FULL-LENGTH ORIGINAL RESEARCH Early and chronic gray matter volume changes in limbic encephalitis revealed by voxel-based morphometry *†Jan Wagner, *†‡Bernd Weber, and *†‡Christian E. Elger Epilepsia, **(*):1–8, 2015 doi: 10.1111/epi.12968 SUMMARY Dr. Jan Wagner is resident, with special interest in imaging, at the Department of Epileptology at the University of Bonn. Objective: Antibody-associated limbic encephalitis (LE) is an increasingly recognized cause of mostly adult-onset temporal lobe epilepsy. Magnetic resonance imaging (MRI) typically shows volume and signal changes of the mesiotemporal structures. However, recent studies indicate that imaging characteristics depend on the type of the associated antibody. The aim of the present study was to investigate early and chronic gray matter (GM) volume changes in LE by means of voxel-based morphometry (VBM). Methods: Optimized VBM analysis was applied to altogether 73 MRI volumes of 55 patients with antibody-associated LE. Based on the time point of MRI acquisition, patients were split into two separate groups to enable the evaluation of early (≤2 years after LE onset) and chronic (>2 years after LE onset) GM volume changes. In addition, separate analyses for the two most common LE subtypes in our study cohort, that is, glutamic acid decarboxylase (GAD)–associated LE and voltage-gated potassium channel (VGKC)-complex–associated LE were performed. Age- and gender-matched healthy subjects served as controls. Results: Referring to the entire LE group, VBM revealed bi-amygdalar GM volume increase in the early disease stage. In the chronic disease stage, amygdala enlargement had resolved and we found GM volume reduction in the right cerebellar hemisphere. In the subgroup analysis, VBM showed corresponding bi-amygdalar GM volume increase in VGKC-complex–associated LE on early MRI, whereas no changes were found in GAD-associated LE. In the chronic disease stage, VBM revealed left frontal GM volume increase in VGKC-complex–associated LE and right frontoparietal GM volume reduction in GAD-associated LE. Significance: The present study provides further evidence of a predominant affection of the amygdala in the early disease stage of LE, which resolves during the later course of the disease. Furthermore, our results show that LE features distinct imaging characteristics depending on the associated antibody and thus may contribute to a better pathophysiologic understanding of this disease. KEY WORDS: Epilepsy, Magnetic resonance imaging, Voxel-based morphometry, Glutamic acid decarboxylase, Voltage-gated potassium channel-complex, Onconeural. Antibody-associated limbic encephalitis (LE) has come up over the past years as an underlying cause of formerly mostly cryptogenic temporal lobe epilepsy.1 Increasingly more autoantibodies are found to be associated with this disorder,2,3 which was described initially as a paraneoplastic syndrome caused by inflammation in limbic structures in adults.4,5 While prognosis in paraneoplastic LE depends mainly on the underlying tumor, clinical outcome in nonparaneoplastic LE seems to be crucially influenced by Accepted February 11, 2015. *Department of Epileptology, University of Bonn, Bonn, Germany; †Department of NeuroCognition/Imaging, Life & Brain Center, Bonn, Germany; and ‡Center for Economics and Neuroscience, University of Bonn, Bonn, Germany Address correspondence to Jan Wagner, Department of Epileptology, University of Bonn, Sigmund-Freud-Str. 25, D-53127 Bonn, Germany. E-mail: jan.wagner@ukb.uni-bonn.de Wiley Periodicals, Inc. © 2015 International League Against Epilepsy 1 2 J. Wagner et al. the associated antibody. Here, LE associated with antibodies against glutamic acid decarboxylase (GAD) usually shows a nonremitting chronic disease course with seizure and antibody persistence and poor responses to immunotherapy, whereas patients who have LE associated with voltage-gated potassium channel (VGKC)-complex antibodies mostly become seizure-free and antibody-negative during follow-up.3,6–8 Typical features of LE on magnetic resonance imaging (MRI) comprise volume and signal changes of the mesiotemporal structures, which have been demonstrated by both conventional9,10 and postprocessing imaging studies.11 In two recently published studies, we could show that the amygdala seems to be primarily affected by the inflammatory process based on automated MRI signal12 and volumetric analyses.6 Furthermore, in the latter study, we could show that distinct volumetric courses of amygdala and hippocampus in the acute disease stage of GAD-associated LE (GAD-LE) and VGKC-complex–associated LE (VGKCLE) corresponded to distinct clinical and paraclinical features in these two LE subforms. Based on the results of these two studies, the aim of the present study was to evaluate gray matter (GM) volume changes in antibody-associated LE by means of voxel-based morphometry (VBM). We