PCOS affects roughly 8 to 13% of women of reproductive age and travels with insulin resistance, missed ovulation and infertility. No GLP-1 drug is approved to treat it. Off-label use has been driven partly by the mechanism and partly by the anecdote of women who had been told they were infertile getting pregnant on a weight drug. Read this list knowing how small it is. The randomized ovulation evidence rests on one trial of 60 women from 2008, the only hard fertility endpoint comes from 28 women, and several of the pilots come from a single research group in Slovenia. One finding complicates the simple story. In the IVF trial, both arms lost the same amount of weight and only the liraglutide arm's pregnancy rate moved, which is not what a purely weight-mediated explanation predicts. For most readers, the practical guidance in the last entry matters more than any effect size here: fertility can return before anyone plans for it, and these drugs need to be stopped well before conception.
Ranked papers
#1
Legro RS, Arslanian SA, Ehrmann DA, et al. · J Clin Endocrinol Metab · 2013
What it measured: Clinical practice guideline (diagnostic + therapeutic recommendations)
Not a GLP-1 paper, but the framework everything else is argued inside. The Endocrine Society's PCOS guideline makes hormonal contraceptives first-line for irregular periods and excess hair or acne, and clomiphene first-line for infertility. Whether losing weight improves PCOS itself is, in its words, uncertain, though it still advises lifestyle change for patients with overweight or obesity because of the other health gains. Off-label GLP-1 use rests on the bet that weight loss helps the syndrome too, which is why the IVF result below is interesting: its effect did not track weight.
PMID 24151290 ↗DOI 10.1210/jc.2013-2350 ↗
#2
Elkind-Hirsch K, Marrioneaux O, Bhushan M, et al. · J Clin Endocrinol Metab · 2008
What it measured: How often periods came, and whether ovulation occurred, over 24 weeks
The first randomized signal, from 2008 and still the anchor citation: 60 women with PCOS and overweight, assigned to exenatide, metformin, or both for 24 weeks. The combination beat either drug alone on menstrual frequency (the primary outcome), ovulation rate, androgen levels, insulin sensitivity and weight, and weight fell further in both exenatide groups than on metformin alone. Only 42 of the 60 finished. The trial had sixty women, ran 24 weeks and used a drug nobody prescribes for this now, but every later review traces back through it.
PMID 18460557 ↗DOI 10.1210/jc.2008-0115 ↗
#3
Salamun V, Jensterle M, Janez A, et al. · Eur J Endocrinol · 2018
What it measured: Pregnancy rate per embryo transfer following IVF
The only trial here with a hard fertility endpoint, and the most surprising result on the page. 28 women with obesity, PCOS and prior IVF failure were given metformin or metformin plus low-dose liraglutide for 12 weeks, then a four-week washout before stimulation. Pregnancy per embryo transfer was 85.7% against 28.6%, and cumulative pregnancy over 12 months 69.2% against 35.7%. The critical detail is that both groups lost the same weight, about 7 kg, so whatever produced the difference, it was not weight loss. With twenty-eight women, treat it as the reason to run a real trial, not as a result to act on.
PMID 29703793 ↗DOI 10.1530/EJE-18-0175 ↗
#4
Jensterle M, Salamun V, Kocjan T, et al. · J Ovarian Res · 2015
What it measured: Body weight change at 12 weeks
A 12-week three-arm pilot from the Slovenian group: 45 women with PCOS and obesity on liraglutide 1.2 mg, roflumilast, or metformin. Liraglutide produced the most weight loss, 3.1 kg on average, against 0.2 kg on metformin. Ovulation was not an endpoint. The study shows that the drug beats metformin at removing weight in this population and says nothing directly about fertility.
PMID 26032655 ↗DOI 10.1186/s13048-015-0161-3 ↗
#5
Wang FF, Wu Y, Zhu YH, et al. · Obes Rev · 2018
What it measured: Body weight reduction across pharmacotherapies (network meta-analysis)
The first formal ranking of drug options for weight loss in PCOS, pooling 23 randomized trials and 941 women on metformin, inositol, liraglutide and orlistat. Liraglutide came out on top for weight and waist circumference. It is heavily weighted by the same small trials listed above, so read it as a tidy summary of a thin literature rather than as independent confirmation. It is nonetheless the citation most often used to justify prescribing.
