Cortical pathology in multiple sclerosis patients with epilepsy: a 3 year longitudinal study
- M Calabrese1,
- P Grossi1,
- A Favaretto1,
- C Romualdi2,
- M Atzori1,
- F Rinaldi1,
- P Perini1,
- M Saladini1,
- P Gallo1
- 1The Multiple Sclerosis Centre of Veneto Region, First Neurology Clinic, Department of Neurosciences, University Hospital of Padova, Padova, Italy
- 2CRIBI, Biotechnology Centre, Department of Biology, University of Padova, Padova, Italy
- Correspondence to Dr M Calabrese, Multiple Sclerosis Centre, First Neurology Clinic, University of Padova, Via Giustiniani 5, Padova 35128, Italy;
Contributors MC and PG: conception and design, analysis and interpretation of the data, drafting the article and revising it critically for important intellectual content and final approval of the version to be published. AF, MA, FR, PP, MS: analysis and interpretation of the data, revising the article critically for important intellectual content and final approval of the version to be published. PG: analysis and interpretation of the cognitive data, revising the article critically for important intellectual content and final approval of the version to be published. CR: statistical analysis and interpretation of the data, drafting the article and revising it critically for important intellectual content and final approval of the version to be published. MC had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
- Received 28 April 2011
- Revised 8 July 2011
- Accepted 13 July 2011
- Published Online First 2 September 2011
Introduction The cause of epilepsy in multiple sclerosis (MS) has not yet been elucidated. The relevance of cortical pathology (cortical lesions and thickness) in MS patients with and without epilepsy was evaluated in a longitudinal study.
Methods 32 relapsing–remitting MS patients with epilepsy (RRMS/E) and 60 matched RRMS patients without epilepsy were included in a 3 year longitudinal study. The following clinical and MR parameters were analysed: Expanded Disability Status Scale (EDSS), cognitive score (CS), cortical lesion (CL) number and volume, grey matter fraction (GMf), global cortical thickness (CTh), T2 white matter lesion volume (T2WMLV), new CLs and new WM lesions.
Results At baseline (T0), CLs were observed in 27/32 (84.4%) RRMS/E and in 26/60 (43.3%) RRMS (p<0.001) patients, and the RRMS/E group had a higher number (10.2±8.9 vs 4.5±2.4; p<0.001) and total volume (2.0±1.3 vs 0.7±0.8 cm3; p<0.001) of CLs compared with the RRMS group. No significant difference in T2WMLV was observed. Global CTh was lower in RRMS/E (2.12±0.19 vs 2.35±0.14 mm; p<0.001), and this group also showed a decline in cognition (CS 10.9±6.3 vs 6.2±3.5; p<0.001). After 3 years (T1), the RRMS/E group had a higher accumulation of new CLs (3.4±3.2 vs 1.2±1.1; p<0.001) and faster reduction of GMf (p=0.022) while the two groups did not differ in the number of new WM and new Gad+ lesions.
Discussion RRMS/E had a more severe and rapidly evolving cortical pathology (CLs and atrophy) compared with RRMS without epilepsy. The RRMS/E group was also characterised by more pronounced cognitive decline, higher EDSS and higher prevalence of men.
Multiple sclerosis (MS) affects both the white matter (WM)1 and grey matter (GM)2 of the CNS. MS related GM pathology can be observed in vivo by means of non-conventional MR sequences that allow the identification of focal inflammatory lesions3–4 and a widespread and early GM atrophy.5–6 By means of double inversion recovery (DIR),3 cortical lesions (CLs) were demonstrated not only in the majority of patients with progressive MS7 but also in those with relapsing–remitting MS (RRMS), even at clinical disease onset,4 with a direct impact on physical8 and cognitive disability.9 Epilepsy occurs more frequently in MS (mean prevalence value 2.3%, range 0.5–8.0%)10–12 than in the general population (range 0.27–1.7%),13 suggesting that the association between epilepsy and MS is not a mere coincidence,14 as recently confirmed in a large MS population.15 Moreover, MS may present with epilepsy and, in some cases, seizures may be the only manifestation of an MS relapse.14
The pathophysiology of epileptic seizures in MS remains a matter of debate. In a previous cross sectional study, we observed that RRMS patients with epilepsy (RRMS/E) had a significant increase in the number and volume of CLs compared with age and sex matched RRMS patients without epilepsy.14
On the basis of these preliminary data, we designed a longitudinal study aimed at analysing the evolution of the clinical and the MR parameters of RRMS with or without epilepsy, with special attention to cortical pathology.