chose VBM, as this is a well-established fully automated processing technique facilitating the detection of GM volume changes in the entire brain, whereas our previous studies were focused mainly on abnormalities of the mesiotemporal structures. The feasibility of VBM in temporal lobe epilepsy has been proven in various studies on hippocampal sclerosis by showing widespread volume reductions even remote from the seizure focus.13 To the best of our knowledge, no studies have applied this technique to LE until now. Because we were particularly interested in early and chronic GM volume changes, two separate study groups were established, depending on the time point of their MRI acquisition. In addition, we performed separate analyses for the two most common LE subtypes in our study cohort, that is, GAD-LE and VGKC-LE. Methods Study groups We retrospectively evaluated all patients diagnosed with antibody-associated LE presenting at the University of Bonn Department of Epileptology, from April 2006 to August 2013. Patients were diagnosed with LE based on the features of a subacute “limbic” syndrome manifesting in adolescence or adulthood (at least one of the following symptoms: seizures of temporal semiology, disturbance of episodic memory, psychiatric symptoms with affective and/ or anxiety disturbances), and positive serologic antibody results (i.e., onconeural, VGKC-complex, GAD). The patient groups in this study were based on the study cohort from our previous report.6 However, as our previous Epilepsia, **(*):1–8, 2015 doi: 10.1111/epi.12968 work focused mainly on the early disease stage, we additionally aimed at investigating chronic changes in GM volume in the present study. Therefore, the following two separate patient groups were established, depending on the time point of their MRI acquisition: (1) the “early” group, consisting of the earliest available MRI of each included patient acquired not later than 2 years after disease onset (in the following referred to as MRI 1); and (2) the “late” group, consisting of the most recent available MRI performed at least 2 years after disease onset (in the following referred to as MRI 2). Hence, patients who were scanned repeatedly could be included in both MRI 1 and MRI 2 groups if their earliest MRI was performed within the first 2 years after onset and if their most recent MRI was acquired at least 2 years after onset. LE onset was defined as the time point of the first symptoms suggestive of LE (seizures and/or psychiatric and/or mnestic disturbances). As mentioned earlier, these selection criteria were chosen to evaluate early and chronic changes in GM volume and were based on experiences from our previous study,6 which focused mainly on the first 2 years after LE onset. Furthermore, the cutoff value of 2 years enabled similar MRI 1 and MRI 2 group sizes. To achieve the best possible age-matching and gendermatching with the patient groups, two separate control groups consisting of healthy subjects with no neurologic or psychiatric disorder were assembled. These controls were recruited from a preexisting in-house database consisting of 1,342 healthy subjects. Age-matching and gender-matching was achieved by building matched pairs for each individual patient. Informed consent was obtained from all study participants and the study was approved by the ethics committee of the University of Bonn. Antibody tests All patients underwent extensive serum antibody testing. Identification of GAD antibodies in serum was performed by radioimmunoprecipitation assay using 125IGAD (normal values ≤1 U/ml; Weatherall Institute, Oxford, United Kingdom, or EUROIMMUN, L€ ubeck, Germany).8 Serum antibodies against the VGKC-complex were assessed by radioimmunoprecipitation assay (normal values <100 pM; Weatherall Institute or EUROIMMUN).14 Antibodies against leucine-rich, glioma inactivated 1 protein (LGI1) and contactin-associated protein 2 (CASPR2) were detected by indirect immunofluorescence using formalin-fixed human embryonic kidney (HEK293) cells containing membrane bound LGI1 (normal values <1:10; EUROIMMUN) or CASPR2 (normal values <1:10; EUROIMMUN).15 The latter two tests were performed from 2010. “Well characterized” onconeural antibodies were tested with use of a commercially available routine test using an immune-dot-blot for Hu, Ma, amphiphysin, and CV2/CRMP5 antibodies (Ravo Diagnostika, Freiburg, Germany). 