PMID 30066361 ↗DOI 10.1111/obr.12720 ↗
#6
Cena H, Chiovato L, Nappi RE · J Clin Endocrinol Metab · 2020
What it measured: A review of the biology and the clinical evidence — not a trial
The mechanistic reference, and unusually candid about its own limits. It lays out the pathways by which these drugs could affect PCOS beyond weight (insulin excess, androgen levels, the hormonal axis governing ovulation, and GLP-1 receptors expressed in ovarian tissue itself), while stating plainly that the fertility evidence at the time amounted to two randomized trials and no approval. That mix of mechanism and candor is why it remains the standard citation.
PMID 32442310 ↗DOI 10.1210/clinem/dgaa285 ↗
#7
Liao M, Li X, Zhang H, et al. · Endocrine · 2024
What it measured: Weight and waist, hormone levels, and blood protein patterns at 12 weeks
One of the few trials to test a GLP-1 against what women with PCOS are usually given rather than against placebo: 60 women with overweight randomized for 12 weeks to metformin with either a GLP-1 or a combined oral contraceptive. The two treatments split the outcomes. The GLP-1 arm did better on weight, waist, insulin resistance and ovulation, and the contraceptive arm did better on testosterone and other androgen measures. That is useful for choosing between them, because it says the choice depends on which problem is bothering the patient most.
PMID 37653215 ↗DOI 10.1007/s12020-023-03487-4 ↗
#8
Austregésilo de Athayde De Hollanda Morais B, Martins Prizão V, de Moura de Souza M, et al. · J Diabetes Complications · 2024
What it measured: BMI, waist, lipids, testosterone and insulin resistance, pooled across trials
The most current pooled analysis, across four placebo-controlled trials and 176 women, mostly on liraglutide or semaglutide. Against placebo, waist circumference fell 5.2 cm, BMI 2.4 kg/m², and triglycerides and total testosterone fell too, but insulin resistance (HOMA-IR) and total cholesterol did not change significantly. Its own limitations section sums up this field well. It lists small samples, short follow-up, geographically concentrated trials, and essentially no data on sustained ovulation or live births.
PMID 39178623 ↗DOI 10.1016/j.jdiacomp.2024.108834 ↗
#9
Sola-Leyva A, Pathare ADS, Apostolov A, et al. · Acta Obstet Gynecol Scand · 2025
What it measured: Mechanistic review of endometrial and implantation effects
A review of the question nobody's trial measured: what these drugs do to the uterine lining, where implantation happens. GLP-1 receptors are present in endometrial tissue, and the signaling could plausibly help (less inflammation, better blood supply) or interfere with implantation. The direction in humans is unknown. The review is speculative by construction, and it is the clearest articulation of why 'it worked, so it must be safe in pregnancy' does not follow.
PMID 39696822 ↗DOI 10.1111/aogs.15010 ↗
#10
Finkle J, Brost BC · Obstet Gynecol · 2025
What it measured: Clinical review of fertility and pregnancy implications
The entry with the most direct bearing on a reader's decisions. An obstetrics-aligned review that treats the surprise-pregnancy phenomenon as real but unquantified, and converts the evidence into instructions: tell patients fertility may return, discuss contraception before starting, allow at least two months between the last dose and trying to conceive, and stop immediately if a pregnancy occurs. If you read one paper on this page, this is the one that changes what anyone should do.
PMID 39847778 ↗DOI 10.1097/AOG.0000000000005825 ↗
About this list
Obesity medicine throws up the same handful of questions over and over, and we answer them with ranked lists anchored to the papers themselves. On 2026-10-30 every citation below was put back through PubMed. Where a trial has a registry entry, the card carries that link too, alongside the PubMed one.
Browse our full index of research lists or our long-form research articles.