Predictive model to estimate the chances of successful sperm retrieval by testicular sperm aspiration in patients with nonobstructive azoospermia

Abstract

Objective: To explore predictors of successful sperm retrieval (SR) and to identify potentially suitable candidates for testicular sperm aspiration (TESA), a more straightforward, less traumatic, and less costly procedure than open surgical SR methods.

Design: Retrospective chart review.

Setting: Academic tertiary medical center.

Patients: A total of 297 patients with nonobstructive azoospermia.

Interventions: All patients underwent full clinical evaluation before undergoing a staged SR procedure, starting with TESA and proceeding to microsurgical testicular sperm extraction (microTESE). Predictors of positive SR with TESA were selected using the least absolute shrinkage and selection operator (LASSO) regression analysis using k-fold cross-validation. The obtained regression coefficients were used to create a predictive model, and a receiver operating characteristic (ROC) curve was obtained to express its predictive ability. Cut-off values for each significant predictor were also identified using ROC analysis.

Main outcome measure(s): Development of a prediction model for positive SR with TESA.

Results: Overall, a positive SR was observed in 23.6% of patients undergoing TESA. Average testis size (P = .017) and serum follicle-stimulating hormone (FSH) level (P < .001) were the significant predictors of positive SR identified by LASSO regression analysis. The predictive model had an AUC of 0.742 with a sensitivity of 73.9% and specificity of 69.9%. Patients presenting with an average testis size >7.75 mL and serum FSH level <8.5 IU/L had a TESA-positive SR of 43%.

Conclusions: TESA may be a suitable alternative to microTESE in selected nonobstructive azoospermia patients presenting with an average testis size >7.75 mL and serum FSH level <8.5 IU/L.

Keywords: Testicular sperm aspiration; microsurgical testicular sperm extraction; predictors; sperm retrieval.