Limbic Encephalitis in Patients with Epilepsy—Is Quantitative MRI Diagnostic?

Clin Neuroradiol (2019) 29:623–630 https://doi.org/10.1007/s00062-018-0705-1

ORIGINAL ARTICLE

Limbic Encephalitis in Patients with Epilepsy—is Quantitative MRI Diagnostic?

Arndt-Hendrik Schievelkamp1 · Alina Jurcoane1 · Theodor Rüber2 ·LeonErnst2 · Andreas Müller1 · Burkhard Mädler3 · Hans Heinz Schild4 · Elke Hattingen1

Received: 14 February 2018 / Accepted: 25 June 2018 / Published online: 16 July 2018 © Springer-Verlag GmbH Germany, part of Springer Nature 2018

Abstract Purpose Limbic encephalitis (LE) is an immune-related disease with limbic symptoms, variable and asymmetric magnetic resonance imaging (MRI) aspects and antibody proﬁles. This study investigated the diagnostic value of quantitative relaxation times T2 (qT2) and MRI signal intensities (SI) in LE. Methods The prospective 3T-MRI study included 39 epilepsy patients with initially suspected LE and 20 healthy controls. Values and asymmetry indices of qT2, T2-weighted (T2-w) and proton density (PD)-w SI of manually delineated and automatically segmented amygdala and hippocampus were measured. Additionally, two raters made a blinded visual analysis on FLAIR (ﬂuid attenuation inversion recovery) and T2-w images. Results According to diagnostic guidelines, 22 patients had probable LE and 17 patients had possible LE. The qT2 was higher (p< 0.01) in patients than in controls (mean± SD, amygdala 98± 7ms vs. 90± 5ms, hippocampus 101± 7ms vs. 92± 3ms), but was not different between probable and possible LE or between sides (left and right). The PD-w SI and T2-w SI were lower in patients than in controls but were not different between patient subgroups or between sides. Diagnostic performance of visual analysis was relatively poor. Conclusions Epilepsy patients with suspected LE had elevated qT2 in amygdala and hippocampus, whereas the expected T2-w SI increase was not found; however, the diagnostic value of qT2 remains questionable since it did not discriminate probable from possible LE.

Keywords Limbic encephalitis · Amygdala · Hippocampus · T2-weighted MRI · T2-relaxation time

Arndt-Hendrik Schievelkamp [email protected]

Alina Jurcoane Elke Hattingen [email protected] [email protected] Theodor Rüber [email protected] 1 Neuroradiology, Department of Radiology, University Hospital Bonn, Sigmund Freud Str. 25, 53127 Bonn, Leon Ernst Germany [email protected] 2 Department of Epileptology, University Hospital Bonn, Bonn, Andreas Müller Germany [email protected] 3 Philips GmbH, UB Healthcare, Burkhard Mädler Lübeckertordamm 5, 20099 Hamburg, [email protected] Germany Hans Heinz Schild 4 Department of Radiology, University Hospital Bonn, [email protected] Sigmund Freud Str. 25, 53127 Bonn, Germany

K 624 A.-H. Schievelkamp et al.

