Meta-analysis of genome-wide dna methylation in schizophrenia: does sex stratification improve discovery?

Sex differences in clinical manifestations, onset, course, and outcome of schizophrenia (SCZ) are well-known, however, research on sex-specific molecular mechanisms is limited. Understanding the molecular mechanisms underlying these sex-specific sub-phenotypes is crucial for precision psychiatry. Here, we aimed to identify sex-specific differentially methylated positions (DMPs) and differentially methylated regions (DMRs) in SCZ and investigate the impact of sex-stratified EWAS meta-analysis on the identification of DMPs and DMRs. We used four DNA methylation datasets of schizophrenia (three publicly available and the TOP cohort) derived from peripheral blood. Our EWAS meta-analyses comprised 1524 cases of SCZ (male = 992, female = 532) and 1729 controls (male = 1001, female = 728). We performed sex-stratified and non-sex-stratified EWAS for each cohort adjusting for smoking scores, age, cell proportions, and methylation PCs. For the total sample, two meta-analyses were carried out, one using sex-stratified EWASs and another using non-sex-stratified ones. Sex-specific meta-analyses were also done. All analyses were performed using the metagen function of the R-package meta. We applied comb-p for genomic regions of 750bp to identify DMRs with at least four probes and Sidak-corrected p-value < 0.001. We examined enrichment of genes annotated to DMPs and DMRs for biological pathways. The number of SCZ-associated DMPs identified was higher for the analysis with sex-stratification EWAS than without (30 vs. 6). We identified 18 SCZ-associated DMRs in the sex-stratified analysis, including DMRs annotated to two imprinted genes BLCAP and NNAT (chr20: 36148457-36149271; 32 probes, Sidak P=1.3E-13), THRB (chr3: 24536077–24537050; 13 probes, Sidak P=2.2E-13) and DIXDC1 (chr11: 111847892–111848400; 8 probes, Sidak P=3.1E-14). In the male-only analysis, we found only one SCZ-associated DMP cg11269166 (METTL8) (P=3.7E-09) and one DMR annotated to WBP1L (chr10: 104535792–104536035, 7 probes, Sidak P=2.9E-06). In female-only analyses, we found one SCZ-associated DMP cg02323699 (NCAM1) (P=7.6E-08) and four DMRs annotated to KDM2B (chr12: 122018897–122019117, 7 probes, Sidak P=1.0E-06), NKAPL and ZKSCAN4 (chr6: 28226885–28227127, 8 probes, Sidak P=9.6E-06), KCNAB3 (chr17: 7832764–7832943, 6 probes, Sidak P=1.4E-05), and SLC44A4 (chr6: 31846797–31847028, 9 probes, Sidak P=2.0E-04). Despite greater statistical power, the findings from the female-only analysis did not reach test-wide significance in the sex-stratified meta-analysis. Our findings suggest that sex-stratified EWAS before meta-analyses may improve the identification of novel SCZ-associated epigenetic changes and hence, advance our understanding of the underlying molecular mechanisms. The identification of female-specific DNA methylation associated with SCZ may partly underlie the sub-phenotypic heterogeneity, and therefore, calls for further research. The identification of sex-specific biomarkers can have potential implications for precision psychiatry and the development of effective treatments for SCZ.