JNS-13415; No of Pages 5 Journal of the Neurological Sciences xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Journal of the Neurological Sciences journal homepage: www.elsevier.com/locate/jns
MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis Shahamat Tauhid, Mohit Neema, Brian C. Healy, Howard L. Weiner, Rohit Bakshi ⁎ Department of Neurology, Brigham and Women's Hospital, Laboratory for Neuroimaging Research, Partners MS Center, Harvard Medical School, Boston, MA, USA
a r t i c l e
i n f o
Article history: Received 13 June 2014 Received in revised form 29 July 2014 Accepted 29 August 2014 Available online xxxx Keywords: MRI Multiple sclerosis Lesions Atrophy Phenotypes Disability
a b s t r a c t Background: While disease categories (i.e. clinical phenotypes) of multiple sclerosis (MS) are established, there remains MRI heterogeneity among patients within those definitions. MRI-defined lesions and atrophy show only moderate inter-correlations, suggesting that they represent partly different processes in MS. We assessed the ability of MRI-based categorization of cerebral lesions and atrophy in individual patients to identify distinct phenotypes. Methods: We studied 175 patients with MS [age (mean ± SD) 42.7 ± 9.1 years, 124 (71%) women, Expanded Disability Status (EDSS) score 2.5 ± 2.3, n = 18 (10%) clinically isolated demyelinating syndrome (CIS), n = 115 (66%) relapsing-remitting (RR), and n = 42 (24%) secondary progressive (SP)]. Brain MRI measures included T2 hyperintense lesion volume (T2LV) and brain parenchymal fraction (to assess whole brain atrophy). Medians were used to create bins for each parameter, with patients assigned a low or high severity score. Results: Four MRI phenotype categories emerged: Type I = low T2LV/mild atrophy [n = 67 (38%); CIS = 14, RR = 47, SP = 6]; Type II = high T2LV/mild atrophy [n = 21 (12%); RR = 19, SP = 2]; Type III = low T2LV/ high atrophy [n = 21 (12%); CIS = 4, RR = 16, SP = 1]; and Type IV = high T2LV/high atrophy [n = 66 (38%); RR = 33, SP = 33]. Type IV was the most disabled and was the only group showing a correlation between T2LV vs. BPF and MRI vs. EDSS score (all p b 0.05). Conclusions: We described MRI-categorization based on the relationship between lesions and atrophy in individual patients to identify four phenotypes in MS. Most patients have congruent extremes related to the degree of lesions and atrophy. However, many have a dissociation. Longitudinal studies will help define the stability of these patterns and their role in risk stratification. © 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-SA license (http://creativecommons.org/licenses/by-nc-sa/3.0/).
1. Introduction While clinically-defined disease categories of multiple sclerosis (MS) are well established [1], there remains heterogeneity of the disease among patients that appears to exist within the confines of these subtypes. This heterogeneity includes the concepts of patientdependent longitudinally-consistent immunopathological lesion subtypes [2] and clustering of patient subgroups based on MRI-defined lesion characteristics and the level of tissue destruction [3,4]. Consistent with the concept of disease heterogeneity, clinical progression and treatment response are difficult to predict in individual patients. Furthermore, there is a topographic uncoupling of structural changes such as the lack of a relationship between brain and spinal cord involvement [5]. Taken
⁎ Corresponding author at: Laboratory for Neuroimaging Research, One Brookline Place, Brookline, MA 02445, USA. Tel.: +1 617 525 6550. E-mail address: [email protected] (R. Bakshi).
together, these data suggest that the pathogenesis of tissue injury may differ among patients with MS. Among the types of CNS tissue injury that typify the disease, the two broadest categories include inflammatory demyelinating foci (lesions) and tissue destruction (neurodegeneration; axonal and neuronal loss). The most commonly-used and readily-available MRI measure of lesion load is hyperintensity on T2-weighted images. To assess overall tissue destruction, MRI-quantified normalized whole brain volume is an established measure of brain atrophy [6]. Several lines of evidence indicate that lesions and atrophy represent somewhat unrelated aspects of the disease process. This includes the observation that in both crosssectional [6,7] and longitudinal [6,8] studies, lesions show a weak relationship to current and subsequent atrophy. Furthermore, the two measures provide complementary information in predicting treatment efficacy [9]. The objective of this study was to assess the ability of MRI-based categorization of the relationship between the severity of cerebral lesions and atrophy in individual patients to identify distinct MS phenotypes. This work was presented in preliminary form at the 2013 annual meeting of the American Academy of Neurology, San Diego.
http://dx.doi.org/10.1016/j.jns.2014.08.047 0022-510X/© 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-SA license (http://creativecommons.org/licenses/by-nc-sa/3.0/).
Please cite this article as: Tauhid S, et al, MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis, J Neurol Sci (2014), http://dx.doi.org/10.1016/j.jns.2014.08.047
2
S. Tauhid et al. / Journal of the Neurological Sciences xxx (2014) xxx–xxx
2. Methods 2.1. Subjects and clinical evaluation Demographic and clinical characteristics of the subjects are summarized in Table 1. This was a single-center cross-sectional retrospective study of 175 consecutive patients who met the following inclusion criteria: 1) Diagnosed with a relapsing form of MS including a clinically-isolated demyelinating syndrome (CIS), relapsing-remitting (RR), or secondary progressive (SP) [1]; 2) Evaluated clinically by an MS specialist provider at the Brigham and Women's Hospital MS clinic (Partners MS Center) within three months of MRI; 3) age 18–60 years. Nearly all patients were receiving disease modifying therapy, as is the common practice at our center [7]. Baseline characteristics, including demographic, clinical, and MRI (Table 1), were comparable to those of large MS cohorts [10,11], indicating that our cohort represented a wide range of the MS spectrum that includes the continuum CIS–RR– SP. We did not include primary progressive MS patients because of the challenging diagnosis [1], the small numbers of those patients available, and that concept that this group may represent a separate entity. This study was approved by our institutional review board.
Fig. 1. Method of creation of MRI-based phenotypes. Patients were stratified by median splits of the entire cohort based on total cerebral MRI-measures of inflammation (T2LV) and atrophy (BPF); the median was used to create bins for each MRI parameter, with patients assigned a score of low or high severity; T2LV = total brain T2 hyperintense lesion volume (cc); BPF = brain parenchymal fraction; A = atrophy; L = T2LV; Type I (mild atrophy/low T2LV); Type II (mild atrophy/high T2LV); Type III (high atrophy/low T2LV); Type IV (high atrophy/high T2LV).
