Resveratrol for PCOS: Why the Meta-Analyses Don’t Match

Resveratrol for PCOS: the interesting part is the disagreement

Resveratrol might help some women with PCOS, but the best current lesson is methodological, not therapeutic. Two independent meta-analyses published in 2023 and 2024 pooled randomized trials of resveratrol in PCOS and did not cleanly land in the same place on testosterone. That matters because the entire clinical evidence base is tiny: one review pooled four randomized controlled trials with about 218 participants, while the other pooled three trials with about 169 participants. When a field is this small, a different choice about which study to include, how to standardize hormone units, or how much weight to give one outlier trial can change the headline.

The 2023 review by Fadlalmola et al. is the most recoverable source. It pooled four randomized trials using resveratrol doses from 800 to 1500 mg per day for 40 days to 3 months. The 2024 review by Larik et al. reportedly pooled three randomized trials, but the database full-text record is misassigned to an unrelated paper, so its included trial list, forest plots, and full statistical methods are not available from the record itself. That is not a footnote. It is the main reason the apparent disagreement cannot be fully resolved.

If you came here hoping for a clean yes-or-no answer, PCOS supplement research has other plans.

What the strongest human trial actually found

The most important single study is the U.S. randomized trial described in the 2023 meta-analysis: Banaszewska 2016, registered as NCT01720459. It enrolled 34 women, used 1500 mg of resveratrol per day, and lasted 3 months. Among the trials included in the review, this was both the highest dose and the longest duration. It reported significant reductions in ovarian testosterone and adrenal DHEAS.

That is the best direct clinical signal for resveratrol as an anti-androgen in PCOS. It also fits a plausible biological story. Preclinical and mechanistic reviews, including Karimi et al. 2022 and Shojaei-Zarghani et al. 2022, describe pathways involving insulin sensitization, SIRT1 signaling, and reduced inflammatory signaling that could indirectly lower ovarian androgen production.

But the trial still had 34 participants. That is enough to be interesting and nowhere near enough to be decisive. In a four-trial meta-analysis, one small positive study can act less like a brick in a wall and more like the entire wall.

The lower-dose, shorter trials tell a narrower story

The other identified trials in the 2023 pool were smaller-dose, shorter-duration studies from Iran. In Brenjian 2018, 44 women received 800 mg per day for 40 days. According to Fadlalmola et al., the focus was anti-inflammatory activity, including NF-kB suppression and reduced endoplasmic reticulum stress in granulosa cells. No clear primary testosterone result is recoverable from the available description.

Bahramrezaie 2019 enrolled 62 women and used 800 mg per day, given as 400 mg twice daily, for 40 days. This was not a typical community PCOS sample. The participants were IVF candidates who had already failed at least six cycles of clomiphene or letrozole and three IUI cycles. The study detected sex hormone changes and effects on HIF1 and VEGF gene expression in granulosa cells, again as summarized in the 2023 review. That population is clinically different from women seeking first-line PCOS management, which makes generalization harder.

The likely fourth study is Mansour et al. 2021, a randomized trial in Clinical Nutrition that examined menstrual cyclicity, hyperandrogenism, and metabolic profile in women with PCOS. It appears to be the missing trial that brings the 2023 meta-analysis total to about 218 participants, but the available summary does not provide enough granular outcome data to show exactly how much it strengthened or diluted the testosterone signal from Banaszewska.

Put those trials together and a pattern emerges. The only clearly documented androgen-lowering signal came from 1500 mg per day over 3 months in a 34-person trial. The 800 mg, 40-day studies looked more anti-inflammatory than anti-androgenic. That does not prove resveratrol needs a high dose and longer course, but it is the most plausible reading of the current human data.

Why two meta-analyses can disagree without anyone cheating

The mismatch between the 2023 and 2024 meta-analyses is the real story because it shows how unstable pooled results become when the trial base is thin.

Start with the sample sizes. One review pooled about 218 participants; the other pooled about 169. The 49-participant gap does not map neatly onto any one named trial from the recoverable 2023 dataset. That means the two reviews may not have had a simple subset relationship. They may have included partially non-overlapping trial sets.

