David M. Yousem, MD #{149}John A. Arrington, MD #{149}S. James Zinreich, MDAshok J. Kumar, MD #{149}R. Nick Bryan, MD, PhD

Pituitary Adenomas: Possible Roleof Bromocriptine in Intratumoral Hemorrhage’

239

Magnetic resonance (MR) images of68 patients examined for adenomas

of the pituitary gland were retro-spectively reviewed for the presenceof hypenintensity on Ti-weightedimages because of recent reportssuggesting that bromocniptine treat-ment may affect Ti values. Twenty-seven patients were examined aftereither surgery or radiation therapy,29 were receiving bromocriptine,

and 22 had not received any treat-ment at the time of MR imaging.MR imaging criteria showed evi-dence of subacute or chronic intra-tumoral hemorrhage in 18 patients,five of whom had hemorrhageproved at surgery. Ten of the 18 pa-tients were asymptomatic from thehemorrhage; eight had headaches,visual field cuts, or cranial nervedeficits. Although an increased fre-quency of intratumoral hemorrhagewas noted in prolactinomas andmacroadenomas and in patients un-dergoing bromocniptine therapy,the effect of bromocniptine onbleeding was the only significantcorrelation (P < .01).

Index terms: Bnomocmiptine #{149}Pituitary, hemom-

nhage. 145.367 #{149}Pituitary, MR studies, 145.1214

#{149}Pituitary, neoplasms, 145.37

Radiology 1989; 170:239-243

I From the Division of Neuroradiology, theRussell H. Morgan Department of Radiology

and Radiological Science, The Johns HopkinsMedical Institutions, Baltimore. From the 1988

RSNA annual meeting. Received June 13, 1988;

revision requested July 13; revision received

August 4; accepted August 8. Address reprint

requests to D.M.Y.. Division of Neuroradiolo-gy, Department of Radiology, Hospital of the

University of Pennsylvania, 3400 Spruce St.

Philadelphia, PA 19104.

� RSNA. 1989

I NCREASED signal intensity on Ti-

weighted images has been attnib-

uted to cholesterol on fat, melanin,

subacute and chronic hemorrhage,

and some proteins. During magnetic

resonance (MR) imaging of 68 pa-

tients with treated on untreated ade-

nomas of the pituitary gland, areas of

hypenintensity within the tumor bed

were frequently noted on Ti-weight-

ed images, particularly in patients me-

ceiving bromocniptine therapy. It has

been proposed that this shortening

of Ti values is due to a decrease in

intracellular protein synthesis and

water content or to a change in the

molecular mobility of the adenoma

cells that involute under bnomocnip-

tine therapy (1). The purpose of this

communication is to evaluate the fre-

quency of high Ti signal within pitu-

itary adenomas, the relation of high

Ti signal to bromocriptine treatment,

and the cause of the increased Ti me-

taxation.

PATIENTS AND METHODS

Oven a 15-month period, 68 patients

with pituitary masses proved on clinical

on pathologic examination to be adeno-

mas were examined. All patients were im-

aged on a 1.5-T Signa (GE Medical Sys-

tems, Milwaukee) superconducting unit

in Ti-weighted coronal and sagittal

planes with a repetition time (TR) of 500-

700 msec, echo time (TE) of 20-30 msec

(TR/TE = 500-700/20-30); 3-5-mm sec-

tion thickness with a 0-1.5-mm gap, de-

pending on tumor size; 256 X 256 matrix;

four excitations; and 16-20-cm fields of

view. Supplemental proton-density and

T2-weighted coronal images were also ob-

tamed (except in nonhemomrhagic mi-

cnoadenomas) with 2,000-3,000/50-100,

3-mm section thickness, i-mm gap, 128 X

256 matrix, one excitation, and 16-20-cm

fields of view. Axial images were ob-

tamed only if deemed necessary to opti-

mize tumor localization. The images were

neviewed by at least two neuronadiolo-

gists (D.M.Y., A.J.K.). The intensity char-

actenistics of the pituitary adenomas were

compared with the intensity of the nor-

mat pituitary gland, which, in the adult,

approaches that of gray matter. The size

of the tumor was measured at its maxi-

mum vertical on horizontal dimension on

the coronal Ti-weighted image.