2.2. MRI acquisition and segmentation Brain MRI was obtained in all subjects at 1.5 T with a 3 mm thick, gapless, T2-weighted, axial dual-echo protocol that included the entire brain. With few exceptions, the protocol was performed on the same scanner and included the following consistent acquisition parameters: repetition time 3000 ms, echo time 30/80 ms, and 0.93 × 0.93 mm pixel size. Using automated template-driven segmentation (TDS +) from the dual echo images, total brain T2 hyperintense white matter lesion volume (T2LV) and normalized whole brain volume [brain parenchymal fraction (BPF) — a surrogate of whole brain atrophy] [6] were automatically determined [12], after which expert manual correction was applied to each output map to ensure accuracy. 2.3. Creation of the MRI-defined phenotypes Patients were stratified into MRI phenotype groups based on a median split (0 or 1) of T2LV and BPF. As shown in Fig. 1, all patients were assigned to one of four unique groups as: Type I = mild atrophy (0) and low T2LV (0) Type II = mild atrophy (0) and high T2LV (1) Type III = high atrophy (1) and low T2LV (0) Type IV = high atrophy (1) and high T2LV (1). 2.4. Statistical analysis The demographic, clinical and MRI characteristics were compared across the four groups using one-way analysis of variance for continuous outcomes (age, disease duration, BPF, and T2LV), a proportional odds model for ordinal outcomes (EDSS) and a chi-square test for nominal Table 1 Demographic and clinical characteristics of the subjects. Number of patients Clinical MS phenotype: • Clinically isolated syndrome • Relapsing-remitting MS • Secondary progressive MS Number of women (n, %) Age (mean ± SD) (range) (years) Expanded Disability Status Scale score Disease duration (years since first symptoms) Total brain T2 hyperintense lesion volume (cc) Brain parenchymal fraction
n = 175 n = 18 (10%) n = 115 (66%) n = 42 (24%) n = 124 (71%) 42.7 ± 9.1 (21–60) 2.5 ± 2.3 (0–8.6) 9.6 ± 8.6 (0.2–48.0) 8.7 ± 10.2 (0.5–57.1) 0.844 ± 0.049 (0.67–0.96)
variables (gender and disease category). If a significant difference was observed in the four group comparison, pairwise group comparisons were completed and Holm's correction for multiple comparisons was applied to adjust the p-values. In addition, we assessed the relationship among the MRI measures and the relationship between MRI and EDSS score using the Spearman's correlation coefficient. These were analyzed for the whole cohort and within each MRI phenotype group. 3. Results Four categories emerged from the median split analysis (Table 2, Fig. 2). Regarding clinical and demographic group comparisons (Table 2), Type IV patients had higher disease duration, higher EDSS scores, and a higher proportion of SP patients than the other three groups (all p b 0.05). Type III and IV patients were each older than the Type I and II patients. Otherwise, there were no differences in age, disease duration, EDSS score, or clinical disease category distribution among groups (p b 0.05). There were no differences in the gender breakdown across the groups (p N 0.05). Regarding the MRI parameters that comprised the phenotypes (Table 2), all pairwise comparisons showed significant differences in T2LV among the groups (p b 0.05), except for Type I vs. Type III (p N 0.05). In addition, all pairwise comparisons showed significant differences in BPF among the groups (p b 0.05) except Type I vs. Type II (p N 0.05). These results confirm the rationale of defining unique MRI phenotypes on the basis of different T2LV and BPF characteristics. We also tested the relationships between lesions and atrophy and between MRI and disability (Table 3). The correlation between T2LV and BPF was significant in the overall cohort (p b 0.0001) and in the Type IV patients (p b 0.0001) but not in the other three groups (p N 0.05). Similarly, the correlation between each of the two MRI parameters (T2LV and BPF) and EDSS score was significant in the overall cohort (p b 0.0001) and in the Type IV patients (p b 0.0001) but not in the other three groups (p N 0.05). 4. Discussion A major finding in this study is that 25% patients had a discordance of where their brain lesions and atrophy fell with regard to the medians of the full cohort. The discordance was equally bidirectional in that about half of the discordant patients had a high degree of atrophy
Please cite this article as: Tauhid S, et al, MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis, J Neurol Sci (2014), http://dx.doi.org/10.1016/j.jns.2014.08.047
S. Tauhid et al. / Journal of the Neurological Sciences xxx (2014) xxx–xxx
3
Table 2 Demographic, clinical and MRI characteristics of the MRI-defined phenotypes.
Number of patients (%) Disease category: • Clinically isolated syndrome • Relapsing-remitting MS • Secondary progressive MS Number of women (n, %) Age (mean ± SD) (range) (years) Expanded Disability Status Scale score Disease duration (years) T2LV (cc) Brain parenchymal fraction
Type I (mild atrophy/ low T2LV)
Type II (mild atrophy/ high T2LV)
Type III (high atrophy/ low T2LV)
Type IV (high atrophy/ high T2LV)
n = 67 (38%)
n = 21 (12%)
n = 21 (12%)
n = 66 (38%)
14 (21%) 47 (70%) 6 (9%) n = 51 (76%) 37.3 ± 8.3 (21.0–54.1) 1.5 ± 1.8 (0–8.5) 5.5 ± 5.7 (0.26–26.0) 2.3 ± 1.0 (0.46–4.21) 0.888 ± 0.025 (0.85–0.93)
0 19 (91%) 2 (9%) n = 16 (76%) 36.9 ± 7.9 (21.7–50.2) 1.9 ± 2.0 (07.5) 8.6 ± 6.6 (0.57–22.0) 8.2 ± 4.9 (4.5–22.2) 0.875 ± 0.025 (0.85–0.93)
4 (19.0%) 16 (76%) 1 (5%) n = 14 (67%) 43.7 ± 7.8 (21.0–54.3) 1.7 ± 1.8 (0–6.5) 8.6 ± 8.1 (0.4–26.0) 2.3 ± 1.1 (0.8–4.1) 0.834 ± 0.016 (0.78–0.85)
0 33 (50%) 33 (50%) n = 43 (65%) 44.0 ± 8.6 (26.5–60.2) 3.9 ± 2.2 (1.0–8.5) 14.3 ± 9.4 (0.25–48.0) 17.3 ± 11.8 (4.4–57.1) 0.791 ± 0.041 (0.67–0.85)
Key: T2LV = total brain T2 hyperintense lesion volume. Regarding the breakdown of women vs. men, no significant differences were observed across the groups (p N 0.05). Regarding age, Types III and IV were significantly older than Types I and II (p b 0.05). Regarding disease category (i.e. clinical phenotype), EDSS, and disease duration, Type IV was significantly more likely to be SPMS, disabled, with higher disease duration than the other three categories (p b 0.05). Regarding T2LV, all pairwise differences except Type I vs III were statistically significant (p b 0.05). Regarding brain parenchymal fraction, all pairwise differences except Type I vs. Type II were statistically significant (p b 0.05).
despite a low lesion load (Type II MRI phenotype) and vice-versa (Type III). This was further underscored by the poor correlation in both groups between T2LV and BPF. Taken together these findings suggest that subgroups of patients have an uncoupling between focal immunemediated demyelination in white matter (inflammation) and overall global cerebral tissue destruction (neurodegeneration). Regarding the Type II patients, who have a high T2LV but low atrophy, this might relate in part to the long held view that T2 hyperintense lesions are non-specific for pathologic changes, reflecting a wide range of severity [13]. We hypothesize that Type II patients have a T2LV that is dominated by mild changes such as mild inflammation, edema, and partial demyelination. Another factor serving to limit overall atrophy in these patients may be the presence of enhanced repair capacity [14] or other resistance to the downstream effects of white matter injury on neuronal and axonal integrity. Regarding the Type III patients, who have a low T2LV but higher degree of atrophy, there may be other contributions to tissue destruction that are not captured by overt lesion load. Considering the prevailing notion that whole brain atrophy in MS is dominated by gray matter rather than white matter atrophy, consistently seen in MRI studies at 1.5 T and 3 T [6,15–17], the Type III patients may be affected by gray
matter injury due to a variety of mechanisms that are separate from white matter lesions (and were not assessed in the present study) such as microglial activation [18], cortical lesions [19,20], meningeal inflammation [21], and iron deposition/oxidative stress [22]. Type 1 and Type IV phenotypes were characterized by a congruent level of lesions and atrophy with each either of a lower (Type I) or higher (Type IV) degree. Comparing these two groups, Type I patients tended to be at an earlier stage of disease on other disease characteristics, showing a lower age, lower disease duration, and lower level of disability. In addition, the Type I group was represented by a preponderance of CIS or RR rather than SP patients. Type IV patients represent more advanced disease, owing to the higher disease duration and higher disability level than the other three groups. Another noteworthy finding in this study is that the MRI phenotypes belie the confines of the current MS disease categories (i.e. clinical phenotypes), readily crossing those boundaries, suggesting that the MRI categorization provides unique information about disease severity. For example, RR and SP patients were identified in each of the four MRI phenotype groups, in a manner that would contradict the intuitive understanding of these clinical stages of MS. In particular, 9% of the Type I (mild) MRI phenotype patients had a SP (advanced) clinical
Fig. 2. Representative proton density MRI scans showing the MRI-defined phenotypes. Four patients are shown as examples of the four phenotypes; Type I (mild atrophy/low T2LV); Type II (mild atrophy/high T2LV); Type III (high atrophy/low T2LV); Type IV (high atrophy/high T2LV); RR = relapsing-remitting; EDSS = Expanded Disability Status Scale score; DD = disease duration (years); T2LV = total brain T2 hyperintense lesion volume (cc); BPF = brain parenchymal fraction.
