Background: This study was undertaken to determine the optimal cut-off value for FSH to predict the presence of spermatogenesis in patients with non-obstructive azoospermia.
Methods: A total of 206 non-obstructive azoospermic men were enrolled in this prospective study. By using receiver operating characteristic (ROC) curves, we determined the optimal cut-off value for FSH and evaluated whether the test could adequately predict successful sperm retrieval.
Results: There were 108 non-obstructive azoospermic patients who had evidence of spermatogenesis (group A) and achieved success in sperm retrieval. Another 98 non-obstructive azoospermic patients (group B) failed in sperm retrieval. The mean value of serum FSH in group B was significantly higher than in group A (28.03 +/- 14.56 mIU/mL vs 7.94 +/- 4.95 mIU/mL, p < 0.01; respectively). The area under the receiver operating characteristic curves were 0.939 +/- 0.02 and a cut-off value of 19.4 mIU/mL discriminated between group A and B with a sensitivity of 70%. The positive predictive value for failed sperm retrieval (group B) can reach 100%.
Conclusions: Elevated plasma levels of FSH of more than 19.4 mIU/mL could be used as a reliable criterion for a trial of sperm retrieval from testes in artificial reproductive techniques.