Teaching Course 11 MAGNIMS Quantitative MR imaging in the management of multiple sclerosis ·...
Transcript of Teaching Course 11 MAGNIMS Quantitative MR imaging in the management of multiple sclerosis ·...
Teaching Course 11 MAGNIMS Quantitative MR imaging in the management of multiple sclerosis
Chairs: A. Rovira (Barcelona, ES)
H. Vrenken (Amsterdam, NL)
23 Focal lesion load quantitative MR measures
H. Vrenken (Amsterdam, NL)
24 Brain and spinal cord volume measurements
A. Rovira (Barcelona, ES)
25 Advanced quantitative brain MR measures
M. Filippi (Milano, IT)
1
Wednesday, 7 October, 10.30 – 12.00
Teaching Course 11 MAGNIMS
Quantitative MR imaging in the management of multiple sclerosis
Chairs:
A. Rovira (Barcelona, ES)
H. Vrenken (Amsterdam, NL)
Focal lesion load quantitative MR measures
H. Vrenken (Amsterdam, NL)
Brain and spinal cord volume measurements
A. Rovira (Barcelona, ES)
Advanced quantitative brain MR measures
M. Filippi (Milan, IT)
2
Teaching course 11: “Quantitative MR imaging in the management of multiple sclerosis”
Focal lesion load quantitative measures
H. Vrenken, VU University Medical Center, Amsterdam, the Netherlands.
Focal lesions in the white matter (WM) were long considered the pathological hallmark of
MS, and play an important role in the diagnostic criteria for MS (Polman, 2011). And
although grey matter lesions and atrophy of both white and grey matter have gained
prominence (e.g., Geurts, 2012; Sastre-Garriga, 2004; Shiee, 2012), focal WM lesions still
remain important to this day. Most work on automated measurement has been done on WM
lesions in the brain, while far less has been done towards contrast-enhancing, grey matter, or
spinal cord lesions. In view of the usability in the clinical management of MS, WM lesions
are also most easily accessible through standard clinical MS imaging protocols. We therefore
focus this lecture on the quantitative measurement of brain WM lesion volumes and their
change over time.
Since many of the disease modifying treatments currently available for MS act primarily on
the inflammatory component of the disease, the focal WM lesions provide an important
indicator of the success of treatment. As we will see in this lecture, the exploitation of
quantitative measures to assess focal lesion loads is lagging.
Current use of MRI of lesions
Diagnosis
As prescribed in the diagnostic criteria of MS (Polman, 2011), the emphasis in the diagnostic
evaluation of focal lesions is on the number of lesions and their locations. Although the size
3
of the lesions plays a role in discriminating MS from other disorders, lesion sizes or total
lesion volumes are not generally quantitatively assessed in routine clinical diagnosis.
Prognosis
While some studies have related (early) lesion load to later clinical worsening (e.g., Popescu,
2013), there are no guidelines on how such information might be used in individual patient
cases to predict the evolution of disease. Leaving aside the problem of how to account for
variable treatment regimens in assessing this relation, the absence of reliable methods to
measure and gauge quantitative focal lesion loads certainly also contributes to this problem.
Treatment response
The latest treatments for MS are successful in reducing the numbers of new relapses and new
lesions. One might therefore argue that in a patient with a substantial amount or volume of
new lesions under treatment, the treatment may be partially or wholly ineffective, and a
change of treatment may be considered. However, it is also possible that the same patient may
have had a much less favorable course without the treatment, and therefore a change of
treatment may not result in any improvement. Here too, there are no guidelines to help the
clinician make their decision based on an observed increase in focal lesion load. And again,
the absence of reliable methods for the assessment of lesion volume change hampers its
clinical use.
Questions
Three key questions may be identified, that are inter-connected, and together summarize the
challenges.
1. Will it be useful to include quantitative measures of focal lesion load?
4
2. Is it possible to measure them in a clinical setting?
3. How will the clinician use the focal lesion load measures?
We will discuss these questions and their answers now.
Will it be useful to include quantitative measures of focal lesion load?
First we have to assess the clinical need for quantitative measures of focal lesion load.
In diagnosing MS, precise volumetric information of the focal lesion volumes does not appear
to add information over the existing information.
By contrast, to assess treatment efficacy, reliable quantitative information on lesion volume
change over time may yield important insights into individual patient pathology worsening –
information that is currently only available in a qualitative sense. Since there are now several
treatments that can reduce the incidence of new lesions, precise quantitative measurement of
lesion change may help to identify the best treatment for an individual patient.
Possibly, quantitative lesion volume and lesion volume change assessment may also help to
improve prognostic accuracy, if the influence of other factors, most notably past and current
medication, can be adequately modeled.
Clearly, on both issues, appropriate steps will also have to be taken in order to allow the
neurologists to use this information in their practice.
Is it possible to quantitatively measure focal lesion load (change) in a clinical setting?
Requirements
If the focal lesion load and its change over time are to be measured in a clinical setting, two
things are required. First, image acquisition should include suitable images for assessing small
lesion changes over time. Preferably these should be of the 3D Flair or 3D T2-weighted type,
with sufficiently small voxel size to allow reliable quantification of small changes between
5
imaging time-points (see, e.g., Vrenken, 2013). The 2015 revision of the recommended
standardized MS imaging protocol specifically emphasizes the use of 3D acquisition for this
purpose (http://www.mscare.org/?page=MRI_protocol; Simon, 2006).
Second, quantification of lesion loads and their change over time should be reliably possible.
An important challenge for both cross-sectional and change measurement, is that the
automated methods should perform well independent of the scanner and pulse sequence used
to acquire the data.
Lesion volume measurement
A very nice review of automated methods for MS lesion segmentation is provided by
Mortazavi and colleagues (Mortazavi, 2012). A fairly large number of methods have been
developed, based on various principles which may, with Mortazavi, be characterized as data
driven, statistical, “intelligent”, or deformable contours methods. Some methods require
additional reference images with lesion labels in order to work, others are fully independent.
Many methods require optimization of parameters to improve performance for the data at
hand.
Despite serious advances over the past years, including recently (e.g., Guizard, 2015; Roura,
2015; Wang, 2015), the performance of automated methods still leaves substantial room for
improvement. The comparison of segmentations by automated methods with those by expert
manual raters shows that at the voxel level, there is substantial disagreement between them.
Dice similarity indices, which can range between 0 and 1 where 1 indicates perfect agreement
and 0 no overlap, seldom reach values above 0.7 (Mortazavi, 2012), implying a volume of
disagreement that is on the order of 30% of the lesion volume. Rates of false positive voxels,
i.e., image voxels labelled as lesion voxels by the automated method but not by the expert, are
generally on the order of 10-20%. The same holds for the rates of false negative voxels, i.e.,
6
those lesion voxels missed by the automated methods; these too are generally around 10-20%
for optimized methods.
It should be noted here that the manual expert labels, often referred to as the “gold standard”,
generally exhibit substantial variability both within and between raters. For example, in Flair
images with 1x1x1 mm resolution, the Dice similarity index between two expert manual
raters was 0.84 (Steenwijk, 2013). This highlights one of the difficulties in further optimizing
automated methods: if the manual reference is not perfect, an automated segmentation that is
superior to the manual one will be interpreted as being inferior because of its disagreement
with the imperfect reference. By combining multiple labels from different raters e.g. using
STAPLE software (Warfield, 2004), such problems may be somewhat alleviated, but the
fundamental problem that one cannot outperform the reference remains.
For individual patient treatment, these errors have to be reduced, or the uncertainty in the
automated methods has to be taken into consideration when interpreting the lesion volumes in
a clinical situation.
Lesion volume change measurement
For a very nice overview of MS lesion change detection and quantification methods, the paper
by Lladó and colleagues is recommended (Lladó, 2012). Methods may be separated into
detection of lesion change, and quantification of the volumetric effects involved, including
both the change of the actual lesions, and the change of the surrounding tissue due to any
mass effects that may be present, e.g., due to edema. In case of two time points, subtraction of
the images obtained on the two time points may be of help, especially for visual change
analysis. In case of multiple time points, modeling of the signal intensities as a function of
time may be more appropriate.
7
Most methods have been tested in relatively small datasets, but have yielded reasonable
reproducibility, with inter-scan coefficients of variation down to just under 1%, and
reasonable agreement with manual assessment yielding Dice similarity coefficients around 0.6
to 0.8. Lladó and colleagues conclude that improvement is needed before clinical application
is feasible. Recently, there have been efforts directed at obtaining more reliable measures of
lesion volume change in MS in a coordinated way (http://iacl.ece.jhu.edu/MSChallenge).
How will the clinician use the focal lesion load measures?
