Causal effects of serum testosterone levels on brain volume: a sex-stratified Mendelian randomization study

To assess the causal effects of serum testosterone and sex hormone-binding globulin levels on brain volumetric measurements in women and men.

We performed a sex-stratified two-sample Mendelian randomization study using the random-effects inverse variance-weighted method as the primary analysis method. Sex-specific genetic instruments were obtained from a study with up to 194,453 men and 230,454 women. For testosterone, variants with dominant effects on both total and bioavailable testosterone but no aggregate effect on sex hormone-binding globulin were used as the main genetic instruments. Sex-specific summary-level data for magnetic resonance imaging brain volumetric measurements were obtained from a study with 11,624 women and 10,514 men.

Analyses showed per standard deviation (approximately 3.7 nmol/L) higher testosterone levels in men were suggestively associated with larger gray matter volume (beta = 0.208, 95% confidence interval = 0.067 to 0.349, p = 0.004). The association remained in sensitivity analyses and multivariable analyses. Further analyses showed the effect was mainly act on peripheral cortical gray matter, but not on subcortical gray matter. Testosterone in men was not associated with hippocampal volume. Testosterone in women and sex hormone binding globulin in both sexes had no effect on all outcomes.

Our findings overall support previous evidence that testosterone might have neuroprotective properties in elderly men. Future larger trials with long duration of intervention are warranted to assess the efficacy of testosterone for elderly men with cognitive impairment, especially in those with hypoandrogenism.