Content uploaded by Livio Casarini
Author content
All content in this area was uploaded by Livio Casarini on Nov 13, 2014
Content may be subject to copyright.
The Polycystic Ovary Syndrome Evolutionary Paradox:
a Genome-Wide Association Studies–Based, in silico,
Evolutionary Explanation
Livio Casarini and Giulia Brigante
Unit of Endocrinology, Department of Biomedical, Metabolic, and Neural Sciences (L.C., G.B.), and
Center for Genomic Research (L.C.), University of Modena and Reggio Emilia, 41121 Modena, Italy
Context: Polycystic ovary syndrome (PCOS) is a common female endocrine disorder characterized
by phenotypes ranging from hyperandrogenism to metabolic disorders, more prevalent in people
of African/Caucasian and Asian ancestry. Because PCOS impairs fertility without diminishing in
prevalence, it was considered an evolutionary paradox. Genome-Wide Association Studies iden-
tified 17 single nucleotide polymorphisms (SNPs) associated with PCOS, with different allele fre-
quencies, ethnicity-related, in 11 susceptibility loci.
Objective: In this study we analyze the PCOS phenotype-genotype relationship in silico, using SNPs
of representative genes for analysis of genetic clustering and distance, to evaluate the degree of
genetic similarity.
Data Source: 1000 Genomes, HapMap, and Human Genome Diversity Project databases were used
as source of allele frequencies of the SNPs, using data from male and female individuals grouped
according to their geographical ancestry.
Setting and Design: Genetic clustering was calculated from SNPs data by Bayesian inference. The
inferred ancestry of individuals was matched with PCOS phenotype data, extracted from a previous
meta-analysis. The measure of genetic distance was plotted against the geographic distance be-
tween the populations.
Results: The individuals were assigned to five genetic clusters, matching with different world
regions (Kruskal-Wallis/Dunn’s post test; P⬍.0001), and converging in two main PCOS phenotypes
in different degrees of affinity. The overall genetic distance increased with the geographic distance
among the populations (linear regression; R
2
⫽0.21; P⬍.0001), in a phenotype-unrelated manner.
Conclusions: Phenotype-genotype correlations were demonstrated, suggesting that PCOS genetic
gradient results from genetic drift due to a serial founder effect occurred during ancient human
migrations. The overall prevalence of the disease supports intralocus sexual conflict as alternative to the
natural selection of phenotypic traits in females. (J Clin Endocrinol Metab 99: E2412–E2420, 2014)
Polycystic ovary syndrome (PCOS) is the most common
endocrinopathy affecting 5–10% of women in re-
productive age worldwide. It is a familial, polygenic con-
dition associated with infertility, irregular menstrual
cycles, anovulation, hyperandrogenism, as well as nonre-
productive health problems depending on genetic back-
ground and affected by lifestyle (1–3).
PCOS phenotypic features
Even if the disease displays a wide variety of charac-
teristics, it is widely accepted that PCOS features decrease
within two main phenotypes. According to the 2003 Rot-
terdam criteria (4), the prevalent clinical symptoms define
the hyperandrogenic or metabolic phenotype (5–10). The
hyperandrogenic PCOS phenotype is defined mainly by
ISSN Print 0021-972X ISSN Online 1945-7197
Printed in U.S.A.
Copyright © 2014 by the Endocrine Society
Received June 19, 2014. Accepted July 29, 2014.
First Published Online August 5, 2014
Abbreviations: GWAS, Genome-Wide Association Studies; PCOS, polycystic ovary syn-
drome; SNP, single nucleotide polymorphis.
JCEM ONLINE
Advances in Genetics—Endocrine Research
E2412 jcem.endojournals.org J Clin Endocrinol Metab, November 2014, 99(11):E2412–E2420 doi: 10.1210/jc.2014-2703
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 07 November 2014. at 07:29 For personal use only. No other uses without permission. . All rights reserved.
hirsutism, androgenic alopecia, and relatively high andro-
gen levels, whereas the metabolic phenotype is character-
ized by metabolic syndrome, insulin resistance, increased
risk for type 2 diabetes, and high body mass index or
central obesity (5–10). It can be approximately established
that the metabolic phenotype is prevalent in Central
Asians and Americans whereas the hyperandrogenic phe-
notype is prevalent in the other world regions. However,
the wide spectrum of secondary disorders results in an
overlap of PCOS characteristics among human popula-
tions, reflecting the polygenic condition of the disease and
genetic admixture. Curiously, no clear differences in the
prevalence of the disease among different ethnic groups
has been identified so far (10–15).
Evolutionary origin of PCOS
Given that the disease impairs fertility without dimin-
ishing its high global prevalence, it was extensively dis-
cussed as an evolutionary paradox. Previous studies at-
tempted to explain how a genetic pattern linked to
metabolic or reproductive disadvantages spread across
continents, generating a dozen theories suggesting differ-
ent explanations for the evolutionary origin of PCOS (16).
