Metabolic & Cardiometabolic
insulin resistance causes hyperandrogenism
Strong support Metabolic & Cardiometabolic 💰 Industry COI noted
RefutedContestedStrong support
consensus score 0.88
📅 Last reviewed: 2026-07-15 ⓘ
Evidence ladder
How far up the ladder this claim has climbed. A high consensus on a low rung means "consistent so far," not "proven in people."
Top evidence so far: Population patterns (Observational)
MechanismIn-vitroAnimalObservationalRCTMeta-analysis
How the studies fall
8 support 0 contradict 0 tested null 1 mixed · 9 sources, 8 independent groups
What the evidence shows
Hyperinsulinemia drives the hyperandrogenism of PCOS — insulin ↑ovarian androgen output and ↓hepatic SHBG (raising free androgens). So the 'reproductive/cosmetic' features are downstream of the metabolic lesion.
The evidence (15)
| Source | Grade | Stance | Quality | Finding |
|---|---|---|---|---|
| Kamalakannan U, Sai Durga Surampudi AV, Racha V, Amanchy R, Jain N, Muralidharan K 2026 · Journal of Pharmacology and Experimental Therapeutics | animal | supports | low | Letrozole-induced PCOS mouse model. Berberine-gallic and berberine-gentisic cocrystals 'significantly improved metabolic parameters and restored insulin responsiveness' (increased AMPK activation, reduced IRS-1 Ser307 phosphorylation, enhanced PI3K/Akt) and 'normalized luteinizing hormone receptor and androgen receptor expression'. Co-movement only, and the androgen-side readout is RECEPTOR EXPRESSION — no post-treatment androgen level is reported in the available text. Causal geometry runs the other way at induction: letrozole raises androgens by aromatase blockade, and 'letrozole administration induced hallmark PCOS features, including weight gain, impaired glucose tolerance, hyperandrogenism, ovarian inflammation, and disruption of hepatic insulin signaling'. Berberine is pleiotropic by the authors' own account ('multitarget agents that integrate insulin sensitization, antiinflammatory activity, and hormonal regulation'), with no mediation or correlation analysis; n, duration, effect sizes, randomization and blinding are all unreported. Abstract only (paywalled). |
| Rosenfield, Ehrmann 2016 · Endocr Rev | observational | supports | moderate | Narrative review (Endocr Rev, 531 refs, no primary data): 'A metabolic syndrome of obesity-related and/or intrinsic insulin resistance occurs in about half of PCOS patients, and the compensatory hyperinsulinism has tissue-selective effects, which include aggravation of hyperandrogenism.' Supports insulin as a contributory cause -- but the review's own thesis is that PCOS is primarily intrinsic ovarian hyperandrogenism, with 'two-thirds of PCOS cases hav[ing] functionally typical FOH' and cultured PCOS theca cells dysregulated with the systemic milieu removed. Abstract-only; no OA full text. |
| szkodziak-2025 2025 | observational | mixed | low | Cross-sectional and directionally agnostic. The one measured datum is 'FAI - HOMA-IR (r = 0.35' (n=200), unadjusted for BMI even though 'A strong positive correlation was found between BMI and HOMA-IR.' (r = 0.72). The authors read that correlation the OTHER way: 'Positive associations among FAI, HOMA-IR, and mFG score support a contribution of ovarian androgens to insulin resistance and hirsutism.' Counter-evidence in the same table: phenotype D is non-hyperandrogenic (mean FAI 6.3, below the paper's FAI > 10 cut-off) yet mean HOMA-IR 2.73 with 51.52% insulin-resistant - IR without androgen excess. The insulin -> androgen arrow appears only as narrative: 'The mechanism described above closes like a carousel, creating a pathological “vicious cycle” model that ultimately leads to insulin resistance in all components of PCOS.' |
