Background Two novel antibodies (abs) directed to γ-aminobutyric acid B receptor (GABABR) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) in patients with limbic encephalitis (LE) were first described by the Philadelphia/Barcelona groups and confirmed by the Mayo group. We present a novel series for further clinical and paraclinical refinement.
Methods Serum and cerebrospinal fluid samples from a diagnostic laboratory were selected if found to be positive for GABABR or AMPAR abs within a broad antineuronal ab panel. Data were retrospectively compiled.
Results In 10 patients, we detected abs to GABABR. Median age was 70 years. Five of them were diagnosed with small cell lung cancer (SCLC). Intrathecal GABABR ab synthesis was found in all six patients with sufficient data available (median ab-index: 76.8). On MRI, we found bilateral mediotemporal and in two cases cortical abnormalities. EEG revealed encephalopathy, partly with epileptiform discharges. Five patients received immunotherapy, two patients tumour treatment and three both therapies. Three patients died, in five patients cognitive functions declined, one patient improved slightly and one patient fully recovered. AMPAR abs were detected in three patients with mnestic disturbances. Median age was 60.7 years. The only female patient was diagnosed with ovarian cancer. None of the patients had intrathecal ab synthesis. MRI findings showed bilateral mediotemporal abnormalities. EEG was normal in all patients. Two of the three immunologically treated patients improved, one patient stabilised on a low level.
Discussion GABABR and AMPAR abs are well associated with LE. GABABR abs lead to severe clinical, neuroradiological and EEG abnormalities with poorer outcome.
- CLINICAL NEUROLOGY
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Autoantibodies to extracellular epitopes of neurons are increasingly recognised in clinical neurology. There appear to be effective treatment options, in particular immunosuppression and (if applicable) tumour treatment. Recognition of these autoimmune diseases therefore should be beneficial for patients.1 Recently, autoantibodies to γ-aminobutyric acid B receptor (GABABR) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) were detected. GABABR antibodies (abs) have been characterised by their discoverers as the causative agents of limbic encephalitis (LE).2 ,3 Single cases have been reported with cerebellar ataxia,3 ,4 opsoclonus-myoclonus syndrome or status epilepticus.3 AMPAR abs directed against a subtype of glutamate receptors have been described as defining another form of immune-mediated LE.5 Both receptors are involved in synaptic plasticity underlying different processes such as learning, behaviour and memory.6 ,7
In this study, we comparatively present clinical, cerebrospinal fluid (CSF), EEG and MRI data of patients with GABABR and AMPAR ab-associated LE. This was done to confirm, extend and refine the data from the original series presented by the Philadelphia/Barcelona groups2 ,5 and confirmed by the Mayo group.8 LE was defined by at least two of the following: seizures of temporal semiology, disturbance of episodic memory, affective disturbances with prominent mood lability, or disinhibition.9 With onconeural abs, the unambiguous identification of different laboratories was once suggested as pre-requisite for ‘well-characterised antibodies’.10
Sera, CSFs and CSF–serum pairs from 4819 patients from centres across Germany, Austria and Switzerland were tested for antineuronal abs from November 2011 to February 2014. A sample of 300 consecutive patients had the following distribution of suspected diagnoses: encephalitis/encephalopathy 42%, epilepsy 22%, cognitive or psychiatric disorder 9%, peripheral neurological disorder 4%, other or no specified information 23%. Data of the patients with abs to GABABR and AMPAR were contributed for clinical discussion by the referring physicians. For the purposes of this study, data were retrospectively completed. All available brain MRI were reviewed by one author (FW). Specific ab indices (titre of specific ab (CSF)/titre of specific ab (serum))/(total IgG concentration (CSF)/total IgG concentration (serum)) were calculated. Values >2 were taken as evidence for intrathecal synthesis of the specific ab.11 For calculating the specific ab synthesis we took the earliest serum/CSF pair available with all necessary information at hand. Albumin CSF/serum ratio (QAlb) was calculated to determine the integrity of the blood–brain barrier (BBB)7 with QAlb=(4+age/15)×10−3 as the upper limit of the reference range.12