Please cite this article as: Tauhid S, et al, MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis, J Neurol Sci (2014), http://dx.doi.org/10.1016/j.jns.2014.08.047
4
S. Tauhid et al. / Journal of the Neurological Sciences xxx (2014) xxx–xxx
Table 3 MRI–MRI and MRI–disability correlations. Correlation examined
Type I (n = 67)
Type II (n = 21)
Type III (n = 21)
Type IV (n = 66)
Entire cohort (n = 175)
T2LV–BPF BPF–EDSS T2LV–EDSS
−0.06 (0.656) −0.35 (0.003)⁎ 0.11 (0.383)
−0.16 (0.491) −0.21 (0.356) 0.18 (0.433)
−0.32 (0.158) 0.30 (0.185) 0.28 (0.225)
−0.49 (b0.0001)⁎ −0.46 (0.0001)⁎ 0.36 (0.003)⁎
−0.66 (b0.0001)⁎ −0.57 (b0.0001)⁎ 0.56 (b0.0001)⁎
Key: rs (p) values are shown for Spearman correlation coefficients; Type I (mild atrophy/low T2LV); Type II (mild atrophy/high T2LV); Type III (high atrophy/low T2LV); Type IV (high atrophy/high T2LV); EDSS = Expanded Disability Status Scale score; T2LV = total brain T2 hyperintense lesion volume (cc); BPF = brain parenchymal fraction. ⁎ p b 0.05.
course, and 50% of the Type IV patients had a RR clinical course (not yet progressed to SP). This paradox is underscored by the fact that the range of EDSS scores was similar between Type I and Type IV groups. Spinal cord involvement (not considered in the present study) might explain the presence of SP patients in the Type I MRI group, given that spinal cord atrophy is a hallmark of SPMS [23] and such involvement is poorly reflected in MRI brain measures of pathology [5]. Why are there so many patients with Type IV pathology in the RR (rather than SP) stage and carrying relatively low levels of disability? These patients may have enhanced cerebral reserve capacity [24], sparing of eloquent substructures/pathways of the brain [25], or relative little spinal cord involvement [5] (analyses which were not performed in the present study). Because this was a cross-sectional study, we were not able to determine if the more mildly affected Type IV patients were at risk for imminent clinical progression. As further evidence of the unique information provided by the MRI phenotype grouping, Type II and Type III patients showed a similar number of RR and SP patients. A longstanding challenge in the MS field has been the poor correlation between brain MRI and neurologic status [26]; this clinical–MRI paradox has traditionally referred to focal brain lesions. While measurement of brain atrophy has improved MRI correlations with disability and cognitive impairment, the association with clinical status has remained inconsistent (and moderate at best) [5–8, 15–17, 19, 26–28]. In the present study, we found a striking difference among groups in modeling the relationship between brain MRI and disability in that three of the groups (Types I, II and III) showed no relationship between either T2LV or BPF and EDSS score; while, Type IV patients showed significant but moderate associations. Our data thus indicate that disease heterogeneity is one factor contribution to the relationship between global MRI and physical disability. Moreover we can bring forward a “two-hit” hypothesis based on these results that both inflammation and neurodegeneration need to be substantially present in a given patient to exceed cerebral reserve capacity and thus cause clinical impairment. Perhaps the first use of a clustering method to classify patients with MS on the basis of MRI-defined inflammation or neurodegeneration was introduced by Bielekova et al. [3]. These investigators used the presence of gadolinium-enhancement on three consecutive monthly scans to initially stratify patients as having either low or high active inflammation. Subjects were then subdivided further by BPF and the ratio of T1 hypointense to T2 hyperintense lesions. Although related, ours was a more simplified classification system requiring only one imaging time point and the availability of T2LV and BPF, which does not consider acute/recent inflammatory activity. We focused disease severity rather than disease activity in creating these MRI phenotypes. Our exploratory study should serve as a basis for confirmation and extension in future studies. A prospective study is warranted to confirm and extend these findings. While the overall sample size was large for an MS-MRI clinical correlation study, the numbers of patients assigned to Type II and Type III were relatively small. It remains to be determined whether consideration of other advanced structural aspects of MRIdefined CNS related damage [7] such as imaging parameters focused on the gray matter, spinal cord, normal-appearing white matter, destructive potential of lesions, or mapping of specific sites/tracts of involvement would help to further refine the MRI phenotype definitions. Given the emerging data suggesting a link between genetics, immune
regulation, and inflammatory or neurodegenerative aspects of MS disease pathogenesis [29–35], it would be of interest to investigate if the MRI phenotypes are associated with specific biomarkers. Because most of our patients were treated with disease modifying medications, this might alter the natural rate or degree of development lesion and atrophy metrics. It would be of interest to assess if these phenotypic definitions persist when considering untreated patients. This may not be feasible in the current treatment era, requiring access to historic data sets. Finally, it remains to be determined in longitudinal studies, the treatment response and rate/factors associated with a change in a patient's MRI phenotype. Conflict of interest and source of funding None. Acknowledgment We are grateful to Antonia Ceccarelli, Velina Sevdalinova Chavarro, Svetlana Egorova, Mark Anderson, and Mariann Polgar-Turcsanyi for the technical assistance and Charles Guttmann for the helpful discussions. References [1] Polman CH, Reingold SC, Edan G, Filippi M, Hartung HP, Kappos L, et al. Diagnostic criteria for multiple sclerosis: 2005 revisions to the “McDonald Criteria”. Ann Neurol 2005;58:840–6. [2] Metz I, Weigand SD, Popescu BF, Frischer JM, Parisi JE, Guo Y, et al. Pathologic heterogeneity persists in early active multiple sclerosis lesions. Ann Neurol 2014; 75:728–38. [3] Bielekova B, Kadom N, Fisher E, Jeffries N, Ohayon J, Richert N, et al. MRI as a marker for disease heterogeneity in multiple sclerosis. Neurology 2005;65:1071–6. [4] Zwemmer JN, Berkhof J, Castelijns JA, Barkhof F, Polman CH, Uitdehaag BM. Classification of multiple sclerosis patients by latent class analysis of magnetic resonance imaging characteristics. Mult Scler 2006;12:565–72. [5] Cohen AB, Neema M, Arora A, Dell'oglio E, Benedict RH, Tauhid S, et al. The relationships among MRI-defined spinal cord involvement, brain involvement, and disability in multiple sclerosis. J Neuroimaging 2012;22:122–8. [6] Bermel RA, Bakshi R. The measurement and clinical relevance of brain atrophy in multiple sclerosis. Lancet Neurol 2006;5:158–70. [7] Bakshi R, Neema M, Healy BC, Liptak Z, Betensky RA, Buckle GJ, et al. Predicting clinical progression in multiple sclerosis with the magnetic resonance disease severity scale. Arch Neurol 2008;65:1449–53. [8] Rudick RA, Fisher E, Lee JC, Simon J, Jacobs L. Multiple Sclerosis Collaborative Research Group. Use of the brain parenchymal fraction to measure whole brain atrophy in relapsing remitting MS. Neurology 1999;53:1698–704. [9] Sormani MP, Arnold DL, De Stefano N. Treatment effect on brain atrophy correlates with treatment effect on disability in multiple sclerosis. Ann Neurol 2014;75:43–9. [10] Jacobs LD, Wende KE, Brownscheidle CM, Apatoff B, Coyle PK, Goodman A, et al. A profile of multiple sclerosis: the New York State Multiple Sclerosis Consortium (NYSMSC). Mult Scler 1999;5:369–76. [11] Holland CM, Charil A, Csapo I, Ichese M, Khoury SJ, Bakshi R, et al. The relationship between normal cerebral perfusion patterns and white matter lesion distribution in 1,249 patients with multiple sclerosis. J Neuroimaging 2012;22:129–36. [12] Wei X, Warfield S, Zou KH, Wu Y, Li X, Guimond A, et al. Quantitative analysis of MRI signal abnormalities of brain white matter with high reproducibility and accuracy. J Magn Reson Imaging 2002;15:203–9. [13] Filippi M, Rocca MA, Barkhof F, Brück W, Chen JT, Comi G, et al. Association between pathological and MRI findings in multiple sclerosis. Lancet Neurol 2012;11:349–60. [14] Sabo JK, Cate HS. Signalling pathways that inhibit the capacity of precursor cells for myelin repair. Int J Mol Sci 2013;14:1031–49. [15] Sanfilipo MP, Benedict RH, Sharma J, Weinstock-Guttman B, Bakshi R. The relationship between whole brain volume and disability in multiple sclerosis: a comparison of normalized gray vs. white matter with misclassification correction. Neuroimage 2005;26:1068–77.