3 VBM in Limbic Encephalitis MRI examinations All included patients underwent routine clinical MRI examinations for the neuroradiologic assessment using a Philips 3 Tesla MRI scanner (Intera; Philips Medical Systems, Amsterdam, The Netherlands) according to a standard protocol.16 The T1-weighted volume datasets of patients and healthy controls that were used for the VBM analysis were acquired independently from the clinical scans at the Life & Brain Center in Bonn using a 3 Tesla scanner (Magnetom Trio; Siemens, Erlangen, Germany). Sequence parameters were as follows: magnetization-prepared rapid acquisition gradient echo (MPRAGE), voxel size 1 9 1 9 1 mm3, repetition time 1,570 msec, echo time 3.42 msec, flip angle 15 degrees, field of view 256 9 256 mm2. All patients and healthy controls were scanned using the same MRI scanner with the same sequence parameters. VBM analysis VBM analysis was performed using Functional MRI of the Brain Software Library (FSL)-VBM, an optimized VBM protocol carried out with FSL tools (Version 5.0; Department of Clinical Neurology, University of Oxford, Oxford, United Kingdom; http://www.fmrib.ox.ac.uk/fsl).17 First, structural images were brain-extracted and GM-segmented before being registered to the Montreal Neurological Institute (MNI) 152 standard space using nonlinear registration. The resulting images were averaged and flipped along the x-axis to create a left-right symmetric, study-specific GM template. Second, all native GM images were nonlinearly registered to this study-specific template and “modulated” to correct for local expansion (or contraction) due to the nonlinear component of the spatial transformation. The modulated GM images were then smoothed with an isotropic Gaussian kernel with a sigma of 2 mm (full width at half maximum ~4.6 mm). Finally, voxelwise statistical analysis was applied using permutation-based nonparametric testing (5,000 permutations) with threshold-free cluster enhancement (TFCE) correcting for multiple comparisons (p < 0.05, family-wise error [FWE] correction). Because this is a rather conservative approach, we also report results using a less stringent, uncorrected threshold with p < 0.001 for the antibody subgroup analyses due to the smaller group sizes. Cross-sectional analyses were performed by comparing patient groups with the corresponding controls. Because these groups were age-matched and gender-matched, we did not include additional covariates of no interest in the general linear model (GLM). Eighteen patients were scanned repeatedly with the appropriate intervals between MRI 1 and MRI 2 (see Results section); therefore, we additionally performed longitudinal VBM analyses in these 18 cases using a singlegroup paired-difference t-test. For details on GLM setup for this analysis see http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM# Single-Group_Paired_Difference_.28Paired_T-Test.29. The time interval between MRI 1 and MRI 2 was added as a covariate of no interest in the GLM to correct for this factor because time intervals differed considerably between these 18 patients (median 2.7 years, min 1.0 years, max 4.8 years, interquartile range [IQR] 1.6 years). Furthermore, VBM correlation analyses between GM volume and disease duration at MRI were performed using the MRI scans of all included patients (irrespective of whether they belonged to MRI 1 or MRI 2 group). Age at MRI was added as covariate of no interest in the GLM to account for age-related GM volume changes. Statistical analysis Statistical analyses of clinical data were performed using SPSS Statistics 21.0 for Mac OS X (IBM, Armonk, NY, U.S.A.). All values throughout this report are given as median unless otherwise stated. For statistical comparisons of independent categorical data, a Fisher’s exact test was performed and for comparisons of independent metrical data, a Mann-Whitney U test was performed. A p-value < 0.05 was regarded as statistically significant using two-tailed tests. Results Entire LE group Clinical data of the patient groups and the corresponding control groups are summarized in Table 1. Eighteen patients were scanned repeatedly with the appropriate intervals between LE onset and MRI, and thus were included in both MRI 1 and MRI 2 group, amounting to altogether 73 MRI studies from 55 patients that were available for further VBM processing. The proportion of patients receiving immunotherapy at the time point of MRI acquisition was significantly higher at MRI 1 compared to MRI 2 (p = 0.001). Tumor searches were performed in all included patients and were positive in one patient of the MRI 1 group (lung adenocarcinoma associated with Hu antibodies) and two patients of the MRI 2 group (small cell lung carcinoma associated with Hu antibodies and acinic cell carcinoma of the cervical lymph nodes associated with Ma antibodies). Cross-sectional VBM analysis of MRI 1 revealed a significant GM volume enlargement in LE relative to the corresponding controls localized in both amygdalae (p < 0.05, FWE-corrected; Fig. 1A). No areas with significant GM volume reduction were found at MRI 1. At MRI 2, GM volume enlargement was no longer present in the amygdala and we found GM volume loss in the right cerebellar hemisphere (Fig. 1B). Longitudinal VBM analysis of repeatedly scanned patients (N = 18) revealed regions with a significant GM volume loss between MRI 1 and MRI 2 including the mesiotemporal