step-Section Preparation

The second cohort of patients underwent radical prostatectomy after imaging. After the radical

prostatectomy, prostate specimens were uniformly processed and entirely submitted for histologic

investigation. After histologic staining, all specimens were evaluated by one expert urological

pathologist (C.A.H.v.d.K. with 17 years of experience). Each individual tumour was graded according

to the 2005 International Society of Urological Pathology Modified Gleason Grading System.

Peripheral zone tumours, with a size of >0.5cc in volume, were divided in two groups, and classified

as low- and high-grade tumours. Tumours with a Gleason grade 4 or 5 component were defined as

high-grade. Low-grade tumours were defined as tumours harboring only Gleason grades 2 and 3.

227


11

Annotation of MR images

All annotations were performed using an in-house developed MR viewing and reporting system. In

the first cohort the center slice of the prostate in the axial direction was used to annotate the

peripheral zone. For this slice the whole peripheral zone was annotated and the median ADC value

was extracted from this annotation.

For the second cohort, ADC maps were acquired in the same orientation and of similar thickness as

the histopathology step-section. A previously described translation technique was used to match

every tumour containing histopathology step-section to a corresponding ADC map (7). Using

histopathology as gold standard, a region of interest (ROI) was placed by one radiologist (T.H. with

four years of experience) and one urologist (with one year of experience) in consensus, on the ADC

maps. The size and extent of the ROI were chosen such that it matched the tumour size and extent

obtained from histological examination as closely as possible. Median ADC values were extracted for

each tumour slice separately. In clinical practice, the ADC slice revealing the lowest signal intensity

for tumour alerts radiologists. Therefore, for each individual PZ tumour, the tumour slice revealing

the lowest ADC values was used for further assessment (7).

Lastly, to determine the effect of incorporating normal PZ ADC values on the prediction of cancer

aggressiveness, an ROI was placed in the normal PZ tissue of every patient. This region was always

selected adjacent to the tumour, in order to be the most representative area of normal PZ ADC value

at the tumour location. This was done to attempt to minimize intra-patient heterogeneity. Median

ADC values were extracted from all ROIs. Median values were used because they are more robust to

image artifacts that might occur due to ADC calculation by the scanner.


Our first hypothesis is that there is a significant degree of inter-patient variation in normal PZ ADC

values. This was assessed using a repeated-measures ANOVA. The repeated measure was the

median ADC value of normal PZ tissue, which was obtained three times for each of the 10 patients in

the first cohort.

228


11

Our second hypothesis is that joint analysis of the normal PZ ADC values and the tumour ADC values

will result in an improved prediction of cancer aggressiveness, because this implicitly corrects for

the inter-patient variations in normal PZ ADC. Multivariate linear logistic regression was used to

test this hypothesis. We can express a regression model of cancer grade as:

Z = C + T ADCT + N ADCN (1)

e z p = (2)

1 + e z

The p indicates the probability that a cancer is high-grade and the ADC variables indicate the

median ADC of the corresponding ROI. Subscripts T and N are tumour and normal PZ respectively.

The beta terms are the regression coefficient corresponding to these variables. Equation 2

represents the conversion from z to the probability p.

The linear logistic regression results in values for and and the significance of these variables

in the regression model. Two regression models were created to compare diagnostic performance:

using only tumour ADC values and using tumour and normal PZ ADC values. SPSS (SPSS, version

16.0.01, Chicago, U.S.A.) was used for the statistical analysis. Furthermore, a visual assessment is

given for the correlation between tumour ADC and normal peripheral zone ADC by plotting the low-

and high-grade tumours with respect to their ADC values and the corresponding normal PZ ADC

values.

Our third hypothesis is that the improved prediction of prostate cancer aggressiveness may result in

a significant improvement in diagnostic accuracy in separating low- and high-grade cancer.

Receiver-operating characteristic (ROC) curves were constructed for a standalone tumour ADC

regression model and the regression model, which incorporates normal PZ ADC values. The areas

under the ROC curves were tested for significant differences using the ROCKIT software package

(Kurt Rossmann Laboratories, University of Chicago, Chicago) (9).

229


11

Nomogram construction

Additionally, the regression model incorporating tumour and normal PZ ADC can be used to

construct a nomogram by evaluating the obtained equation for a range of ADC values. The ranges

used to construct the nomogram are 0.5 1.7 x10-3 mm2/s for the tumour ADC value and 0.8 2.2

x10-3 mm2/s for the normal PZ ADC value. These ranges are slightly larger than the ranges found in

this study to accommodate more extreme values.

RESULTS

Assessment of inter-patient variation of normal PZ ADC values

Normal PZ ADC values differed significantly between patients relative to measurement variability

(p-value < 0.001) as assessed using the repeated measures ANOVA. The ADC measurements are

plotted in figure 1.

Figure 1: Three median ADC measurements of the peripheral zone of 10 patients. The black

dots represent the individual measurements, the vertical axis shows the median ADC value,

the horizontal axis shows to which patient the measurement belongs

230


11

Effect of including normal PZ ADC values in the prediction of cancer aggressiveness

Normal peripheral zone ADC correlates with ADC of high-grade tumours. Its addition to the

regression model results in a significantly improved prediction of aggressiveness (p = 0.01). This

was determined using the logistic regression procedure; the results are summarized in table 3.

Included parameters in the model

Tumour median ADC Tumour and normal PZ median ADC

9.103 0.000 -18.82 0.003

- - 13.43 0.013

C 10.76 0.000 0.126 0.978

Table 3: Result of the linear logistic regression for three regressions based on equation 1 and

2. Regressions performed: using only tumour ADC and using tumor and normal PZ median

ADC. The second row shows the values used in each regression. The regression parameters

are presented in the bottom three rows, their value and significance respectively for each

regression.

Both regression models show a significant contribution of the tumour ADC (p = 0.003). Normal PZ

ADC values also show a significant contribution to the regression model (p = 0.01).

The regression model using standalone tumour ADC values can then be expressed as:

z = 10.76 9.103 ADCT (3)

and the model combining tumour and normal PZ ADC values can be expressed as:

z = 0.126 18.82 ADCT + 13.43 ADCN (4)

231


11

In combination with equation 2 these models result in a probability that a given sample is a high-

grade cancer. The model incorporating normal PZ ADC (Eq. 4), together with the data used in the

regression, is shown in figure 2. This plot indicates that a relatively high tumour ADC value might

still constitute a high-grade tumour if the normal PZ ADC is high. In addition one can appreciate that

using a static threshold on tumour ADC (a vertical line/contour in figure 2) to determine cancer

aggressiveness could result in incorrect diagnosis in some patients.

Figure 2: Decision Boundary at p=.5 of the logistic regression model. The line represents the

decision boundary, the green dots the low-grade cancer and the red dots the high-grade

cancers.

232


11

Diagnostic performance of the regression models

Including normal PZ, significantly (p = 0.04) improved the diagnostic accuracy. The ROC curves for

the regression models in equations 3 and 4 are shown in figure 3. The area under the curve

increases from 0.91 to 0.96.

Figure 3: ROC curve of the regression models. The red line shows the diagnostic accuracy

when including the adjacent PZ tissue median ADC in addition to the tumor ADC, the blue line

show the diagnostic accuracy when only using tumor ADC

233


11

Nomogram

The constructed nomogram is shown in figure 4. This nomogram can be used in a clinical setup to

quickly lookup the change of a certain region with the peripheral zone being a aggressive cancer.

Figure 4: Contour of the probabilities of having an aggressive cancer given the adjacent PZ

tissue ADC (vertical axis) and the tumor ADC (horizontal axis). The point corresponding to

these two values will correspond to the probability of a high-grade cancer. The probability

values are specified along the contours and in the color bar on the right of the figure.

234


11

DISCUSSION

In this study we have shown that there is a significant inter-patient variation in normal peripheral

zone ADC values (1.2 2.0 x10-3 mm2/s), which cannot be solely attributed to measurement

variability (average measurement standard deviation 0.068 ± 0.027 x10-3 mm2/s). We hypothesize

that the inter-patient variations arise from natural variations in prostate physiology.

Secondly, adding normal PZ ADC values to the linear logistic regression, results in a significantly

improved prediction of cancer aggressiveness (p = 0.01). This suggests that tumour ADC values

normal PZ tissue composition.

Thirdly, the improvement also results in an increased area under the ROC curve, from 0.91 to 0.96

(p < 0.05), thus an improved diagnostic accuracy.

This study has a number of limitations. First, the use of ADC to assess aggressiveness of transition

zone tumours has not been investigated in this study. Second, this study was limited to the

peripheral zone. This was done because it is known that ADC in peripheral and transition zone

tumours can differ substantially. However, the majority of prostate tumours arise in the PZ. Third,

the annotation of ROIs was performed by a single observer; the effect of the inter-observer

variability on the regression model was not assessed. Our nomogram must be tested and validated

in a prospective multi-reader study.

CONCLUSION

In conclusion, there is a large inter-patient variation in prostate peripheral zone ADC values. This

variation propagates into tumour ADC values. Compensating for this variation by combining tumour

and normal PZ ADC when assessing cancer grade significantly increases diagnostic performance.

235


11

REFERENCES

1. Jemal A, Siegel R, Xu J, and Ward E. Cancer statistics, 2010. CA Cancer J Clin, 60:277 300, 2010.

2. Blute M.L, Bergstralh E.J, Iocca A, Scherer B, and Zincke H. Use of gleason score, prostate specific antigen, seminal vesicle and margin status to predict biochemical failure after radical prostatectomy. J Urol, 165(1):119 125 2001.

3. Egevad L, Granfors T, Karlberg L, Bergh A, and Stattin P. Percent gleason grade 4/5 as prognostic factor in prostate cancer diagnosed at transurethral resection. J Urol, 168(2):509 513 2002.

4. Narain V, Bianco F.J, Grignon D.J, Sakr W.A, Pontes J.E, and Wood D.P. How accurately does prostate biopsy gleason score predict pathologic findings and disease free survival? Prostate, 49(3):185 190 2001.

5. Itou Y, Nakanishi K, Narumi Y, Nishizawa Y, and Tsukuma H. Clinical utility of apparent diffusion coefficient (ADC) values in patients with prostate cancer: can ADC values contribute to assess the aggressiveness of prostate cancer? J Magn Reson Imaging, 33(1):167 172 2011.

6. Turkbey B, Pinto P.A, Mani H, Bernardo M, Pang Y, McKinney Y.L, Khurana K, Ravizzini G.C, Albert P.S, Merino M.J, and Choyke P.L. Prostate cancer: value of multiparametric mr imaging at 3 t for detection histopathologic correlation. Radiology, 255(1):89 99 2010.

7. Relation of Apparent Diffusion Coefficient Values at 3 Tesla Magnetic Resonance Imaging with Prostate Cancer Gleason Grade in the Peripheral Zone. Hambrock T, Somford D, Huisman H, van Oort I, Witjes J, Hulsbergen van de Kaa C, Scheenen T, Barentsz JO. Radiology 2011 May; 259 (2): 453-461

8. Vargas H.A, Akin O, Franiel T, Mazaheri Y, Zheng J, Moskowitz C, Udo K, Eastham J, and Hricak H. Diffusion-weighted endorectal MR imaging at 3 T for prostate cancer: Tumour detection and assessment of aggressiveness. Radiology 259(3):775-784 2011.

9. Metz C.E, Herman B.A, and Roe C.A. Statistical comparison of two roc-curve estimates obtained from partially-paired datasets. Med Decis Making, 18(1):110 121, 1998.

236


11


11

CHAPTER 12

CAD of prostate cancer using 3T MP-MRI: Effect on Observer

Performance

Thomas Hambrock; Pieter Vos; Christina Hulsbergen et al.


Computer-aided Diagnosis of Prostate Cancer using Multiparametric 3 Tesla Magnetic Resonance Imaging: Effect on Observer Performance


Hambrock T, Vos P, Hulsbergen-van de Kaa C, Barentsz J, Huisman J

Advances in knowledge: Development of a Computer Assisted Diagnosis (CAD) method using quantitative features

obtained from 3T DWI and DCE MR imaging for prostate lesions characterization, is feasible.

The stand-alone performance (AUC=0.90) of a Computer Assisted Diagnosis (CAD) method

using quantitative features obtained from 3T Diffusion Weighted and Dynamic Contrast

Enhanced MR imaging for prostate lesions characterization, is similar to prostate

experienced radiologists (AUC=0.88).

The addition of CAD significantly improved lesion discriminating performance for less-

experienced radiologists both for the peripheral zone (p<0.001) as well as the transition

zone (p=0.01).

After CAD, less-experienced radiologists (AUC=0.91) reached similar performances as

experienced radiologists (AUC=0.93).


CAD methods that aid radiologists especially those less experienced in prostate MRI, may

expedite utilization of multi-parametric MR imaging for accurate detection and localization

of prostate cancer.

Summary Statement

The addition of CAD for evaluation of prostate cancer suspicious regions on 3T multi-parametric MRI, significantly improves the discriminating

performance for less experienced observers for both the peripheral and transition zone.

238

Computer Aided Diagnosis of Prostate Cancer using 3T Multiparametric MRI: Effect on Observer Performance

12

ABSTRACT

Purpose: To determine the effect of computer-aided diagnosis (CAD) on less-experienced and

experienced observer performance in differentiating benign and malignant prostate lesions on 3T

multi-parametric MRI (MP-MRI).

Materials and Methods: The institutional review board waived the need for informed consent.

Retrospectively 34 patients were included that had: prostate cancer, received an MP-MRI including

T2-weighted, diffusion weighted and dynamic contrast enhanced MRI prior to radical

prostatectomy. Six less-experienced and 4 experienced prostate radiologists were asked to

characterize different cancer suspicious regions as benign or malignant on MP-MRI, first without

and subsequently with the use of CAD software. The effect of CAD was statistically analyzed using a

multiple-reader, multi-case, receiver operating characteristic analysis and linear mixed-model

analysis.

Results: In 34 patients, 206 pre-annotated regions, including 67 malignant and 64 benign regions in

the peripheral zone (PZ) and 19 malignant and 56 benign regions in the transition zone (TZ) were

evaluated. Stand-alone CAD had an overall area-under the receiver operating characteristic curve

(AUC) of 0.90. For PZ and TZ lesions, the AUC’s were 0.92 and 0.87 respectively. Less-experienced

observers (LEO) had a 0.81 overall pre-CAD AUC which significantly increased to 0.91(p=0.001)

post-CAD. For experienced observers (EO) the pre-CAD AUC was 0.88 which increased to

0.91(p=0.17). For PZ lesions, LEOs increased their AUC from 0.86 to 0.95(p<0.001) post-CAD. EOs

showed an increase from 0.91 to 0.93(p=0.13). For TZ lesions, LEOs significantly increased their

performances from 0.72 to 0.79(p=0.01) post-CAD and EOs from 0.81 to 0.82(p=0.42).

Conclusion: The addition of CAD significantly improved the performance of less experienced

observers in distinguishing benign and malignant lesions, who when using CAD, reached similar

performance as experienced observers. The stand-alone performance of CAD was similar to

experienced observers.

239


12

INTRODUCTION

Despite high-resolution imaging, the signal intensities on anatomic T2-weighted MR images (T2-w)

show overlap between prostate cancer, post-biopsy hematoma, benign prostatic hypertrophy,

fibrosis and prostatitis. Therefore, functional imaging modalities like dynamic contrast enhanced

MR imaging (DCE-MRI), diffusion weighted MR imaging (DWI) and spectroscopic MR imaging have

been employed. Multi-parametric MR imaging both at 1.5T and 3T, has proven to be valuable in

detection, localization and characterization of prostate cancer(1-4).

High-temporal resolution DCE-MRI, allows both qualitative and quantitative estimations of

perfusion, capillary surface area and extravascular extracellular space. All these parameters are

modified by angiogenesis. Different techniques for compartment modeling, to determine arterial

input function (AIF) and gadolinium concentration, have been studied in order to improve the

objectivity and reproducibility of quantitative pharmacokinetic parameters(5;6). Reference tissue

techniques have shown promising results in generating more robust and accurate estimations of the

AIF(7;8). DWI which measures the restriction of free proton movement, has been used increasingly

in prostate MR imaging, not only for detection and localization, but also to assess the degree of

aggressiveness of the lesion(9;10). The apparent diffusion coefficient (ADC) values calculated from

the DWI, allow a more objective quantitative assessment of the tissue micro-environment.

However, despite the quantitative nature of pharmacokinetic DCE MRI and ADC values, prostate

cancer analysis on MP-MRI is still challenging, as it requires a high level of expertise and suffers

from observer variability(11). Therefore a need exists to help radiologists improve their assessment

of MP-MRI and to reduce their inter-observer variability.

Computer assisted diagnosis (CAD) techniques for the radiological assessment of various

malignancies such as for breast cancer(12), lung cancer(13;14) and colorectal cancer(15) have been

developed. CAD appears to be particularly helpful to less-experienced radiologists in improving

their ability to detect tumors(16;17). Studies that incorporated various features of MRI did show

the feasibility of CAD to discriminate benign and malignant lesions in the prostate with reported

high accuracies(18-20). However, there has been no evaluation of a CAD system in a reader study

240


12

with experienced and less experienced radiologists to determine the effect of CAD on observer

performance in for distinguishing benign from malignant prostate lesions.

The purpose of this study was to determine the effect of computer-aided diagnosis (CAD) on less-

experienced and experienced observer performance in differentiating benign and malignant

prostate lesions on 3T multi-parametric MRI (MP-MRI).


Study population

The institutional review board waived the need for informed consent. Between Jan 2008 and

January 2009, 50 consecutive patients with biopsy proven prostate cancer who were scheduled for

radical prostatectomy, were referred from the department of urology at the Radboud University

Nijmegen Medical Centre, Nijmegen, Netherlands. All patients received a clinically routine MP-MRI

for tumor localization and staging prior to radical prostatectomy. Inclusion criteria for the study

were: a) MR imaging performed at 3 Tesla using an endorectal coil combined with pelvic phased

array coils; b) MR imaging included T2-w imaging in 3 planes, DWI and DCE-MRI; c) prostatectomy

performed in our institution and analyzed by one expert prostate pathologist. Exclusion criteria

were: previous hormonal or radiotherapy or substantial imaging artifacts related to patient

movement.

10 Patients were excluded because imaging was performed without the endorectal coil, 4 due to

severe imaging artifacts and 2 because no DCE-MRI was performed. Finally 34 patients were

included with a mean age of 64 years (range 53-74) and mean PSA of 7.5 ng/ml (range 3.4 – 21.8).

241


12

Figure 1: Flow chart diagram of patient inclusion. PZ=peripheral zone; TZ=Transition zone;

DWI=Diffusion weighted imaging; DCE=Dynamic contrast enhanced MRI; T2-w=T2 weighted

imaging.

Radical Prostatectomies

2008-2009

Endorectal 3T MR imaging

including T2-w,

DWI and DCE

n= 34 patients

Exclusion:

No combined DWI

and/or DCE

n= 2No endorectal coil

n= 10

Severe motion artifacts

n= 4

TZ lesions

n= 75

PZ tumors

n= 67

TZ benign lesions n= 56

PZ lesions

n= 131

PZ benign lesions n= 64

TZ tumors

n= 19

242


12

MR Imaging

MR imaging was performed using a 3T MR scanner (Siemens, Trio Tim, Erlangen, Germany) with the

use of an endorectal coil (Medrad, Pittsburgh) and pelvic phased array coil. The endorectal coil was

filled with a 40-mL perfluorocarbon preparation (Fomblin;Solvay-Solexis, Milan, Italy). Peristalsis

was suppressed with an intramuscular administration of 20-mg Butylscopolaminebromide

(Buscopan; Boehringer-Ingelheim, Ingelheim, Germany) and 1 mg of glucagon (Glucagen; Nordisk,

Gentofte, Denmark). The imaging sequence parameters are shown in Table 1. Gadopentetate

dimeglumine (Dotarem; Guerbet, Paris, France), of which 15 ml was administered with a power

injector (Spectris; Medrad) at 2.5mL/s and followed by a 30-mL saline flush, was used as contrast

agent.

Sequence Type

Slice thickness

Slices In-plane resolution


T2W Axial TSE 3 mm 15-19 0.4x0.4 mm 4260 ms 99 ms 2 - -

T2W Coronal TSE 3 mm 15-19 0.5x0.5 mm 3590 ms 98 ms 2 - -

T2W Sagital TSE 3 mm 15-19 0.5x0.5 mm 4290 ms 105 ms 2 - -

DWI SE-EPI 3 mm 15-19 1.5x1.5 mm 2800 ms 81 ms 10 3 0, 50, 500, 800 mm2/s

T1-w DCE GRE FLASH 3 mm 16 1.5x1.5 mm 34 ms 1.4 ms 1 2 -

Table 1. MR imaging sequence parameters. T2-w: T2-weighted; T1-w: T1-weighted; DWI:

Diffusion weighted imaging; DCE: Dynamic contrast enhanced imaging; TSE: Turbo Spin Echo;

SE-EPI: Spin Echo Echo Planar Imaging; TR: Repetition Time; TE: Echo Time; GRAPPA:

Parallel imaging factor; GRE: Gradient echo imaging; FLASH: Fast low angle shot imaging.

243


12

Pharmacokinetic MRI

Pharmacokinetic maps were generated offline by an in-house developed system in three steps.

Firstly, kinetic DCE imaging parameters were estimated by fitting each MR signal enhancement-time

curve to a general exponential signal enhancement model as described previously(21). Secondly,

the signal enhancement-time curves were converted to tracer concentration [mmol/ml] time curves

by applying the approach of Hittmair et al.(22). Thirdly, inter-patient plasma profile variability was

removed using the reference tissue method presented by Kovar et al.(23). The reference tissue

method assumes that a tissue area within the patient is available with a known tissue model based

on literature values. For this experiment, the reference tissue was manually determined in

consensus by two radiologists, by placing a region of interest (ROI) of 5x5 mm in the normal

appearing peripheral zone which was visually characterized by a high-signal intensity on T2-w and

ADC as well as homogeneous enhancement after contrast. Estimation of pharmacokinetic

parameters was thereafter performed, conform to the theoretical derivations(24). An extensive

description of the method can be found in Vos et al.(25). The pharmacokinetic parameters estimated

are Ve, kep and Ktrans (=Ve x kep) and Washout, where Ve is an estimate of the extravascular

extracellular volume (expressed as a percentage), Ktrans is the volume transfer constant (1/minute),

Kep is the rate constant (1/minute) between the extracellular extravascular space and the plasma

space and Washout (semi-quantitative) the slope of the curve following peak enhancement

(1/minute).