Abbreviations changes, especially in the group II anti-N-methyl-D-aspar- FLAIR Fluid attenuation inversion recovery tate (NMDA) receptor and VGKC(Voltage-gated potassium FOV Field of view channel)-associated encephalitis [23–25]. Concordant with LE Limbic encephalitis the literature, many LE patients from our large epilepsy LI Laterality index center had normal MR appearance of the amygdala and NMDA N-methyl-D-aspartate receptor hippocampus or they only had subtle signal changes of PD Proton density these mesiotemporal regions on T2-weighted and FLAIR PD-w Proton density weighted images. This experience prompted this study that examined qT2 Quantified T2-relaxation time diagnostic value of T2-relaxometry in LE compared to con- ROI Region of interest ventional MRI. SI Signal intensity The quantitative relaxation time T2 (qT2) is a major tis- T2-w T2-weighted sue-related physical MRI parameter, which dominates the TE Echo time signal of T2-w and T2-FLAIR images. Longer qT2 as well TR Repetition time as higher density of water protons yield higher SI on T2- VGKC Voltage-gated potassium channel w and the T2-FLAIR images. Therefore, the measurement of the qT2 may have a better diagnostic accuracy in patients with suspected LE compared to visual assessment Introduction or measurement of signal intensities on conventional MRI. To prove this hypothesis, conventional MRI sequences and Limbic encephalitis (LE) is an immune-related syndrome quantitative T2 maps were analyzed in patients admitted in with subacute onset of neurological symptoms assigned to a large epilepsy center with suspected diagnosis of LE. the limbic system [1, 2]. According to the German guidelines, the patient should have been symptomatic within the last 5 years with at least one limbic symptom, e.g. Methods dysfunction of the episodic memory, seizures with temporal semiology, or psychiatric symptoms with affect lability Subjects [3]. In addition to these limbic symptoms, positive autoimmune antibodies in the cerebrospinal fluid (CSF), tumor All participants in this prospective observational, non-inter- disease or/and typical magnetic resonance imaging (MRI) ventional cohort study signed an institutional review board- features or alternatively typical histopathological findings approved informed consent prior to enrolment. The study are required to make the diagnosis of LE [3]. Typical an- was approved by the local ethics committee (ref. nr 372/15) tibodies in LE are either antibodies targeting intracellular and performed in accordance with the ethical standards antigens, which are mainly caused by cytotoxic T-cell me- laid down in the 1964 Declaration of Helsinki and its later diated cross-reactions between oncogens and intracellular amendments. neuronal antigens (group I) [1, 4–7] and with non-parane- During a 21-month period 39 patients (mean age 39.4 oplastic antibodies [1, 8–10] targeting antigens on the cell years, range 9–75 years; 15 female) with the initial sus- surface (group II) [1, 4–6, 11, 12]. While group I antibod- pected diagnosis of LE from a large epilepsy center were ies can also occur in cancer patients without paraneoplastic enrolled. Exclusion criteria were MRI contraindications syndromes [1, 7, 13], group II antibodies are more specific (such as pacemakers, foreign metals, claustrophobia, elec- for autoimmune encephalitis [14, 15]. These antibodies in- tronic implants). teract with membrane receptors or ion channels which are From patients with suspected LE, the definitive diag- mainly involved in the synaptic transmission and in home- nosis of LE was made according to the national guide- ostasis of the membrane potential [6, 16, 17]. If no antibod- lines of LE [3] by the treating physician without interfer- ies can be identified, the case is classified as antibody-nega- ence from the radiologist. The diagnosis considered clinical tive LE [3]. Thus, antibody detection has limited diagnostic symptoms including seizures of temporal semiology, mem- value so that characteristic features on MRI are pivotal to ory disturbances and electroencephalogram (EEG) findings, diagnose LE [18, 19]. These MRI features are increased therapy response after anti-autoimmune treatment, autoim- signal intensities (SI) in T2-weighted spin-echo or FLAIR mune antibody profile, and/or MR imaging on admission images mainly of the amygdala, but also of the hippocam- and under treatment. Clinical criteria (seizures with tem- pus and insula. Initially, these structures are swollen but poral semiology and neuropsychological testing) and an- within weeks to months, swelling may disappear and pro- tibody status were essential for the diagnosis as the MRI gressive atrophy or sclerosis may occur [20–22]; however, findings were often only suggestive of limbic pathology MRI frequently shows normal results or only minor signal or even inconclusive. Depending on whether criteria were

