In this article, we described 9 anti-AMPAR encephalitis patients and summarized their clinical characteristics. The patients were generally middle-aged women with an average onset age of 59 years old and a female-to-male ratio of 3.5: 1. Onset modes varied from acute, subacute to chronic. Three clinical pictures, including limbic encephalitis, pure amnesia and fulminant encephalitis, were identified with LE as the majority. Brain MRI were abnormal in 75% of the patients with no specific patterns recognized. All patients have positive blood AMPAR antibodies, and 67% of them have paired antibodies in CSF. 67% percent of the patients had tumors, lung cancers or thymomas. After immunotherapy and oncotherapy, partial improvement of symptoms was observed among all 6 patients during their hospitalization. During follow-up, 3 patients had marked decrease of mRS score, 1 patient had unchanged mRS score, 4 patients died and 1 was lost.
The demographic characteristics of anti-AMPAR encephalitis revealed by this research were similar to that of other studies [4,5,6,7]. In the original study that identified AMPAR as a novel antigen in 10 limbic encephalitis patients, the median age was 60 and 9 of the 10 patients were female. Additionally, in the recent systemic review covering 55 cases of anti-AMPAR encephalitis, the median age was 53.2 years old (range14–92 years) and the female-to-male ratio was 36 to 19 [4, 7].
Our research expanded the clinical features of Anti-AMPAR encephalitis. Anti-AMPAR encephalitis was initially recognized as limbic encephalitis, and the following research identified more clinical patterns [4]. Hoftberger et al. summarized four clinical modes in 22 patients, including limbic encephalitis (12 patients), diffuse encephalitis (8 patients), limbic encephalitis preceded by motor deficits (1 patient), and pure psychosis (1 patient) [5]. Similarly, Joubert et al. identified four main modes according to the prominent onset symptoms in a seven-patient cohort, including confusion (3 patients), isolated epileptic (1 patient), isolated amnestic (1 patient) and fulminant encephalitis (2 patients) [6]. Our study observed similar patterns, with limbic encephalitis in 7 patients, purely amnestic in 1 patient, and fulminant encephalitis in 1 patient. However, the clinical presentations were highly variable, ranging from the commonly seen symptoms of limbic encephalitis as psychosis, confusion, and amnesia, to the infrequent symptoms of seizure, dysautonomia, ataxia or other cerebellar symptoms, insomnia, involuntary movements, dysarthria, and sensory symptoms. We expanded the clinical spectrum of anti-AMPAR encephalitis by adding dysphagia and deafness. The bilateral deafness developed as a prominent symptom during the disease course without prior identifiable risks, such as ototoxic drugs administration. This manifestation was also observed in a recently diagnosed patient, which is not included in this series. The expanded profiles will help clinicians accurately recognize patients with atypical presentations, and reduce the rate of misdiagnosis and missed diagnosis.
Despite the diverse symptoms mentioned above in anti-AMPAR encephalitis, limbic encephalitis remains the majority. Additionally, clinicians should always meticulously rule out anti-AMPAR encephalitis in patients with pure amnesia or psychosis, as the disease is treatable and may be comorbid with tumors.
Limbic encephalitis is frequently seen in autoimmune encephalitis, such as anti-NMDAR, GABAB-R, CASPR2, LGI1, and AMPAR encephalitis, suggesting common mechanisms underlined. Limbic encephalitis in anti-AMPAR encephalitis is thought to be caused by antibody-mediated internalization of AMPAR clusters at synapses [4]. Increased availability of AMPAR clusters is critical for long-term potentiation in the hippocampus, and therefore for memory consolidation [8]. Specifically, GluA2 antibodies resulted in reduction of synaptic GluA2-containing AMPARs, impairment of long-term synaptic plasticity in vitro, and damaged learning and memory in vivo [9]. This explains amnesia in anti-AMPAR encephalitis patients and provides insights into the symptomatic overlap with LGI1 encephalitis, as LGI1-ADAM22 complex interacts with PSD95 and stabilizes AMPARs in the postsynaptic membrane [10, 11]. On the other hand, the glutamate hypothesis of psychosis indicates that hypofunction of GABAergic neurons may account for psychiatric symptoms in some autoimmune encephalitis. Indeed, internalization of NMDARs by GluN1 antibodies and AMPARs by GluA1/GluA2 antibodies affects the activities of cortical networks [3, 12]. Epilepsy is another commonly encountered symptom in autoimmune encephalitis. One possible mechanism is that increased seizure susceptibility is caused by reduced inhibitory neurotransmission, as indicated in GABAA-R, GABAB-R, or GAD65 encephalitis [13,14,15,16]. In hippocampal pyramidal neurons treated with CSF of anti-AMPAR encephalitis patients, patch-clamp revealed decreased miniature excitatory postsynaptic currents (EPSCs), which seemed paradoxical to seizures in patients [17]. Explanation was that decreased EPSCs resulted in decreased inhibitory synaptic transmission and increased intrinsic excitability, predisposing patients to epilepsy [17]. However, seizures were relatively rarely observed in anti-AMPAR encephalitis compared with other autoimmune encephalitis mentioned above. The discrepancy of seizure incidence and type in different antibody-mediated encephalitis remains unexplained.