So far, we have addressed requirements and performance of current methods from a technical
perspective. An equally important perspective perhaps, is how clinicians would use the
information on lesion volume and change of lesion volume if reliable measurements were to
become available. We leave aside here, apart from this brief mention, the necessity to have
analysis methods developed according to certified protocols, and regulatory approval for their
use in a clinical setting, according to standardized guidelines. What is of concern to us now, is
that such guidelines for clinical use of lesion volumes and lesion volume changes, have to be
developed in order to give the treating physician sufficient guidance to make well-informed
decisions incorporating the lesion volume information. It will likely not be possible to provide
precise information on how to interpret an observed lesion volume increase in a patient for
any disease duration, and any possible combination of past treatments, relapses, and
progression of disability. Nevertheless, it is important that at least basic clinical guidelines
become available. Even if imaging-related technical challenges are all overcome, use of lesion
volume measures in the clinic can only take place if supported by such guidelines.
Conclusion
8
While much work has been done over the past decades, there are still no completely
satisfactory automated methods for MS WM lesion volume measurement or for lesion volume
change measurement. Remaining disagreements with an, albeit imperfect, manual reference,
generally remain at least 10-20% of the lesion volume or lesion volume change under
investigation, warranting some caution for individual patient application. Robustness across
scanners and pulse sequences remains an important requirement to provide a feasible clinical
tool.
References
Geurts JJ, Calabrese M, Fisher E, Rudick RA. Measurement and clinical effect of grey
matter pathology in multiple sclerosis. Lancet Neurol. 2012;11(12):1082-92.
Guizard N, Coupé P, Fonov VS, Manjón JV, Arnold DL, Collins DL. Rotation-
invariant multi-contrast non-local means for MS lesion segmentation. Neuroimage
Clin. 2015 May 13;8:376-89. doi: 10.1016/j.nicl.2015.05.001. eCollection 2015.
Jain S, Sima DM, Ribbens A, Cambron M, Maertens A, Van Hecke W, De Mey J,
Barkhof F, Steenwijk MD, Daams M, Maes F, Van Huffel S, Vrenken H, Smeets D.
Automatic segmentation and volumetry of multiple sclerosis brain lesions from MR
images. Neuroimage Clin. 2015 May 16;8:367-75. doi: 10.1016/j.nicl.2015.05.003.
eCollection 2015.
Lladó X, Ganiler O, Oliver A, Martí R, Freixenet J, Valls L, Vilanova JC, Ramió-
Torrentà L, Rovira A. Automated detection of multiple sclerosis lesions in serial brain
MRI. Neuroradiology. 2012 Aug;54(8):787-807. doi: 10.1007/s00234-011-0992-6.
Epub 2011 Dec 20.
Polman CH, Reingold SC, Banwell B, Clanet M, Cohen JA, Filippi M, Fujihara K,
Havrdova E, Hutchinson M, Kappos L, Lublin FD, Montalban X, O'Connor P,
9
Sandberg-Wollheim M, Thompson AJ, Waubant E, Weinshenker B, Wolinsky JS.
Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria.
Ann Neurol. 2011 Feb;69(2):292-302. doi: 10.1002/ana.22366.
Popescu V, Agosta F, Hulst HE, Sluimer IC, Knol DL, Sormani MP, Enzinger C,
Ropele S, Alonso J, Sastre-Garriga J, Rovira A, Montalban X, Bodini B, Ciccarelli O,
Khaleeli Z, Chard DT, Matthews L, Palace J, Giorgio A, De Stefano N, Eisele P, Gass
A, Polman CH, Uitdehaag BM, Messina MJ, Comi G, Filippi M, Barkhof F, Vrenken
H; MAGNIMS Study Group. Brain atrophy and lesion load predict long term
disability in multiple sclerosis. J Neurol Neurosurg Psychiatry. 2013 Oct;84(10):1082-
91. doi: 10.1136/jnnp-2012-304094. Epub 2013 Mar 23.
Roura E, Oliver A, Cabezas M, Valverde S, Pareto D, Vilanova JC, Ramió-Torrentà
L, Rovira À, Lladó X. A toolbox for multiple sclerosis lesion segmentation.
Neuroradiology. 2015 Jul 31. [Epub ahead of print]
Sastre-Garriga J, Ingle GT, Chard DT, Ramió-Torrentà L, Miller DH, Thompson AJ.
Grey and white matter atrophy in early clinical stages of primary progressive multiple
sclerosis. Neuroimage 2004;22(1):353-9.
Shiee N, Bazin PL, Zackowski KM, Farrell SK, Harrison DM, Newsome SD,
Ratchford JN, Caffo BS, Calabresi PA, Pham DL, Reich DS. Revisiting brain atrophy
and its relationship to disability in multiple sclerosis. PLoS One. 2012;7(5):e37049.
Simon JH, Li D, Traboulsee A, Coyle PK, Arnold DL, Barkhof F, Frank JA,
Grossman R, Paty DW, Radue EW, Wolinsky JS. Standardized MR imaging protocol
for multiple sclerosis: Consortium of MS Centers consensus guidelines. AJNR Am J
Neuroradiol. 2006 Feb;27(2):455-61.
Steenwijk MD, Pouwels PJ, Daams M, van Dalen JW, Caan MW, Richard E, Barkhof
F, Vrenken H. Accurate white matter lesion segmentation by k nearest neighbor
10
classification with tissue type priors (kNN-TTPs). Neuroimage Clin. 2013 Oct
14;3:462-9. doi: 10.1016/j.nicl.2013.10.003. eCollection 2013.
Vrenken H, Jenkinson M, Horsfield MA, Battaglini M, van Schijndel RA, Rostrup E,
Geurts JJ, Fisher E, Zijdenbos A, Ashburner J, Miller DH, Filippi M, Fazekas F,
Rovaris M, Rovira A, Barkhof F, de Stefano N; MAGNIMS Study Group.
Recommendations to improve imaging and analysis of brain lesion load and atrophy in
longitudinal studies of multiple sclerosis. J Neurol. 2013 Oct;260(10):2458-71. doi:
10.1007/s00415-012-6762-5. Epub 2012 Dec 21. Review.
Wang R, Li C, Wang J, Wei X, Li Y, Zhu Y, Zhang S. Automatic segmentation and
volumetric quantification of white matter hyperintensities on fluid-attenuated
inversion recovery images using the extreme value distribution. Neuroradiology. 2015
Mar;57(3):307-20. doi: 10.1007/s00234-014-1466-4. Epub 2014 Nov 19.
Warfield SK, Zou KH, Wells WM. Simultaneous truth and performance level
estimation (STAPLE): an algorithm for the validation of image segmentation. IEEE
Trans Med Imaging. 2004 Jul;23(7):903-21.
1
Teaching Course 11
MAGNIMS Quantitative MR imaging in the management of MS
“Brain and spinal cord volume measurements in MS”
Àlex Rovira Unitat de RM. Servei de Radiologia
Hospital Vall d’Hebron Barcelona
www.magnims.eu
New Therapies Allow Re-evaluation of Treatment Goals and Greater Consideration of the Patient Perspective
1. IFNB MS Study Group. Neurology 1993;43:655-61; 2. PRISMS Study Group. Lancet 1998;352:1498-504; 3. Kappos L et al. N Engl J Med 2010;362:387-
401; 4. Cohen JA et al. Lancet 2012;380:1819-28; 5. Coles AJ et al. Lancet 2012;380:1829-39; 6. O'Connor P et al. N Engl J Med 2011;365:1293-303; 7. Twyman C et al. AAN 2013. Poster. P07.098
Current Goals3-7 Traditional Goals1,2
Delay disability progression
Reduce relapse frequency
Freedom from clinical
disease activity
Freedom from MRI disease
activity
MRI disease activity
Prevent new /enlarging T2 lesion and new Gd+ T1 lesions
Clinical disease activity
New Therapies Allow Re-evaluation of Treatment Goals and Greater Consideration of the Patient Perspective
1. IFNB MS Study Group. Neurology 1993;43:655-61; 2. PRISMS Study Group. Lancet 1998;352:1498-504; 3. Kappos L et al. N Engl J Med 2010;362:387-
401; 4. Cohen JA et al. Lancet 2012;380:1819-28; 5. Coles AJ et al. Lancet 2012;380:1829-39; 6. O'Connor P et al. N Engl J Med 2011;365:1293-303; 7. Twyman C et al. AAN 2013. Poster. P07.098
Current Goals3-7 Traditional Goals1,2
Delay disability progression
Reduce relapse frequency
Freedom from clinical
disease activity
Freedom from MRI disease
activity
MRI disease activity
Prevent new /enlarging T2 lesion and new Gd+ T1 lesions
Clinical disease activity
Improve disability
Improve QOL and Patient reported
outcomes
Emerging Goals1,2
Reduction in brain atrophy
MR imaging Biomarkers in Multiple Sclerosis
Purpose Measure
Disease activity New/enlarging T2 lesions Gadolinium enhancing lesions CUAL (combined unique activity lesions)
Disease burden or load T1/T2 lesion load, MTR
Demyelination / Remyelination MTR DTI (FA, MD) T2 mapping
Neuroaxonal loss Atrohy: BPF, NBV (global, regional), PBVC, UCCA T1 lesion load MRS (NAA) DTI (FA, MD)
Brain plasticity Functional (BOLD) MRI
Iron content R2 maps, QSM
Two overlapping pathogenetic components of MS
Preclinical Relapsing–remitting Secondary progressive
Lesion load T2
Brain volume atrophy measures
Disability
Time (years)
Inflammation Neurodegeneration
Anti-inflammatory/ immunomodulatory
therapies Myelin/neural repair/ neuroprotection
Active lesions
Weight of evidence suggests that strongest surrogate marker correlation with disability in MS is brain (spinal cord) atrophy
MS, multiple sclerosis. Modified from Roland Martin, personal communication.