Thirty-two consecutive patients with RRMS (RRMS/E) who presented, during the course of the disease, with epileptic seizures that could not be explained by any cause other than MS, were enrolled between 1 April 2007 and 31 December 2007, into a pre-planned 3 year follow-up study. Twenty of these patients were included in a preliminary cross sectional study previously published.14 Table 1 reports the demographic and clinical characteristics of these patients. All patients had at least two clinically documented (ie, clinical diary, emergency room reports, general practitioner communications) epileptic seizures. In four patients (12.5%), epilepsy was retrospectively considered as the initial clinical manifestation of MS. For the remaining 28 patients, the first epileptic seizure occurred a mean of 7.1 years and a median of 8.0 years after MS onset. Eleven patients had partial seizures which consisted essentially of motor seizures and 21 had generalised tonic–clonic seizures (among these, however, 12 were retrospectively considered as secondary generalised).
As a control MS population, we randomly selected a group of 60 RRMS patients with no history of epilepsy and matched to the RRMS/E group with regard to age and disease duration (table 1).
At study entry, all RRMS/E and RRMS patients were receiving immunomodulatory therapy: 19 RRMS/E and 27 RRMS received subcutaneous interferon β-1a 44 μg three times weekly; five RRMS/E and 24 RRMS received intramuscular interferon β-1a 30 μg once weekly; and eight RRMS/E and nine RRMS received subcutaneous glatiramer acetate 20 μg once daily. Thirty RRMS/E were treated with antiepileptic drugs (22 with monotherapy and eight with polytherapy) while two did not receive antiepileptic treatment since the last seizure had occurred several years before (>3 years) and therapy was discontinued.
AT T0 and after 3 years (T1) (median 36 months; range 36–38 months), all of the patients were assessed using the Expanded Disability Status Scale (EDSS)16 and underwent MR examination, as below described. At T1, a patient was considered clinically worsened if an EDSS score of ≥1.0 point was observed and confirmed by a second visit after 6 months. All MR examinations were performed at least 1 month after the last relapse and/or the last high dose steroid course. The study was approved by the local ethics committee and informed consent was obtained from all patients.
At baseline (T0), EEG registered far-off epileptic fits were available for each patient. The EEG was normal in nine (28.1%) patients whereas 23 (71.9%) showed focal changes consistent with epilepsy: slow activity background was present in seven (21.2%) patients, focal spikes in 15 (46.9%) and focal slow waves in 13 (40.7%) patients.
At T0 and T1, a neuropsychological assessment was performed, within the same month as the MR examination, by a neuropsychologist blinded to the MR results. The Rao's Brief Repeatable Battery of Neuropsychological Test (BRBNT), version A (at T0) and version B (at T1) were used. As previously described,17 a patient failed a test if the score in the test was below 2 SDs of the normal control mean value; therefore, a cognitive score (CS) was elaborated17 and the difference in CS between T1 and T0 (Δ-CS) was calculated (table 2).
All images were acquired using a 1.5 T machine (Achieva, Philips Medical Systems, Best, The Netherlands) with a 16 channel head coil. The following sequences were obtained from all subjects.
DIR. Repetition time (TR)=15631 ms, echo time (TE)=25 ms, inversion time (TI)=3400 ms, delay =325 ms, echo train length (ETL)=17, 50 contiguous axial slices with a thickness=3 mm, a matrix size=130×256 and FOV=250×200 mm.
