Clinical features, prognostic factors, and antibody effects in anti-mGluR1 encephalitis

Abstract

Objective

To clinically characterize patients with anti-metabotropic glutamate receptor (mGluR) 1 encephalitis, to identify prognostic factors, and to study the immunoglobulin G (IgG) subclasses and effects of antibodies on neuronal mGluR1 clusters.

Methods

Clinical information on new and previously reported patients was reviewed. Antibodies to mGluR1 and IgG subclasses were determined with brain immunohistochemistry and cell-based assays, and their effects on mGluR1 clusters were studied on rat hippocampal neurons.

Results

Eleven new patients were identified (10 adults, 1 child);4 were female. In these and 19 previously reported cases (n = 30, median age 55 years), the main clinical manifestation was a subacute cerebellar syndrome that in 25 (86%) patients was associated with behavioral/cognitive changes or other neurologic symptoms. A tumor was found in 3 of 26 (11%). Brain MRI was abnormal in 7 of 19 (37%) at onset and showed cerebellar atrophy in 10 of 12 (83%) at follow-up. Twenty-five of 30 (83%) patients received immunotherapy. Follow-up was available for 25: 13 (52%) had clinical stabilization; 10 (40%) showed significant improvement; and 2 died. At the peak of the disease, patients with bad outcome at 2 years (modified Rankin Scale score > 2, n = 7) were more likely to have higher degree of initial disability, as reflected by a worse Scale for Assessment and Rating of Ataxia score, and more frequent need of assistance to walk. Antibodies to mGluR1 were mainly IgG1 and caused a significant decrease of mGluR1 clusters in cultured neurons.

Conclusions

Anti-mGluR1 encephalitis manifests as a severe cerebellar syndrome, often resulting in long-term disability and cerebellar atrophy. The antibodies are pathogenic and cause significant decrease of mGluR1 clusters in cultured neurons.

Metabotropic glutamate receptors (mGluRs) are G-protein–coupled glutamate receptors that mediate excitatory neurotransmission in the CNS and peripheral nervous system. They are involved in a variety of functions such as memory, learning, anxiety, and pain perception.¹ There are 8 different types of mGluRs (mGluR₁–mGluR₈) divided into 3 groups, which share similar mechanisms of action and synaptic localization (presynaptic or postsynaptic).^1,2 mGluR1 and mGluR5 belong to group 1; they are localized mainly postsynaptically, and their activation results in potentiation of NMDA receptor activity and excitotoxicity.^2,3 We recently reported the clinical features and outcome of patients with anti-mGluR5 encephalitis and suggested a pathogenic role of these autoantibodies, which cause a significant decrease of mGluR5 from the surface of cultured live neurons⁴ and loss of memory and increased anxiety in a mouse model.⁵ Similarly, a few single case reports and small case series have investigated the main clinical syndrome associated with anti-mGluR1 encephalitis.^6–15 These studies suggested that, despite a common clinical presentation as cerebellar ataxia, disease progression and outcome can be variable and difficult to predict on an individual basis. In particular, it is unclear why some patients respond to immunotherapy and return to their baseline functional status,^7,13 whereas others show no response to treatment and are left with severe cerebellar symptoms and long-term neurologic dysfunction.^7,8,14,15 Our study aimed to characterize the long-term outcome of patients with anti-mGluR1 encephalitis and to identify predictive factors of response to immunotherapy. Moreover, similar to what has been reported for mGluR5 autoantibodies,^4,5 there is evidence that mGluR1 antibodies may be pathogenic; they have been shown to alter Purkinje cells function in cerebellar slices⁶ and to cause motor incoordination when injected in the cisterna magna of mice.⁷ However, the mechanisms by which these antibodies alter neuronal function are unknown. Therefore, here we also explored whether patients' antibodies alter the surface density of mGluR1 in cultured neurons.

