Long-term Functional Outcomes and Relapse of Anti-NMDA Receptor Encephalitis
A Cohort Study in Western China

This study updated the clinical features of patients with anti-NMDAR encephalitis in Western China in a large cohort and prospectively reported the long-term functional outcomes and relapses. The main findings of the study are as follows: (1) There were differences in demographic data between our cohort and Western countries, with more males and a lower tumor frequency prevailing in our cohort. The male patients were much younger and showed more abnormal ancillary tests and less frequent tumors than females at disease onset. (2) Symptom presentation varied between children and adults, and most patients achieved substantial recovery during the long-term follow-up, which was in line with the studies of Western countries. The overall prognosis continued to improve through 42 months after onset. (3) The patients with disturbance of consciousness at onset remained in a poor neurologic status in the long term. (4) Relapses occurred in approximately one-sixth of patients and were generally milder than at onset. In particular, delayed treatment and female sex were risk factors for relapse.

According to most previous studies, patients with anti-NMDAR encephalitis are predominantly young females. However, no sex difference was found in our cohort, which is in agreement with previous Asian countries' reports on adult-onset patients in Korea⁶ and on children with anti-NMDAR encephalitis in China.²⁴ In addition, we found that male patients presented much younger and less frequently exhibited tumors than females at disease onset. Abnormal MRI, EEG, and CSF analyses were reported in 42.21%, 77.04%, and 47.54% of the patients in our study, respectively, and male patients appeared to have more abnormal ancillary test results than female patients. Signals of atypical MRI findings, such as meninx enhancement, pituitary lesions, and nonspecific periventricular white matter lesions, were also included in our study, which did not have clinicopathologic significance (table e-4, links.lww.com/NXI/A412). In our study, many more male patients lived in rural areas, were exposed to poor sanitation and living conditions, and were susceptible to viral infection. In addition, most male patients had smoking habits. All these factors might be associated with frequent ancillary test abnormalities. Compared with that in Western study populations,^3,7,25 the prevalence of tumors (8% before 2015¹⁰ and 15.57% through 2019) was still much lower in our study, but it was similar to that reported in other Asian populations (Lim et al., 22.7%; Xu et al., 19.5%; and Wang et al., 1.9%).^6,10,14,26 These phenomena and discrepancies between different populations thereby indicate that genetic background characteristics (e.g., sex hormones, tumors, and human leukocyte antigens) may attribute to susceptibility to anti-NMDAR encephalitis, and viral infection could also trigger the disease.^5,7,26–29

Symptom presentation varied between children and adults. The most common clinical profile of the adolescents with anti-NMDAR encephalitis in our study was more neurologic disorders (e.g., fever, sleep disorder, seizures, and disturbance of consciousness), whereas adults presented with more psychiatric and cognitive disorders, which aligns with previous findings.^7,24,28 In a previous study, ataxia, dyskinesias, and movement disorders were more frequently exhibited in children. However, there were no differences in our study, probably because our cohort did not include patients younger than 9 years. We reported that 48.4% of patients displayed autonomic symptoms, and other studies on anti-NMDAR encephalitis also reported elevated incidences of autonomic dysfunction (Titulaer: 37%–48%; Dalmau: 69%).^5,7

In this series of patients, 84.8% showed clinical improvement within 4 weeks after beginning immunotherapy, and 80.7% and 85.7% exhibited substantial recovery (i.e., mild or no residual symptoms) at 12 and 24 months, respectively. The overall prognosis continued to improve through 42 months after onset, which also agrees with the findings in other previous studies (75%–93%).^7,14 In Western countries, general clinical improvements of 50%–60% within 4 weeks have been reported. This is likely because the baseline characteristics of the Western cohort (the median mRS on admission was 5, and the ICU admission rate was 50%–77%) were more severe than those in our cohort (the median baseline mRS was 4, and only 15% were admitted to the ICU). Thus, our results might not be applicable to Western populations. Moreover, different populations with varied genetic backgrounds and other external factors (e.g., lifestyle, metabolism, and diet) might also respond differently to treatment. Therefore, more research is required to understand the mechanisms underlying disease initiation, aggravation, and progression in different populations.

The mortality of anti-NMDAR encephalitis in our cohort was also low (only 6.9% in our study; 7% in the study by Titulaer⁷). In previous studies, the predictors of good functional outcomes were the lack of ICU admission and early treatment,⁷ whereas the poor functional outcome predictors were memory deficits, disturbance of consciousness, young age, and high CSF and serum titers.^6,7,13 In our study, disturbance of consciousness (an indication of ICU admission) was an independent predictor of poor functional outcomes, which was in line with previous research, whereas young age and cognitive disorders did not predict poor functional outcomes. Our study showed that patients of older ages at disease onset had higher frequencies of ICU admission, and older age was associated with worse functional outcomes. However, only children from 9 years of age and older were included because younger children were all referred to other children's specialist hospitals. This fact might also relate to the lack of significance between being of a younger age and poor functional outcome.