Please cite this article as: Tauhid S, et al, MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis, J Neurol Sci (2014), http://dx.doi.org/10.1016/j.jns.2014.08.047
S. Tauhid et al. / Journal of the Neurological Sciences xxx (2014) xxx–xxx [16] Shiee N, Bazin PL, Zackowski KM, Farrell SK, Harrison DM, Newsome SD, et al. Revisiting brain atrophy and its relationship to disability in multiple sclerosis. PLoS One 2012;7:e37049. [17] Klawiter EC, Ceccarelli A, Arora A, Jackson J, Bakshi S, Kim G, et al. Corpus callosum atrophy correlates with gray matter atrophy in patients with multiple sclerosis. J Neuroimaging [Epub ahead of print May 9, 2014; doi: http://dx.doi.org/10.1111/ jon.12124] [18] Politis M, Giannetti P, Su P, Turkheimer F, Keihaninejad S, Wu K, et al. Increased PK11195 PET binding in the cortex of patients with MS correlates with disability. Neurology 2012;79:523–30. [19] Calabrese M, Agosta F, Rinaldi F, Mattisi I, Grossi P, Favaretto A, et al. Cortical lesions and atrophy associated with cognitive impairment in relapsing-remitting multiple sclerosis. Arch Neurol 2009;66:1144–50. [20] Nelson F, Datta S, Garcia N, Rozario NL, Perez F, Cutter G, et al. Intracortical lesions by 3 T magnetic resonance imaging and correlation with cognitive impairment in multiple sclerosis. Mult Scler 2011;17:1122–9. [21] Howell OW, Reeves CA, Nicholas R, Carassiti D, Radotra B, Gentleman SM, et al. Meningeal inflammation is widespread and linked to cortical pathology in multiple sclerosis. Brain 2011;134:2755–71. [22] Bermel RA, Puli SR, Rudick RA, Weinstock-Guttman B, Fisher E, Munschauer FE, et al. Prediction of longitudinal brain atrophy in multiple sclerosis by gray matter magnetic resonance imaging T2 hypointensity. Arch Neurol 2005;62:1371–6. [23] Horsfield MA, Sala S, Neema M, Absinta M, Bakshi A, Sormani MP, et al. Rapid semiautomatic segmentation of the spinal cord from magnetic resonance images: application in multiple sclerosis. Neuroimage 2010;50:446–55. [24] Weiss S, Mori F, Rossi S, Centonze D. Disability in multiple sclerosis: when synaptic long-term potentiation fails. Neurosci Biobehav Rev 2014;43C:88–99. [25] Llufriu S, Martinez-Heras E, Fortea J, Blanco Y, Berenguer J, Gabilondo I, et al. Cognitive functions in multiple sclerosis: impact of gray matter integrity. Mult Scler 2014;20:424–32.
5
[26] Zivadinov R, Leist TP. Clinical-magnetic resonance imaging correlations in multiple sclerosis. J Neuroimaging 2005;15(4 Suppl):10S–21S. [27] Amato MP, Bartolozzi ML, Zipoli V, Portaccio E, Mortilla M, Guidi L, et al. Neocortical volume decrease in relapsing-remitting MS patients with mild cognitive impairment. Neurology 2004;63:89–93. [28] Bakshi R, Thompson AJ, Rocca MA, Pelletier D, Dousset V, Barkhof F, et al. MRI in multiple sclerosis: current status and future prospects. Lancet Neurol 2008;7: 615–25. [29] Eikelenboom MJ, Petzold A, Lazeron RH, Silber E, Sharief M, Thompson EJ, et al. Multiple sclerosis: neurofilament light chain antibodies are correlated to cerebral atrophy. Neurology 2003;60:219–23. [30] Rinaldi L, Gallo P, Calabrese M, Ranzato F, Luise D, Colavito D, et al. Longitudinal analysis of immune cell phenotypes in early stage multiple sclerosis: distinctive patterns characterize MRI-active patients. Brain 2006;129:1993–2007. [31] Zivadinov R, Uxa L, Bratina A, Bosco A, Srinivasaraghavan B, Minagar A, et al. HLADRB1*1501, -DQB1*0301, -DQB1*0302, -DQB1*0602, and -DQB1*0603 alleles are associated with more severe disease outcome on MRI in patients with multiple sclerosis. Int Rev Neurobiol 2007;79:521–35. [32] Greer JM, Csurhes PA, Muller DM, Pender MP. Correlation of blood T cell and antibody reactivity to myelin proteins with HLA type and lesion localization in multiple sclerosis. J Immunol 2008;180:6402–10. [33] Dinacci D, Tessitore A, Russo A, De Bonis ML, Lavorgna L, Picconi O, et al. BDNF Val66Met polymorphism and brain volumes in multiple sclerosis. Neurol Sci 2011; 32:117–23. [34] Magraner MJ, Bosca I, Simó-Castelló M, García-Martí G, Alberich-Bayarri A, Coret F, et al. Brain atrophy and lesion load are related to CSF lipid-specific IgM oligoclonal bands in clinically isolated syndromes. Neuroradiology 2012;54:5–12. [35] DeLuca GC, Alterman R, Martin JL, Mittal A, Blundell S, Bird S. Casting light on multiple sclerosis heterogeneity: the role of HLA-DRB1 on spinal cord pathology. Brain 2013;136:1025–34.