Then there is the issue of measurement. Testosterone was measured in different units across studies, including nmol/L in European work and ng/dL in Iranian studies. Fadlalmola et al. used standardized mean differences to harmonize outcomes across units. That is a reasonable method, but it comes with a catch: standardized mean differences depend on within-study variability. With small trials, that variability can bounce around enough to make a borderline result look significant or not significant depending on the pooling approach. If the 2024 review handled those data as mean differences or used a different conversion strategy, the pooled testosterone result could shift even if the underlying participants did not.

Duration also matters. PCOS-related hyperandrogenism often tracks insulin resistance, and insulin resistance does not usually move in 40 days just because a supplement label says it should. If resveratrol works mainly by improving insulin sensitivity, then short studies could miss the downstream endocrine effect and bias the pooled estimate toward no androgen benefit. That possibility is discussed in mechanistic reviews such as Karimi et al. 2022 and Shojaei-Zarghani et al. 2022.

And finally, one study probably carries too much narrative weight. Exclude or down-weight the 34-person, 1500 mg, 3-month Banaszewska trial and the case for testosterone lowering gets much thinner. In a field with four trials, that is not a minor sensitivity analysis. It is the whole ballgame.

Metabolic outcomes are plausible, but the published summaries are thin

Resveratrol has a coherent mechanistic pitch in PCOS. It may improve insulin sensitivity, modulate inflammatory signaling, and affect mitochondrial function. A 2024 randomized, triple-blind placebo-controlled trial in women with PCOS undergoing assisted reproduction, Ajabi Ardehjani et al., reported improved mitochondrial biogenesis markers and better reproductive outcomes. That supports biological activity. It does not prove routine androgen control in the broader PCOS population, especially because the trial was conducted in an assisted-reproduction setting.

The same caution applies to combination therapy. In Hassan et al. 2023, a double-blind randomized trial in 110 obese, oligo-anovulatory women with PCOS, combined resveratrol plus myoinositol outperformed combined metformin plus pioglitazone on endocrine, metabolic, and stress-related outcomes after 12 weeks. That is interesting, but it is not a monotherapy result. Myoinositol has its own insulin-sensitizing effects, so the benefit cannot be pinned on resveratrol alone.

Broader supplement evidence also cuts against treating resveratrol as uniquely special. A 2024 meta-analysis of polyphenols in PCOS, Balieiro et al., included 15 randomized trials with 916 patients and found that polyphenols as a class significantly reduced serum insulin and BMI compared with placebo. A 2025 review of dietary polyphenols, Jian et al., assessed hormonal, glycolipid, inflammatory, and oxidative stress outcomes across PCOS studies. Those reviews suggest polyphenol effects may be real at the class level, but they do not isolate resveratrol as the star of the show.

That matters because when four small resveratrol trials generate multiple reviews, the bottleneck is no longer enthusiasm. It is primary data.

What a careful reader should conclude

Resveratrol has a plausible mechanism in PCOS and one small, higher-dose, longer-duration trial with a meaningful androgen signal. That is not the same thing as reliable proof that it lowers testosterone across the broader PCOS population.

The evidence supports three cautious conclusions.

• Resveratrol at 1500 mg per day for 3 months produced the clearest documented reduction in testosterone and DHEAS, but that finding comes from a single 34-person randomized trial.
• Resveratrol at 800 mg per day for 40 days has been studied more than once, but those trials mainly support anti-inflammatory or cell-level effects, not a consistent androgen-lowering result.
• The pooled literature is too small and too heterogeneous to support a confident claim that resveratrol is a dependable standalone treatment for androgen excess in PCOS.

For someone with PCOS who is already working on insulin resistance through diet, exercise, weight management, or medication, resveratrol may be a biologically plausible add-on. The current human evidence does not justify selling it as a primary fix for hyperandrogenism. The disagreement between the meta-analyses is the warning label: if a result cannot reproduce cleanly across two pooling exercises built on essentially the same tiny evidence base, confidence should stay modest.

Right now, resveratrol in PCOS looks less like a settled therapy than a hypothesis with one decent clinical clue and several reasons to stay skeptical.