GLP-1 Drugs for PCOS: Eleven Trials, and Two Analyses That Disagree

No GLP-1 medicine is approved anywhere for polycystic ovary syndrome. Every prescription written for it is off-label, resting on a randomized evidence base that amounts to eleven trials.^[1] What those trials show is a modest weight effect graded low-certainty, an insulin-sensitivity effect that looks better than metformin’s, and — depending which analysis you read — either a meaningful improvement in periods and pregnancy rates, or not enough evidence to say. Both readings are published. The newer and more careful one is the cautious one.

What PCOS treatment is actually aiming at

This matters before any number, because PCOS is not one problem and a drug can help with one part while doing nothing for another. The things people want changed are usually some combination of irregular or absent periods, difficulty conceiving, insulin resistance, excess hair growth, and weight that is unusually hard to shift.

Our PCOS condition page covers the syndrome itself; this one stays on the drug evidence. Weight loss is the intervention with the longest history here, because losing a relatively small amount often restores ovulation. That is the mechanism through which a weight-loss drug could plausibly help with the reproductive side at all — not a direct effect on the ovary. Keep that in mind reading what follows, because it is what the trials are really testing.

What the newest review found

In 2026 the European Journal of Endocrinology published a registered systematic review and meta-analysis of GLP-1 drugs in women with PCOS. It screened 9,654 studies, read 365 in full, and found 11 randomized trials worth including.^[1]

That bottom row is worth reading twice. Not “no benefit found” — nobody has looked. For a condition where the psychological burden is a large part of what people want treated, and where cost decides who can have the drug at all, the literature is silent.

The analysis that reads more positively

A 2023 meta-analysis in BMC Endocrine Disorders pooled 11 randomized trials covering 840 women and reported something the 2026 review did not: an improvement in the natural pregnancy rate, at a risk ratio of 1.72 (95% CI 1.22 to 2.43).^[2] It also reported improvements in menstrual cyclicity, insulin sensitivity, body measurements and hormonal indexes.

So two pooled analyses of a similar-sized literature reach different levels of confidence about the reproductive outcomes. That is not a scandal; it is what happens when the underlying trials are small and measure things differently. The reason to weight the more cautious one is not that it is newer. It is that it registered its protocol in advance and graded the certainty of each finding, which is the machinery that stops a review from reading its own hopes into thin data.

Two analyses, one literature, different verdicts on the outcome people care most about.

Against metformin, which is what most people are already on

Metformin has been the default drug in PCOS for decades, so the practical question is rarely “GLP-1 or nothing”. A 2019 meta-analysis of 8 randomized trials compared them directly.^[3] Whether taking the two together adds anything is a separate question, covered in metformin and a GLP-1 together.

• Insulin sensitivity improved more on a GLP-1 (standardized mean difference −0.40, 95% CI −0.74 to −0.06).
• BMI fell further on a GLP-1 (−1.02, 95% CI −1.85 to −0.19), as did abdominal girth.
• Nausea and headache were more common on the GLP-1 than on metformin.
• Everything else measured showed no significant difference between them.

The authors’ own summary is that a GLP-1 might be a good choice for women with PCOS and obesity, particularly with insulin resistance — and that the available evidence is inconclusive.^[3] Both halves are theirs.

The timing problem

Here is the part that gets left out, and it is the part most likely to matter to you.

PCOS is among the most common causes of infertility, so a large share of the women taking these drugs for it are doing so because they want to conceive. The Wegovy label puts a number on the gap it wants: two months, minimum, between the last dose and trying — semaglutide clears slowly, and the label sets that interval for exactly that reason.^[4]

There is a second-order version of the same issue. Weight loss can restore ovulation before anyone plans for it, and cycles that were unpredictable for years becoming regular is exactly the situation in which an unplanned pregnancy happens. If that is not what you want yet, contraception is part of starting the drug, not a later conversation — and if you take an oral contraceptive, our birth control piece covers which GLP-1 interferes with it and for how long.

What nobody can tell you yet

• Whether it helps the symptoms you may care about most. Hirsutism and menstrual regularity both sit in the insufficient-evidence column of the newest review.^[1]
• Whether it helps how you feel. No trial has measured quality of life or mental health in this population at all.^[1]
• What happens long term. The trials are short. PCOS is not.
• Whether it is worth the money for you. No cost-effectiveness study exists,^[1] and off-label use makes coverage harder — our insurance section tracks what plans publish and the prior authorization letter generator assembles a submission.