Our study showed a relapse rate of 15.9%. Most (82.0%) patients experienced an initial relapse during the first 24 months. Titulaer et al. followed their cohort for a median time of 2 years, and 12% of the patients with anti-NMDAR encephalitis experienced clinical relapse. The relapse rate reported in other studies varied between 20% and 25%,^12,13 even reaching 36.4%.¹⁴ With regard to treatment modalities, the risk of further relapse was elevated in those who did not receive immunotherapy at onset in the cohort described by Gabilondo et al.³⁰ and in patients who did not receive aggressive immunotherapies at the first episode as described by Nosadini et al.²¹; these results also echoed those observed in our cohort (delayed treatment [HR: 0.41, 95% CI: 0.20–0.75; p = 0.01]; figure 3B). Therefore, timely and aggressive immunotherapies might be beneficial for patients with anti-NMDAR encephalitis. Notably, females had an increased risk of recurrence in our follow-up study, which indicated that sex hormones and other genetic background factors may also affect susceptibility to relapses. Compared to Western populations, lack of treatment and lack of second-line immunotherapies during the initial episode were not statistically associated with an increased relapse frequency. These differences were probably due to the small number of patients not receiving immunotherapy (n = 10) or using second-line immunotherapies and plasma exchange at onset (n = 22), as well as due to reduced tumor incidences (n = 38), in our cohort.

However, the current proposed definition of clinical relapse is not based on the presence of antibodies in the CSF but rather on observations and manifestations of clinical symptoms.^7,12–14,21 Therefore, the reported 15.9% relapse rate in our study, along with other earlier studies (12%–35%), might have been diagnosis bias partially if antibodies in the CSF were lacking because the disease is caused by antibodies. In this cohort, we observed 2 patients with antibody negativity and 3 patients without CSF retesting who experienced early relapse (within 3–6 months). These patients performed a clinical relapse but lacked antibody evidence; thus, pseudorelapse may have occurred because of viral infection, incomplete initial treatment, and so on. Therefore, we divided the subjects into those who relapsed early and those who relapsed later in the hope of revealing more convincing predictive factors and potentially better aligning with relapse data from other studies.

In our study, different treatment modalities at disease onset were not considered to be predictors of time to relapse because such an analysis would have increased the false positive rate. In addition, selective treatment bias occurred in our cohort because of the observational and nonrandomized controlled nature of the study. In our cohort, different treatment modalities were statistically associated with sex (as shown in table e-1, links.lww.com/NXI/A412). The mRS scores at baseline were also significantly different among the IV methylprednisolone (IVMP) alone, IVIg alone, and IVIg + IVMP groups (analysis of variance: p < 0.05).

Despite substantial progress in understanding this disease, many questions remain, and our study has several limitations. First, there was no significant difference in the subsequent recurrence rates between second-line and MMF/tacrolimus-treated patients and other patients in our study, and only a relatively small proportion of patients received second-line or long-term noncorticosteroid regimen therapy. This may be because CTX and RTX are off-label drugs in China and patients fear their cost or adverse effects. Second, it is inevitable that in clinical practice, patients with anti-NMDAR encephalitis and neurologic worsening are more likely to be classified as having autoimmune-mediated relapse rather than viral infection. Thus, to validate relapse and rule out other potential causes, we need to both dynamically observe autoantibodies in the CSF and perform PCR testing to detect herpes simplex virus. Finally, mRS as a primary outcome measure might overshadow findings on relapse and opens the study up to criticisms of the relevance of identifying predictors of poor outcomes in the setting of an incomplete outcome measure. Data have shown that mRS outcomes do not correlate with cognitive, behavioral, psychosocial, or quality of life outcomes.^11,17,19 To better assess and detect potential risk factors related to prolonged cognitive deficits, neuropsychiatric symptoms, or seizure recurrence symptoms, using a more specific questionnaire and detailed evaluations is necessary (e.g., Neuropsychiatric Inventory, Cognitive Impairment Rating Scale, Montreal Cognitive Assessment, and National Hospital Seizure Severity Scale). Despite these limitations, we describe the clinical profiles, long-term outcomes, and relapse in the largest group of Chinese patients with anti-NMDAR encephalitis in China (n = 244) examined to date, which promotes a deeper understanding of the prognosis of this disease.

The authors thank all participants for their participation in this study.