Methods

Identification of patients, sample collection, and clinical information

We investigated sera and/or CSF samples sent to our laboratory for antibody studies in patients with suspected autoimmune neurologic disorders. All samples were screened by immunohistochemistry for reactivity against neuropil of rat brain and cell-based assays (CBAs) for NMDA receptor (NMDAR), as reported.^16,17 Samples showing neuropil reactivity different from that of NMDAR antibodies were then investigated with CBAs for antibodies against mGluR1 and other neuronal targets (mGluR5,⁴ AMPA receptor,¹⁷ GABA_B receptor,¹⁸ GABA_A receptor,¹⁹ LGI1,²⁰ CASPR2,²⁰ DPPX,²¹ GlyR,²² IgLON5,²³ dopamine2 receptor,²⁴ and neurexin-3α²⁵). Moreover, all samples were screened for onconeuronal and GAD65 antibodies using previously described techniques.^22,26,27

Clinical information was obtained retrospectively from medical records or from structured questionnaires completed by the referring physician after antibody results were known. This included prodromal symptoms, neurologic manifestations, autoimmune and tumor comorbid conditions, results from CSF analysis, EEG and brain MRI, type of immunotherapy and cancer treatment, and outcome at last follow-up. Symptom severity was measured with the modified Rankin Scale (mRS)²⁸ and the Scale for the Assessment and Rating of Ataxia (SARA)²⁹ at the peak of disease and at last follow-up. One patient has been published in a short video.⁸ For this patient, we obtained additional clinical information about the onset and peak of the disease, as well as long-term outcome that was not available for the initial report.

Review of previously reported cases with mGluR1 antibodies and outcome study

To analyze the neurologic symptoms, tumor association, CSF, EEG and MRI features, treatment response, and outcome of patients with anti-mGluR1 encephalitis, we reviewed the current data along with all previously reported cases with mGluR1 antibodies^7–15 for whom the following information was available (as published data or obtained as additional/updated information provided by the authors or the referring physician after publication): (1) symptoms at onset and severity at peak of disease, (2) CSF and/or MRI results, and (3) clinical outcome at last follow-up, including mRS score, with a minimum follow-up of 6 months from encephalitis onset. Patients for whom a formal mRS evaluation was not reported were included only if the clinical information at last follow-up allowed the authors (M.S., F.G., and J.D.) to confidently evaluate the mRS score. For example, a patient reported to have persistent severe cerebellar ataxia and who was unable to walk unassisted⁷ was assigned an mRS score of 4.

Patients were considered to have a good outcome if the mRS score at the last follow-up was ≤2 and to have a bad outcome if the mRS score was >2.

mGluR1 antibody detection, CBA, and determination of immunoglobulin G subclasses

Samples were suspected to harbor mGluR1 antibodies on the basis of the characteristic neuropil and cerebellum staining in rat brain immunohistochemistry, as previously described,¹⁴ and were confirmed by CBA. Human embryonic kidney cells were transfected with green fluorescent protein–tagged mGluR1 cDNA plasmid (RG214400, Origene, Herford, Germany) and, 24 hours after transfection, were incubated with patients' serum (1:40) or CSF (1:5) for 1 hour at 37°C. The cells were then washed and fixed with 4% paraformaldehyde for 5 minutes, permeabilized with 0.3% Triton X-100 for 5 minutes, and incubated with secondary goat anti-human immunoglobulin G (IgG) Alexa Fluor 594 (1:1,000, A11014, Invitrogen, Carlsbad, CA) or, for determination of IgG subclasses, with specific biotinylated mouse anti-human IgG1 (B6775, Sigma-Aldrich, St. Louis, MO, or MCA4774, Bio-Rad, Hercules, CA), IgG2 (B3398, Sigma-Aldrich or 555873, BD Bioscience, East Rutherford, NJ), IgG3 (B3523, Sigma-Aldrich, or 5247-9850, Bio-Rad), and IgG4 (B3648, Sigma-Aldrich, or 555881, BD Bioscience) (all 1:1,000), followed by incubation with rhodamine avidin (1:1,000, A-2005, Vector Laboratories, Burlingame, CA). Reactivity was determined with a fluorescence microscope and Zeiss Axiovision software, as reported.^16,17 Intensity of reactivity was assessed by 2 investigators (M.S. and J.D. for all CBAs; J.H. and S.M.-C. for IgG subclass only) independently and classified as mild (+), moderate (++), or strong (+++) (table 1).

Table 1.

