Serum and CSF Anti-NMDAR Antibody Testing in Psychiatry

Anti-N-methyl-d-aspartate receptor (NMDAR) encephalitis is an autoimmune disorder that most commonly presents with rapid-onset psychiatric symptoms, such as severe behavioral disturbance and psychosis (1). In most cases, there is overt evolution of the disease with seizures, movement disorders, and autonomic disturbance (2). However, an isolated psychiatric form is seen in up to 5% of cases, and in others, subtle neurologic symptoms can be missed, making differentiation from primary psychiatric disorders, such as schizophreniform psychosis or mania, difficult (3–5). Informed by several studies screening patients with first-episode psychosis for the presence of the anti-NMDAR antibody (NMDAR Ab), as well as clinical practice guidelines, the proportion of psychiatric patients being tested has increased in recent years (6–8). Typically, serum testing on commercially available assays is first performed, with only serum-positive cases proceeding to CSF sampling for diagnostic confirmation. This approach, however, creates diagnostic and therapeutic uncertainty if serum and CSF markers of disease render discrepant results.

While the specificity of serum NMDAR Ab testing using standard diagnostic assays is high, it is unavoidable that this test renders some false-positive results if used as a screening tool for large numbers of patients (9). It is generally accepted that anti-NMDAR encephalitis requires the presence of CSF immunoglobulin G (IgG) antibodies, which are directly pathogenic, and large observational studies have demonstrated markedly greater sensitivity of CSF antibodies compared with serum (2, 10, 11). Conversely, several case series have described successful diagnosis and treatment of anti-NMDAR encephalitis on the basis of positive serum testing alone, with negative or absent confirmatory CSF testing (12, 13). There is a need to develop a prompt and accurate diagnostic process for anti-NMDAR encephalitis, particularly in atypical cases involving presentation to psychiatric services. Early treatment of anti-NMDAR encephalitis with immunomodulatory therapies and removal of any associated tumors results in markedly improved long-term outcomes (2, 14). Additionally, treating encephalitic patients with antipsychotics could result in severe extrapyramidal side effects and lower seizure threshold. However, delaying traditional psychiatric treatment for a primary psychiatric disorder, or exposing patients to unnecessary immunotherapy, may also have significant consequences.

Our aim in the present study was to comprehensively describe and compare the symptoms that occurred on presentation, during the course of illness, and after treatment in an IgG serum or CSF NMDAR-Ab-positive sample of patients who presented to psychiatric care. It was hypothesized that clinical features on initial presentation at hospitalization could differentiate anti-NMDAR encephalitis from primary psychiatric illness and guide accurate diagnosis and timely institution of management.

Methods

Ethical approval for this study was granted from Metro South Addiction and Mental Health Services Human Research and Ethics Committee. A retrospective chart review of a statewide cohort of anti-NMDAR-Ab-positive cases was performed in Queensland, Australia, up to and including December 2018. Since 2010, all NMDAR Ab testing for the state has been performed in a single laboratory (Pathology Queensland, Brisbane, Queensland, Australia), which is a diagnostic referral laboratory offering specialized neuroimmunology testing. Anti-NMDA Abs are detected through a two-person blinded reading of a commercial fixed-cell assay containing four biochips of primate hippocampus, primate cerebellum, NR1-transfected human embryonic kidney 293 (HEK293) cells, and nontransfected control HEK293 cells (Euroimmun, Lübeck, Germany). Two neuroimmunology clinics in the public health care system are responsible for the population studied. Cases were identified by the laboratory records and review of clinic records. Adult patients were included if there was a positive IgG NMDAR Ab in either serum or CSF and the first presentation of illness was to psychiatric services. Equivocal or indeterminate NMDAR Ab cases, on the basis of weak immunofluorescence staining, which are typically repeated for accuracy, were also included but compared separately. Individuals were excluded if they were <18 years old, residing out of state, or tested by a private physician.

