Elevated FSH is associated with a higher utilization of assisted reproduction (IUI OR IVF) but similar birth rates after vasectomy reversal

Abstract

OBJECTIVE: We test the hypothesis that elevated pre-op FSH levels predict a lower paternity rate in men with suspected impaired sperm production after vasectomy reversal. DESIGN: A retrospective review of vasectomy reversal patients at our institution. MATERIALS AND METHODS: Preop FSH levels were obtained in 249 of 946 reversal patients chosen because of suspected impaired spermatogenesis for any of the following reasons: no prior proven fertility, history of infertility, or suspected subfertility based on physical exam. Men were divided into normal FSH (<10, n=194) and high FSH (≥ 10, n=55) groups. Multiple logistic regressions evaluated for predictors of assisted reproduction. RESULTS: Mean follow-up was 2.5 ± 3.1 years. Mean FSH for the normal group was 5.1 ± 2.1 U/L and for the high group was 16.1 ± 6.0 U/L. Baseline characteristics were similar for the two groups except for higher LH levels and lower testis volumes in the high FSH group. Post-op live birth rates were similar in both groups (40.2% vs. 32.7%, p=0.32). However, in the high FSH group, there was a higher utilization of assisted reproduction (IUI or IVF, 72.7% vs. 45.3%, p=0.0014) and longer mean time to pregnancy (2.3 vs 1.5 years, p=0.02). On multivariate analysis, only obstructed interval greater than 10 years (OR 1.92, 95% CI: 1.03-3.59) and FSH >10 U/L (OR 1.92, 95%CI: 1.15-5.62) were associated with higher use of assisted reproduction. CONCLUSION: Men with high FSH have similar baseline characteristics except for slightly lower testicular volume and higher LH, which is consistent with poorer testis function. Live birth rates were lower than our previously reported rates which is not surprising since FSH levels were only obtained preoperatively in men with suspected of impaired fertility. Elevated FSH, was associated with higher rates of assisted reproduction, even after controlling for confounding covariates.