The patients’ hospital charts, office

notes, pituitary hormonal levels, patho-

logic specimens, and surgical notes were

reviewed to attempt to conclusively de-

tenmine a cause of hypenintensity on Ti-

weighted images.

Of the 68 patients examined for pitu-

itany adenomas, 17 were men and 51 were

women, with an average age of 38 years

(range, 10-85 years). Pituitary hormonal

analysis revealed the following distnibu-

tion of secreted hormones: pnolactin, 39

(57%) patients; adrenoconticotrophic hon-

mone, two (3%) patients; growth hon-

mone, two (3%) patients; multiple hon-

mones, two (3%) patients; and nonsecret-

ing adenomas, 23 (34%) patients (Table 1).

Seventeen patients were treated with

surgery and/on radiation therapy alone.

In ten patients, bromocniptine was added

to the surgery on radiation therapy proto-

cot. Nineteen patients were treated solely

with bnomocmiptine, and 22 patients me-

ceived no therapy before undergoing MR

imaging (Table 2).

Thirty-seven patients had macnoadeno-

mas defined as tumors with a maximum

linear dimension of 1 cm on greaten in the

vertical on horizontal plane. Fifteen of the

patients with macnoadenomas were treat-

ed with bnomocniptine, and 14 of the 31

patients with microadenomas received

bromocniptine (Table 3).

The differences in intratumonal hemon-

rhage between the proportions of pnolac-

tinomas and nonsecreting adenomas,

macnoadenomas and microadenomas, and

patients receiving and not receiving bro-

mocniptine were tested with the two-

tailed critical ratio test for P values.

RESULTS

Twenty-eight of the 68 patients

with pituitary adenomas showed evi-

dence of hypenintensity on Ti-

weighted images (Table 4). In nine of

these patients, who had undergone

Abbreviations: TE echo time, TR repeti-tion time.

Table I . . Table 2

Treatment Modalities in Patients with Pituitary Adenomas at Time of MRImaging

Note-BC = bmomocniptine. ACTH = adme-nocorticotrophic hormone.

* Numbers in parentheses are percentages.� t Both patients received bromocniptine.

. I One patient received bromocriptine for� nonsecreting adenoma that did not bleed.

Cause of

No. of

Bright Area

on Ti

Shortened Ti

PostoperativeTreatment Patients Images Fat Packing Bleeding

SurgeryAlone 6 4 4 0(0)With radiation therapy 10 6* 4 1 (10)WithBC 0 0 0 0(0)With radiation therapy and BC 10 1 1 0 (0)

Radiation therapyAlone 1 0 0 0(0)WithBC 0 0 0 0(0)

BConlyNo treatment

1922

134

00

13(68)4 (18)

Total receiving BCTotal not receiving BC

Total

2939

1414

18

i3 (45)5 (13)

68 28* 9 18(26)

I Note-Numbers in parentheses are percentages.

* One patient had ACTH-secreting macroadenoma containing cholesterol.

.- . ,.� . . � � �. . .� �.. _-:��� !. � � -.� � � � � � � .�. .

� Tabte3

� Size of Lesion and Relationship to Intratumoral Hemorrhage

Patient CategoryMacroadenomas Average

(0 37) Size (cm)Microadenomas Average

(n=3i) Size(cm)

Receiving BCReceiving BC and bled

Not receiving BCNot receiving BC and bled

157 (47%)

224 (18%)

2.12.32.02.0

146 (43%)

171 (6%)

0.570.670.580.60

Note-Eleven of 37 (30%) macroadenomas bled and averaged 2.2 cm in size. Seven of 31 (23%)� microadenomasbled and averaged 0.66cm insize(P .697). � � � � � ,, .. � . . �:

Table5

Symptoms of Patients withIntratumoral Hemorrhage

240 . Radiology January i989

Hormonal Analysis, BromoTherapy, and Intratumoral

criptine

Hemorrhage

Hormone Secmeted*No. of

Patients

Prolactin (n = 39)Receiving BCReceiving BC and bled

2613 (50)

NotreceivingBCNot receiving BC and bledTotalbled

130 (0)

13(33)ACTH (,, = 2)

Totalbled 0(0)Growth hormone (n = 2)