Histopathological evaluation of prostatectomies


submitted for histological investigation. Immediately after surgical resection, specimens were fixed

in 10% neutral-buffered formalin, using fine needle formalin injections and fixation overnight.

Subsequently, the entire surface was marked with ink using three different colours, after which the

entire prostate specimen was cut into serial transverse 4 mm thick slices, perpendicular to the

dorsal-rectal surface in the same plane the axial MR images were acquired. All slices were

macroscopically photographed with a CCD-camera. After histological staining all specimens were

244


12

evaluated by one expert urological pathologist (C.A.H.K, 17 years experience). Tumors were

outlined on the microscopic slides and subsequently mapped on the macroscopic photographs to

allow depiction of tumor extent and multi-focality. From prostatectomy the calculated median

tumor volume was 2.34 cc (range 0.5-32) and median Gleason score 7 (range 5-9) in the PZ while

the median tumor volume in the TZ was 2.5 cc (range 0.5-12.48) and median Gleason score was 6

(range 5-7). Fifty-six percent of patients had stage pT2 and 44% stage pT3.

Standard of Reference

Histology tumor maps were used as ground truth for cancerous regions. Annotations of MR images

were performed in consensus by two radiologists (T.H and P.V. both with 6 years experience in

prostate MRI). The morphology of the central gland, peripheral zone, cysts, calcifications, and

urethra were used as landmarks to find the corresponding MRI slices. Translation techniques as

described previously were used(26). The anatomy of the prostate is best depicted on the axial T2-w

images. These were compared with the histopathologic slices. First, based on histopathology, all

tumor regions were identified and a region of interest (ROI) placed in the peripheral and transition

zone corresponding to tumor extent on histopathology. Only tumors >0.5 cc were annotated. Only

based on imaging findings, additional benign regions were annotated when: a) focal low-signal

intensity on T2-w images and/or b) relative focal low signal on ADC maps and/or c) suspicious

irregular focal enhancing areas were evident but the underlying histopathology revealed no cancer.

Therefore, all cancerous areas and areas minimal to strongly suspicious of malignancy based on

current known features on multi-parametric imaging were annotated. To allow exact spatial

matching of the different imaging sets, a manual registration was applied for all patients, to correct

for patient related movement. In the 34 patients, 120 benign and 86 malignant lesions were

annotated and evaluated. Of the 120 benign lesions, 64 were in the PZ and 56 in the TZ. Of the 86

malignant lesions, 67 were in the PZ and 19 in the TZ. The median number of ROIs per patient was 5

(range 2 to 8).

245


12

CAD system

An in-house developed CAD system was used to assist the radiologists in the diagnosis of prostate

lesions. An extensive description of the system can be found in previous publications(27;28).

Briefly, the CAD system can visualize multi-parametric MRI and derived maps simultaneously in

multiple linked views either as background or as transparent color-coded overlays. Figure 2

demonstrates the CAD system with a dedicated prostate hanging protocol as it was used in the

experiment. The CAD system characterizes a region of interest by extracting a relevant feature set

from the available quantitative DCE and ADC maps. The extracted set of features is presented to a

trained classifier that calculates the likelihood of malignancy. Thereafter, the calculated likelihood is

presented to the radiologist to assist in their diagnosis, as shown in Figure 3. For this observer

study, two linear discriminant analysis classifiers were trained separately for the PZ and TZ. For the

two classifiers, the selection of features was carried out by Sequential Forward Floating Selection

(SFFS)(29) to establish the most discriminant features. The SFFS procedure uses leave-one-patient-

out training and testing with the area under the Receiver Operating Characteristic (ROC) curve as

the criterion to be optimized. For the PZ, the 25th percentile of ADC values, 75th percentiles of Ktrans

and Ve and 25th percentile of WashOut were selected. For the TZ, the 25th percentile of ADC values

and 25th percentile of WashOut were selected. The bootstrap resampling approach with 1000

iterations was used for estimating the bootstrap mean area-under-the (AUC) receiver operator

characteristic curve as well as the 95% confidence intervals(30).

246


12

Figure 2: Example patient case of the observer study. The MRCAD observer hanging protocol

shows on the top row from left to right a) T2-w axial b) T2-w coronal c) T2-w sagittal and d)

ADC map. On the bottom row T1-w axial images are displayed with the pharmacokinetic

maps as transparent color overlays representing e) Ktrans , f) Ve, g) WashOut and h) native T1-

w image prior to contrast. The separate window shows the scoring interactive screen tool

that the observer uses to enter a malignancy likelihood for the provided region of interest

(see figure 3 for more info).

247


12

Observer Study

The anonimized studies of all patients and the resultant ROIs were shown in identical order to 10

observers. The observers varied in their level of experience: 6 less experienced observers in multi-

parametric MR imaging of the prostate (<50 prostate MRIs evaluated) and 4 prostate experienced

observers (>100 prostate MRIs evaluated). Observers were informed that all patients had biopsy

proven prostate cancer followed by prostatectomy. The CAD system was designed to include an

experimental environment where predefined ROIs were automatically displayed for

characterization by the observers. For each ROI to be evaluated, the axial, coronal and sagital T2-w

images, the ADC map, the pre-contrast T1-w images and as color-coded transparent overlays, the

DCE parameters: Ktrans, WashOut and Ve were shown. An automated hanging protocol ensured that

all images and maps of each patient were synchronized to display each ROI. Lookup tables provided

encoding of scalar values. Window and level settings were automatically set and fixed to a

predefined intensity range: ADC (0.5 – 1.5x10-3 mm/s2) and Pharmacokinetic: Ktrans (1-3/s);

Washout (-1 – -10/s) and Ve (20 – 70%). See 2 for an example.

For every ROI shown, the observer was instructed to first provide an estimate of the likelihood of

malignancy on a scale of 0-100%. An interactive tool was displayed on top of the CAD system to

guide the observer through the successive ROIs of each patient. The tool recorded a (pre-CAD)

malignancy likelihood entered by the observer for a given ROI. Hereafter the ROI CAD likelihood

was displayed to the observer in combination with a distribution of the predicted likelihoods that

was obtained during training of the classifier in relation to their reference standard (Figure 3).

Subsequently, the observer entered an additional (post-CAD) malignancy likelihood before the next

ROI was shown. Observers were provided with the stand-alone performance of the of the CAD

system as the area-under the receiver operating characteristic curve (AUC) in the PZ and TZ

respectively. Prior to the study, all observers were trained and familiarized with the CAD system,

evaluating 4 cancer patients with a total of 25 different ROIs in the PZ and TZ with and without CAD.

248


12

Figure 3: The interactive screen tool used for scoring in this observer study. The observer is

asked to provide a malignancy likelihood for the provided region after which the observer

presses the Score button. After the observer provides the pre-CAD malignancy likelihood, the

CAD system calculates a malignancy score which is presented (dashed vertical line) in a

density plot. The green area in the density plot summarizes the smoothed distribution of all

the calculated likelihoods for all benign regions from the database used for training the

classifier. Likewise, the blue and red area corresponds with the normal and malignant

regions, respectively. Hereafter, the observer can enter a post-CAD malignancy likelihood

while taking the CAD prediction into account.

249


12


A multiple-reader multiple-case (MRMC) ROC analysis (DBM MRMC 2.2, Kurt Rossmann

Laboratories, Chicago, U.S.A) was performed. The average AUC for less experienced observers and

experienced observers for the whole prostate as well as the PZ and TZ separately, were established

before and after CAD. As MRMC analysis cannot provide statistical significance in repeated

observational studies, additional linear mixed model analysis (using SPSS version 17) was

performed to determine the significance. P-values less than .05 were considered to indicate a

significant difference.

RESULTS

CAD Stand-Alone Performance

The overall CAD stand-alone AUC was 0.90 (CI 0.83-0.96) while for the PZ and TZ this was

respectively 0.92 (CI 0.88-0.96) and 0.87 (CI 0.78-0.96).

Observer Performance without CAD

Less-experienced observers had an overall pre-CAD AUC of 0.81 (CI 0.76-0.85); for the PZ this was

0.86 (CI 0.83-0.88) and for the TZ, 0.72 (CI 0.66-0.77). Experienced observerss had an overall AUC

of 0.88 (CI 0.85-0.93), for the PZ and TZ this was 0.91 (CI 0.89-0.93) and 0.81 (CI 0.69-0.94)

respectively.

Observer Performance with CAD

When the observers were allowed to change their ratings depending on CAD predictions, the overall

average AUC for less-experienced observers improved significantly to 0.91 (CI 0.90-0.93; p=0.001)

and for experienced observers to 0.91 (CI 0.86-0.97; p=0.17). For less-experienced observers this

was more evident for the PZ where the average AUC increased to 0.95 (CI 0.94–0.95; p<0.001)

compared to the TZ, where the average AUC improved to 0.79 (CI 0.76-0.83; p=0.01). Experienced

250


12

observers revealed no significant improvement in overall PZ (post-CAD AUC=0.93 [0.90-0.97];

p=0.13) or TZ lesion characterization (post-CAD AUC=0.82 [0.68-1.00]; p=0.42). A summary of the

pre- and post-CAD performances are shown in Table 1 and 2 as well as Figure 4.

Pre-CAD performance (AUC)

Post-CAD performance (AUC)

Significance (p-values)

Less-experienced observers

Overall

PZ

TZ

0.81 (0.76-0.85)

0.86 (0.83-0.88)

0.72 (0.66-0.77)

0.91 (0.90-0.93)

0.95 (0.94-0.95)

0.79 (0.76-0.83)

0.001 *

< 0.001*

0.01*

Experienced observers

Overall

PZ

TZ

0.88 (0.85-0.93)

0.91 (0.89-0.93)

0.81 (0.69-0.94)

0.91 (0.86-0.97)

0.93 (0.90-0.97

0.82 (0.68-1.00)

0.17

0.13

0.42

Table 2. Summary of mean pre- and post-CAD performances for readers grouped into less-

experienced and experienced readers. * denotes a statistical significance. PZ=Peripheral

zone; TZ=Transitional zone; AUC=area-under the receiver operating characteristics curve.

DISCUSSION

In our study, we have demonstrated the effectiveness of CAD in aiding radiologists in the

characterization of prostate lesions as benign or malignant using information obtained from

quantitative pharmacokinetic DCE parameters and ADC values. The performance in discriminating

lesions in the PZ and TZ significantly improved for less experienced observers when assisted by

CAD. Furthermore, when CAD was used, the performance of the less experienced observers were

comparable to that of the experienced observers CAD however, did not significantly improve the

performance of experienced observers. Furthermore it was evident that the inter-observer

251


12

variability for less experienced observers was large (CI 0.76-0.85) and that CAD not only improved

the overall performance to the level of experienced observers, but also appears to decrease inter-

observer variability (CI 0.90-0.93).

Multi-parametric MR imaging is the most accurate imaging technique available to detect, localize

and stage prostate cancer(31-35). The additional value of information obtained from DWI is also

gaining importance as a tumor aggressiveness biomarker, especially at 3T. T2-w imaging can still

be regarded as the cornerstone of prostate evaluation as anatomy is exquisitely well depicted

although low-grade tumors might not be depicted and therefore detected as easily as high-grade

tumors(36). The overall accuracy of T2-w imaging has therefore been rather low with AUC ranging

between 0.68 and 0.81(37;38). DCE-MR imaging suffers from a similar lack of specificity as

prostatitis, high-grade PIN and normal BPH can also show increased vascularization and perfusion.

For DWI, BPH and fibrosis also reveal increased proton movement restriction. Therefore, a multi-

parametric approach combining all three imaging modalities has been shown to be most

optimal(39;40). These authors showed that information obtained from the combination of the

different imaging techniques provides a better discriminating performance than each technique

individually. Yet, MP-MRI evaluation remains challenging, is largely dependent on experience and

substantial inter-observer variability in interpretation exists.

As most MR imaging modalities lack specificity, the goal of our CAD approach was to improve the

ability to differentiate benign and malignant lesions using multi-parametric information from DCE

and DWI. This was done using linear discriminant analysis to determine the likelihood that a region

represents malignancy or not. The current standard paradigm for the use of CAD systems is to use

CAD as a second reader. After the radiologist has evaluated the multiple imaging sets, CAD indicates

the likelihood that a given suspicious region is malignant, thus aiding in differentiation. The CAD

system we used has a performance of (AUC TZ: 0.87, PZ: 0.92) in discriminating benign from

malignant lesions, similar to that of an experienced radiologist (AUC TZ: 0.81, PZ: 0.91). In routine

clinical practice, many radiologists tend to evaluate MR images without quantitative analysis. For

example, on DCE-MRI, enhancement patterns that indicate the presence of tumor are compared to

the relative enhancement of the normal surrounding prostate tissue. In addition, ADC maps are

252


12

visually evaluated by looking at a focal area of relative restriction that may be indicative of tumor

although absolute values are also often used. The lack of consensus on standardized cut-off values

both for DCE and ADC limits widespread utilization and uniformity of results. The evaluation of

multi-parametric MRI requires a high level of experience and induces observer variability(41).

Figure 4: Average receiver operating characteristics (ROC) curves for the less experienced

(A. and B.) and experienced readers (C. and D.). For the peripheral zone (A. and C.) and

transition zone (C. and D.) the pre-CAD and post-CAD curves as well as the area-under the

ROC (AUC) values are provided. The dotted green lines donates the 0.5 value. The dotted

blue line represents the pre-CAD curve and the continuous black line represents the post-

CAD curve.

253


12

Histologically the normal PZ consists of more glandular components than the TZ, which due to

common BPH formation has a larger stromal component including compact muscle fibers. On MR

imaging this results in a lower T2-w signal and ADC values compared to the normal PZ where higher

values are seen. In addition, the higher vascularity of BPH nodules result in enhancement patterns

on DCE that are similar to that of cancer(42). The differences in MR appearances of TZ cancers

compared to PZ cancers have been reported before(43;44). Transition zone tumors are known to

have different genetic mutations, biological behavior and overall prognosis (being more favorable)

compared to PZ tumors(45;46). TZ tumors are also often larger in volume and are associated with

higher PSA values, yet these are often of lower grade and more likely to be confined to the prostate.

For this reason, the CAD system we used, consists of two separate classifiers for characterization of

the PZ and TZ lesions. The results of our study confirm that the TZ is indeed a more challenging

location to evaluate than the PZ, since lower stand-alone CAD performance (AUC 0.87 vs. 0.91 in the

PZ) and lower overall observer performance (AUC 0.72-0.81) were observed compared to the PZ

(AUC 0.86-0.91).

Our study has a number of limitations. First, we have used multiple ROI observations per patient,

which may hamper a straightforward interpretation of the results. A linear mixed model analysis

which incorporates findings from multiple observations in the same patient was used to determine

significance. Secondly, the number of TZ tumors was fairly low compared to PZ tumors. This is

consistent with the known overall lower prevalence (30%) of these tumors identified clinically(47).

Therefore, the overall performance of both the CAD system as well as that of the readers may rather

constitute a PZ dominated result. Thirdly, as an integral part of DCE MRI quantification, the

reference tissue calibration method requires an annotation of normal PZ. For this study, this was

done manually prior to the experiment. Ideally, such annotation should be performed automatically

without requiring any user interaction. Previous studies have shown that despite the fact that

automatic calibration performs better compared to a general estimate, manual calibration is still

superior at this stage(7). Fourthly, all regions scored by the observer were annotated and

predefined beforehand. Although this is substantial drawback, primarily evaluating the potential of

a CAD system for prostate cancer evaluation on MRI requires limitating the observer variability

254


12

caused by self annotating. Subsequent studies should first identify the improvement when readers

identify potential suspicious regions themselves and then obtain the CAD support for that region. If

this proves to be useful as well, CAD systems should ideally calculate tumor probability maps, but

this requires further prostate segmentation algorithms which are part of ongoing developments.

Exact geometrical alignment of histopathology and MR imaging is considered very difficult but a

number of strategies have been implemented incl. using percentiles (25/75) to capture hot-spot

features within ROI’s. A final limitation relates to the fact that hydrogen spectroscopic MR imaging

was not included in our CAD system but future systems should be developed using these additional

features as well.

CONCLUSIONS

In conclusion, we have shown that the addition of CAD on multi-parametric 3T MP-MRI, significantly

improves the discriminating performance for less experienced observes for both the PZ and TZ.

Furthermore less experienced observers assisted by CAD, reached similar performance compared to

experienced observers. Therefore, CAD appears to be a promising method for implementation into

routine clinical environment for the MR imaging assessment of suspected prostate cancer.

255


12

APPENDIX

Pre-CAD performance

(AUC)

Post-CAD performance

(AUC) Less-experienced observers

PZ - Reader 1 - Reader 2 - Reader 3 - Reader 4 - Reader 5 - Reader 6

TZ - Reader 1 - Reader 2 - Reader 3 - Reader 4 - Reader 5 - Reader 6

0.83 0.84 0.84 0.87 0.88 0.87

0.73 0.72 0.65 0.65 0.80 0.74

0.95 0.95 0.93 0.95 0.95 0.94

0.81 0.77 0.80 0.77 0.85 0.76

Experienced observers

PZ - Reader 7 - Reader 8 - Reader 9 - Reader 10

TZ - Reader 7 - Reader 8 - Reader 9 - Reader 10

0.92 0.89 0.91 0.91

0.89 0.87 0.80 0.71

0.95 0.94 0.96 0.91

0.92 0.88 0.88 0.69

Table 3. Summary of the individual performances pre- and post-CAD for all readers.

256


12

REFERENCES


2. Kitajima K, Kaji Y, Fukabori Y, Yoshida K, Suganuma N, Sugimura K. Prostate cancer detection with 3 T MRI: comparison of diffusion-weighted imaging and dynamic contrast-enhanced MRI in combination with T2-weighted imaging. J.Magn Reson.Imaging 2010 Mar;31(3):625-31.

3. Puech P, Potiron E, Lemaitre L, Leroy X, Haber GP, Crouzet S, Kamoi K, Villers A. Dynamic contrast-enhanced-magnetic resonance imaging evaluation of intraprostatic prostate cancer: correlation with radical prostatectomy specimens. Urology 2009 Nov;74(5):1094-9.


5. Hittmair K, Gomiscek G, Langenberger K, Recht M, Imhof H, Kramer J. Method for the quantitative assessment of contrast agent uptake in dynamic contrast-enhanced MRI. Magn Reson.Med. 1994 May;31(5):567-71.

6. Tofts PS. Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. J.Magn Reson.Imaging 1997 Jan;7(1):91-101.

7. Vos P, Hambrock T, Futterer J, Hulsbergen-van de Kaa CA, Barentsz J.O., Huisman HJ. Effect of calibration on computerized analysis of prostate lesions using quantitative dynamic contrast-enhanced magnetic resonance imaging. In: Maryellen L.Giger NK, editor. 2007.

8. Kovar DA, Lewis M, Karczmar GS. A new method for imaging perfusion and contrast extraction fraction: input functions derived from reference tissues. J.Magn Reson.Imaging 1998 Sep;8(5):1126-34.

9. Vargas HA, Akin O, Franiel T, Mazaheri Y, Zheng J, Moskowitz C, Udo K, Eastham J, Hricak H. Diffusion-weighted Endorectal MR Imaging at 3 T for Prostate Cancer: Tumor Detection and Assessment of Aggressiveness. Radiology 2011 Jun;259(3):775-84.



12. Karssemeijer N, Otten JD, Verbeek AL, Groenewoud JH, de Koning HJ, Hendriks JH, Holland R. Computer-aided detection versus independent double reading of masses on mammograms. Radiology 2003 Apr;227(1):192-200.

13. Ge Z, Sahiner B, Chan HP, Hadjiiski LM, Cascade PN, Bogot N, Kazerooni EA, Wei J, Zhou C. Computer-aided detection of lung nodules: false positive reduction using a 3D gradient field method and 3D ellipsoid fitting. Med.Phys. 2005 Aug;32(8):2443-54.

14. de HB, De Boo DW, Gietema HA, van HF, Mearadji B, Schijf L, van GB, Prokop M, Schaefer-Prokop C. Computer-aided detection of lung cancer on chest radiographs: effect on observer performance. Radiology 2010 Nov;257(2):532-40.

15. Graser A, Kolligs FT, Mang T, Schaefer C, Geisbusch S, Reiser MF, Becker CR. Computer-aided detection in CT colonography: initial clinical experience using a prototype system. Eur.Radiol. 2007 Oct;17(10):2608-15.

16. Baker ME, Bogoni L, Obuchowski NA, Dass C, Kendzierski RM, Remer EM, Einstein DM, Cathier P, Jerebko A, Lakare S, et al. Computer-aided detection of colorectal polyps: can it improve sensitivity of less-experienced readers? Preliminary findings. Radiology 2007 Oct;245(1):140-9.

257


12

17. Szucs-Farkas Z, Patak MA, Yuksel-Hatz S, Ruder T, Vock P. Improved detection of pulmonary nodules on energy-subtracted chest radiographs with a commercial computer-aided diagnosis software: comparison with human observers. Eur.Radiol. 2010 Jun;20(6):1289-96.

18. Chan I, Wells W, III, Mulkern RV, Haker S, Zhang J, Zou KH, Maier SE, Tempany CM. Detection of prostate cancer by integration of line-scan diffusion, T2-mapping and T2-weighted magnetic resonance imaging; a multichannel statistical classifier. Med.Phys. 2003 Sep;30(9):2390-8.

19. Langer DL, van der Kwast TH, Evans AJ, Trachtenberg J, Wilson BC, Haider MA. Prostate cancer detection with multi-parametric MRI: logistic regression analysis of quantitative T2, diffusion-weighted imaging, and dynamic contrast-enhanced MRI. J.Magn Reson.Imaging 2009 Aug;30(2):327-34.

20. Vos P, Hambrock T, Futterer JJ, Hulsbergen-Van de Kaa CA, Barentsz JO, Huisman HJ. Computerized analysis of prostate lesions using quantitative dynamic contrast enhanced magnetic-resonance imaging. 2007. Ref Type: Generic

21. Huisman HJ, Engelbrecht MR, Barentsz JO. Accurate estimation of pharmacokinetic contrast-enhanced dynamic MRI parameters of the prostate. J.Magn Reson.Imaging 2001 Apr;13(4):607-14.

22. Hittmair K, Gomiscek G, Langenberger K, Recht M, Imhof H, Kramer J. Method for the quantitative assessment of contrast agent uptake in dynamic contrast-enhanced MRI. Magn Reson.Med. 1994 May;31(5):567-71.

23. Kovar DA, Lewis M, Karczmar GS. A new method for imaging perfusion and contrast extraction fraction: input functions derived from reference tissues. J.Magn Reson.Imaging 1998 Sep;8(5):1126-34.