For detection of abs, commercially available postfixed sagittal mouse brain sections and cell-based assays in the form of commercially available biochips (Euroimmun, Lübeck, Germany) were used. The assays consist of human embryonic kidney (HEK 293) cells transfected with plasmids encoding the following antigens (subsequent fixation with the substances given in brackets): N-methyl-d-aspartate receptor (NMDAR; consisting of NR1 subunits only), NMDAR (NR1-NR2), glutamic acid decarboxylase (GAD) GAD65, GAD67 (acetone), leucine-rich glioma inactivated1 protein (LGI1), contactin-associated protein-like2 (CASPR2), AMPAR1, AMPAR2, GABABR, glycine receptor (GlyR) (paraformaldehyde). Their preparation has been described before (cf. ref. 13) according to the method given in ref. 14. The protocol for indirect immunofluorescence follows the recommendations given by Euroimmun (FA 112d-1005-1, IgG) with few modifications: serum samples were diluted to 1:15 before incubation (Euroimmun: 1:10); the buffer was the phosphate buffer solution (PBS; Euroimmun: PBS-Tween); the secondary system consisted of a goat-antihuman IgG (heavy and light chain) ab conjugated with DyLight 594 produced by Jackson ImmunoResearch, West Grove, Pennsylvania, USA, Code No. 109-515-088 at a dilution of 1:100 (Euroimmun: goat-antihuman IgG, conjugated with fluorescein, no further information given); nuclear counterstaining with Hoechst 33 342 at 1:10 000 (Euroimmun: no nuclear staining); embedding with 1,4-diazabicyclo[2.2.2]octan (Euroimmun: glycerol). The stained biochips were examined using a fluorescence microscope (Leica DM 2000; Wetzlar, Germany) with excitation at 592 nm and emission filter at 616 nm for bound ab and 350/462 nm for the nuclear counterstain. The decision if an ab was present in the tested material was made by one of the three investigators (CGB, CK and MDO) using the signal of the surrounding (supposedly negative) fields as respective negative controls.
An end point titration was performed by serial dilutions in a multiple of 1:2. The titre is the concentration at which a signal is still just detectable in comparison with adjacently stained non-transfected cells. Each titration is rated by two of the four investigators (CGB, CK, MDO or an experienced technician) independently. In cases of divergent ratings, the mean of the two ratings is recorded.
Using indirect immunofluorescence on mouse brain tissue and transfected HEK293 cells, we found 10 patients to harbour GABAB1+2R abs and 3 AMPAR2 abs (figure 1). AMPAR1 abs were not found in any of our patients. (Besides this, 206 patients were identified as being positive for abs to the following other antigens: GAD65 N=67; NMDAR N=55; LGI1 N=44; CASPR2 at titres >1:100 N=14; glycin receptor N=6; Hu N=5; Ma N=1; amphiphysin N=2; Yo N=4; Sox-1 N=4; amphiphysin and Hu N=1; Hu and Sox-1 N=3.) In the eight patients with CSF–serum pairs available, the abs were found in both materials. All GABABR and all AMPAR ab-positive cases had predominantly limbic syndromes: only one patient with GABABR abs (and Sox-1 abs) had a pure Lambert Eaton myasthenic syndrome; this case is not further considered here and has been reported separately.15
The median age was 70 years (range 51–75), three of the patients were women. Eight patients presented with generalised convulsive seizures being the primary reason for admittance to hospital. Among these eight patients two presented with status epilepticus and one with a series of generalised tonic–clonic seizures. Seizures were accompanied by memory deficits, confusion, in part also disorientation and behavioural abnormalities such as aggressiveness and agitation. One patient developed severe immobility and a catatonic-like condition. Apraxia and aphasia were observed in three patients. Five patients showed cerebellar dysfunction. Three of them presented with gait and stance ataxia, one with limb ataxia and bradydiadochokinesis, two with intentional tremor and two with pathological nystagmus. Two patients (4 and 6) had prodromal symptoms consisting of vertigo and malaise 1–4 days before the onset of neurological symptoms.
In five patients, small cell lung cancer (SCLC) was detected, in all of them after manifestation of the neurological symptoms within a median of 18 days (range 0–35). In five patients, tumour searches (CT of the chest and abdomen or whole body fluorodeoxyglucose positron emission tomography (FDG-PET), at minimum two examinations) were initiated without an indicative result. Follow-up period in these patients covered in median 3 months (range 2–12).
None of the 10 patients had a history of autoimmune disease. Four patients had additional abs against Sox-1 (6, 8 and 10) and Hu (3; see table 1). These patients did not show significant differences regarding clinical, paraclinical features, response to therapy and outcome compared with the patients without the evidence of intracellular antigens.
All patients in our series had CSF studies. Lymphocytic pleocytosis (>4 cells/µL) was found in 6/9 patients with available data (median white cell count was 13/μL; range 2–78) at first spinal tapping. These data were compiled within a median of 4 days after onset of symptoms (range 0–33). Intrathecal IgG synthesis evidenced by elevated IgG indices or unmatched oligoclonal bands was found in 7/10. Extensive microbiology studies were without abnormalities in all cases.