Case identification and data collection were performed separately by two authors (N.W. and A.S.). Variables collected included demographic characteristics, medical and psychiatric history, clinical features of presentation, course of illness and recovery, results of investigations, and treatment received. These variables were preselected and based on previous review of published cases (5). The timing of each major symptom type, investigations, and treatment commencement was noted. Initial and peak Bush-Francis Catatonia Rating Scale scores were recorded, but catatonia was noted if DSM-5 criteria were met (15, 16).

Symptoms were categorized into the following clusters for analysis: psychiatric, movement, seizures, autonomic, and cognitive. For clinical features, missing data were recorded as absence of that feature. For investigations, missing data were recorded as not tested. Treatment response and outcome were categorized as resolved, improved, no improvement, or deterioration/death, with posttreatment symptoms recorded.

Descriptive statistics were used for the initial analysis of the data. The CSF NMDAR-Ab-positive and CSF NMDAR-Ab-negative cases were compared by using Fisher’s exact test or t test as appropriate. The prespecified variables for comparison were chosen to aid psychiatric decision making in the acute phase (prodromal features: flu-like symptoms, insomnia, and cognitive deficit; psychiatric symptoms: severe agitation, odd/unusual behavior, affective lability, catatonia, and speech disturbance). The presence of neurological symptoms (e.g., seizures or movement disorder), autonomic disturbance, or loss of consciousness should prompt further investigation and thus not included in the comparison. Correction for multiple comparisons was performed with the Benjamini-Hochberg procedure, with a false discovery rate of 5% (17).

Results

Twenty-four case patients, who presented to psychiatric services from 2011 to 2018, were identified with positive serum or CSF NMDAR Ab. This represents 60% (N=24/40 cases) of all positive NMDAR Ab results tested by the laboratory, with the remaining case patients presenting to neurology or general medicine. An additional three case patients with equivocal serum NMDAR Ab, who presented to psychiatry, were identified. All subjects had both serum and CSF antibody testing, except for one equivocal NMDAR Ab case in which repeated lumbar puncture was unsuccessful. The mean time to notify the treating clinician after a positive test was 2.8 days (SD=1.5).

CSF NMDAR Ab-Positive Cases

There were 16 CSF NMDAR Ab-positive case subjects (mean age=32.7 years [SD=13.2]); female, N=10; male, N=6) (Table 1). The mean average time for testing was 21 days after index admission (SD=27), with 11 out of 16 cases tested in <7 days. A prior psychiatric history was present for seven case patients, four of whom had a past history of anxiety or depression. Three case patients had an existing diagnosis of schizophrenia (mean time=3.3 years [SD=0.5]), with no prior neurological or speech disturbance, catatonia, cognitive deficits, or antipsychotic sensitivity; however, they had no prior anti-NMDA Ab testing. One case patient had a history of epilepsy, and four had a history of head injury. Prodromal insomnia (N=12, 75%), cognitive deficits (N=14, 87.5%), headache (N=3, 18.8%), weight loss (N=3, 18.8%), and flu-like illness (N=1, 6.3%) were reported. The average time of symptoms before presentation was 33 days (SD=41.5).

All case subjects had psychotic symptoms on admission to hospital. Delusions were described as persecutory (N=13, 81.3%), with additional unusual (N=3), religious (N=1), thought alienation (N=1), reference (N=1), and delusional jealousy (N=1) themes noted. Hallucinations were reported for 10 patients (62.5%)—auditory (N=8) and visual (N=7)—and one patient also reported olfactory, tactile, and gustatory perceptual disturbance. Disorganized thought disorder was described in 11 cases (68.8%). Affective lability (N=11, 68.8%) was more common than sustained depressed mood (N=1, 6.3%) or elevated mood (N=2, 12.5%). Affect was most commonly reported as perplexed (N=9, 56.3%). Other psychiatric features described include change in personality (N=13, 81.3%), anxiety (N=12, 75%), odd/unusual behavior (N=11, 68.8%), and pathological laughter/crying (N=4, 25%). Agitation (N=13, 81.3%), suicidal ideation/attempts (N=4, 25%), and aggression/harm to others (N=4, 25%) were frequently reported. These psychiatric features were all present on hospital admission, with only one patient showing further symptom development (auditory hallucinations commencing during the second week of admission).