24. Tofts PS, Brix G, Buckley DL, Evelhoch JL, Henderson E, Knopp MV, Larsson HB, Lee TY, Mayr NA, Parker GJ, et al. Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J.Magn Reson.Imaging 1999 Sep;10(3):223-32.

25. Vos PC, Hambrock T, Hulsbergen-Van de Kaa CA, Futterer JJ, Barentsz JO, Huisman HJ. Computerized analysis of prostate lesions in the peripheral zone using dynamic contrast enhanced MRI. Med.Phys. 2008 Mar;35(3):888-99.

26. Hambrock T, Somford DM, Huisman HJ, van O, I, Witjes JA, Hulsbergen-Van de Kaa CA, Scheenen T, Barentsz JO. Relationship between apparent diffusion coefficients at 3.0-T MR imaging and gleason grade in peripheral zone prostate cancer. Radiology 2011 May;259(2):453-61.

27. Vos PC, Hambrock T, Hulsbergen-Van de Kaa CA, Futterer JJ, Barentsz JO, Huisman HJ. Computerized analysis of prostate lesions in the peripheral zone using dynamic contrast enhanced MRI. Med.Phys. 2008 Mar;35(3):888-99.

28. Vos PC, Hambrock T, Barenstz JO, Huisman HJ. Computer-assisted analysis of peripheral zone prostate lesions using T2-weighted and dynamic contrast enhanced T1-weighted MRI. Phys.Med.Biol. 2010 Mar 21;55(6):1719-34.

29. Pattern Recognition Letters. Vol. 15, 1994.

30. Obuchowski NA. Determining sample size for ROC studies: what is reasonable for the expected difference in tests' ROC areas? Acad.Radiol. 2003 Nov;10(11):1327-8.


32. Futterer JJ, Heijmink SW, Scheenen TW, Jager GJ, Hulsbergen-Van de Kaa CA, Witjes JA, Barentsz JO. Prostate cancer: local staging at 3-T endorectal MR imaging--early experience. Radiology 2006 Jan;238(1):184-91.


258


12

34. Hambrock T, Somford DM, Hoeks C, Bouwense SA, Huisman H, Yakar D, van O, I, Witjes JA, Futterer JJ, Barentsz JO. Magnetic resonance imaging guided prostate biopsy in men with repeat negative biopsies and increased prostate specific antigen. J.Urol. 2010 Feb;183(2):520-7.


36. Wang L, Mazaheri Y, Zhang J, Ishill NM, Kuroiwa K, Hricak H. Assessment of biologic aggressiveness of prostate cancer: correlation of MR signal intensity with Gleason grade after radical prostatectomy. Radiology 2008 Jan;246(1):168-76.

37. Delongchamps NB, Rouanne M, Flam T, Beuvon F, Liberatore M, Zerbib M, Cornud F. Multiparametric magnetic resonance imaging for the detection and localization of prostate cancer: combination of T2-weighted, dynamic contrast-enhanced and diffusion-weighted imaging. BJU.Int. 2010 Nov 2.



40. Turkbey B, Pinto PA, Mani H, Bernardo M, Pang Y, McKinney YL, Khurana K, Ravizzini GC, Albert PS, Merino MJ, et al. Prostate cancer: value of multiparametric MR imaging at 3 T for detection--histopathologic correlation. Radiology 2010 Apr;255(1):89-99.


42. Engelbrecht MR, Huisman HJ, Laheij RJ, Jager GJ, van Leenders GJ, Hulsbergen-Van de Kaa CA, de la Rosette JJ, Blickman JG, Barentsz JO. Discrimination of prostate cancer from normal peripheral zone and central gland tissue by using dynamic contrast-enhanced MR imaging. Radiology 2003 Oct;229(1):248-54.


44. Kayhan A, Fan X, Oommen J, Oto A. Multi-parametric MR imaging of transition zone prostate cancer: Imaging features, detection and staging. World J.Radiol. 2010 May 28;2(5):180-7.

45. Guo CC, Zuo G, Cao D, Troncoso P, Czerniak BA. Prostate cancer of transition zone origin lacks TMPRSS2-ERG gene fusion. Mod.Pathol. 2009 Apr 24.

46. Noguchi M, Stamey TA, Neal JE, Yemoto CE. An analysis of 148 consecutive transition zone cancers: clinical and histological characteristics. J.Urol. 2000 Jun;163(6):1751-5.


259


12

PART FIVE

DISCUSSION, CONCLUSIONS AND

FUTURE PERSPECTIVES

Discussion, Conclusions and Future Perspectives

T. Hambrock


DISCUSSION

March 2014, Netherlands:

[UROLOGIST] : “Mr. v. S, your PSA levels have been rising over the last 2 years from 2 ng/ml to

7 ng/ml. The wisest next step is to perform an MRI to exclude significant

prostate cancer.

[PATIENT MR. v.S]: “My brother received 12 biopsies of his prostate 3 years ago. Is this still

necessary?”

[UROLOGIST]: “No, if the MRI reveals no significant prostate cancer we can safely observe you

with confidence. When the MRI detects suspicious lesions, targeted biopsies

with a maximum of 4 biopsies will be performed. The MRI with targeted

biopsies would not only provide us with the information, whether you have

cancer or not, but also give an indication of the aggressiveness of the cancer and

assess as to whether there has been spread of tumour outside the prostate or

not”

[PATIENT MR. v.S]: “Amazing, all that information with an MRI”

[UROLOGIST]: “Yes! We will be able to have a much clearer picture of what treatment will be

most required in your particular case.”

The author of this thesis predicts that this hypothetical conversation taking place in the year

2014 (that is if the Maya’s weren’t correct about the end of the world occurring on 21st Dec

2012) could be a reality. However, this will imply that a PARADIGM SHIFT has occurred.

For a PARADIGM SHIFT to occur, many stones need be dislodged. The author is undoubtedly

aware of the fact that the anticipated paradigm shift is a very daunting and verly “bold”. He also

fully acknowledges that this statement will be criticized by many physicians and that the Battle

of Anhiari will still rage in its fiercest moment and much blood, sweat, tears and heated

arguments will still occur. But the end of the Battle of Anhiari is unlikely to occur in the

262

Discussion, Conclusions and Future Perspectives 13

lifetime. It will only change to a more structured organized and precisely orientated battle with

only the true enemy destroyed while collateral damage to the civilians minimized.

From the philosophical conclusion drawn from the current thesis, we can move to a more

scientific discussion of its merits and weaknesses: In the introduction a number of clinical

problems were stated, both faced by clinicians directly dealing with patients and some faced by

radiologists evaluating MR images.

Problems faced by Clinicians:

Clinical Problem 1 : Patients with an elevated/elevating PSA value but the TRUS biopsies remain

negative, are a considerable concern. Does the patient have cancer or not? Should further TRUS

biopsies be performed or not?

Clinical Problem 2 : If MRI is accurate in identifying a tumour location, what effective method is

available to reliably obtain histological proof of this location?

Background

Currently prostate cancer detection is performed by using tools with limited accuracies

(PSA, DRE and TRUS biopsies). Because the specificity of PSA measurement is low, it is often

the case that many unnecessary repeat systematic random biopsies are performed.

Additionally, because an inherent sampling error occurs with systematic biopsies, tumours

which are present can be missed.

Multiparametric MR imaging techniques including anatomical T2-weighted imaging and

functional techniques such as DCE, DWI and MR spectroscopic imaging have been shown in

combination, to have considerable value in prostate cancer detection. Because these MR

techniques have a relatively high specificity in comparison with PSA measurement, they

could prevent unnecessary systematic random biopsies. These techniques in combination

also have a high sensitivity. In a recent evaluation of multiparametric MR imaging at 3T, the

addition of dynamic contrast-enhanced and/or DWI to T2-weighted MR imaging

significantly improved prostate cancer detection sensitivity from 63% to 80% in the

peripheral zone, while maintaining a stable specificity. Multiparametric MR imaging

techniques may also contribute in detection of transition zone prostate cancers (chapter 6).

263


The combined use of DWI, DCE and T2-weighted MR imaging leads to an increased accuracy

in detection of transition zone cancer especially for high grade tumours with a 91%

detection rate vs. 47% for low-grade tumours.

Solution to clinical problem 1 and 2 : Multiparametric MR imaging at 3T incl. T2-weighted

imaging, DWI and DCE-MRI is an accurate technique for the detection of significant tumour in

these patients. Furthermore, the use of an MR guided biopsy device, to perform targeted biopsies

towards tumour suspicious regions, is a useful and accurate technique to make a definite diagnosis.

A considerable number of patients with persistent abnormal PSA values harbor tumour and should

therefore receive further evaluation with MRI.

Discussion

In Chapter 3 of this thesis, the author tested the technique and feasibility of translating

tumor suspicious region maps of the prostate, obtained by multiparametric, anatomical and

functional 3T MRI data, to imaging at a separate MR imaging session for directing MR

guided biopsies. Furthermore, the practicability of MR guided biopsy on a 3T MR scanner

using a 32-channel coil and an MR compatible biopsy device was determined. It was shown

that a basic translation technique can be developed to transfer information from the initial

detection MRI to a subsequent MR biopsy session. It was also shown that the procedure

was feasible to perform in a clinically acceptable time frame. The initial patient cohort

consisted of patients in whom at least two prior negative TRUS biopsies had been

performed, but PSA was still elevated. Using the above developed translation and biopsy

technique a 38% cancer detection yield was determined.

264


.

.

Figure 1. Principal aspects of Chapter 3: Top two images – schematic presentation of the

procedure. Bottom three: Detection – Translation – MR biopsy.

In Chapter 4, the developed technique of MR guided biopsies was performed in a large

clinical cohort of patients with persistently elevated PSA and at least two prior negative

TRUS biopsy sessions. The purpose of this study was to determine the maximum tumour

detection rate using 3T multiparametric MRI and MR-GB in a cohort mentioned above. A

low threshold for evaluating the images was applied with any regions vaguely to strongly

suspicious for cancer eventually biopsied. In a consecutive group of 68 men we were able to

diagnose cancer in 59% of the patients. Furthermore we determined that these tumours

265


were deemed clinically significant in 93% of the cases. Additionally, we established that

tumours missed with serial TRUS biopsies are located in regions not explicitly sampled by

TRUS biopsy schemes. The results were compared to the tumour detection rates at the 2nd

and 3rd TRUS biopsy in a separate cohort. MR guided biopsies significantly outperformed

conventional TRUS biopsy detection rates for different PSA, prostate volume and PSA

density subgroups. MRI should therefore be part of any workup protocol of patients who

are suspected of harboring malignancy but who have successive negative biopsies. Because

of the low numbers of cores needed, MR guided biopsies are an appealing alternative to

procedure such as saturation biopsies.

266


Figure 2. Principal findings of Chapter 4: Tumor detection rates of MR-guided biopsies vs. 2nd

and 3rd TRUS biopsy session sub grouped according to PSA (top left) and PSA density (top right).

Location map of tumour identified with MR-GB.

Clinical problem 3 : After radiation therapy for PCa, diagnosing local recurrence vs. metastatic

disease when the PSA starts rising again, is challenging.

Background

External beam radiotherapy is performed as first line treatment in around 30% of patients

with prostate cancer. As the prostate is not destroyed completely, viable normal prostate

tissue remains which can continue producing PSA. The utilization of PSA to detect

recurrence of radiation therapy is therefore hampered. Based on the current knowledge,

biochemical failure is defined in these patients when three consecutive rises in PSA occur

after reaching a nadir. Furthermore the PSA value should be > 2 ng/ml. This however is

hampered by the fact that biochemical failure does not define if tumour production of PSA is

due to local recurrence or metastases to lymph nodes or the skeleton. The implication of

accurately determining the location of tumour in this scenario is that salvage therapy (cryo-,

laser-, HIFU-therapy or salvage radical prostatectomy can be offered for patients with local

recurrence only.

Due to radiation effects, the normal prostate undergoes shrinkage, fibrosis and normal

glandular function ceases, all factors which make imaging using conventional MR imaging

problematic. On T2-weighted images, the whole prostate exhibits low signal intensity.

Similarly on ADC maps, a high diffusion restriction occurs in normal tissue (due to loss of

glandular luminal spaces and fibrosis) and on spectroscopic imaging, the normal citrate

production is significantly reduced. Initial reports indicated that DCE-MRI and MRSI may

have a potential to detect local recurrence.

Solution to clinical problem 3: 3T DCE MRI followed by MR guided biopsies is a feasible and very

useful technique for diagnosing local recurrence of prostate cancer following external beam

radiotherapy.

267


Discussion

In Chapter 5 MR imaging at 3T with DCE-MRI was performed in a group of 24 patients with

a PSA biochemical failure following external beam radiotherapy. Initial MRI or bone

scintigraphy detected metastatic disease in 4 patients. In the remaining patients DCE-MRI

was performed and in all patients focal abnormalities on DCE-MRI were identified. All these

focal regions were subsequently biopsied using MR-guided biopsies. The conclusion drawn

from this chapter was that a definite diagnosis of local recurrence was made in 75%

(15/20) of these patients. Only 3 biopsy cores were taken and the median MR biopsy time

was 30 min. Overall, the positive predictive value of focal DCE-MRI abnormalities in the

post-radiated prostate was 68% with the vast majority of the remaining non-malignant

enhancing regions representing radiation induced reactive atypia in preexisting glandular

tissue (with or without some inflammation). Furthermore, the strict criteria of biochemical

recurrence especially the absolute PSA value did not appear to hold true in all cases. Large

volume Gleason 5+5=10 cancer was detected in a patient with a single PSA rise from 0 to 0.4

ng/ml. Thus in patients with elevated PSA following external radiotherapy, DCE MRI can

play an important role in diagnosing local recurrence and therefore guide therapy for

systemic vs. local therapy.

Figure 3. Principal findings of Chapter 5: Tumor detection with MR-guided biopsy of abnormal

region on DCE-MRI (but normal on T2-weighted imaging) in a patient with biochemical

recurrence.

268


Clinical problem 4 : Pretreatment identification of prostate cancer aggressiveness is crucial for

management and prognostication. The current methods to determine this are inaccurate. What in

vivo methods are available to reliably predict the nature of the PCa?

Background

The histologically determined Gleason score of prostate cancer is probably the most

important determinant of its biological activity and aggressiveness. A vast body of

literature has established Gleason score as one of the most important markers in predicting

disease outcome in prostate cancer. In fact, this grading scheme has now become so

important that it is often used as an integral piece of information in both management and

treatment stratification of patients with prostate cancer before and after definitive therapy.

Pre-treatment knowledge of true Gleason score would be an important advance, but

currently, such information remains elusive. This is explained by the fact that biopsy

determination of Gleason score often undergrades and therefore gives a poor reflection of

true Gleason score, determined at prostatectomy. Partin tables and risk stratification

schemes that incorporate information from biopsy Gleason scores into decision making are

therefore rendered less accurate and less reliable. Therefore, a definite need for a more

accurate and non-invasive methods exists to improve the accuracy of determination of true

pretreatment Gleason score.

Diffusion Weighted MR imaging measures the diffusivity of water molecules in tissue and

quantifies this random Brownian motion property of water molecules (diffusion) in tissue.

The degree of restriction to water diffusion in biologic tissue is inversely correlated to

tissue cellularity and the integrity of cell membranes. The principal microscopic differences

that are evident for low Gleason grade tumours compared to higher grades, is that the size

of the ductal lumina decrease (the diffusion space reduces) and the cellular density is

markedly increased as the Gleason score increase.

Proton magnetic resonance spectroscopic imaging provides spatial mapping of the tissue

levels of the metabolites citrate, choline and creatine in the whole prostate gland. Prostate

cancer tissue is characterized by lower citrate levels and/or higher choline levels compared

to normal tissue resulting in the ratio of choline and creatine to citrate (Cho+Cr/Cit) as a

269


marker for prostate cancer. The choline/creatine (Cho/Cr) ratio is also of note since choline

increases in malignant tissue due to altered phospholipid metabolism. Furthermore

increased choline to creatine ratios have been identified in high-resolution ex vivo Magic-

Angle-Spinning NMR of prostate biopsies which showed a significant correlation of the

Gleason score to the Cho/Cr ratio, among other ratios.

Solution to clinical problem 4 : ADC values determined from DWI MR imaging and spectroscopic

imaging at 3T both represent useful biomarkers for prostate cancer aggressiveness.

Discussion

In Chapter 7 the author evaluated the hypothesis that the increase in cellular density

associated with increasing tumour Gleason grades should be reflected by an increasing

water movement restriction and therefore decreasing ADC values. In 51 patients who

received 3T DWI MRI prior to radical prostatectomy a separate step-section by ADC slice

analysis of all peripheral zone tumour were made. For each slice on pathology and

corresponding ADC slice, a meticulous analysis of the proportion of Gleason grade

compositions and the matching ADC values were determined. The results from this analysis

showed that the ADC values of prostate cancer in the peripheral zone inversely relate to

prostate cancer Gleason grades, with low-, intermediate- and high-grade tumors showing

significant differences in ADC values. Furthermore using the median ADC values of the most

aggressive tumor regions a high discriminatory accuracy is achieved for discerning low-

grade (Gleason grade 2/3) from combined intermediate- and high-grade cancers (Gleason

grade 4/5) with an AUC of 0.90. Non-invasive prediction of peripheral zone Gleason grades

may improve patient management by more accurate risk-stratification.

As tumours originating in the transition zone are known to have different behavioral

characteristics, different genetic mutations and originate in a different background of

“normal” tissue, these were excluded from the primary analysis. However ADC values of

transition zone tumours were analyzed (but not presented in this thesis) and no correlation

between ADC values and Gleason grades observed. For still incompletely understood

reasons (probably a decrease in stroma), low grade transition zone tumors especial Gleason

270


2+3=5 tumours often show significant restriction on ADC making a discrimination with

more aggressive tumours using this technique unsuitable.

Figure 4. Principal findings of Chapter 7: The relationship between tumor Gleason grades and

ADC values from 3T DWI in the peripheral zone.

In Chapter 9 a further potentially MR biomarker was evaluated to obtain insight as to its

usefulness for the assessment of aggression. The study was performed to validate the

performance of MR spectroscopic imaging of the prostate at 3T to assess tumour

aggressiveness, based on the choline plus creatine to citrate ratio (Cho+Cr/Cit) and choline

to creatine rato (Cho/Cr), using the Gleason score of the radical prostatectomy (RP)

specimen as the gold standard. We assessed the tumour aggressiveness differentiation by

the AUC of the ROC curves and this gave similar results when using either the Cho+Cr/Cit

(0.70) or the Cho/Cr (0.74) ratio. The performance of combining both ratios was better

(0.78). A significant correlation was found between the maximum Cho+Cr/Cit ratio and the

aggressiveness classes. The comparison of the medians of the three aggressiveness classes

revealed a significant difference between the low and high grade tumours. The maximum

Cho/Cr ratio also correlated significantly with the aggressiveness classes and the median

Cho/Cr of the low grade tumours was significantly different from the high grade tumours. A

Cho/Cr adaptation level of 2.3 for the standardized threshold approach gave the highest

AUCs when discriminating high- from low- and intermediate-grade tumours (AUC=0.73)

and low- from the combined high- and intermediate-grade tumours (AUC=0.78).

271


MR spectroscopic imaging for the evaluation of aggressiveness reflects different aspects of

tumour biology to DWI. However, some features of a tumour which make them amenable

for DWI evaluation, also reflect changes observed in the spectroscopic analysis. The

principal metabolites in spectroscopic imaging of the prostate are choline, creatine and

citrate. The later is part of the normal excretory production of prostate epithelial cells and

is mainly present in the intraluminal space of the ducts and serves as energy supply for the

sperm. The relative ductal size in normal prostate tissue is large, being reflected by a high

citrate signal on MR. Reduction of the citrate levels in tumour relates to two changes that

occur. The less differentiated the malignant epithelial cells become, the less of the normal

exocrinic function remains and the production of citrate decreases or is used for de novo

lipid synthesis for tumour growth Furthermore, as discussed previously, the size of the

luminal spaces decrease as the tumour progresses from Gleason grade 3 to 5. Therefore the

primary compartment where citrate is located also decreases independent of the reduction

in the production thereof. The other two metabolites, choline and creatine are

intracellularly located in the epithelial cells. The increased signal from choline in malignant

tissue reflects the turnover of cell membrane synthesis. Cellular density increases with

increasing Gleason grade. Therefore the same effects that make ADC a useful marker of

cellularity and Gleason grade estimation, also partially explain the changes observed in

spectroscopic imaging.

A standardized threshold approach was developed to make the evaluation of spectroscopic

data in a clinical setting more robust and improve the inter-reader agreement of the

evaluation. This approach also appeared to offer an equally good assessment of

aggressiveness compared to a more quantitative evaluation of the absolute ratios of

different metabolites. For the above mentioned study, spectroscopic data was pooled both

for transition zone and peripheral zone tumours because of the relative small sample size

especially for transition zone. Therefore, a potential value for the transition zone (TZ) vs.

peripheral zone (PZ) tumours in aggressiveness assessment could not be assessed.

In a subsequent study (not included in this thesis) which has recently been accepted for

publication in Radiology, the value of combined DWI and MRS for TZ vs. PZ aggressiveness

assessment was evaluated. The principal conclusion drawn from that study was that the

combination of both techniques did not improve overall aggressiveness prediction.

272


However, it did determine that DWI was the best modality for PZ tumours grade assessment

while MRS, the most suited for the TZ. Quantitative DCE-MRI and quantitative T2 would

constitute further interesting research topics on their value in assessment of prostate

cancer aggressiveness.

Figure 5. Principal findings of Chapter 9: Metabolic rations and malignancy rating score using

MRS for differentiation of aggressiveness groups of prostate cancer (combined PZ and TZ).

Clinical problem 5 : When a patient is diagnosed with PCa Gleason Score 3+3=6 on biopsy, is

there a method available to reliably aid in differentiating those patients where biopsies represent

an undergrading (and therefore need more radical therapy) from those where it is a correct

prediction (and therefore may be managed more conservatively)?

