Bidirectional two-sample Mendelian randomization study of differential white blood cell count and schizophrenia

BackgroundSchizophrenia and white blood cell count (WBC) are both complex and polygenic disease/traits. Previous evidence suggested WBC can lead to higher all-cause mortality, and some other evidence found elevated WBC in first-episode psychosis and chronic schizophrenia patients. However, prior observational findings may be confounded by antipsychotic exposures and their effects on WBC. Mendelian randomization (MR) is a useful method to examine the directional causal relationship between schizophrenia and WBC.MethodsWe performed a two-sample MR using summary statistics of the Psychiatric Genomics Consortium Schizophrenia Workgroup (N=130,644) and the Blood Cell Consortium (N=563,085). The MR methods included inverse variance weighted, ME Egger, weighted median, and MR-PRESSO, contamination mixture, and a novel approach called mixture model reciprocal causal inference (MRCI). False discovery rate was employed to correct for multiple testing.ResultsAfter correcting for horizontal pleiotropy, the MRCI method demonstrated that schizophrenia can cause by elevated lymphocyte count (OR=1.080; FDR adjusted p-value=0.026) and eosinophil count (OR=1.050; FDR adjusted p-value=0.026). The contamination mixture method showed that schizophrenia may cause elevated neutrophil counts (Beta=0.011 in unit of standard deviation of mean absolute neutrophil count; FDR adjusted p-value=0.045) and reduction of eosinophil count (Beta=-0.013 in unit of standard deviation of mean absolute eosinophil count; FDR adjusted p-value=0.045). Some further significant findings had been identified by conventional MR approaches and MR-PRESSO, but we interpreted those with cautious due to substantial heterogeneity and plausible pleiotropic effects identified.ConclusionThis MR study provided evidence that schizophrenia has causal relationships with altered differential WBC. Our findings support the role of WBC in influencing schizophrenia risk, and may concur with the hypothesis of neuroinflammation in schizophrenia.