Totalbled 0(0)Multiple hormones (it 2)t

Total bled 0 (0)Nonsecmeting adenomas

(n 23)1Total bled 5 (22)

Total 68

surgery, fat packed within the selladuring tnanssphenoidal hypophysec-

tomy accounted for the short Ti vat-

ue on the MR images. This was con-

firmed by the decrease in signal in-

tensity on the second echo of the

proton-density/T2-weighted se-

quence (Fig 1). Of the remaining 19

patients in whom bright Ti foci ap-

peamed, six underwent subsequent

surgery. In five patients, subacute on

chronic hemorrhage was discovered,

and in one patient a fatty substance

exuded from the adenoma (Figs 2-4).In the images of 13 patients who did

not undergo surgery, hypenintense

areas were thought to be caused by

intmatumoral hemorrhage, since the

hemorrhages were hypenintense on

T2-weighted images with the intensi-

ty greater than that of neighboring

fat. This same pattern was seen in the

five patients with intmatumomal hem-

ommhage confirmed at surgery. In the

one patient with the oily tumonal ex-

cnetion, the T2-weighted intensity

decreased and paralleled that of fat.

Of the five patients with intratu-

moral hemorrhage confirmed at

pathologic examination and the 13

patients with suspected hemorrhage,13 (72%) had pnolactin-secmeting ade-

nomas and five (28%) had nonfunc-

tioning adenomas (Table 1). Overall,

57% of pituitary adenomas were pro-

lactin-secreting and 34% were non-

functioning, so an increased frequen-cy of bleeding was seen in prolacti-

nomas. One-third of prolactinomas

bled, whereas 22% of nonsecreting

adenomas bled. This difference was

not significant (P .5 16).

Table 4

Causes of Ti Hypenintensity in 28

Patients

No.of

PatientsCause

Postoperative fat packingFat containing adenomaSubacute or chronic intratumoral

91

hemorrhageConfirmed at surgeryPresumed on the basis of Ti

5

and 12 signal characteristics 13

Total 28

All 13 patients with prolactinomas

that had intratumorat hemorrhage

were receiving bmomocniptine. Thus,

13 of 29 (45%) patients that received

bmomocmiptine had intmatumorat

hemorrhage, compared with five of

39 (i3%) who did not receive bromo-

cniptine (Table 2). This difference

was statistically significant (P < .01).

Thirteen patients with prolactinomas

had not received bromocniptine, and

none of these pmotactinomas showed

intratumonal hemorrhage (Table 1).

The duration of bromocniptine

treatment before hemorrhage was de-

tected ranged from 1 month to 4

SymptomNo. of

Patients

NosymptomsHeadache

106

New visual field cut 3*New cranial nerve deficit 2*Hypopituitarism 0

* Three of these five patients also had head-

� aches.

I, �

years; six patients with hemorrhage

were detected before 12 months of

treatment and seven were detected

after that time. Since the imaging in-

tenvals were not standardized (but

imaging was performed annually in

most asymptomatic patients), the

hemorrhage could have occurred at

any point within these intervals. Ad-

ditionatty, in nearly alt patients, the

preceding study was computed to-

mognaphy (CT), which may not havebeen sensitive to hemorrhage.

Eleven of the i8 (6i%) patientswith intnatumomat bleeding had mac-

moadenomas, a figure that compares

Figure 2. Images of hemorrhagic prolactin-secmeting microadenoma in a 37-year-old worn-

an with amenorrhea and galactonrhea who underwent bromocniptine therapy for 2 years.

The 7-mm lesion is hypenintense just to the right of midline (arrow) on all three coronal se-

quences: (a) Ti-weighted, (b) proton-density, and (c) 12-weighted. The persistent hypenin-

tensity on 12-weighted images suggests subacute on chronic intnatumoral hemorrhage.

b.