Background

Radical prostatectomy and external beam radiotherapy have been the primary treatment

methods for many patients diagnosed with prostate cancer for a considerable time. With

the increasing awareness that many tumours do not progress or cause disease related

morbidity and mortality, a more conservative approach which includes active surveillance

has been advocated for these tumours. Yet, the current methods to reliable determine

which patients need radical therapy from those where a conservative approach is desirable

remains elusive. As mentioned in a previous discussion, the Gleason grade and therefore

Gleason Score remains one of the most crucial components for this assessment. In the

current

being less than 0.5 cc in volume are classified as being indolent. It however needs to be

273


emphasized that future dedifferentiation into a more malignant phenotype cannot be

assessed as yet. The grey area of larger > 0.5 cc, Gleason Score

more uncertain und no clear consensus on the management of these tumours exist.

of a

biopsy core with tumour < 33% and PSA levels < 10 ng/ml are currently used to predict

patients that harbor indolent tumour. As PSA levels are related to prostate volume, a

correction has also been implemented whereby a PSA density of < 0.2 ng/ml/cc is also

included in the above mentioned criteria. The principal shortcoming of the above

mentioned strategy is that tumours are often missed and located in regions, sampled

inadequately with systematic biopsy (as described in Chapter 4). It is also known from

literature that undergrading of biopsy schemes is substantial (up to 40%). Therefore any

information obtained from biopsy cores may result in non-optimal management of the

patient. Especially patients who harbor aggressive tumours (but have false low Gleason

score 3+3=6 on biopsy) need additional assessment for the presence of extracapsular

extension or metastatic disease. It is therefore of paramount importance to determine if

biopsy findings reflect the true state prior to any further treatment.

Solution to problem 5: 3T DWI MR imaging is a very valuable technique for accurately

identifying patients in whom biopsies represent an underestimation of prostate cancer

aggressiveness.

Discussion

In Chapter 8, the initial experience of patients with a biopsy Gleason Score

evaluated. To test the hypothesis from previous knowledge that DWI is useful biomarker

for prostate cancer aggressiveness and that it may be of value in indentifying patients with

undergrading, MRI was performed prior to radical prostatectomy. This was a retrospective

study but for testing this hypothesis a definite gold standard (prostatectomy) was needed.

Of the 23 included patients with TRUS biopsy Gleason Score

identified undergrading in 48% (11/23). A further histological finding of considerable

concern was that 82% (9/11) of the patients with undergrading also had extracapsular

extension (Stage pT3). This underlines the considerable problem using biopsy Gleason

grade for risk assessment and prognostication. In the remaining 12 patients with a correct

identification of their Gleason score, none revealed extraprostatic growth. On ADC maps,

274


measurement of tumour ADC were determined and for final evaluation only the ADC slice

per patient where the tumour showed the lowest ADC values was used for further

assessment. The reasoning behind this approach is that if ADC is to be used in a

prospective manner, radiologists will identify the hotspot (of low ADC) and use these values

to infer prediction of the Gleason grades present in the tumour. When using only the most

abnormal slice on ADC and comparing the ADC values for the 12 patients with no

undergrading vs. the 11 patients with undergrading, a significant difference was identified.

For ADC values the AUC of the ROC analysis was 0.88. AUC ROC analysis however does not

determine sensitivity or specificity. For cut-off purposes the desired sensitivity or

specificity level need to be identified. In this clinical setting a high sensitivity is most

desirable. This chapter therefore shows that DWI can indeed be of clinical value in patients

emphasizes that MR imaging SHOULD be part of any

patient with a diagnosis of prostate cancer.

Figure 6. Principal findings of Chapter 8: ADC values of tumour in patients with no-

undergrading vs. undergrading of their TRUS

Clinical problem 6 : Transrectal ultrasound guided biopsies only reflect the true aggressiveness

i.e. Gleason grade in about 60% of patients. Are there any methods to improve the tumour

aggressiveness representativeness in biopsy samples on which further management decisions can

be based?

275


Background

As discussed previously, undergrading of biopsy determined Gleason Score is a

considerable problem faced in diagnosis and management of prostate cancer. Methods to

improve this have included increasing the number of biopsies, adjusting the biopsies

according to prostate volume or even performing saturation biopsies. Of paramount

importance in evaluating the current literature on the concordance between biopsy Gleason

Score and radical prostatectomy Gleason score, is the inherent prevalence of different

Gleason Score in the series evaluated. Overgrading (a higher Gleason Score found in biopsy

compared to radical prostatectomy), is of lesser importance as this usually constitutes less

than 10% of the discordance in most series. In many reports on improved overall

concordance rates using different biopsy schemes, a meticulous analysis of the data actually

reveals a high proportion of Gleason Score 3+3=6 tumours in prostatectomy. As these are

not undergraded on biopsy, the overall concordance rates give a false impression of the

exact undergrading of the higher Gleason grades, which were shown to still range between

30-40%. Grey-scale transrectal ultrasound, power Doppler ultrasound and contrast-

enhanced ultrasound are inherently insensitive in detecting and localizing tumour although

improvements have been found with the latter two techniques.

Solution to clinical problem 6: MR guided biopsies targeted towards the most abnormal regions

on DWI MR imaging represent a substantially improved prospective method for assessment of true

tumour aggressiveness.

Discussion

The findings and the study shown in Chapter 10 probably represent the crown of the

current thesis as it incorporates findings of multiple different studies into one single

conclusion. This prospective study incorporates the high diagnostic accuracy of

multiparametric, T2-w, DWI and DCE-MRI at 3T for tumour detection, the value of DWI in

determining the most aggressive component in the tumour, as well as MR guided biopsies

not only for diagnosis of malignancy but also for representing a much improved method to

obtain a reliable estimation of the true aggressiveness with a minimum of just 3 biopsy

cores. This was shown in a cohort of 34 patients with a prior negative TRUS biopsy and

276


elevated PSA who received MR-DWI guided biopsies and a comparison cohort of 64 patients

in whom a tumour diagnosis was established using a 10-core TRUS biopsy scheme. In both,

radical prostatectomy served as gold standard to determine to what degree the highest

Gleason grade (HGG) identified in the respective biopsy techniques matched the highest

Gleason grades present in prostatectomy. MR guided biopsies targeted towards the most

abnormal regions on DWI were shown to have a vastly superior accuracy for determining

the true HGG compared to TRUS. When patients findings were sub grouped into patients

having a HGG of 2/3 vs. 4/5, MR guided biopsies had a 95% accuracy in defining the true

HGG group in prostatectomy, compared to 54% for routine 10-core TRUS biopsies.

The presence of Gleason grade 5 components in a tumour infers a particularly bad

prognosis for the patient. Identification of this component at biopsy is therefore considered

essential. When the overall accuracy rate for this component in biopsy-prostatectomy is

evaluated, conventional TRUS biopsies only identify 30% of grade 5 components, compared

to 73% for MR guided biopsy. A substantial undergrading in biopsy would occur if the HGG

on biopsy was 3 while in prostatectomy a HGG of 5 was identified. This occurred 57% of

patients when conventional TRUS biopsy was performed and did not occur in any patients

using MR guided biopsies. In the 27% of MR patients where the biopsy did not represent a

HGG of 5 but the prostatectomy did, the HGG was 4 on biopsy, representing a more

favorable underestimation.

Figure 7. Principal findings of Chapter 10. Performance of MR-GB and 10-core TRUS

biopsy to predict the true highest Gleason grade (HGG) in prostatectomy.

277


Problems faced by Radiologists:

Clinical problem 7 : No prostate looks alike. The transition zone is an especially chaotic region.

What multiparametric MR imaging features and techniques are available and what is the

diagnostic accuracy for these techniques at 3T.

Background

The transition zone (TZ) exhibits age related changes in prostate anatomy. Under the

influence of dihydrotestosterone, the TZ undergoes hyperplastic changes with various

degrees of glandular hyperplasia and/or fibromuscular hyperplasia giving the TZ of the

aged man (usually > 50 years) a typical heterogeneous appearance. Adenocarcinomas

developing within the TZ often have different behavioral characteristics being more often

confined to the gland, are associated with higher PSA levels and often tend to be larger at

diagnosis compared to PZ tumours. Furthermore it is also known that other types of genetic

mutations may be present in these tumours. Nevertheless, they still cause considerable

diagnostic challenges but they are important to find as Chapter 4 indicated that the

majority of tumours in patients with persistent elevated/elevating PSA levels and negative

TRUS biopsies are located in the TZ. From a histopathological point of view, it is also

important to consider the fact that the growth pattern of most prostate tumours does not

consist of a mere spherical enlargement of a tumour mass but represent a mixture of tumor

growing in-between normal pre-existing ducts and stroma. Apart from the more sheet like

growth pattern of extremely high-grade components (Gleason grade 5), the visibility of

tumours on any imaging modality for that sake, is influenced by the relative contribution of

normal and intermixed tumour tissue within a particular region. The peripheral zone,

containing numerous large ducts filled with fluid, usually has a high signal intensity on T2-

weighted imaging. Identification of tumour is therefore more readily possible as lower-

signal intensity regions of tumour relative to the high signal intensity surrounding it is

present. For the TZ, this is much more difficult as the fibromuscular regions of BPH tissue

in the TZ also show low signal intensity on T2-w imaging as well as water movement

restriction on DWI. Furthermore, the TZ is a highly vascularized region and enhancement

patterns in “normal” TZ tissue are similar to peripheral zone tumours. This makes the

278


evaluation of the TZ for the presence of tumour much more challenging. Identifying tumour

in the TZ is best summoned up by what a dear colleague, Dr. Joe Bush from Chattanooga,

Tennessee says about finding tumour there:

TZ is identified by observing disorganized chaos within a region of organized

Furthermore one needs to have an artistic approach as well, as hobby artist, Prof. Barentsz

appreciates the “erased charcoal sign” on T2-w, as a strong indicator of TZ tumour. This

refers to the visual effect of smearing fine charcoal on a painting or artwork.

This challenge has long been a reason why radiologists not familiar with prostate MR

imaging have been reluctant to become more involved in prostate imaging.

Solution to clinical problem 7 : Multiparametric 3T MR imaging is a very accurate technique to

detect and localize clinically significant aggressive transition zone cancer.

Discussion

Contrary to common belief, detection of prostate cancer in the TZ is possible with a high

diagnostic accuracy using multiparametric MR imaging. In Chapter 6 we evaluated the

performance of detecting and localizing TZ cancer using 4 experienced prostate radiologists.

It is of crucial importance to consider the prevalence of low-grade (Gleason grade 2/3) vs.

high-grade (Gleason 4/5) tumours in any analysis of the performance in detecting,

localizing, staging as well as the concordance rate between biopsy and prostatectomy. As

discussed in previous chapters, a high prevalence in a cohort of low-grade tumours for

example, will result in a good concordance between biopsy Gleason grade and

prostatectomy Gleason grade. Furthermore it will result in a good staging accuracy (low

prevalence of pT3 disease) on MR imaging but will result in poor detection and localization

accuracies. This is probably one of the most important reason why results differ so much in

literature for the accuracies of biopsy techniques and MR for localization. It is therefore

important to consider the specific value of MRI for low-grade disease vs. high-grade disease,

similar to Chapter 10, instead of reporting on the overall value.

279


From the study reported in Chapter 6 a number of important points came to light. Firstly,

high-resolution T2-w imaging remains the cornerstone for evaluation of any prostate. To

the trained eye, features indicative of tumour in the TZ can sufficiently serve to allow

detection (86%) of high-grade tumours. The addition of functional imaging DCE and DWI,

does improve the overall detection rate, but does so only moderately (91%). On the

contrary, low-grade tumors are notoriously undetectable using T2-w imaging alone (24%).

The addition of functional imaging however improves this detection to 47%.

Two specific points need further explanation. Detection vs. localization. In the study

reported in Chapter 6, detection refers to: does the patient have a TZ tumour – yes/no

while localization refers to the exact mapping of tumour in the TZ in 6 different regions.

While the first is sufficient to determine if a patient has tumour and where to direct the

needle for biopsy purposes, the localization is needed for optimal surgical planning,

planning of intensity modulated radiotherapy (IMRT) and for low-grade tumours, also the

planning of brachytherapy, cryotherapy or laser therapy. The study evaluated both.

Although the addition of functional imaging only slightly, but not significantly, improved the

detection accuracy of high-grade tumours, it did however improve the overall localization

accuracy from an AUC of 0.91 to 0.94 (p=0.01). For low-grade tumours, the detection rates

increased substantially from 24% to 47% (p=0.02) while the overall localization improved,

however not significantly, from an AUC of 0.56 to 0.64.

Not detecting low-grade tumours and therefore not overdiagnosing indolent (clinically

insignificant) tumours would according to current knowledge and understanding, be ideal.

-grade tumours

should be ideal candidates for active surveillance. A finding of the above mentioned study

however does put a question mark on the validity of these criteria for TZ tumours. Our

study did confirm prior reports that TZ tumours tend to be large. Of note was that the

median tumour volume of low-grade tumours in our study was 3.5 cm3 (range 0.5 – 22 cm3)

compared to 7.4 cm3 (0.5 -15.7 cm3) for high-grade tumours. Therefore all the TZ tumours

in our study would therefore be considered as not being indolent (i.e. not suitable for active

surveillance). The truly massive sizes of some of the low-grade tumours would evoke the

question on what the management should be of these tumours? Although the absolute

number of patients with TZ tumours in our study was rather small, it is important to

280


consider that 22% (2/9) of the patients with low-grade TZ revealed extracapsular extension

(one pT3a and one pT4) while 72% (9/11) of the high-grade tumours where of stage pT3/4.

Multiparametric imaging misses 50% of low-grade tumours. Based on the above findings,

the author is of the opinion that the Gleason Score should not be the most important feature

to consider for the best management of TZ tumours. Future research should definitely

focus on the behavioral characteristics especially as some of these “benign” tumours were

associated with locally advanced disease (extracapsular extension).

Figure 8. Principal findings of Chapter 6. Detection rates (top) and localization AUC for

multiparametric MRI for low-grade and high-grade TZ tumours.

Clinical problem 8 :

prostate tissue in a dif

quantitative measurements and our assessment of what is malignant?

281


Background

What do we usually mean by 'normal'?

“Referring to humans, we mean that he or she is like everyone else, behaves as most people

behave, and stays within current conventions. But we have a problem, because now-a-days

the idea of what is normal changes from one decade to another. Fortunately the popular

press keeps us up-to-date by surveys and advice columns. Normality for humans in fact has

nothing to do with statistics. It refers to a norm, a model of perfection, an example to be

followed. It indicates what we should be. Normality is therefore something to strive for,

something at which to aim, it is not what most people do. It is what they would do if they

lived up to their human potential.”

“Every normal person, in fact, is only normal on the average. His ego approximates to that of

the psychotic in some part or other and to a greater or lesser extent.” ( )

“To study the abnormal is the best way of understanding the normal. “

“Normal is in the eye of the beholder. Normal is nothing more than a cycle on a washing

machine. “

Based on the above mentioned opinions about normality it is clear that from a philosophical

point it is quite bothersome to define what is normal. From a scientific point of view

especially looking at physical characteristics of the prostate one can however get closer to

defining normality. Any pathologist and radiologist dealing with the prostate will attest that

no prostate looks alike needless to say, no tumour looks exactly alike. Based on our

findings in Chapter 7 it was observed that the median ADC values for the “normal”

peripheral zone varied substantially between patients. Tumours in the prostate are not

mass-like, expansile growing lesions but intermixed with various amounts of “normal”

background tissue and develop out of pre-existing glandular tissue. Since normal prostate

PZ tissue fluctuates significantly in ADC value, the ADC values of an aggressive tumour may

show similar fluctuations. If normal PZ and tumour ADC are correlated, considering both

simultaneously, may lead to better estimates of aggressiveness. Inter-patient ADC variation

could therefore affect the discriminative power of ADC both for prostate cancer localization

as well as for the determination of prostate cancer aggressiveness.

282


Solution to problem 8 : Peripheral zone ADC show a significant inter-patient variation, which has

a significant effect on the prediction of prostate cancer aggressiveness. Correcting this effect on a

per patient basis results in a significant increase in diagnostic accuracy.

Discussion

In Chapter 11 we have evaluated normality and “normalness” of the peripheral zone. In a

first set of 10 patients who received 3 separate 3T MR imaging sets at different time points,

the variation of ADC values for a fixed given region of “normal” appearing peripheral zone

was measured. Here we showed that the normal PZ ADC differed significantly between

patients relative to measurement variability (p<0.01). This significant inter-patient

variation in normal peripheral zone ADC values (1.2 – 2.0 x 10-3 mm2/s), is something we

could not solely attribute to measurement variability (average measurement SD 0.068 ±

0.027 x 10-3 mm2 /s). We therefore hypothesize that the inter-patient variations arise from

natural variations in prostate physiology.

In the same chapter another cohort of 51 patients were used and we have determined that

adding normal PZ ADC values to the tumour ADC values (using linear regression analysis),

resulted in a significantly improved prediction of cancer aggressiveness (p=0.01). This

suggests that tumour ADC values should not be considered absolute but that these values

are influenced by “background” variation of normal PZ tissue composition. Adding the

information of normal PZ ADC, increased the AUC from 0.91 to 0.96 (p<0.05) in separating

low-grade from combined intermediate- and high-grade prostate cancer.

Figure 9. Clinical useful nomogram for assessment PZ cancer aggressiveness.

283


Clinical problem 9 : Prostate multiparametric MR imaging should be left to the experts. The

prostate is too complex, too many imaging modalities are needed and tumours are very

heterogeneous. Is there any help for the non-expert?

Background

Numerous publications have shown the necessity of performing multiparametric MR

imaging to improve the tumour localization capabilities. As discussed previously, not only

a natural heterogeneity of the normal prostate anatomy is present, also tumours reveal

heterogeneous features. Furthermore benign conditions like prostatitis, high-grade PIN,

atrophy, atypical adenomatous hyperplasia, BPH and fibrosis can all reveal features that

mimic tumour on all MR imaging modalities. Currently, quantitative cut-off values on MRI

are not used routinely to differentiate tumour from benign or normal changes in the

prostate. Images are evaluated qualitatively that is, a relative abnormality identified in

relation to a normal surrounding tissue. T2-w imaging is purely qualitative and further

affected by coil profile changes. Therefore signal intensity as such on T2-w imaging cannot

reliably be used to differentiate tumour from non-tumour tissue. Quantitative T2-w

imaging is possible but is of excessive imaging duration and of low spatial resolution. ADC

maps obtained from DWI imaging are quantitative but from literature no clear cut-off values

have reliably been reproduced as b-factor differences, TE differences and imaging at 1.5T vs.

3 T resulting in different ranges are being reported. Absolute values for DCE-MRI are also

difficult to determine as correct assessment of the arterial input function as well as the

normal heterogeneity of microvascular features in tissue, affecting the absolute values. All

these factors have lead to a great reluctance on part of radiologists to evaluated MR imaging

of the prostate. Additionally substantial reader variability especially for less-experienced

observers has been reported. Therefore techniques that can aid radiologists in improving

their evaluation of the prostate and reducing the inter-reader variability are in great need.

Solution to problem 9 : The addition of computer aided diagnosis (which incorporates

quantitative ADC values and quantitative pharmacokinetic DCE values) for evaluation of prostate

cancer suspicious regions on 3T MRI, significantly improves the discriminating performance for less

experienced observers for both the peripheral and transition zone.

284


Discussion

In Chapter 12 we report on the development of a computer aided diagnosis (CAD)

technique for prostate cancer evaluation in the PZ and TZ. A novel technique of

determining quantitative pharmacokinetic DCE parameters using the normal peripheral

zone for normalization and calibration was used. This has the advantage that both the

arterial input for the prostate itself as well as the inherent per patient variability in normal

microvascular features, are compensated for. Combining these features with ADC values, a

computer was trained to discriminate tumour lesions and benign but tumour suspicious

lesions (on MRI) both in the PZ and TZ. The results presented in Chapter 12 indicate that

such a CAD technique can be developed and revealed a stand-alone AUC ROC performance

of 0.92 for the PZ and 0.87 for the TZ.

Using such a CAD technique to aid radiologists in improving their evaluation of

multiparametric MRI of the prostate was then further evaluated. In this study 6 less-

experienced and 4 experienced prostate radiologists evaluated 206 different prostate

regions in 34 patients using 3T multiparametric MRI. For each region, a tumour likelihood

had to be given before and after a CAD tumour likelihood was shown to the reader. In this

study it was firstly shown that the stand-alone performance of CAD is similar to prostate

experienced radiologists (overall AUC of CAD 0.90 vs. AUC of 0.88 for experienced

radiologists). Secondly, the addition of CAD significantly improved lesion discriminating

performance for less experienced radiologists both for the peripheral zone (pre-CAD AUC

0.86 to post-CAD AUC of 0.95) as well as the transition zone (pre-CAD AUC 0.72, post-CAD

0.79). A final important finding of this study indicated that after CAD, less experienced

radiologists reached similar performances as experienced radiologists in evaluating the

prostate. It therefore appears promising that in due time, by using CAD, more radiologists

would be willing and able to use MRI for prostate analysis. This will therefore result in

better support for urologists and radiotherapists in finding solutions and answers to most

of their eminent problems in the diagnosis and management of prostate cancer.

285


Figure 10. Principal findings of Chapter 12. AUC for less-experienced and experienced

observers grouped according to overall, PZ and TZ performances.

286


LIMITATIONS

The vast majority of any PhD research work is devoted to the positive and “ground-breaking”

discoveries presented in one’s thesis. This is an inherent human error. Sir Bertrand Russell has

strikingly identified this by stating:

work is terribly

The author cannot but stress that this thesis is but a drop of water in the vast ocean of

knowledge. A mere drop and nothing more. The information presented in this thesis, might

mean life or death, hope or despair, meaning or nihility to many patients, but in the vastness of

space, time and things to come, it will not have any substantial influence whatsoever.

Having a state-of-the art 3T MRI scanner, the availability of MR guided biopsy equipment, in-

house dedicated analytical software, motivated urologists, radiologists, pathologists, physicists

and off-course the immense importance of funding for research, is an exceptional combination

that is (and for the time being) of particular scarcity. Therefore, the value of 3T MRI in the

diagnosis and management of prostate cancer will for a number of years to come be reserved to

a few centers of excellence. All the links in the above mentioned chain need be in place to fully

show its true value.

Another important consideration is to acknowledge the fact that making mistakes and being

uncertain in evaluating the prostate will definitely happen on some occasions - even to the most

hardened prostate radiologists. MRI is an exceptionally good technique, yet it is not perfect. In

the pressure of “normal” routine radiological work, outside an experimental research setting,

results will always be worse than the most optimistic research. That is a known fact. Also, it

does not matter how many patients one has seen, how many images one has viewed or number

of publications written, in medicine exceptions are the rule. The author can truly attest that

tumours in some patients can be seen clearly in others not at all. The one is regarded as highly

malignant, while in reality it is benign.

