“It’s Genetic”: The Convenient Inheritance of Hair Loss

The idea that male pattern hair loss is inherited goes back at least to 1916, when Dorothy Osborn published Inheritance of Baldness in the Journal of Heredity [1]. Osborn drew on information from 22 families, although the published paper illustrated only five family trees and some of those pedigrees were incomplete.

She proposed that common baldness could be explained by a single autosomal factor that behaved as dominant in men and recessive in women. Under this historical model, men with either BB or Bb were predicted to develop baldness, while women were predicted to develop it only if they inherited BB.

This was an autosomal model, not an X-linked model. Both men and women could inherit BB, Bb or bb. The letters described genotype rather than severity, so BB did not mean severe baldness and Bb did not mean mild baldness.

Osborn was writing during the early Mendelian era, when researchers were attempting to organise visible traits into clean dominant and recessive patterns. Similar sex-influenced models had already been proposed for traits such as horns in sheep.

Osborn was writing in the early Mendelian era, when genetics was still in its confident, pattern-hunting phase and researchers were keen to map visible traits into clean dominant and recessive rules. Mendel’s work had recently been re-established in scientific thinking, and family tree analysis became a popular method for inferring inheritance patterns from visible traits, often favouring simplified dominant vs recessive interpretations. A few years earlier, sex-influenced inheritance models had already been discussed for traits such as sheep horns, reinforcing the idea that a trait could behave like “dominant in one sex, recessive in the other.”

Osborn’s paper became historically influential, but its evidence could not support such a confident inheritance model. Family-tree analysis was an important tool at the time, yet incomplete pedigrees and a limited number of illustrated families made it difficult to establish one reliable pattern.

The limitation was not the suggestion that heredity mattered. It was expecting one dominant or recessive rule to explain a complex condition that develops differently according to sex, age and individual biology.

Later research moved away from the single-gene model. Androgenetic alopecia is now understood as polygenic, meaning that many genetic variants can each contribute to susceptibility. Some are located on the autosomes and others on the X chromosome. Together, they influence risk without allowing one gene to determine the outcome.

The 1916 view was too simple.  Osborn’s model treated male pattern baldness as a single inherited trait, whereas later work attempted to show that androgenic alopecia is not a single gene condition, it could potentially be better understood as polygenic.  Meaning that multiple genetic variants may each make small contributions to susceptibility rather than one gene determining the outcome.

A 2017 study [2] used genetic data from 52,874 White British men aged 40-69 years who self-reported androgenic alopecia in four different categories: no hair loss, slight hair loss, moderate hair loss, and severe hair loss. That analysis identified 287 independent genetic signals associated with male pattern baldness. These were not 287 genes related to baldness, but statistical association signals across the entire genome, often tagging wider regions of linked DNA that may include regulatory, non-coding sequences as well as nearby genes.  The researchers split the men into two groups: one group to identify the signals and build a polygenic risk score, and a second to test how well that score could distinguish men with more severe hair loss from those with none.

This paper had several important strengths, including a large dataset, genome-wide coverage and a separate group in which the polygenic score could be tested. However, the results need to be interpreted carefully.

The combined genetic score produced an AUC of 0.78 when distinguishing severe hair loss from no hair loss. This does not mean that the model correctly diagnosed 78 out of every 100 men. It means that if one man with severe hair loss and one man with no hair loss were selected at random, the man with severe hair loss would receive the higher genetic score approximately 78 per cent of the time.

Performance fell to an AUC of 0.68 for moderate hair loss versus no hair loss and 0.61 for slight hair loss versus no hair loss. The genetic score was therefore better at separating the clearest extremes than distinguishing subtler differences.

There was also considerable overlap between the groups. Among men with a polygenic score below the median, 14 per cent still reported severe hair loss and 39 per cent reported no hair loss. Among those whose scores fell within the highest 10 per cent, 58 per cent reported moderate to severe hair loss.

Genetic risk altered the probability of hair loss, but it did not determine the outcome.