region (especially in both amygdalae), the basal ganglia (especially caudate nucleus), the thalamus, the parietooccipital cortex, and the cerebellum (p < 0.05, Epilepsia, **(*):1–8, 2015 doi: 10.1111/epi.12968 4 J. Wagner et al. Table 1. Clinical and serologic data of the two LE groups and the corresponding control groups Demographical data N (male) Age at MRI, years, median, IQR (range) Age at LE onset, years, median, IQR (range) Disease duration at MRI, years, median, IQR (range) Antibody GAD, N (%) VGKC-complex, N (%) LGI1, N (%) CASPR2, N (%) Onconeural, N (%) Seizures Per month, median, IQR (range) Seizure-free, N (%) Immunotherapy, N (%) LE MRI 1 CON MRI 1 LE MRI 2 CON MRI 2 36 (14) 48.5, 32.8 (17–73) 46.7, 33.0 (16–72) 0.5, 0.7 (0.0–2.0)a 36 (14) 48.0, 32.6 (16–74) NA NA 37 (17) 44.8, 27.0 (17–76) 43.0, 26.7 (12–72) 3.9, 2.2 (2.3–6.1) 37 (17) 43.0, 28.2 (17–72) NA NA 16/36 (44) 18 (50) 4 (11) 1 (3) 2 (6) NA NA NA NA NA 18/37 (49) 16 (43) 1 (3) 4 (11) 3 (8) NA NA NA NA NA 12.5, 31.7 (0–900) 8/36 (22) 31/36 (86)b NA NA NA 5.5, 10.0 (0–180) 11/37 (30) 18/37 (49) NA NA NA CON, control group; IQR, interquartile range; NA, not available. a p < 0.001 comparing LE MRI 1 with LE MRI 2. b p < 0.01 comparing LE MRI 1 with LE MRI 2. FWE-corrected; Fig. 1C). We did not find any regions with a significant GM volume increase between MRI 1 and MRI 2. VBM correlation analysis of GM volume with disease duration at MRI in all included patients (N = 73) did not reveal any regions with a significant positive or negative correlation when using an FWE-corrected threshold of p < 0.05. LE subgroup analysis Clinical data of the LE subgroups and the corresponding control groups are summarized in Table 2. The overlap of patients who were included in both the MRI 1 and MRI 2 groups amounted to eight GAD-LE cases and nine VGKCLE cases. We found highly significant differences concerning age at LE onset and age at MRI at both MRI 1 and MRI 2 between GAD-LE and VGKC-LE (all p < 0.001). Furthermore, significant differences between the two LE subgroups were found in seizure frequency (p < 0.001), the proportion of seizure-free patients (p < 0.001), and the proportion of antibody-negative patients at MRI 2 (p = 0.001). Based on a conservative approach using an FWE-corrected threshold of p < 0.05, no significant differences in the cross-sectional analyses between GAD-LE and VGKCLE and controls were found at MRI 1 and MRI 2. However, by using a less stringent uncorrected threshold with p < 0.001, we found a significant bi-amygdalar GM volume enlargement in VGKC-LE relative to the corresponding controls at MRI 1. No differences between GAD-LE and controls were found at this time point. At MRI 2, we found a circumscribed GM volume reduction in the right frontoparietal operculum in GAD-LE, whereas a small area of GM volume increase located in the left frontal operculum was present in VGKC-LE. Results of the cross-sectional analyses are illustrated in Figure 2A–C. Epilepsia, **(*):1–8, 2015 doi: 10.1111/epi.12968 In the longitudinal analyses, no significant changes were detected in either GAD-LE (N = 8) or VGKC-LE (N = 9), even when using an uncorrected threshold of p < 0.001, which is most probably caused by the relatively small patient groups. VBM correlation analysis revealed a bilateral cluster located in the amygdala with a negative correlation between GM volume and disease duration in the VGKC-LE subgroup (N = 34; p < 0.001 uncorrected; Fig. 2D), which is in line with the findings of the cross-sectional analyses. No significant correlations were found in the GAD-LE subgroup. Discussion To the best of our knowledge, this study represents the first application of VBM in antibody-associated LE. To assess early and chronic changes in GM volume, two separate study groups were established based on the time point of the MRI acquisition. In addition to cross-sectional analyses in which patient groups were compared with matched controls, we performed longitudinal VBM analyses in repeatedly scanned patients and VBM correlation analyses between GM volume and disease duration. Our cross-sectional results demonstrate bilateral GM volume increase in the amygdala in the early disease stage, whereas no regions with significant GM volume reduction were found at this time point. This volume increase was no longer detectable during the later course of the disease, and we found right cerebellar GM volume loss at MRI 2. In line with these cross-sectional results, a bilateral mesiotemporal GM volume loss located mainly in the amygdalae and a bilateral cerebellar GM volume loss could be detected in the longitudinal VBM analysis. However, we found additional regions with a longitudinal GM volume loss including the basal gan- 5 VBM in Limbic Encephalitis A B C Figure 1. (A) Cross-sectional VBM results of the entire LE group at MRI 1 using an FWE-corrected threshold of p < 0.05 demonstrating