Non-linear association between androgen receptor CAG repeat length and risk of male subfertility--a meta-analysis.

The CAG repeat in the androgen receptor (AR) has been widely studied in association with male infertility, but the results are conflicting. In a recent meta-analysis, infertile men had <1 repeat longer CAG stretch than fertile men when analysed in a linear regression model assuming that AR function diminishes with increasing CAG length. However, in vitro, a non-linear activity pattern was recently demonstrated so that ARs containing short and long stretches, respectively, displayed lower activity than the AR of median length. These results prompted us to explore the possible association between CAG number and male infertility risk in a stratified manner on the basis of data from the mentioned meta-analysis and subjects from our clinical unit. The study population included 3915 men, 1831 fertile and 2084 infertile. Data were divided into three categories: CAG<22, CAG 22-23 (reference) and CAG>23 and analysed in a binary logistic regression model. Men with CAG<22 and CAG>23 had 20% increased odds ratio of infertility compared with carriers of the median lengths [for CAG<22: p=0.03, 95% confidence interval (CI): 1.02-1.39; for CAG>23: p=0.02, 95% CI: 1.03-1.44]. These results show that an alternative model to a linear one for the genotype-phenotype association in relation to AR CAG repeats is likely, as lengths close to the median confine lowest risk of infertility.