Suicidal Thoughts and Behaviors in Anti-NMDA Receptor Encephalitis: Psychopathological Features and Clinical Outcomes

We conducted an observational and longitudinal study that was approved by the Institutional Research Committee (protocol 53/16) and by the Ethics Committee of the National Institute of Neurology and Neurosurgery of Mexico (NINN) and was conducted in accordance with the Declaration of Helsinki of 1975 (as revised in 2008). Informed consent was obtained from the patients. The anonymity of all patients was preserved. All the diagnostic and therapeutic procedures described in this article were necessary to fulfill clinical standards of care.

This cohort included hospitalized patients with definite ANMDARE who attended the NINN between 2014 and 2021. Sampling was consecutive, and patients were included in the study if they fulfilled the Graus criteria for definite ANMDARE (panel 4), including the presence of one or more of six major groups of symptoms, a positive determination of immunoglobulin G antibodies against the NR1 subunit of NMDA receptor in cerebrospinal fluid (CSF), and a reasonable exclusion of other disorders (12). CSF anti-NMDA receptor antibodies were processed at Labco Nous Diagnostics (Barcelona, Spain), with the use of rat brain immunohistochemistry and cell-based assays with NMDA-expressing cells, to prevent the false-positive or false-negative results that are frequently obtained with analyses of a patient’s serum (12, 13). All participants underwent a complete neurological and psychiatric examination, including laboratory tests and brain imaging, to rule out other causes of their symptoms, as recommended in the current diagnostic criteria (12). These tests included examinations for systemic autoimmune diseases, such as anti-double-stranded DNA antibodies, antinuclear antibodies, antineutrophil cytoplasmic antibodies, anti-beta-2-glycoprotein antibodies, and antiphospholipid antibodies; metabolic and endocrine diseases due to disruptions in vitamin B12, thyroid-stimulating hormone, and T4; and infectious diseases, such as HIV and syphilis. In addition, we ruled out viral and bacterial central nervous system (CNS) infections: in the current episode of all patients, CSF polymerase chain reaction results were negative for herpes simplex virus types 1 and 2; cytomegalovirus; Epstein-Barr virus; varicella zoster virus; human herpes virus types 6, 7, and 8; enterovirus; toxoplasma; parvovirus B19; and lymphocytic choriomeningitis virus. Other antibodies known to be related to autoimmune encephalitis were not included due to financial limitations at the study site (in Mexico).

All clinical measures were obtained before and after immunotherapy. Clinical data were prospectively recorded with a structured format that included an assessment of a broad scope of neurological and psychiatric features observed in patients with ANMDARE, as detailed elsewhere (4). Items assessing suicidal thoughts and behaviors are part of this instrument. These thoughts and behaviors were classified in accordance with the Self-Directed Violence Classification System, developed and promulgated by the U.S. Department of Veterans Affairs and Department of Defense (14). Information regarding suicidal ideation and self-directed violence was collected through clinical interviews with relatives and direct clinical assessment with each patient. Assessments occurred at admission (before receiving pharmacological treatment and immunotherapy) and at discharge. We recorded thoughts of engaging in suicide-related behavior if the patient declared those thoughts explicitly and unambiguously. We classified these thoughts as either suicidal ideation with suicidal intent or suicidal ideation without suicidal intent. Suicidal self-directed violence occurring before the hospitalization was recorded and based on information provided by a relative and included a description of the specific behavior. Because patients with encephalitis may engage in erratic behaviors accidentally leading to nonsuicidal self-directed violence, suicidal self-directed violence was considered only if the self-harm behavior was accompanied by an explicit verbal exchange revealing thoughts of engaging in suicide-related behavior. Acts of self-directed violence during hospitalization were also recorded. We used the DSM-5 criteria to diagnose psychiatric disorders, such as depression, catatonia, and delirium. The full neuropsychiatric assessment of these patients included a series of psychometric scales and inventories, in addition to cognitive screening tests (discussed elsewhere) (4). Sociodemographic variables were also recorded and included socioeconomic status determined by the social work department (a score of 1 indicated a very low socioeconomic status; 2, low status; 3, middle-low status; 4, middle status; 5, middle-high status; and 6, high status).

Descriptive statistics were calculated, and Kolmogorov-Smirnov normality tests were conducted. Wilcoxon tests and chi-square tests were used to compare data from patients with and without suicidal thoughts and behaviors. Bonferroni corrections for multiple comparisons were used to reduce the probability of type I errors. Data analysis was performed with SPSS (version 21).

We included 120 patients with definite ANMDARE: 15 patients (13%) had some form of suicidal thoughts or behaviors (Table 1). This subgroup had a median age of 32 (with a range between 19 and 48), and 53% were female.

As seen in Table 1, suicidal ideation was present for each of the 15 patients; of these patients, 12 patients had suicidal ideation with intent, three patients engaged in preparatory behaviors, and seven patients engaged in suicidal self-directed violence.

