Testosterone-to-Follicle Stimulating Hormone Ratio Predicts Sperm Retrieval Success in Men with Idiopathic Non-Obstructive Azoospermia

OBJECTIVE: To evaluate novel preoperative hormonal predictors of sperm retrieval success in men with idiopathic non-obstructive azoospermia (NOA) undergoing microsurgical testicular sperm extraction (mTESE). DESIGN: Multi-institutional retrospective cohort study conducted between 2014 and 2024. SUBJECTS: 60 men with strictly defined idiopathic NOA from multiple sites participating in the MOBYUS Male Infertility Research Consortium were included, comprising a prediction cohort (n=30) and an independent multi-institutional validation cohort (n=30). Idiopathic NOA was defined by azoospermia on two separate semen analyses without prior paternity, evidence of obstruction, prior genitourinary surgery, documented genetic abnormalities, or use of empirical medical therapy within one year prior to surgery. Sensitivity analyses were performed by excluding men with FSH <7.6 IU/L and men with hypospermatogenesis on pathology. EXPOSURE: Preoperative serum follicle-stimulating hormone (FSH), luteinizing hormone (LH), and total testosterone (T) levels measured within one year prior to mTESE. Hormone-derived ratios, including T/FSH, T/LH, and LH/FSH, were subsequently derived for analysis. MAIN OUTCOME MEASURES: Successful sperm retrieval at mTESE served as the primary outcome. Predictive performance was evaluated using logistic regression to assess hormonal predictors, area under the receiver operating characteristic curve (AUC), and partition analysis to identify optimal clinical thresholds for sperm retrieval prediction. RESULTS: Men with successful sperm retrieval had significantly lower gonadotropin levels and a higher T/FSH ratio (43.0 vs. 20.9, p=0.004). Among all hormonal metrics, T/FSH demonstrated the strongest discriminative ability (AUC=0.81). The significance remained after excluding men with hypospermatogenesis noted on testicular pathology. In external validation, T/FSH was the only variable that maintained statistical significance (p=0.03; AUC=0.74). Partition analysis further revealed a two-step model (T/FSH>20 and LH<10) that achieved 93% sensitivity and 81% specificity for predicting sperm retrieval success. CONCLUSION: This multi-institutional study demonstrates that the T/FSH ratio is a reproducible and clinically meaningful predictor of sperm retrieval success in men with idiopathic NOA. Across two independent and diverse cohorts, this ratio consistently outperformed single-hormone predictors and remained the only externally validated preoperative predictor of mTESE success.