Please cite this article as: Tauhid S, et al, MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis, J Neurol Sci (2014), http://dx.doi.org/10.1016/j.jns.2014.08.047
Contents lists available at ScienceDirect
Journal of the Neurological Sciences journal homepage: www.elsevier.com/locate/jns
MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis Shahamat Tauhid, Mohit Neema, Brian C. Healy, Howard L. Weiner, Rohit Bakshi ⁎ Department of Neurology, Brigham and Women's Hospital, Laboratory for Neuroimaging Research, Partners MS Center, Harvard Medical School, Boston, MA, USA
a r t i c l e
i n f o
Article history: Received 13 June 2014 Received in revised form 29 July 2014 Accepted 29 August 2014 Available online xxxx Keywords: MRI Multiple sclerosis Lesions Atrophy Phenotypes Disability
a b s t r a c t Background: While disease categories (i.e. clinical phenotypes) of multiple sclerosis (MS) are established, there remains MRI heterogeneity among patients within those definitions. MRI-defined lesions and atrophy show only moderate inter-correlations, suggesting that they represent partly different processes in MS. We assessed the ability of MRI-based categorization of cerebral lesions and atrophy in individual patients to identify distinct phenotypes. Methods: We studied 175 patients with MS [age (mean ± SD) 42.7 ± 9.1 years, 124 (71%) women, Expanded Disability Status (EDSS) score 2.5 ± 2.3, n = 18 (10%) clinically isolated demyelinating syndrome (CIS), n = 115 (66%) relapsing-remitting (RR), and n = 42 (24%) secondary progressive (SP)]. Brain MRI measures included T2 hyperintense lesion volume (T2LV) and brain parenchymal fraction (to assess whole brain atrophy). Medians were used to create bins for each parameter, with patients assigned a low or high severity score. Results: Four MRI phenotype categories emerged: Type I = low T2LV/mild atrophy [n = 67 (38%); CIS = 14, RR = 47, SP = 6]; Type II = high T2LV/mild atrophy [n = 21 (12%); RR = 19, SP = 2]; Type III = low T2LV/ high atrophy [n = 21 (12%); CIS = 4, RR = 16, SP = 1]; and Type IV = high T2LV/high atrophy [n = 66 (38%); RR = 33, SP = 33]. Type IV was the most disabled and was the only group showing a correlation between T2LV vs. BPF and MRI vs. EDSS score (all p b 0.05). Conclusions: We described MRI-categorization based on the relationship between lesions and atrophy in individual patients to identify four phenotypes in MS. Most patients have congruent extremes related to the degree of lesions and atrophy. However, many have a dissociation. Longitudinal studies will help define the stability of these patterns and their role in risk stratification. © 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-SA license (http://creativecommons.org/licenses/by-nc-sa/3.0/).
1. Introduction While clinically-defined disease categories of multiple sclerosis (MS) are well established [1], there remains heterogeneity of the disease among patients that appears to exist within the confines of these subtypes. This heterogeneity includes the concepts of patientdependent longitudinally-consistent immunopathological lesion subtypes [2] and clustering of patient subgroups based on MRI-defined lesion characteristics and the level of tissue destruction [3,4]. Consistent with the concept of disease heterogeneity, clinical progression and treatment response are difficult to predict in individual patients. Furthermore, there is a topographic uncoupling of structural changes such as the lack of a relationship between brain and spinal cord involvement [5]. Taken
⁎ Corresponding author at: Laboratory for Neuroimaging Research, One Brookline Place, Brookline, MA 02445, USA. Tel.: +1 617 525 6550. E-mail address: [email protected] (R. Bakshi).
together, these data suggest that the pathogenesis of tissue injury may differ among patients with MS. Among the types of CNS tissue injury that typify the disease, the two broadest categories include inflammatory demyelinating foci (lesions) and tissue destruction (neurodegeneration; axonal and neuronal loss). The most commonly-used and readily-available MRI measure of lesion load is hyperintensity on T2-weighted images. To assess overall tissue destruction, MRI-quantified normalized whole brain volume is an established measure of brain atrophy [6]. Several lines of evidence indicate that lesions and atrophy represent somewhat unrelated aspects of the disease process. This includes the observation that in both crosssectional [6,7] and longitudinal [6,8] studies, lesions show a weak relationship to current and subsequent atrophy. Furthermore, the two measures provide complementary information in predicting treatment efficacy [9]. The objective of this study was to assess the ability of MRI-based categorization of the relationship between the severity of cerebral lesions and atrophy in individual patients to identify distinct MS phenotypes. This work was presented in preliminary form at the 2013 annual meeting of the American Academy of Neurology, San Diego.
http://dx.doi.org/10.1016/j.jns.2014.08.047 0022-510X/© 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-SA license (http://creativecommons.org/licenses/by-nc-sa/3.0/).
Please cite this article as: Tauhid S, et al, MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis, J Neurol Sci (2014), http://dx.doi.org/10.1016/j.jns.2014.08.047
2
S. Tauhid et al. / Journal of the Neurological Sciences xxx (2014) xxx–xxx
2. Methods 2.1. Subjects and clinical evaluation Demographic and clinical characteristics of the subjects are summarized in Table 1. This was a single-center cross-sectional retrospective study of 175 consecutive patients who met the following inclusion criteria: 1) Diagnosed with a relapsing form of MS including a clinically-isolated demyelinating syndrome (CIS), relapsing-remitting (RR), or secondary progressive (SP) [1]; 2) Evaluated clinically by an MS specialist provider at the Brigham and Women's Hospital MS clinic (Partners MS Center) within three months of MRI; 3) age 18–60 years. Nearly all patients were receiving disease modifying therapy, as is the common practice at our center [7]. Baseline characteristics, including demographic, clinical, and MRI (Table 1), were comparable to those of large MS cohorts [10,11], indicating that our cohort represented a wide range of the MS spectrum that includes the continuum CIS–RR– SP. We did not include primary progressive MS patients because of the challenging diagnosis [1], the small numbers of those patients available, and that concept that this group may represent a separate entity. This study was approved by our institutional review board.
Fig. 1. Method of creation of MRI-based phenotypes. Patients were stratified by median splits of the entire cohort based on total cerebral MRI-measures of inflammation (T2LV) and atrophy (BPF); the median was used to create bins for each MRI parameter, with patients assigned a score of low or high severity; T2LV = total brain T2 hyperintense lesion volume (cc); BPF = brain parenchymal fraction; A = atrophy; L = T2LV; Type I (mild atrophy/low T2LV); Type II (mild atrophy/high T2LV); Type III (high atrophy/low T2LV); Type IV (high atrophy/high T2LV).