Fast fluid attenuated inversion recovery (FLAIR). TR=10000 ms, TE=120 ms, TI=2500 ms, ETL=23, 50 contiguous axial slices with a thickness=3.0 mm, a matrix size =172×288 and FOV=250×200 mm.
Three-dimensional fast field echo (FFE) sequence. TR=25 ms, TE=4.6 ms, 120 contiguous axial slices with slice thickness=1.2 mm, flip angle=30°, a matrix size=256×256 and FOV=250×250 mm2;
Post-contrast T1-weighted spin echo. TR=618 ms, TE=10 ms, 20 slices with a thickness=5.5 mm, flip angle 90°, a matrix size=224×256 and a FOV=230×230 mm. This sequence was acquired 5 min after gadolinium-EDTA (0.1 mmol/kg) intravenous administration.
Subjects were carefully repositioned according to published guidelines for serial MRI studies of MS.18
Cortical lesion number and volume
All images were assessed by consensus by two experienced observers who were blinded to patient identity. At baseline and at the 3 year follow-up, the number of CLs and number of new CLs were assessed on DIR images (figure 1) following the recent recommendations for CLs scoring in patients with MS.19 CL volume was calculated using a semiautomatic thresholding technique based on Fuzzy C mean algorithm,20 21 included in a software developed at the National Institutes of Health, Medical Images Processing, Analysis and Visualisation (MIPAV) (http://mipav.cit.nih.gov).
WM lesion number and volume
The same procedure was applied to FLAIR images, to identify and segment WM lesions, thus obtaining a T2 hyperintense WM lesion volume (T2WMLV) at baseline and follow-up. Changes in T2WMLV (Δ-T2WMLV) were calculated as (T2WMLV follow-up−T2WMLV baseline)/T2WMLV baseline. The number of contrast enhancing lesions (CELs) was evaluated only at baseline.
GM fraction evaluation
The three-dimensional FFE images (T0 and T1) were first registered using FMRIB's Linear Image Registration Tool (FLIRT, ie, a part of FSL software, http://www.fmrib.ox.ac.uk/fsl/flirt/) and then segmented into WM, GM and CSF using SPM8, as previously described.22 23 Lesion masks were obtained by contouring lesions on the three-dimensional FFE scans; then, these masks were subtracted from WM, GM and CSF maps to obtain four mutually exclusive tissue masks with their volumes in millilitres. Total intracranial volume was calculated as the sum of WM+GM+CSF+lesion mask volumes. GM fraction (GMf) was calculated as GM volume/total intracranial volume at T0 and T1. The annualised percentage of GM volume change between T0 and T1 (Δ-GMf) was derived as follows: (GMf-T0−GMf-T1/GMf-T0)/number of years.
Cortical thickness evaluation
Global and regional CTh (mean of right and left hemispheres) was performed at study entry on three anatomical FFE images by means of the Freesurfer image analysis suite, available online (http://surfer.nmr.mgh.harvard.edu/), as described elsewhere.17 All images were accurately controlled for errors/artefacts by an experienced neurologist (MC). A first visual inspection was necessary after the skull stripping to visualise areas not completely removed while a second visual inspection was needed after WM/GM segmentation, to visualise possible misclassifications (especially due to MS lesions). In one patient, a manual and automated correction of topological defects was required.
Differences between the RRMS/E and RRMS groups were assessed using analysis of variance (ANOVA) followed by pairwise post hoc comparison using Tukey's HSD procedure to account for multiple comparisons. Therefore, a t test with Bonferroni's correction was applied on regional CTh variables separately to test differences between patient groups. Pearson χ2 was applied to test the difference between the two groups in terms of percentages of female/male, patients with CLs and CELs at study entry and patients with new CLs and new WM lesions at follow-up.
