A Prediction Model of Sperm Retrieval in Males with Idiopathic Non-obstructive Azoospermia for Microdissection Testicular Sperm Extraction.

Patients with Idiopathic non-obstructive azoospermia (iNOA) can achieve fertility by extracting testicular sperm through microdissection testicular sperm extraction (mTESE). But more than half of iNOA patients still cannot benefit from mTESE. In recent years, some studies had reported that serum hormones may be related to the outcome of sperm retrieval, but few had been verified. We hope to obtain a predictive method that is convenient for clinical application and can help judge the outcome of sperm extraction before implementing mTESE. We performed a retrospective analysis of NOA patients who underwent mTESE in the same andrology center from June 2020 to November 2022. A total of 261 patients with complete data were collected, logistic regression analysis was performed and a predictive model was constructed. Then, from December 2022 to May 2023, one prospective cohort of 48 NOA patients who met the inclusion criteria from the same center was recruited to validate the risk prediction model. We successfully constructed a logistic regression model to predict the outcome of iNOA patients undergoing mTESE and found that higher serum anti-Müllerian hormone (AMH) levels were associated with failure sperm retrieval, resulting in an AMH cut-off of 2.60 ng/ml. The area under the receiver operating curve was 0.811, the sensitivity was 0.870, and the specificity was 0.705. Decision curve analysis demonstrated that the threshold probability was above 4%, and unnecessary mTESE could be reduced using this model. In a prospective cohort at the same center, 85.42% (41/48) of iNOA patients correctly identified the mTESE outcome using this model. A logistic regression model with AMH as an independent predictor can predict mTESE outcomes in iNOA patients. Preoperative selection of mTESE in patients with iNOA using this model had clinical benefit in reducing unnecessary surgery. The model demonstrated good accuracy in a small prospective cohort validation. [ABSTRACT FROM AUTHOR]