The study recorded each participant’s hair pattern at the time of assessment and did not establish the age at which his hair loss began. Participants also selected from four images rather than undergoing clinical grading, trichoscopy or standardised scalp photography. These limitations do not invalidate the findings, but they restrict what the score can tell us about the onset and future progression of hair loss in an individual.

A 2018 UK Biobank study analysed genetic data from 205,327 European men who rated their hair loss using four pictures [3].

However, its headline estimate that inherited factors explained 62% of the variation did not come from the whole group. It was calculated from 4,428 pairs of brothers.

The same method produced an estimate of 41% from 789 father-and-son pairs and 59% when all first-degree relatives were combined.

Brothers may share genetic variants, including variants inherited from their mother’s X chromosome. However, they may also share childhood environment and other exposures. The researchers acknowledged that shared environment, complex genetic effects and statistical variation could have contributed to the higher brother estimate.

A separate analysis of unrelated men found that measured common genetic variants explained approximately 39% of the variation in age-adjusted, self-reported hair-loss scores.

Understanding Polygenic Risk

A polygenic score can tell us that inherited genetic variation contributes to risk, yet it still fails to predict individual outcomes reliably, and twin studies makes the limitation obvious: identical genetic background does not guarantee identical hair loss. 

Genetics may shape baseline sensitivity, but systemic factors such as insulin resistance, inflammatory tone, microvascular function, and hormonal signalling influence whether that susceptibility is expressed, how early it appears, and how aggressively it progresses. In other words, the strongest reading of the evidence is not that androgenic alopecia “is genetic,” but that it is genetically influenced and metabolically modulated, which is a very different claim and far more clinically useful.

One genome multiple phenotypes

Cells may contain the same DNA but behave very differently because they do not express the same genes in the same way. Genes can be switched on or off at different times, in different patterns, and at different intensities depending on the cell type, tissue environment, hormones, inflammation, age, and other biological signals.  Epigenetics describes how the environment chemically modifies the machinery that reads DNA so that the ways genes are expressed changes without changing the underlying sequence.  

Science shows a genetic susceptibility dependent on environmental factors, what it does not show is genetic predetermination.  If hair loss were genetically predetermined in any simple sense, identical twins would be expected to lose hair in the same way and to the same degree, yet twin data show otherwise, with differences in whether hair loss occurs at all and with severity linked to factors such as smoking, stress, alcohol exposure, multiple marriages and BMI.  The broader hair loss literature supports this non-deterministic view. Twin data is important in genetic research because identical twins provide a literal test of genetic determinism: if the same DNA does not reliably produce the same outcome, genes cannot be at the determining factor.

A 2013 study examined 92 identical male twins using questionnaires, testosterone measurements and standardised photographs [4]. Although each twin pair shared almost identical DNA, differences in frontal, temporal and vertex hair loss were associated with several reported exposures.

Longer smoking duration was associated with greater frontal and vertex hair loss, while longer periods of stress were associated with greater vertex hair loss. Other findings varied according to scalp region and were sometimes difficult to interpret. Both alcohol abstinence and drinking more than four alcoholic drinks per week were associated with greater vertex hair loss.

The researchers also found an association between spending more money on hair-loss products and greater temporal hair loss. The most likely explanation is reverse causation: men experiencing more hair loss may have spent more money attempting to treat it.

This was a relatively small observational study, so it cannot prove that these factors caused or prevented androgenetic alopecia. It does show that individuals with almost identical genomes can develop different visible hair-loss patterns.

A separate study by Fraga and colleagues examined gene regulation in identical twins rather than hair loss specifically [5]. The researchers found that older twins tended to show greater differences in DNA methylation and histone modification, particularly when their lifestyles and life histories had diverged.

Epigenetics describes regulatory processes that influence gene activity without altering the underlying DNA sequence. These include DNA methylation, histone modification and regulatory RNA activity.

The Fraga study cannot prove that epigenetic changes cause androgenetic alopecia. It provides a broader example of how people with nearly identical DNA can develop differences in gene regulation over time.

Genetics and gene expression are connected parts of the same biology. Inherited variants can influence how a follicle responds to androgens, while age, hormones, inflammation and the local tissue environment can affect how those responses are expressed.