• Global marker of neuroaxonal loss in MS1,2
• Occurs in all clinical stages of MS at a rate of 0.5–1.3%/year vs. 0.1–0.3%/year in healthy subjects1,2
• Measures of global brain atrophy are robust, sensitive and relatively easy to standardise2
• Correlates with and predicts disability progression on group-wise level
• Generally measured on 2D/3D T1-weighted images
BPF, brain parenchymal fraction; SIENA, Structural Image Evaluation, using Normalisation, of Atrophy; VBM, voxel-based morphometry 1. Filippi M & Agosta F. J Magn Reson Imaging 2010;31:770-88; 2. Giorgio A, et al. Neuroimaging Clin N Am 2008;18:675-86
1997 2002 2003 1995 2005
Road map for validating brain atrophy as a proxy for (long-term) disability progression
2
• BVL measurement is prone to variability due to factors such as the pseudoatrophy effect, steroid use, dehydration, comorbidities (smoking, alcohol, BMI) BMI, body mass index; BVL, brain volume loss. 1. De Stefano N et al. Oral presentation S13.006 at AAN 2014; 2. Filippi M, Agosta F. J Magn Reson Imaging 2010; 3. Giorgio A et al. Neuroimaging Clin N Am 2008; 4. Barkhof F et al. Nat Rev Neurol 2009; 5. Bermel RA, Bakshi R. Lancet Neurol 2006; 6. Kappos L et al. Oral FC1.5 presented at ECTRIMS-ACTRIMS 2014
Threshold of -0.4% / year proposed as it falls between the ranges of mean annual BVL in healthy adults and patients with MS6
Healthy adults (-0.1 to -0.3% per year)1-3
An
nu
al B
VL
(%)
Untreated patients with MS (-0.5 to -1.35% per year)4,5
-0.4% per year
Brain volume loss Healthy adults vs. multiple sclerosis
-1.0
-0.5
0.0
0.5
1.0
All patientsN=176
2nd attackN=76
MRI-only MSN=32
Pure CISN=68
**
p=0.003 Brain volume change rates
(mean with 95% confidence interval)
Mult Scler J 2013
Clinical impact of early brain atrophy in CIS
BPF, brain parenchymal fraction; EDSS, Expanded Disability Status Scale; IFN β, beta-interferon; MSFC, Multiple Sclerosis Functional Composite; RRMS, relapsing–remitting MS.
1. Fisher E, et al. Mult Scler 2000; 2. Fisher E, et al. Neurology 2002.
Change in BPF from baseline to year 2 (%) (N = 138)
1.3 to –0.26 –0.27 to –0.89 –0.90 to –1.7 –1.7 to –6.5
Patients with EDSS score ≥ 6 at year 8 (%)2
14.3 29.4 28.6 55.9
Measures of overall brain atrophy predict disability and disability progression
• Follow-up study ≈ 8 years of 160 RRMS patients included in a 2-year Phase III trial of IFNβ-1a
– BPF was significantly correlated with EDSS score and MSFC score at baseline, and after 1, 2 and 8 years1
– The change in BPF during the 2-year study significantly correlated with EDSS1 and MSFC1,2 at year 8
• Multi-centre study in 977 patients with MS
• GM, but not WM, volumes decreased with advancing disease stage
CIS, clinically isolated syndrome; EDSS, Expanded Disability Status Scale; GM, grey matter; nGMV, normalized grey matter volume; nWMV, normalized white matter volume; PASAT, Paced Auditory Serial Addition Test;
PP, primary progressive; RR, relapsing–remitting; SP, secondary progressive; WM, white matter. Roosendaal SD, et al. Mult Scler 2011
* p < 0.01
** p < 0.001
• GM volume predicted physical disability (EDSS) and cognitive impairment (PASAT) better than WM volume or T2 lesion volume
• Disability and cognitive impairment are better predicted by GM than WM volume
nW
MV
CIS RRMS SPMS PPMS
0.85
0.80
0.75
0.70
0.65
** *
nG
MV
CIS RRMS SPMS PPMS
0.90
0.85
0.80
0.75
0.70
** **
** *
Clinical relevance of WM and GM atrophy
• Regions of GM loss (A) participants with RRMS + CIS
• (B) RRMS only subgroup
Deep grey matter bilaterally / pre- and post-central bilaterally / cingulate
bilaterally
CIS, clinically isolated syndrome; GM, grey matter; RRMS, relapsing–remitting MS. Lansley J, et al. Neurosci Biobehav Rev 2013.
A. RRMS including CIS*
B. RRMS without CIS
*Ceccarelli et al. 2008; Henry et al. 2008; Raz et al. 2010; Audoin et al. 2010
Localized grey matter atrophy in multiple sclerosis: A meta-analysis of
voxel-based morphometry studies and associations with functional disability
Lansley J, Mataix-Cols D, Grau M, Radua J, Sastre-Garriga J
GM atrophy in MS
Among all GM structures, the thalamus is the most consistently implied in MS
Thalamic atrophy • Occurs within the first five years of MS onset, when most patients are still minimally
disabled (Davies et al., 2005; Henry et al., 2009).
• Inversely correlated with lesion load in MS (Shiee et al., 2012)
• Associated with a wide range of clinical manifestations: cognitive decline, motor deficits, disability, fatigue, painful syndromes, and ocular motility disturbances (Minagar et al., 2013; Shiee et al., 2012, Batista et al., 2012; Hulst et Geurts, 2012).
• Associated with conversion from CIS to definite MS over 2 years (Zivadinov et al., 2013;
Vaneckova et al., 2013).
Subcortical GM atrophy in MS
NeuroQuant FIRST-FSL Freesurfer
3
• EDSS score:
– In early RRMS patients (N = 425): most significant correlations between cortical thickness and EDSS score for the anterior pole of the left inferior temporal gyrus, the middle and superior frontal gyri bilaterally, and the bilateral anterior cingulate cortex and left post-central cortical area
• Cognitive impairment:
– Hippocampal atrophy associated with impaired memory encoding and retrieval
– Atrophy of the putamen predicts poor performance in tests assessing cognitive performance
• Fatigue:
– Atrophy in the frontal lobes and progression of corpus callosum atrophy over 5 years associated with severity of fatigue
EDSS, Expanded Disability Status Scale; RRMS, relapsing–remitting MS. Charil A, et al. NeuroImage 2007; Rocca MA, et al. J Neurol 2012.
