Influence of methodological and patient factors on serum NMDAR IgG antibody detection in psychotic disorders: a meta-analysis of cross-sectional and case-control studies

Abstract

Antibodies targeting the N-methyl-D-aspartate receptor (NMDAR) have been detected in patients with psychosis. However, studies measuring the IgG subclass in serum have provided variable estimates of prevalence, and it is unclear whether these antibodies are more common in patients than controls. Because these inconsistencies could be due to methodological approaches and patient characteristics, we aimed to investigate the effect of these factors on heterogeneity. We searched Web of Science and Ovid (MEDLINE and PsycINFO) for cross-sectional and case-control studies published between Jan 1, 2000, and May 5, 2019, that reported NMDAR IgG antibody seropositivity in patients with psychosis. Pooled proportions and odds ratios (ORs) were derived using random-effects models. We estimated between-study variance (τ2) and the proportion of observed variance due to heterogeneity (I2). We then used univariable random-effects meta-regression analysis to investigate the effect of study factors on heterogeneity of proportions and ORs. Our protocol was registered on PROSPERO (CRD42018099874). Of 1276 articles in the initial search, 28 studies were eligible for inclusion, including 14 cross-sectional studies and 14 case-control studies. In cross-sectional studies, NMDAR IgG antibodies were detected in 0·73% (95% CI 0·09-1·38; I2 56%; p=0·026) of patients with psychosis, and in case-control studies, patients with psychosis were not significantly more likely to be seropositive than healthy individuals (OR 1·57, 95% CI 0·78-3·16; I2 15%; p=0·20). Meta-regression analyses indicated that heterogeneity was significantly associated with assay type across both study designs, illness stage in cross-sectional studies, and study quality in case-control studies. Compared with studies using a fixed cell-based assay, cross-sectional and case-control studies using the live method yielded higher pooled prevalence estimates (0·36% [95% CI -0·23 to 0·95] vs 2·97% [0·70 to 5·25]) and higher ORs (0·65 [0·33 to 1·29] vs 4·43 [1·73 to 11·36]). In cross-sectional studies, the prevalence was higher in exclusively first-episode samples than in multi-episode or mixed samples (2·18% [0·25 to 4·12] vs 0·16% [-0·31 to 0·63]), and in case-control studies, higher ORs were reported in low-quality studies than in high-quality studies (3·80 [1·47 to 9·83] vs 0·72 [0·36 to 1·42]). Higher estimates of NMDAR IgG antibody prevalence have been obtained with the live cell-based assay, and studies using this method find that seropositivity is more common in patients with psychosis than in controls. The effects of illness stage and study quality on heterogeneity were not consistent across study designs, and we provide clear recommendations for clinicians and researchers regarding interpreting these findings. None.