Nomogram based on a circular RNA biomarker for predicting the likelihood of successful sperm retrieval via microdissection testicular sperm extraction in patients with idiopathic non-obstructive azoospermia.

Abstract

Many circular RNAs (circRNAs) are specifically expressed in the testes and seminal plasma of patients with non-obstructive azoospermia (NOA), highlighting them as potential predictors of microdissection testicular sperm extraction (micro-TESE) outcomes. Although research has indicated that circular RNA monoglyceride lipase (circ_MGLL) is highly expressed in the testicular tissues of patients with NOA, the association between circ_MGLL expression and sperm retrieval outcomes (SROs) in patients with idiopathic non-obstructive azoospermia (iNOA) receiving micro-TESE remains unclear. This single-center, retrospective cohort study enrolled 114 patients with iNOA who underwent micro-TESE at Northwest Women's and Children's Hospital from January 2017 to November 2021.A logistic regression model was used to examine associations between SRO and circ_MGLL expression in testicular tissues, the results of which were used in conjunction with previous findings to establish a nomogram. The predictive performance of the circ_MGLL-based nomogram was evaluated via calibration curves, receiver operating characteristic curves, and decision curve analysis (DCA) using an internal validation method. The generalized additive model indicated that the probability of successful SRO for micro-TESE decreased as circ_MGLL expression increased in testicular tissues. Across the entire cohort, univariate logistic regression analysis revealed that circ_MGLL expression was inversely associated with SRO in patients with NOA. This trend did not change after stratification according to age, body mass index, testicular volume, follicle-stimulating hormone (FSH) level, luteinizing hormone (LH) level, testosterone (T) level, or pathological type (or after adjusting for these confounders) (odds ratio <1, P < 0.001). A nomogram was then generated by integrating circ_MGLL, pathological types, and FSH, LH, and T levels. The circ_MGLL-based predictive model achieved satisfactory discrimination, with an area under the curve of 0.857, and the calibration curves demonstrated impressive agreement. The DCA indicated that the net clinical benefit of the circ_MGLL-based predictive model was greater than that of circ_MGLL alone. circ_MGLL is significantly associated with the SRO of micro-TESE in patients with iNOA. The circ_MGLL-based nomogram developed in the current study can predict successful SRO with high accuracy.