Clinical correlates of regional brain atrophy
Oral drugs Fingolimod FREEDOMS Ph. III, n=1033, 2y
FTY 0.5mg vs FTY 1.25mg vs placebo PBVC (SIENA)
Overall, 30% reduction in pooled FTY-treated arms
TRANSFORMS Ph. III, n=1153, 1y FTY 0.5mg vs FTY 1.25mg vs im IFN-β-1a
PBVC (SIENA)
FTY 0.5mg: 31% reduction FTY 1.25mg: 33% reduction
Ph. III, n=799, 1y extension phase FTY 0.5mg and FTY 1.25mg open-label
PBVC (SIENA)
FTY 0.5mg: 51% reduction FTY 1.25mg: 62% reduction
FREEDOMS II Ph. III, n=1033, 2y FTY 0.5mg vs FTY 1.25mg vs placebo
PBVC (SIENA)
Overall, 45% reduction in FTY-treated arms
Dimethyl-fumarate DEFINE Ph. III, n=540, 2y Dimethyl-fumarate BID vs TID vs placebo
PBVC (SIENA)
6mo. to 2y: 30% reduction in the BID dose, negative results for the TID dose
Teriflunomide TEMSO Ph. III, n=1074, 2y Teriflunomide 7mg vs 14mg vs placebo
BPF, GMF, WMF changes
BPF and GMF change: No significant differences WMF: 83% (7mg) and 164% (14mg) relative change in teriflunomide-treated arms
Monoclonal antibodies
Natalizumab AFFIRM Ph. III, n=942, 2y Natalizumab vs placebo
BPF change
BL to 2y: no significant differences BL to 1y: 40% greater atrophy in NAT-treated arm 1 to 2y: 44% reduction in NAT-treated arm
SENTINEL Ph. III, n=1003, 2y IFN-β-1a im+natalizumab vs IFN-β-1a im+placebo
BPF change
BL to 2y: no significant differences BL to 1y: 19% greater atrophy in NAT-treated armf 1 to 2y: 23% reduction in NAT-treated arm
Alemtuzumab CAMSS223 Ph. II, n=334, 3y Alemtuzumab 12mg vs 24 mg vs sc IFN-β-1a 44 g
BPF change
72% reduction in alemtuzumab-treated arm g
CARE-MS-I Ph. III, n=581, 2y Alemtuzumab 12mg vs sc IFN-β-1a 44 g
BPF change
42% reduction in alemtuzumab-treated arm
CARE-MS-II Ph. III, n=840, 2y Alemtuzumab 12mg vs 24 mg vs sc IFN-β-1a 44 g
BPF change
24% reduction in pooled alemtuzumab-treated arms
Drug Clinical Trial Characteristics Measure a Results b, c
Reported data from CTs
Effects of DMTs on Brain Atrophy (RRMS)
Vidal-Jordana et al. J Neurol 2015
Capturing temporal patterns of structural brain changes requires: • adequate MRI protocols • accurate and robust image analysis tools
It might be challenging the use of MRI-based assessments on individual basis to reliably classify subjects into patient versus normal, as opposed to assessing significant group differences
Atrophy measurements
Registration-based
Global or regional automated segmentation
BPF=brain parenchymal fraction BPF: brain volume/ intracranial volume
NBV= Normalized brain volume
single time-point
PBVC=percentage of brain volume loss
two time-points
IC volume
Brain volume
MRI Time 1
MRI Time 2
Co-registered image
Segmentation-based
• Heterogeneous results • Influenced by the quality of T1-
weigthed images • Not recommended for longitudinal
studies • Designed to analyse regional volumes
• Sensitive to changes over time • Robust and less sensitive to the quality
of MR imaging acquisitions • Recommended for longitudinal studies • Not usually designed to analyse regional
volume changes over time
Brain atrophy in MS: MRI-based methods
Measurement errors
0.62% 0.82% 0.13% 0.17% 0.77% 1.06% 0.23% median
Courtesy of Wim Van Hecke. Icometrix
0.5–1.3%/year MS
0.1–0.3%/year in HC
Multiple regression model used to calculate the normalised brain volume (NBV) according to baseline characteristics (age, gender, disease duration, EDSS, T2LL)
NBV distance from expected = NBV observed-NBV expected
Small NBV: • Higher risk of disability progression • Higher treatment effect
Sormani MP, et al. ECTRIMS 2014 PS 12.3
Brain atrophy in MS: segmentation-based
Placebo group Fingolimod group
EDSS, Expanded Disability Status Scale; LL, lesion load
4
3D T1
Brain regional atrophy measurement (regional)
GM
WM
CSF
CSF, cerebrospinal fluid; GM, grey matter; SIENA, Structural Image Evaluation, using Normalisation, of Atrophy; WM, white matter.
GM and WM volumes can also be calculated separately, using SIENAx, SPM
Lesion filling and GM atrophy measure
SPM 52% of lesion voxels misclassified as GM (Chard, et al. Brain 2002) Effect: Lesion volume GM volume
0
2
4
6
8
0 50 100 150
lesion volume (ml)
err
or
in g
ray m
att
er
vo
lum
e (
%)
test-retest
CIS
RR
PP
Error in GM volume estimation as a function of lesion volume for all patients included.
D. Pareto. Hospital Vall d’Hebron
Lesion volume errors: 0.3-2.5%
Chard, et al. 2010; Battaglini et al. 2011; Popescu et al. 2014
Methods
Expectation maximization segmentation (Van Leemput et al., 2001)
Lesion-TOADS (Shiee et al., 2010)
Lesion segmentation toolbox (LST) (Schmidt et al., 2012)
CASCADE (Damangir et al., 2012)
kNN-Tissue Type Prior (Steenwijk et al., 2013)
HAMMER-White matter lesion (Yu et al., 2002)
Lesion segmentation tool for 3D slicer (Scully et al., 2010)
Msmetrix (Jain et al., 2015)
Lesion segmentation
Lesion segmentation toolbox (LST)
Lesion filling and GM atrophy measure
SPM 52% of lesion voxels misclassified as GM (Chard, et al. Brain 2002) Effect: Lesion volume GM volume
FLAIR+LST mask
Segmentation (without and with LST)
Baseline 1 year 2 years
SIENA [Structural Image Evaluation using Normalization of Atrophy]
Registration-based measures
Baseline 1 year 2 years
SIENA (Structural Image Evaluation, using
Normalisation, of Atrophy)
SIENA (Structural Image Evaluation, using
Normalisation, of Atrophy)
aPBVC first year: -1,6% aPBVC second year: -1,8%
SIENA [Structural Image Evaluation using Normalization of Atrophy]
Registration-based measures
5
Segmentation-based measures
MSmetrix
baseline 1 year 5 years
RRMS. EDSS 1.5 aPBVC: -1.51%
aPBVC (grey matter): -1.60% T2LL: 60cc to 85cc
EDSS 3.0
Segmentation/registration-based measures
De Stefano N, et al. J Neurol Neurosurg Psychiatry 2015
Cut-off values that identify a PBVC distinguishing healthy controls from MS patients
Brain atrophy in MS: registration-based
In 0-7 years: 46%-8% (Rotstein et al. JAMA Neurol 2014)
NEDA at 2 years had a positive predictive value (78%) for no progression at 7 years
Associated with lower brain volume loss
Concept of no evidence of disease activity (NEDA)
Free of gad
and new T2
lesions
Free of
disability
progression
Free of
disease
Free of
relapses
Free of clinical activity/progression
Free of MRI activity
NEDA 3
Increase in EDSS score at 6 months of: • 1.5 points if baseline score is 0 • 1.0 point if baseline score is ≥1.0 • 0.5 points from a baseline score of >5.0
Free of clinically confirmed
relapses
Brain volume loss rates might be helpful in expanding the concept of NEDA
Concept of no evidence of disease activity (NEDA)
Free of
disability
progression
Free of
disease
Free of
relapses
Free of clinical activity/progression
Brain atrophy
NEDA 4 No volume loss >0.4% per year
Free of gad
and new T2
lesions Free of MRI activity
Free of MRI progression
Increase in EDSS score at 6 months of: • 1.5 points if baseline score is 0 • 1.0 point if baseline score is ≥1.0 • 0.5 points from a baseline score of >5.0
Free of clinically confirmed
relapses*
*New neurological symptoms present for at least 24 hours, in the absence of fever or infection, manifesting ≥30 days from onset of a preceding demyelinating event and confirmed by an independent evaluating physician within 7 days of onset.
Pipelines • Non-standardised • Variability • Off-line • Resources
MRI-related factors • Variation in imaging protocols • Artefacts: signal heterogeneity, spatial distortions, motion, etc. • Lack of normative data acquired with the same MR protocol
MS-related factors • Pseudoatrophy effect (long-term follow-up required)
Non-MS related factors (accelerate brain volume decrease) • High body mass index • Genetic factors: glycated haemoglobin, APOE Σ4
• High alcohol consumption • Smoking • Dehydration • Cardiovascular risk factors
Brain atrophy in MS: translation to clinical practice
Challenges
6
PBVC, percentage brain volume change Modified from Claudia Wheeler-Kingshoot, London, personal communication
PB
VC
/ y
ear
–0.8 (IQR –1.3; –0.3)
–1.3 (IQR –1.7; –0.5)
No disability progression
Disability progression
–3
–2
–1
0
1
shortcut
Pre-processing of data
Segmentation and registration of data
Strategies for extracting atrophy
Intra- and inter-subject comparison
Errors Technically demanding
Time consuming Cost (reimbursement?)