GABABR abs were found in serum samples of all patients. In one patient, titration was not possible due to lack of material (5). Median GABABR titre in serum was 1:1000 (range 1:100–1:50 000). Interestingly, patients with a tumour had titre values that were about 50-fold (median value 26 000 (500–50 000)) higher than in patients without a tumour (median value 1000 (100–2000)).
GABABR abs were found in CSF samples of six patients with a median titre of 1:250 ranging from 1:32 to 1:1000. As far as data were available, median titre of abs in patients with a tumour was 1:375 (1:250–1:1000) and without a tumour 1:225 (1:32–1:750). In three patients we could not determine GABABR titres in CSF as we did not receive this material. In these three cases diagnostic results were exclusively based on serum tests (table 1).
Intrathecal GABABR ab synthesis was found in five of the six patients (1, 3, 6, 8, 9 and 10) with sufficient data available. Median disease duration at determination of the ab-index was 8.5 days (range 4–33). Median-specific index for GABABR abs was 76.8, ranging 57.9–100.
Nine of the 10 patients had MRI lesions that were thought to be related to encephalitis. At the earliest available study (median 4 days, range 1–129 days after manifestation), we saw mediotemporal fluid-attenuated inversion recovery (FLAIR)/T2 volume and signal increase (three bilateral, six unilateral, figure 2). Hippocampi (in part, the amygdalae, too) initially exhibited features of acute oedema.
We had follow-up examinations of nine patients. There was increasing atrophy and signal increase on T2/FLAIR-weighted images in five of them accounting for evolution of mediotemporal sclerosis. Interestingly, on the outer aspects of the hippocampus on the border to the temporal horn of the lateral ventricle, T1-weighted not-enhanced images convey the impression of laminar necrosis that is usually seen as an MRI correlate of necrosis of neurons and astrocytes in vulnerable brain regions.16 ,17 This feature was observed in 3/10 patients (4, 6 and 10), 2 weeks to 4 months into the disease (shown for an AMPAR ab case, figure 2C2). Involvement of extramediotemporal areas was found in 2/10 patients, one in other parts of the limbic system, that is, the insula and the fronto-orbital cortex (8, figure 2A), one had diffuse hyperintensity involving the whole temporal lobe (2; not shown).
EEG revealed generalised or focal slowing in six patients, in two cases combined with foci of epileptic activity in temporal lobe or more extended in one hemisphere. Epileptic activity without slowing occurred in two patients (1 and 6; table 2).
In three serum or CSF samples we found abs to AMPAR. All reacted with the subunit 2. Looking at the age, there were no significant differences among the patients. The median age was 61 years (range 60–62), one of the patients was a woman. This patient was the only one with a tumour.
The most prominent symptoms were short-term memory deficits. In all three patients, a two-stage progression of disease could be observed. Two patients (12 and 13) showed initially mild, partially transient and later on prolonged, more prominent symptoms.
The only female patient (13) presented Hashimoto thyroiditis and showed ovarian cancer in the diagnostic work-up. After resection of the tumour, memory worsened and a psychosyndrome combined with anxiety and thoughts of suicide manifested itself. Immunotherapy and chemotherapy were applied. After 14 months memory deficits and the psychosyndrome regressed. The patient was able to live independently.
The other two patients (11, 12) did not show a concomitant autoimmune disorder, a tumour in the diagnostic work-up or additional autoimmune-abs. Patient 12 presented a multiphase course of disease with fluctuation of his psychosyndrome and mood disorder. He was treated with corticosteroids and rituximab without significant effect. After a period of 6 months follow-up symptoms persisted without any fluctuations.
Patient 11 showed complete resolution of his mnestic difficulties after treatment with corticosteroids and azathioprine as immunosuppressant agents. Interestingly, this patient had a prodromal episode consisting of nausea and abdominal pain 10 days before the beginning of neurological disease (ie, a potentially triggering infection).
In all three patients CSF samples were collected after a median time of 4 days, ranging 0–9 days. One of the three patients (11) had lymphocytic pleocytosis (16 cells/µL), none had intrathecal IgG synthesis.
Owing to the missing data regarding CSF values we could only ascertain specific AMPAR ab-index in one patient (11). It did not indicate intrathecal AMPAR ab synthesis.
All patients showed exclusively MRI changes of the mediotemporal lobes. From the beginning, two patients showed bilateral abnormalities (figure 2B). Patient 11 had initially left-sided abnormalities which became bilateral over time (not shown).