A general linear model was applied to evaluate the relation of all MRI parameters (T2WMLV, CEL number, CLs number, CLs volume, new WM lesion, new CLs, global CTh, GMf and Δ-GMf) and age, considered as independent variables, and the presence/absence of epilepsy (dependent variable=group). An automatic model selection procedure (stepwise regression) was adopted to choose the best variables combination. The prediction capability of the estimated model (goodness of fit) is given by the error rate of classifying a patient as RRMS/E and RRMS.
The EDSS change was adjusted for the baseline EDSS.
Data were considered significant at the 0.05 level and were analysed using SPSS V.18 for Windows.
Prevalence of epilepsy in MS and gender ratio
At study initiation, among a cohort of 1500 MS patients followed at the MS Centre in Padova, 38 patients with epilepsy were identified; thus the prevalence of epilepsy in our MS population was 2.5%. Interestingly, while the female/male ratio was 2.3 in our general RRMS population, the female/male ratio in RRMS/E was 0.9 (see table 1).
The RRMS/E group had a higher CL number and volume than the RRMS group (p<0.001 for both comparisons). GMf and global CTh were decreased in RRMS/E compared with RRMS (p=0.004 and p<0.001). Four CLs, all in the RRMS/E group, were enhanced by gadolinium but no difference was observed in terms of CEL (p=0.098) or T2WMLV (p=0.071) between the two groups (more details are available in table 1). No significant difference in cortical pathology (both CL number/volume, global CTh and GMf) was observed between RRMS/E patients with normal or abnormal EEG, or between RRMS/E patients receiving polytherapy or monotherapy (data not shown).
Nine of 60 (15.0%) RRMS patients failed in only one test while 5/60 (8.3%) failed in two or more tests of the BRBNT. In contrast, 7/32 (21.9%) RRMS/E patients failed in only one test (p=0.40) and 10 (31.3%) in two or more tests (p=0.004). CS was 10.9±6.3 (range 1–24) in RRMS/E and 6.2±3.5 (range 0–15.0) in RRMS, and the difference was highly significant (p<0.001).
Difference in CTh between RRMS and RRMS/E
Global CTh was lower in RRMS/E than in RRMS (p<0.001, see table 1). After Bonferroni correction, areas of significant cortical thinning were observed in the frontal and temporal brain regions of both hemispheres, with no difference between the left and right sides. No significant difference in CTh between groups was found in any of the occipital or parietal cortical areas analysed, with the exception of both the left and right postcentral gyrus (p=0.001 for both areas) (detailed data on regional analysis are reported in table 3).
Twenty-eight (87.5%) RRMS/E and 45 (76.0%) RRMS patients experienced one or more clinical relapses (p=0.15) and all were treated with high dose steroids (1 g daily for 5 days). Among these, 18/28 (64.2%) RRMS/E and 12/45 (26.7%) RRMS patients showed EDSS worsening (p=0.001). Twenty-three (71.9%) RRMS/E patients had one or more epileptic seizures (mean 1.7±0.8, range 1–5); 16 of these had a partial seizure whereas two had a complex partial seizure and five had a generalised seizure. Secondary generalisation occurred in eight out of 16 patients with partial seizures. Among the 23 patients with epileptic seizures, 19 also had an MS relapse but no temporal correlation was observed between relapses and epileptic seizures (mean time between relapse and epileptic seizure 186±62 days, range 42–394 days). No epileptic seizures were observed in RRMS patients.
At T1, RRMS/E patients showed a higher accumulation of new CLs (p<0.001) than RRMS patients, and patients with new CLs were almost double in RRMS/E (84.4%) than in RRMS (43.3%, p<0.001) (table 2). Δ-GMf was higher in RRMS/E than in RRMS (p=0.002) while no significant difference was found in the accumulation of WM lesions (new WM lesions or Δ-T2WMLV) between the two groups (see table 2).