All 15 patients had psychotic features: 10 patients had psychotic depression, 14 patients had delusions (persecutory delusions, N=9; grandiose delusions, N=6, nihilistic delusions, N=5; jealous delusions, N=3), and 11 patients reported hallucinations (visual hallucinations, N=10; auditory hallucinations, N=9). Delirium and catatonia were observed in 13 and eight patients, respectively. Interestingly, five of the 15 patients were experiencing a relapse of ANMDARE.

Regarding their psychiatric history, one patient had received specialized attention due to self-harm and a history of repeated sexual abuse. Three patients had suffered intimate partner violence, and five patients reported alcohol or tobacco abuse. Eight of the 15 patients had an unstable family situation, and five of the 15 patients had a background of domestic violence.

Routine examinations showed that seven of the 15 patients had indicators of inflammation in the CSF; eight of the 15 patients had abnormalities on MRI involving the medial temporal lobe; and all 15 patients had abnormal electroencephalography results (generalized slowing was present in all patients).

As seen in Table 2, the patients with suicidality were older and more frequently married (or within a stable relationship) than those without suicidal behaviors. Psychotic depression, insomnia, and impulsivity, as well as being diagnosed with a relapse of the disease, were significantly associated with suicidality. However, after Bonferroni correction for multiple comparisons, only differences in psychotic depression (67% vs. 10%; χ²=28.7, df=1; p<0.001) and impulsivity (87% vs. 43%; χ²=10.1, df=1; p=0.001) remained significant.

During hospitalization, four of the 15 patients engaged in repetitive acts of self-directed violence, including head banging, cutting their own neck or forearms, trying to hang themselves with sheets or towels, and injuring themselves with objects such as pens. Thirteen of the 15 patients received treatment with intravenous methylprednisolone, 10 patients received plasmapheresis, and seven patients received intravenous immunoglobulin. Regarding psychopharmacological agents, 14 of the 15 patients received an antipsychotic, 12 patients received lorazepam, and four patients received dexmedetomidine. At discharge, self-directed violent thoughts and behaviors remitted completely in 14 patients. Long-term follow-up confirmed the absence of suicidal thoughts and behaviors after hospitalization and treatment. Follow-up visits occurred with a median of 43 months (with a range between 12 and 92 months). One patient (patient 13) had episodic psychiatric disturbances after hospitalization, including thoughts of engaging in suicide-related behavior and acts of nonsuicidal self-directed violence.

Patient 13 had a relevant psychosocial and behavioral history, including repeated sexual abuse in her childhood and nonsuicidal self-directed violence (cutting). However, she had never attempted suicide until the onset of the psychotic phase of ANMDARE. Although significant improvements were observed after immunotherapy, residual symptoms, including mild anxiety and depression, remained and were exacerbated by interpersonal stress, which led to a chronic, episodic course of psychosis with auditory verbal hallucinations, thoughts of engaging in suicide-related behavior, and nonsuicidal self-directed violence. Only a partial response was observed after psychotherapy and pharmacological therapy (including an antipsychotic and an antidepressant).

In our sample of patients with ANMDARE, 13% presented some form of suicidality, which is consistent with the study by Zhang et al., where suicidality was observed in 13% of 133 ANMDARE patients from China (11). In that study, seven patients (41%) had suicidal ideation, and eight patients (46%) had a suicidal attempt; strikingly, 12% of this subgroup of patients died by suicide (11). No patients died by suicide in our sample. In the Zhang et al. study, patients with suicidality initially presented with more prominent psychiatric symptoms, including delusions, mania, insomnia, aggression, and depression, than those without suicidality (11). In our sample, all patients exhibited psychotic symptoms, highlighting the concept of autoimmune psychosis (15), which provides an operational approach to identify patients with psychosis who are at a high risk of having an immunological condition.

In the bivariate analysis, patients with suicidal thoughts and behaviors from our sample were characterized by a later age of onset and a higher frequency of psychotic depression, delusions, insomnia, and impulsivity. On the basis of results after correcting for multiple comparisons, our current hypothesis is that psychotic depression and impulsivity are psychopathological features that are significantly related to suicidal thoughts and behaviors in patients with ANMDARE.

In our study, suicidal thoughts and behaviors remitted completely in 14 of 15 patients after immunotherapy and long-term follow-up. This suggests a causal mechanism of the anti-NMDAR antibodies in most patients from our sample, and it also highlights the need for more research on the relationship between suicidal behavior, inflammation, and neurological disease (16, 17). Activation of the kynurenine pathway and modification of different metabolites (quinolinic acid and kynurenic acid) with effects on glutamate transmission have been hypothesized to be biological mechanisms related to suicidal behavior. Quinolinic acid, a potent NMDA receptor agonist, is increased in the CNS of suicidal patients, providing a neurobiological hypothesis for the rapid effect of ketamine on suicidal behavior (16). Although the neurobiological mechanism underlying suicidal behavior among patients with ANMDARE remains to be elucidated, our findings support the hypothesis that antibodies may increase or decrease biological signaling mediated by NMDA receptors in different stages of the disease (18).