Genetics vs Gene Expression 

Genes can be upregulated and downregulated depending upon the environment.  Most genes are switched on and off, turned up and down, depending on context. The technical term is gene expression. A gene can be present but barely used, or heavily used, depending on signals the cell is receiving.

Cells respond to their environment through signalling pathways. Those signals can come from:

• hormones (androgens, cortisol, insulin, thyroid hormones)

• inflammatory cytokines

• oxidative stress and nutrient status

• mechanical stress

• circadian rhythm and sleep disruption

• medications and toxins

These signals change which genes are expressed, that is why two people can carry identical genes, but express it differently.  Epigenetics is the layer of chemical regulation that influences gene expression without changing the physical DNA. 

The big mechanisms are:

• DNA methylation

• histone modifications

• regulatory RNAs

They do not rewrite the genome, but they can make certain genes easier or harder to read. Some epigenetic marks are relatively stable, some are dynamic and change with environment and age. In complex traits, epigenetics is one of the key ways “environment” gets under the skin and into our biology.

So what does it mean?

That brings us to the real issue with the phrase “it’s genetic.” In common use, it implies a specific gene or a fixed polygenic inheritance that explains androgenic alopecia. That is not what science shows. No single gene reliably determines how hair loss appears or progresses as in a true single-gene disorder, and even the polygenic evidence points to risk association rather than determining outcome.

Instead, there are statistical associations, heritable tendencies, and risk-modifying variants with incomplete and inconsistent predictive value. The phrase therefore becomes clinically blunt and scientifically misleading. It compresses a dynamic, multifactorial process into a story of inherited inevitability.

Osborn, who introduced the concept of genetic inheritance, concluded that heredity “explains away the difficulties,” which is revealing in the worst way.  It suggests heredity is being used as a neat bow on a messy problem rather than being rigorously weighed against alternative explanations.  

When we look at a patient with hair loss, research has shown a broader physiology that often travels with it: insulin resistance, chronic inflammation, high cholesterol, increased visceral fat, low vitamin D, low zinc, fibrosis of the scalp, altered androgen metabolism, oxidative stress, and microvascular strain.  In that context, hair loss becomes the visible output of modifiable biological drivers acting on a susceptible system.  

Genetics may increase susceptibility, while onset, expression and severity are shaped by multiple interacting factors that have nothing to do with genetics.

It’s not as catchy or convenient, but it’s accurate.  

1. Osborn D. Inheritance of baldness. Journal of Heredity. 1916 Aug 1;7(8):347-55.
2. Hagenaars SP, Hill WD, Harris SE, Ritchie SJ, Davies G, Liewald DC, Gale CR, Porteous DJ, Deary IJ, Marioni RE. Genetic prediction of male pattern baldness. PLoS genetics. 2017 Feb 14;13(2):e1006594.
3. Yap CX, Sidorenko J, Wu Y, Kemper KE, Yang J, Wray NR, Robinson MR, Visscher PM. Dissection of genetic variation and evidence for pleiotropy in male pattern baldness. Nature communications. 2018 Dec 20;9(1):5407.
4. Gatherwright J, Liu MT, Amirlak B, Gliniak C, Totonchi A, Guyuron B. The contribution of endogenous and exogenous factors to male alopecia: a study of identical twins. Plastic and Reconstructive Surgery. 2013;131(5):794e–801e.
5. Fraga, M.F., Ballestar, E., Paz, M.F., Ropero, S., Setien, F., Ballestar, M.L., Heine-Suñer, D., Cigudosa, J.C., Urioste, M., Benitez, J., Boix-Chornet, M., Sanchez-Aguilera, A., Ling, C., Carlsson, E., Poulsen, P., Vaag, A., Stephan, Z., Spector, T.D., Wu, Y.Z., Plass, C. and Esteller, M. (2005) ‘Epigenetic differences arise during the lifetime of monozygotic twins’, Proceedings of the National Academy of Sciences, 102(30), pp. 10604–10609. doi: 10.1073/pnas.0500398102.