Predicting live birth for poor ovarian responders: the PROsPeR concept

Area of Science:

• Reproductive Endocrinology
• Infertility Treatment
• Clinical Prediction Modeling

Background:

• Existing live-birth predictive models demonstrate poor performance and are rarely utilized.
• No predictive models specifically address poor ovarian responders (PORs).

Purpose of the Study:

• To develop a clinically applicable tool for predicting live birth in PORs undergoing treatment with recombinant human FSH (r-hFSH).

Main Methods:

• Developed a predictive model using data from the ESPART trial for PORs receiving r-hFSH.
• Created separate models for patients with prior assisted reproductive technology (ART) cycles and ART-naïve patients.
• Derived the simplified Poor Responder Outcome Prediction (PROsPeR) concept.

Main Results:

• The PROsPeR model categorizes PORs into three risk scores, with live-birth rates decreasing across categories.
• Achieved a superior area under the receiver operating characteristic curve (AUC_ROC) of 0.84 (0.79 to 0.88) when adequately calibrated.
• Observed lower discrimination (AUC_ROC 0.66-0.72) when applied to broader patient groups or those receiving additional treatments.

Conclusions:

• The PROsPeR model is specific for PORs receiving r-hFSH.
• Offers an optimal balance of precision and simplicity for routine clinical use.
• Represents the best available tool for predicting live birth in this patient population.