Anti-NMDA receptor encephalitis in Japan

In recent years, a severe but often reversible encephalitis of unknown etiology that predominantly affects young women has been increasingly recognized in Japan.¹ The disorder has received several names, including acute diffuse lymphocytic meningoencephalitis,^2,3 acute reversible limbic encephalitis,⁴ acute juvenile female nonherpetic encephalitis,¹ or juvenile acute nonherpetic encephalitis.⁵ No association has yet been made with infections, cancer, or specific autoantibodies, but given that most patients develop a prodromic viral-like illness, a postinfectious immune-mediated etiology has been postulated.¹ We were impressed by the phenotypic similarities between thisdisorder and the recently characterized ovarian teratoma associated encephalitis (OTE).^6-12 This is a treatment-responsive paraneoplastic disorder that occurs in association with antibodies to NR1/NR2 heteromers of the NMDA receptor (NMDAR).^11,12 Therefore, we reasoned that similar antibodies could be associated with juvenile acute nonherpetic encephalitis in Japan. To test this hypothesis, we reassessed four women who developed the disorder 4 to 7 years ago¹³ and since then have had regular clinical follow-up.

METHODS

Since 1999, we have identified at Kitasato University Hospital four young women with a clinical picture¹³ similar to the recently described OTE.^6-12 Dyskinesias were carefully monitored with a digital video recorder with permission of the patients' families.

RESULTS

Unresponsive phase.

On admission, patients were mute, akinetic, and unresponsive to verbal commands while keeping their eyes open, resembling catatonia.¹⁵ Bizarre and inappropriate smiling was noted on admission in two patients. Athetoid dystonic postures (figure 1A), echo phenomenon (mimicking the examiner's movements), and catalepsy-like symptoms (figure e-1A) were also noted. Brainstem reflexes were normal, but patients did not move the eyes spontaneously or close them to a visual threat. Paradoxical findings raised the possibility of a psychogenic reaction or malingering; e.g., some patients were unresponsive to painful stimuli but resisted passive opening of the eyes. However, diffuse delta slowing on EEG was not compatible with a psychogenic reaction or malingering.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Static video clip images and MRI of Patient 3

This patient gradually developed orolingual dyskinesia and athetoid dystonic postures (A), followed by bizarre involuntary movements including, clenching of the teeth (B), disconjugate ocular deviation (B, covered by black bar), rhythmic contractions of upper extremities synchronized with orofacial dyskinesias (B), unsynchronized arrhythmic wiggling of the hands (C), and forceful jaw opening (D). Panel D shows four sequential static video images taken every 0.5 seconds revealing continuous jaw opening movements mimicking jaw-opening dystonia. Video clips were recorded on the 3rd day of the hospital stay (A), the 23rd day (B), the 35th day (C), and the 32nd day (D). Serial fluid-attenuated inversion recovery MRI shows progressive brain atrophy in the mesiotemporal and frontal lobes (E, F, G). Brain MRIs were obtained on the 4th day (E), 48th day (F), and 11th month after admission (G).

Findings at reassessment: NMDAR antibodies and pelvic tumor.

Archived sera and CSF from the time of symptom onset revealed in all patients the presence of antibodies to NR1/NR2B heteromers of the NMDAR (figure 2). In three patients with available serum and CSF, the presence of intrathecal synthesis of antibodies was confirmed (CSF antibody titers measured with HEK293 cells transfected with NR1/NR2B heteromers ranged from 1:500 to 1:2,000). These findings prompted a recall of the patients (year 2007) for serum antibody titers and tumor screening, including MRI of the pelvis. These studies showed that serum antibodies to NR1/NR2 heteromers were not longer detectable, but three of the patients had findings compatible with a teratoma of the ovary¹⁶ (figure 3). In retrospect, one of these patients had a small, 2-cm ovarian cyst detected by CT at the time of symptom presentation that on the current study had grown to 6 cm. Removal of the tumor in the three patients confirmed the diagnosis of mature cystic teratoma, which contained neural tissue in all three cases.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Antibodies to NR1/NR2 heteromers of NMDA receptor