Atrophy measurements: off-line
• Fully automated pipelines: o Cross-sectional/longitudinal volumetric analysis o Automated lesion filling o Quality control o Normative data
• Commercially available (NeuroQuant®; IcoMetrix®)
• Integrated in all major MR vendors post-processing software • Information transferred into PACS • Reimbursement (public health systems, private insurance)
shortcut
Atrophy measurements: actions
IcoMetrix /MSMetrix
NeuroQuant
ANTs /Atropos-DIReCT
Freesurfer
Software Library/ Tool
FSL /SIENAX – SIENA – FLIRT – FIRST
CIVET
TOADS, CRUISE, Lesion
SPM /LST, Segment
IcoMetrix, Leuven, Belgium
Advanced Radiology, Baltimore, Maryland, USA
Penn Image Computing and Science Laboratory, University of Pennsylvania, Philadelphia, USA
Athinoula A. Martinos Center for Biomedical Imaging, Harvard Medical School/Massachusetts General Hospital, Charlestown, USA;
FMRIB Centre, John Radcliffe Hospital, Oxford, UK
McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University, Montreal, Canada
Johns Hopkins University, Baltimore, USA
Wellcome Trust Centre for Neuroimaging,, London, UK
Commercial
Commercial
Open source
Open source
Open source
Free
Freeware
Open source
web-based interface or installed on Mac
OS
C++ / depends on the Insight Toolkit
C
C++
web-based interface - Perl scripts/ C/C++
JAVA, plug-ins for the MIPAV
software package
matlab
Institution License Language
Pipelines for atrophy measurements (in MS)
Mechanisms producing brain volume loss
• Tissue loss (i.e., loss of myelin, glial cells, neurons and axons due to the inflammatory demyelination and neurodegeneration),
Brain atrophy in MS
Atrophy
Brain atrophy in MS Brain atrophy in MS
Mechanisms producing brain volume loss
• Tissue loss (i.e., loss of myelin, glial cells, neurons and axons due to the inflammatory demyelination and neurodegeneration),
• Change in non-tissue components (i.e., fluids shifts)
The respective contribution of each component to brain volume loss may depend on many factors, such as disease stage, brain region affected, type of pharmacological
treatment, presence of comorbidities and other factors unrelated to the disease
Brain atrophy in MS
Pseudoatrophy Atrophy
Brain atrophy in MS
Ventricular volume change, obtained using FreeSurfer segmentation results, in percent in comparison to normal hydration (set to 100%)
Streitbürger DP, et al. Plos One 2012.
Ventricular volume change
Measurement timepoints
Pe
rce
nta
ge c
han
ge o
f ve
ntr
icu
lar
volu
me
re
lati
ve t
o n
orm
al h
ydra
tio
n
Hyperhydr. 1st dehydr. 2nd dehydr. 3rd dehydr. –5
–4
–3
–2
–1
0
1
2
3
4
5
Brain volume loss: hydration
7
Reported data from CTs
Effects of DMTs on Brain Atrophy (RRMS)
Drug Global effect on
Brain Volume 1
Immediate effect
on Brain Volume 2
Delayed effect
on Brain Volume 3
Able to cross Blood-
Brain Barrier
Placebo-controlled studies Interferon beta 1a No No Yes No Glatiramer Acetate No Noa NAb No Fingolimod Yes Yes Yes Yes Dimethyl-fumarate Yesc Nod Yesd No Teriflunomide No No No No Laquinimod Yes NA NA Yes Natalizumab No No Yes No
Active comparator4 Interferon vs glatiramer acetate
Yes (GA)e Yes (GA)e Yes (GA)e No
Fingolimod vs im IFN-β-1a
Yes (FTY) Yes (FTY) NAf Yes (FTY)
Alemtuzumab vs sc IFN-β-1a 44 g
Yes (AL) NAg NAg No
Vidal-Jordana et al. J Neurol 2015
Fingolimod treatment has a consequence on brain volume loss
* p < 0.05 for fingolimod vs. placebo; ** p < 0.01 for fingolimod vs. placebo; *** p ≤ 0.001 for fingolimod vs. placebo; Rank analysis of covariance (ANCOVA) adjusted for treatment group,
country and baseline normalized brain volume; n = number of patients with non-missing value Presented by Radue et al. ENS 2012. 219.
24.7% reduction
27.6% reduction
n=395 384 383
n=383 371 358
n=357 332 331
n=290 252 254
n=115 77 93
Point of switch from placebo–fingolimod
-0.44–0.45% / year treated -0.69% / year placebo -0.1-0.3% / year HC (expected) M
ean
pe
rce
nt
chan
ge in
b
rain
vo
lum
e f
rom
bas
elin
e
Brain atrophy data from FREEDOMS extension study
Brain atrophy data from CARE MS I study
*Alemtuzumab vs SC IFNB-1a, p<0.0001 BPF= brain parenchymal fraction; CI=confidence interval; SC IFNB-1a=subcutaneous interferon beta-1a
Arnold D, Cohen J, Barkhof F et al. Neurology 2014;83;e35.
• Alemtuzumab slowed the reduction in BPF over 3 years, despite 82% of patients only receiving the initial 2 courses of alemtuzumab at Month 0 and Month 12 in the core study
• The median yearly rate of brain volume loss with alemtuzumab decreased progressively over time
CARE MS I: Percentage Change in Brain Parenchymal Fraction
Alemtuzumab 12 mg SC IFNB-1a 44 µg
-1.8
-1.6
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2 0.0
Me
dia
n C
han
ge F
rom
B
ase
line,
% (
95%
CI)
*
*
42% slowing of brain volume loss vs SC IFNB-1a at Year 2
0 1 2 3 Year
171 – 185 176
351 323 371 367
No. of patients
Median yearly BPF change
‒0.59% ‒0.25% ‒0.19% ‒
‒0.94% ‒0.50% – –
Alemtuzumab treatment has a consequence on brain volume loss
Reported data from CTs
Drug Global effect on
Brain Volume 1
Immediate effect
on Brain Volume 2
Delayed effect
on Brain Volume 3
Able to cross Blood-
Brain Barrier
Placebo-controlled studies Interferon beta 1a No No Yes No Glatiramer Acetate No Noa NAb No Fingolimod Yes Yes Yes Yes Dimethyl-fumarate Yesc Nod Yesd No Teriflunomide No No No No Laquinimod Yes NA NA Yes Natalizumab No No Yes No
Active comparator4 Interferon vs glatiramer acetate
Yes (GA)e Yes (GA)e Yes (GA)e No
Fingolimod vs im IFN-β-1a
Yes (FTY) Yes (FTY) NAf Yes (FTY)
Alemtuzumab vs sc IFN-β-1a 44 g
Yes (AL) NAg NAg No
Vidal-Jordana et al. J Neurol 2015
Effects of DMTs on Brain Atrophy (RRMS)
• Particularly relevant in the first months of treatment • More evident with highly anti-inflammatory DMDs • Evident in patients with inflammatory activity (Gad) • Mainly affect white matter (grey matter is insensitive) • Not demonstraed in the spinal cord
Ne
t b
rain
par
en
chym
al
volu
me
loss
Treatment duration (months)
PREVENTION
PSEUDOATROPHY
IRREVERSIBLE ATROPHY
>–2.0%
–1.5%
–1.0%
–0.5%
0 12 36 6 24
Untreated
Low-dose IFN
Natalizumab
GA
High-dose IFN / Chemotherapeutics
Zivadinov R, et al. J Neurol 2008;Suppl 1:61-74
Vidal-Jordana A, et al. Mult Scler 2013; 19:1175-81
DMD-induced pseudoatrophy effect
Linear regression analyses were performed to predict PBVC, WMF and GMF evolution using the following independent variables: age, disease duration, presence of Gd-enhancing lesions at baseline MRI, corticosteroid treatment one month prior to baseline MRI and corticosteroid treatment between baseline and one-year MRI.
Pseudatrophy effect in the first year of treatment with Natalizumab
Vidal-Jordana A, et al. Mult Scler 2013; 19:1175-81
8
n=39
Presence of Gd+ at baseline
Partial correlation= -0.37
p=0.012
Linear regression analyses were performed to predict PBVC, WMF and GMF evolution using the following independent variables: age, disease duration, presence of Gd-enhancing lesions at baseline MRI, corticosteroid treatment one month prior to baseline MRI and corticosteroid treatment between baseline and one-year MRI.
Month 0 to 12
Vidal-Jordana A, et al. Mult Scler 2013; 19:1175-81
Pseudatrophy effect in the first year of treatment with Natalizumab
• Exclude brain atrophy changes during the first months after treatment onset
• Grey matter assessment (insensitive to fluctuations in water content) • Normalisation
o Degree of inflammatory activity (Gad lesions) o White matter T2 relaxometry
Pre-treatment (DX, FU) Baseline 6 months Follow-up every 12 months
-1 0 1 Treatment onset
Minimise pseudoatrophy effect
Alternatives to brain volume assessment (as a measure of atrophy)
oCorpus callosum index
o Spinal cord area
Figueira et al, Arq Neuropsiq 2007
a
b a'
b' c'
c
CCI: aa’ + bb’ + cc’ /ab
UCCA, Upper cervical cord cross-sectional area; Lukas, et al. Radiology 2013
UC
CA
mm
2
Corpus callosum atrophy
Corpus callosum index
a
b
a'
b'
c'
c
CCI: aa’ + bb’ + cc’ /ab
Modyfied from Figueira et al. Arq Neuropsiquiatr 2007
Corpus callosum area
CIS CDMS
Decrease in corpus callosum area at 6 months ≥ 1%, and baseline T2 lesion volume ≥ 5 cm3
CDMS, clinically definite MS; CIS, clinically isolated syndrome; GM, grey matter; MRI, magnetic resonance imaging; WM, white matter.
Kalincik T, et al. PLoS One 2012.