2.2. MRI acquisition and segmentation Brain MRI was obtained in all subjects at 1.5 T with a 3 mm thick, gapless, T2-weighted, axial dual-echo protocol that included the entire brain. With few exceptions, the protocol was performed on the same scanner and included the following consistent acquisition parameters: repetition time 3000 ms, echo time 30/80 ms, and 0.93 × 0.93 mm pixel size. Using automated template-driven segmentation (TDS +) from the dual echo images, total brain T2 hyperintense white matter lesion volume (T2LV) and normalized whole brain volume [brain parenchymal fraction (BPF) — a surrogate of whole brain atrophy] [6] were automatically determined [12], after which expert manual correction was applied to each output map to ensure accuracy. 2.3. Creation of the MRI-defined phenotypes Patients were stratified into MRI phenotype groups based on a median split (0 or 1) of T2LV and BPF. As shown in Fig. 1, all patients were assigned to one of four unique groups as: Type I = mild atrophy (0) and low T2LV (0) Type II = mild atrophy (0) and high T2LV (1) Type III = high atrophy (1) and low T2LV (0) Type IV = high atrophy (1) and high T2LV (1). 2.4. Statistical analysis The demographic, clinical and MRI characteristics were compared across the four groups using one-way analysis of variance for continuous outcomes (age, disease duration, BPF, and T2LV), a proportional odds model for ordinal outcomes (EDSS) and a chi-square test for nominal Table 1 Demographic and clinical characteristics of the subjects. Number of patients Clinical MS phenotype: • Clinically isolated syndrome • Relapsing-remitting MS • Secondary progressive MS Number of women (n, %) Age (mean ± SD) (range) (years) Expanded Disability Status Scale score Disease duration (years since first symptoms) Total brain T2 hyperintense lesion volume (cc) Brain parenchymal fraction
n = 175 n = 18 (10%) n = 115 (66%) n = 42 (24%) n = 124 (71%) 42.7 ± 9.1 (21–60) 2.5 ± 2.3 (0–8.6) 9.6 ± 8.6 (0.2–48.0) 8.7 ± 10.2 (0.5–57.1) 0.844 ± 0.049 (0.67–0.96)
variables (gender and disease category). If a significant difference was observed in the four group comparison, pairwise group comparisons were completed and Holm's correction for multiple comparisons was applied to adjust the p-values. In addition, we assessed the relationship among the MRI measures and the relationship between MRI and EDSS score using the Spearman's correlation coefficient. These were analyzed for the whole cohort and within each MRI phenotype group. 3. Results Four categories emerged from the median split analysis (Table 2, Fig. 2). Regarding clinical and demographic group comparisons (Table 2), Type IV patients had higher disease duration, higher EDSS scores, and a higher proportion of SP patients than the other three groups (all p b 0.05). Type III and IV patients were each older than the Type I and II patients. Otherwise, there were no differences in age, disease duration, EDSS score, or clinical disease category distribution among groups (p b 0.05). There were no differences in the gender breakdown across the groups (p N 0.05). Regarding the MRI parameters that comprised the phenotypes (Table 2), all pairwise comparisons showed significant differences in T2LV among the groups (p b 0.05), except for Type I vs. Type III (p N 0.05). In addition, all pairwise comparisons showed significant differences in BPF among the groups (p b 0.05) except Type I vs. Type II (p N 0.05). These results confirm the rationale of defining unique MRI phenotypes on the basis of different T2LV and BPF characteristics. We also tested the relationships between lesions and atrophy and between MRI and disability (Table 3). The correlation between T2LV and BPF was significant in the overall cohort (p b 0.0001) and in the Type IV patients (p b 0.0001) but not in the other three groups (p N 0.05). Similarly, the correlation between each of the two MRI parameters (T2LV and BPF) and EDSS score was significant in the overall cohort (p b 0.0001) and in the Type IV patients (p b 0.0001) but not in the other three groups (p N 0.05). 4. Discussion A major finding in this study is that 25% patients had a discordance of where their brain lesions and atrophy fell with regard to the medians of the full cohort. The discordance was equally bidirectional in that about half of the discordant patients had a high degree of atrophy
Please cite this article as: Tauhid S, et al, MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis, J Neurol Sci (2014), http://dx.doi.org/10.1016/j.jns.2014.08.047
S. Tauhid et al. / Journal of the Neurological Sciences xxx (2014) xxx–xxx
3
Table 2 Demographic, clinical and MRI characteristics of the MRI-defined phenotypes.
Number of patients (%) Disease category: • Clinically isolated syndrome • Relapsing-remitting MS • Secondary progressive MS Number of women (n, %) Age (mean ± SD) (range) (years) Expanded Disability Status Scale score Disease duration (years) T2LV (cc) Brain parenchymal fraction
Type I (mild atrophy/ low T2LV)
Type II (mild atrophy/ high T2LV)
Type III (high atrophy/ low T2LV)
Type IV (high atrophy/ high T2LV)
n = 67 (38%)
n = 21 (12%)
n = 21 (12%)
n = 66 (38%)
14 (21%) 47 (70%) 6 (9%) n = 51 (76%) 37.3 ± 8.3 (21.0–54.1) 1.5 ± 1.8 (0–8.5) 5.5 ± 5.7 (0.26–26.0) 2.3 ± 1.0 (0.46–4.21) 0.888 ± 0.025 (0.85–0.93)
0 19 (91%) 2 (9%) n = 16 (76%) 36.9 ± 7.9 (21.7–50.2) 1.9 ± 2.0 (07.5) 8.6 ± 6.6 (0.57–22.0) 8.2 ± 4.9 (4.5–22.2) 0.875 ± 0.025 (0.85–0.93)
4 (19.0%) 16 (76%) 1 (5%) n = 14 (67%) 43.7 ± 7.8 (21.0–54.3) 1.7 ± 1.8 (0–6.5) 8.6 ± 8.1 (0.4–26.0) 2.3 ± 1.1 (0.8–4.1) 0.834 ± 0.016 (0.78–0.85)
0 33 (50%) 33 (50%) n = 43 (65%) 44.0 ± 8.6 (26.5–60.2) 3.9 ± 2.2 (1.0–8.5) 14.3 ± 9.4 (0.25–48.0) 17.3 ± 11.8 (4.4–57.1) 0.791 ± 0.041 (0.67–0.85)
Key: T2LV = total brain T2 hyperintense lesion volume. Regarding the breakdown of women vs. men, no significant differences were observed across the groups (p N 0.05). Regarding age, Types III and IV were significantly older than Types I and II (p b 0.05). Regarding disease category (i.e. clinical phenotype), EDSS, and disease duration, Type IV was significantly more likely to be SPMS, disabled, with higher disease duration than the other three categories (p b 0.05). Regarding T2LV, all pairwise differences except Type I vs III were statistically significant (p b 0.05). Regarding brain parenchymal fraction, all pairwise differences except Type I vs. Type II were statistically significant (p b 0.05).
despite a low lesion load (Type II MRI phenotype) and vice-versa (Type III). This was further underscored by the poor correlation in both groups between T2LV and BPF. Taken together these findings suggest that subgroups of patients have an uncoupling between focal immunemediated demyelination in white matter (inflammation) and overall global cerebral tissue destruction (neurodegeneration). Regarding the Type II patients, who have a high T2LV but low atrophy, this might relate in part to the long held view that T2 hyperintense lesions are non-specific for pathologic changes, reflecting a wide range of severity [13]. We hypothesize that Type II patients have a T2LV that is dominated by mild changes such as mild inflammation, edema, and partial demyelination. Another factor serving to limit overall atrophy in these patients may be the presence of enhanced repair capacity [14] or other resistance to the downstream effects of white matter injury on neuronal and axonal integrity. Regarding the Type III patients, who have a low T2LV but higher degree of atrophy, there may be other contributions to tissue destruction that are not captured by overt lesion load. Considering the prevailing notion that whole brain atrophy in MS is dominated by gray matter rather than white matter atrophy, consistently seen in MRI studies at 1.5 T and 3 T [6,15–17], the Type III patients may be affected by gray
matter injury due to a variety of mechanisms that are separate from white matter lesions (and were not assessed in the present study) such as microglial activation [18], cortical lesions [19,20], meningeal inflammation [21], and iron deposition/oxidative stress [22]. Type 1 and Type IV phenotypes were characterized by a congruent level of lesions and atrophy with each either of a lower (Type I) or higher (Type IV) degree. Comparing these two groups, Type I patients tended to be at an earlier stage of disease on other disease characteristics, showing a lower age, lower disease duration, and lower level of disability. In addition, the Type I group was represented by a preponderance of CIS or RR rather than SP patients. Type IV patients represent more advanced disease, owing to the higher disease duration and higher disability level than the other three groups. Another noteworthy finding in this study is that the MRI phenotypes belie the confines of the current MS disease categories (i.e. clinical phenotypes), readily crossing those boundaries, suggesting that the MRI categorization provides unique information about disease severity. For example, RR and SP patients were identified in each of the four MRI phenotype groups, in a manner that would contradict the intuitive understanding of these clinical stages of MS. In particular, 9% of the Type I (mild) MRI phenotype patients had a SP (advanced) clinical
Fig. 2. Representative proton density MRI scans showing the MRI-defined phenotypes. Four patients are shown as examples of the four phenotypes; Type I (mild atrophy/low T2LV); Type II (mild atrophy/high T2LV); Type III (high atrophy/low T2LV); Type IV (high atrophy/high T2LV); RR = relapsing-remitting; EDSS = Expanded Disability Status Scale score; DD = disease duration (years); T2LV = total brain T2 hyperintense lesion volume (cc); BPF = brain parenchymal fraction.
