Significant ethnic variation in total and free testosterone concentration

OBJECTIVE Measurement of serum testosterone is an integral part of the assessment of men presenting to endocrine clinics. Little is known about the variation of total bound or bioavailable testosterone by ethnic group. The principal determinant of testosterone bioavailability is SHBG, which itself is a marker for insulin sensitivity. Our aim was to examine variations in testosterone and SHBG levels across three ethnic groups in relation to ethnic differences in insulin sensitivity. DESIGN Men of three ethnic groups living in Manchester, UK, were sampled randomly from population registers being of white European (n = 55), Pakistani (n = 50) and African-Caribbean (AfC) origin (n = 75). Circulating serum testosterone and SHBG concentrations were measured and free testosterone calculated. Insulin sensitivity (HOMA-S) and insulin secretory capacity (HOMA-B) were determined from fasting plasma intact insulin and glucose values. RESULTS Testosterone levels were lower in Pakistani men (mean 14.6 nmol/l, 95% confidence interval 12.6-16.6 nmol/l) than in Europeans (18.7, 16.8-20.6 nmol/l) or AfCs (18.0, 16.4-19.6 nmol/l) (F = 4.8, P = 0.009). Despite SHBG levels also being lower in Pakistani men (22.9, 19.4-26.5 nmol/l) compared with Europeans (28.7, 25.7-31.8 nmol/l) and AfCs (26.9, 23.9-30.0 nmol/l) (F = 3.0, P < 0.05), circulating free testosterone was significantly lower in the Pakistani group (367, 326-408 pmol/l) than in Europeans (455, 416-494 pmol/l) or AfCs (458, 424-492 pmol/l) (F = 6.8, P = 0.001). Pakistani men were on average 4 cm shorter than other groups. However, the lower free testosterone persisted even when adjusted for height or waist-hip ratio. The lower SHBG in the Pakistani men was paralleled by a lower HOMA-S (0.40, 0.25-0.56) compared with Europeans (0.77, 0.61-0.93) and AfCs (0.80, 0.66-0.93) (F = 8.2, P < 0.0001). SHBG correlated positively with HOMA-S (rho = 0.28, P < 0.001) and strongly with total testosterone (rho = 0.54, P < 0.001). There was no difference in insulin secretory capacity (HOMA-B) in Pakistani men compared with Europeans and AfCs. Multiple linear regression analysis showed that total testosterone was independently and negatively related to ln fasting insulin (beta = -0.28, P < 0.001) and age (beta =-0.17, P = 0.02) and positively to ln SHBG (beta = 0.23, P < 0.001) and height (beta = 0.22, P = 0.001). There was no relationship with ethnicity or waist-hip ratio. CONCLUSION Both total bound and calculated free testosterone were lower in Pakistani men. SHBG levels were also lower in Pakistani men, in keeping with poorer insulin sensitivity. We propose that further work is necessary to establish ethnic-specific ranges for the interpretation of total circulating and free testosterone levels in men.