When puberty starts is heritable, polygenic, and moved by years within a century — which tells you how much of it is not genetic. One of the variants here is the same position this site carries for obesity, and that overlap is the mechanism, not a coincidence.
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.
Menarche — a first period — is the milestone used to mark the timing of puberty in girls, because it is the one that can be dated. It varies widely between people, it runs in families, and it has been studied at enormous scale, which makes it one of the better-understood common traits in human genetics.
The largest study looked at about 370,000 women and found 389 independent signals associated with age at menarche. Together they explain roughly 7.4% of the variation in when it happens — about a quarter of what is thought to be inherited at all.
That is the shape of the trait: hundreds of positions, each worth days rather than years, and most of the inherited component still unaccounted for. Anyone offering to predict when a girl will start puberty from a genotype is offering something the data does not support.
rs9939609, in FTO, is the best-known common obesity variant, and it is also a menarche variant. That is not two unrelated findings that happen to share an rsID.
Puberty starts when the brain decides the body is ready, and body composition is part of what it reads. When the 2010 meta-analysis found its 30 new positions, four of them were already known as body-mass positions — in or near FTO, SEC16B, TRA2B and TMEM18 — and three more sat near genes involved in energy balance, including MCHR2, which is the second position on this page.
So the overlap is the mechanism showing through. Genes that influence how the body stores and senses energy also influence when it starts puberty. Our obesity page covers the other half of that same variant.
All three come from a meta-analysis of 32 genome-wide studies in 87,802 women, with replication in up to 14,731 more.
None of them is close to the strongest signal for this trait. That is LIN28B, at a p-value of 5 x 10-60 — one of the most reproducible associations in the whole of human genetics.
Almost everywhere else on this site, an allele is an allele. Here, for a few positions, it matters whether it arrived from the mother or the father. The signals for this trait cluster in imprinted regions — parts of the genome where one parent's copy is switched off — and rare variants near MKRN3 and DLK1 have large effects specifically when inherited from the father.
This is unusual and worth pausing on. Most of genetics treats the two copies as interchangeable. In this corner of it, they are not.
Average age at menarche fell substantially across the twentieth century, and it did so far too fast for genes to be the reason. Nutrition and childhood health did most of that, and they still do most of the difference between populations today.
So the accurate statement is that within a group of girls growing up in similar conditions, some of the difference is inherited — and between generations, almost none of it is.
Puberty timing is associated in population studies with several later outcomes, and genetic analyses support some of those links being causal rather than coincidental: earlier puberty with slightly higher risk of breast and endometrial cancer in women, and the corresponding pattern for prostate cancer in men — independent of body mass index.
Two things follow, and they are both important. The association is real at the level of populations. And it is small, it is one thread among many, and there is nothing here for an individual to act on. Nobody's screening or care changes because of when they started puberty, and certainly not because of these three positions.
Very early puberty (before about eight) and puberty that has not started by the mid-teens are both reasons to see a doctor, and both are assessed clinically — growth, examination, hormone levels, sometimes imaging. Neither is diagnosed from a genotype, and nothing on this page substitutes for that assessment.
Architecture. Age at menarche is highly polygenic. The 2017 analysis in approximately 370,000 women identified 389 independent genome-wide significant signals, which explained about 7.4% of population variance in Icelandic data — roughly 25% of the estimated heritability. Around 250 genes were implicated through coding variation or associated expression, with significant enrichment in neural tissues, consistent with hypothalamic control of pubertal onset rather than a peripheral mechanism.
Imprinting and parent-of-origin effects. The 2014 analysis of 182,416 women found 123 signals at 106 loci and reported enrichment in imprinted regions, with parent-of-origin-specific associations at DLK1-WDR25, MKRN3-MAGEL2 and KCNK9 concordant with known parental expression patterns. Rare variants near MKRN3 and DLK1 show large effects when paternally inherited. Loss-of-function variants in MKRN3 are an established cause of central precocious puberty, again with paternal transmission — the same genes appearing as common modifiers of normal timing and as rare causes of a clinical disorder.
The three variants here. All are from Elks et al. 2010: a meta-analysis of 32 GWAS in 87,802 women of European descent with replication in up to 14,731. rs9939609 (FTO) was one of four newly identified menarche loci already known for body mass index, alongside SEC16B, TRA2B and TMEM18; rs4840086 (MCHR2) is one of three near genes implicated in energy homeostasis, with BSX and CRTC1; rs1364063 is recorded against NFAT5. Per-allele effects for common menarche variants are on the order of weeks, and the strongest known locus, LIN28B, was already established (P = 5.4 x 10-60) before that study.
Downstream associations. Mendelian randomization in the 2017 study supported causal inverse associations, independent of BMI, between puberty timing and risk of breast and endometrial cancer in women and prostate cancer in men. Observational work has also linked earlier menarche with obesity, type 2 diabetes, cardiovascular disease and all-cause mortality. None of this is used clinically at the level of an individual, and no guideline modifies screening on the basis of menarcheal age or of these genotypes.
Clinical boundary. Central precocious puberty and delayed puberty are clinical diagnoses made from examination, growth data, bone age and hormonal assessment; their monogenic causes (MKRN3, DLK1, KISS1R, TAC3 and others) are distinct from the common variants recorded here, which are not diagnostic and carry no screening role.
What a 23andMe/AncestryDNA export or raw VCF can and can't tell you about Age at Menarche comes down to these specific, well-studied positions — not a diagnosis.
No. The largest study found 389 positions that together explain about 7.4% of the variation — a quarter of what is inherited at all. Growth, family history and a doctor watching development say far more than a genotype can.
Because it is the same biology, not a coincidence. Puberty starts when the brain judges the body ready, and body composition is part of what it reads. Four of the menarche positions found in 2010 were already known as body-mass positions, and three more sat near genes for energy balance.
For a few positions here, yes — which is rare. Menarche signals are enriched in imprinted regions, where one parent's copy is switched off, and rare variants near MKRN3 and DLK1 have large effects specifically when inherited from the father.
At the level of populations there is a real association, and genetic analyses support part of it being causal — earlier puberty with slightly higher breast and endometrial cancer risk, independent of weight. For one person it is one small thread among many, no screening decision turns on it, and these three variants say nothing about it.
Puberty starting before about eight, or not having started by the mid-teens, is worth a medical opinion. Both are assessed from growth, examination and hormone levels — never from these positions.
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.