Suicidal ideation persisted during follow-up for one patient (6%) who had a relevant psychiatric history (repeated sexual abuse in childhood). Although our patient had a history of nonsuicidal self-directed violence, the psychopathological features of the patient changed significantly after the onset of ANMDARE, with the appearance of auditory verbal hallucinations and suicidal thoughts and behaviors and with a worsening of the self-directed violence. Zhang et al. also reported two patients who presented suicidal thoughts and behaviors after the acute phase of the disease (11), which would be in line with the increased frequency of affective and cognitive symptoms observed following ANMDARE (19, 20). These findings also call for an awareness of the potential interactions between specific neurobiological effects of anti-NMDAR antibodies and psychosocial backgrounds. A subset of the patients in our sample had relevant personal histories, including repeated sexual abuse (one of 15 patients), intimate partner violence (three of 15), and domestic violence (five of 15). Although the rate of suicidality relapse in our sample was low (one of 15 patients) and even though ANMDARE is a well-defined neurological disease, we cannot rule out the possibility that there might be biopsychosocial interactions in a subset of patients leading to atypical and unfavorable outcomes.

Although suicidality in ANMDARE is likely triggered by different pathophysiological mechanisms, it can occur in different phases of the disease course, particularly during the acute phase and relapses (11). In our study, 10 of the patients (66%) had some form of suicidality during the first episode of the disease, but 5 of the patients (33%) had suicidality during relapse of the disease. Similarly, Zhang et al. reported suicidality across different stages of the disease: 10 of 17 patients had suicidal thoughts and behaviors prior to admission; three, during the hospital stay; two, after discharge; and two, during relapse. Because suicidal thoughts and behaviors represent a potentially lethal risk, clinicians evaluating patients with ANMDARE or suspicion of ANMDARE must strictly and routinely screen and assess suicide risk across the different phases of the disease course. In addition, staff with mental health experience should be involved whenever necessary.

According to our results and those of previous studies (9–11), ANMDARE shares an increased risk of suicidality with other neurological diseases. This increased risk is consistent with the research conducted with patients with neurological diseases that frequently lead to cognitive dysfunction and psychotic features. For instance, the risk of suicide was increased among patients in the first year after receiving a diagnosis of dementia compared with the general geriatric population in a large cohort study (21). A recent Danish study by Erlangsen et al., including more than 7.3 million individuals, found a suicide rate of 44 per 100,000 person-years among patients with a diagnosed neurological disorder compared with 20.1 per 100,000 person-years among individuals without a diagnosed neurological disorder (17). Importantly, patients with encephalitis had a higher suicide rate (fully adjusted incidence rate ratio [faIRR]=1.7, 95% CI=1.3–2.3) than patients with stroke (faIRR=1.3, 95% CI=1.2–1.3) but a rate similar to those with epilepsy (faIRR=1.7, 95% CI=1.6–1.8) and Parkinson’s disease (faIRR=1.7, 95% CI=1.5–1.9) (17). Interestingly, suicidal thoughts and behaviors have been reported following other forms of encephalitis such as encephalitis due to herpes simplex virus, encephalitis lethargica, and other forms of autoimmune encephalitis (22, 23). These findings highlight the need to ensure more routine suicide risk screening and assessment among patients with neurological conditions.

The presence of psychosis in all the patients with suicidal thoughts and behaviors from our sample is consistent with the well-replicated epidemiological relationship between suicide and psychotic disorders (6). Recent studies show that suicide rate among those with schizophrenia is 352 per 100,000 person-years, whereas among those with bipolar depression, the suicide rate is 237 per 100,000 person-years (6), highlighting the fact that both affective and nonaffective psychoses are associated with an increased risk of suicide.

Overall, the present study provides important insights into the mental health of those in the acute stage of ANMDARE. Raising awareness among clinicians and encouraging risk assessments in this patient population can help reduce patient morbidity and mortality. The impact of suicide on families, friends, and communities is devastating and far reaching. The World Health Organization has also outlined the importance of suicide prevention not only for the directly affected individuals and families but also for the benefit and well-being of communities, our health care systems, and society at large (24).

Cognitive dysfunction and catatonic states interfered with our capacity to conduct in-depth interviews that could have provided a better understanding of the patients’ motivations for suicidal and nonsuicidal self-directed violence. After the acute episode of encephalitis, most patients were unable to provide a clear recollection of their mental state during the psychiatric episodes in which suicidal thoughts and behaviors appeared. The observational nature of the design is not appropriate for making strong inferences regarding the neurobiological mechanisms related to suicidality or inferences about the efficacy of immunotherapy in this population, and the small sample size limits generalizability of the results.