An empirical study of sex-stratified vs. sex-combined analysis in the presence of genetic effect heterogeneity

The effect of a genetic variant on a complex trait may differ between female and male, and in the presence of such genetic effect heterogeneity, sex-stratified analysis is often used. For example, genetic effects are sex-specific for testosterone levels, and sex-stratified analysis of testosterone in literature provided easy-to-interpret, sex-specific effect size estimates. However, from the perspective of association testing power, sex-stratified analysis may not be the best approach. As sex-specific genetic effect implies SNPxSex interaction effect, jointly testing SNP main and SNPxSex interaction effects may be more powerful than sex-stratified analysis or the standard main-effect testing approach. Moreover, since individual data may be unavailable, it is then of interest to study if the interaction analysis can be derived from sex-stratified summary statistics. We considered several different sex-combined methods and evaluated them through extensive simulation studies. We observed that a) the joint SNP main and SNPxSex interaction analysis is most robust to a wide range of genetic models, and b) this joint interaction testing result can be obtained by quadratically combining sex-stratified summary statistics (i.e. squared sum of the sex-stratified summary statistics). We then reanalysed the testosterone levels of the UK Biobank data using sex-combined interaction analysis, which identified 27 new loci that were missed by the sex-stratified approach and the standard sex-combined analysis. Finally, we provide supporting association evidence for nine new loci, uniquely identified by the sex-combined interaction analysis, from earlier association studies of either testosterone level or steroid biosynthesis pathway where testosterone is synthesized. We thus recommend sex-combined interaction analysis, particularly for traits with known sex differences, for most powerful association testing, then followed by sex-stratified analysis for effect size estimation and interpretation.