Clinical features, antibody results, treatment and outcome of 11 patients with anti-mGluR1 encephalitis

Cultured hippocampal neurons and study of mGluR1 antibodies effects

Rat hippocampal neurons were prepared from E18 embryos and cultured for 15 days, as reported.^16,17,30 To confirm expression of mGluR1, live neurons were fixed with 4% paraformaldehyde for 5 minutes, incubated with a polyclonal rabbit anti-mGluR1 antibody (1:100, BML-SA610-0050, Enzo Life Sciences, Farmingdale, NY) for 1 hour followed by goat anti-rabbit IgG Alexa Fluor 594 (1:1,000, A11012, Invitrogen) for 1 hour, and then mounted with ProlongGold (P36931, Thermo Fisher, Waltham, MA) and examined with a confocal microscope (LSM710, Carl Zeiss, Jena, Germany).

To determine the effect of the patient's antibodies on mGluR1 expressed by neurons, we used the CSF from a representative patient (patient 6) that showed a typical pattern of reactivity with rat brain tissue, strong reactivity with live neurons, lack of concurrent antibodies, and sufficient amount of sample available. Live neurons were incubated for 48 hours at 37°C with CSF (dilution 1:10) from patient 6 or with control CSF, washed with fresh media, and incubated with anti-mGluR1 from a positive CSF (1:20, used here as mGluR1 biomarker) for 1 hour at 4°C (low temperature and short incubation time were chosen to reduce the chance of mGluR1 internalization during staining), followed by washing and incubation with a secondary goat anti-human IgG Alexa Fluor 488 (1:1,000, A11013, Invitrogen). Neurons were then fixed with 4% paraformaldehyde for 5 minutes, permeabilized with 0.3% Triton X-100 for 5 minutes, blocked with 1% bovine serum albumin, and sequentially incubated with a polyclonal rabbit anti-PSD95 antibody (1:400, AB18258, Abcam, Cambridge, UK) for 1 hour at room temperature, followed by goat anti-rabbit IgG Alexa Fluor 594 (1:1,000, A11012, Invitrogen) for 1 hour. Coverslips were mounted with ProlongGold (P36931, Thermo Fisher), and images were taken under a confocal microscope (LSM710, Carl Zeiss). Images were deconvoluted with Huygens Essential software (Scientific Volume Imaging, Hilversumm, the Netherlands), and cell-surface mGluR1 and intracellular PSD95 clusters were analyzed with Imaris suite 8.2 (Bitplane AG, Zürich, Switzerland), as reported.³¹

Statistical analyses

Descriptive statistics of demographics and clinical features and inference statistics about outcome and molecular studies on cultured neurons were performed with GraphPad Prism (version 8.2.0 [272], 2019), Excel (version 16.35), and R (version 3.6.2 [2019-12-12]). To identify prognostic factors, we compared clinically meaningful variables (such as age, sex, presence of a tumor, CSF analysis, MRI findings, time from onset to peak, severity scales at peak of disease, immunotherapy treatment, and others) in patients with bad vs good outcome, estimated the difference between the 2 groups using medians (calculated as Hodges-Lehmann estimate) and odds ratios, and reported the corresponding 95% confidence intervals (CIs) (using the Hodges-Lehmann method for medians).^32,33 We estimated the correlation between continuous variables (such as time from onset to peak and from onset to start of immunotherapy) using Spearman correlation coefficients. Analysis of the effects of antibodies on neurons was performed with the Mann-Whitney U test (2 sided). Results <0.05 were regarded as statistically significant. Given the exploratory nature of the study and the small sample size, we did not perform a multivariate regression analysis.

Standard protocol approvals, registrations, and patient consents

This study was approved by the Institutional Review Board of the Hospital Clinic (Barcelona, Spain). All patients gave written informed consent for use of samples and clinical information.

Data availability

Any data not published within the article are available and will be shared anonymized by request from any qualified investigator.

