A Personalized FSH Dosing Strategy for Women with Polycystic Ovary Syndrome Undergoing GnRH Antagonist Protocols

Background: Polycystic ovary syndrome (PCOS) is characterized by substantial inter-individual variability in ovarian sensitivity to recombinant follicle-stimulating hormone (rFSH) during controlled ovarian stimulation (COS). Clinically applicable tools for personalized dosing in this population remain limited. Methods: This retrospective single-center study (2013-2024) analyzed 369 PCOS patients undergoing GnRH antagonist protocols who achieved optimal ovarian responses (10-20 oocytes with at least 40% of follicles ≥ 16 mm in diameter on trigger day). The final retrospective dataset was randomly split into modeling (n = 258) and validation (n = 111) groups. A multivariate linear regression model incorporating age, BMI, basal FSH, basal LH, AMH, and AFC was developed to estimate the average daily rFSH dose. Model performance was evaluated using correlation analysis, prediction error metrics, and calibration assessment. Results: Age, BMI, and basal FSH were positively associated with average daily rFSH dose, whereas basal LH, AMH, and AFC were negatively associated. The model explained 40.4% of the variability in average daily rFSH dose. In the modeling cohort, 77.9% of estimated doses fell within ±20% of the observed values, with a moderate correlation between predicted and observed doses (ρ = 0.646). In the validation cohort, 67.6% of estimates met the predefined accuracy threshold (ρ = 0.676). Calibration analyses demonstrated robust agreement between predicted and observed doses. Conclusions: By integrating endocrine markers, ovarian reserve indicators, and clinical characteristics, this study provides a practical example of personalized medicine in COS in women with PCOS. The internally validated approach may support individualized rFSH dosing during COS and serve as a basis for future development of decision support tools in this specific population.