PCOS Linked to Higher Miscarriage Risk After IVF/ICSI

Key Takeaways

• Among women who conceived via IVF/ICSI, miscarriage was more frequent in those with PCOS than in controls, at 18.4% versus 11%.
• The excess pregnancy loss in the PCOS group was reported to be driven mainly by early miscarriage.
• Biochemical pregnancy was higher and live birth was lower in the PCOS group than in controls.
• PCOS remained significantly associated with miscarriage after multivariable adjustment for maternal and treatment-related factors.
• Mendelian randomization analyses pointed in the same direction as the clinical cohort, and sensitivity analyses did not show significant heterogeneity or horizontal pleiotropy.

Whether miscarriage risk in polycystic ovary syndrome (PCOS) reflects the syndrome itself beyond age, adiposity, and the fertility-treatment setting remains an important clinical question. In pregnancies achieved through in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI), separating syndrome-related risk from treatment-related influences is especially relevant. This question framed an evaluation of both pregnancy outcomes in an assisted-reproduction cohort and a genetic causal-inference analysis of PCOS liability and miscarriage risk.

In the PCOS and miscarriage risk analysis, investigators conducted a retrospective cohort of 445 women who conceived via IVF or ICSI, including 191 women with PCOS and 254 controls, with miscarriage as the primary outcome. Logistic regression adjusted for maternal age, body mass index (BMI), endometrial thickness on human chorionic gonadotropin (hCG) day, number of embryos transferred, and embryo transfer and fertilization methods. A second component used two-sample Mendelian randomization (MR) to examine whether genetically predicted PCOS liability tracked with sporadic miscarriage risk, and subgroup analyses explored the association across key demographic and clinical variables.

Miscarriage occurred in 18.4% of women with PCOS versus 11% of controls (p=0.029), and the difference was reported to be driven mainly by early miscarriage at 14.2% versus 8.3% (p=0.048). Other reproductive outcomes moved in the same direction, with higher biochemical pregnancy in the PCOS group at 5.2% versus 1.2% (p=0.012) and lower live birth at 73.3% versus 81.9% (p=0.03).

After multivariable adjustment, PCOS remained associated with miscarriage, with an adjusted odds ratio of 1.851 (95% CI 1.019-3.362; p=0.043). In the genetic component, the IVW MR estimate for PCOS liability and sporadic miscarriage was OR 1.08 (95% CI 1.01-1.14; p=0.0203). The replicated MR analysis showed similar effect sizes, and sensitivity analyses found no significant heterogeneity or horizontal pleiotropy.

These findings are bounded to women who conceived through IVF or ICSI, so the same magnitude of miscarriage risk should not be assumed across all pregnancies. Subgroup analyses were described as directionally consistent across key demographic and clinical variables, but not every stratum was statistically significant. For U.S. clinicians, the results are most directly relevant to counseling within assisted-reproduction settings rather than to spontaneous conception. Overall, the study linked PCOS with higher miscarriage risk in this IVF/ICSI cohort, and that association persisted after multivariable adjustment. The Mendelian randomization findings pointed in the same direction, supporting a possible causal signal without establishing causality across all pregnancy settings.

Clinician Questions

Which factors were accounted for before PCOS remained linked to miscarriage?

The adjusted analysis accounted for maternal age, body mass index, endometrial thickness on hCG day, number of embryos transferred, and embryo transfer and fertilization methods. The persistence of the association after adjustment suggests that the reported signal was not fully explained by those measured maternal and treatment-related factors.

How did the Mendelian randomization results support the PCOS-miscarriage association?

The Mendelian randomization analysis linked genetically predicted PCOS liability with sporadic miscarriage risk, adding a causal-inference signal that aligned directionally with the cohort findings. The replicated MR analysis showed similar effect sizes, and sensitivity analyses found no significant heterogeneity or horizontal pleiotropy, which supports the robustness of the genetic signal without proving causality on its own.

Were the subgroup findings uniform across demographic and clinical strata in women with PCOS?

Subgroup analyses were described as showing consistent directional trends across key demographic and clinical variables, but some strata were not statistically significant.