Predicting pregnancy and spermatogenesis by survival analysis during gonadotrophin treatment of gonadotrophin-deficient infertile men

BACKGROUND: Predictors of fertility or spermatogenesis during gonadotrophin therapy of gonadotrophin-deficient men remain poorly defined. METHODS AND RESULTS: In order to evaluate potential predictors, this study evaluated 29 consecutive gonadotrophin-deficient men all desiring paternity who received 43 courses of therapy in one centre between 1982 and 1998. The Kaplan-Meier survival analysis estimates of median (SE) time to a sperm concentration of >0, >5 and >20 x 10(6)/ml were 5.5 (1.1), 12.4 (2.3) and 29.1 (1.9) months respectively. Conception occurred in 22/43 cycles (with eight men achieving two pregnancies) with a median (SE) Kaplan-Meier estimate of 20.5 (4.7) months. The median sperm concentration at conception was 5.0 (SE 2.0; range 0.0-59.5) x 106/ml. Multivariate correlated Cox proportional hazards models predicting these same sperm thresholds and conception were developed by forward stepwise variable selection with verification of the model by backward stepping. Larger testicular volume, prior gonadotrophin therapy, completion of puberty, older age, the absence of adverse fertility factors and the absence of multiple pituitary hormone deficiency predicted a favourable response. Multivariate modelling suggests that the two most important predictors of sperm output are testicular volume and pubertal status. The most important potentially modifiable predictor was prior gonadotrophin therapy. The efficacy of recombinant and urinary FSH were similar. Prior androgen therapy and partner's age did not appear to be significant. CONCLUSIONS: Since prolonged treatment may be required to induce spermatogenesis, attention to these predictors may allow appropriate early use of advanced reproductive technologies.