K Limbic Encephalitis in Patients with Epilepsy—is Quantitative MRI Diagnostic? 625 fulﬁlled or not, patients were divided in patients with prob- 3874ms, matrix size 512× 512, FOV 240× 240mm, 4mm able LE and possible LE, related to the proposal of Graus slice thickness. et al. [26]. The diagnosis and the classiﬁcation into pos- For the healthy controls, only the multi-echo sequence sible or probable LE were therefore independent of the to calculate the T2-relaxation time was acquired. radiology-focused analysis performed for this study. Ad- ditionally, 20 healthy age-matched volunteers (mean age T2-Relaxometry and Conventional MRI 39.9 years, range 22–63 years; 11 female) were enrolled as control group. Additional exclusion criteria for controls Based on the signal changes of the multi-echo sequence, (except the MRI contraindications) were any neurological quantitative T2-maps of T2 relaxation time (qT2) were or psychiatric symptoms. calculated by linear matrix inversion of a mono-exponen- tial model. Two sequences of the multi-echo data set with MR Imaging TE= 15 (predominantly PD-w) and with TE= 105 (predominantly T2-w) were also analyzed. Prior to analysis, PD-w Protocol and T2-w images were intensity normalized by setting the whole image intensity average to an arbitrary value of 100 All patients with suspected LE underwent the same MRI [22]. Regions of interest (ROIs) were manually placed in protocol on an Achieva 3.0T TX, Philips Healthcare (Best, both the amygdala and hippocampus (Fig. 1) and extracted The Netherlands), assigned for all patients with epilepsy the T2 relaxation time qT2 and the PD-w and T2-w signal which included among others a 3D T1-weighted (T1-TFE intensities (SI) from these ROIs. 3D) MRI measurement (TE= 3.8ms, TR= 8.2ms, matrix size 256× 256, FOV 256× 256mm). Segmentation of Amygdala and Hippocampus In addition, a multi-echo sequence with increasing equidistant echo times (TE= 15, 30, 45, 60, 75, 90, 105 In the patient group, for which the 3D T1-w sequence was and 120ms) was acquired to calculate the T2-relaxation available, the right and left amygdala and hippocampus time. This sequence was angulated perpendicular to the were also automatically segmented with FIRST (https://fsl. mesiotemporal structures; additional parameters were TR fmrib.ox.ac.uk/fsl/fslwiki/FIRST), and their volumes were

Fig. 1 Regions of interest (ROIs) were manually placed with the ImageJ software (U.S. National Institutes of Health, Bethesda, Maryland, USA) in the right (1)andleft(2) amygdala (a–c), and hippocampus (d–f) and MRI parameters (qT2, PD-w SI and T2-w SI) were extracted from these ROIs

K 626 A.-H. Schievelkamp et al. noted (normalized to the subject’s total intracranial vol- statistical analyses were performed with R Statistics (ver- ume). These 3D T1-w images were then linearly registered sion 3.2.4;2016-03-10, http://www.R-project.org/)software with reg_aladin (http://cmictig.cs.ucl.ac.uk/wiki/index.php/ package. Group results are presented as mean± standard Reg_aladin) to the TSE sequences, their qT2, PD-w SI and deviation and the level of statistical signiﬁcance was set at T2-w SI were extracted and compared with those from the p< 0.05. manually segmented ROIs.

Laterality Results

For each subject, parameter (qT2, PD-w SI, T2-w SI, vol- Subjects ume) and structure (amygdala, hippocampus) the absolute laterality index (LI) was calculated as follows: A total of 39 consecutive patients were included with initial suspected diagnosis of LE. According to the abovemen- jLeft − Rightj LI = tioned criteria, based on national guidelines of LE, 22 pa- .Left + Right/ tients with probable LE (mean age 42 years, 9 females), and The modulus of the difference between left and right was 17 patients with possible LE (mean age 36.1, 6 females) used so that the value for the LI would always be positive. were investigated. The control group of 20 age-matched healthy controls had a mean age of 39.9 years and 11 were Visual Analysis females.