Time (months)
Pat
ien
ts w
ith
CD
MS
(%)
HR = 1.2 (0.7–2.2)
HR = 6.5 (3.4–12)
Cumulative risk of CDMS
100
90
70
60
80
10
0
20
40
30
50
0 6 12 18 24
Number of MRI predictors:
2 1 0
0 12 6 18 24
p = 10–4
–5
–4
–3
–2
–1
0
1
Corpus callosum area change
Time (months)
%
Brain volume change
0 12 6 18 24 Time (months)
%
–2
–1.5
-1
0
–0.5
0 12 6 18 24
WM volume change
Time (months)
%
–2
–1.5
–1
0
–0.5
0.5
0 12 6 18 24
0
–1
–2
–3
GM volume change
Time (months)
%
Dashed lines delineate 95% confidence intervals
Value of corpus callosum (CC) area change for predicting conversion to CDMS
• Assessment of relationship between MRI measures and clinical status in 181 highly active RRMS patients from Avonex-Steroid-Azathioprine study (mean age 31.2 yr, disease duration 5.5 yr, EDSS 1.9) over 9 yr
• Multivariate analysis: 2 MRI markers evaluated during the 1st yr on treatment were predictive of sustained disability progression over 9 yr:
– T1 lesion volume at mo 12 (HR 1.16; P=0.009)
– CC change mo 12-0 (HR 0.92; P=0.009)
• Individual predictive accuracy (ROC analysis):
CC volume and T1 lesion volume seem to predict about half of pts who will have sustained disability progression over 9 year of interferon β
Value of corpus callosum (CC) volume change for predicting disability progression after 9 yr
Horakova D. Neurology 2014;82(10 Suppl):P6.117
9
• Affects the majority of MS patients (++ SPMS, + PPMS), progressive over time
(Stevenson, 1998; Lin, 2003; Furby, 2008, 2010; Rocca, 2011, Lukas 2013, Lukas 2014) • Moderate association with clinical status (Losseff, 1996; Nijeholt, 1998; Edwards, 1999, Furby, 2008; Rocca, 2011, Lukas 2013) • Faster in MS patients with disability progression over time (Lukas 2014) • Relevant for disability in the long term (Kearney, 2014)
Spinal cord atrophy in MS
440 MS patients
UCCA, Upper cervical cord cross-sectional area; Lukas, et al. Radiology 2013
MR imaging–derived UCCA was found to be the most significant spinal cord parameter for explaining EDSS score
UC
CA
mm
2
P=0.30
MRI parameters predicting EDSS P value
UCCA <0.001
T1 lesion volume 0.001
Diffuse spinal cord abnormalities 0.002
Number of spinal cord segments 0.004
P<0.001 Multivariate linear regression
Spinal cord atrophy in MS: correlation with clinical disability
352 MS patients (two centers)
UCCA, Upper cervical cord cross-sectional area; Lukas, et al. JNNP 2014
Spinal cord atrophy in MS: changes over time
UC
CA
mm
2
•SC atrophy develops mostly independently from brain pathology, but not to number of SC segments involved by lesions and the presence of diffuse SC abnormalities •aUCCA was significantly higher in patients with disease progression (−2.3% per year) than in stable patients (−1.2% per year; p=0.003) •aPBVC did not differ between subtypes (RRMS: −0.42% per year; SPMS −0.6% per year; PPMS: −0.46% per year) nor between progressive and stable patients ( p=0.055)
aUCCA: -2.2%
aUCCA: -1.3%
aUCCA: -1.5%
113 MS patients; 20 controls
UCCA, Upper cervical cord cross-sectional area; Schlaeger, et al. Ann Neurol 2014
Grey matter spinal cord atrophy in MS: correlation with clinical disability
PSIR control
Mild RRMS
Progressive MS
Spearman Rank Correlations between PSIR and Clinical Measures
manually segmented
3D T1w brain Dedicated 3D T1w c-spine
Spinal cord atrophy in MS: brain/spinal 3D T1-PSIR
PSIR Advantages:
• Easy to measure
• “Summary measure” of the irreversible/destructive pathological process of MS
• Whole brain atrophy is highly reproducible and sensitive to disease-related changes
• Correlates with disability, cognitive impairment and fatigue
Disadvantages:
• End-stage phenomenon
• Pseudoatrophy effect (first 6–12 months)
• Fluctuations: steroids, hydration
• Co-morbidities: smoking, alcohol, body mass…
• Technically demanding / time consuming: reimbursement?
• Not enough evidence to use atrophy measures to assess and predict individual treatment response
Brain/spinal cord atrophy assessment in MS
10
Neuroimmunology Unit MRI Neuro Unit
Special thanks to: Radiology department • Cristina Auger • Deborah Pareto • Raquel Mitjana • Elena Huerga • Juan F. Corral • Xavier Aymerich • Manel Alberich Neurology department • Xavier Montalban • Jaume Sastre-Garriga • Mar Tintore • Jordi Rio • Carmen Tur • Anka Vidal
Advanced quantitative brain MR measures Massimo Filippi Neuroimaging Research Unit, Institute of Experimental Neurology, Division of Neuroscience, San Raffaele Scientific Institute, Vita-Salute San Raffaele University. E-mail address: [email protected]
Introduction
Conventional magnetic resonance imaging (MRI) sequences (i.e., dual-echo, FLAIR, pre- and post-
contrast T1-weighted scans) provide important pieces of information for diagnosing multiple
sclerosis (MS), understanding its natural history, and assessing treatment efficacy. Disappointingly,
the strength of the association between conventional MRI findings and the clinical manifestations of
the disease remains modest in patients with MS. This is likely due to the relative lack of specificity
of conventional MRI in the evaluation of the heterogeneous pathological substrates of the disease,
its inability to provide accurate estimates of damage outside focal lesions, and the fact that it cannot
provide information on central nervous system (CNS) functional reorganization after tissue injury
has occurred. Several advanced MRI techniques have been applied to estimate overall MS burden in
the different phases of the disease. Their use has allowed to grade in vivo the heterogeneity of MS
pathology in focal lesions, normal-appearing white matter (NAWM) and gray matter (GM).
Heterogeneity of WM lesions
Variable degrees of magnetization transfer ratio (MTR) reduction have been reported in acute and
chronic MS lesions, with the most prominent changes found in T1-hypointense lesions. Recently,
MRI contrast agents composed of iron particles, known as ultrasmall particles of iron oxide
(USPIO) or super-paramagnetic iron particles of oxide (SPIO), have been introduced to monitor
different aspects of the MS inflammatory process. These particles are taken by cells of the
monocyte/macrophage system. As a consequence, USPIO-enhancement reflects cellular infiltration
and may complement Gd-enhancement.[1]
High-field (3.0-4.0 Tesla) and ultra-high field (7.0 Tesla and higher) MRI scanners are becoming
progressively available and contribute to detect a greater number and a larger volume of T2 and
enhancing brain lesions. These MR scanners also provide a better definition of lesions in terms of
location in the WM and GM, morphology, architecture and association with vasculature, at a
resolution which resembles that of pathological assessment.[2-4] Due to the better definition of the
relationship between demyelinating lesions and the deep venous system, several studies have shown
that MS plaques form around the microvasculature.[4, 5] This feature has been reinforced by the
investigation of blood-brain barrier abnormalities in MS at 7.0 Tesla, which showed that the
majority of enhancing lesions are perivenular, and that the smallest lesions have a concentric pattern
of enhancement, suggesting that they grow outward from a central vein.[5] The presence of a
central small vein and a rim of hypointensity on 7.0 Tesla T2*-weighted magnitude images can also
assist in the differentiation of WM lesions found in MS patients from those of patients with other
neurological conditions that can mimic MS.
Cortical lesions
Although imaging the cortex is technically difficult, due to its morphology and location, and the
nature of the pathology within this structure, the development of specialized MRI sequences that
can suppress the signal from WM and cerebrospinal fluid simultaneously, named “double inversion
recovery” (DIR), has enabled substantial improvements in cortical lesions (CLs) detection.[6] Using
DIR, MRI studies revealed that CLs develop early in MS and increase in number and size with
progression of the disease.[7] On the other hand, they are sparse in benign forms of the disease[8]
and in children with MS.[9] CLs contribute to explain both locomotor disability and cognitive
impairment,[10] and their presence is associated with other MRI indexes of damage such as T2
lesion load and WM and GM atrophy.[11] In patients presenting with a clinically isolated syndrome
(CIS) suggestive of MS the accuracy of MRI diagnostic criteria for MS increases when considering
CLs on baseline scans.[12] Importantly, CLs have not been identified in migraine patients with
multiple brain WM hyperintensities and patients with neuromyelitis optica, thus suggesting a high
specificity for MS. Measuring CLs might also represent a valuable marker to monitor treatment
effect.[13]
Diffuse NAWM damage
Outside T2 lesions, quantitative MRI techniques allow estimating the presence and extent of
abnormalities in the NAWM of MS patients. Several studies with serial MR investigations have
shown that, at least in some cases, subtle WM changes can be seen in areas which days to weeks
later develop into classical enhancing lesions. These changes consist of a reduction of MTR,
increase in mean diffusivity (MD), and mild to moderate reduction of N-acetylaspartate. NAWM
MTR histogram-derived measures evolve at different rates in the major MS clinical phenotypes.