Most these hypotheses prompt a balancing mechanism
between viability selection and metabolic thrift against
fertility disadvantages associated with this condition. For
example, androgenisation and insulin resistance may con-
fer survival benefit to females and improve the glucose
availability for ovulatory functions in hunter-gatherer so-
cieties (17). Moreover, metabolic thrift and increased fat
storage are advantages for mother and fetus under low
food conditions (18). However, the effect on the individ-
ual fitness of the PCOS phenotype during the evolution of
humans is not understood, and no evolutionary advantage
for the PCOS genotype carriers has been proven. Surpris-
ingly, all previous theories about PCOS consider evolu-
tionary dynamics involving only females, not considering
the contribution of the male in the genotype-phenotype
inheritance and evolution. All the genetic and evolution-
ary analyses of PCOS were carried out on a sample of
female individuals, presumably resulting in biased evalu-
ations and in an overall loss of genetic information.
Genetic markers of the disease
Previous studies identified a hundred candidate genes
associated with PCOS and several genetic markers affect-
ing the pathogenesis, phenotype, and prevalence of the
disease have been proposed (19). Recently, two Genome-
Wide Association Studies (GWAS) performed in Han Chi-
nese women identified 11 new risk loci for PCOS (20, 21),
which count 17 single nucleotide polymorphisms (SNPs)
leading to genetic variants strongly associated with the
disease. The gene sequences located within the PCOS sus-
ceptibility loci are involved in the ovarian response to the
gonadotropic hormones, in the metabolism of glucose and
lipids, and in cell cycle regulation. This finding is corrob-
orated by other studies showing the association between
these markers and the disease (22–27). The results of these
two statistically powerful analyses were confirmed by
other works performed in populations of non-Chinese an-
cestry (19, 28, 29), showing a common genetic risk profile
across human populations.
All the genes falling within the susceptibility loci iden-
tified by GWAS may potentially be implicated in the mod-
ulation of the PCOS phenotype and its severity. These
genes are FSHR,LHCGR,DENND1A,THADA,
C9orf3,YAP1,HMGA2,RAB5B/SUOX,INSR,TOX3
and SUMO1P1; their potential relation with PCOS was
described separately (Supplemental Discussion).
Geography of PCOS genetic markers
The distribution of the allelic variants associated to
PCOS is different among the populations worldwide, as
observable by Human Genome Diversity Project (HGDP)
selection browser (http://hgdp.uchicago.edu/cgi-bin/gbrowse/
HGDP) (30–32), a web-based software, which calculates
the geographic distributions of user-selected markers from
Stanford SNP genotyping data (33, 34). A different genetic
pattern distribution of PCOS markers could be reflected in
different phenotypic features of the disease, resulting from
adaptive evolution (19, 28, 35) or from genetic drift gen-
erated by a serial founder effect occurring during the an-
cient human migrations out of Africa (36). Accordingly,
the decay of expected heterozygosity as measure of genetic
variation accompanies the increase of genetic and geo-
graphic distance from Africa (33, 34, 37). But the overall
constant prevalence of the disease remains unexplained.
The determination of the genetic background and its
relationship with the phenotype may be relevant to opti-
mize the pharmacological treatment of the disease and
protocols for assisted reproduction. To define the degree
of similarity of the PCOS genotypes, we show a popula-
tion genetics analysis by Bayesian clustering and an eval-
uation of pairwise genetic distance using SNPs data from
different populations, available in online databases. The
genotype-phenotype link and the correlation between ge-
netic and geographic data are discussed from an evolu-
tionary point of view.
Materials and Methods
A detailed description of the methods is available as Supple-
mental Materials and Methods.
doi: 10.1210/jc.2014-2703 jcem.endojournals.org E2413
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 07 November 2014. at 07:29 For personal use only. No other uses without permission. . All rights reserved.
SNP selection
The genetic analyses were performed using the frequencies
data of the 17 PCOS-related SNPs of individuals from human
populations sampled in different world regions. The SNP panel
(Supplemental Materials and Methods) was taken from two
GWAS (20, 21), which found a strong association between
PCOS and the 17 genetic markers. The SNP data were obtained
from the HGDP-CEPH (Centre d’Etude du Polymorphisme Hu-
main) Stanford (33, 34), 1000 Genomes (The 1000 Genomes
Project Consortium, 2010) (38) and in part from the HapMap
(The International HapMap Consortium, 2003) (39) panels,
which provide SNP frequencies and geographical coordinates of
a wide number of human populations worldwide, to ensure a
good coverage in the territorial distribution among the conti-
nents. Based on their geographic coordinates, the populations
were grouped by continent (Supplemental Table 1). All the ge-
netic data are from both male and female individuals unselected
for PCOS, ensuring that the analysis takes into account males as
carrier of a PCOS-linked genotype and avoiding the bias arising
from the use of only PCOS patients.