Two neuroradiologists blinded for the clinical data (with 15 MR Imaging and 8 years of experience, respectively) independently evaluated the T2-fluid-attenuated inversion recovery (FLAIR) Visual Analysis and T2-w MR images and assessed the MR signs of LE. They rated the presence or absence of swelling and/or signal For the visual rating, the more experienced rater had increase in the amygdala and hippocampus, mainly on T2- a higher accuracy than the less experienced rater. Although FLAIR, and the loss of the hippocampal internal structure on individual basis there were patients with clear imaging best seen on high-resolution T2-w images. findings (Fig. 2), the overall performance of both raters was rather poor (rater 1 versus rater 2—sensitivity 0.83 Statistics vs. 0.45, specificity 0.65 vs. 0.88, positive predictive value 0.75 vs. 0.83, negative predictive value 0.73 vs. 0.56 and Differences between groups were evaluated using a one- accuracy 0.74 vs. 0.64). The inter-rater agreement was also way analysis of variance (ANOVA) with side (left or right), poor, Cohen’s kappa was 0.25 for the presence of LE, 0.4 method (manual or automatic) and cortisone therapy (yes for the amygdala volume increase, 0.43 for the amygdala or no) as covariates. Post hoc pairwise comparisons were signal increase, 0.19 for the hippocampi volume increase, done with the Tukey test. In patients, the manual analy- 0.37 for the hippocampi signal increase, 0.36 for the wash- sis was correlated with the segmentation approach with the out of the hippocampal internal structure and 0.18 for the Spearman’s rho correlation coefficient. For the visual anal- hippocampal sclerosis. ysis the performance of each rater relative to the clinical diagnosis of LE was evaluated and the inter-rater agreement was measured with the Cohen’s Kappa coefficient. All

Fig. 2 Clear imaging ﬁnding in a patient with limbic encephalitis. Especially on the left side the amygdala is enlarged and shows a higher SI on the coronal (a) and the axial (b) T2-w images

K Limbic Encephalitis in Patients with Epilepsy—is Quantitative MRI Diagnostic? 627

Table 1 ANOVA results of the ROI-based analysis (means and conﬁdence intervals), for groups and subgroups, right and left sides averaged. Post hoc pairwise comparisons were done with the Tukey test (1) Controls (2) LE patients (2.1) Possible LE (2.2) Probable LE ANOVA n=20 n=39 n=17 n=22 Amygdala qT2 (ms) 90± 5 98± 7 99± 8 98± 5 F(1.57)= 24.4, p< 0.001* F(2.56)= 12.4, p< 0.001* (1)–(2) p< 0.001* (1)–(2.1) p< 0.001* (1)–(2.2) p< 0.001* (2.1)–(2.2) p=0.72 PD-w SI 206± 17 190± 27 186± 28 193± 17 F(1.57)= 5.8, p= 0.02* F(2.56)= 3.3, p= 0.046 (1)–(2) p= 0.02* (1)–(2.1) p= 0.04* (1)–(2.2) p=0.2 (2.1)–(2.2) p=0.66 T2-w SI 194± 21 179± 29 173± 32 183± 21 F(1.57)= 4.3, p= 0.042* F(2.56)= 2.9, p= 0.061 (1)–(2) p= 0.04* (1)–(2.1) p= 0.05 (1)–(2.2) p= 0.4 (2.1)–(2.2) p= 0.45 Hippocampus qT2 (ms) 92± 3 101± 7 101± 7 100± 3 F(1.57)= 24.7, p< 0.001* F(2.56)= 12.4, p< 0.001* (1)–(2) p< 0.001* (1)–(2.1) p< 0.001* (1)–(2.2) p< 0.001* (2.1)–(2.2) p=0.79 PD-w SI 221± 17 198± 29 198± 29 201± 17 F(1.57)= 10.6, p= 0.002* F(2.56)= 5.7, p= 0.005* (1)–(2) p< 0.001* (1)–(2.1) p= 0.01* (1)–(2.2) p= 0.04* (2.1)–(2.2) p=0.62 T2-w SI 212± 21 190± 32 190± 32 195± 21 F(1.57)= 7.8, p= 0.007* F(2.56)= 4.7, p= 0.013* (1)–(2) p= 0.01* (1)–(2.1) p= 0.01* (1)–(2.2) p=0.14 (2.1)–(2.2) p=0.44 SI is unitless ANOVA analysis of variance, F F-test, PD-w proton density weighted, qT2 quantiﬁed T2-relaxation time, SI signal intensity, T2-w T2-weighted, (1)-(2) p-values for comparison of group (1) with group (2)