Results

Case description: A child with anti-mGluR1 encephalitis

Two months after recovering from a streptococcal pharyngitis, a 6-year-old boy (table 1, patient 11) presented with fever, headache, nausea, and vomiting, followed 2 days later by a cerebellar syndrome that worsened over 1 week. He had a broad-based gait allowing only 6 to 8 steps unassisted, severe truncal ataxia making it impossible to sit unassisted, dysarthria, dysmetria, tremor, involuntary choreiform movements of the face, and small, uncoordinated jerky movements of the fingers. His SARA score was 21 of 40 and mRS score was 4. Tumor screening was negative. CSF analysis showed pleocytosis (125 white blood cells [WBC]/mm³), normal proteins, CSF-specific oligoclonal bands, and increased IgG index. Infectious workup was negative, including CSF PCR for herpes simplex virus 1 and 2, cytomegalovirus, Epstein-Barr virus, and vesicular stomatitis virus. Brain MRI was normal at onset and showed mild cerebellar edema on day 12; magnetic resonance spectroscopy showed decreased n-acetylacetate asparte/creatine ratio and increased lactate, suggesting inflammatory edema (figure 1A). Antibodies targeting mGluR1 were found in CSF, but not in serum, confirming the diagnosis of anti-mGluR1 encephalitis. Immunotherapy was started 10 days after onset, consisting of IV immunoglobulin and IV methylprednisolone for 2 days and IV cefotaxim for 8 days, followed by oral penicillin V. The child showed a rapid response to treatment and completely recovered within 10 weeks from onset (SARA score 0 of 40; mRS score 0).

Figure 1. MRI of patients with anti-mGluR1 encephalitis.

A) Patient 11. Initial MRI was normal, but on day 12, repeat MRI showed edema of cerebellar sulci predominantly involving the right hemisphere (not shown). Single-voxel magnetic resonance spectroscopy of the right cerebellar lesion (red polygon on upper and lower brain image) shows a decreased n-acetylacetate asparte/creatine (NAA/Cr) ratio, a slightly increased choline/creatine (Cho/Cr) ratio, and elevated lactate peak (downward doublet peak), compatible with inflammation (cerebellitis). (B–D) Brain MRI of patient 3 showing fluid-attenuated inversion recovery hyperintensity of the right cerebellar hemisphere at disease onset (B) that resolved on short-term follow up MRI (not shown) and then showed progressive cerebellar atrophy at 5 weeks (C) and 9 years (D) of follow-up. mGluR1 = metabotropic glutamate receptor 1.

Clinical features, results from ancillary tests, and outcome

We identified 11 patients (10 adults, 1 child) with anti-mGluR1 encephalitis. Detailed clinical features, results from CSF analysis and MRI, antibody results, and outcomes are shown in table 1.

We combined the information of these 11 patients with that of 19 previously published cases from 9 studies^7–15 (total 30 patients). The median age at onset of anti-mGluR1 encephalitis was 55 years (interquartile range [IQR] 43–64 years); 43% were female (table 2). Anti-mGluR1 encephalitis manifested with a cerebellar syndrome (median time from onset to peak 3 months, IQR 0.37–23 months) in 29 of 30 (97%) patients; in 7 (23%) of these patients, the cerebellar symptoms progressed over >3 months (median time from onset to peak 19.5 months, IQR 8–48 months). There was a positive correlation between time from onset to peak and delay to antibody test (Spearman r = 0.85, p = 0.0001), indicating a more rapid diagnosis in patients with acute presentation. Overall, symptoms included gait and trunk instability (24 of 24), cerebellar dysarthria (18 of 22, 81%), oculomotor abnormalities (16 of 18, 89%), and limb ataxia (17 of 19, 89%) (video 1). Prodromal symptoms such as headache, weight loss, fatigue, nausea or flu-like syndrome occurred in 7 of 17 (41%) patients and preceded neurologic symptoms by a median of 30 days. The first neurologic manifestation was cerebellar ataxia in 17 of 25 (68%) patients; the remaining patients had extracerebellar symptoms at onset. During the course of the disease, the cerebellar syndrome remained the main clinical manifestation, which was isolated in 4 of 29 (14%) patients, whereas in the other 25 of 29 (86%) patients, it was accompanied by behavioral changes (6 of 25, 24%), cognitive deficits (11 of 25, 44%), or other symptoms (dysgeusia, dysphagia, dysautonomia, seizures, sleep disorders, and/or movement disorders). Behavioral changes ranged from irritability, apathy, and mood and personality changes to frank psychosis with hallucinations and catatonia. Cognitive deficits included memory problems, executive dysfunction, and spatial orientation deficits. Movement disorders occurred in 5 patients and were characterized by myoclonus or dystonia in adults, whereas the only identified child (patient 11) showed facial and appendicular choreoathetosis. Seizures were uncommon and occurred in 2 patients: 1 patient (10) had seizures associated with cognitive and behavioral changes but never developed a cerebellar syndrome; the other patient (6) developed partial complex seizures 24 months after cerebellar symptom onset. Other infrequent clinical manifestations included visual loss and motor weakness.