Although MR imaging even at 1.5T is generally available, MR guided biopsy equipment is not

(yet). Despite being commercially available, investing in such equipment requires a dedicated

team of urologists and radiologists. To the opinion, MR imaging should only performed

287


by experienced and trained radiologists in this field, who are expected to deal with prostate MR

imaging on a daily basis. In this thesis a number of publications dealt with MR guided biopsies

(Chapter 3, 4, 5, 10). One might argue that with dedicated fusion software, biopsies might be

performed under ultrasound guidance. As for now, at least the gold standard of biopsying areas

abnormal on MRI, has been determined. If in future, MR guided ultrasound biopsies may prove

to be equally accurate and may result in a future financial advantage. Although MR biopsies are

performed within 30 minutes, MR time remains valuable.

Chapters 7, 8 and 9 were chapters dealing with the evaluation of MR imaging and

aggressiveness using radical prostatectomy as gold standard. These studies were retrospective

in nature. It is of paramount importance that such findings should also be validated in a

prospective ideally multi-centre setup. For Chapter 10 this was attempted, as DWI was used to

guide the needle towards the darkest most “aggressive” spot. The results prospectively indicate

that at least DWI allows sampling of tissue being truly representative of the most aggressive

component. Furthermore slight imaging differences exist between different MRI vendors and

more importantly differences exist between 1.5T and 3T. Quantitative values on the one scanner

do not necessarily equate to the same quantitative values on the scanner of a different vendor.

The presented results should therefore also be evaluated between different vendors and for

different field strength. Chapter 11 did reveal that inter-patient variations for PZ ADC exist.

However no satisfactory explanation could be identified to explain the intra-patient temporal

variation. Further research should focus on addressing the potential temporal effect of

ejaculation, amount of bicycle riding etc. as external / internal influences of our absolute

measurements.

Chapter 12 described the use of a CAD system to aid radiologists in improving their

performance. A drawback of the current setup was that tumour suspicious regions were

identified upfront and radiologists were asked to characterize those presented regions only.

This was done to test the potential value of a CAD system. In the manner that our current CAD

system works, a widespread implementation at this stage is not possible. In future, this system

will be changed (work in progress) to make the radiologists themselves, identify the region of

concern and then obtain a second opinion from CAD or alternatively, the CAD should project

tumor probability maps.

288


A final thought on the limitations of the current thesis. Although some important clinical

questions have been answered, a legion more, still exist. For the prostate gland itself, local

staging is another further crucial point especially to aid in decisions on preservation of

neurovascular bundles and to reduce positive surgical margins. The predecessors have

performed initial studies in this regard, but the final word has not been spoken yet. The current

thesis dealt insufficiently with methods to determine aggressiveness of TZ tumours and

hopefully current ongoing work will find some solution for this.

289


FUTURE PERSPECTIVES

Local prostate : Diagnostics

Active surveillance is becoming of increasing importance. There is tremendous value for

MR imaging in identifying patients who are not true candidates for active surveillance as

well as identifying patients over time, when their tumours are differentiating into more

aggressive phenotypes that need active intervention.

Tumor mapping – although information on tumour localization accuracies using MRI are

published, information on the capabilities of MRI based exact mapping of tumour (for

focal therapy), is still lacking and further work is needed prior to full implementation of

prostate sparing therapy.

Currently, many histopathological features underlying the visibility and effect that these

features have on multiparametric imaging have not been studied sufficiently. It is crucial

for any scientist working on developing new techniques for imaging to understand the

pathophysiological processes underlying imaging parameters. The Gleason Score is not

the only marker for aggressiveness and other histological markers of aggressiveness

should also be evaluated and correlated to MRI in future. Novel techniques are rapidly

gaining importance for more reliable predictors of tumour characteristics and behavior.

The most important at this stage appears to be the upcoming techniques in genomics and

proteomics whereby tissue (preferably only a biopsy sample) analyzed, can be

characterized on the presence of pathological mutations and molecular compositions

which may facilitate decision making on management beyond and in addition to the most

widely used Gleason Score. MRI facilitated (either via MR-guided biopsies or MR-TRUS

biopsies) targeting of tissues for these purposes would be a further important future

development.

Screening with MRI – To many a blasphemous statement! The author believes this to

hopefully be the greatest contribution of MRI in the next few years. Having a short (15

min), highly sensitive (for aggressive tumours) imaging protocol, will certainly have its

merit in identifying patients with an elevated PSA who should or should not receive

targeted biopsies. The author hopes to eventually see the day when this paradigm shift

has occurred.

290


MR imaging at > 3 Telsa field strength – Researchers will always find such things

interesting. The value? The future will tell.

Local prostate therapeutics

Focal therapy – many very exciting MR guided focal therapy techniques are being

developed and tested. Of special note are MR compatible cryotherapy, laser therapy and

also high intensity focused ultrasound. These would hopefully establish themselves as

useful techniques for managing small, relative benign tumours or they will serve a

valuable role as salvage therapy after prior brachytherapy or radiotherapy and maybe

also prostatectomy.

Metastatic disease diagnostics

MR lymphography – with the loss of the only proven lymphnode imaging agent Sinerem,

the future management of advanced prostate cancer received a major blow and setback.

Currently however, new lymphotrophic particles are in development and hopefully will

emerge again, like the Phoenix from the ashes, in not too distant future.

Skeletal metastasis – currently 99mTc bone scintigraphy is still heavily relied upon for

exclusion or identification of bone metastases. Unfortunately the sensitivity and

specificity are generally not good. Recent publications have identified that MRI is

superior in identifying skeletal metastasis in prostate cancer. Ongoing research which

includes “whole-body” diffusion weighted imaging might eventually prove to be the

modality of choice for skeletal metastasis assessment and potentially make 99mTc bone

scinitigraphy obsolete, at least for the establishment of the presence of prostate

metastasis or not. The comparative value of MRI vs. 18F (NaF) PET or 11C

(Choline/Acetate) PET in detection of tumour recurrence and metastasis is still an open

undecided battlefield.

Metastatic disease therapeutics

With improved identification of metastatic lymph nodes and bony metastases,

therapeutic modalities including intensity modulated radiotherapy, proton therapy or

even selective surgery may become MR guided and hopefully provide a valuable

contribution in future to reduce morbidity and mortality.

291


FINALE

Following the conclusions made in this thesis, there are but a few obstacles in the way for

clinicians and patients to have an extremely useful weapon in their arsenal against prostate

cancer. Now it is time that treating clinicians should be made aware of the great potential of

MRI, that radiologists learn and become confident and good in the evaluation thereof. Patients

should be comfortable with the knowledge that they are receiving top of the range diagnostics.

Lack of awareness is therefore one of the greatest obstacles.

The cautioned reader would certainly raise the valid question of cost? Our health systems are

already overburdened with medical costs and reductions. The author believes that cost will

always be an issue. There will never be enough funding to deliver all the groundbreaking

discoveries/techniques and treatments to patients. That is unfortunately the world we live in.

Cost however, should not necessarily be the principal obstacle. In the long run, the improved

early diagnosis and more efficient management will (daunting to say) reduce morbidity and

mortality and be more cost-efficient. The large European trial on using PSA for PCa screening

has already revealed a 20-30% reduction in mortality. Undoubtedly adding MRI on top of this,

hopefully not only mortality will be reduced but also unnecessary morbidity.

Prior studies at 1.5T necessitating the endorectal coil for adequate imaging seemed a major

hurdle for widespread utilization. For accurate determination of minimal extracapsular

extension, even at 3T, the endorectal coil is still needed and remains a valuable tool, despite its

drawbacks. It is however, currently feasible to perform MR imaging without the endorectal coil

incl. high-resolution T2-weighted imaging, DWI and DCE in 20 min duration, making

implementation into routine clinical use (here this is already the case) absolutely feasible.

Imaging is however performed at 3T but it is merely a matter of time that 3T MR imaging will be

the standard MR performed in most hospitals and that 1.5T will become increasingly obsolete.

Like the old English saying: ”Time and tide waiteth for no man!” so does routine MRI for prostate

cancer diagnosis and management – it will come!

Yes We Scan!

292


English Summary - Nederlandse Samenvatting -

T. Hambrock; J. Barentsz

Auguste Rodin The Thinker


English Summary

nce Imaging for the Detection and Aggressiveness


- From Theory to Practice -

SUMMARY STATEMENT OF THESIS

Every man with a DIAGNOSIS of prostate cancer SHOULD receive a Multi-

parametric MRI

Every man with the persistent SUSPICION for having prostate cancer

SHOULD receive a Multi-parametric MRI, and

Every man who needs correct AGRESSIVENESS assessment of his prostate

cancer SHOULD receive an MRI!

Prostate cancer has become a substantial burden to society. It has surpassed lung cancer as the

number one most commonly diagnosed malignancy in men. Furthermore it is the number 2

killer of men due to cancer. The U.S.A. has an annual rate of new prostate cancer (PCa)

diagnoses of around 220.000. This fact adds a substantial burden to doctors, patients and the

society, notably also as a psychological and financial burden.

Despite being a malignancy of predominantly older men, it appears from postmortem studies,

that the prevalence of prostate cancer in 30-40 year olds is high (25%), increasing to 60% at the

age of 60-70 year old,, and to >80% at 80 years. Nonetheless, most men die with prostate

cancer, and only 15% from it. Of further note is that 2/3rd of death occurs in elderly patients

(>75 years) . This translates to the following: of all patients diagnosed with PCa, only 5% (in the

U.S., equating to 11 000 deaths) of PCa deaths are in young men. Although disease specific

mortality for PCa is only 15%, a substantial additional number of patients suffer significant

morbidity i.e. impotence, incontinence, cystitis, proctitis, lymph edema and wound infections

following attempted radical treatment. The immense psychological trauma, like anxiety,

294

English Summary – Nederlandse Samenvatting – 14

uncertainty, worry and often depression, experienced by patients who have increasing PSA

values (thereby suggesting local disease recurrence or metastases) after attempted curative

radical treatment, cannot be denied.

Most prostate cancers are low-aggressive, and do not cause morbidity or even mortalitity. Until

now the determination of the tumour aggression is done by pathologic evaluation of tissue

samples, and expressed in the so called Gleason grade or scores. The correct establishment of

the Gleason grade/score is therefore of utmost importance for prognostication. It is associated

with advanced disease stage, associated with disease recurrence and overall survival. Herein

lies of one of the greatest pitfalls in current diagnosis and management of PCa: The current

random trans-rectal ultrasound guided biopsy techniques performed to make a definite

histological diagnosis of prostate cancer, do not only miss a substantial number of tumours but

also often miss (up to 40%) the most aggressive part of the tumour, resulting not only in an

under diagnosis of cancer but also in an under grading of tumours. Furthermore, following

radical treatment, clinicians often struggle to identify the sources of tumour tissue recurrence

following a post-treatment rise in PSA.

On a routine basis, patients undergo “screening” with blood serum Prostate Specific Antigen

(PSA) quantification. In general, PCa is associated with higher PSA values, yet PSA is non-

specific (specificity of 63%) due to other non-malignant disease processes also increasing blood

values. This poses a particular challenge to physicians. An arbitrarily PSA cut-off value is most

often used to decide on obtaining histological sampling of the prostate. At the cut-off value of 4

ng/ml, the sensitivity for detecting tumour is around 85%. Under this value around 15% of

tumours are missed while of these again, 15% are considered high-grade, representing tumours

that warrant definite treatment. For the values above 4 ng/ml, especially between 4 – 10 ng/ml

a low specificity is found resulting in a substantial amount of patients undergoing unnecessary

invasive diagnostic procedures. For a definite diagnosis of prostate cancer, clinicians most often

utilize transrectal ultrasound guided, random but systematic biopsies techniques, using 10-12

biopsy cores, spaced throughout the prostate gland. Such “brutal and archaic” methods are still

the cornerstone in current day and age. Fortunately, females have chosen the path of

enlightenment much earlier than men. Random multi-core sampling of breast tissue for

detection of malignancy is not part of the 21st century clinical practice or even the 20th century

practice for that matter.

295


The author of this thesis therefore envisioned that a change to the initial management and

diagnosis of prostate cancer had to occur. A PARADIGM SHIFT is the only way forward. This

thesis therefore evaluated the important challenges faced by clinicians and patients and the

author sought to develop new techniques, apply current, recently developed techniques into

routine PCa care and determine the value that these can have on patient management. The aim

was to evaluate how it is possible reduce the number of biopsy needles and to increase its yield

and aggressiveness representation. This thesis was set up and evaluated these aspects from

“Theory to Practice”.

The following summary findings and conclusions can be drawn from this thesis:

In Chapter 3 the principal aim of our study was to test the technique and feasibility of

identifying tumour suspicious region maps obtained by multi-parametric 3T MRI and a 32

channel coil for directing MR guided biopsies. Furthermore, we evaluated the practicability of

MR guided biopsy on a 3T MR scanner using a MR compatible biopsy device. In this chapter we

concluded that multi-parametric MR imaging at 3T using T2-weighted imaging, DWI and DCE-

MRI is an accurate technique for the detection of tumour in these patients. Furthermore the use

of an MR guided biopsy device, to perform targeted biopsies towards tumour suspicious regions,

is a useful and accurate technique to make a definite diagnosis. A considerable number of

patients with persistent abnormal PSA values harbor tumour and should therefore receive

further evaluation with MRI.

In Chapter 4, the technique developed in chapter 3, using MR guided biopsies was performed in

a large clinical cohort of patients with persistently elevated PSA and at least two prior negative

TRUS biopsy sessions. The purpose of this chapter was to determine the absolute tumour

detection rate possible using 3T multi-parametric MRI and MR-GB in this cohort. Furthermore

we aimed to determine whether the detected tumours were clinically significant. In a

consecutive group of 68 men we were able to diagnose cancer in 59% of the patients.

Furthermore we determined that these tumours were deemed clinically significant in 93% of the

cases. Additionally, we established that tumours missed with serial TRUS biopsies are located

in regions not explicitly sampled by TRUS biopsy schemes. MR guided biopsies also significantly

outperformed conventional TRUS biopsy detection rates for different PSA, prostate volume and

PSA density subgroups. We concluded that MRI should therefore be part of any workup protocol

296


of patients who are suspected of harboring malignancy but who have successive negative

biopsies.

In Chapter 5 we evaluated the feasibility of the combination of MR guided biopsy and diagnostic

3T MR imaging, in localization of PCa local recurrence after external beam radiation therapy. We

have shown that the positive predictive value of focal abnormal DCE enhancement for local

recurrence was 75% and concluded that 3T DCE-MRI followed by MR guided biopsies is a

feasible and very useful technique for diagnosing local recurrence of prostate cancer following

external beam radiotherapy.

The detection and localization of transition zone tumours is a particular challenge to

radiologists. Therefore, in Chapter 6 we sought to retrospectively determine Gleason grade 2/3

(low-grade) and Gleason Grade 4/5 (high-grade) transition zone cancer detection and

localization accuracies, using individual and combined multi-parametric MR imaging. Firstly, we

identified that high-resolution T2-w imaging remains the cornerstone for evaluation of any

prostate. For the experienced radiologist, T2-w imaging alone allows detection of 86% of high-

grade tumours. The addition of functional imaging: DCE and DWI, improves the overall

detection rate to 91%. On the contrary, low-grade tumours were found to be notoriously

undetectable using T2-w imaging alone (24%). The addition of functional imaging however

improved this detection to 47%. Our conclusion was that multi-parametric 3T MR imaging is a

very accurate technique to detect and localize clinically significant, aggressive transition zone

cancer.

The importance of accurately and non-invasively assessing the aggressiveness (i.e. Gleason

Score) of a tumour has many important implications in the clinical management of patients. Up

to now this has been elusive, therefore a great need exists for techniques to be developed to

assess this more accurately. In Chapter 7 we evaluated the relationship between ADC values

from 3T DWI-MRI and peripheral zone prostate cancer Gleason grades determined from step-

section specimens after prostatectomy. As the physical principals underlying diffusion weighted

imaging are sensitive to regions with high cellularity, our hypothesis was that this technique

could be useful to quantify and correlate tumour Gleason Score, which is known to be related to

cellularity. We showed that ADC values of prostate cancer in the peripheral zone inversely

related to prostate cancer Gleason grades, with low-, intermediate- and high-grade tumours

showing significant differences in ADC values (p<0.001). Furthermore, using the median ADC

297


values of the most aggressive tumour regions, a high discriminatory accuracy was achieved for

discerning low-grade from combined intermediate- and high-grade cancers (AUC=0.90). We

therefore concluded that ADC values determined from DW-MR imaging at 3T represents a useful

biomarker for prostate cancer aggressiveness in the peripheral zone.

In Chapter 9 an additional potential biomarker for prostate cancer aggressiveness at 3 Tesla

was evaluated. Proton magnetic resonance spectroscopic imaging (MRSI) provides spatial

mapping of the tissue levels of the metabolites citrate, choline and creatine in the whole prostate

gland. Prostate cancer tissue is characterized by lower citrate levels and/or higher choline levels

compared to normal tissue resulting in the ratio of choline and creatine to citrate (Cho+Cr/Cit)

as a marker for prostate cancer. We validated the performance of MR spectroscopic imaging of

the prostate at 3T to assess tumour aggressiveness, based on the choline plus creatine to citrate

ratio (Cho+Cr/Cit) and choline to creatine rato (Cho/Cr), using the Gleason score of the radical

prostatectomy (RP) specimen as the gold standard. The metabolite ratios Cho+Cr/Cit and

Cho/Cr resulted in a tumour aggressiveness differentiation AUC of 0.70 and 0.74 respectively

and a combined AUC performance of 0.78. A significant correlation was found between the

maximum Cho+Cr/Cit ratio and the aggressiveness classes. A standardized threshold approach

was also developed to make the evaluation of spectroscopic data in a clinical setting more robust

and improve the inter-reader agreement of the evaluation. We concluded that 3T 1H-MRSI offers

potential for in vivo non-invasive assessment of prostate cancer aggressiveness.

To evaluate the potential clinical impact of our identified non-invasive aggressiveness

biomarker, DWI-MRI, we performed two different studies. The first study, being retrospectively

performed on a prostatectomy series of patients with prior multi-parametric MRI, is presented

in Chapter 8. An important clinical problem is that patients diagnosed with PCa Gleason Score

3+3=6 on biopsy, often represent an undergrading (and therefore need more radical therapy)

from those where it is a correct prediction (and therefore may be managed more

conservatively). In our study of 23 patients with TRUS biopsy Gleason Score

prostatectomy identified undergrading in 48% (11/23). This furthermore emphasizes and

underlines the considerable problem using biopsy Gleason grade for risk assessment and

prognostication. On ADC maps, measurements of tumour ADC were determined and for final

evaluation only the ADC slice per patient where the tumour showed the lowest ADC values was

used for further assessment. When using only the most abnormal slice on ADC and comparing

298


the ADC values for the 12 patients with no undergrading vs. the 11 patients with undergrading, a

significant difference was identified. For ADC values the AUC of the ROC analysis was 0.88. This

chapter therefore showed that DWI can indeed be of clinical value in patients with Gleason Score

whom biopsies represent an underestimation of prostate cancer aggressiveness. This represents

another important emphasis that MR imaging SHOULD be part of any patient with a diagnosis of

prostate cancer.

The findings and the study shown in Chapter 10 probably represent the crown of the current

thesis as it incorporates findings of multiple different studies into one single conclusion. This

prospective study incorporated the high diagnostic accuracy of multi-parametric, T2-w, DWI and

DCE-MRI at 3T for tumour detection, the value of DWI in determining the most aggressive

component in the tumour, as well as MR guided biopsies not only for diagnosis of malignancy

but also for representing a much improved method to obtain a reliable estimation of the true

aggressiveness with a minimum of just 3 biopsy cores. MR guided biopsies targeted towards the

most abnormal regions on DWI were shown in this chapter to have a vastly superior accuracy

for determining the true highest Gleason grades compared to conventional 10-core transrectal

guided biopsies. We showed that MR guided biopsies had a 95% accuracy to represent the true

highest Gleason grade group (low vs. high) in prostatectomy, compared to 54% for routine 10-

core TRUS biopsies.

Based on our findings in Chapter 7,

peripheral zone varied substantially between patients. Since normal prostate PZ tissue

fluctuates significantly in ADC value, the ADC values of tumour may show similar fluctuations. If

normal peripheral zone and tumour ADC are correlated, we hypothesized that considering both

simultaneously, might lead to better estimates of aggressiveness. Inter-patient ADC variation

could therefore affect the discriminative power of ADC both for prostate cancer localization as

well as for the determination of prostate cancer aggressiveness. In Chapter 11 we have

We measured the intra-patient

variation of . Here we

showed that the normal PZ ADC differed significantly between patients relative to measurement

variability per patient (p<0.01). A significant inter-patient variation in normal peripheral zone

ADC values (1.2 – 2.0 x 10-3 mm2/s) was noted and we attributed this to natural variations in

299


prostate physiology. We then further determined that adding normal PZ ADC values to the

tumour ADC values (using linear regression analysis), resulted in a significantly improved

prediction of cancer aggressiveness (p=0.01). This suggested that tumour ADC values should

not be considered absolute but that these values are influenced by “background” variation of

normal PZ tissue composition. Adding the information of normal PZ ADC, increased the AUC

from 0.91 to 0.96 (p<0.05) in separating low-grade from combined intermediate- and high-grade

prostate cancer. We therefore concluded that peripheral zone ADC shows a significant inter-

patient variation, which had a significant effect on the prediction of prostate cancer

aggressiveness. Correcting this effect on a per patient basis resulted in a significant increase in

diagnostic accuracy.

Chapter 12, being last but not least in the line, represented an important aspect to aid bringing

MR imaging of the prostate into routine clinical use. Despite the apparent “ease” and “value” of

MRI shown in this thesis, evaluation of the multi-parametric images is notoriously challenging

and often a reason why many a radiologist is reluctant and not willing to take up the challenge of

learning prostate MR interpretation. To aid radiologists, especially the unacquainted and

inexperienced ones in improving their confidence and accuracy when reporting on multi-

parametric MRI, we have developed a novel computer aided diagnosis (CAD) system which

utilizes quantitative pharmacokinetic parameters derived from DCE-MRI, in combination with

absolute ADC values from DWI to differentiate tumour from benign but tumour suspicious

regions on MRI. In this chapter we presented the results which indicated that such a CAD

technique could be developed and revealed a stand-alone AUC ROC performance of 0.92 for the

peripheral zone characterization and 0.87 for the transition zone. We then sought to determine

the effect of computer-aided diagnosis (CAD) on less-experienced and experienced observer

performance in differentiating benign and malignant prostate lesions on 3T multi-parametric

MRI (MP-MRI). The addition of CAD significantly improved lesion discriminating performance

for less- experienced radiologists both for the peripheral zone (p<0.001) as well as the transition

zone (p=0.01). After CAD, less-experienced radiologists (AUC=0.91) reached similar

performances as experienced radiologists (AUC=0.93). CAD methods that aid radiologists

especially those less experienced in prostate MRI, may expedite utilization of multi-parametric

MR imaging for accurate detection and localization of prostate cancer.

