In 1935, Irving Stein and Michael Leventhal first described a series of seven women sharing the features of a clinical triad of polycystic ovaries, hirsutism and irregular or absent menses.1 In the 85 years since this publication, the condition which came to bear their name has become the commonest cause of anovulatory infertility globally, affecting one in every 7–17 of all women,2 and is an important driver of the $32 billion global infertility treatment industry.3 Renamed, the polycystic ovary syndrome (PCOS) is now the subject of more than 4000 scientific publications annually,4 evidence of an intense effort to unravel the complex problems of definition, fundamental physiology and genetics, the effectiveness of treatment and the association of this condition with metabolic disturbance, cardiovascular disease and cancer.

PCOS has also attracted the close attention of evolutionary biologists, drawn to the exquisite evolutionary paradox that the condition represents – a very common and highly heritable condition that affects all human populations, and is an important global cause of infertility. They are also drawn to the curious phenotypic juxtaposition in PCOS of metabolic and reproductive traits. This is because in evolutionary biology, resource availability, procurement and allocation play a large part in shaping reproductive success and somatic evolution.5 6 7 As Darwin famously said, this struggle for resources, and indeed existence, applies with ‘manifold force to the whole animal and vegetable kingdoms.’8

RA Fisher, one of the architects of the modern synthesis of the theory of evolution, devoted an entire chapter of his seminal book, The Genetical Theory of Natural Selection,9 to the power of fertility selection; selection based upon the differential effects of genes on fertility:
‘The intensity of fertility selection is sufficient to produce considerable evolutionary changes in relatively short periods.’

Accordingly, fertility selection, if unimpeded, should act swiftly to reduce the prevalence of the genetic determinants of PCOS.

Speculation as to why this has not occurred and why PCOS is so common has been imaginative and agile, and the subject of several reviews.10 11 12 13 These evolutionary hypotheses can be grouped as responses to fundamental questions about any biological trait: are they due to phylogeny, growth and development, or adaptation?14

Phylogenetic analyses of PCOS vulnerability are rare. Barnett has proposed that successful female reproductive adaptations to accommodate the growth demands of large-brained primate fetuses – pre-implantation endometrial proliferation and a rapidly invading placenta – have facilitated a particular vulnerability of higher primates to hypergonadotropic disruption of ovulatory function, as found in PCOS.15Another interesting analysis of global genotypic and phenotypic variation in PCOS has suggested that an intralocus sexual conflict, (that is, a fertility disadvantage in women balanced by a fertility advantage in men) may be present in humans and their hominid ancestors.16

Developmental programming of the hypothalamic–pituitary control of LH by in utero and prepubertal exposure to androgens enhances visceral fat distribution.17 In challenging nutritional environments, this mechanism may confer survival and fertility benefits for the infant such as increased fat storage and increased follicular readiness in adulthood.18

Hypotheses that imagine the selective advantages in ancient times of the PCOS phenotype abound.19 20 21 22 23 24 25 They include: kin selection,26 delayed menopause,27 increased muscularity and resistance to infection,28

and, most commonly, the fertility benefits of insulin resistance and a relative hyperinsulinaemia.29 30 31 32 33 34 These metabolic traits have been shown  to foreshorten the duration of lactational amenorrhoea,35 which was the most important determinant of lifetime reproductive success in hunter-gatherer and agrarian populations.36

Conversely, PCOS evolution need not have been driven by adaptive evolutionary mechanisms. Genetic drift due to a serial founder effect and population balance due to sexually antagonistic selection could equally account for contemporary PCOS patterns of occurrence.37 38

Notwithstanding this abundance of ideas, there is general agreement that whatever the genetic legacy of our pre-industrial past, the modern phenotype has come about because of a profound mismatch between traits that have evolved to optimise fertility and survival in ancient times with modern energetic and living conditions.39 In the ancient world, average BMI of women was 18–21, total fertility rate was 6–7 births per woman and births were highly seasonal and tightly linked to rainfall, agricultural cycles and the price of grain. The unprecedented changes in average body weight, fertility and mortality, caloric intake and average levels of physical activity began in Western Europe over 200 years ago and now apply to more than 80% of the world’s population.40 41

There are two issues underpinning these hypotheses that need to be critically examined.

The first is whether there are ethnic differences in PCOS occurrence. Many of these hypotheses are premised on a view that prevalence is unvarying in human populations. Ethnic differences in disease incidence and prevalence have been used to infer adaptation to secular or geographic differences in environmental exposures or living conditions. The most pertinent example is the striking ethnic differences seen in the occurrence of type 2 diabetes with increasing body weight.42 43 Explanations for this hierarchy of susceptibility, and in particular the comparatively low prevalence in people of European descent, have included the duration of exposure to an agrarian, rather than a hunter-gatherer, diet, the length of time elapsed since nutritional and demographic transition, and monsoon-driven agricultural cycles.

It is unclear whether PCOS mirrors this pattern of difference, and in particular, the relative steepness of the rise in prevalence with increasing body weight. A recent review concluded that PCOS prevalence was lower in Asians compared to Europeans and Africans, and that Indigenous Australians had a very high prevalence of the disease.44 However, these comparisons are beset with the following problems of measurement error and bias:

  • The paucity of population-based estimates of PCOS prevalence and referral bias in clinic-based studies.
  • The inclusion of hirsutism as one of the diagnostic criteria potentially underestimates PCOS prevalence in Asian women.45
  • The almost complete absence of estimates of prevalence in women with a BMI less than 20 limits any meaningful comparison with average body weight in the past.
  • Changes in diagnostic criteria limit the ability to meaningfully compare prevalence studies over time.

The second is whether there is any evidence in contemporary populations of selection against the PCOS phenotype. Genetically based differences in lifetime reproductive success (LRS), or number of births per woman per lifetime, is a precondition for selection on the basis of fertility to occur. The cumulative probability of childbirth and the proportion achieving desired family size in women with PCOS over the longer term might be similar to that in women without PCOS;46 47 48 however, there are differences in parity and LRS. A recent large study using linked population data in Sweden and a clinical diagnosis rather than self-reported symptoms enabled precise estimation of LRS.49 Women with and without PCOS had an LRS of 0.8 births compared to 0.9 births per woman per lifetime, respectively. Furthermore, LRS differences were likely to be greater in women who did not undergo fertility treatment. While these differences may seem modest, they translate into a negative or purifying selection intensity of 10% per generation and, if sustained, would rapidly diminish the prevalence of the genetic determinants of PCOS. But even if rapid natural or purifying fertility selection against a PCOS phenotype had been happening over the last 1–200 years, it would have been tempered by the speed with which average body weight, and therefore expression of the PCOS phenotype, had been increasing.

Evolutionary medicine is a relatively new field that recognises that diseases need both proximate explanations of bodily mechanisms and evolutionary explanations of why natural selection has left the body vulnerable to disease. The insights gained have already contributed to important developments in the application of phylogenetics to biology of pathogen resistance and population genetics, and hold much promise in areas such as cancer and infectious disease control, reproductive medicine
and public health.50

The puzzle of the evolutionary paradox of PCOS remains unsolved, but evolutionary speculation suggests that studies which elucidate the physiology and epidemiology of insulin resistance and fertility in natural fertility populations may point to a solution. These studies may also help to resolve the question as to whether prolonged breastfeeding mitigates the risk of the development of subsequent type 2 diabetes in women with gestational diabetes.