Skeletal traits are exemplars of sexual dimorphism, with consistent sex differences across populations in bone mineral density (BMD), bone geometry, and osteoporotic fracture risk.(1–3) Heritability of these traits has also been demonstrated, and many common single nucleotide polymorphisms (SNP) have pointed to genes accounting for variability in BMD and fracture risk.(4–6) These observations beg the question whether genetic variation affects the female skeleton differently than it does the male. Because testing such gene-by-sex interactions for common SNPs in genome-wide association studies requires extraordinarily large sample sizes, the analysis has only recently become possible with the collaboration of investigators in the GEFOS consortium and other independent studies.(7) Their report follows on recent triumphs such as the GWAS meta-analysis of BMD in more than 80,000 men and women in GEFOS/GENOMOS cohorts(6) that identified 56 genes, several of which were also associated with fracture. Gene-by-sex interactions could identify new signals not detected in the sex-combined analyses and indicate important sex-dependent skeletal biology. Liu and colleagues(7) report that, in fact, no sex differences in the effects of autosomal SNPs on BMD were found in an analysis of more than 50,000 men and women. The contributions of this well designed study and what remains to be explored—beyond SNPs and BMD—to explain genetic causes of skeletal sexual dimorphism are the focus of our discussion. Multiple reports of sex differences in heritability and genetic associations with human skeletal traits have been reported (reviewed in Karasik and Ferrari(8)), including a large meta-analysis of BMD quantitative trait loci (QTL) in humans and SNP association studies for important candidates such as LRP5 and ESR1. There are well recognized limitations of this body of evidence, several of which are addressed by the study from Liu and colleagues.(7) First, although genetic associations discovered in only one sex might, in fact, point to sex-dependent effects, they may also be the result of a lack of power to detect what is truly the same underlying effect in each of the sex strata. In many cases, sex-specific effects have been reported without formally testing the difference in effect between the sexes. In the current study, the investigators have conducted the most comprehensive and well designed study to test that genetic polymorphisms have a different effect on BMD in men than they have in women. Second, as with many studies of sex differences, few replication studies have been published.(9) Publication bias is a well recognized impediment to the consideration of negative results (ie, findings of no sex differences in QTLs or genetic associations) and, in fact, a few other studies have found no significant sex differences in BMD genetic variance(10) or in genotypic effects on BMD.(7) This bias is a problem that meta-analyses, such as the one by Liu and colleagues, take strides against by aligning multiple cohorts and reporting no evidence of gene-by-sex interactions. It is tempting to imagine that with even larger sample sizes statistically significant gene-by-sex interactions might emerge in human cohorts. However, given the small detectable effect sizes noted by Liu and colleagues (explaining <1% of variation in BMD), it is likely to be more rewarding to turn to explanations for the dimorphism that are not based on differential effects of common autosomal SNPs between the sexes. If sequence analysis delivers on the promise of larger effect sizes, it might indeed be possible to document gene-by-sex interactions in future studies. Studies of several genetic mechanisms that were not explored in the study by Liu and colleagues might provide additional insight.
Read full abstract