Please cite this article as: Tauhid S, et al, MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis, J Neurol Sci (2014), http://dx.doi.org/10.1016/j.jns.2014.08.047
4
S. Tauhid et al. / Journal of the Neurological Sciences xxx (2014) xxx–xxx
Table 3 MRI–MRI and MRI–disability correlations. Correlation examined
Type I (n = 67)
Type II (n = 21)
Type III (n = 21)
Type IV (n = 66)
Entire cohort (n = 175)
T2LV–BPF BPF–EDSS T2LV–EDSS
−0.06 (0.656) −0.35 (0.003)⁎ 0.11 (0.383)
−0.16 (0.491) −0.21 (0.356) 0.18 (0.433)
−0.32 (0.158) 0.30 (0.185) 0.28 (0.225)
−0.49 (b0.0001)⁎ −0.46 (0.0001)⁎ 0.36 (0.003)⁎
−0.66 (b0.0001)⁎ −0.57 (b0.0001)⁎ 0.56 (b0.0001)⁎
Key: rs (p) values are shown for Spearman correlation coefficients; Type I (mild atrophy/low T2LV); Type II (mild atrophy/high T2LV); Type III (high atrophy/low T2LV); Type IV (high atrophy/high T2LV); EDSS = Expanded Disability Status Scale score; T2LV = total brain T2 hyperintense lesion volume (cc); BPF = brain parenchymal fraction. ⁎ p b 0.05.
course, and 50% of the Type IV patients had a RR clinical course (not yet progressed to SP). This paradox is underscored by the fact that the range of EDSS scores was similar between Type I and Type IV groups. Spinal cord involvement (not considered in the present study) might explain the presence of SP patients in the Type I MRI group, given that spinal cord atrophy is a hallmark of SPMS [23] and such involvement is poorly reflected in MRI brain measures of pathology [5]. Why are there so many patients with Type IV pathology in the RR (rather than SP) stage and carrying relatively low levels of disability? These patients may have enhanced cerebral reserve capacity [24], sparing of eloquent substructures/pathways of the brain [25], or relative little spinal cord involvement [5] (analyses which were not performed in the present study). Because this was a cross-sectional study, we were not able to determine if the more mildly affected Type IV patients were at risk for imminent clinical progression. As further evidence of the unique information provided by the MRI phenotype grouping, Type II and Type III patients showed a similar number of RR and SP patients. A longstanding challenge in the MS field has been the poor correlation between brain MRI and neurologic status [26]; this clinical–MRI paradox has traditionally referred to focal brain lesions. While measurement of brain atrophy has improved MRI correlations with disability and cognitive impairment, the association with clinical status has remained inconsistent (and moderate at best) [5–8, 15–17, 19, 26–28]. In the present study, we found a striking difference among groups in modeling the relationship between brain MRI and disability in that three of the groups (Types I, II and III) showed no relationship between either T2LV or BPF and EDSS score; while, Type IV patients showed significant but moderate associations. Our data thus indicate that disease heterogeneity is one factor contribution to the relationship between global MRI and physical disability. Moreover we can bring forward a “two-hit” hypothesis based on these results that both inflammation and neurodegeneration need to be substantially present in a given patient to exceed cerebral reserve capacity and thus cause clinical impairment. Perhaps the first use of a clustering method to classify patients with MS on the basis of MRI-defined inflammation or neurodegeneration was introduced by Bielekova et al. [3]. These investigators used the presence of gadolinium-enhancement on three consecutive monthly scans to initially stratify patients as having either low or high active inflammation. Subjects were then subdivided further by BPF and the ratio of T1 hypointense to T2 hyperintense lesions. Although related, ours was a more simplified classification system requiring only one imaging time point and the availability of T2LV and BPF, which does not consider acute/recent inflammatory activity. We focused disease severity rather than disease activity in creating these MRI phenotypes. Our exploratory study should serve as a basis for confirmation and extension in future studies. A prospective study is warranted to confirm and extend these findings. While the overall sample size was large for an MS-MRI clinical correlation study, the numbers of patients assigned to Type II and Type III were relatively small. It remains to be determined whether consideration of other advanced structural aspects of MRIdefined CNS related damage [7] such as imaging parameters focused on the gray matter, spinal cord, normal-appearing white matter, destructive potential of lesions, or mapping of specific sites/tracts of involvement would help to further refine the MRI phenotype definitions. Given the emerging data suggesting a link between genetics, immune
regulation, and inflammatory or neurodegenerative aspects of MS disease pathogenesis [29–35], it would be of interest to investigate if the MRI phenotypes are associated with specific biomarkers. Because most of our patients were treated with disease modifying medications, this might alter the natural rate or degree of development lesion and atrophy metrics. It would be of interest to assess if these phenotypic definitions persist when considering untreated patients. This may not be feasible in the current treatment era, requiring access to historic data sets. Finally, it remains to be determined in longitudinal studies, the treatment response and rate/factors associated with a change in a patient's MRI phenotype. Conflict of interest and source of funding None. Acknowledgment We are grateful to Antonia Ceccarelli, Velina Sevdalinova Chavarro, Svetlana Egorova, Mark Anderson, and Mariann Polgar-Turcsanyi for the technical assistance and Charles Guttmann for the helpful discussions. References [1] Polman CH, Reingold SC, Edan G, Filippi M, Hartung HP, Kappos L, et al. Diagnostic criteria for multiple sclerosis: 2005 revisions to the “McDonald Criteria”. Ann Neurol 2005;58:840–6. [2] Metz I, Weigand SD, Popescu BF, Frischer JM, Parisi JE, Guo Y, et al. Pathologic heterogeneity persists in early active multiple sclerosis lesions. Ann Neurol 2014; 75:728–38. [3] Bielekova B, Kadom N, Fisher E, Jeffries N, Ohayon J, Richert N, et al. MRI as a marker for disease heterogeneity in multiple sclerosis. Neurology 2005;65:1071–6. [4] Zwemmer JN, Berkhof J, Castelijns JA, Barkhof F, Polman CH, Uitdehaag BM. Classification of multiple sclerosis patients by latent class analysis of magnetic resonance imaging characteristics. Mult Scler 2006;12:565–72. [5] Cohen AB, Neema M, Arora A, Dell'oglio E, Benedict RH, Tauhid S, et al. The relationships among MRI-defined spinal cord involvement, brain involvement, and disability in multiple sclerosis. J Neuroimaging 2012;22:122–8. [6] Bermel RA, Bakshi R. The measurement and clinical relevance of brain atrophy in multiple sclerosis. Lancet Neurol 2006;5:158–70. [7] Bakshi R, Neema M, Healy BC, Liptak Z, Betensky RA, Buckle GJ, et al. Predicting clinical progression in multiple sclerosis with the magnetic resonance disease severity scale. Arch Neurol 2008;65:1449–53. [8] Rudick RA, Fisher E, Lee JC, Simon J, Jacobs L. Multiple Sclerosis Collaborative Research Group. Use of the brain parenchymal fraction to measure whole brain atrophy in relapsing remitting MS. Neurology 1999;53:1698–704. [9] Sormani MP, Arnold DL, De Stefano N. Treatment effect on brain atrophy correlates with treatment effect on disability in multiple sclerosis. Ann Neurol 2014;75:43–9. [10] Jacobs LD, Wende KE, Brownscheidle CM, Apatoff B, Coyle PK, Goodman A, et al. A profile of multiple sclerosis: the New York State Multiple Sclerosis Consortium (NYSMSC). Mult Scler 1999;5:369–76. [11] Holland CM, Charil A, Csapo I, Ichese M, Khoury SJ, Bakshi R, et al. The relationship between normal cerebral perfusion patterns and white matter lesion distribution in 1,249 patients with multiple sclerosis. J Neuroimaging 2012;22:129–36. [12] Wei X, Warfield S, Zou KH, Wu Y, Li X, Guimond A, et al. Quantitative analysis of MRI signal abnormalities of brain white matter with high reproducibility and accuracy. J Magn Reson Imaging 2002;15:203–9. [13] Filippi M, Rocca MA, Barkhof F, Brück W, Chen JT, Comi G, et al. Association between pathological and MRI findings in multiple sclerosis. Lancet Neurol 2012;11:349–60. [14] Sabo JK, Cate HS. Signalling pathways that inhibit the capacity of precursor cells for myelin repair. Int J Mol Sci 2013;14:1031–49. [15] Sanfilipo MP, Benedict RH, Sharma J, Weinstock-Guttman B, Bakshi R. The relationship between whole brain volume and disability in multiple sclerosis: a comparison of normalized gray vs. white matter with misclassification correction. Neuroimage 2005;26:1068–77.