Comparatively, five of the 12 patients with catatonia presented with symptoms on admission, and the other seven developed symptoms during the first 2 weeks in the hospital. The mean average on the Bush-Francis Catatonia Rating Scale at presentation was 4.3 [SD=3.4], with a peak at 10.9 [SD=4.4], at a mean average of 10.1 days (SD=10.5). The most common catatonic symptoms were as follows (in descending order of their frequency): immobility, perseveration, staring, withdrawal, impulsivity, mannerisms, stereotypy, increased activity/excitement, and mutism. Catatonia was reported as rapidly fluctuating in six patients (54.5% of catatonia cases).

Other key features on hospital admission were disturbance of speech and cognition. Speech disturbance (N=12, 75%) was most commonly described as reduced output (N=8), dysphasic speech (N=8), or mutism (N=5), with some patients having more than one disturbance. Cognitive deficits on admission were observed in 14 patients (87.5%), with some patients having more than one deficit: confusion (N=12), memory disturbance (N=10), executive dysfunction (N=10), and disorientation (N=7). In an additional case, significant memory and executive dysfunction were described 14 days after admission.

Further progression of illness with neurological symptoms occurred in 13 patients (81.3%). Seizures developed in 10 patients (62.5%), with a mean average time of 20.1 days (SD=20.4) from hospital admission. Movement disorders were noted in 11 patients (68.8%), with a mean average time of 15.6 days (SD=10.9) from hospital admission, including dystonia (N=7), rigidity (N=7), orofacial dyskinesia (N=5), hyperreflexia (N=5), and chorea (N=4). Autonomic disturbance was reported in nine patients (56.3%)—most commonly alterations to blood pressure (N=5), tachycardia (N=4), and hypoventilation (N=4)—with a mean average time of 16.2 days (SD=7.2) from hospital admission. Episodic or sustained decreased consciousness occurred in seven patients (43.8%). All anti-NMDAR encephalitis symptom clusters were present in six patients (37.5%), and in two of these cases (12.5%), the symptoms were not neurological or autonomic and considered isolated psychiatric presentations.

The mean average time to CSF diagnosis from presentation was 37.1 days (SD=46.9). Both serum and CSF were positive in 15 out of 16 cases (93.8%). It is noteworthy that one patient who was negative on initial CSF NMDAR Ab testing became CSF positive after 90 days. CSF pleocytosis was observed in four patients (25%), elevated protein was observed in two (12.5%), and oligoclonal bands were seen in two out of seven tested. No patient had other antineuronal antibody positivity or an elevated IgG index. Brain MRI was abnormal in five out of 13 patients (38.5%; three patients with temporal T₂-fluid-attenuated inversion recovery abnormality, one patient with cerebellar atrophy, and one patient with hippocampal dysembryoplastic neuroepithelial tumor), and EEG was abnormal in six out of 15 (40%), all with diffuse mild-to-moderate slowing. An ovarian teratoma was found in two patients (12.5%) with no other malignancies.

Fourteen patients (87.5%) were taking antipsychotic medications—most commonly olanzapine (N=8), risperidone (N=6), quetiapine (N=4), and aripiprazole (N=4)—with 10 patients (62.5%) requiring more than one antipsychotic. Extrapyramidal side effects were noted in five patients (31.3%), all of whom were prescribed olanzapine, and suspected neuroleptic malignant syndrome was noted in two cases (12.5%; risperidone, N=1; zuclopenthixol acetate, N=1). There was minimal to no improvement of symptoms with antipsychotics. Lorazepam challenge was given unsuccessfully for five patients with catatonia; of these, two patients showed partial improvement after ECT.