In the follow-up MRI examinations, hippocampal and amygdalar scleroses (11, 13; no follow-up images for patient 12) manifested themselves. T1-weighted sequences of patient 11 displayed a laminar necrosis-like pattern in the affected hippocampus (figure 2C2). EEG recordings of all three patients were unremarkable (table 2).
GABABR and AMPAR abs are rare causes for LE. GABABR ab-associated LE seems to occur more frequently than AMPAR ab-induced LE. AMPAR-associated symptoms of the CNS seem to have a more favourable outcome with higher tendency of remission as has been described before.5 ,18 We extend the existing knowledge by describing details on CSF–serum studies of samples collected early in the disease course and novel observations on the morphology of hippocampal changes during disease progression.
MRI changes in both groups were similar. We found hippocampal sclerosis in GABABR-positive and AMPAR-positive patients. Interestingly, four patients of both groups (4, 6, 10 and 13) showed laminar necrosis-like pattern in cornu ammonis (CA) 1 and CA2 of the hippocampus. Two of them had no seizures (6, 13). In literature, this pattern has been described as a result of hypoxia, cerebral ischaemia and prolonged seizures or other metabolic disturbances. Evolution of this pattern has not been fully understood yet. Histopathologically, it is characterised by ‘pan-necrosis’, that is, necrosis of neurons, glial cells and blood vessels.16 The existence of this pattern in brain MRI may contribute to the recognition of LE. Of note, this pattern has been documented but not explicitly described in LE with other abs.19–21
In this series, abs could be found in both materials of patients with CSF–serum pairs studied and in the serum samples when only these were available. Some authors made similar observations.5 ,22 Others, however, described patients with abs in CSF only.2 ,3 ,23 To the best of our knowledge, no data on patients with serum-only GABABR or AMPAR abs have been published.
CSF data available to us revealed that GABABR-induced LE is associated with intrathecal synthesis of specific abs in all patients without a tumour independent of the latency between onset of symptoms and lumbar puncture. It is conceivable that the ab-producing plasma cells cross the BBB at an early stage when the latter is leaky. Breach of integrity of the BBB due to local inflammation and due to different modulators has been described before.24
One patient with a tumour (3) showed oligoclonal bands and intrathecal synthesis of GABABR abs 9 days after onset of neurological symptoms, another one (6) had only slight intrathecal production of specific ab but without general intrathecal IgG production 4 days after symptom onset. We would suggest that the initial peripheral, tumour-driven production of abs later on may expand to the brain compartment. Such expansion from peripheral ab synthesis towards production of abs within the CSF within ≈3 weeks of clinical disease onset has been suggested in anti-NMDAR encephalitis.25 We assume that patient 6 is in a transitional stage to more pronounced intrathecal ab synthesis. Understanding temporospatial dynamics of ab production in CSF or in the periphery may help making rational therapy choices. In contrast to plasma exchange and intravenous immunoglobulins with their primarily peripheral effect, cyclophosphamide and rituximab have been suggested to exert an inhibitory effect on the intrathecal production of abs.26 ,27
Previous publications expanded the clinical picture of GABABR abs to cerebellar ataxia, opsoclonus-myoclonus syndrome and progressive encephalomyelopathy.3 ,4 ,8 We found some of these symptoms in parallel with LE but not in isolation. Consistent with previous findings, diagnosis of a tumour is more likely in older patients with GABABR. In our series, 30% died as compared with 25–43% in previous series.2 ,3 ,8 Furthermore, four of our 10 patients had additional onconeural abs (with no clinical or paraclinical features distinguishing them from the ‘GABABR ab-only’ patients). In previous reports, the proportion of additional onconeural abs had been in the range 7–35% only.2 ,3 ,8 Two previous reports found tumours in all patients with additional onconeural abs.2 ,3 In contrast to that, we found two patients ( 8, 10) with an additional onconeual ab (Sox-1) without evidence of a tumour; the follow-up intervals, however, were limited (2 and 3 months, respectively). Two of our patients with onconeural abs (6 and 10, both Sox-1) improved with immunotherapy (and chemotherapy in case of 6). An improvement of a patient with onconeural ab was not seen in one group,2 ,3 but in another.8 All our paraneoplastic GABABR cases were associated with SCLC. Breast carcinoma, rectum carcinoma, multiple myeloma, carcinoid of the thymus and melanoma have been described as rarer tumour types in GABABR-associated LE.8 AMPAR ab-associated LE is characterised by relapses that show good response to immunotherapy similar to the first episode.5 We could confirm these observations. All of our three patients had at least a two-stage course of disease. Two of them showed improvement of symptoms during immunotherapy, in one case combined with chemotherapy. We identified two male and one female patient with AMPAR abs. In the literature, more women than men have been described so far.5 ,18 ,23 We found only one female patient with a tumour (ovarian cancer). Previous reports had shown higher tumour rates, up to 70% with a broad spectrum of neoplasms (lung, breast and thymus).5 ,23 Deaths did not occur in our series in contrast to previous reports.5 None of our patients had onconeural abs in contrast to the findings of previous reports.5
A limitation of this study is that the patients studied at our laboratory are biased towards encephalitides, encephalopathies and epilepsies; peripheral or cerebellar syndromes are rare. This may lead to an ascertainment bias of the frequency of GABABR and AMPAR abs in those disorders. Concerning the methodology, we did not have facilities to use unfixed cells or cultures to confirm binding of the abs to the surfaces of living cells.