Panels A and C show nonpermeabilized cultures of hippocampal neurons incubated with sera of Patients 1 (A) and 3 (C) obtained at the time of symptom presentation. Panels B and D are similar cultures incubated with sera from the same patients obtained at follow-up 7 years 3 months and 6 years later. Note that sera obtained at the time of symptom presentation show intense immunolabeling of the cell surface of neurons and neuronal processes, whereas sera obtained at follow-up are not reactive. Insets correspond to sera from the same patients incubated with HEK293 cells expressing NR1/NR2B heteromers of the NMDA receptor. Sera from the time of symptom presentation (insets in A and C), but not from follow-up (insets in B and D), react with cells expressing NR1/NR2B heteromers (green cell membrane and cytoplasmic staining). The expression of NR1/NR2B heteromers was confirmed with double labeling with commercially available antibodies to NR1 and NR2B (not shown). Sera from controls (blood donors, lupus, and ovarian teratoma patients without encephalitis) did not show any reactivity with neurons or NR1/NR2B expressing cells (not shown). The methods used for these assays are identical to those previously reported.¹¹ In all studies, sera were used at a dilution of 1:250. Immunofluorescence technique (A through D × 800, oil lens); all insets × 400. Nuclei of neurons and HEK293 cells visualized in blue with 4′,6-diamidino-2-phenylindole (DAPI).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Pelvic MRI of the patients obtained 4 to 7 years after symptom presentation

An ovarian tumor is demonstrated in three patients (Patients 1, 2, and 4) on T1-weighted fat suppression images (FSIs, A), T1-weighted MRI (out-of-phase image, B), and T2-weighted MRI (C, arrows). Note that the T1-weighted hyperintensity in the tumors was suppressed on FSI, indicating the presence of fat, compatible with ovarian teratoma. In Patient 3, the T1-weighted hyperintensity in the left ovary was not suppressed on FSI (arrowhead), suggesting a follicular or corpus luteum hemorrhage; however, the possibility of bleeding from a cystic teratoma cannot be completely ruled out. No calcification was found on pelvic CT (not shown).

DISCUSSION

We report four Japanese women who were initially diagnosed with juvenile acute nonherpetic encephalitis,¹³ a disorder for which an etiology has remained unknown until now. Their clinical pictures had such similarity with the recently reported anti-NMDAR encephalitis in patients with ovarian teratoma¹¹ that it led us to investigate this possible association. Hence, using sera and CSF archived since the time of symptom presentation, up to 7 years ago, we have now confirmed the presence of NMDAR antibodies in all patients. Analysis of the target antigens was performed as previously reported, demonstrating high expression in the hippocampus, detectable with immunohistochemical but not immunoblot techniques, expressed on the cell surface of neuronal cultures, and the identity unambiguously established with immunocytochemistry of HEK293 cells transfected to express functional NR1/NR2B heteromers of the NMDAR.¹¹ In addition, the antibody binding required the presence of NR1 (NR1 dependent), differentiating these antibodies (NR1/NR2B) from the less-specific antibodies reported against linear epitopes of NR2B.¹¹ These findings provide strong evidence of an immune mediated etiology for juvenile acute nonherpetic encephalitis.^1-5 The detection of antibodies to NR1/NR2B also led to the subsequent demonstration of pelvic tumors in three patients, all with pathologically confirmed mature ovarian teratoma containing neural tissue. The experience gained from the close clinical monitoring and long-term follow-up of these cases has important implications.

The clinical features of these patients emphasize the concept that antibodies to NR1/NR2 heteromers of the NMDAR associate with a characteristic syndrome that develops in several predictable stages. The disorder usually presents with a prodromic episode of fever, headache, or malaise, followed a few days later by mood and behavioral changes (sometimes severe short-term memory loss), psychiatric symptoms suggesting schizophrenia or catatonia, decline of level of consciousness, hypoventilation, hyperkinesias, and eventual recovery or death. Compared with previously reported patients,^11,12 whose tumors were removed at early stages of the disease, none of our patients had early tumor removal.

Recently, the “NMDAR hypofunction hypothesis” of schizophrenia has emerged primarily based on observations that NMDAR antagonists exacerbate psychiatric symptoms in schizophrenia and induce schizophrenic symptoms in healthy individuals whereas agents that enhance NMDAR function ameliorate schizophrenic symptoms.^17-22 The psychomimetic effects of NMDAR antagonists have been attributed to the functional blocking of NMDAR in presynaptic γ-aminobutyric acid–mediated (GABAergic) interneurons of the thalamus and frontal cortex, causing a decrease of release of GABA. This results in disinhibition of postsynaptic glutamatergic transmission, excessive release of glutamate in the prefrontal cortex, and glutamate and dopamine dysregulation.²² Although NR1/NR2 heteromers are expressed in the entire nervous system, the main epitope targets of patients with NMDAR autoimmunity are contained in NR1/NR2B heteromers.¹¹ These are preferentially expressed in the adult forebrain, including the prefrontal cortex, hippocampus, amygdala, and hypothalamus, all structures involved in our patients (figure e-3). Therefore, we postulate that NMDAR antibodies may cause inhibition, rather than stimulation, of the NMDAR, contributing to the development of schizophrenia-like symptoms.