T2-relaxometry and Conventional MRI but no difference between patient subgroups (p=0.72for amygdala and p= 0.79 for hippocampus) (Table 1;Fig.3). The group analysis demonstrated a signiﬁcant difference The signals of the PD-w and T2-w images in both of the in the T2 relaxation times of the hippocampus and amyg- amygdala and the hippocampus were lower in patients than dala among the three groups (ANOVA, p< 0.001). Tukey’s in controls (see Table 1) with no difference between pa- pairwise post hoc comparisons showed a signiﬁcantly pro- tient subgroups. The segmentation based analysis delivered longed qT2 in patients compared with controls (p< 0.001), similar results as the ROI-based analysis. The two methods

K 628 A.-H. Schievelkamp et al.

Fig. 3 ROI analysis for both right and left sides, for qT2, PD-w SI and T2-w SI in the amygdala (upper row) and hippocampus (lower row). Signif- icant differences are marked with asterisk (*), non-signiﬁcant differences are marked with NS. For p-values see Table 1

correlated significantly (p< 0.001, Spearman rho between ization of SI or qT2 in any of the (sub)groups. Cortisone 0.74 and 0.98) and ANOVA analysis showed no significant administration had no influence on the results (ANOVA differences between methods for qT2, PD-w SI and T2-w p> 0.05). SI.

Volumetry Discussion

Amygdala volume of possible LE patients was marginally The visual assessment of conventional MRI focused on higher than that of probable LE patients (1.21± 0.18ml vs. FLAIR and T2-w images did not prove to be diagnostic 1.1± 0.15ml, ANOVA p= 0.049). Hippocampus volumes for the differentiation between healthy subjects and pa- were not different between the two subgroups. tients with suspected LE, whereas T2-relaxation times of amygdala and hippocampus seemed to be useful. The T2- Laterality relaxation times quantified in ms of the amygdala and hippocampus were higher in epilepsy patients with suspected No significant differences were found between right and LE compared to healthy controls. In contrast, the dimen- left sides (ANOVA p> 0.05); however, in the hippocampus, sionless and normalized measurement of the SI on PD-w patients with probable LE showed more qT2 lateralization and T2-w images was lower in patients than in controls. than the controls (mean absolute laterality± SD 0.04± 0.04 The segmentation-based analysis delivered similar results vs. 0.02± 0.01, p= 0.03) and more T2-w lateralization as the ROI-based analysis of qT2 and of the signal in- (0.04± 0.04) than the possible LE patients (0.02± 0.02, tensities. Automatically segmented amygdala volume was p= 0.05). In the amygdala, there was no significant lateral-