| Tienforti D et al 2026 · Clin Endocrinol (Oxf) | meta-analysis | supports | low | OFF-SCOPE for this claim, and quality high -> low. Systematic review/meta-analysis of 9 RCTs, 440 women with PCOS (NIH/Rotterdam/AE-PCOS). What it measured: 'Primary outcomes were serum total testosterone (TT), calculated free testosterone (cFT), free androgen index (FAI), and sex hormone-binding globulin (SHBG).' What it found: 'Inositol supplementation was associated with significant reductions in TT (SMD -1.30; 95% CI -2.17 to -0.42), cFT, and FAI, together with an increase in SHBG.' It does NOT measure insulin, fasting insulin or HOMA-IR as an outcome, so it cannot show that lowering insulin lowered androgens -- inositol's insulin-sensitizing action is assumed, not demonstrated here. Where measured insulin resistance does enter, the result cuts the other way: 'In subgroup analyses, the largest and most consistent reduction in TT was observed among normal-weight women (BMI < 25 kg/m2) (SMD -2.97; 95% CI -3.78 to -2.16), with minimal heterogeneity (I2 = 9%). No significant improvements were detected in overweight or obese women, nor in women with insulin resistance when considered independently of BMI.' The insulin-sensitizer worked least in the women with the most insulin resistance. Quality drops to low independently of the scope call: 9 small trials, 440 women, 'Substantial heterogeneity was observed for several outcomes', 'These subgroup findings should be interpreted as exploratory', and the authors themselves frame the work as a 'hypothesis-generating framework' needing 'adequately powered, phenotype-stratified trials'. No PROSPERO ID, GRADE rating or risk-of-bias summary visible at abstract grade. 'The authors declare no conflicts of interest.' |
| Ji C et al 2026 · Front Endocrinol | observational | supports | low | Full text read. n=1 case report: de novo INSR c.3328G>C (p.Asp1110His) in a non-obese 12-year-old girl (BMI 17.85). Baseline HOMA-IR 30.2, fasting insulin 153.1 mu/L (post-load >300 at 30-180 min), with total testosterone 2026.6 pg/ml (ref <750), androstenedione 2642.3 pg/ml (ref 420-1000), mFG 16 and acanthosis nigricans. After 3 months of metformin plus a resistant-starch diet, fasting insulin fell to 60.80 mu/L and testosterone normalized to 16.60 ng/dl (ref 7-27.57) with menses restored. Direction matches the claim in both the genetic natural experiment and the within-patient reversal, but n=1, uncontrolled, unblinded, and the insulin-lowering is confounded with metformin's own anti-androgen action, so this stays low-quality observational. |
| Weidner, Vann, Hodowanec, Ivey, Roy, Sechrist, Cole, Astapova 2026 · Endocrinology | animal | supports | moderate | [FT-verified] Weidner 2026 Endocrinology, DHT-pellet mouse PCOS model. Ovarian theca/granulosa cells stay fully insulin sensitive while hepatocytes from the same mice are insulin resistant; 'PCOS TCs produced significantly more testosterone when stimulated with both hCG and insulin' (P<.01, n=6-10 mice/group). CONDITIONAL: 'TCs from both groups did not respond to stimulation with insulin alone' — the androgen rise needs insulin PLUS gonadotropin. ANIMAL, single lab, bidirectional caveat (in vivo arm runs hyperandrogenism -> insulin resistance); model lacks the high-LH state of human PCOS. |
| Xu J et al 2026 · J Ovarian Res | observational | mixed | moderate | Narrative review (abstract only; no primary data, no stated search strategy) whose thesis runs the opposite direction to this claim: it 'synthesize[s] evidence demonstrating that hyperandrogenism promotes organ-specific lipotoxicity and insulin resistance', contributing to 'visceral adiposity, hepatic steatosis (and its progression to metabolic dysfunction-associated steatotic liver disease), and skeletal muscle dysfunction', and concludes that 'redefining hyperandrogenism as a primary metabolic driver is crucial'. Insulin resistance appears only as the other half of a 'vicious cycle with insulin resistance' and in the therapeutic line that insulin-sensitizing agents 'disrupt the hyperandrogenism-insulin resistance cycle' - framing and strategy, not a measured androgen outcome. Off-scope for this claim's direction; it is a review-grade vote on the reverse arrow. |
| Andrade LJdO et al 2026 · Research Square | observational | supports | low | Research Square PREPRINT (not peer-reviewed), reanalysis of public multi-ethnic datasets: admixture mapping in admixed women with HAIR-AN syndrome found local African ancestry at 19p13.2 (INSR locus / insulin signalling) associated with both hyperandrogenism and IR traits; fine-mapped with eQTL/pathway annotation. Genetically-informed association consistent with insulin signalling driving hyperandrogenism; MR-analogous, not interventional. |
