Introduction: While the clinical utility of FSH is well-established in the diagnosis and management of azoospermic men, the utility of FSH in non-azoospermic men is unclear. First, we characterized the distribution of FSH in fertile and subfertile non-azoospermic men. Second, we sought to determine an optimal FSH threshold that would distinguish between these groups. Methods: We performed a retrospective analysis of men presenting for fertility evaluation between 2002 and 2020. Men with at least two semen analyses (SA) over a one-month period and one FSH level within 6 months of their index SA were included. Men with semen volume < 1.0cc, azoospermia, or men on hormonal medications were excluded. Men with at least one SA with total motile sperm count (TMSC) = 20 million were classified as fertile, while men who had <20 million TMSC on both SAs were classified as subfertile. FSH was evaluated as an independent variable to determine a threshold that would predict fertility status. A multivariable logistic regression using dichotomized FSH with a broad range of FSH thresholds and other clinical variables was used to predict fertility status. We determined the area under the curve (AUC) for each FSH threshold via 5-fold cross-validation, repeated 1000 times. The optimal FSH threshold was selected according to the mean AUC. Results: Among 1876 men, 1115 (59.4%) were fertile and 275 (14.7%) were subfertile. Age, varicocele, and testosterone were similar between groups. Median FSH was 4.00 vs 5.96 (p < 0.001) among fertile vs subfertile men. AUC analysis was performed for FSH thresholds corresponding to the 5th, 10th, 25th, 50th, 75th, 90th, and 95th percentile FSH level for each cohort. While the FSH threshold of 5.6 had the highest AUC for prediction of fertility status (0.663), multiple additional thresholds ranging from 4.0 to 9.4 had very similar AUCs. Positive predictive value of FSH increased with higher FSH thresholds. Conclusions: This is the first large-scale study to characterize FSH values in non-azoospermic men. While there are significant differences in FSH among fertile and subfertile non-azoospermic men, no single FSH cutoff had high accuracy for prediction of fertility status. These findings suggest that the utility of FSH in non-azoospermic men may be greatest in identifying men with extreme FSH values, which may guide further clinical management. SOURCE OF Funding: None