K Limbic Encephalitis in Patients with Epilepsy—is Quantitative MRI Diagnostic? 629 marginally higher in possible LE patients than in probable tient group. The PD-SI is mainly influenced by the amount LE patients. of water in the examined tissue. The measured decrease This result was unanticipated since the increase of T2 of the PD-SI could be explained by infiltrating T-cells or relaxation times would be expected to result in an increase activated microglia in the course of the inflammation that of the signal in T2-w images instead of the decrease found might displace interstitial water [28]. in this study. This result is also discrepant to the previously Thirdly, MRI signals are influenced by inhomogeneities reported imaging features in LE patients. Another unex- of the magnetic field B0 and inhomogeneities from the coil pected result was the lack of laterality. There was a signifi- profile B1, whereby the visual assessment is additionally cant asymmetry (laterality index) in the LE patients only for compromised. This is the main difference to the quantita- the qT2 and the SI on T2-w images in the hippocampus but tive measure qT2, because tissue contrast on these maps not in the amygdala. The MRI features in LE are reported is purely influenced by the relaxation time T2, minimizing as asymmetric swelling and signal increase on T2-w and influences of B0 and B1 inhomogeneities. FLAIR images of the amygdala and, less pronounced, of Fourthly, previous therapy with steroids might also alter other limbic structures [20–22]. The study of Wagner et al. the MRI signal. As dexamethasone can reduce the perme- showed higher signal intensities on FLAIR images in the ability of the blood-brain barrier, it might decrease the PD amygdala of LE patients with seizures. Still our results are signal by decreasing tissue water [29]; however, we can discrepant to this FLAIR-analysis. A possible explanation exclude effects of a previous therapy with steroids on our beyond the different sequence type might be the conducted measurements, as a group comparison of our patient group bias correction and intensity normalization of the FLAIR showed no significant differences between patients who had images in this study which is not a routine procedure [22]. a received therapy with steroids and patients without steroid We did not quantitatively evaluate the FLAIR-SI since therapy (results not shown). Changes in MRI can also be we extracted the T2-w and PD-w images from the same seen after recent seizures [30]. As our patients did not have multi-echo sequence. This procedure should guarantee the seizures recently before the examination or during the ex- best comparability between conventional SI and qT2 mea- amination, we can exclude that our measurements are in- sures. We further had to expect patients’ movements due fluenced by seizures. to the long acquisition time of our sequences that could in- In contrast to the SI of FLAIR and T2-w images, qT2 fluence the regional SI (caused by partial volume effects to values are not influenced by field inhomogeneities. Differ- the CSF or different B1 inhomogeneities, for instance). Al- ences in qT2 were not only found in patients with prob- though FLAIR sequences are commonly considered as T2- able LE but also in patients with possible LE. Therefore, weighted imaging with dark cerebrospinal fluid, their signal this quantitative measure may help to separate healthy from changes are more ambiguous since they are also influenced symptomatic subjects, but it seems not to be diagnostic to by mild T1-weighted contrast produced by the long TI [27]. verify the LE diagnosis. Finally, this is a group analysis Various other reasons for this discrepancy need to be dis- which means that most of the patients may fail to have cussed. Firstly, we exclusively analyzed patients admitted clear imaging criteria; however, this does not exclude that to a large epilepsy center presenting with seizures mostly on an individual basis, there are patients with clear imaging of temporal semiology yielding the suspected diagnosis LE. findings (Fig. 2). The LE patients with epilepsy seldom have an underlying malignancy, in contrast to LE patients investigated in previous MRI studies and reports [1, 6, 10]. All but one of our Conclusion patients had non-paraneoplastic LE with different antibody profiles compared to paraneoplastic LE. In our experience The MRI criteria reported in the literature do not seem to non-paraneoplastic forms tend to have a milder presentation accurately ensure the diagnosis of LE in epileptic patients in in MRI compared to the paraneoplastic forms. Also many whom LE is suspected. In contrast, elevated T2 relaxation of the patients had a longer history of seizures, meaning times measured in the amygdala and hippocampus seem to that the LE is also not in the acute stage when discovered be helpful; however, it remains questionable if quantitative having a possible milder presentation in MRI [20–22]. MR is diagnostic since it did not discriminate probable from Secondly, the signal on T2-weighted images is not only possible LE. influenced by the T2-relaxation times but also by the proton Conflict of interest A.-H. Schievelkamp, A. Jurcoane, T. Rüber, density, where the higher content of protons yields higher L. Ernst, A. Müller, B. Mädler, H.H. Schild, E. Hattingen and declare signals. We found a lower SI on PD-w images in the hip- that they have no competing interests. pocampus and also in the amygdala of LE patients compared to healthy controls, which might explain why also the signal on T2-weighted images was not increased in our pa-

K 630 A.-H. Schievelkamp et al.

