Low testosterone and the risk of dementia in elderly men: Impact of age and education.

OBJECTIVE

The objective of this study was to examine the association of plasma estradiol and testosterone with risk for dementia in elderly men.

METHODS

Within the population based Three-City study, including 3650 men age 65 years and older, a case-cohort design was set up after 4-years of follow-up. Baseline plasma levels of total 17-β estradiol (Total-E2), total testosterone (total-T) and bioavailable testosterone (bio-T) were measured for all cases of incident dementia (n=105) and for a random sample of the cohort (n=413). Cox regression models were used to estimate multivariate steroid sex hormone-associated hazard ratios (HR) and 95% confidence intervals of dementia.

RESULTS

There was a reverse J-shaped relationship between total-T and risk for dementia (P=.007). Compared with the median tertile, the HRs associated with total-T in the lower and upper tertile were increased (HR, 2.33; P=.026; HR, 1.9, P=.126; respectively). Low bio-T was associated with a greater risk for dementia (HR for one standard deviation of decreasing log(bio-T), 1.29; 95% confidence interval, 1.03-1.62). An interaction was found between bio-T and age (P<.0001), and bio-T and education (P=.044). Risk for dementia associated with low bio-T was greater in older men (80 years or older) than in younger men (younger than 80 years; HR, 3.11; P=.011 vs. HR, 1.07, P=.715, respectively) and in men with high level of education compared with those with low level of education (HR, 2.32; P=.0002 vs. HR, 0.95; P=.790, respectively). No significant association was found between Total-E2 and dementia.

CONCLUSIONS

Low levels of testosterone are associated with a risk for dementia in elderly men. The association between low bio-T and dementia may be more relevant to men 80 years or older and men with a high level of education.