300


In chapter 13 the previous ones are discussed. The general conclusion is that with multi-

parametric MRI and MR-biopsy, less needles are used with an increased yield and with the

added bonus (because of lesion targeting), a more accurate presentation of the true tumour

aggression (Gleason grades). With this technique, insignificant cancers can be differentiated

from significant ones. The MRI-technique is substantially superior to TRUS-biopsy for this

purpose.

Future perspectives:

1. Multi-parametric MRI should be implemented as fast as possible. Implementation

barriers should be defined and solved.

2. The value of multi-parametric MRI in screening should be investigated. This technique

in combination with PSA, potentially can lead to implementation of prostate screening.

3. Also the role of multi-parametric MRI in Active Surveillance should be evaluated

4. The same is true for MR-guided minimal invasive focal therapy, as the most aggressive

tumour part can be mapped.

5. CAD will enhance implementation of multi-parametric MRI, however, a substantial

amount of research in this regard is needed.

6. The potential, additional value of 7T MRI should be explored.

A final word of consideration:

According to the author, sufficient stones have been dislodged to aid in bringing about a

PARADIGM SHIFT in diagnosis and management of prostate cancer. The thunder sounds of the

“Battle of Anghiari” are audible. May this battle turn out to be a victorious one for patients, their

families and clinicians alike!

Let us as clinicians never forget the purpose of all our work and endeavor:

301


Nederlandse Samenvatting

MRI bij de detectie en de bepaling van de aggressiviteit van

- Van Theorie tot Praktijk

SAMENVATTING VAN DE STELLINGEN OP BASIS VAN DIT

PROEFSCHRIFT:

Iedere man met de DIAGNOSE van prostaatkanker MOET een multi-

parametrische MRI ondergaan

Iedere man met een blijvende VERDENKING op het hebben van

prostaatkanker MOET een multi-parameterische MRI ondergaan

Iedere man bij wie een nauwkeurige bepaling van de AGRESSIVITEIT van

prostaatkanker nodig is MOET een multi-parametrische MRI ondergaan!

Prostaatkanker is een aanzienlijk maatschappelijk probleem geworden. Deze vorm van kanker

heeft longkanker van de eerste plaats van de meest voorkomende maligniteiten verdrongen.

Bovendien is het de 2e doodsoorzaak ten gevolge van kanker bij de man. In Nederland worden

er jaarlijks 9.000 nieuwe diagnoses gesteld. Naast een groot maatschappelijk probleem is dit een

aanzienlijke belasting voor patiënten en artsen, met aanzienlijke psychologische effecten en

hoge kosten van de zorg.

Ondanks het feit, dat prostaatkanker een ziekte van de “oude man” is, laten post-mortem studies

zien, dat deze vorm van kanker al bij 25% voorkomt bij mannen van 30-40 jaar, toenemend tot

>80% bij mannen van 80 jaar. Toch gaan meer mannen dood “met” dan “aan” prostaatkanker.

Bovendien treed 2/3de van de sterfte op bij de oudere patiënten (>75 jaar). Dit houdt in, dat

slechts 5% van de sterfte door prostaatkanker gezien wordt bij jonge mannen. Dit zijn er

302


jaarlijks 450. Alhoewel de ziekte specifieke sterfte slechts 15% is, is de morbiditeit na therapie

(zoals impotentie, incontinentie, blaasontsteking, gezwollen benen en infectie) aanzienlijker.

Hier komen de gevolgen van een PSA-stijging na een radicale behandeling (zoals angst

onzekerheid, zorg, depressie) nog eens bij. Deze worden vaak niet voldoende herkend en

gewaardeerd.

De meeste prostaatkankers zijn laag agressief en veroorzaken geen klachten en de man gaat er

niet aan dood. Daarom worden deze kankers ook wel niet-significant genoemd. De agressie van

de kanker wordt tot nu toe bepaald door pathologisch onderzoek van tumorweefsel, verkregen

met een echografisch geleid biopt via de anus en uitgedrukt in de zogenaamde Gleason

gradering. De juiste bepaling hiervan is van zeer groot belang voor de bepaling van de juiste

therapie en de prognose. En dit is nu net de grootste valkuil! Met de tot nu toe gebruikelijke

rectale echogeleide biopsie (het wegnemen van stukjes weefsel via de anus) wordt niet alleen

een aanzienlijk aantal prostaatkankers gemist, maar wordt ook het meest agressieve deel van de

kanker in 40% gemist, hetgeen leidt tot het later ontdekken van de tumor of tot ondergradering

van de agressie ervan. Het behoeft geen uitleg, dat dit leidt tot onderbehandeling of inefficiënte

behandeling.

Op dit moment laat meer dan de helft van de mannen hun PSA testen. Helaas is niet alleen de

PSA verhoogd bij kanker, maar wordt dit ook vaak gezien bij goedaardige prostaataandoeningen

(specificiteit van de test is 63%). Wanneer het PSA niveau boven een bepaalde waarde komt,

wordt overgegaan tot de echografische biopten. Bij een drempelwaarde van 4 ng/ml wordt 15%

van de kankers gemist. Van de ontdekte tumoren is weer 15% agressief, en behoeven

behandeling. Daarentegen hebben de meeste mannen met een licht verhoogd PSA (4-10 ng/ml)

door de lage specificiteit van de PSA-test geen prostaatkanker, en ondergaan derhalve een

onnodig echo-biopt. Omdat met de echografie de kanker veelal niet zichtbaar is maar de prostaat

wel, wordt met het echografisch biopt op 10 tot 12 stelselmatige plekken uit de prostaat weefsel

weggegnomen. Deze “blinde” biopsie wordt alleen nog maar bij de prostaat gedaan. Zo

ondergaan bijvoorbeeld vrouwen met een mogelijke borstkanker gerichte weefselafname geleid

door goede beeldvorming.

De schrijver van dit proefschrift heeft zich daarom als doel gesteld, de huidige diagnostiek van

303


prostaatkanker te verbeteren. Een “PARADIGMA SHIFT” creëren, is de enige weg vooruit!

Daarom wordt in dit proefschrift uitvoerig en nauwgezet ingegaan op de uitdagingen voor de

patiënt en hun artsen. Er wordt gezocht naar nieuwe technieken, deze worden gevalideerd en

vervolgens geïmplementeerd in de routine zorg. Het doel van dit proefschrift, was niet alleen om

te zien welke technieken leiden tot minder biopsie naalden met betere opbrengst maar ook een

verbetering in de representativiteit van meest agressieve component aangwezig in de tumor.

Dit proefschrift onderzoekt derhalve het pad van “van theorie tot praktijk”.

Samenvattend is het volgende gevonden:

In hoofdstuk 3 worden de haalbaarheid en de nauwkeurigheid van multi-parametrische MRI –

bestaande uit anatomische T2-gewogen beelden en functionele dynamische contrast en diffusie

MRI- onderzocht om voor tumor verdachte gebieden te identificeren. Er werd gebruik gemaakt

van een magneetveldsterkte van 3T en een 32-kanaals oppervlaktespoel. Ook werd gekeken of

deze biopten met een MR-compatibel biopsie apparaat genomen konden worden. Multi-

parametrische MRI bleek zeer nauwkeurig te zijn om prostaatkanker op te sporen. Bovendien

bleek de MR-geleide biopsie, gericht op de multi-parametrische MRI verdachte afwijking, goed

uitvoerbaar en waardevolle informatie op te leveren over de aanwezigheid van prostaatkanker.

Daarom zouden een aanzienlijk deel van de mannen met een verhoogd PSA een multi-

parametrisch onderzoek moeten ondergaan.

In hoofdstuk 4 wordt de in hoofdstuk 3 ontwikkelde techniek in een grotere patiëntengroep met

patiënten met een persisterend verhoogd PSA en minimaal 2 negatieve voorafgaande

echografische bioptie-sessies toegepast. Het doel was om te kijken bij hoeveel patiënten multi-

parametrische MRI en MR-biopsie, ondanks voorafgaande negatieve echo-biopsie, toch

prostaatkanker ontdekt kan worden. Bovendien werd gekeken, bij hoeveel patiënten er een

klinisch significante kanker aanwezig was (dat zijn kankers, die agressief zijn en dus behandeld

moet worden). Bij 40/ 68 (59%) patiënten werd prostaatkanker gevonden. Hiervan hadden er

37 (93%) een significant carcinoom. Ook werd gezien, dat de meeste tumoren buiten het gebied

van de echografisch biopsie lagen. Ook in een subgroep analyse –voor PSA waarde, prostaat

volume en PSA densiteit- bleek de MR-techniek significant betere resultaten op te leveren in

vergelijking met de echo-biopsie. Deze bevindingen onderstrepen de CBO-2007 richtlijn, dat

304


patiënten met een blijvende klinisch verdenking op prostaatkanker na een negatieve echo-

biopsie een multi-parametrische MRI en MR-biopsie moeten ondergaan.

In hoofdstuk 5 werd de waarde van multi-parametrische MRI en MR-biopsie onderzocht bij

patiënten met een mogelijk recidief na een voorafgaande externe radiotherapiebehandeling.

Deze techniek had bij deze patiëntengroep een positief voorspellende waarde van 75% en bleek

ook hier succesvol te zijn.

Vooral het ontdekken en lokaliseren van kankers in de zogenaamde is een

uitdaging. Daarom werd in hoofdstuk 6 in een retrospectieve studie gekeken hoe nauwkeurig

multi-parametrische MRI is om de laaggradige (Gleason graad 2/3) en agressievere (Gleason

graad 4/5) kankers in de transitie zone op te kunnen sporen. Er werd gevonden dat de

anatomische T2-gewogen beelden het beste waren om dit te doen: de ervaren radioloog

ontdekte 86% van de agressieve carcinomen. Indien ook de functionele technieken (diffusie en

contrast) werden toegevoegd, nam de nauwkeurigheid maar weinig toe tot 91%. De laaggradige

tumoren werden echter met de T2-techniek slecht ontdekt (24%) en nam meer toe met de

functionele beeldvorming erbij (47%). De T2-gewogen techniek bleek dus een nauwkeurige

techniek te zijn om agressieve kankers in de transitie zone op te sporen.

De belangrijkste uitdaging is gelegen in het niet invasief nauwkeurig kunnen voorspellen of een

tumor niet significant of significant is. Tot nu toe blijkt dit met bestaande technieken niet goed

mogelijk. Daarom wordt in hoofdstuk 7 in de perifere zone van de prostaat de correlatie tussen

de apparent diffusion coefficient (ADC) waarden van 3T diffusie MRI met de Gleason gradering

in het operatiepreparaat onderzocht. Het bleek, dat de ADC waarden een omgekeerde correlatie

met de Gleason gradering vertoonde, en dat de ADC waarden van de laag-, intermediair- en de

hooggradige tumoren significant verschilden (p<0.001). Met de gemiddelde ADC waarde van een

bepaald gebied bleek het mogelijk te zijn, de laaggradige (niet significante) van de agressieve

(significante) tumoren te onderscheiden (AUC 0.90). Er werd daarom geconcludeerd, dat

diffusie MRI, gebruik makend van de ADC waarden een goede fenotypische biomarker is om niet

significante prostaatkaker van significante te onderscheiden.

In hoofdstuk 9 werd de waarde van een additionele functionele MRI-techniek, magnetic

305


resonance spectroscopic imaging (MRSI) bij 3T onderzocht voor de differentiatie tussen laag en

hoger agressieve tumoren. Er werd een correlatie gevonden tussen de Gleason graad en de

Choline + Creatine over Citraat- en de Choline over Citraat ratio’s (AUC 0.70 en 0.74). De

correlatie tussen Cho+Cr/Cit en tumor agressie was significant. Ook werd de MRSI techniek

gestandaardiseerd teneinde deze in de praktijk beter te kunnen gebruiken. MRSI levert op grond

van dit onderzoek een extra mogelijkheid om de tumor agressie niet invasief met MRI te

bepalen.

Een belangrijk probleem is, dat de patiënten met een laaggradige tumor (Gleason 3+3) vaak

ondergegradeerd worden met het echo-biopt. Eigenlijk moeten deze patiënten een meer

agressieve behandeling ondergaan, echter de “fout lage” gradering leidt tot onderbehandeling.

Graag zouden we de “ware” Gleason 3+3 patiënten onderscheiden van die met een meer

agressieve tumor. Om de klinische waarde van diffusie MRI hiervoor te onderzoeken, zijn er 2

verschillende studies gedaan. De eerste studie, is een retrospectieve, beschreven in Hoofdstuk

8. De tweede in hoofdstuk 10. Bij 11/23 patiënten met op echo-biopsie uitslag van een

laaggradige tumor (Gleason score 3+3=6), liet het operatiepreparaat toch een agressievere

tumor zien. Deze patiënten (48%) hadden dus op de echo-biopsie een ondergradering. Diffusie

MRI liet een significant verschil zien in de groep van patiënten die met echografie

ondergradering toonden (11/23) versus die patiënten, die een terecht laag stadium hadden

(12/23). ROC analyse toonde een AUC van 0.88 voor de ADC waarde. Deze resultaten tonen aan,

dat diffusie MRI van grote klinische waarde is bij patiënten, bij wie de echo-biopsie een

laaggradige tumor (Gleason score ) aantoont. De ondergradering wordt dankzij multi-

parametrische MRI aanzienlijk gereduceerd. Daarom ZOU deze MRI techniek een onderdeel

MOETEN uitmaken bij patiënten met (de verdenking op ) prostaatkanker.

De bevindingen in hoofdstuk 10 vormen de kroon op dit proefschrift. Deze prospectieve studie

vergelijkt in 2 gelijkwaardige patiëntengroepen (matched-cohorts) de uitkomsten voor de

detectie van het meest agressieve tumor deel in het operatie preparaat met multi-parametrisch

MRI + MR-biopsie in de ene, en met echo-biopsie in de andere groep. Het bleek dat MR-biopsie

gericht op de meest afwijkende laesie op de multi-parametrische MRI superieur was. De MRI

methode liet in 95% een exacte overeenkomst zien met het operatiepreparaat voor wat betreft

het aantonen van de agressieve tumoren (hoogste Gleason graad). Bij het echo-biopt was dit

306


slechts in 54% het geval. Bovendien waren met MRI slechts 3 naalden nodig voor de biopsie,

tegen 10 met het echo-biopt. Dit onderstreept, dat met MRI met minder naalden meer informatie

wordt verkregen, en daarom onderdeel moet uitmaken bij de diagnostiek van prostaatkanker.

In hoofdstuk 7 werd gezien, dat de gemiddelde waarde van de ADC voor de “normale” perifere

zone tussen patiënten variatie vertoont. Op grond daarvan kan verwacht worden, dat ook de

waarde van de tumor varieert. Deze inter-patiënt variatie kan nadelig werken op de voorspelling

van de tumor agressie. Daarom werd deze variatie in hoofdstuk 11 onderzocht. Inderdaad bleek

de ADC waarde voor normale perifere zone significant tussen patiënten te verschillen (p<0.001).

Dit is waarschijnlijk het gevolg van natuurlijke variatie in de fysiologie van de prostaat. Indien

voor deze “normale” variatie gecorrigeerd werd, bleek de voorspelling van de tumor agressie

significant te verbeteren (p=0.001). De ADC waarden moeten dus gezien worden tegen het licht

van de ”achtergrond” variatie van de “normale” prostaat.

In hoofdstuk 12 gaat in op het belangrijke aspect, hoe de multi-parametrische MRI bij de in de

radiologische praktijk algemeen geïmplementeerd kan worden. In dit proefschrift lijkt de multi-

parametrische MRI makkelijk toe te passen, maar de praktijk is weerbarstiger. Het kost de

beginnende radioloog moeite om de juiste beoordelingstechniek van de MRI-beelden te leren.

Om dit vergemakkelijken, en om de variatie tussen de radiologen bij de beoordeling te

reduceren is er een computer assisted diagnosis (CAD) techniek ontwikkeld en uitgetest. Deze

techniek maakt gebruik van de functionele parameters van contrast en diffusie MRI. Toevoeging

van CAD verbeterde de herkenning van prostaatkanker door de niet ervaren radioloog

significant voor zowel de perifere (p<0.001) als de transitie zone (p=0.001). Met gebruik van

CAD bereikte de niet ervaren radioloog (AUC 0.91) bijna het niveau van de expert radioloog.

(AUC 0.93).

In hoofdstuk 13 worden voorgaande hoofdstukken in verband gebracht en bediscussieerd. De

algehele conclusie is, dat met multi-parametrische MRI en MR-biopsie met minder naalden een

betere diagnose gesteld kan worden. Dat wil zeggen insignificante kanker kan worden

onderscheiden met deze techniek van de significante. De techniek is significant beter dan wat er

op dit moment –met de echo-biopsie- gedaan wordt.

307


Toekomst-persopectieven:

1. Multi-parametrische MRI moet zo snel mogelijk worden geïmplementeerd. De

implementatie barrières moeten worden geïdentificeerd en overwonnen.

2. Er moet worden onderzocht, wat de waarde is van deze MRI techniek bij prostaatkanker

screening. Op grond van de bewijzen uit dit proefschrift, lijkt het aannemelijk dat

invoering van screening met PSA+ multi-parametrische MRI mogelijk en wenselijk is.

3. De rol van multi-parametriche MRI bij active survellance moet eveneens worden

gevalueerd.

4. De rol van MRI bij lokale therapie is eveneens veelbelovend. Immers met multi-

parametrische MRI kan de agressieve tumor component in kaart worden gebracht.

5. CAD zal leiden tot snellere implementatie van Multi-parametrische MRI en de kwaliteit

van de beoordeling verbeteren, echter er moet nog veel aan CAD ontwikkeld worden,

voordat iedereen dit kan gaan gebruiken.

6. De potentiële meerwaarde van hogere veldsterkten (bijvoorbeeld 7T) zal moeten

worden onderzocht.

Enkele slotopmerkingen:

Er zijn voldoende stenen verplaatst om de PARADIGMA SHIFT in de diagnostiek en de daarop

volgende behandeling te laten plaatsvinden. Het gedonder van de “Battle of Anghiari” is

hoorbaar. Laat deze oorlog uitdraaien op overwinning voor de patiënt met prostaatkanker, zijn

familie en zijn behandelaar.

Clinici, vergeet echter nooit het doel van ons werk en inspanningen:

atiënt is onze grootste roeping!

308


PART SIX

POSTLUDE

Postlude

A. LIST OF PUBLICATIONS

2012

1. Value of 3T multiparametric MR Imaging and MR guided biopsy for early risk re-

stratification in active surveillance of low-risk prostate cancer: a prospective

multicentre cohort study. C. Hoeks, D. Somford, T. Hambrock, C. Hulsbergen-van de

Kaa, J.O. Barentsz (Submitted)

2. Differentiation of Prostatitis and Prostate Cancer using Diffusion Weighted Imaging

and MR-guided Biopsy at 3T. M. Schouten, K. Nagel, B ten Haken, C. Hoeks, G. Litjens T.

Hambrock, J.O. Barentsz, J.J. Fütterer. Radiology (Accepted awaiting publication)

3. Evaluation of Diffusion-Weighted MR Imaging (DWI) at Inclusion in an Active

Surveillance Protocol for Low-Risk Prostate Cancer. D.M. Somford, C. M. Hoeks, C.A.

Hulsbergen-van de Kaa, T. Hambrock, J.J. Futterer, J. A. Witjes, C. H. Bangma, H. Vergunst,

G.A. Smits, J. R. Oddens, I.M. van Oort , J.O. Barentsz. Invest Radiol (Accepted awaiting

publication)

4. Computer-aided diagnosis of prostate cancer using multiparametric 3T MR

imaging : Effect on Observer Performance. Hambrock T, Vos P, Hulsbergen-van de

Kaa C, Barentsz J, Huisman HJ Radiology (Accepted awaiting publication)

5. The effect of inter-patient normal peripheral zone Apparent Diffusion Coefficient

variation on the Prediction of Prostate Cancer Aggressiveness. Litjens G, Hambrock

T, Barentsz JO, Huisman HJ. Radiology 2012 Oct 265 (1):260-3

6. MR Spectroscopy and Diffusion Weighted Imaging at 3T for in vivo Assessment of

Prostate Cancer Aggressiveness. Kobus T, Vos P, Hambrock T, Hulsbergen-van de Kaa

C, Barentsz JO, Scheenen T. Radiology. 2012 Nov;265(2):457-67

7. Simulated required accuracy of image registration tools for targeting high-grade

cancer components with prostate biopsies". Van de Ven W, Hulsbergen-van de Kaa C,

Hambrock T, Barentsz JO, Huisman HJ Eur Radiol. (Accepted awaiting publication)

310311

Postlude

8. 3T Magnetic Resonance-Guided Prostate Biopsy in Men with Increased PSA and

Repeated Negative Random Systematic Transrectal Ultrasound Biopsies: Detection

of Clinically Significant Prostate Cancers. Hoeks C, Schouten MG, Bomers JG,

Hoogendoorn SP, Hulsbergen-van de Kaa CA, Hambrock T, Vergunst H, Sedelaar JP,

Fütterer JJ, Barentsz JO. Eur Urol. 2012 Nov;62(5):902-9

9. Initial Experience With Identifying High-Grade Prostate Cancer Using Diffusion-

W

Schematic TRUS-Guided Biopsy: A Radical Prostatectomy Correlated Series.

Somford DM, Hambrock T, Hulsbergen-van de Kaa CA, Fütterer JJ, van Oort IM, van

Basten JP, Karthaus HF, Witjes JA, Barentsz JO. Invest Radiol. 2012 Mar;47(3):153-8.

10. Functional MRI techniques demonstrate early vascular changes in renal cell cancer

patients treated with sunitinib: a pilot study. Desar IM, ter Voert EG, Hambrock T, van

Asten JJ, van Spronsen DJ, Mulders PF, Heerschap A, van der Graaf WT, van Laarhoven

HW, van Herpen CM.Cancer Imaging. 2012 Jan 12;11:259-65.

11. Prospective Assessment of Prostate Cancer Aggressiveness using 3 Tesla Diffusion

Weighted MR Imaging Guided Biopsies versus a systematic 10-core Transrectal

Ultrasound Prostate Biopsy Cohort. Hambrock T, Hoeks C, Hulsbergen-van de Kaa C,

Scheenen J, Oort I, Fütterer JJ, Huisman H, Barentsz J. Eur Urol. 2012 Jan;61(1):177-84.