References Titulaer MJ, Balice-Gordon R, Graus F, Dalmau J. Encephali- tis with refractory seizures, status epilepticus, and antibodies to the GABAA receptor: a case series, characterisation of the anti- 1. Kelley BP, Patel SC, Marin HL, Corrigan JJ, Mitsias PD, Grif- gen, and analysis of the effects of antibodies. Lancet Neurol. fith B. Autoimmune encephalitis: pathophysiology and imaging 2014;13:276–86. review of an overlooked diagnosis. AJNR Am J Neuroradiol. 18. Leypoldt F, Wandinger K-P, Bien CG, Dalmau J. Autoimmune en- 2017;38:1070–8. cephalitis. Eur Neurol Rev. 2013;8:31–7. 2. da Rocha AJ, Nunes RH, Maia AC Jr, do Amaral LL. Recognizing 19. Titulaer MJ, McCracken L, Gabilondo I, Armangué T, Glaser C, autoimmune-mediated encephalitis in the differential diagnosis of Iizuka T, Honig LS, Benseler SM, Kawachi I, Martinez-Hernandez limbic disorders. AJNR Am J Neuroradiol. 2015;36:2196–205. E, Aguilar E, Gresa-Arribas N, Ryan-Florance N, Torrents A, Saiz 3. Bien CG, Deutsche Gesellschaft für Neurologie. S1-Leitlinie: Im- A, Rosenfeld MR, Balice-Gordon R, Graus F, Dalmau J. Treatment munvermittelte Erkrankungen der grauen ZNS-Substanz sowie and prognostic factors for long-term outcome in patients with anti- Neurosarkoidose. 2012. https://www.dgn.org/leitlinien/2396-ll-32- NMDA receptor encephalitis: an observational cohort study. Lancet 2012-immunvermittelte-erkrankungen-der-grauen-zns-substanz- Neurol. 2013;12:157–65. sowie-neurosarkoidose. Accessed 31 Aug 2017. 20. Urbach H, Soeder BM, Jeub M, Klockgether T, Meyer B, Bien CG. 4. Dalmau J, Rosenfeld MR. Autoimmune encephalitis update. Neuro Serial MRI of limbic encephalitis. Neuroradiology. 2006;48:380–6. Oncol. 2014;16:771-8. 21. Wagner J, Witt J-A, Helmstaedter C, Malter MP, Weber B, Elger 5. Graus F, Saiz A, Dalmau J. Antibodies and neuronal autoimmune CE. Automated volumetry of the mesiotemporal structures in anti- disorders of the CNS. J Neurol. 2010;257:509–17. body-associated limbic encephalitis. J Neurol Neurosurg Psychiatr. 6. Bien CG, Vincent A, Barnett MH, Becker AJ, Blümcke I, Graus 2015;86:735–42. F, Jellinger KA, Reuss DE, Ribalta T, Schlegel J, Sutton I, Lass- 22. Wagner J, Schoene-Bake JC, Malter MP, Urbach H, Huppertz HJ, mann H, Bauer J. Immunopathology of autoantibody-associated en- Elger CE, Weber B. Quantitative FLAIR analysis indicates predom- cephalitides: clues for pathogenesis. Brain. 2012;135:1622–38. inant affection of the amygdala in antibody-associated limbic en- 7. Dalmau J, Bataller L. Clinical and immunological diversity of lim- cephalitis. Epilepsia. 2013;54:1679–87. bic encephalitis: a model for paraneoplastic neurologic disorders. 23. Irani SR, Bera K, Waters P, Zuliani L, Maxwell S, Zandi MS, Hematol Oncol Clin North Am. 2006;20:1319–35. Friese MA, Galea I, Kullmann DM, Beeson D, Lang B, Bien CG, 8. Mittal MK, Rabinstein AA, Hocker SE, Pittock SJ, Wijdicks EF, Vincent A. N-methyl-D-aspartate antibody encephalitis: temporal McKeon A. Autoimmune encephalitis in the ICU: analysis of progression of clinical and paraclinical observations in a pre- phenotypes, serologic findings, and outcomes. Neurocrit Care. dominantly non-paraneoplastic disorder of both sexes. Brain. 2016;24:240–50. 2010;133:1655–67. 