Sex effects on DNA methylation affect discovery in epigenome-wide association study of schizophrenia

Abstract Sex differences in the epidemiology and clinical characteristics of schizophrenia are well-known; however, the molecular mechanisms underlying these differences remain unclear. Further, the potential advantages of sex-stratified meta-analyses of epigenome-wide association studies (EWAS) of schizophrenia have not been investigated. Here, we performed sex-stratified EWAS meta-analyses to investigate whether sex stratification improves discovery, and to identify differentially methylated positions (DMPs) and regions (DMRs) in schizophrenia. Peripheral blood-derived DNA methylation data from 1519 cases of schizophrenia (male n = 989, female n = 530) and 1723 controls (male n = 997, female n = 726) from three publicly available datasets, and the TOP cohort were meta-analyzed to compare sex-specific, sex-stratified, and sex-adjusted EWAS. The predictive power of each model was assessed by polymethylation risk score (PMRS). The number of schizophrenia-associated DMPs identified was higher for the sex-stratified model than for the sex-adjusted one (25 vs. 5). We identified 19 schizophrenia-associated DMRs in the sex-stratified analysis. PMRS from sex-stratified analysis outperformed that from sex-adjusted analysis in predicting schizophrenia. Notably, PMRSs from the sex-stratified and female-only analyses, but not those from sex-adjusted or the male-only analyses, significantly predicted schizophrenia in males. The findings suggest that sex-stratified EWAS meta-analyses improve the identification of schizophrenia-associated epigenetic changes and are consistent with the notion that methylation effects in schizophrenia are stronger in females than in males. Sex-specific DNA methylation may have potential implications for precision psychiatry and the development of stratified treatments for schizophrenia.