Lipid-related metabolic indices and in vitro fertilization outcomes in women with polycystic ovary syndrome: a retrospective cohort study with mediation analysis

Abstract

To examine the association between composite lipid-related metabolic indices and embryological outcomes in women with polycystic ovary syndrome (PCOS) undergoing in vitro fertilization/intracytoplasmic sperm injection (IVF/ICSI), and to test whether body mass index (BMI) and serum total cholesterol (TC) mediate this association in opposing directions. Retrospective cohort study with mediation analysis. A total of 800 women with PCOS were categorized by BMI as underweight ( n = 29), normal weight ( n = 378), overweight ( n = 287), and obese ( n = 106). Metabolic and lipid profiles worsened progressively with rising BMI, and all four composite indices—the atherogenic index of plasma (AIP), the triglyceride–glucose (TyG) index, the metabolic score for insulin resistance (METS-IR), and the non–high-density lipoprotein cholesterol to high-density lipoprotein cholesterol ratio (NHHR)—increased across BMI categories (all P < 0.001). The number of available embryos fell from 11.0 in underweight to 5.0 in obese women ( P < 0.001) and high-quality embryos from 7.0 (normal weight) to 5.5 (obese) ( P = 0.028), whereas the available embryo rate was unchanged ( P = 0.170), indicating fewer retrieved oocytes rather than reduced per-oocyte developmental efficiency. In the fully adjusted model, higher BMI independently predicted fewer available embryos (β = − 0.13, 95% CI − 0.21 to − 0.06, P < 0.001; P for trend = 0.002). Higher AIP was associated with fewer available and high-quality embryos in the minimally and intermediately adjusted models but attenuated after full adjustment; the TyG index was inversely associated with high-quality embryos, whereas NHHR and METS-IR showed no consistent association, and no index was associated with the available embryo rate. In mediation analysis, TC consistently acted as a compensatory (opposing) mediator, partially offsetting the adverse effect on high-quality embryos for BMI (proportion mediated − 21.0%), AIP (–17.0%), and the TyG index (–51.2%), and on available embryos for BMI (–10.2%) and AIP (–13.4%) (all indirect effects P < 0.05). Conversely, BMI acted as an amplifying (same-direction) mediator, significantly mediating the adverse effect of AIP on available embryos (proportion mediated 54.2%, 95% CI 13.0% to 179.9%). Thus, within the same exposure–outcome pathway (AIP → number of available embryos), BMI amplified (+ 54.2%) and TC attenuated (–13.4%) the adverse effect. Among 31 women who underwent fresh embryo transfer, no lipid-related index or BMI was significantly associated with clinical pregnancy. In women with PCOS undergoing IVF/ICSI, higher lipid-related metabolic indices, most consistently AIP and higher BMI are associated with fewer available and high-quality embryos. BMI and serum total cholesterol mediate these associations in opposite directions, with BMI amplifying and total cholesterol partially compensating for the adverse effect. These findings support weight optimization as a preconception target and indicate that the contribution of circulating cholesterol is compensatory rather than uniformly harmful.