A condition defined by a committee, with no single test — and the genetics still found something. The strongest locus reached an odds ratio of 0.67, which for a common variant is large.
Solid lines are connections this site curates. Dashed lines mean the two ends share a research paper — worth knowing, and not a claim that one explains the other.
Polycystic ovary syndrome is common, it affects fertility, metabolism and skin, and it does not have a defining test. Diagnosis is made from a combination of irregular or absent ovulation, clinical or biochemical signs of excess androgens, and the appearance of the ovaries on ultrasound — with other causes excluded.
A syndrome defined by a checklist is a hard thing to run genetics on. The cases are not guaranteed to be one biological thing.
A 2011 study in Han Chinese women began with 744 cases and 895 controls and replicated in two independent cohorts — 2,840 cases and 5,012 controls from northern China, then 498 cases and 780 controls from southern and central China.
Three loci came through with strong evidence:
Odds ratios of 0.67 and 1.34 are substantial for common variants. Most common-variant findings sit near 1.05. Whatever the diagnostic fuzziness, something real is underneath.
The other position on this page, rs10818854, is in DENND1A at the 9q33.3 locus. The study's own index SNP for that region is rs2479106; ours is a different marker in the same gene, and this page does not transfer the published odds ratio onto it.
DENND1A is involved in how cells take receptors back in from their surface — a general piece of cell machinery rather than anything reproductive. And the clinical picture supports the wider reading: insulin resistance, weight, and cardiovascular risk are part of this condition, not complications bolted onto it.
PCOS is a metabolic condition that presents through the reproductive system, and the genetics reads that way too.
All three cohorts were Han Chinese. That matters in both directions: it is a large, well-replicated study in a population that GWAS has historically under-served, and it is a single ancestry, so the effect sizes should not be assumed to transfer unchanged.
Not diagnose. PCOS is diagnosed on the clinical criteria, and no genotype substitutes for them or predicts who will develop it.
What the genetics does support is a change in how the condition is framed. Long treated as a gynaecological problem to be managed cycle by cycle, PCOS looks — from the loci and from the clinical course — like a systemic metabolic condition. That framing changes what gets monitored over a lifetime, and it is a change worth more to patients than a risk score would be.
Source. Chen et al. (Nat Genet 2011) conducted a genome-wide association study of PCOS in Han Chinese. The discovery set comprised 744 cases and 895 controls; replication used 2,840 cases and 5,012 controls from northern Han Chinese and 498 cases and 780 controls from southern and central Han Chinese. Three loci showed strong evidence of association: 2p16.3 (rs13405728, meta P = 7.55 × 10⁻²¹, OR 0.71), 2p21 (rs13429458, meta P = 1.73 × 10⁻²³, OR 0.67) and 9q33.3 (rs2479106, meta P = 8.12 × 10⁻¹⁹, OR 1.34).
Positions listed here. rs13429458 is the study's own 2p21 index SNP. rs10818854 was recorded by the collector against DENND1A at the 9q33.3 locus and is not the index SNP named in the abstract for that region; the odds ratio above belongs to rs2479106 and is not transferred to it here.
Diagnosis. Made on established consensus criteria requiring a combination of oligo- or anovulation, clinical or biochemical hyperandrogenism, and polycystic ovarian morphology, with exclusion of other causes of hyperandrogenism and ovulatory dysfunction. There is no diagnostic laboratory test and no genotype has diagnostic or predictive use.
Metabolic dimension. Insulin resistance, impaired glucose tolerance and adverse cardiometabolic risk profiles are recognised features rather than incidental comorbidities, and long-term management reflects that.
Population. Discovery and both replication cohorts were Han Chinese. Effect sizes and marker–allele relationships should not be assumed to transfer to other ancestries without direct study.
What a 23andMe/AncestryDNA export or raw VCF can and can't tell you about Polycystic Ovary Syndrome comes down to these specific, well-studied positions — not a diagnosis.
It is a fair objection, and the answer is that the signal came through anyway. Fuzzy case definitions weaken a study, they do not fabricate findings — and this one replicated in two independent cohorts at odds ratios far larger than common-variant genetics usually produces.
No. Diagnosis rests on the clinical criteria: ovulation, androgen excess and ovarian appearance, with other causes excluded. No genotype substitutes for them, and none predicts who will develop the condition.
Because insulin resistance and cardiometabolic risk are features of it rather than complications of it, and the genes found point at general cellular machinery rather than at anything reproductive. It presents through the reproductive system; it is not confined to it.
Partly, and honestly nobody should promise more than that. It is a large and well-replicated study in a population genetics has under-served, and it is one ancestry. Effect sizes and the relationship between a marker and the causal variant can both differ elsewhere.
Free to reuse. This page's text is original writing from freely-available research, licensed CC BY 4.0 — reuse it, including commercially, with attribution to MyGeneLog. It's general research-derived information, not medical advice or a diagnosis — see Terms of Use.