Table 2.

Clinical features of 30 patients^a with anti-mGluR1 encephalitis

Video 1

(Patient 1). This patient developed a severe cerebellar syndrome in February 2015 that persisted until the last follow-up 4 years later (April 2019) despite aggressive immunotherapy. The video shows severe appendicular ataxia and cerebellar tremor of right > left arm, truncal and neck hypotonia with persistent/continuous head titubation, inability to stand without support, and severe gait instability requiring bilateral help.Download Supplementary Video 1 ^{(91.1MB, mp4)} via http://dx.doi.org/10.1212/010854_Video_1

Encephalitis was considered paraneoplastic (tumor developing <5 years before or after encephalitis onset) in 3 of 26 (11%) cases: 1 patient had cutaneous T lymphoma and prostate adenocarcinoma 36 months before and 18 months after encephalitis onset, respectively, and 2 patients had Hodgkin lymphoma, one 18 months and the other 24 months before encephalitis onset.

CSF analysis was abnormal in 19 of 25 (76%) patients, including pleocytosis in 11 of 25 (44%, median 27 WBC/mm³, IQR 8–125 WBC/mm³) and oligoclonal bands or increased IgG index in 11 of 20 (55%).

EEG was usually not obtained; however, it was abnormal in 5 of 8 (62%) patients, showing focal bilateral frontal or temporal slowing, associated with interictal epileptiform discharges in 1 patient with clinical seizures.

Information on brain MRI at onset was available from 19 patients. The first MRI studies were obtained within a median of 30 days from disease onset and were abnormal in 7 (37%), showing cerebellar T2/fluid-attenuated inversion recovery (FLAIR) hyperintensities (figure 1B) or leptomeningeal gadolinium enhancement in 3 and nonspecific small ischemic-like subcortical lesions in the other 4. Antibody studies were performed after a median of 9 months (IQR 1–25 months) from onset and were done in serum in all 30 patients and in CSF in 19. mGluR1 antibodies were positive in 29 of 30 (97%) serum samples and in 18 of 19 (95%) CSF samples. Matched samples were available from 19 patients: 17 were positive in both serum and CSF; 1 was positive only in serum¹¹; and 1 was positive only in CSF (patient 11).

Twenty-five of 30 (83%) patients received immunotherapy after a median of 77 days (IQR 11–632 days) from onset: 24 of 25 (96%) received first-line immunotherapy including IV methylprednisolone, oral prednisone, plasma exchange, or IV immunoglobulins; 14 of 24 (58%) patients received >2 of these treatments. In addition, 14 of 25 (56%) patients received second-line immunotherapy including one or more of the following: cyclophosphamide, rituximab, mycophenolate mofetil, azathioprine, or tacrolimus. Time from symptom onset to start of immunotherapy had a positive correlation with time from onset to antibody test, reflecting time to diagnosis (Spearman r = 0.79, p = 0.001). Only 5 of 12 patients were empirically started on immunotherapy before antibody results (4 of them had an acute presentation of symptoms); the remaining 7 patients received immunotherapy after antibody results (5 of them had a subacute presentation).

At last follow-up (median 24 months), 12 of 16 (75%) patients had abnormal MRI, showing cerebellar atrophy in 10 (figure 1, C and D), which was predominantly in the vermis in 8 of them. Overall, 13 of 25 (52%) had clinical stabilization or mild improvement, whereas 10 of 25 (40%) had significant improvement or complete resolution of symptoms. Two patients died. One patients¹⁴ initially improved after 6 cycles of monthly corticosteroid treatment, stabilized for several years, and then died (cause of death unknown) 4.5 years after onset. The other patient (patient 4) showed progressive worsening of ataxia over several years and died of sepsis and multiorgan failure 14 years after disease onset. Autopsy was not available for either patient.