Intensive care admission was required in four cases (25%). One patient did not receive immunotherapy and showed resolution of psychiatric and mild movement disorder symptoms with psychotropics. The other 15 patients were treated with intravenous immunoglobulin (IVIg) (N=14), intravenous steroids (N=8), plasmapheresis (N=4), rituximab (N=9), cyclophosphamide (N=4), or mycophenylate (N=2).

Full recovery was noted in five cases (31.3%), with the remaining patients reporting partial improvement (N=11, 68.8%). Ongoing symptoms included behavioral disturbance (N=8), cognitive deficits (N=6), and neurological deficits (N=3). Amnesia for the acute illness was reported for seven patients (43.8%). Recurrence of encephalitis occurred in four cases (25%). The mean length of follow-up was 573 days (SD=550).

CSF NMDAR Ab Negative Cases

There were eight patients (women, N=5; men, N=3)—with a mean age of 32.4 years (SD=9.8)—in which a serum NMDAR Ab was detected with a negative CSF NMDAR Ab test. The mean average time to initial testing was 2.8 days from index hospital admission (SD=2.8) and CSF testing at 18 days (SD=11), with six out of eight patients having further repeated serum antibody testing later in their illness course. There were an additional three patients who had equivocal serum NMDAR Ab positivity and a negative CSF NMDAR Ab.

The pattern of clinical presentation was clinically distinct in two of these cases, with a clinical picture of encephalitis (Table 1). These two patients presented with an acute onset of symptoms with no psychiatric history but a prodromal flu-like illness and insomnia in the weeks before hospital presentation. Both patients experienced rapid development of catatonia within the first week of admission (peak mean Bush-Francis Catatonia Rating Scale score, 14), and one patient reported rapidly fluctuating catatonia. Both patients described reduced and dysphasic speech as well as significant cognitive (memory and disorientation) deficits. Within the second week of admission, these two patients had developed a movement disorder and features of autonomic instability. Extrapyramidal side effects with antipsychotic medication were reported for both patients. For one patient, the medication was discontinued as a result of suspected neuroleptic malignant syndrome. This patient later required intensive care admission for decreased consciousness. Neither patient had any abnormal investigation (CSF, MRI, or EEG). Both patients received immunotherapy (one with IVIg alone and one with IVIg and steroids) demonstrating rapid improvement of symptoms with only mild cognitive ongoing deficits.

The other six patients reported only psychiatric symptoms with no development of catatonia (mean average peak on the Bush-Francis Catatonia Rating Scale, 1.6), cognitive concerns, or speech disturbance. Two patients had a history of a psychiatric disorder (schizophrenia, N=1; emotionally unstable personality disorder, N=1). Three patients had an abnormal EEG with increased fast activity that was believed to be medication related. There were no abnormal CSF or MRI investigations. There was improvement of symptoms with antipsychotic medications in five of the six cases, with no side effects noted. For the one patient whose symptoms did not improve with antipsychotic medication alone, IVIg was trialed with no benefit. The other patients did not receive immunotherapy. The mean average follow-up time was 319 days (SD=248.9).

There were three patients with equivocal serum NMDAR Ab positivity with a negative CSF antibody test. All three patients had a history of schizophrenia or schizophreniform psychosis and were admitted with a psychotic deterioration with associated depressed mood or anxiety. None of the three developed catatonia, speech disturbance, cognitive deficits, seizures, movement disorders, or autonomic symptoms. MRI and EEG were normal; however, two patients had elevated serum antinuclear antibodies, and the other had elevated IgG4. All patients were successfully treated with antipsychotic medications alone. Given the co-occurring antibodies, these three patients were further excluded from analysis.

