Development and Validation of a Clinical Pregnancy Failure Prediction Model for Poor Ovarian Responders During IVF/ICSI

Abstract

Backgrounds: Despite the great advances in assisted reproductive technology (ART), poor ovarian response (POR) is still one of the most challenging tasks in reproductive medicine. This predictive model we developed aims to predict the individual probability of clinical pregnancy failure for poor ovarian responders (PORs) under in vitro fertilization/intracytoplasmic sperm injection (IVF/ICSI).

Methods: The nomogram was developed in 281 patients with POR according to the Bologna criteria from January 2016 to December 2019, with 179 in the training group and 102 in the validation group. Univariate and multivariate logistic regression analyses were used to identify characteristics that were associated with clinical pregnancy failure. The nomogram was constructed based on regression coefficients. Performance was evaluated using both calibration and discrimination.

Results: Age >35 years, body mass index (BMI) >24 kg/m², basic follicle-stimulating hormone (FSH) >10 mIU/ml, basic E2 >60 pg/ml, type B or C of endometrium on human chorionic gonadotropin (hCG) day, and the number of high-quality embryos <2 were associated with pregnancy failure of POR patients. The area under the receiver operating characteristic curve (AUC) of the training set is 0.786 (95% confidence interval (CI): 0.710-0.861), and AUC in the validation set is 0.748 (95% CI: 0.668-0.827), showing a satisfactory goodness of fit and discrimination ability in this nomogram.

Conclusion: Our nomogram can predict the probability of clinical pregnancy failure in PORs before embryo transfer in IVF/ICSI procedure, to help practitioners make appropriate clinical decisions and to help infertile couples manage their expectations.

Keywords: IVF/ICSI; clinical pregnancy failure; nomogram; poor ovarian response; predictive model.