Relapses were reported in 6 patients, always in the context of immunotherapy discontinuation. Symptoms reappeared between 3 weeks and 3 months after treatment was discontinued, and improved or resolved after immunotherapy was resumed. Only 1 patient⁹ had multiple relapses, some of them occurring after 12 to 28 months of symptom remission. This patient received several cycles of first-line immunotherapy and as second-line immunotherapy only received 1 dose of rituximab and then tacrolimus and azathioprine as maintenance therapy, which might have resulted in less effective and durable B-cell depletion. No information about the presence/absence of disease recurrence was available for the remaining 24 patients.

Outcome study

Detailed outcome information was available from 19 patients (table 3). Compared to patients with good outcome (mRS score ≤2, n = 12), those with bad outcome (mRS score >2, n = 7) had higher neurologic disability at the peak of the disease, as reflected by higher SARA score (median 29 vs 17), and were more likely to need assistance to walk (100% vs 33%). There were no significant differences between the 2 groups in terms of age, sex, clinical presentation, tumor association, brain abnormalities at initial MRI, CSF analysis, immunotherapy, or delay from onset to treatment.

Table 3.

Comparison of patients according to good or bad outcome

mGluR1 antibody subclasses and effects on cultured neurons

Serum or CSF samples were available for antibody subclass studies from 11 patients (8 from the current study, 3 from previously published reports,^13–15 including 4 paired serum and CSF samples, 1 CSF alone, and 6 serum alone). The main IgG subclass of mGluR1 antibodies was IgG1, which was found in all patients, either alone (n = 5) or associated with IgG3 (n = 3) or IgG2, IgG3, and IgG4 (n = 1) (figure 2). In 2 patients, the IgG subclass remained undetermined (table 1). All patient samples were negative for onconeuronal antibodies.

Figure 2. mGluR1 IgG subtypes in a representative patient.

(A) Reactivity of a representative patient's (patient 6) serum with human embryonic kidney cells expressing metabotropic glutamate receptor 1 (mGluR1) and immunoglobulin G (IgG) subclasses. Scale bar = 10 μm. (B) Venn diagram of the distribution of IgG subclasses in 9 patients with anti-mGluR1 encephalitis. All patients had IgG1 antibodies, which either were the only subclass (n = 5) or were accompanied by IgG3 alone (n = 3) or by IgG2, IgG3, and IgG4 (n = 1).

Compared with control CSF, incubation of neuronal cultures with patient CSF for 48 hours caused a significant decrease of both total and synaptic cell-surface mGluR1 clusters without affecting the density of clusters of PSD95 (figure 3).

Figure 3. mGluR1antibodies cause reduction of mGluR1 clusters in cultured neurons.

(A) Decrease of the density of total and synaptic cell-surface metabotropic glutamate receptor 1 (mGluR1) clusters (green) in neurons treated for 48 hours with patient's CSF compared with neurons treated with control CSF, from representative confocal images (40 dendrites per condition). There was no change in the density of PSD95 clusters (red) in neurons treated with patient or control CSF. (B–D) Quantification analysis of total (B) and synaptic (C) mGluR1 clusters and PSD95 (D). Scale bar = 10 μm. Statistical analyses by Mann-Whitney U test; median, interquartile range, and minimum/maximum values are plotted; ****p < 0.0001.

Discussion

This study confirms and expands the clinical features of anti-mGluR1 encephalitis, identifies predictors of bad outcome, and provides further evidence that patients' antibodies are pathogenic.