Given the small number of CSF-negative/encephalitis-suspected cases, these were excluded from comparison. When comparing CSF NMDAR-Ab-positive and CSF NMDAR-Ab-negative/encephalitis-nonsuspected cases, three significant differences were identified: prodromal cognitive deficits (p=0.0003), speech disturbance (p=0.003), and catatonia (p=0.012). All of these symptoms were more common in the CSF NMDAR-Ab-positive group.

Discussion

Testing for the NMDAR Ab is increasingly performed in psychiatric settings. It is a sensitive and specific test, and false-positive results are thought to be rare when standard transfected cell-based assays are used in experienced laboratories (9, 18, 19). However, antibody presence does not necessarily confirm disease, and the ongoing clinical management of patients with atypical features can be challenging for psychiatrists. This study illustrates a detailed clinical profile that may differentiate patients with anti-NMDAR encephalitis from those with primary psychiatric disorders.

The clinical presentation, progression of illness, response to antipsychotics, and proportion of abnormal investigations of the CSF NMDAR Ab cases reflect previously described cohorts and case series of anti-NMDAR encephalitis (2, 5, 20). The majority responded to immunomodulatory therapy and/or tumor removal with a recurrence rate and recovery symptom profile that were, again, consistent with those for anti-NMDAR encephalitis. There was a higher proportion of patients with an isolated psychiatric presentation, likely attributable to case selection and the smaller numbers of this cohort (3). This may also explain the higher rates of psychosis (5). This study demonstrated that the majority of psychiatric symptoms—such as psychosis, agitation, mood disturbance, and anxiety—occurred on presentation to hospital, with only peak catatonic features developing later in the second week of admission. The psychiatric symptoms did not change until immunomodulation commenced, and over this time, there was high psychiatric risk, with half of the patients reporting thoughts or intent to harm themselves or others. Even with improvement after appropriate treatment, many patients still had behavioral disturbance or cognitive deficits in the extended recovery and rehabilitation phase.

The CSF NMDAR-Ab-negative cases appeared to cluster into two distinct symptom profiles. There were two patients with a clinical syndrome consistent with a classical anti-NMDAR encephalitis presentation with neurological symptoms and autonomic disturbance, responding to immunological therapy after unsuccessful psychiatric treatment. It is possible that for these patients, there may have been later CSF NMDAR Ab positivity that was not assessed, similar to the delayed CSF-positive case described in this cohort. The previously present CSF antibodies may have diminished with preserved serum antibodies, which has been rarely seen in long-term follow-up of patients with anti-NMDAR encephalitis (2). Alternatively, immunoprecipitation of the antibody at the blood-brain barrier has been hypothesized to account for this situation (21). The remaining patients, including those with equivocal NMDAR Ab results, did not develop any further neurological symptoms. It is noteworthy that only one of these patients was trialed, without effect, on immunological therapy, but all of these patients showed improvement with traditional psychiatric care. It is possible that patients with serum NMDAR Ab may be at risk of developing anti-NMDAR encephalitis if there is future disruption of the blood-brain barrier or some other “second hit” phenomena. Additionally, it is possible that these patients may have been tested too early, given the difference in initial timing of the antibody testing between CSF-positive and CSF-negative cases. However, reassuringly, the majority of CSF-negative cases have had subsequent antibody testing.