Eleven out of 60 (18.3%) RRMS patients failed in only one test and 7/60 (11.6%) failed in two or more tests of the BRBNT. Ten of 32 (31.2%) RRMS/E patients failed in only one test (p=0.15) and 14/32 (43.7%) in two or more tests (p<0.001). Δ-CS (table 2) in RRMS/E (4.0±3.7, range 0–10) was higher than in RRMS (1.3±2.0, range 0–6; p=0.002).
Stepwise regression procedure revealed a significant contribution from new CL (p<0.001), Δ-GMf (p<0.001) and CL volume (p<0.001) as independent predictors of the presence of epilepsy, while all other variables, including age, were removed. The estimated model correctly identified as epileptic 29/32 of the RRMS/E and as not epileptic 57/60 of the RRMS patients. The prediction capability of such a model (goodness of fit) can be recovered by the error rate of classifying patients (6.5%).
The results of our longitudinal study strongly confirm a previous preliminary observation about the association between the severity of cortical pathology—namely, number and volume of CLs (figures 1 and 2)—and the occurrence of epilepsy in patients with MS.14 Analysis of the so-called NAGM (including the evaluation of GM volume and cortical thickness) further reveals that, even at study entry, significant cortical atrophy characterises RRMS/E compared with RRMS patients. Moreover, the regional analysis discloses a significant cortical thinning in the frontotemporal areas, already identified as areas of marked atrophy in MS24 and sites of accumulation of CLs.25 Atrophy of these areas has recently been found to be associated with early cognitive dysfunction in MS.17 Indeed, we would like to point out that the CS of the RRMS/E group was about two times higher than that of the RRMS group, and that the percentage of RRMS/E affected by significant cognitive dysfunction (failure in at least two cognitive tests) was four times higher than in the RRMS group.
Follow-up of RRMS/E patients reveals a higher progression rate of cortical pathology in these patients. Indeed, both Δ-GMf and the accumulation of new CLs were significantly higher in RRMS/E than in RRMS, whereas no significant difference was observed in the appearance of new WM lesions. These findings confirm previous studies suggesting that WM and GM pathology may proceed, at least partly, independently.26–28
Another interesting finding of our study is the faster cognitive decline observed in RRMS/E compared with RRMS patients, despite the short follow-up period (3 years) and the intrinsic limitations of BRBNT. This result further underlines the close relationship between cognitive impairment and cortical damage in MS, as observed in previous studies.9 17
Taken together, the results of our longitudinal study indicate that RRMS/E constitute a subgroup of MS patients characterised by a more pronounced and selective involvement of the GM that determines a severe (cortical) clinical picture. Whether epilepsy and cognitive impairment in these patients are the consequence of a more severe cortical pathology, as suggested by our findings, rather than an accidental coincidence,15 needs to be investigated further.
An intriguing observation of our study is the unexpected high frequency of males in the RRMS/E group (female/male ratio 0.9) compared with the RRMS group (female/male ratio 2.3), a finding observed in our previous study14 that does not agree with the available literature.15 This discrepancy might be due to the small number of RRMS/E patients included in our study. However, we would like to point out that men suffering from MS tend to develop a more severe cortical pathology than women.4 The occurrence of epilepsy several years after disease onset, therefore, might be explained by the increase in CL number and volume. The lack of any difference in cortical pathology (both CL number/volume and GMf/global CTh) observed between RRMS/E patients with normal or abnormal EEG, as well as between RRMS/E receiving polytherapy or monotherapy, could be explained by the low number of patients studied.
In summary, epilepsy in RRMS patients seems to be the consequence of a particularly severe and rapidly evolving cortical pathology (CLs and atrophy). These patients are also characterised by more pronounced cognitive impairment and higher physical disability. Thus they constitute a peculiar subgroup of MS patients whose cortical pathology merits further investigation.
Competing interests None.
Ethics approval The study was approved by the local ethics committees (Ethic Committee of Azienda Ospedaliera Of Padova).
Provenance and peer review Not commissioned; externally peer reviewed.