Anti-mGluR1 encephalitis occurs mainly in middle-aged adults, although we now report that it can rarely occur in children (patient 11). Patients manifest with a cerebellar syndrome that usually does not remain isolated but is accompanied by other neurologic symptoms, in particular behavioral changes and cognitive impairment. The child described here also developed prominent orofacial and appendicular choreoathetosis that was not observed in any of the adults. The fact that the spectrum of symptoms in children is different from that of the adults has been observed in other antibody-mediated autoimmune encephalitis. For example, children with anti-GABA_A receptor encephalitis or autoimmune encephalitis after herpes simplex encephalitis are more likely to have seizures and movements disorders, in particular choreoathetosis, whereas adults more frequently show cognitive and behavioral changes.^19,34

The onset of anti-mGluR1 encephalitis is usually subacute over several weeks (median time from onset to peak of symptoms 3 months). However, some patients have an acute presentation, over 2 to 7 days, as shown here in 2 new patients (patients 2 and 11) and a previously reported case.¹⁵ These 3 patients showed significant higher CSF pleocytosis (214, 125, and 190 WBC/mm³) compared to those with subacute onset (median 190 vs 9 WBC, 95% CI of difference in medians 98–207) and had FLAIR/T2 cerebellar hyperintensities or leptomeningeal gadolinium enhancement on brain MRI performed within the first 15 days from onset. However, in most of the other patients, the CSF and MRI findings were usually normal, as occurs in some patients with other types of autoimmune encephalitis (e.g., NMDAR, LGI1).^16,35 Unlike anti-mGluR5 encephalitis, which is associated with cancer in more than half of the patients,⁴ anti-mGluR1 encephalitis is rarely paraneoplastic (<15% of the cases).

The cerebellar syndrome at the peak of the disease is usually severe (video 1); 11 of 19 (58%) patients required a walker or wheelchair. In our study, the main prognostic factor for bad outcome was the severity of neurologic disability at the peak of the disease and inability to walk unassisted. SARA scores showed a good correlation with mRS scores (Spearman r = 0.85 p < 0.001, data not shown) and thus might represent a useful clinical tool to assess and monitor cerebellar symptoms in patients with anti-mGluR1 encephalitis. Further studies are needed to determine whether patients with higher disability at the peak of the disease might benefit from a more aggressive or prolonged immunotherapy and whether this strategy might improve long-term outcome.

Overall, only 40% of patients showed significant clinical improvement after immunotherapy, and ≈60% had an mRS score ≤2 at the last follow-up. Relapses were observed in a few patients who discontinued immunotherapy early (<3 months), but clinical symptoms improved or resolved after immunotherapy was resumed. This outcome at last follow-up is worse than that observed in anti-mGluR5 encephalitis and other more frequent autoimmune encephalitis such as anti-NMDAR or anti-LGI1 encephalitis, in which ≈80% of the patients have a favorable outcome.^4,36,37 The reasons for this worse outcome are unclear but may be related to a delayed diagnosis and treatment or to irreversible neuronal damage. Longer time to diagnosis and to the start of immunotherapy correlated with longer time from symptom onset to peak of disease, suggesting that patients with subacute presentation are more likely to receive a delayed diagnosis and treatment. Although our study was unable to detect significant differences in treatment delay between patients with good vs bad outcome (median 71 vs 224 days, respectively; 95% CI of difference in medians −647 to 357), likely due to the small sample size, the overall time from symptom onset to start of immunotherapy was longer (median 77 days) than that reported in other autoimmune encephalitis such as anti-NMDAR encephalitis (median 21 days, IQR 14–46 days).³⁶ On the other hand, a less favorable outcome might be due to irreversible cerebellar damage, as suggested by the loss of Purkinje cells observed in the autopsy of a patient with mGluR1 antibodies.⁶ In other cerebellar ataxias associated with antibodies against neuronal surface antigens such as voltage-gated calcium channels, the overall clinical outcome is worse than that of encephalitis associated with neuronal cell-surface antibodies such as LGI1 or NMDAR.^35–38 For example, although in anti-NMDAR encephalitis the occurrence of cerebellar symptoms is infrequent, patients who develop cerebellar ataxia usually have progressive cerebellar atrophy. The brain atrophy observed in some patients with anti-NMDAR encephalitis is often reversible, but the cerebellar atrophy is irreversible.³⁹

Previous studies have shown that mGluR1 antibodies can block glutamate-mediated cellular functions in nonneuronal cells expressing mGluR1⁷ and can prevent induction of long-term depression, which is a neuronal mechanism involved in synaptic homeostasis, in cultured rodent Purkinje cells.⁶ Moreover, direct infusion of these antibodies in the CSF around the cerebellum caused motor incoordination in mice.⁶ Our study provides insights into the molecular mechanisms involved, showing that the patient's antibodies cause a significant and specific decrease of total and synaptic mGluR1 clusters in cultured neurons without affecting the density of other synaptic proteins such as PSD95.