bi-amygdalar GM volume increase in the early disease stage. (B) At MRI 2, amygdala volume increase was no longer detectable and we found GM volume reduction in the right cerebellar hemisphere. (C) Longitudinal VBM results showing relatively widespread GM volume loss between MRI 1 and MRI 2 including the mesiotemporal region (especially in both amygdalae), basal ganglia (especially caudate nucleus), thalamus, parietooccipital cortex, and cerebellum. No regions with a GM volume increase between MRI 1 and MRI 2 were found. Red/orange, regions with significant volume increase; blue, regions with significant volume reduction. Epilepsia ILAE glia (especially caudate nucleus), the thalamus, and the parietooccipital cortex. Although a correction for the different time intervals between MRI 1 and MRI 2 was performed in this analysis, we suppose that this discrepancy to the cross-sectional results is at least partially caused by agerelated effects that may have interfered with disease-related changes. Supporting this hypothesis, significant physiologic age-related atrophy of the above-mentioned regions has been reported in previous volumetric and VBM studies.18,19 Taking both cross-sectional and longitudinal results into account, we could largely reproduce the results of our previous study6 using a different approach applied to a larger patient group. In particular, in our previous study, we found a bilateral amygdala enlargement in the initial disease stage of antibody-associated LE by means of automated mesiotemporal volumetry. Corresponding to our current results, amygdala volume also decreased during follow-up and did not differ from controls during the further course of the disease in our previous study. However, in contrast to our previous results, we did not find hippocampal volume reduction at MRI 2 using cross-sectional VBM despite a larger patient group in the current study. Hippocampal atrophy in the convalescent phase of LE has also been reported in a study comprising eight patients with LGI1 and VGKC-complex antibodies in comparison to healthy controls.11 That we did not find hippocampal atrophy at MRI 2 in the crosssectional analysis of the present study, could be a result of methodologic factors, as it is known that VBM features only a limited sensitivity in detecting subtle hippocampal atrophy.13,20 Cerebellar GM volume reduction at MRI 2 may reflect postinflammatory changes because cerebellar involvement has been described especially in GAD-associated3,21 and paraneoplastic neurologic disorders.3,22 Analog to our results, cerebellar GM volume reduction has also been described in various VBM studies on hippocampal sclerosis.13,23–30 Most commonly, this finding has been attributed to either chronic medication effects25 or excitotoxic damage30 due to neural pathways between the hippocampus and the cerebellum,31,32 both of which may be also attributable to our findings in LE. Apart from studies on LE, amygdala enlargement is increasingly recognized as a morphologic correlate of mesial temporal lobe epilepsy on MRI.33–40 In summary, all patients in these studies showed isolated and unilateral enlargement of the amygdala with or without accompanying signal hyperintensity. Of interest, in most cases, seizure onset was relatively late and seizure focus was usually ipsilateral to the enlarged amygdala. However, the etiology of amygdala enlargement remains unclear in most of these studies. Antibody testing, cerebrospinal fluid (CSF) investigations, and follow-up MRIs were not performed in most cases. Furthermore, histologic investigations in operated patients report inconsistent or unspecific results. Given the late seizure onset in most of these patients, an inflammatory etiology of the epilepsy may at least be possible in some of the reported cases. Amygdala enlargement seems to be a sensitive marker for LE but above-cited studies raise the question regarding the specificity of this finding on MRI. From the current point of view, the underlying cause of amygdala enlargement on MRI seems to comprise several etiologies including inflammation, tumor,38 dysplasia,38,39 and hypertrophic neurons.39,40 Bilateral pathology, additional hippocampal affection, and dynamic clinical and imaging course during follow-up (especially volume decrease) may be indicators for an inflammatory etiology, as dysplastic lesions or tumors are usually unilateral and do not shrink over time. Amygdala enlargement on MRI should prompt antibody testing in serum (and ideally also in CSF) to identify an inflammatory etiology as early as possible. Early initiation of immunosuppressive therapy may prevent irreversible damage to the mesiotemporal structures.11 In addition, antibody testing is of particular importance if Epilepsia, **(*):1–8, 2015 doi: 10.1111/epi.12968 6 J. Wagner et al. Table 2. Clinical and serologic data of the LE subgroups and the corresponding control groups GAD-LE