Polycystic ovary syndrome affects roughly 11% to 13% of women, is defined by two of three features — hyperandrogenism, ovulatory dysfunction and polycystic ovarian morphology or a raised anti-Müllerian hormone — and is the leading cause of anovulatory infertility.[1] It is also tightly bound to insulin resistance and to visceral fat, which is why a drug class built for weight keeps being pointed at it.
No GLP-1 receptor agonist is approved for PCOS by any regulator. Every seller on this site dispenses for weight, under a weight indication, and the PCOS evidence sits alongside that purchase rather than inside it. The distance between a PCOS trial and the headline obesity figures in the semaglutide weight-loss article is the first thing worth measuring.
The dedicated PCOS literature is small, and it studies one phenotype
A 2026 systematic review searched three databases for randomized trials of GLP-1 receptor agonists in PCOS against no treatment, placebo, lifestyle intervention, combined oral contraceptives, antiandrogens or metformin. It screened 9,654 records, read 365 in full and included 11 randomized controlled trials. Pooled as an add-on, the class reduced body-mass index by 1.38 kg/m² (95% CI −2.39 to −0.38), graded low certainty.[2]
A 2024 meta-analysis restricted to placebo-controlled trials in women with PCOS and obesity diagnosed by Rotterdam criteria pooled four trials and 176 participants — 103 of them on liraglutide, 23 on semaglutide. Body-mass index fell by 2.42 (95% CI −3.10 to −1.74) and waist circumference by 5.16 cm (95% CI −6.11 to −4.21).[3]
The largest single PCOS trial in the class is a 32-week phase 3 study of liraglutide 3 mg against placebo, with lifestyle intervention in both arms, in 82 women diagnosed by NIH criteria with obesity. Body weight changed by −5.7% on liraglutide against −1.4% on placebo, and 25 of 44 completers reached a 5% reduction against 5 of 23.[4] That is 55 women on the drug. Every number on this page comes from cohorts of that order, in women who had PCOS and overweight or obesity, and none of it describes a lean PCOS phenotype.
Whether anything improves beyond the weight
This is where the literature stops agreeing with itself. The placebo-controlled pooling found total testosterone fell by 1.33 (95% CI −2.55 to −0.12, P = 0.03) while HOMA-IR did not move: mean difference −0.30, 95% CI −0.92 to 0.32, P = 0.35, with total cholesterol also unchanged.[3] A 2026 pooling of seven liraglutide trials in 330 overweight or obese women reached the opposite conclusion on the same measure, with HOMA-IR improving at Hedges’ g −0.52 (95% CI −0.83 to −0.22) alongside a menstrual-frequency effect of g 1.76 (95% CI 0.28 to 3.24) whose heterogeneity the authors call high.[5]
A confidence interval running from 0.28 to 3.24 is not a measurement of menstrual regularity; it is a statement that the trials disagree. The underlying observation is real enough — a 26-week randomized trial of liraglutide 1.8 mg in 72 overweight women with PCOS improved the bleeding ratio by 0.14 more than placebo (95% CI 0.03 to 0.24, P < 0.05), with sex hormone-binding globulin up 7.4 nmol/L and ovarian volume down 1.6 ml.[6] It is the size of the pooled effect, not its direction, that nobody can pin down.
The 2026 systematic review that graded certainty formally came to the blunt version: for glucose, insulin, hirsutism and menstrual regularity, the evidence was insufficient to draw a conclusion, and no trial at all had measured quality of life, mental health or cost-effectiveness.[2] Weight is the outcome that survives grading. The rest is a signal in low-quality data.
Change the comparator and the androgen benefit disappears
Almost every favorable PCOS figure above was produced against placebo or against nothing. Metformin is what these women are usually already offered, and the comparison against it is far less flattering. A 2023 meta-analysis of 11 randomized trials covering 840 women found that GLP-1 monotherapy was not superior to metformin for free testosterone, for DHEAS or for the free androgen index.[7]
A 2026 network meta-analysis put the question to 29 randomized trials and 1,771 women with PCOS, ranking GLP-1 receptor agonists against SGLT-2 inhibitors and DPP-4 inhibitors on the same outcomes. SGLT-2 inhibitors ranked first for body-mass index, at −4.26 kg/m² against placebo (95% CI −6.01 to −2.52, SUCRA 91%), and first for HOMA-IR at −2.56 (95% CI −4.53 to −0.59, SUCRA 89%). Metformin combined with a GLP-1 agonist ranked first for total testosterone and free androgen index. GLP-1 monotherapy carried the highest odds of nausea and vomiting in the network, and certainty was moderate to very low for most comparisons.[8]
So in the only ranking exercise the PCOS literature has produced, the class sold on weight-loss telehealth did not come first on weight or on insulin resistance. It came first as metformin’s partner on androgens, and first on vomiting. The general tolerability picture is in what the trials recorded on side effects.