Figure 3. Images of intratumoral hemorrhage in a chromophobe macmoadenoma in a 55-

year-old man with visual field deficits and palsy of the night third cranial nerve. The tumor

is hypemintense (arrow) on sagittal (a) and coronal (b) TI-weighted images.

with 54% overall. Thirty percent of

macroadenomas bled, versus 23% ofmicnoadenomas (Table 3). The differ-

ence between the frequency of bleed-

ing in macroadenomas and that in

micmoadenomas was not statistically

significant (P = .697). The average

size of the macnoadenomas that bled

Volume i70 #{149}Number i Radiology S 241

.‘ ..*‘ S ��#{149}��‘*

Figure 1. (a) Preoperative Ti-weighted coronal image of a 44-year-old woman with a hypointense prolactinoma demonstrates bowing ofthe optic chasm (arrows) by the sellar mass. (b-d) Images after transsphenoidal hypophysectomy demonstrate typical MR features of sellar

fat packing. Fat (large arrow) in b is hypemintense on the coronal Ti-weighted sequence. Fat within the sphenoid sinus is also present (smallarrow) in b. Fat (arrow) in c is hypenintense on the proton-density image but shows signal loss (arrow) in d on the second-echo 12-weighted

sequence.

(2.2 cm) was nearly identical to that

of nonhemonrhagic macroadenomas

(2.1 cm). Similarly, the size of mi-

cnoadenomas that bled (0.66 cm) was

less than 0.1 cm greaten than the size

of those that did not bleed (0.57 cm).

The average size of bromocniptine-

treated, bleeding macroadenomas

(2.3 cm) and micmoadenomas (0.67

cm) was comparable to the bleeding

tumors that were not treated with

bromocniptine (2.0 and 0.6 cm, me-

spectively) (Table 3).In most patients with intratumoral

hemorrhage (Table 5), the finding at

MR imaging of blood within the tu-

mon was serendipitous (ten of 18 pa-

tients). Six patients had recent onset

or exacerbation of headaches, three

had new visual field deficits, and two

had extraocular neuromuscular

changes. Three of these patients had

more than one symptom. No patients

had acute change in hormonal func-

tion as their presenting symptom or a

catastrophic presentation.

DISCUSSION

In our institution, fat and fascia are

routinely packed in the sella during

transsphenoidal hypophysectomy.

The fat packed in the sella displays

imaging characteristics comparable

to those of neighboring bone marrow

fat and appears hypenintense on Ti-weighted and proton-density images,

with the intensity decreasing on sec-

ond-echo T2-weighted images.

Chemical shift artifacts from the fat

may also suggest its presence (2).

Thus, after surgery in nine patients,

the bright areas on Ti-weighted im-

ages were confirmed to be fat pack-

ing.

Several reports have described the

histopathologic characteristics of pi-tuitany adenomas after treatment

with bromocniptine, an ergot deniva-

a. c.

242 #{149}Radiology January 1989

b.

Figure 4. Sagittal Ti-weighted (a), coronal Ti-weighted (b), and cononal 12-weighted (c) images demonstrate a hypenintense 3-cm sellan mass

in a 22-year-old man with hypenprotactinemia, long-standing peniphenal visual field cut, and worsening severe headache. He had begun treat-

rnent with bromocniptine 2 months previously, and a CT scan at that time showed no hemorrhage within this macnoadenorna. A chronic clot

within a macnopnolactinoma was found at surgery.

tive that functions as a dopaminergic

agonist that decreases prolactin se-

cretion by the pituitary gland (3-6).

After short-term treatment with bro-

mocniptine, histologic examination

reveals a decrease in cell volume,

particularly within the cytoplasmic

compartment of the cell. There is a

concomitant decrease in cellular nibo-

somes, rough endoplasmic reticulum,

and Golgi complexes, which come-

sponds to a decline in protein syn-

thesis. No definite cell necrosis, vas-

culan thrombosis, pituitary infanc-

tion, or hemorrhage has been

reported from short-term bmomocnip-

tine therapy, although one report

does raise the possibility of central

tumor necrosis after long-term bro-

mocniptine treatment (3-6). Maxi-

mum shrinkage of tumor volume ap-

pears to occur within 3 weeks of in-

stitution of the drug.