Interestingly, patients who later developed an encephalitic profile and responded to immunomodulatory therapy, independent of the CSF NMDAR Ab result, were older and had higher rates of prodromal cognitive deficits, catatonia, speech disturbance, and antipsychotic sensitivity, all evident before neurological deterioration. Other features—such as a viral prodrome, insomnia, severe agitation, pathological laughter/crying, unusual behavior, an atypical pattern of psychosis, and imaging—did not discriminate. Although further prospective confirmation is required, this suggests reassurance that, in the setting of a positive serum but negative CSF NMDAR Ab result, if there are no key features of catatonia, speech disturbance, cognitive deficits, or antipsychotic sensitivity, traditional psychiatric care can proceed without delay. Remaining alert to changes in the clinical presentation and repeated antibody testing are important. There may also be a role for EEG evidence of encephalopathy in assisting decision making (2). This is of consequence, as increased duration of untreated or partially treated psychosis and other psychiatric conditions are associated with higher psychiatric morbidity and poorer long-term outcomes (22). Confusion, stigmatization, and fracture of the therapeutic relationship can occur with repeated questioning of a diagnosis. Additionally, trialing immunomodulatory therapy, such as IVIg, in significantly suggestive cases, even without CSF-antibody presence, is a reasonable and appropriate management option (23).

The psychiatric symptom profile in the encephalitic cases likely represents the direct pathological effect of the NMDAR Ab and its impact on other neurotransmitters and neuronal networks, which can be extrapolated from the known effects of ketamine, a noncompetitive NMDAR antagonist. At lower doses, ketamine increases the release of excitatory neurotransmitters such as glutamate and dopamine, which causes agitation, positive psychotic symptoms, impairment of executive function, and repetitive or stereotyped motor behaviors in animal models (10, 24, 25). This represents NMDAR Ab titer in the initial phases of illness and postimmunomodulatory treatment. Higher ketamine doses result in catatonia, sedation, and unresponsiveness (10, 25). Rapidly fluctuating catatonia, between hyperactive and withdrawn/hypoactive states, is an unusual feature and could be pathognomonic for anti-NMDAR encephalitis, potentially representing a fluctuating state of NMDAR dysfunction or associated instability of the inhibitory and excitatory networks normally controlled by NMDA (26).

A number of limitations to the study should be considered. The majority of patients with anti-NMDAR encephalitis were tested in response to severe symptoms or rapid clinical deterioration. In addition, a previous local first-episode psychosis study as well as Royal Australia and New Zealand College of Psychiatrists 2016 guidelines recommending first-episode psychosis neuronal antibody testing may have increased the frequency of testing in this population (6, 27). Therefore, there may be a bias against testing psychogeriatric, chronic psychosis, and mood-predominant presentations. It is important to consider that more than 10% of patients demonstrate a positive CSF/negative serum pattern (5). There was one patient in the present cohort in which rapid and severe deterioration prompted CSF testing, despite a negative serum NMDAR-Ab result. However, in general, patients with negative serum tests are not subjected to lumbar puncture; potentially, this may lead to an underestimation of the prevalence of this disorder and an incomplete understanding of the psychiatric phenotype.

The clinical data were recorded retrospectively, and although we were involved in the care of the majority of patients, missing or inaccurately recorded data may have occurred. The methodology of recording absence of a feature when not documented will potentially underestimate the proportion of that feature, which is particularly significant for cognitive deficits and specific catatonic signs that are often poorly documented. No children were included in this study, which is an important consideration given that around a third of anti-NMDAR encephalitis cases occur in individuals <18 years of age, who may have a different clinical presentation (5, 28). Finally, the uncommon nature of anti-NMDAR encephalitis limits the size of this study cohort, which may affect its clinical application.

This study highlights the importance of careful psychiatric clinical assessment and formulation in determining the ongoing management after NMDAR-Ab testing. We suggest that traditional psychiatric care is appropriate for patients with isolated psychiatric symptoms who have positive serum NMDAR testing when CSF testing is negative and there are no key clinical features suggesting encephalitis. Further larger cohorts are required for confirmation, and we emphasize the importance of ongoing follow-up of the untreated CSF NMDAR-Ab-negative patients. Criteria established to clinically diagnose anti-NMDAR encephalitis independent of antibody status may be beneficial in this situation; however, the applicability of these criteria in a psychiatric setting has not yet been determined (23). Clinicians should remain alert for any clinical changes that are an indication for repeated CSF testing but should not otherwise delay psychiatric care.