Fertility: a real signal, and a washout inside every trial
The 2023 pooling of 840 women found natural pregnancy rate improved at a risk ratio of 1.72 (95% CI 1.22 to 2.43, P = 0.002, I² = 0%) — a tight, consistent result. In the same analysis there was no significant difference in total pregnancy rate or in IVF pregnancy rate.[7] A 2023 meta-analysis of nine trials comparing exenatide with metformin in 785 women reported a pregnancy risk ratio of 1.93 (95% CI 1.28 to 2.92) and an ovulation risk ratio of 1.41 (95% CI 1.11 to 1.80), and its authors describe the evidence as inconclusive given the bias in the included studies.[9]
One study separates the fertility effect from the weight loss. A pilot randomized trial enrolled 28 infertile women with PCOS and obesity who had responded poorly to first-line reproductive treatment, assigning them metformin or metformin plus low-dose liraglutide for 12 weeks, followed by a four-week medication-free period before ovarian stimulation. Weight loss was statistically indistinguishable: 7.0 kg against 7.5 kg, P = 0.246. Pregnancy rate per embryo transfer was 85.7% against 28.6% (P = 0.03), and the cumulative 12-month rate 69.2% against 35.7%.[10]
That is the finding that resists the obvious reading. Equal weight loss, very different implantation outcomes, in 28 women at one center. It is far too small to act on and too clean to dismiss, and the authors nominate endometrial receptivity as the mechanism to go look for.
Now the design detail that almost no seller page mentions. A 2025 open-label trial randomized 100 overweight or obese women with PCOS to metformin or metformin plus semaglutide 1 mg weekly for 16 weeks; the combination arm lost 6.09 ± 3.34 kg against 2.25 ± 4.27 kg. Between weeks 16 and 40 every participant was moved to metformin alone, and the natural pregnancy rate over that window was 35% against 15%.[11] The pregnancies were counted after the GLP-1 had been withdrawn — exactly as in the liraglutide pilot, which built in its own four-week gap.[10]
The collision a buyer needs to see
Both trial designs encode the same clinical fact: these drugs are not for use in pregnancy, so a course of treatment has to end before conception is attempted. What the PCOS literature actually supports is a sequence — treat, stop, then conceive — and not a pregnancy achieved on the drug. The labeled timing rules, and what the observational record shows after an unplanned exposure, are set out in the pregnancy and contraception article.
The risk runs the other way too. Restored ovulation is the mechanism behind that pregnancy risk ratio, and it arrives in women whose cycles have been irregular for years and who may have stopped treating contraception as necessary. The menstrual-frequency effect was the largest in the liraglutide pooling and also the least stable,[5][6] which means it can appear fast and without warning in an individual. Hormonal absorption is a separate question from ovulation, and the pharmacokinetic data on it are in the article on sex differences.
A planned stop is also a planned pause, and pausing has a documented consequence: weight returns after the drug is withdrawn. For someone treating PCOS through weight, that pause lands precisely in the months when conception is being attempted.
What is actually approved, and how the old options compare
Nothing in this class carries a PCOS indication, and metformin does not carry one either — it is long-standing off-label practice rather than an approval. The systematic review commissioned to inform the 2023 international PCOS guideline found eleven trials of anti-obesity agents, 545 participants in intervention arms and 451 in control arms, and could pool only two comparisons. Exenatide against metformin showed no difference on anthropometric, biochemical hyperandrogenism or metabolic outcomes, apart from fasting blood glucose, which was lower on metformin (mean difference 0.10 mmol/L, 95% CI 0.02 to 0.17, two trials).[12]
The combined oral contraceptive pill is the other long-standing option, and its head-to-head record against metformin is a Cochrane review of 44 randomized trials in 2,253 women. Metformin may be less effective than the pill for hirsutism in the 25 to 30 kg/m² band (mean difference 1.92, 95% CI 1.21 to 2.64), while below 25 and above 30 the comparison is uncertain. Severe gastrointestinal adverse events were more common on metformin (Peto odds ratio 6.42, 95% CI 2.98 to 13.84) and severe other adverse events less common (Peto odds ratio 0.20, 95% CI 0.09 to 0.44). Evidence quality ranged from very low to low.[13]
A 2026 evidence map of incretin drugs in PCOS summarizes where each molecule stands: liraglutide has the densest PCOS-specific record, semaglutide data are sparse, and tirzepatide has no PCOS-specific evidence at all and cannot be recommended beyond extrapolation from obesity and diabetes trials.[14] Its authors position these drugs as powerful but adjunctive weight-loss agents rather than PCOS-modifying therapies.
What this means at a telehealth checkout
A cash-pay service prescribes for weight. It is not running a PCOS protocol, it is not measuring a free androgen index, and it is not positioned to manage a fertility plan. Compounded semaglutide and tirzepatide, which is what most of these sellers dispense, are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they are dispensed — see what a compounded vial contains and the sellers collected on the compounded semaglutide board.
Three questions belong to the prescriber holding the chart rather than to a checkout page: whether metformin has been tried, what contraception is in place if ovulation returns, and how a planned pregnancy will be sequenced around stopping. How thoroughly an intake asks any of that varies by seller, which is what the provider write-ups record.
The honest summary is narrow. In women with PCOS and obesity, these drugs reduce weight, and the weight reduction is the finding that survives quality grading. Androgens improve against placebo and not reliably against metformin. Insulin resistance improves in some poolings and not others. Ovulation and natural conception improve in short trials that stopped the drug before anyone conceived. None of that is an approval, and none of it is a reason to start or stop a medication without a prescriber.