Please cite this article as: Tauhid S, et al, MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis, J Neurol Sci (2014), http://dx.doi.org/10.1016/j.jns.2014.08.047
S. Tauhid et al. / Journal of the Neurological Sciences xxx (2014) xxx–xxx [16] Shiee N, Bazin PL, Zackowski KM, Farrell SK, Harrison DM, Newsome SD, et al. Revisiting brain atrophy and its relationship to disability in multiple sclerosis. PLoS One 2012;7:e37049. [17] Klawiter EC, Ceccarelli A, Arora A, Jackson J, Bakshi S, Kim G, et al. Corpus callosum atrophy correlates with gray matter atrophy in patients with multiple sclerosis. J Neuroimaging [Epub ahead of print May 9, 2014; doi: http://dx.doi.org/10.1111/ jon.12124] [18] Politis M, Giannetti P, Su P, Turkheimer F, Keihaninejad S, Wu K, et al. Increased PK11195 PET binding in the cortex of patients with MS correlates with disability. Neurology 2012;79:523–30. [19] Calabrese M, Agosta F, Rinaldi F, Mattisi I, Grossi P, Favaretto A, et al. Cortical lesions and atrophy associated with cognitive impairment in relapsing-remitting multiple sclerosis. Arch Neurol 2009;66:1144–50. [20] Nelson F, Datta S, Garcia N, Rozario NL, Perez F, Cutter G, et al. Intracortical lesions by 3 T magnetic resonance imaging and correlation with cognitive impairment in multiple sclerosis. Mult Scler 2011;17:1122–9. [21] Howell OW, Reeves CA, Nicholas R, Carassiti D, Radotra B, Gentleman SM, et al. Meningeal inflammation is widespread and linked to cortical pathology in multiple sclerosis. Brain 2011;134:2755–71. [22] Bermel RA, Puli SR, Rudick RA, Weinstock-Guttman B, Fisher E, Munschauer FE, et al. Prediction of longitudinal brain atrophy in multiple sclerosis by gray matter magnetic resonance imaging T2 hypointensity. Arch Neurol 2005;62:1371–6. [23] Horsfield MA, Sala S, Neema M, Absinta M, Bakshi A, Sormani MP, et al. Rapid semiautomatic segmentation of the spinal cord from magnetic resonance images: application in multiple sclerosis. Neuroimage 2010;50:446–55. [24] Weiss S, Mori F, Rossi S, Centonze D. Disability in multiple sclerosis: when synaptic long-term potentiation fails. Neurosci Biobehav Rev 2014;43C:88–99. [25] Llufriu S, Martinez-Heras E, Fortea J, Blanco Y, Berenguer J, Gabilondo I, et al. Cognitive functions in multiple sclerosis: impact of gray matter integrity. Mult Scler 2014;20:424–32.
5
[26] Zivadinov R, Leist TP. Clinical-magnetic resonance imaging correlations in multiple sclerosis. J Neuroimaging 2005;15(4 Suppl):10S–21S. [27] Amato MP, Bartolozzi ML, Zipoli V, Portaccio E, Mortilla M, Guidi L, et al. Neocortical volume decrease in relapsing-remitting MS patients with mild cognitive impairment. Neurology 2004;63:89–93. [28] Bakshi R, Thompson AJ, Rocca MA, Pelletier D, Dousset V, Barkhof F, et al. MRI in multiple sclerosis: current status and future prospects. Lancet Neurol 2008;7: 615–25. [29] Eikelenboom MJ, Petzold A, Lazeron RH, Silber E, Sharief M, Thompson EJ, et al. Multiple sclerosis: neurofilament light chain antibodies are correlated to cerebral atrophy. Neurology 2003;60:219–23. [30] Rinaldi L, Gallo P, Calabrese M, Ranzato F, Luise D, Colavito D, et al. Longitudinal analysis of immune cell phenotypes in early stage multiple sclerosis: distinctive patterns characterize MRI-active patients. Brain 2006;129:1993–2007. [31] Zivadinov R, Uxa L, Bratina A, Bosco A, Srinivasaraghavan B, Minagar A, et al. HLADRB1*1501, -DQB1*0301, -DQB1*0302, -DQB1*0602, and -DQB1*0603 alleles are associated with more severe disease outcome on MRI in patients with multiple sclerosis. Int Rev Neurobiol 2007;79:521–35. [32] Greer JM, Csurhes PA, Muller DM, Pender MP. Correlation of blood T cell and antibody reactivity to myelin proteins with HLA type and lesion localization in multiple sclerosis. J Immunol 2008;180:6402–10. [33] Dinacci D, Tessitore A, Russo A, De Bonis ML, Lavorgna L, Picconi O, et al. BDNF Val66Met polymorphism and brain volumes in multiple sclerosis. Neurol Sci 2011; 32:117–23. [34] Magraner MJ, Bosca I, Simó-Castelló M, García-Martí G, Alberich-Bayarri A, Coret F, et al. Brain atrophy and lesion load are related to CSF lipid-specific IgM oligoclonal bands in clinically isolated syndromes. Neuroradiology 2012;54:5–12. [35] DeLuca GC, Alterman R, Martin JL, Mittal A, Blundell S, Bird S. Casting light on multiple sclerosis heterogeneity: the role of HLA-DRB1 on spinal cord pathology. Brain 2013;136:1025–34.
Please cite this article as: Tauhid S, et al, MRI phenotypes based on cerebral lesions and atrophy in patients with multiple sclerosis, J Neurol Sci (2014), http://dx.doi.org/10.1016/j.jns.2014.08.047