Central hypoventilation is an important feature of anti-NMDAR encephalitis, having been reported in 14 of 17 cases (82%), including our 4 patients.^11,12 When we temporarily turned off ventilatory support, spontaneous breathing did not return, but the orofacial-limb dyskinesias persisted, suggesting that the hypoventilation was independent of the dyskinesias. Of interest, NR1 knockout animals die of hypoventilation.²³ Because reactivity of patients' antibodies depends on the presence of NR1 subunits,¹¹ the frequent hypoventilation of patients with this disorder can be explained by the immune response. In fact, recent studies demonstrate that patients' antibodies react not only with NR2B, but also with NR1, and that the reactivity with NR2B largely depends of coexpression of functional NR1/NR2B heteromers (Dalmau et al., unpublished data).

The long-lasting dyskinesias of our patients posed a major clinical problem. They presented gradually as orolingual dyskinesias with athetoid postures, and evolved to intractable bizarre involuntary movements, some of which resembled complex partial seizure status epilepticus or a psychogenic disorder. Extensive monitoring with EEG showed absence of paroxysmal discharges, suggesting that the movements were not epileptic. After unsuccessful trials with multiple antiepileptics and sedatives, the dyskinesias responded to propofol and midazolam. The mechanism of these dyskinesias remains unclear. Although one may consider the possibility of a side effect of the antipsychotic medication, many patients with anti-NMDAR encephalitis (even those who do not receive antipsychotics as our patients) develop dyskinesias, suggesting that they are part of the immune disorder.¹¹ We do not have neuropathologic studies in the present patients, but other Japanese patients with similar encephalitis^2,3 had pathologic findings closely resembling those of patients with anti-NMDAR encephalitis.¹¹ These studies emphasized the paucity of inflammatory infiltrates relative to the severity of the neurologic deficits.^1,3 These findings, the normal or mild changes in the initial MRI of our patients, and preservation of drug-induced fast EEG activity (implying that cortical oscillations mediated by GABAergic inhibitory interneurons are preserved)^24,25 all concur with the potential reversibility of anti-NMDAR encephalitis.

Our study indicates that anti-NMDAR encephalitis and juvenile acute nonherpetic encephalitis or a subgroup of this disorder are the same. The reasons why this disorder is more frequent (or has received more attention) in Japan are unclear. The 4 patients reported here were seen in the past 8 years at a single institution. Another 11 patients with a similar syndrome were reported from another institution (antibodies to NR1/NR2 heteromers were not studied).¹ These experiences, as well as the fact that some patients survive without recognizing the tumor and others may die without a diagnosis (i.e., some cases¹¹ were diagnosed postmortem), suggest that the frequency of anti-NMDAR encephalitis is underestimated.

Our findings raise two questions. First, how often is anti-NMDAR encephalitis paraneoplastic? After a rapid increase of the number of identified cases, we have found that 26 of 33 patients (79%), including 3 reported here, had tumors or cysts usually in the pelvic region (20 pathologically confirmed teratomas of the ovary, 1 in the mediastinum, and 5 ovarian cysts under study). No tumor has been identified in the other 7 cases (3 of them men), who are being closely followed up for a possible occult neoplasm (Dalmau et al., unpublished data).

Second, is it necessary to remove the tumor or use immunotherapy to treat this disorder? A previously reported patient (Case 8 of reference ¹¹ and the current four patients demonstrate that the disorder may resolve without tumor removal. However, the duration of intensive care and ventilatory support in two of these patients (6 to 9 months) was significantly longer than that of all known cases that had tumor resection and immunotherapy (median 12 weeks, longest time of ventilation 4 months). Furthermore, the severity of this disorder, which may result in death or severe complications, and the reported responses to tumor resection and immunotherapy strongly support the use of these treatments.

Anti-NMDAR encephalitis is different from other types of paraneoplastic encephalitis in several ways: it results in a highly characteristic syndrome, usually affects young women, is treatment responsive, and associates with tumors that can be benign. Another difference shown here is that despite the presence of the tumor, the immune response is not maintained. This brings into consideration a contributory role of the prodromal “viral-like” disorder, which by itself or in combination with a teratoma sets off or enhances the autoimmune response. The answers to these questions will come from more experience and work in progress to model the disorder.

ACKNOWLEDGMENT

The authors thank Dr. Kimiyoshi Arimura and Dr. Osamu Watanabe (Department of Neurology and Geriatrics, Kagoshima University Graduate School of Medical and Dental Sciences), Dr. Yukitoshi Takahashi (National Epilepsy Center, Shizuoka MIND), and Dr. Keiko Tanaka (Department of Neurology, Brain Research Institute, Niigata University) for measuring antineuronal antibodies. The authors thank Dr. Atsuko Hara (School of Medicine, Kitasato University) for her pathologic assistance. They also thank Dr. Erdem Tüzün and Myrna R. Rosenfeld for their critical review of the manuscript and Mr. Jeffrey Rossi for his excellent technical assistance (University of Pennsylvania).