9. Gultekin SH, Rosenfeld MR, Voltz R, Eichen J, Posner JB, Dalmau 24. Florance NR, Davis RL, Lam C, Szperka C, Zhou L, Ahmad S, J. Paraneoplastic limbic encephalitis: neurological symptoms, im- Campen CJ, Moss H, Peter N, Gleichman AJ, Glaser CA, Lynch munological findings and tumour association in 50 patients. Brain. DR, Rosenfeld MR, Dalmau J. Anti-N-methyl-D-aspartate receptor 2000;123(Pt 7):1481–94. (NMDAR) encephalitis in children and adolescents. Ann Neurol. 10. Dalmau J, Lancaster E, Martinez-Hernandez E, Rosenfeld MR, 2009;66:11–8. Balice-Gordon R. Clinical experience and laboratory investiga- 25. Kotsenas AL, Watson RE, Pittock SJ, Britton JW, Hoye SL, Quek tions in patients with anti-NMDAR encephalitis. Lancet Neurol. AM, Shin C, Klein CJ. MRI findings in autoimmune voltage-gated 2011;10:63–74. potassium channel complex encephalitis with seizures: one poten- 11. Vincent A. Autoimmune channelopathies: new antibody-mediated tial etiology for mesial temporal sclerosis. AJNR Am J Neuroradiol. disorders of the central nervous system. F1000 Biol Rep. 2009;1:61. 2014;35:84–9. 12. Tüzün E, Dalmau J. Limbic encephalitis and variants: classification, 26. Graus F, Titulaer MJ, Balu R, Benseler S, Bien CG, Cellucci T, diagnosis and treatment. Neurologist. 2007;13:261–71. Cortese I, Dale RC, Gelfand JM, Geschwind M, Glaser CA, Hon- 13. Pittock SJ, Kryzer TJ, Lennon VA. Paraneoplastic antibodies co- norat J, Höftberger R, Iizuka T1, Irani SR, Lancaster E, Leypoldt F, exist and predict cancer, not neurological syndrome. Ann Neurol. Prüss H, Rae-Grant A, Reindl M, Rosenfeld MR, Rostásy K, Saiz 2004;56:715–9. A, Venkatesan A, Vincent A, Wandinger KP, Waters P, Dalmau J. A 14. Nunez-Enamorado N, Camacho-Salas A, Belda-Hofheinz S, clinical approach to diagnosis of autoimmune encephalitis. Lancet Cordero-Castro C, Simon-De Las Heras R, Saiz-Diaz R, Martinez- Neurol. 2016;15:391–404. Sarries FJ, Martinez-Menendez B, Graus F. Fast and spectacular 27. Essig M, Knopp MV, Schoenberg SO, Hawighorst H, Wenz F, De- clinical response to plasmapheresis in a paediatric case of anti- bus J, van Kaick G. Cerebral gliomas and metastases: assessment NMDA encephalitis. Rev Neurol. 2012;54:420–4. with contrast-enhanced fast fluid-attenuated inversion-recovery MR 15. Gresa-Arribas N, Titulaer MJ, Torrents A, Aguilar E, McCracken imaging. Radiology. 1999;210:551–7. L, Leypoldt F, Gleichman AJ, Balice-Gordon R, Rosenfeld MR, 28. Khan NL, Jeffree MA, Good C, Macleod W, Al-Sarraj S. Histopathol- Lynch D, Graus F, Dalmau J. Antibody titres at diagnosis and dur- ogy of VGKC antibody-associated limbic encephalitis. Neurology. ing follow-up of anti-NMDA receptor encephalitis: a retrospective 2009;72:1703–5. study. Lancet Neurol. 2014;13:167–77. 29. Kotsarini C, Griffiths PD, Wilkinson ID, Hoggard N. A systematic 16. Moscato EH, Jain A, Peng X, Hughes EG, Dalmau J, Balice-Gor- review of the literature on the effects of dexamethasone on the brain don RJ. Mechanisms underlying autoimmune synaptic encephalitis from in vivo human-based studies: implications for physiological leading to disorders of memory, behavior and cognition: insights brain imaging of patients with intracranial tumors. Neurosurgery. from molecular, cellular and synaptic studies. Eur J Neurosci. 2010;67:1799–815. discussion 1815. 2010;32:298–309. 30. Cole AJ. Status epilepticus and periictal imaging. Epilepsia. 17. Petit-Pedrol M, Armangue T, Peng X, Bataller L, Cellucci T, Davis 2004;45(Suppl 4):72–7. R, McCracken L, Martinez-Hernandez E, Mason WP, Kruer MC, Ritacco DG, Grisold W, Meaney BF, Alcalá C, Sillevis-Smitt P,