In conclusion, patients with suspected LE should undergo broad ab testing including GABABR and AMPAR abs, ideally in CSF and serum to maximise sensitivity. Recognition of these abs should prompt tumour search. GABABR or AMPAR ab-positive patients should undergo immunotherapy even though GABABR abs may be associated with an unfavourable outcome. CSF findings may become more important as we gain more insights into the dynamics underlying the temporospatial processes in autoimmune encephalitides.
The authors would like to thank Professor Theodor May for his support with the preparation of the raw data and Mrs Jutta Potthoff and Mrs Dagmar Bonhaus-Liebmann for excellent technical assistance.
Contributors The study was conceptualised and initiated by MDO and CGB. Antibody determination was done by CGB, CK and MDO who also searched the medical records and compiled the data. MRI were viewed and interpreted by FGW. Patients were diagnosed and treated by DD, CAH, HJH, SH, SI, HL, RM, SP, WRS, KS, MS, FS, ET, TJvO and HW. MDO wrote the manuscript draft. It was reviewed and refined by all authors for important intellectual content.
Competing interests MDO received a travel grant from Fresenius Medical Care (Bad Homburg, Germany). Her employer (Krankenhaus Mara, Bielefeld, Germany) runs a laboratory for the detection of autoantibodies including those described in this study; external senders are charged for antibody diagnostics. CK received a congress travel grant from Eisai (Frankfurt, Germany). Her employer (Krankenhaus Mara, Bielefeld, Germany) runs a laboratory for the detection of autoantibodies including those described in this study; external senders are charged for antibody diagnostics. FS received travel grants from Biogen, Novartis, Bayer (Germany), obtained honoraria for talks from Boehringer, Serono, Biogen, Genzyme and gave scientific advice to Genzyme and Boehringer. HW receives compensation for serving on Scientific Advisory Boards for Bayer Healthcare, Biogen Idec, Genzyme, Merck Serono, Novartis and Sanofi Aventis. He receives speaker honoraria and travel support from Bayer Vital GmbH, Bayer Schering AG, Biogen Idec, CSL Behring, Fresenius Medical Care, Genzyme, Glaxo Smith Kline, GW Pharmaceuticals, Merck Serono and Novartis, Sanofi Aventis and compensation as a consultant from Biogen Idec, Merck Serono, Novartis and Sanofi Aventis. He receives research support from Bayer Vital, Biogen Idec, Merck Serono, Novartis, Sanofi Aventis Germany, Sanofi US. CGB gave scientific advice to Eisai (Frankfurt, Germany) and UCB (Monheim, Germany), undertook industry-funded travel with support of Eisai (Frankfurt, Germany), UCB (Monheim, Germany), Desitin (Hamburg, Germany) and Grifols (Frankfurt, Germany), obtained honoraria for speaking engagements from Eisai (Frankfurt, Germany), UCB (Monheim, Germany), Desitin (Hamburg, Germany), diamed (Köln, Germany), Fresenius Medical Care (Bad Homburg, Germany), and received research support from Astellas Pharma (München, Germany), Octapharma (Langenfeld, Germany), diamed (Köln, Germany) and Fresenius Medical Care (Bad Homburg, Germany). His employer (Krankenhaus Mara, Bielefeld, Germany) runs a laboratory for the detection of auto-antibodies including those described in this study; external senders are charged for antibody diagnostics. He received research support from diamed (Köln, Germany), Fresenius Medical Care (Bad Homburg, Germany), Astellas Pharma (München, Germany) and Octapharma (Langenfeld, Germany).
Provenance and peer review Not commissioned; externally peer reviewed.