Selection of PCOS phenotypes
In order to evaluate the link between genotype and ethnicity,
the geographical distribution of the different PCOS phenotypes
was evaluated by analyzing the clinical data registered in the
scientific literature (Supplemental Materials and Methods). Phe-
notypic data are shown in a world map (Figure 1).
Human genetic clustering analysis
Genetic clustering analysis assigns the individuals to the
group (cluster) that best represents their genetic background,
calculated using the frequencies of PCOS markers. To this end,
SNPs data from individuals were used and the genetic population
stratification was inferred by the Bayesian analysis implemented
in STRUCTURE 2.3.4 software (Pritchard Lab) (40). The num-
ber of subpopulations (K) in which the individuals were assigned
was selected using the ⌬K method (41) (Supplemental Figure 1).
The degree of affinity of the populations to the resulting genetic
clusters is expressed as a numeric value (Q value) by the software.
Thus, Q values define an estimation of ancestry, inferred by the
SNP frequencies. Then, Q values were grouped for world area,
and used for a graphical representation together with geograph-
ical and clinical data.
The genotype-phenotype link was obtained from the analysis
of geographical data and genetic clustering. It was confirmed by
principal component analysis implemented in National Institute
on Aging (NIA) Array Analysis software (42) using the SNP
frequency data.
Evaluation of the genetic drift
To evaluate the contribution of the genetic drift in the estab-
lishment of the modern PCOS markers distribution, a linear re-
gression of the expected/observed heterozygosity and genetic
against geographic distance was performed (36). The SNPs panel
was used to obtain the heterozygosity data and to calculate the
Figure 1. World distribution of the affinity to the genetic clusters and PCOS phenotypes prevalence. A, Bar plots of individual Q values calculated
by the STRUCTURE software assign each individual to different subpopulations that matches the main world areas, with a certain degree of
admixture. The analysis was performed differentially for the HGDP and the 1000 Genomes merged together with the HapMap samples. Each color
indicates the membership of individuals in a genetic cluster (K ⫽5); Af, African; CA, Central Asian; ME, Mediterranean/Middle Eastern; Eu,
European; Am, American; Oc, Oceanian. B, Pie chart of cluster affinity among continents, indicating the frequencies of the PCOS susceptibility
markers. The charts were obtained as the means of the Q values by merging the HGDP, 1000 Genomes, and HapMap populations for each
genetic cluster (colors of the pie charts do not refer to panel A). The overall prevalence of a genetic cluster is different between the geographic
area, suggesting a link with the corresponding PCOS phenotype and clinic features, which were obtained by a review of the literature; green,
cluster 1; red, cluster 2; blue, cluster3; yellow, cluster 4; magenta, cluster 5; n.a., data not available or not assessed.
E2414 Casarini and Brigante PCOS Genotypes and Phenotypes J Clin Endocrinol Metab, November 2014, 99(11):E2412–E2420
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 07 November 2014. at 07:29 For personal use only. No other uses without permission. . All rights reserved.
fixation index (Fst) as a measure of pairwise genetic distance (43)
resulting by comparing each world population vs Africans. A
scatter plot of Fst and heterozygosity against the geographic dis-
tance was illustrated by the waypoints-method previously de-
scribed (36, 44). Briefly, the geographic distance between pop-
ulations was calculated using Addis Ababa in Ethiopia as the
starting point and taking into account the kilometers covered by
humans during their expansion worldwide through migratory
waypoints. The waypoints were chosen simulating the migratory
routes (Supplemental Table 2), assuming that humans bypassed
the natural obstacles during migrations, such as oceans and high
mountains. The radius of the Earth was also considered for the
calculation of the geographic distance.
Statistical analysis and image software
ANOVA, ttests, or linear regression analysis was applied as
appropriate and indicated in the figure legends. The statistical anal-
ysis was performed by GraphPad Prism software (GraphPad).
Results
Geographical distribution of PCOS genotypes and
phenotypes
Bayesian genetic clustering analysis assigned the indi-
viduals to five subpopulations (K ⫽5) obtained from the
estimation of the proportion of ancestry inferred by the
SNP frequencies. It can be defined by the Q value as the
degree of affinity of the population to each genetic cluster
by the SNPs combination of the individuals within the
indicated geographic area. Thus, the degree of prevalence
of each cluster is variable among the world continents,
revealing that human populations could be divided into
five groups with geographically different, non-homoge-
neous genetic background calculated using the frequency
of PCOS markers, though a degree of admixture exists
(Figure 1A). The proportion of affinity of the populations
to each genetic cluster is also illustrated for each world
area, which is represented by the prevalent PCOS geno-
type and phenotype (Figures 1B and 2). The metabolic
phenotype characterized by insulin resistance, metabolic
syndrome, type 2 diabetes, hypertension, and acanthosis
nigricans is dominant in Asia, especially Central Asia, and
in America. The hyperandrogenic phenotype predomi-
nates in European, Mediterranean, and Middle Eastern
patients, who show the most severe hirsutism. Consider-
ing African and Oceanian PCOS–affected women, the
phenotype is characterized by both hirsutism and insulin
resistance with a predominant role of the latter feature.