On the basis of these histopatho-

logic observations, Weissbuch pro-

posed an explanation for the MR im-

aging signal changes in pituitary ad-

enomas after bnomocniptine therapy

(1). He stated that a reduction in Ti

values may be due to a decrease in

cellular and tissue water seen in in-

voluting cells or to a change in mo-

lecutar mobility that causes the fre-

quency of water-proton collisions to

occur closer to the Lanmom frequency

(1). Bromocniptine treatment can

change molecular mobility by alter-

ing intracellular contents because of

the decreased protein synthesis and

fewer protein-synthesizing organ-

elles. Weissbuch further postulated

that necrotic or edematous cells, in-

creased extnacellular water, and

leaked proteins from dying cells

could alter the Ti-weighted MR im-

aging signal characteristics in the in-

voluting prolactin-secneting adeno-

ma (1). Weissbuch’s explanation of-

fers several obtuse factors that may

contribute to MR imaging signal

changes but does not satisfactorily

clarify why alt adenomas treated

with bromocniptine do not exhibit

this Ti hypenintensity.

Hemorrhage is cleanly one of the

pathologic occurrences that creates a

complex interplay of Ti and T2 relax-

ation times. Intnatumonal hemon-

rhage is common in pituitary adeno-

mas and was suggested in five pa-

tients with nonfunctioning tumors in

this series. In 13 patients treated with

bromocniptine, surgical proof or MR

imaging signal characteristics sug-

gested intratumorat hemorrhage. The

surgical proof in alt patients was ob-

tamed with visual inspection of the

tumor. In many patients, the pitu-

itamy adenoma is soft and friable after

bromocniptine therapy, and micro-

scopic evidence of hemorrhage after

histologic fixing may be attributed to

bleeding at surgery. Nonetheless,

pathologic correlation with intraop-

emative assessment by the surgeon for

intratumomat hemorrhage was ob-

tamed. No area of fat, melanin, or

protein deposition was seen on

pathologic examination; one oily tu-

mom exudate was high in cholesterol

concentration.

The competing effects of intnacellu-

tar and extnaceltutan methemoglobin,

deoxyhemoglobin, hemosidenin, and

magnetic field heterogeneities caus-

ing local magnetic gradients may leadto hypointense on isointense Ti signal

acutely. However, the overwhelming

effect of free extraceltutar methemo-

globin causes Ti shortening in the

subacute and chronic course. T2 hy-

penintensity tends to trail the Ti

shortening and is of less value in de-

tecting hemonrhage but helps in dif-

ferentiating it from fat. Decreased sig-

nat due to the panamagnetic field ef-

fects of hemosidemin is frequently

seen in intracerebral hemorrhage but

much less frequently in intratumonat

hemorrhage in pituitary adenomas,

since these extmacemebnal tumors tack a

blood-brain barmier, and the accumu-lation of hemosidenin-laden macro-

phages does not usually occur (7-9).

In several other reports of findings

at MR imaging in pituitary adeno-

mas, anecdotal cases of postbnomo-

cniptine intnatumorat bleeding have

been described (10-13). To our

knowledge, ours is the first lange se-

nies to emphasize this occurrence in

bnomocniptine-treated pnolactinomas.

The frequency of intratumonal hem-

omnhage in all pituitary adenomas has

been gathering increased attention,

as MR imaging has made hemonnhag-

ic foci easier to detect. Overall, 18 of

68 (26%) pituitary adenomas showed

evidence of hemorrhage on MR im-

ages. Thirteen of 29 (45%) patients

treated with bromocniptine showed

intnatumonat hemorrhage, white only

five of 39 (13%) patients not treated

with bromocriptine had evidence of

hemorrhage (P < .01). These num-

bems compare with a surgical report

of intnatumoral hemorrhage in i6.i%

of 560 cases of pituitary adenoma

(14). In the study by Wakai et at, no

significant correlation with size, hon-

monal secretion, or sex was found.

They also raised the possibility of the

influence of bromocniptine in the de-

velopment of intnatumonat hemom-

rhage (14).

