In March 2013, we published a study in JAMA Psychiatry assessing the prevalence of anti–N-methyl-D-aspartate glutamate receptor (NMDAR) serum autoantibodies in unmedicated acutely ill patients with schizophrenia, major depression (MD), borderline personality disorder (BLPD), and healthy control individuals without neuropsychiatric disorder.1 The laboratory analyses of this set of samples were done in 2010. N-methyl-D-aspartate glutamate receptor antibodies were identified in 15 patients, primarily in those with a clinical diagnosis of schizophrenia (12 of 121, 9.9%) as opposed to those with MD (2 of 70, 2.9%), those with BLPD (0 of 38, 0%), and control individuals (1 of 230, 0.4%). Two patients were initially misdiagnosed as having catatonic or disorganized schizophrenia. Retrospectively, these cases were reclassified as cases with NMDAR encephalitis (presence of specific IgG anti-NR1a autoantibodies in both serum and cerebrospinal fluid).
Driven by a recent publication of Hammer et al2 showing a higher seroprevalence of anti-NMDAR autoantibodies in healthy control individuals compared with our publication (137 of 1272, 10.8% vs 1 of 230, 0.4%), we set out to assess whether this finding might be reproducible in our local center. Accordingly, we reanalyzed all currently available blood samples from acutely ill unmedicated psychiatric patients and matched control individuals from the scientific biobank at the University of Magdeburg’s Department of Psychiatry (Table, eTable in Supplement).1 The sample collection was approved by the University of Magdeburg ethics committee and written informed consent was obtained from patients.1 Current commercial indirect immunofluorescence BIOCHIP assays (Neurology-Mosaic-6 and Neurology-Validation-Mosaic) were applied at the reference laboratory of EUROIMMUN.
The Table summarizes the results from cell-based assays (CBAs) with fixed NR1a-expressing cells. The percentage of blood samples with IgA, IgG, or IgM NR1a anti-NMDAR autoantibodies again tended to be higher in patients with schizophrenia (18 of 184, 9.9%) compared with those with MD (5 of 99, 5.1%), those with BLPD (1 of 42, 2.4%), and healthy control individuals (25 of 357, 7.0%). However, the diagnostic group difference were not statistically significant (χ23,682 = 0.260). This finding was mainly attributable to the detection of a higher seroprevalence of NMDAR antibodies in healthy control participants than in our previous publication.1 The results from respective CBAs with fixed NR2a- or NR2b-expressing cells were negative in all tested samples.
Random sampling effects might have contributed to the different results in previously (2010) and recently (2013) tested healthy control samples. However, even if the prevalence of anti-NMDAR autoantibodies was higher in the nonoverlapping vs overlapping control cohort (8.7% vs 4.0%, Table), more antibody-positive cases have been detected in the retested control subcohort than previously (4.0% vs 0.4%). Thus, alternatively, it is conceivable that an improved sensitivity of recent commercial indirect immunofluorescence BIOCHIP assays (EUROIMMUN) used now might underlie this difference. Overall, the test-retest agreement regarding the presence or absence of NMDAR antibodies in the peripheral blood was 95.0% (211 of 222) among all patients with psychiatric disorders and 94.4% (119 of 126) in healthy control individuals. Within the group of patients with psychiatric disorders, we observed a test-retest reliability of 92.3% (108 of 117) in schizophrenia, 98.6% (69 of 70) in MD, and 97.1% (34 of 35) in BLPD. Supporting the idea of an influence of assay sensitivity, anti-NMDAR autoantibody titers of 1:320 or greater were more likely (test-retest reliability of 83.3%) than low antibody titers (1:10, 1:32, and 1:100; test-retest reliability of 10.5%) to show reproducibility in the present test and previous analyses.
Using up-to-date commercial assays, IgA, IgG, and IgM NR1a anti-NMDAR autoantibodies showed a broader distribution than previously assumed and can be detected in the peripheral blood of a considerable percentage of healthy control individuals.2 Notably, low IgG anti-NR1a autoantibody titers (1:10 and 1:32) have been observed in 2 healthy individuals from our retest sample. These antibodies were previously believed to be 100% specific for NMDAR encephalitis.3 However, in line with recent observations of antibodies found in lupus erythematosus (antinuclear antibodies) or type 1 diabetes mellitus (islet cell and insulin antibodies), low serum autoantibody titers can be present in a considerable percentage of healthy people, while constantly higher autoantibody titers predispose to the clinical manifestation of an autoimmune disorder.4,5 Future research is needed to define a cutoff threshold for NMDAR autoantibody titers. Notably, this threshold may vary depending on the test systems applied. For instance, CBAs using fixed cells (eg, EUROIMMUN BIOCHIP assay system) appear to be more sensitive than CBAs using live transfected human embryonic kidney cells.6 On the other hand, fixed cells do not have intact neuronal cell surfaces. Therefore, the NMDAR autoantibody binding seen in these cells could be less specifically targeted against cell surface epitopes of the NMDAR (intracellular epitopes may be detected).
Based on previous publications, we assumed that the presence of disorganized or catatonic symptoms in young female psychiatric patients should be considered as an alarm signal to think of NMDAR encephalitis as differential diagnosis, particularly when patients do not respond sufficiently to neuroleptic medication.1,7 However, blood IgA, IgG, or IgM NR1a anti-NMDAR autoantibody titers do not per se justify the diagnosis of NMDAR encephalitis and the initiation of immunosuppressive treatment. An interplay of several factors, such as blood-brain barrier integrity and immunological activity, is apparently crucial for the development of central nervous system pathology and the consecutive emergence of mental illness in anti-NMDAR–seropositive cases.2 Accordingly, data from patients with NMDAR encephalitis suggest that cerebrospinal fluid anti-NMDAR autoantibody titers show a much better correlation with clinical symptoms than blood autoantibody titers.1,3,6
Corresponding Author: Johann Steiner, MD, PhD, Department of Psychiatry, University of Magdeburg, Leipziger Str 44, D-39120 Magdeburg, Germany (firstname.lastname@example.org).
Published Online: May 28, 2014. doi:10.1001/jamapsychiatry.2014.469.
Author Contributions: Drs Steiner and Schiltz had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Steiner, Schiltz, Bernstein, Stoecker, Bogerts.
Acquisition, analysis, or interpretation of data: Steiner, Teegen, Schiltz, Stoecker.
Drafting of the manuscript: Steiner, Teegen, Schiltz, Bernstein.
Critical revision of the manuscript for important intellectual content: Steiner, Schiltz, Stoecker, Bogerts.
Statistical analysis: Schiltz.
Administrative, technical, or material support: Teegen, Schiltz, Stoecker, Bogerts.
Study supervision: Steiner, Bernstein, Stoecker.
Conflict of Interest Disclosures: Drs Teegen and Stoecker are full-time employees of and hold stock in EUROIMMUN AG. No other disclosures were reported.
Additional Contributions: We are very grateful to Henrik Dobrowolny, Dipl-Ing (Department of Psychiatry, University of Magdeburg), for his skillful assistance in database management and statistical analysis.
eTable. Demographic and Clinical Data of Patients and Individuals Who Served as Controls
Steiner J, Teegen B, Schiltz K, Bernstein H, Stoecker W, Bogerts B. Prevalence of N-Methyl-D-Aspartate Receptor Autoantibodies in the Peripheral BloodHealthy Control Samples Revisited. JAMA Psychiatry. 2014;71(7):838-839. doi:10.1001/jamapsychiatry.2014.469