MRI 1 Demographical data N (male) Age at MRI, years, median, IQR (range) Age at LE onset, years, median, IQR (range) Disease duration at MRI, years, median, IQR (range) Antibody Concentration, median, IQR (range) Antibody negative, N (%) Seizures Per month, median, IQR (range) Seizure-free, N (%) Immunotherapy, N (%) MRI 2 Demographical data N (male) Age at MRI, years, median, IQR (range) Age at LE onset, years, median, IQR (range) Disease duration at MRI, years, median, IQR (range) Antibody Concentration, median, IQR (range)b Antibody negative, N (%)c Seizures Per month, median, IQR (range) Seizure-free, N (%) Immunotherapy, N (%) VGKC-LE 16 (5) 32.5, 17.1 (17–58)a 31.4, 17.8 (16–58)a 0.4, 0.7 (0.1–2.0) All >1,000 U/ml 0/16 (0) 18 (9) 60.6, 18.9 (20–73) 60.4, 18.8 (19–72) 0.6, 0.7 (0.0–1.9) 621 pM, 536 (127–7,655) 0/18 (0) GAD-CON VGKC-CON 16 (5) 32.8, 16.2 (16–58) NA NA 18 (9) 60.7, 17.2 (20–74) NA NA NA NA NA NA 9.5, 40.0 (0–150) 5/16 (31) 12/16 (75) 23.0, 10.5 (0–900) 3/18 (17) 17/18 (94) NA NA NA NA NA NA 18 (7) 26.6, 19.8 (17–61)a 23.2, 20.8 (12–55)a 4.3, 2.4 (2.3–6.1) 16 (10) 52.9, 10.3 (22–76) 48.3, 11.1 (19–72) 3.9, 2.0 (2.5–6.1) 18 (7) 26.8, 16.8 (17–61) NA NA 16 (10) 53.5, 11.9 (21–72) NA NA NA NA NA NA NA NA NA NA NA NA All >1,000 U/ml 0/16 (0)d 451 pM, 312 (119–1,929) 8/14 (63) 7.5, 25.3 (0.5–150)a 0/18 (0)a 7/18 (39) 0.0, 1.3 (0–4) 10/16 (63) 8/16 (50) a p < 0.001 comparing GAD-LE with VGKC-LE. Antibody-negative patients excluded from this analysis. No antibody tests performed in two patients in each LE subgroup at MRI 2. d p < 0.01 comparing GAD-LE with VGKC-LE. b c epilepsy surgery is considered. A subsidiary affection of the contralateral mesiotemporal structures after surgery bears the risk of severe mnestic deficits and may furthermore lead to seizure recurrence. In the subgroup analyses, cross-sectional VBM revealed bi-amygdalar GM volume enlargement in VGKC-LE at MRI 1, but no significant changes in GAD-LE at this disease stage. Hence, VGKC-LE seems to show more prominent volume abnormalities in the acute disease stage, which additionally confirms the results of our previous study6 and has been found to mirror distinct clinical courses with a more severe initial clinical symptomatology regarding seizure, mnestic, and psychiatric disturbances in VGKC-LE compared to GAD-LE. At MRI 2, amygdala enlargement in VGKC-LE could no longer be detected and we found GM volume increase located in the left frontal opercular region, while no regions with GM volume reduction were present at this time point. Although frontal GM increase has also been reported in several VBM studies on temporal lobe epilepsy with and without hippocampal sclerosis,25,27,30 this finding is nevertheless slightly surprising given that Irani et al.11 found reduced global brain volumes in eight patients with LGI1 and VGKC-complex antibodies in the convalescent stage of the disease. Frontal GM increase in temporal lobe epilepsy has been attributed to diminished gray–white Epilepsia, **(*):1–8, 2015 doi: 10.1111/epi.12968 matter demarcation in previous studies25,30 and may also be the causative factor in VGKC-LE, although this remains speculative. In agreement with the cross-sectional results, bilateral clusters located in the amygdala with a negative correlation between GM volume and disease duration were detected in the VGKC-LE subgroup using VBM correlation analysis. These findings additionally confirm that the amygdala seems to be predominantly affected in LE, and volume changes are particularly pronounced in the VGKC-complexassociated subgroup. In GAD-LE, GM volume reduction located in the right frontoparietal operculum was found in the cross-sectional analysis at MRI 2, which is also a common finding in hippocampal sclerosis13,23,41,42 and has been attributed mostly to network damage. Furthermore, this finding may also result from ongoing seizure activity in GADLE. This would also explain the fact that no volume loss was present in VGKC-LE, as seizure burden differed significantly between these two groups at MRI 2. Supporting this hypothesis, a study on drug-resistant and drug-responsive temporal lobe epilepsy found more extensive GM atrophy in the group with ongoing seizures.42 Finally, antibody persistence in GAD-LE may represent another potential causative factor leading to chronic inflammation and consecutive GM atrophy in this group. In contrast to this, the majority of VGKC-LE patients became antibody negative at MRI 2. 