Our study has some limitations, including the retrospective collection of clinical information, the small sample size, and the fact that we did not address the reversibility of antibody effects on cultured neurons. Future in vitro studies and animal models should address this and investigate the mechanisms that lead to frequent irreversible cerebellar deficits in patients with mGluR1 antibodies. Our current findings, with only 40% of patients showing substantial improvement, suggest that patients with cerebellar ataxia and mGluR1 antibodies should be treated with more aggressive approaches than those used for autoimmune encephalitis, perhaps considering second-line immunotherapies such as rituximab or cyclophosphamide as part of upfront treatments. Only prospective, multicenter studies can address this question in the future.

Glossary

CBA

cell-based assay

CI

confidence interval

FLAIR

fluid-attenuated inversion recovery

IgG

immunoglobulin G

IQR

interquartile range

mGluR

metabotropic glutamate receptor

mRS

modified Rankin Scale

NMDAR

NMDA receptor

SARA

Scale for the Assessment and Rating of Ataxia

WBC

white blood cells

Appendix. Authors

Study funding

This study was supported in part by Instituto Carlos III/FEDER (FIS 17/00234, J.D.) and CIBERER No. CB15/00010, J.D.; NIH RO1NS077851, J.D.; Neuroscience Research Training Scholarship by the American Academy of Neurology, M.S.; Fondation pour la Recherche Médicale (SPF201909009269, France), M.P.P.; Basque Government Doctoral Fellowship Program (PRE_2019_1_0255), E.M.; BETPSY project, a public grant overseen by the French National Research Agency (ANR), as part of the second “Investissements d´Avenir” program (ANR-18-RHUS-0012), J.H.; Jubiläumsfonds der Österreichischen Nationalbank (project 16919), R.H.; Safra Foundation, and Fundació CELLEX, J.D.

Disclosure

C. Kornblum received travel funding and/or speaker honoraria from Sanofi Genzyme, Novartis, Santhera, Fulcrum Therapeutics’ acknowledges financial support as advisory board member and/or primary investigator for Stealth BioTherapeutics, Inc, Amicus Therapeutics, Roche Pharma AG, Fulcrum Therapeutics; and received research support from BMBF, Marigold Foundation Canada, and DGM e.V. C.G. Bien received speaker honoraria from UCB (Monheim, Germany), Desitin (Hamburg, Germany), and Euroimmun (Lübeck, Germany). He receives research support from Deutsche Forschungsgemeinschaft (German Research Council, Bonn, Germany) and Gerd-Altenhof-Stiftung (Deutsches Stiftungs-Zentrum, Essen, Germany). F. Leypoldt receives speaker honoraria from Biogen, Roche, Merck, Fresenius, Teva, Bayer, Novartis, and Alexion and advisory board memberships Roche, Biogen, and Alexion; he works for an academic institution offering commercial antibody testing. M.J. Titulaer receives research grants from Euroimmun AG and Novartis and for consultancy from Guidepoint Global LLC. In addition, M.J. Titulaer has a patent on “Methods for Typing Neurologic Disorders and Cancer, and Devices for Use Therein” pending to Erasmus Medical Center. J. Honnorat receives royalties from Athena Diagnostics, Euroimmun, and RAVO diagnostika for the use of CV2/CRMP5 as an autoantibody test. M.R. Rosenfeld receives royalties from Athena Diagnostics for the use of Ma-2 as an autoantibody test and from Euroimmun for the use of NMDAR as an autoantibody test. F. Graus receives royalties from Euroimmun for the use of IgLON5 as an autoantibody test. J. Dalmau receives royalties from Athena Diagnostics for the use of Ma-2 as an autoantibody test and from Euroimmun for the use of NMDA, GABA_B receptor, GABA_A receptor, DPPX, and IgLON5 as autoantibody tests. The rest of the authors have no disclosures relevant to the manuscript. Go to Neurology.org/N for full disclosures.