NAWM MTR reduction has also been shown to predict the accumulation of clinical disability over
the subsequent five years in patients with definite MS. A voxelwise assessment of the regional
distribution of abnormalities of diffusion tensor (DT) MR indexes has been shown to be a
rewarding strategy for understanding the heterogeneity of MS clinical phenotypes. In a recent
study,[14] this method has been applied to 172 MS patients with different clinical phenotypes and it
has been found that compared with healthy controls, CIS patients had significantly increased MD,
axial diffusivity, and radial diffusivity in the majority of WM tracts of the brain. Compared to
controls, patients with primary progressive (PP) MS also showed a diffuse fractional anisotropy
(FA) decrease involving the majority of WM tracts. No relevant difference in diffusivity measures
was found between CIS and relapsing-remitting (RR) MS patients. Compared with benign MS
patients, those with RRMS had reduced FA values in all WM tracts and a decreased axial
diffusivity in the majority of the tracts. Secondary progressive (SP) MS had a pronounced damage
to the majority of tracts and, compared with benign MS, a pronounced FA alteration of the tracts
relevant for motor impairment. Significant DT MRI abnormalities of brain WM tracts, which were
only partially correlated with focal WM lesions, were also found in very young pediatric MS
patients.[15]
Several approaches have been developed to investigate damage to selected WM tracts with the
ultimate goal of improving the correlation with clinical measures. In patients with CIS, diffusivity
measures of the corticospinal tract correlate with clinical measures of motor impairment.[16] In
patients with optic neuritis, reduced structural connectivity values in the optic radiations compared
with controls have been shown.[17] Diffusivity abnormalities in optic radiations have been related
to trans-synaptic degeneration secondary to optic nerve damage and Wallerian degeneration due to
local lesions in a recent study in which patients were classified according to the presence of
previous optic neuritis and lesions along these tracts.[18] In cognitively impaired MS patients
damage of specific corticothalamic tracts explained global cognitive dysfunction and impairment of
selected cognitive domains better than MRI measures of thalamic and cortical damage.[19]
Advances in DT MRI and tractography have spurred the development of brain connectivity
techniques, which allow defining and quantifying anatomical links between remote brain
regions.[20] The use of these approaches has revealed reduced network efficiency in the WM
structural networks of MS patients,[21] including those at the earliest stages of the disease.[22]
Diffuse GM damage
The application of modern MR techniques can contribute to the assessment of different aspects of
GM damage, including the presence of diffuse disease-related abnormalities (measured using
quantitative techniques such as MT and DT MRI), metabolic abnormalities (measured by means of
proton MR spectroscopy), irreversible tissue loss (atrophy) and iron deposition (quantified using
T2/T2*-weighted imaging), thought to reflect neurodegeneration. Several studies have demonstrated
reduced MTR and increased MD in the GM of patients with different MS phenotypes including
those at the earliest clinical stages of the disease.[23, 24] These abnormalities are more severe in
patients with the progressive disease phenotypes. Analogous findings have been shown when
measuring cortical atrophy.[25] Diffuse cortical damage is not stable, but tends to worsen over time,
independent of the progression of damage within the WM.[26] The clinical relevance of measuring
such a damage has been demonstrated by several studies which have shown correlations with
clinical disability and cognitive impairment.[25] A longitudinal study has found an increased rate of
cortical tissue loss in patients with progressing disability compared to those with stable disease,[27]
whereas another study demonstrated that progressive neocortical loss is relevant to MS-associated
cognitive impairment.[28, 29] A recent 13-year longitudinal study demonstrated that GM damage is
one of the key factors associated with both long-term accumulation of disability and cognitive
impairment in MS patients.[29]
Analysis of the spatial distribution of GM damage has demonstrated that different regions might
have different vulnerabilities to MS-related pathological processes. Overall, MRI studies have
agreed in identifying the frontal, temporal and parietal lobes as the most affected cortical regions in
MS patients. However, the patterns of GM loss differs between the major MS clinical
phenotypes.[30] The quantification of diffuse GM damage provides robust prognostic measures of
disease progression. In RRMS patients, GM MTR was found to be an independent predictor of the
accumulation of disability over the subsequent eight years[31] and baseline GM fraction was found
to be an independent predictors of the development of SPMS over a 13-year period,[29] while in
PPMS, GM MD predicted the accumulation of disability over a five-year period.[26]
Assessment of atrophy of strategic GM structures could contribute to explain deficits in selective
cognitive domains, as well as the occurrence of specific symptoms and disability progression.
Hippocampal atrophy has been associated with deficits in memory encoding and retrieval.[32] In
this perspective, a recent study showed that hippocampal subregions have a different vulnerability
to MS-related damage, with a relative sparing of the head of the left hippocampus.[33] Atrophy of
the frontal and parietal lobes has been correlated with the presence and severity of fatigue.[34]
Thalamic atrophy has been correlated with accumulation of disability after a eight year follow up
period in patients with relapse-onset MS[35] and after a five year follow up period in PPMS
patients.[36]
Brain functional reorganization
Studies with functional MRI (fMRI) of the visual, cognitive and motor systems have consistently
demonstrated functional cortical changes in all MS phenotypes, with altered activation of regions
normally devoted to the performance of a given task and/or the recruitment of additional areas in
comparison to healthy subjects.[37] fMRI abnormalities in MS patients occur relatively early in the
course of the disease, even in patients with CIS and pediatric MS,[38] and tend to vary over the
course of the disease, not only after an acute relapse, but also in clinically stable patients.[37]
Functional and structural MRI abnormalities in MS patients are strictly correlated,[37] suggesting
that increased recruitment of “critical” cortical networks helps to limit the functional impact of MS-
related damage. However, increased cortical recruitment cannot continue indefinitely, and a lack of,
or exhaustion of, the “classical” adaptive mechanisms has been considered as a possible factor
responsible for unfavorable clinical evolution or accelerated cognitive decline.[37, 39]
Several studies have attempted to develop sophisticated statistical approaches to establish the
strength of activations and the synchrony between specific cortical areas, through the analysis of
functional and effective connectivities.[40] The combination of measures of functional connectivity
with measures of structural damage to specific WM fiber tracts is also likely to improve our
understanding of the relationship between structural and functional abnormalities.
Recently, the analysis of brain activity at rest has shown an increased synchronization of the
majority of the resting-state networks (RSN) in patients with CIS,[41] a reduced functional
connectivity (FC) of anterior regions of the default-mode network in patients with progressive
MS[42] and cognitive impairment,[40] and a complex reorganization of the visual network in
normal-sighted patients who recovered from a previous optic neuritis.[43] Distributed abnormalities
of RS FC within and between large-scale neuronal networks have been shown in RRMS and
pediatric MS patients and have been related to the extent of T2 lesions.[44, 45] Active and RS fMRI
have also been applied to assess modifications of the patterns of activations and FC following
cognitive rehabilitation in a few single-centre studies.[46, 47] A recent study demonstrated that
changes in RS FC of cognitive-related networks contributes to the persistence of the effects of
cognitive rehabilitation after six months in RRMS patients.[48]
Graph theory is a mathematical framework which allows to describe a network as a graph
consisting of a collection of nodes (i.e., brain regions) and edges (i.e., structural and functional
connections). The application of this approach to RS fMRI data has allowed to investigate the
topological organization of functional brain network connectivity. In a recent study an impairment
of global integration, that likely reflects a reduced competence in information exchange between
distant brain areas, has been demonstrated in MS patients. Such a modification of regional network
properties contributed to cognitive impairment and phenotypic variability of MS.[49]
Conclusions
Conventional and advanced MRI techniques have been applied extensively to the study of MS and
such an effort has contributed to improve our ability to diagnose and monitor the disease, as well as
our understanding of its pathophysiology and treatment efficacy. Nevertheless, many challenges
remain. Quantitative, metabolic and functional imaging techniques need to be optimized and
standardized across multiple centres to monitor adequately disease evolution. In addition, from the
large body of available literature, it results clear that none of these quantitative techniques, taken in
isolation is able to provide a complete picture of the complexity of the MS process. A
multiparametric approach, combining aggregates of different MR quantities might improve our
ability to monitor the disease. With the increased availability of high field and ultra-high field MR
scanners, such an issue is now becoming extremely critical. Furthermore, the practical utility of
modern MR approaches from a research setting to daily-life clinical practice still needs to be
investigated.
References
1. Vellinga, M.M., et al., Pluriformity of inflammation in multiple sclerosis shown by ultra-small iron oxide particle enhancement. Brain, 2008. 131(Pt 3): p. 800-7.