7 VBM in Limbic Encephalitis specific subset of antibody-associated LE, as no patients with antibodies to the alpha-amino-3-hydroxy5-methyl-4-isoxazolepropionic acid (AMPA) receptor, c-aminobutyric acid receptor B (GABAB) and metabotropic glutamate receptor 5 were identified in our institution during the time of this study. A Conclusion B By means of VBM, our study provides further evidence of a predominant pathology of the amygdala in the early disease stage of antibody-associated LE and supports the results of our previous studies on LE. This bi-amygdalar volume enlargement seems to be caused primarily by VGKC-LE and resolves during the later course of the disease. In the chronic stage of LE, small areas of GM volume reduction were found when referring to the entire LE group and the GAD-LE subgroup, whereas a small area of GM volume increase located in the left frontal opercular region was found in VGKC-LE. C Acknowledgments D JW was supported by the Gerok Program of the BONFOR Commission, University of Bonn. BW was supported by the Deutsche Forschungsgemeinschaft (DFG) with a Heisenberg grant (BW: WE 4427/3-2). Disclosure Figure 2. VBM results of the LE subgroups using an uncorrected threshold of p < 0.001. (A) Cross-sectional analysis in GAD-LE revealed a small area of GM volume reduction located in the right frontoparietal operculum at MRI 2 (cluster size 42 continuous voxels = 336 mm3), whereas no changes were present at MRI 1. (B) Cross-sectional results in VGKC-LE showing GM volume increase in both amygdalae at MRI 1. (C) At MRI 2, amygdala enlargement was no longer detectable and we found GM volume increase located in the left frontal operculum. (D) VBM correlation analysis between GM volume and disease duration revealed two negatively correlated clusters located in both amygdalae, supporting the findings of the cross-sectional analyses. Red/orange, regions with significant volume increase; blue, regions with significant volume reduction. Epilepsia ILAE Limitations It should be noted that the results of the subgroup analyses are based on a p-value < 0.001 not corrected for multiple comparisons. Although this is a commonly used threshold in many VBM studies, the results should nevertheless be interpreted with relative caution in this regard. Furthermore, only a few onconeural cases were included in the study and longitudinal data were available in only 18 patients. This should also be taken into account when interpreting our findings. Finally, our results are limited to a None of the authors has any conflict of interest to disclose, which is relevant to this research activity. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. References 1. Bien CG, Urbach H, Schramm J, et al. Limbic encephalitis as a precipitating event in adult-onset temporal lobe epilepsy. Neurology 2007;69:1236–1244. 2. Lancaster E, Martinez-Hernandez E, Titulaer MJ, et al. Antibodies to metabotropic glutamate receptor 5 in the Ophelia syndrome. Neurology 2011;77:1698–1701. 3. Vincent A, Bien CG, Irani SR, et al. Autoantibodies associated with diseases of the CNS: new developments and future challenges. Lancet Neurol 2011;10:759–772. 4. Brierley JB, Corsellis JAN, Hierons R, et al. Subacute encephalitis of later adult life mainly affecting the limbic areas. Brain 1960;83:357– 368. 5. Corsellis JA, Goldberg GJ, Norton AR. “Limbic encephalitis” and its association with carcinoma. Brain 1968;91:481–496. 6. Wagner J, Witt JA, Helmstaedter C, et al. Automated volumetry of the mesiotemporal structures in antibody-associated limbic encephalitis. J Neurol Neurosurg Psychiatry 2014 Sep 2 [Epub ahead of print]. 7. Frisch C, Malter MP, Elger CE, et al. Neuropsychological course of voltage-gated potassium channel and glutamic acid decarboxylase antibody related limbic encephalitis. Eur J Neurol 2013;20:1297– 1304. 8. Malter MP, Helmstaedter C, Urbach H, et al. Antibodies to glutamic acid decarboxylase define a form of limbic encephalitis. Ann Neurol 2010;67:470–478. 9. Urbach H, Soeder BM, Jeub M, et al. Serial MRI of limbic encephalitis. Neuroradiology 2006;48:380–386. Epilepsia, **(*):1–8, 2015 doi: 10.1111/epi.12968 8 J. Wagner et al. 10. Baumgartner A, Rauer S, Mader I, et al. Cerebral FDG-PET and MRI findings in autoimmune limbic encephalitis: correlation with autoantibody types. J Neurol 2013;260:2744–2753. 11. Irani SR, Stagg CJ, Schott JM, et al. Faciobrachial dystonic seizures: the influence of immunotherapy on seizure control and prevention of cognitive impairment in a broadening phenotype. Brain 2013;136:3151–3162. 12. Wagner J, Schoene-Bake JC, Malter MP, et al. Quantitative FLAIR analysis indicates predominant affection of the amygdala in antibodyassociated limbic encephalitis. Epilepsia 2013;54:1679–1687. 13. Keller SS, Roberts N. Voxel-based morphometry of temporal lobe epilepsy: an introduction and review of the literature. Epilepsia 2008;49:741–757. 14. Vincent A, Buckley C, Schott JM, et al. Potassium channel antibodyassociated encephalopathy: a potentially immunotherapy-responsive form of limbic encephalitis. Brain 2004;127:701–712. 15. Irani SR, Alexander S, Waters P, et al. Antibodies to Kv1 potassium channel-complex proteins leucine-rich, glioma inactivated 1 protein and contactin-associated protein-2 in limbic encephalitis, Morvan’s syndrome and acquired neuromyotonia. Brain 2010;133:2734–2748. 