2011

12. Value of 3 Tesla Endorectal Coil Magnetic Resonance Imaging in Local Staging of

Prostate Cancer. Hamoen E, Hambrock T, Witjes J, Barentsz J (Submitted)

13. High-risk prostate cancer: value of multi-modality 3T MRI-guided biopsies after

previous negative biopsies. Fütterer JJ, Verma S, Hambrock T, Yakar D, Barentsz JO.

Abdom Imaging. 2011 Oct 29.

14. Prostate cancer: multiparametric MR imaging for detection, localization, and

staging. Hoeks CM, Barentsz JO, Hambrock T, Yakar D, Somford DM, Heijmink SW,

Scheenen TW, Vos PC, Huisman H, van Oort IM, Witjes JA, Heerschap A, Fütterer JJ.

Radiology. 2011 Oct;261(1):46-66. Review.

311312

Postlude

15. F-18 FDG PET/CT as a Crucial Guide Toward Optimal Treatmet Planning in a Case

of Postirradiation Sarcoma 10 Years after Primary Bone Lymphoma of the Pelvis.

de Rooy, JW, Hambrock T, Vriends D, Flucke UE, van der Geest IC, van de Luijtgaarden AC,

Schreuder BW, de Geus-Oei LF. Clin Nucl Med 2011 Jul; 36 (7): 565-7.

16. Relationship between apparent diffusion coeeffcients at 3.0-T MR imaging and

Gleason Grade in Peripheral Zone Prostate Cancer. . Hambrock T, Somford D,

Hoeks C, Hulsbergen-vandeKaa C, Scheenen T, Huisman HJ, van Oort I, Witjes JA,

Barentsz J. Radiology 2011 May; 259(2):453-61

17. In Vivo Assessment of Prostate Cancer Aggressiveness Using Magnetic Resonance

Spectroscopic Imaging at 3T with an Endorectal Coil. Kobus T, Hambrock T,

Hulsbergen-van de Kaa CA, Wrigth AJ, Barentsz JO, Heerschap A, Scheenen TW. Eur Urol.

2011 Nov;60(5):1074-80

18. Prostate cancer detection and dutasteride: utility and limitations of prostate-

specific antigen in men with previuos negative biopsies. Van Leeuwen PJ, Kölbe K,

Huland H, Hambrock T, Barentsz J, Schröder FH. Eur Urol 2011 Feb; 59(2): 183-190.

2010:

19. Magnetic Resonance Imaging guided prostate biopsy in men with repeat negative

biopsies and increased prostate specific antigen. Hambrock T, Somford DM, Hoeks C,

Bouwense SA, Huisman HJ, Yakar D, van Oort IM, Witjes JA, Fütterer JJ, Barentsz JO.

J Urol 2010 Feb;183(2):520-7.

20. Relationship of Apparent Diffusion Coefficient Values at 3T and prostate cancer

Gleason grades in the peripheral zone. Hambrock T, Somford D, Hoeks C, Hulsbergen-

vandeKaa C, Scheenen T, Huisman HJ, van Oort I, Witjes JA, Barentsz J. Radiology. 2011

May;259(2):453-61

21. Feasibility of 3T Dynamic Contrast-Enhanced Magnetic Resonance-Guided Biopsy

in Localizing Local Recurrence of Prostate Cancer after External Beam Radiation

Therapy. Yakar D, Hambrock T, Huisman H, Hulsbergen-van de Kaa CA, van Lin E,

Vergunst H, Hoeks CM, van Oort IM, Witjes JA, Barentsz JO, Fütterer JJ. Invest Radiol. 2010

Mar ;45(3):121-5.

312313

Postlude

2009:

22. Automated calibration for computerized analysis of prostate lesions using

pharmacokinetic magnetic resonance images. Vos P, Hambrock T, Barentsz J,

Huisman HJ. MICCAI 2009, September 20-24, 2009, Proceedings, Part II, Volume

5761/2009, p 836-843

2008:

23. Magnetic Resonance Imaging guided biopsies of the prostate: Technique,

feasibility and clinical applications. Yakar D, Hambrock T, Hoeks C, Barentsz JO,

Fütterer JJ. Top Magn Reson Imag. 2008 Dec;19(6):291-5

24. Diffusion and Perfusion MR imaging of the Prostate. Somford DM, Futterer JJ,

Hambrock T, Barentsz JO. Magn Reson Imaging Clin N Am. 2008 Nov;16(4):685-95.

25. MR-Guided Biopsy of the Prostate: An Overview of Techniques and a Systematic

Review. Pondman KM, Fütterer JJ, Ten Haken B, Schultze Kool LJ, Witjes JA, Hambrock T,

Macura KJ, Barentsz JO. Eur Urol. 2008 Sep;54(3):517-27

26. 32-Channel Coil 3T MR Guided Biopsies of Prostate Tumor Suspicious Regions

Identified on Multi-Modality 3T MR Imaging Technique and Feasibility.

Hambrock T, Futterer J, Huisman H, Oort I, Witjes J, Van Basten J,Barentsz J; Invest

Radiology 2008 Oct;43(10):686-94.

27. Computerized analysis of prostate lesions in the peripheral zone using dynamic

contrast enhanced MRI. Vos P, Hambrock T, Hulsbergen-vandeKaa C, Futterer J,

Barentsz J, Huisman J; Med Phys. 2008 Mar;35(3):888-99

2007:

28. Local Staging of Prostate Cancer using Endorectal Coil MR Imaging. Hambrock T,

Barentsz JO, Futterer JJ. Cancer Imaging Vol. II : Instrumentation and Applications,

H.M. Hayat (Editor), Acad. Press, Ch. 46 p.641 655

313314

Postlude

29. Prostate cancer: body-array versus endorectal coil MR imaging at 3T

comparison of image quality, localization and staging performance. Heijmink SW,

Futterer JJ, Hambrock T, Huisman HJ, Hulsbergen-Van deKaa CA, Knipscheer BC,

Kiemeney LA, Witjes JA, Barentsz JO. Radiology 2007 Jul;244(1):184- 95.

2006:

30. Dynamic Contrast Enhanced MR Imaging in the Diagnosis and Management of \

Prostate Cancer. Hambrock T, Padhani A, Tofts P, Vos P, Huisman H, Barentsz JO.

Categorical Course in Genitourinary Imaging, RSNA 2006.

2002:

31. Screening for Chilhood Anemia using Coppersulfate Densitometry. Funk M,

Hambrock T, van Niekerk GC, Wittenberg DF. S Afr Med J 2002. Dec;92(12):978-82.

314315

Postlude

B. LIST OF PRESENTATIONS SCIENTIFIC PAPER PRESENTATIONS

1. 2011 Oct International Cancer Imaging Society Kopenhagen, Denmark:

"The Value of MR guided Biopsies in Prostate Cancer”

2. 2010 Mar Society of Computed Body Tomography and MR San Diego, U.S.A

"Correlation between 3T DWI-ADC and tumor Gleason score in Prostatectomy Specimens”

3. 2010 Mar European Society of Radiology Vienna, Austria:

"Concordance between MR-guided biopsy determined Gleason Score and Prostatectomy GS”

4. 2010 Mar European Society of Radiology Vienna, Austria:

"Correlation between 3T DWI-ADC and tumor Gleason score in Prostatectomy Specimens”

5. 2008 Dec Radiological Society of North America annual meeting Chicago, U.S.A :

The Value of 3 Tesla Magnetic Resonance Imaging Guided Prostate Biopsies in Men with

Repetitive Negative Biopsies an elevated PSA

6. 2008 Oct Nederlandse Radiologendagen Rotterdam, Netherlands:

Effect van Computer-Aided Diagnosis op de Karakterisatie van Prostaat Laesies op Dynamische

Contrast MRI

7. 2008 Oct Nederlandse Radiologendagen Rotterdam, Netherlands :

Waarde van MR Geleide Bioptie van de Prostaat op 3 Tesla

8. 2008 Oct International Cancer Imaging Society annual meeting Bath, England :

Correlation between 3T DWI-ADC and tumor Gleason Score in Prostatectomy specimens

9. 2008 Oct International Cancer Imaging Society annual meeting Bath, England :

Effect of Computer Assisted Diagnosis to characterized prostate tumor suspicious regions on

DCE-MRI

10. 2008 Oct International Cancer Imaging Society annual meeting Bath, England:

DCE-MRI & MR Guided Biopsy for Detection of Prostate Cancer Recurrence following

Radiotherapy”

315316

Postlude

11. 2008 Oct International Cancer Imaging Society annual meeting Bath, England:



12. 2008 Sep European Society of Uroradiology annual meeting Munic, Germany:



13. 2008 Sep European Society of Uroradiology annual meeting Munic, Germany:

Correlation between 3T MRI Apparent Diffusion Coefficient Values and Prostate Cancer Gleason

Score in Prostatectomy Specimen

14. 2007 Jun International Society of Magnetic Resonance in Medicine / European

Society of Magnetic Resonance in Medicine and Biology joint meeting Berlin, Germany

“32-Ch MR Guided Biopsy of tumor suspicious regions on multi-modality 3T MRI of the prostate

– Initial Experience”

15. 2007 Apr European Society of Uroradiology (ESUR) / Society of Uroradiology

(SUR) joint annual meeting Bonita Springs, U.S.A.

“32-Channel MR Guided Biopsy of tumor suspicious regions on multi-modality 3T MR imaging of

the prostate – Initial Experience”

16. 2006 Nov Radiological Society of North America (RSNA) meeting Chicago, U.S.A

-mapping vs T2* mapping of prostate cancer lesions : Elektronic poster presentation

316317

Postlude

C. LIST OF PRESENTATIONS PRESENTATIONS ON INVITATION

1. 2012 Oct Radiologendagen

“MRI van de prostaat: Pro en contra”

2. 2012 May Oncology 2.0 Apeldoorn, Netherlands:

“The Value of MRI prior to Radical Prostatatcomy”

3. Feb 2012 Pathologie Onderwijs St. Radboud, Nijmegen

“De waarde MRI bij prostaatkanker diagnostiek”

4. Sept 2011 Kernspintomografie Fortbildung – Münster, Germany

5. Jul 2011 Radiologen Fortbildungscongress Düsseldorf, Germany

6. Jun 2011 Urologen Fortbildungscongress Zürich, Switserland

7. Okt 2010 - Patient Prostata Krebs Selbshifegruppe Düsseldorf, Germany

“The value of MRI in screening and detection of prostate cancer“

8. 2008 Jan - Abteilung Radiologie/Urologie/Radiotherapie Vienna, Austria

“The value of MRI in screening and detection of prostate cancer“

9. 2007 Nov - Afdeling Radiologie/Urologie Hirslanden Hospital, Zürich, Zwitserland

“The role of Magnetic Resonance Imaging in the diagnosis and management of prostate cancer”

10. 2007 Aug Afdeling Urologie UMCN St. Radboud Nijmegen

"De rol van MRI in de diagnostiek en behandeling van prostaatkanker”

317318

Postlude

D. LIST OF AWARDS

2010:

1. Lauterbur Award from the Society of Computed Body Tomography and Magnetic Resonance (SCBT-MR), San Diego, U.S.A for best MR paper entitled:

"Scientific presentation on Correlation of 3T DWI-ADC and prostate cancer Gleason Score”

2008 :

2. First Prize Award from the International Cancer Imaging Society (ICIS) Bath, United Kingdom for best paper entitled:

"Scientific presentation on Correlation of 3T DWI-ADC and prostate cancer Gleason Score”

3. Cum Laude Award from the Society of Computed Body Tomography and Magnetic

Resonance (SCBT-MR), Charleston, U.S.A (presented by Prof.dr.J.O.Barentsz) for:

"Scientific Presentation on MR imaging guided biopsy of the prostate at 3T using a 32- channel phased array coil"

2007 :

4. First Prize Award from the Society of Urogenital Radiology and European Society of Urogenital Radiology, Bonita Springs, U.SA. for best scientific paper entitled: "Scientific Presentation on MR imaging guided biopsy of the prostate at 3T using a 32- channel

phased array coil"

318319

Postlude

E. CURRICULUM VITAE

Thomas Hambrock werd geboren onder heldere Afrika zon op de eerste dag van lente in het

zuidelijke halfrond. Dit was op 1 september 1978 in Pretoria, Zuid-Afrika. Hij groeide in een

bijzonder liefdevolle gezin op en woonde met zijn ouders Eckehard en Karin alsook zijn twee

broers Norbert en Bernard in het dorp Vereeniging. Met 5 jarige leeftijd besloot hij arts te

worden. Zijn interesses in de wetenschap, onderzoek en avontuur zijn op verscheidene

manieren door zijn liefdevolle ouders gestimuleerd. Achter zijn Zeiss microscoop, zijn

chemie experiment set, zijn edelgesteente, munt, postzegel en fossiel verzameling alsook

jagen met zijn luchtgeweer heeft hij veel uren van zijn jongen leven verbracht. Zijn latere

leven was gekenmerkt door wandeltochten met rugsak, tent en slaapzak in de

Drakensbergen van Zuid-Afrika. Zijn grootste hobby is en blijft het intensief bestuderen van

de geschiedenis en genealogie van zijn familie. Zowel zijn basisschool opleiding (bij

Laerskool Vryheidsmonument) alsook zijn middelbare schoolopleiding (bij Hoërskool

Vereeniging) werd in Vereeniging met succes afgerond. Vanaf 1997-2002 studeerde hij

geneeskunde aan de Universiteit van Pretoria, Zuid-Afrika waar hij ook zijn artsen beul, Cum

Laude heeft behaald. Dezelfde rusteloosheid en zin in avontuur die ooit zijn 72 voorouders

heeft gedreven om hun tuiste in Europa achter te laten, heeft hem weer terug naar Europa

gebracht. Hij werkte in totaal 3 jaar als arts in de Verenigde Koninkrijken, eerst binnen de

heelkunde in Stoke-on-Trent (Engeland), dan interne geneeskunde in Bangor (Wales).

Vervolgens werkte hij in Kirckaldy en Dunfermline (Schotland), eerste binnen de

kindergeneeskunde en dan binnen de spoedeisende hulp. In 2005 verhuisde hij opnieuw, dit

keer naar Nederland waar hij een bijzondere PhD onderzoek bij de gerenommeerde prostaat

centrum in Nijmegen, onder leiding van Prof. Barentsz, kon beginnen. Vanaf oktober 2009 is

hij in opleiding tot radioloog bij het Universitair Medisch Centrum St. Radboud, Nijmegen,

Nederland. Op 29 maart 2012 trouwde hij in Zwitserland met de vrouw van zijn dromen,

Nadia.

319

E. CURRICULUM VITAE

Thomas Hambrock werd geboren onder de heldere Afrika zon, op de eerste dag van lente in het

zuidelijke halfrond. Dit was op 1 september 1978 in Pretoria, Zuid-Afrika. Hij groeide in een

bijzonder liefdevolle gezin op en woonde met zijn ouders Eckehard en Karin alsook zijn twee

broers Norbert en Bernard in het dorp Vereeniging. Met 5 jarige leeftijd besloot hij arts te

worden. Zijn interesses in de wetenschap, onderzoek en avontuur zijn op verscheidene

manieren door zijn liefdevolle ouders gestimuleerd. Achter zijn Zeiss microscoop, zijn chemie

experiment set, zijn edelgesteente, munt, postzegel en fossiel verzameling alsook jagen met zijn

luchtgeweer heeft hij veel uren van zijn jongen leven verbracht. Zijn latere leven was

gekenmerkt door wandeltochten met rugsak, tent en slaapzak in de Drakensbergen van Zuid-

Afrika. Zijn grootste hobby is en blijft het intensief bestuderen van de geschiedenis en genealogie

van zijn familie. Zowel zijn basisschool opleiding (bij Laerskool Vryheidsmonument) alsook zijn

middelbare schoolopleiding (bij Hoërskool Vereeniging) werd in Vereeniging met succes

afgerond. Vanaf 1997-2002 studeerde hij geneeskunde aan de Universiteit van Pretoria, Zuid-

Afrika waar hij ook zijn artsen beul, cum laude heeft behaald. Dezelfde rusteloosheid en zin in

avontuur die ooit zijn 72 voorouders hebben gedreven om hun tuiste in Europa achter te laten,

hebben hem weer terug naar Europa gebracht. Hij werkte in totaal 3 jaar als arts in de Verenigde

Koninkrijken, eerst binnen de heelkunde in Stoke-on-Trent (Engeland), dan interne

geneeskunde in Bangor (Wales). Vervolgens werkte hij in Kirckaldy en Dunfermline (Schotland),

eerst binnen de kindergeneeskunde en dan binnen de spoedeisende hulp. In 2005 verhuisde hij

opnieuw, dit keer naar Nederland waar hij een bijzondere PhD onderzoek bij het gerenommeerd

prostaat centrum in Nijmegen, onder leiding van Prof. Barentsz, kon beginnen. Vanaf oktober

2009 is hij in opleiding tot radioloog bij het Universitair Medisch Centrum St. Radboud,

Nijmegen, Nederland. Op 29 maart 2012 trouwde hij in Zwitserland met de vrouw van zijn

dromen, Nadia.

320

E. CURRICULUM VITAE

Thomas Hambrock werd geboren onder de helder Afrika zon, op de eerste dag van lente in het

zuidelijke halfrond. Dit was op 1 september 1978 in Pretoria, Zuid-Afrika. Hij groeide op in een

bijzonder liefdevolle gezin en woonde met zijn vader Eckehard en moeder Karin alsook zijn twee

broers: Norbert en Bernard, in het dorp Vereeniging. Met 5 jarige leeftijd besloot hij arts te

worden. Zijn interesses in de wetenschap, onderzoek en avontuur zijn op verschillende wijzen

door zijn ouders gestimuleerd. Achter zijn Zeiss microscoop, zijn chemie experiment set, zijn

edelgesteente, munt, postzegel en fossiel verzameling alsook jagen met zijn luchtgeweer, heeft

hij veel uren van zijn jongen leven verbracht. Zijn latere leven was gekenmerkt door

wandeltochten met rugzak, tent en slaapzak in de Drakensbergen van Zuid-Afrika. Zijn grootste

hobby is en blijft het intensief bestuderen van de geschiedenis en genealogie van zijn familie.

Zowel zijn basisschool opleiding (aan de Laerskool Vryheidsmonument) alsook zijn

middelbare schoolopleiding (aan de Hoërskool Vereeniging) werd met succes afgerond. Vanaf

1997-2002 studeerde hij geneeskunde aan de Universiteit van Pretoria, Zuid-Afrika waar hij ook

zijn artsen beul, cum laude heeft behaald. Dezelfde rusteloosheid en zin in avontuur die ooit zijn

72 voorouders hebben gedreven om hun tuiste in Europa achter te laten, hebben hem weer

terug naar Europa gebracht. Hij werkte in totaal 3 jaar als arts in de Verenigde Koninkrijken.

Eerst binnen de heelkunde in Stoke-on-Trent (Engeland), dan interne geneeskunde in Bangor

(Wales). Vervolgens werkte hij in Kirckaldy en Dunfermline (Schotland), eerst binnen de

kindergeneeskunde en dan binnen de spoedeisende hulp. In 2005 verhuisde hij opnieuw, dit

keer naar Nederland, waar hij een bijzondere PhD onderzoek kreeg bij het gerenommeerd

prostaat centrum in Nijmegen, onder leiding van Prof. Barentsz. Vanaf oktober 2009 is hij in

opleiding tot radioloog aan het Universitair Medisch Centrum St. Radboud, Nijmegen, Nederland.

Op 29 maart 2012 trouwde hij in Zwitserland met de vrouw van zijn dromen, Nadia.

Postlude

E. DANKWOORD

Hoewel het dankwoord vaak de eerste en ook enige is wat mensen ooit in een proefschrift

lezen, zo heeft het dankwoord to grote belang. Een proefschrift is nooit het werk van een

mens. Veel, veel mensen hebben direct of indirect hiertoe bijgedragen. Het is dus van cruciaal

belang erkenning te geven aan die wie erkenning toekomt.

Indien ik moet terugkijken op mijn tijd als onderzoeker en ja, zelf terug op het leven zo kan ik

niet anders dan dezelfde woorden gebruiken die mijn betovergrootvader, H.W. Stumpf

(1841-1920) ooit schreef over zijn eigen leven:

“An mir und meinem Leben ist nichts auf dieser Erd. Was Christus mir gegeben, das ist des Dankes wert!”

Ein ganz besonderen Dank richte ich euch mein lieber Vater und Mutter. Eure besondere

Fürsorge, Liebe, Geduld und Unterstützung durch die Jahre hindurch haben mich mein Leben

begleitet. Wäre ist nicht für euer Einsatz und vor allem die Motivation um das Beste aus uns

heraus zu fordern, wäre es mir niemals gelungen zu erreichen was ich hab. In gute und

schlechte Zeiten wart ihr immer da als Stützen. Ein ewiger Dank.

An meine Brüdern, Norbert und Bernard. Ihr seid die Säulen in meinem Leben. Ich konnte

immer auf eure Hilfe und Unterstützung rechnen. Vor allem haben eure Gebete mich stets

wieder neue Kraft verliehen. Ihr seid nicht nur meine Brüder, ihr seid auch meine besten

Freunde. Unser unerschütterlicher Glaube in Gott ist ein weiterer Band der unsere Familie

zusammen hält.

Geachte Prof. Barentsz, beste Jelle. Het is uiterst moeilijk te beginnen om mijn dank aan jou

uit te spreken. Jij bent de grootste inspiratie voor mijn proefschrift geweest. Jouw passievolle

liefde voor jouw prostaatkanker werk en je buitengewone gave om met mensen om te gaan

zijn voor vele een bijzondere motivatie. Je hebt niet alleen de ziekte voor ogen maar je ziet

telkens de mens achter de ziekte. Zonder jouw motivatie zou dit proefschrift nooit tot stand

gekomen zijn. Dank, dat ik je niet alleen als hoogleraar, promotor maar ook als vriend mag

waarderen. In de donkerste tijden was jij altijd een helpende hand. Ik zie er na uit, dat wij in

de toekomst nog een lange weg samen zullen volgen in de grote strijd om prostaatkanker.

Beste Henkjan, van dag een, toen ik nog geen Nederlands sprak tot aan het einde was je er

altijd geweest. Jouw perfectionisme, kritische en vooral scherpe ideeën zijn in al mijn

320

E. DANKWOORD

Hoewel het dankwoord vaak de eerste en ook enige is wat mensen ooit in een proefschrift lezen,

zo heeft het dankwoord toch grote belang. Een proefschrift is nooit het werk van een mens.