2. Hammond, K.E., et al., Quantitative in vivo magnetic resonance imaging of multiple sclerosis at 7 Tesla with sensitivity to iron. Ann Neurol, 2008. 64(6): p. 707-13.
3. Ge, Y., V.M. Zohrabian, and R.I. Grossman, Seven-Tesla magnetic resonance imaging: new vision of microvascular abnormalities in multiple sclerosis. Arch Neurol, 2008. 65(6): p. 812-6.
4. Absinta, M., et al., Seven-tesla phase imaging of acute multiple sclerosis lesions: a new window into the inflammatory process. Ann Neurol, 2013. 74(5): p. 669-78.
5. Gaitan, M.I., et al., Initial investigation of the blood-brain barrier in MS lesions at 7 tesla. Mult Scler, 2013. 19(8): p. 1068-73.
6. Geurts, J.J., et al., Intracortical lesions in multiple sclerosis: improved detection with 3D double inversion-recovery MR imaging. Radiology, 2005. 236(1): p. 254-60.
7. Calabrese, M., et al., A 3-year magnetic resonance imaging study of cortical lesions in relapse-onset multiple sclerosis. Ann Neurol, 2010. 67(3): p. 376-83.
8. Calabrese, M., et al., Evidence for relative cortical sparing in benign multiple sclerosis: a longitudinal magnetic resonance imaging study. Mult Scler, 2009. 15(1): p. 36-41.
9. Absinta, M., et al., Cortical lesions in children with multiple sclerosis. Neurology, 2011. 76(10): p. 910-3.
10. Calabrese, M., et al., Cortical lesions and atrophy associated with cognitive impairment in relapsing-remitting multiple sclerosis. Arch Neurol, 2009. 66(9): p. 1144-50.
11. Calabrese, M., et al., Cortical lesions in primary progressive multiple sclerosis: a 2-year longitudinal MR study. Neurology, 2009. 72(15): p. 1330-6.
12. Filippi, M., et al., Intracortical lesions: relevance for new MRI diagnostic criteria for multiple sclerosis. Neurology, 2010. 75(22): p. 1988-94.
13. Rinaldi, F., et al., Natalizumab strongly suppresses cortical pathology in relapsing-remitting multiple sclerosis. Mult Scler, 2012. 18(12): p. 1760-7.
14. Preziosa, P., et al., Intrinsic damage to the major white matter tracts in patients with different clinical phenotypes of multiple sclerosis: a voxelwise diffusion-tensor MR study. Radiology, 2011. 260(2): p. 541-50.
15. Rocca, M.A., et al., Diffusion tensor magnetic resonance imaging in very early onset pediatric multiple sclerosis. Mult Scler, 2015.
16. Pagani, E., et al., A method for obtaining tract-specific diffusion tensor MRI measurements in the presence of disease: application to patients with clinically isolated syndromes suggestive of multiple sclerosis. Neuroimage, 2005. 26(1): p. 258-65.
17. Ciccarelli, O., et al., Optic radiation changes after optic neuritis detected by tractography-based group mapping. Hum Brain Mapp, 2005. 25(3): p. 308-16.
18. Rocca, M.A., et al., Wallerian and trans-synaptic degeneration contribute to optic radiation damage in multiple sclerosis: a diffusion tensor MRI study. Mult Scler, 2013. 19(12): p. 1610-7.
19. Bisecco, A., et al., Connectivity-based parcellation of the thalamus in multiple sclerosis and its implications for cognitive impairment: A multicenter study. Hum Brain Mapp, 2015. 36(7): p. 2809-25.
20. Guye, M., et al., Graph theoretical analysis of structural and functional connectivity MRI in normal and pathological brain networks. MAGMA, 2010. 23(5-6): p. 409-21.
21. Bozzali, M., et al., Anatomical brain connectivity can assess cognitive dysfunction in multiple sclerosis. Mult Scler, 2013. 19(9): p. 1161-8.
22. Li, Y., et al., Diffusion tensor imaging based network analysis detects alterations of neuroconnectivity in patients with clinically early relapsing-remitting multiple sclerosis. Hum Brain Mapp, 2013. 34(12): p. 3376-91.
23. Fernando, K.T., et al., Magnetization transfer histograms in clinically isolated syndromes suggestive of multiple sclerosis. Brain, 2005. 128(Pt 12): p. 2911-25.
24. Ramio-Torrenta, L., et al., Abnormalities in normal appearing tissues in early primary progressive multiple sclerosis and their relation to disability: a tissue specific magnetisation transfer study. J Neurol Neurosurg Psychiatry, 2006. 77(1): p. 40-5.
25. Filippi, M., et al., Imaging cortical damage and dysfunction in multiple sclerosis. JAMA Neurol, 2013. 70(5): p. 556-64.
26. Rovaris, M., et al., Grey matter damage predicts the evolution of primary progressive multiple sclerosis at 5 years. Brain, 2006. 129(Pt 10): p. 2628-34.
27. Chen, J.T., et al., Relating neocortical pathology to disability progression in multiple sclerosis using MRI. Neuroimage, 2004. 23(3): p. 1168-75.
28. Amato, M.P., et al., Association of neocortical volume changes with cognitive deterioration in relapsing-remitting multiple sclerosis. Arch Neurol, 2007. 64(8): p. 1157-61.
29. Filippi, M., et al., Gray matter damage predicts the accumulation of disability 13 years later in MS. Neurology, 2013. 81(20): p. 1759-67.
30. Ceccarelli, A., et al., A voxel-based morphometry study of grey matter loss in MS patients with different clinical phenotypes. Neuroimage, 2008. 42(1): p. 315-22.
31. Filippi, M. and M.A. Rocca, Multiple sclerosis: Linking disability and spinal cord imaging outcomes in MS. Nat Rev Neurol, 2013. 9(4): p. 189-90.
32. Sicotte, N.L., et al., Regional hippocampal atrophy in multiple sclerosis. Brain, 2008. 131(Pt 4): p. 1134-41.
33. Longoni, G., et al., Deficits in memory and visuospatial learning correlate with regional hippocampal atrophy in MS. Brain Struct Funct.
34. Sepulcre, J., et al., Fatigue in multiple sclerosis is associated with the disruption of frontal and parietal pathways. Mult Scler, 2009. 15(3): p. 337-44.
35. Rocca, M.A., et al., Thalamic damage and long-term progression of disability in multiple sclerosis. Radiology, 2010. 257(2): p. 463-9.
36. Mesaros, S., et al., Thalamic damage predicts the evolution of primary-progressive multiple sclerosis at 5 years. AJNR Am J Neuroradiol, 2011. 32(6): p. 1016-20.
37. Filippi, M. and M.A. Rocca, Functional MR imaging in multiple sclerosis. Neuroimaging Clin N Am, 2009. 19(1): p. 59-70.
38. Rocca, M.A., et al., Altered functional and structural connectivities in patients with MS: a 3-T study. Neurology, 2007. 69(23): p. 2136-45.
39. Cirillo, S., et al., Abnormal cerebellar functional MRI connectivity in patients with paediatric multiple sclerosis. Mult Scler, 2015.
40. Rocca, M.A., et al., Default-mode network dysfunction and cognitive impairment in progressive MS. Neurology, 2010. 74(16): p. 1252-9.
41. Roosendaal, S.D., et al., Resting state networks change in clinically isolated syndrome. Brain, 2010. 133(Pt 6): p. 1612-21.
42. Bonavita, S., et al., Distributed changes in default-mode resting-state connectivity in multiple sclerosis. Mult Scler, 2011. 17(4): p. 411-22.
43. Gallo, A., et al., Visual resting-state network in relapsing-remitting MS with and without previous optic neuritis. Neurology, 2012. 79(14): p. 1458-65.
44. Rocca, M.A., et al., Large-scale neuronal network dysfunction in relapsing-remitting multiple sclerosis. Neurology, 2012. 79(14): p. 1449-57.
45. Rocca, M.A., et al., Intranetwork and internetwork functional connectivity abnormalities in pediatric multiple sclerosis. Hum Brain Mapp, 2014. 35(8): p. 4180-92.
46. Filippi, M., et al., Multiple sclerosis: effects of cognitive rehabilitation on structural and functional MR imaging measures--an explorative study. Radiology, 2012. 262(3): p. 932-40.
47. Sastre-Garriga, J., et al., A functional magnetic resonance proof of concept pilot trial of cognitive rehabilitation in multiple sclerosis. Mult Scler, 2011. 17(4): p. 457-67.
48. Parisi, L., et al., Changes of brain resting state functional connectivity predict the persistence of cognitive rehabilitation effects in patients with multiple sclerosis. Mult Scler, 2014. 20(6): p. 686-94.
49. Rocca, M.A., et al., Impaired functional integration in multiple sclerosis: a graph theory study. Brain Struct Funct, 2014.