| Olewi NZ, Hassan AF 2026 · Naunyn-Schmiedeberg's Archives of Pharmacology | animal | supports | low | Testosterone-propionate rat PCOS model (30 prepubertal Wistar rats, n = 6/group, 30-day induction then 14-day treatment; abstract-only, paywalled). Co-improvement vote: 'Tirzepatide induced significant improvements in estradiol, testosterone, LH, FSH, and LH/FSH ratio' while 'A greater reduction in body weight, cholesterol, triglycerides, fasting glucose, insulin, and HOMA-IR was achieved with the higher dose of tirzepatide.' Quality moderate -> LOW: (1) tirzepatide is not an isolated insulin sensitizer — weight, lipids and ovarian TNF-α/IL-6/caspase-3 fell in the same animals, each a rival mediator, and no correlation or mediation analysis links the HOMA-IR fall to the androgen fall; (2) the model imposes androgen excess as the upstream cause — 'Testosterone propionate resulted in pronounced hyperandrogenism and an increased LH/FSH ratio, dyslipidemia, hyperglycemia and insulin resistance' — so the paper evidences androgen -> IR directly and only co-movement for IR -> androgen; (3) n = 6/group, no blinding or randomization stated, no effect sizes/p-values available. Two corrections to the prior extract: metformin's HOMA-IR effect is NEVER reported (metformin equivalence is stated only for TNF-α/IL-6/caspase-3), and the stated dose-dependence covers the metabolic panel and histology, not the hormone panel. |
| Shorakae S et al 2018 · Clin Endocrinol (Oxf) 2018 Nov;89(5):628-633 | observational | supports | moderate | Cross-sectional substudy nested in an RCT, community-recruited: 49 women with PCOS (Rotterdam criteria; age 30 ± 6, BMI 29 ± 5) vs 23 controls (age 29 ± 8, BMI 33 ± 7). Insulin resistance measured as fasting insulin and glucose, hyperandrogenism as testosterone/SHBG/Free Androgen Index. 'Testosterone and FAI both correlated positively with insulin resistance in women PCOS.' CORRELATIONAL ONLY, and the authors' own model is a competing one: 'Chronic low-grade inflammation potentially mediates the effect of sympathetic dysfunction on hyperandrogenism and insulin resistance' — i.e. the two may be co-effects of an upstream driver rather than cause and effect. No correlation coefficient, CI or P value is reported for the finding, and no adjustment is stated for it (the age/BMI adjustment attaches to the PCOS-status model). Abstract-grade: paywalled, not in PMC. |
| Serin S, Celikaslan H, Cangi S, Dokuyucu R 2026 · Sci Rep | animal | supports | low | Letrozole-induced rat PCOS (60 Wistar Albino rats, six groups); ALA + vitamin D lowered insulin resistance and testosterone together — 'The combination therapy group demonstrated the most pronounced improvements, with HOMA-IR (2.43 ± 0.47), MDA (2.52 ± 0.39 nmol/mg), testosterone (1.29 ± 0.27 ng/mL), and LH/FSH ratio (1.03 ± 0.19) values approaching control levels (all p < 0.01 vs. PCOS)'. Co-improvement only, and weak: letrozole raises androgens BY CONSTRUCTION (aromatase blockade), so the induction phase demonstrates androgen excess -> insulin resistance, the reverse of this claim. MDA fell 5.84 -> 2.52 and the LH/FSH ratio normalised in the same sentence, making redox and gonadotropin routes equally viable mediators; ALA is an antioxidant, not an insulin sensitiser. The reported 'significant interaction effects for HOMA-IR, MDA, testosterone, and LH/FSH ratio' are the ALA x vitamin-D factorial interactions, not an insulin x androgen link, and no correlation or mediation analysis is reported. Prior extract's 'metformin arm showing similar co-improvement' is unanchored — no metformin-group value appears in the retrievable text. |
| zhu-shbg-2019 2019 | observational | supports | low | [abstract-only] Narrative review, no primary data: 'Low serum SHBG levels are considered a biomarker of abnormal metabolism and are related to insulin resistance (IR), compensatory hyperinsulinemia' and 'A reduction in plasma sex hormone-binding globulin (SHBG) ... is often used as an indicator of hyperandrogenism in women with PCOS' — directionally consistent with the claim but stated as association only; the authors hedge the causal route: 'it remains unclear whether HNF-4alpha is indirectly involved in the pathogenesis of PCOS via regulation of hepatic SHBG synthesis.' |
