Women with polycystic ovary syndrome who added a GLP-1 drug to their treatment were about three times as likely to get pregnant on their own as women on the standard pill, metformin. The same drugs did nothing once those women went into an IVF lab.

That split is the useful part of a new meta-analysis published October 15 in the European Journal of Obstetrics and Gynecology and Reproductive Biology ↗, led by a team spanning Harvard Medical School and several Istanbul fertility centers. It is the first attempt to pool the randomized trials on a question thousands of PCOS patients are already asking their doctors: does a weight-loss shot help me conceive?

What they pooled

Polycystic ovary syndrome is the most common hormonal cause of infertility, and it is tangled up with insulin resistance and weight. GLP-1 drugs attack exactly that metabolic knot, which is why they get prescribed off-label to women trying to conceive. The authors searched the literature through February 2026 and found only three randomized trials, four published reports, totaling 252 women, that tested a GLP-1 drug ↗, which acts through the GLP-1 receptor ↗, against metformin alone, the drug obstetricians have leaned on for PCOS for two decades.

Pooled together, the picture breaks in three. Overall pregnancy, counting every route, did not clearly improve (odds ratio 1.41, meaning a 41 percent relative bump, but with a confidence interval from 0.81 to 2.45 that crosses the line of no effect). Natural conception, women getting pregnant without assisted reproduction, did improve, and sharply: an odds ratio of 3.10, with an interval of 1.74 to 5.49 that stays well clear of no effect. Pregnancies achieved through IVF showed nothing at all (odds ratio 1.01, dead on the null).

Why the split is the story

Read those three numbers together and they tell you something the headline "GLP-1 helps fertility" would flatten. The benefit lives entirely in natural conception. That fits the mechanism. By trimming weight and easing insulin resistance, the drugs appear to restore ovulation, so more women release an egg and conceive the usual way. Once a patient is in an IVF cycle, where ovulation is driven by injected hormones and the egg is retrieved regardless, that lever has nothing left to pull. The drug cannot help with a step the clinic has already taken over.

The authors rated their confidence as moderate for the natural-conception finding, helped by a trial sequential analysis that suggested the result was not a fluke of small numbers, and low for everything else. And there is a hole they name themselves: not one of the trials reported live births. An odds ratio on pregnancy is not an odds ratio on a baby carried to term, and in a population with higher miscarriage risk, that distinction is not academic.

What it changes

This platform covered a different PCOS readout this month ↗, a single Phase 2 where the dual agonist ribupatide cut weight and made menstrual cycles more frequent, with the reproductive gain riding the weight loss. The new meta-analysis is the evidence-grade version of that same idea, now measured against the actual standard of care rather than placebo, and now with pregnancy as the endpoint rather than a proxy. The signal survives the comparison, but only for the women conceiving on their own, and only until someone runs a trial big enough to count the births. For a clinician deciding whether to reach for a GLP-1 drug like liraglutide ↗ over metformin in a PCOS patient, the honest answer from 252 women is: maybe, if she is trying naturally, and not yet proven where it counts most.