Brain MRI, EEG, CSF study, and antibody test are the main diagnostic tools for anti-AMPAR encephalitis. Brain MRI is considered as a sensitive but not specific diagnostic tool for anti-AMPAR encephalitis. According to the systemic review with the largest anti-AMPAR encephalitis cohort (55 participants), up to 86% of the patients had abnormal brain MRI with a predilection of bilateral temporal lobes, which was related to topography of GluA1 and GluA2 expression [7]. Seventy-five percent of our patients had abnormal brain MRI, with no preference for specific brain sites. However, it should be noted that patients with anti-AMPAR encephalitis may have completely normal brain MRI as indicated by our patients and the imaging abnormalities may spread to unexpected sites, like basal ganglia, cerebellum, and even posterior temporal and parieto-occipital regions [18]. Therefore, for patients with nonspecific MRI manifestations but with typical symptoms, anti-AMPAR encephalitis should be cautiously differentiated. EEG was less sensitive than brain MRI and only 44% of patients had EEG abnormalities [7]. EEG was also nonspecific, varying from nonspecific slow waves, epileptiform activities, to normal. The most common EEG abnormality in our patients was nonspecific slowing. Cerebrospinal fluid study has limited significance for differential diagnosis. Systemic analysis revealed that inflammatory CSF changes, defined as pleocytosis, increased CSF protein levels, and/or oligoclonal band, were rather frequent seen in NMDAR, GABABR, AMPAR, and dipeptidyl-peptidase-like protein 6 (DPPX) encephalitis. While in autoimmune encephalitis with either CASPR2, LGI1, GABAA, or glycine receptor antibodies, CSF findings were generally normal [19]. In accordance with this systemic review, 5 patients (63%) in our study had elevated CSF protein and 1 (13%) patient showed pleocytosis in CSF. Blood and CSF AMPAR antibodies were the definitive diagnostic markers for anti-AMPAR encephalitis. Different from other studies, our study showed that the positivity rate of AMPAR antibodies was higher in serum than in CSF and that only GluA2 antibodies were detected. The difference might be accounted by the relatively small number of patients included. These findings suggest that for patients suspicious of anti-AMPAR encephalitis, both serum and CSF should be sent for antibody tests. Interestingly, the difference of clinical profiles between patients with antibodies against GluA1 and GluA2 was not clear yet.
Anti-AMPAR encephalitis can be paraneoplastic. Forty-eight to 70 % of patients were found to have tumors, mostly lung, thymus, breast, and ovarian tumors [4,5,6,7]. 6 (67%) patients in our study had tumors, 3 of which had lung cancers and the rest had thymomas. Additionally, rare tumors including medullary thyroid cancer, malignant melanoma, and Ewing’s Sarcoma were also reported in anti-AMPAR encephalitis cases [20,21,22]. Psychiatric symptoms at presentation predicted the presence of tumors [7]. These findings suggest the necessity of extensive tumor screening in patients with psychiatric symptoms. In addition to tumors, patients of anti-AMPAR encephalitis seem to have a predisposition to other autoimmune diseases. Systemic lupus erythematosus, Hashimoto’s thyroiditis, and myasthenia gravis were reported to be concurrent with anti-AMPAR encephalitis [23,24,25]. Therefore, signs of autoimmune diseases should also be paid attention to when anti-AMPAR encephalitis is suspected.
Treatment of anti-AMPAR encephalitis includes immunotherapy and oncological treatment if tumors are comorbid. Immunotherapy is composed of first-line therapies (IVIG, steroids, and plasmapheresis), and second-line therapies (rituximab and immunosuppressants, etc.). Treatment response, defined as mRS score decrease with an mRS score ≤ 3 at the last follow-up, was observed in 3 of 8 patients in our cohort and in 71% of the patients reported in the literature [5]. The overall survival rate of patients with and without tumors showed no significant difference as indicated by our study and literature [5]. The poor prognosis is unlikely related to delayed diagnosis and treatment, as indicated by the 3 patients who died despite of prompt diagnosis and adequate treatment. The presence of psychiatric symptoms and concurrent onco-neuronal antibodies were associated with poorer outcomes while younger age and confusion at presentation were linked with favorable prognosis [5, 7]. This was also observed in patients of anti-AMPAR encephalitis with concurrent CRMP5 antibodies [26]. Additionally, fulminant encephalitis was associated with a poor prognosis. Whether or not the presence of tumors or onco-neuronal antibodies predicts relapse remains elusive. Cases have been reported that anti-AMPAR encephalitis and the comorbid tumors relapsed after immunological and oncological treatments but not in our patients [27]. It seems that patients who received aggressive therapy (chemotherapy and rituximab) were unlikely to have relapses than those who did not [5]. Therefore, it’s important to closely monitor the patients after treatment.
Taken together, our study characterized a series of anti-AMPAR encephalitis patients from China and expand the clinical features of anti-AMPAR encephalitis. However, the limitations of the study are obvious. First, the small sample size limits the statistical power, therefore hindering a firm conclusion. Second, the clinical information was not documented completely, with the results of certain tests not available. Third, only patients clinically suspicious of encephalitis were tested for anti-neuronal antibodies. This selective bias may underestimate the unusual symptoms in anti-AMPAR encephalitis. Fourth, patients No.1 and patient No.9 were concurrent with positive blood anti-Hu antibodies, which complicated the diagnosis of anti-AMPAR encephalitis despite of paired positivity of anti-AMPAR antibodies in both serum and CSF. Fifth, for patients No. 2 and No. 4, despite of typical symptoms and the high positive predictive value of our method, low titers of anti-AMPAR antibodies were tested in only serum, raising the possibility of false-positive. Better understandings of this disease, including its symptoms, comorbidities, prognosis and development of better diagnostic and therapeutic maneuvers, rely on deeper investigations into the pathological mechanisms and the accumulation of patient cohorts.