| Pappas D, Kandaraki E, Kalogirou A, Diamanti-Kandarakis E 2026 · Expert Rev Endocrinol Metab. 2026 Jul;21(4):253-269 | observational | supports | moderate | Narrative review: insulin resistance and compensatory hyperinsulinemia alter ovarian steroidogenesis and impair granulosa cell aromatase activity, driving hyperandrogenism; literature synthesis 1985-2025. |
| La Vignera S, Condorelli RA 2026 · Nutrients | observational | supports | low | [FT-verified, PMC13259202 JATS] Narrative review, 2 authors, Nutrients (MDPI), 37 included studies, search to March 2025, NOT PROSPERO-registered; the paper states 'No new data were created or analyzed in this study.' Zero-weight under CONVENTIONS §5 (a review is not evidence) — retained for transparency only. What it says on this claim: 'Insulin resistance is a central feature of PCOS and is mechanistically linked to hyperandrogenism through multiple pathways: insulin directly stimulates ovarian androgen production, reduces hepatic SHBG synthesis (increasing free androgen levels), and potentiates LH effects on theca cells' — sourced to reference 31, Diamanti-Kandarakis & Dunaif, Endocr Rev 2012, itself a review, so the chain to primary data is two secondary layers deep. And 'SHBG production by the liver is suppressed by insulin and androgens, contributing to hyperandrogenism in PCOS'. The one claim-relevant line from the review's own corpus is a co-movement, not a causal test: 'In PCOS studies, reductions in HOMA-IR (insulin resistance index) correlated with decreased FAI and improved reproductive outcomes' — a single 6-week single-arm study in 30 women (Güven 2023) where FAI fell, SHBG rose, 'The percentage of patients with hyperandrogenism decreased significantly' and 'menstrual cycles normalized in over 70% of patients', but where 'These improvements occurred alongside reductions in BMI, waist–hip ratio, and HOMA-IR (insulin resistance index)', i.e. weight loss is an uncontrolled co-mediator and no mediation analysis is reported. No effect size or confidence interval appears anywhere in the review; every result is a bare p-value. Three further reasons to read this as transparency rather than support: the review's PCOS section rests on Ranneh 2025, which the vault already counts as a meta-analysis (sources/ranneh-2025.md) — the double-count §5 exists to prevent; the review grades that same section both 'MODERATE—supported by one RCT' and '[STRONG] ... supported by multiple RCTs, meta-analyses, and systematic reviews'; and it promotes its one primary from 'a 6-week intervention study ... in 30 women' to 'one RCT' within three paragraphs. Funding: 'This research received no external funding.' COI: 'The authors declare no conflicts of interest.' |
| Barbieri, Makris, Ryan 1984 · Obstet Gynecol 1984;64(3 Suppl) (symposium supplement) | in-vitro | supports | low | Ex-vivo human ovarian explants: 'Insulin (500 ng/mL) alone stimulated androstenedione and testosterone accumulation, but not progesterone or estradiol accumulation.' Selectivity for the androgen arm is the mechanistic point. Quality low, not moderate: tissue came from only two donors — 'a woman with hyperandrogenism, insulin resistance, and acanthosis nigricans, and from a normally cycling woman' — no statistics are reported, 500 ng/mL is ~2 orders of magnitude above physiologic insulin (IGF-1R cross-reactivity at that dose), and the venue is a symposium supplement. Foundational mechanism, not evidence of effect size in humans. Read from abstract only (2026-08-19); the prior '[FT-verified]' tag is not supported by any full text in reach. |
Disagree, or know a study we missed?
We grade by evidence, not opinions. The way to weigh in is to point us to a study we haven't cited (check the evidence table above first), or to flag a problem with one we have. Every submission is reviewed; if it holds up, the grade updates and shows in Science Changes Its Mind.
Opens a short form. You'll sign in with Google so submissions are tied to a real account — we don't display your identity, and we only accept a link we can verify (PubMed, DOI, ClinicalTrials.gov).
Educational only, not medical advice. Grades and scores reflect published evidence weighted by study design and quality; see the methodology.