16. Urbach H. High-field magnetic resonance imaging for epilepsy. Neuroimaging Clin N Am 2012;22:173–189, ix–x. 17. Smith SM, Jenkinson M, Woolrich MW, et al. Advances in functional and structural MR image analysis and implementation as FSL. NeuroImage 2004;23(Suppl. 1):S208–S219. 18. Sullivan EV, Rosenbloom M, Serventi KL, et al. Effects of age and sex on volumes of the thalamus, pons, and cortex. Neurobiol Aging 2004;25:185–192. 19. Giorgio A, Santelli L, Tomassini V, et al. Age-related changes in grey and white matter structure throughout adulthood. NeuroImage 2010;51:943–951. 20. Labate A, Cerasa A, Gambardella A, et al. Hippocampal and thalamic atrophy in mild temporal lobe epilepsy: a VBM study. Neurology 2008;71:1094–1101. 21. Saiz A, Blanco Y, Sabater L, et al. Spectrum of neurological syndromes associated with glutamic acid decarboxylase antibodies: diagnostic clues for this association. Brain 2008;131:2553–2563. 22. Gultekin SH, Rosenfeld MR, Voltz R, et al. Paraneoplastic limbic encephalitis: neurological symptoms, immunological findings and tumour association in 50 patients. Brain 2000;123(Pt 7):1481–1494. 23. Scanlon C, Mueller SG, Cheong I, et al. Grey and white matter abnormalities in temporal lobe epilepsy with and without mesial temporal sclerosis. J Neurol 2013;260:2320–2329. 24. Brazdil M, Marecek R, Fojtikova D, et al. Correlation study of optimized voxel-based morphometry and (1)H MRS in patients with mesial temporal lobe epilepsy and hippocampal sclerosis. Hum Brain Mapp 2009;30:1226–1235. 25. Riederer F, Lanzenberger R, Kaya M, et al. Network atrophy in temporal lobe epilepsy: a voxel-based morphometry study. Neurology 2008;71:419–425. 26. Mueller SG, Laxer KD, Cashdollar N, et al. Voxel-based optimized morphometry (VBM) of gray and white matter in temporal lobe epilepsy (TLE) with and without mesial temporal sclerosis. Epilepsia 2006;47:900–907. Epilepsia, **(*):1–8, 2015 doi: 10.1111/epi.12968 27. Cormack F, Gadian DG, Vargha-Khadem F, et al. Extra-hippocampal grey matter density abnormalities in paediatric mesial temporal sclerosis. NeuroImage 2005;27:635–643. 28. Keller SS, Wilke M, Wieshmann UC, et al. Comparison of standard and optimized voxel-based morphometry for analysis of brain changes associated with temporal lobe epilepsy. NeuroImage 2004;23:860– 868. 29. Keller SS, Wieshmann UC, Mackay CE, et al. Voxel based morphometry of grey matter abnormalities in patients with medically intractable temporal lobe epilepsy: effects of side of seizure onset and epilepsy duration. J Neurol Neurosurg Psychiatry 2002;73:648– 655. 30. Keller SS, Mackay CE, Barrick TR, et al. Voxel-based morphometric comparison of hippocampal and extrahippocampal abnormalities in patients with left and right hippocampal atrophy. NeuroImage 2002;16:23–31. 31. Oganesian EA, Melik-Musian AB, Fanardzhian VV, et al. Morphofunctional analysis of the nature of cerebello-hippocampal connections. Fiziol Zh Im I M Sechenova 1980;66:1632–1639. 32. Bertashius KM. Propagation of human complex-partial seizures: a correlation analysis. Electroencephalogr Clin Neurophysiol 1991;78:333–340. 33. Bower SP, Vogrin SJ, Morris K, et al. Amygdala volumetry in “imaging-negative” temporal lobe epilepsy. J Neurol Neurosurg Psychiatry 2003;74:1245–1249. 34. Mitsueda-Ono T, Ikeda A, Inouchi M, et al. Amygdalar enlargement in patients with temporal lobe epilepsy. J Neurol Neurosurg Psychiatry 2011;82:652–657. 35. Coan AC, Morita ME, de Campos BM, et al. Amygdala enlargement in patients with mesial temporal lobe epilepsy without hippocampal sclerosis. Front Neurol 2013;4:166. 36. Coan AC, Morita ME, Campos BM, et al. Amygdala enlargement occurs in patients with mesial temporal lobe epilepsy and hippocampal sclerosis with early epilepsy onset. Epilepsy Behav 2013; 29:390–394. 37. Takaya S, Ikeda A, Mitsueda-Ono T, et al. Temporal lobe epilepsy with amygdala enlargement: a morphologic and functional study. J Neuroimaging 2014;24:54–62. 38. Kim DW, Lee SK, Chung CK, et al. Clinical features and pathological characteristics of amygdala enlargement in mesial temporal lobe epilepsy. J Clin Neurosci 2012;19:509–512. 39. Kimura Y, Sato N, Saito Y, et al. Temporal lobe epilepsy with unilateral amygdala enlargement: morphometric MR analysis with clinical and pathological study. J Neuroimaging 2014 Mar 5 [Epub ahead of print]. 40. Minami N, Morino M, Uda T, et al. Surgery for amygdala enlargement with mesial temporal lobe epilepsy: pathological findings and seizure outcome. J Neurol Neurosurg Psychiatry 2014 Sep 15 [Epub ahead of print]. 41. Bernasconi N, Duchesne S, Janke A, et al. Whole-brain voxel-based statistical analysis of gray matter and white matter in temporal lobe epilepsy. NeuroImage 2004;23:717–723. 42. Bilevicius E, Yasuda CL, Silva MS, et al. Antiepileptic drug response in temporal lobe epilepsy: a clinical and MRI morphometry study. Neurology 2010;75:1695–1701.