Bijzonder veel mensen hebben direct of indirect hieraan bijgedragen. Het is dus van cruciaal

belang erkenning te geven aan die wie erkenning toekomt. Indien ik moet terugkijken op mijn

tijd als onderzoeker en ja, zelf terug op het leven, zo kan ik niet anders dan dezelfde woorden

uitspreken die mijn betovergrootvader, H.W. Stumpf (1841-1920) ooit schreef over zijn eigen

leven:

“An mir und meinem Leben ist nichts auf dieser Erd. Was Christus mir gegeben, das ist des Dankes

wert!”

Ein ganz besonderen Dank sage ich euch, mein lieber Vater und meine Mutter. Eure besondere

Fürsorge, Liebe, Geduld und Unterstützung durch die Jahre hindurch, haben mir in meinem Leben

eine besondere Stärke verliehen. Wäre ist nicht für euer Einsatz und vor allem die Motivation um

das Beste aus uns heraus zu fordern, wäre es mir niemals gelungen zu erreichen was ich hab. In

guten und schlechten Zeiten wart ihr immer da, als Stützen. Ein ewiger Dank.

An meinen Brüdern, Norbert und Bernard. Ihr seid die Säulen in meinem Leben. Ich konnte immer

auf eure Hilfe und Unterstützung rechnen. Vor allem haben eure Gebete mich stets wieder neue

Kraft verliehen wenn diese Doktorarbeit, sich lange hinaus gezögert hat. Ihr seid nicht nur meine

Brüder, ihr seid auch meine besten Freunde. Unser unerschütterlicher Glaube in Gott ist ein

wichtiger Band der unsere Familie zusammen hält.

Geachte Prof. Barentsz, beste Jelle. Het is uiterst moeilijk te beginnen om mijn dank aan jou uit te

spreken. Jij bent de grootste inspiratie voor mijn proefschrift geweest. Jouw passievolle liefde

voor jouw prostaatkanker werk en je buitengewone gave om met mensen om te gaan, zijn voor

vele een bijzondere motivatie. Je hebt niet alleen de ziekte voor ogen maar je ziet telkens de

mens achter de ziekte. Zonder jouw motivatie zou dit proefschrift nooit tot stand gekomen zijn.

Dank, dat ik je niet alleen als hoogleraar, promotor maar ook als vriend mag waarderen. In de

donkerste tijden was jij altijd een helpende hand. Ik zie er na uit, dat wij in de toekomst nog een

lange weg samen zullen volgen in de grote strijd tegen prostaatkanker.

Beste Henkjan, van dag een, toen ik nog geen Nederlands sprak tot aan het einde was je er altijd

geweest. Jouw perfectionisme, kritische en vooral scherpe ideeën zijn in al mijn artikelen terug

te vinden. De wereld heeft te weinig onderzoekers van jouw kwaliteit. Ik kan zonder twijfel

toegeven dat mijn proefschrift veel, veel magerder en oninteressanter geweest zou zijn, had jij

321

E. DANKWOORD

Hoewel het dankwoord vaak de eerste en ook de enige is wat mensen ooit in een proefschrift

lezen, zo heeft het dankwoord toch grote belang. Een proefschrift is nooit het werk van een

mens. Bijzonder veel mensen hebben direct of indirect hieraan bijgedragen. Het is dus van

cruciaal belang, erkenning te geven aan die wie erkenning toekomt. Indien ik moet terugkijken

op mijn tijd als onderzoeker en ja, zelf terug op het leven, zo kan ik niet anders dan dezelfde

woorden uitspreken die mijn betovergrootvader, H.W. Stumpf (1841-1920) ooit schreef over

zijn eigen leven:

“An mir und meinem Leben ist nichts auf dieser Erd. Was Christus mir gegeben, das ist des Dankes

wert!”

Ein ganz großen Dank sage ich euch, mein lieber Vater und meine Mutter. Eure besondere

Fürsorge, Liebe, Geduld und Unterstützung durch die Jahre hindurch, haben mir in meinem Leben

eine besondere Stärke verliehen. Wäre ist nicht für euer Einsatz und vor allem die Motivation um

das Beste aus uns heraus zu fordern, wäre es mir niemals gelungen zu erreichen was ich hab. In

guten und schlechten Zeiten wart ihr immer da, als Stützen. Ein ewiger Dank.

An meinen Brüdern, Norbert und Bernard. Ihr seid die Säulen in meinem Leben. Ich konnte immer

auf eure Hilfe und Unterstützung rechnen. Vor allem haben eure Gebete mich stets wieder neue

Kraft verliehen wenn diese Doktorarbeit, sich lange hinaus gezögert hat. Ihr seid nicht nur meine

Brüder, ihr seid auch meine besten Freunde. Unser unerschütterlicher Glaube hält uns zusammen.

Geachte Prof. Barentsz, beste Jelle. Het is uiterst moeilijk te beginnen om mijn dank aan jou uit te

spreken. Jij bent de grootste inspiratie voor mijn proefschrift geweest. Jouw passievolle liefde

voor prostaatkanker diagnostiek en je buitengewone gave om met mensen om te gaan, zijn voor

vele een echte motivatie. Je hebt niet alleen de ziekte voor ogen maar je ziet telkens de mens

achter de ziekte en wat je altijd deed was voor de patiënt en niet voor jezelf. Zonder jouw

inspiratie zou dit proefschrift nooit tot stand gekomen zijn. Dank, dat ik je niet alleen als

hoogleraar, promotor, maar ook als vriend mag waarderen. In de donkerste tijden was jij altijd

een helpende hand. Ik zie er na uit, dat wij in de toekomst nog een lange weg samen zullen

volgen in de grote strijd tegen prostaatkanker.

Beste Henkjan, van dag een, toen ik nog geen Nederlands sprak tot aan het einde was je er altijd

geweest en door jouw verdiensten, was de KWF aanvraag waarop ik kon promoveren, tot stand

gekomen. Jouw perfectionisme, kritische en vooral scherpe ideeën zijn in al mijn artikelen terug

te vinden. De wereld heeft te weinig onderzoekers van jouw kwaliteit. Ik kan zonder twijfel

toegeven dat mijn proefschrift veel, veel magerder en oninteressanter geweest zou zijn, had jij

Postlude

artikelen terug te vinden. De wereld heeft te weinig onderzoekers van jouw kwaliteit. Ik kan

zonder twijfel toegeven dat mijn proefschrift veel, veel magerder en oninteressanter geweest

zou zijn, had jij niet de kwaliteit omhoog gedreven. Jij moest vaak achter mij aanzetten om

mijn artikelen uiteindelijk af te ronden. Ik denk dat jij trots kunt zijn. Het is uiteindelijk

allemaal goed gelukt.

Beste Christina. Wij hebben vele, vele uren samen achter de microscoop gezeten om

prostaten te bekijken. Hoewel je veel moeite en tijd heb geofferd hiervoor, had je dit altijd

met grote bereidwilligheid en vreugde gedaan. Bedankt hiervoor. Zonder aarzelen kan ik

zeggen dat jij de beste prostaat patholoog bent die ik ken. Helaas is klonen nog geen

mogelijkheid. Jouw enthousiasme voor de pathologie heeft mij ook geraakt. Ik kan eerlijk

zeggen dat indien de mooiste richting binnen de geneeskunde (radiologie natuurlijk) niet

bestond en ik ook niet kleurblind was, was ik direct patholoog geworden.

Beste Tom, alias Tomaso. Onze wegen kruisten al vroeg. Eerst in Praag en toen als buur

kamergenoten in het F.C. Donders instituut. Jij bent een onuitputbaar kennis van MRI

techniek en Fysica. Jij hebt mij bijna alles geleerd wat ik over MRI weet. Ik besef dat het

dagelijkse lastig vallen met mijn vragen over dit en dat, zeer zeker uitputbaar voor jou was.

Het was onontbeerlijk voor mij en daardoor kon ik zelf achter the scanner aan de slag. Dus

indirect heb jij vele patiënten geholpen.

Beste Pieter, alias Pietro. Dit proefschrift was absoluut niet gelukt zonder jou vele, vele uren

werk in MRCAD. Wij hebben samen duizenden uren verbracht om alles goed te krijgen. De

kroon van onze harde werk is zonder meer onze laatste artikel in Radiology (hoofdstuk 12).

Hier komt alle eer jou ook toe. Bedankt voor alle gezellige tijden en grappen. Hoop dat wij nog

in de toekomst samen kunnen werken aan projecten en ook buiten werk verband kunnen

blijven verkeren.

Beste Ritse. Ik moet eerlijk toegeven dat mijn Nederlandse woordenschat voor 80% van jouw

afkomstig is. Als (gezellige) kamergenoot moest je vele onderbrekeningen en vragen vanuit

mijn kant verdragen zodat ik de Nederlandse taal wat onder beheer kon krijgen. Wij hebben

vele leuke gesprekken: van de samenstelling van het atoom, beklimmen van Mount Everest,

tot de zin van het leven, bij een glaasje bier in de Esculaaf gehad. Ik verlang naar deze tijden.

Beste Monique. Jij was de derde persoon in onze onderzoekers kamer. Jij bent een echt

bijzonder mens en jouw uitgesproken vermogen tot efficiënte multi-tasking heeft mij sterk

321

niet de kwaliteit omhoog gedreven. Jij moest vaak achter mij aanzetten om mijn artikelen

uiteindelijk af te ronden. Ik denk dat jij trots kunt zijn. Het is uiteindelijk allemaal goed gelukt.

Beste Christina. Wij hebben vele, vele uren samen achter de microscoop gezeten om prostaten te

bekijken. Hoewel je veel moeite en tijd heb geofferd hiervoor, had je dit altijd met grote

bereidwilligheid en vreugde gedaan. Bedankt hiervoor. Zonder aarzelen kan ik zeggen dat jij de

beste prostaat patholoog bent die ik ken. Helaas is klonen nog geen mogelijkheid. Jouw

enthousiasme voor de pathologie heeft mij ook geraakt. Ik kan eerlijk zeggen dat indien de

mooiste richting binnen de geneeskunde (radiologie natuurlijk) niet bestond en ik ook niet

kleurblind was, was ik wel patholoog geworden.

Beste Tom, alias Tomaso. Onze wegen kruisten al vroeg. Eerst in Praag en toen als buurmannen

in het F.C. Donders instituut. Jij bent een onuitputbaar kennis van MRI techniek en fysica. Jij hebt

mij bijna alles geleerd wat ik over MRI weet. Ik besef dat het dagelijkse lastig vallen met mijn

vragen over dit en dat, zeer zeker uitputbaar voor jou was geweest. Het was onontbeerlijk voor

mij en daardoor kon ik zelf makkelijker achter the scanner aan de slag. Dus indirect heb jij vele

patiënten geholpen.


werk in MRCAD. Wij hebben samen duizenden uren verbracht om alles goed te krijgen. De kroon

van onze harde werk is zonder meer onze laatste artikel in Radiology (hoofdstuk 12). Hier komt

alle eer jou ook toe. Bedankt voor alle gezellige tijden en grappen. Hoop dat wij nog in de

toekomst samen kunnen werken aan projecten en ook buiten werk verband, kunnen blijven

verkeren.

Beste Ritse. Ik moet eerlijk toegeven dat mijn Nederlandse woordenschat voor 80% van jouw

afkomstig is. Als (gezellige) kamergenoot moest je vele onderbrekingen en vragen vanuit mijn

kant verduren zodat ik de Nederlandse taal wat onder beheer kon krijgen. Wij hebben vele

leuke gesprekken bij een glaasje bier in de Esculaaf gehad: van de samenstelling van het atoom,

beklimmen van Mount Everest, tot de zin van het leven. Ik verlang naar deze tijden.

Beste Monique. Jij was de derde persoon in onze onderzoekers kamer. Jij bent een echt bijzonder

mens en jouw uitgesproken vermogen tot efficiënte multi-tasking heeft mij sterk onder de

indruk gebracht. Ik denk dat jij een van de weinige uitstekende arts-onderzoeker-moeder

combinaties maakt. Je zult zonder twijfel een succes van je toekomstige leven maken.

Beste Caroline. Jij kwam wat later in mijn onderzoekers tijd erbij. Van alle prostaat onderzoekers

is jij beslist de meest enthousiaste en kundige op dit gebied. Jouw creativiteit en kennis bij het

322

niet de kwaliteit omhoog gedreven. Jij moest vaak achter mij aanzetten om mijn artikelen

uiteindelijk af te ronden. Ik denk dat jij trots kunt zijn. Het is uiteindelijk allemaal goed gelukt.

Beste Christina. Wij hebben vele, vele uren samen achter de microscoop gezeten om prostaten te

bekijken. Hoewel je veel moeite en tijd heb geofferd hiervoor, had je dit altijd met grote

bereidwilligheid en vreugde gedaan. Bedankt hiervoor. Zonder aarzelen kan ik zeggen dat jij de

beste prostaat patholoog bent die ik ken. Helaas is klonen nog geen mogelijkheid. Jouw

enthousiasme voor de pathologie heeft mij ook geraakt. Ik kan eerlijk zeggen dat indien de

mooiste richting binnen de geneeskunde (radiologie natuurlijk) niet bestond en ik ook niet

kleurblind was, was ik wel patholoog geworden.

Beste Tom, alias Tomaso. Onze wegen kruisten al vroeg. Eerst in Praag en toen als buurmannen

in het F.C. Donders instituut. Jij bent een onuitputbaar bron van kennis: van MRI techniek en

fysica. Jij hebt mij bijna alles geleerd wat ik over MRI weet. Ik besef dat het dagelijks lastig vallen

met mijn vragen over dit en dat, zeer zeker uitputtend voor jou was geweest. Het was

onontbeerlijk voor mij en daardoor kon ik zelf makkelijker achter the scanner aan de slag. Dus

indirect heb jij vele patiënten geholpen.


werk in MRCAD. Wij hebben samen honderden uren verbracht om alles goed te krijgen. De

kroon van onze harde werk is zonder meer onze laatste artikel in Radiology (hoofdstuk 12). Hier

komt alle eer jou ook toe. Bedankt voor alle gezellige tijden en grappen. Hoop dat wij nog in de

toekomst samen kunnen werken aan projecten en ook buiten werk verband, kunnen blijven

ontmoeten.

Beste Ritse. Ik geef toe dat mijn Nederlandse woordenschat voor 80% van jouw afkomstig is. Als

(gezellige) kamergenoot moest je vele onderbrekingen en vragen vanuit mijn kant verduren

zodat ik de Nederlandse taal wat onder beheer kon krijgen. Wij hebben vele leuke gesprekken

bij een glaasje bier in de Esculaaf gehad: van de samenstelling van het atoom, beklimmen van

Mount Everest, tot de zin van het leven. Ik zal deze tijden zeer zeker missen.

Beste Monique. Jij was de derde persoon in onze onderzoekers kamer. Jij bent een echt

inspirerend mens en jouw uitgesproken vermogen tot efficiënte multi-tasking heeft mij sterk

onder de indruk gebracht. Ik denk dat jij een van de weinige uitstekende arts-onderzoeker-

moeder combinaties maakt. Je zult zonder twijfel een succes van je toekomstige leven maken.

Beste Caroline. Jij kwam wat later in mijn onderzoekers tijd erbij. Van alle prostaat onderzoekers

ben jij beslist de meest enthousiaste en kundige op dit gebied. Jouw creativiteit en kennis bij het

Postlude

onder de indruk gebracht. Ik denk dat jij een van de weinige uitstekende arts-onderzoeker-

moeder combinaties maakt. Je zult zonder twijfel een succes van je toekomstige leven maken.

Beste Caroline. Jij kwam wat later in mijn onderzoekers tijd bij. Van alle prostaat

onderzoekers is jij beslist de meest enthousiaste en kundige op dit gebied. Jouw creativiteit

en kennis bij het aanpakken van onderzoek heeft mij ook veel geholpen. De aantal

wederzijdse mede-auteurschappen getuigt van een goede spanpoging. Dank hiervoor.

Aan de kinderen van de prostaat team (jonge onderzoekers): Mathijn, Joyce, Thiele, Geert en

Wendy met wie ik enigszins heb samengewerkt. Jullie zijn beslist de beste kandidaten om het

belangrijke werk op prostaat kanker beeldvorming over te nemen. Succes ermee.

Jurgen en Stijn. Van jullie heb ik geleerd om een MRI scanner te gebruiken en zelf te kunnen

scannen. Dit was een essentieel onderdeel van mijn onderzoek. Jullie hebben ook het

baanbrekende voorwerk van MRI bij prostaatkanker gedaan. Zonder dit voorwerk was geen

van mijn eigen onderzoek mogelijk.

Beste Solange. Een echte perfectionistische secretaresse. Streng, exact en je doet je werk met

absolute overgave. De menigte VIP patiënten zijn jou in het bijzonder dankbaar voor de

excellente service die zij hebben gehad. Je vriendelijkheid en de glimlach iedere dag op jouw

gezicht betoverd iedereen. Mocht Jelle eendaags aftreden, zou ik jou (mocht ik in een

dergelijke positie verkeren) graag als mijn secretaresse willen aanstellen.

Beste M&M (Manita en Marijke). Wie van jullie de gele pinda en wie de rode een is, kunnen

jullie zelf uitvechten. Jullie harde werk was onontbeerlijk voor mijn onderzoek. De duizenden

patiënten die jullie voor mij hebben ingepland was vaak tijdrovend. Bedankt in ieder geval.

Bedankt ook dat ik altijd jullie kamer kon binnenlopen en een gezellig gesprekje kon maken.

Last but not least. My liewe vrou. Nadi, jy het ongelukkig eers teen die einde van my

navorsingsteit in my lewe weer gekom. Jammer dat dit nie eerder was nie. Ons kan ons daaraan

troos dat die lewe veel eensamer was gewees as jy aand vir aand wat ek moes scan alleen tuis

sou gesit het. Jy het my weer moet, lus en sin vir die lewe voorentoe gegee. Ek is God innig

dankbaar dat hy ons paaie weer laat kruis het. Hoewel die afgelope jaar een met vele stene in

die weg was, weet ek dat ons 2013 jaar die mooiste van ons lewe sal wees. Van jou het ek geleer:

“Carpe Diem”, gryp die dag. Laat ons mekaar motiveer om meer volluit te lewe. Jy is die

waardevolste wat ek het!

322

aanpakken van onderzoek heeft mij ook veel geholpen. De aantal wederzijdse

medeauteurschappen getuigt van een goede spanpoging. Dank hiervoor.

Aan de kinderen van de prostaat team (jonge onderzoekers): Mathijn, Joyce, Thiele, Geert en

Wendy met wie ik enigszins heb samengewerkt. Jullie zijn beslist de beste kandidaten om het

belangrijke werk op prostaat kanker beeldvorming over te nemen. Succes ermee.



baanbrekende voorwerk van MRI bij prostaatkanker gedaan. Zonder dit voorwerk was geen van

mijn eigen onderzoek mogelijk.



excellente service die zij hebben gehad. Je vriendelijkheid en de glimlach iedere dag op

jouwgezicht betoverd iedereen. Mocht Jelle eendaags aftreden, zou ik jou (mocht ik in een

dergelijke positie verkeren) graag als mijn secretaresse willen aanstellen.

Beste M&M (Manita en Marijke). Wie van jullie de gele pinda en wie de rode een is, kunnen jullie

zelf uitvechten. Jullie harde werk was onontbeerlijk voor mijn onderzoek. De duizenden



Last but not least. My liewe vrou. Nadi, jy het ongelukkig eers teen die einde van my navorsingsteit

in my lewe teruggekeer. Jammer dat dit nie eerder was nie. Ons kan ons daaraan troos dat die lewe

veel eensamer was gewees as jy aand vir aand wat ek moes scan alleen tuis sou gesit het. Jy het my

weer moet, lus en sin vir die lewe voorentoe gegee. Ek is God innig dankbaar dat hy ons paaie weer

laat kruis het. Hoewel die afgelope jaar een met vele stene in die weg was, weet ek dat ons 2013

jaar die mooiste van ons lewe sal wees. Van jou het ek geleer: “Carpe Diem”, gryp die dag. Laat ons

mekaar motiveer om meer voluit te lewe. Jy is diewaardevolste wat ek het!

323

aanpakken van onderzoek heeft mij ook veel geholpen. De aantal wederzijdse

medeauteurschappen getuigen van een goede spanpoging. Dank hiervoor.

Beste Rik, de enige uroloog waarmee ik direct heb samengewerkt. Wij hebben menigte avonden

achter het computer programma gezeten om het mysterie van prostaat

kanker agressiviteit op MRI te ontrafelen. Vaak wilden wij de handdoek in de strijd gooiden,

omdat het zo traag en moeizaam verliep. Maar Perseverance mooie artikelen

zijn het uiteindelijk geworden. Bedankt voor jou enorme inzet in dit verband.

Aan de kinderen van de prostaat team (jonge onderzoekers): Martijn, Joyce, Thiele, Geert, Esther

en Wendy met wie ik enigszins heb samengewerkt. Jullie zijn beslist de beste kandidaten om het

belangrijke werk van prostaat kanker beeldvorming over te nemen. Succes ermee.



baanbrekende voorwerk van MRI bij prostaatkanker gedaan. Zonder dit voorwerk was geen van

mijn eigen onderzoek mogelijk.



excellente service die zij hebben gehad. Je vriendelijkheid en de glimlach iedere dag op jouw

gezicht betoverd iedereen. Mocht Jelle eendaags aftreden, zou ik jou (mocht ik in een dergelijke

positie verkeren) graag als mijn secretaresse willen aanstellen.

Beste M&M (Manita en Marijke). Wie van jullie de gele pinda en wie de rode een is, kunnen jullie

zelf uitvechten. Jullie harde werk was onontbeerlijk voor mijn onderzoek. De duizenden



Last but not least. My liewe vrou. Nadi, jy het ongelukkig eers teen die einde van my navorsingstyd

in my lewe teruggekeer. Jammer dat dit nie eerder was nie. Ons kan ons daaraan troos dat die lewe

vir jou veel eensamer was gewees as jy aand vir aand wanneer ik tot laat moes werk om te scan

alleen tuis sou gesit het. Jy het my weer moet, lus en sin vir die lewe voorentoe gegee. Ek is God innig

dankbaar dat hy ons paaie weer laat kruis het. Hoewel die afgelope jaar een met vele stene in die

weg was, weet ek dat ons 2013 jaar die mooiste van ons lewe sal wees. Van jou het ek geleer: “Carpe

Diem”, gryp die dag. Laat ons mekaar motiveer om meer voluit te lewe. Jy is die waardevolste wat

ek het!

Sir Bertrand Russels (1872-1970)

323

The Author

324

The Value of 3 Tesla Magnetic................

Documents

Transcript of The Value of 3 Tesla Magnetic................