Volume 170 #{149}Number 1 Radiology #{149}243

The time course of hemorrhage af-

ten institution of bromocniptine then-

apy was variable; the patients in our

study were nearly evenly split into

those treated less than and more than

1 year. Of interest were the clinical

findings that 55% of the patients

were asymptomatic. Wakai et at

found that 43% of 93 patients with

intratumonal hemorrhage were

asymptomatic (i4). Only 39% of their

patients had a major episode associat-ed with bleeding, and, as in our

study, the symptoms were ‘due to

mass effect (eg, headache, visual

changes, on new cranial nerve abnor-

malities [i4J). No symptoms due to

hormonal abnormalities on panhypo-

pituitanism were found. The “catas-

trophe” of pituitary apoplexy was

not the presenting complaint in any

of the patients in our study. This

leads one to wonder if the headache

associated as a common side effect of

bromocniptine therapy may be attnib-

utable to microscopic bleeding into

the tumor. Could the headache be a

sign that these patients are at risk for

subsequent macroscopic intratumorat

hemorrhage?

Because the results of this prelimi-

nary paper suggest that bromocnip-

tine may induce intratumonal hemon-

rhage, we believe a prospective study

should be undertaken in patients

with pituitary adenomas who are me-

ceiving this drug. The intensity

changes before and after institution

of bnom#{244}cziptine, the frequency of

bromocniptine-induced headaches

with and without evidence of intra-

tumonal hemorrhage, and the fre-

quency of headaches with and with-

out bnomocniptine therapy should be

examined. The high frequency of in-

tnatumoral hemorrhage in pituitary

adenomas treated and not treatedwith bromocriptine should be con-

nobonated. U

Acknowledgments: Many thanks go to Betty

Bmandt and Valerie Tsafos for the preparationof the manuscript and to Harold Kundell, MD,

for assistance with statistical analysis.

References1. Weissbuch SS. Explanation and implica-

tions of MR signal changes within pitu-itary adenomas after bromocniptine thera-py. AJNR 1986; 7:214-216.

2. Nishimura K, Fujisawa I, Togashi K, et al.Posterior lobe of the pituitary: identifica-

tion by lack of chemical shift artifact inMR imaging. J Comput Assist Tomogr1986; 10:899-902.

3. Anniko M, Wersall J. Morphologicalchanges in bromocriptine-treated pitu-itary tumoums. Acta Otolamyngol 1983;96:337-353.

4. Barrow DL, Tindall GT, Kovacs K, ThomnerMO, Horvath E, Hoffman JC Jr. Clinicaland pathological effects of bromocniptine

on prolactin-secreting and other pituitarytumors. J Neumosumg 1984; 60:1-7.

5. Gen M, Uozumi T, Ohta M, Ito A, KajiwaraH, Mon S. Necrotic changes in prolacti-nomas after long term administration ofbmomocriptine. J Clin Endocrinol Metab1984; 59:463-470.

6. Rengachary 55, Tornita 1, Jeffenies BF, Wa-tanabe I. Structural changes in humanpituitary tumor after bromocriptine thera-py. Neurosurgery 1982; 10:242-251.

7. Gomoni JM, Grossman RI, Goldberg HI,Zimmerman RA, Bilaniuk LT. Intracrani-al hernatomas: imaging by high-field MR.Radiology 1985; 157:87-93.

8. Zimmerman RD. Deck MDF. Intracranialhematomas: imaging by high-field MR(letter). Radiology 1986; 159:565.

9. Zimmerman RD. Heiem LA, Snow RB, LiuDPC, Kelly AB, Deck MDF. Acute intra-cranial hemorrhage: intensity changes onsequential MR scans at 0.5 I. AJR 1988;150:651-661.

10. Kamnaze MG, Sartor K, Winthrop JD, GadoMH, Hodges FJ III. Suprasellar lesions:evaluation with MR imaging. Radiology1986; 161:77-82.

1 1. Kucharczyk W, Davis DO, Kelly WM, SzeG, Norman D, Newton TH. Pituitary ade-nomas: high-resolution MR imaging at 1.5T. Radiology 1986; 161:761-765.

12. Kulkamni MV, Lee KF, McArdle CB, Yeak-ley JW, Haar FL. 1.5-1 MR imaging of pi-tuitamy microadenomas: technical consid-

emations and CT correlation. AJNR 1988;9:5-11.

13. Pojunas KW, Daniels DL, Williams AL,Haughton VM. MR imaging of prolactin-secreting microadenomas. AJNR 1986;7:209-213.

14. Wakai 5, Fukushima T, Teramoto A, SanoK. Pituitary apoplexy: its incidence and

clinical significance. J Neurosurg 1981;55:187-193.