Indeed, a certain degree of mixed PCOS phenotypes co-
Figure 2. Box and whiskers plot of the overall continental membership in each genetic cluster. The continents are represented by the distribution
of the Q values calculated by the STRUCTURE software. The charts, A, were obtained as the means of the Q values by merging the HGDP, 1000
Genomes, and HapMap populations for each genetic cluster. Each cluster is peculiar for a specific geographic area because of the
nonhomogeneous distribution of PCOS susceptibility marker frequencies. The acronym above the bars indicates a significant difference vs Africa,
Af; America, Am; Central Asia, CA; Europe, Eu; East Asia, EA; Middle East/Mediterranean area, ME; and Oceania, Oc. Kruskal-Wallis and Dunn’s
post-test (P⬍.0001). B, Relationship between genetic background and PCOS phenotype. The genetic clusters are represented by colored ovals
and located in the position of the corresponding phenotype (Figure 1). Prevalence of each cluster in each world area is proportional to the area of
the oval, calculated using Q values; green, cluster 1; red, cluster 2; blue, cluster3; yellow, cluster 4; magenta, cluster 5. C, Clustering by principal
component (PC) analysis performed using SNP genotypes in relation to the metabolic and hyperandrogenic PCOS phenotypes (cumulative
percentage of data set coverage from PC1 and PC2 ⫽54.187%). The populations are represented by points colored depending on their
geographical origin (left panel) or prevalent phenotype (right panel).
doi: 10.1210/jc.2014-2703 jcem.endojournals.org E2415
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 07 November 2014. at 07:29 For personal use only. No other uses without permission. . All rights reserved.
exists in each world population, as a result of genetic ad-
mixture (Figure 1).
Genotype-phenotype link
The analysis of the geographical distribution of the
merged Q values reveals that the affinity to a predominant,
peculiar genetic cluster is typical for each continent (Figure
2). Thus, the genetic background inferred by PCOS mark-
ers is different and nonhomogeneous among the human
populations from diverse world areas (Figure 2A). Cluster
1 is predominant in Africans. Cluster 2 is typical in Central
Asian, European, and Mediterranean/Middle Eastern
people, but with a lesser extent of membership to the clus-
ter 1, revealing that these populations share a similar ge-
netic background, calculated using PCOS markers. How-
ever, Central Asians share a certain degree of membership
also to Cluster 5 together with East Asians. Cluster 3 is
typical of Americans and Eastern Asians. Cluster 4 is typ-
ical of Oceanians and partly of Africans. The degree of
severity of the two main PCOS phenotypes results from the
contribution of the genetic background. Each PCOS ge-
notype is linked to its geographic area with a peculiar
combination of prevalent PCOS features. This is shown by
cluster memberships, calculated as the Q value of the prev-
alent clusters in each geographic area (Figure 2B). Indeed,
cluster 1 is well represented in European, Mediterranean/
Middle Eastern, African, and American people, where hy-
perandrogenic hirsutism is mid/severe. Independently of
the PCOS features, cluster 2 is prevalent where the phe-
notype is more severe. Clusters 3 and 5 are represented in
association with the metabolic phenotype, which mainly
differs for the overall severity and type of its features. Af-
rican and Oceanian share a similar phenotype (cluster 4)
characterized by mid degrees of metabolic risk and hy-
perandrogenism, suggesting a genetic similarity for PCOS
markers between these populations, according to the re-
sult of genetic clustering at K ⫽4 (HGDP samples, Sup-
plemental Figure 1). The link between PCOS genotype and
phenotype was confirmed by principal component anal-
ysis (Figure 2C), given that the population characterized
by the metabolic or hyperandrogenic phenotype shares
peculiar graph areas. However, all clusters are represented
in each geographic area (Figures 1 and 2), probably con-
tributing to the PCOS features and its severity (Figure 2).
Genetic diversity resulting from PCOS markers
The measure of the genetic distance, Fst, calculated for
the panel of PCOS susceptibility markers reveals a strong
diversity in the genetic background among human popu-
lations, grouped by continents (Figure 3). The pairwise Fst
of the non-Africans vs Africans increases together with the
geographic distance from Africa, producing R
2
⫽0.21
Figure 3. Analysis of the genetic distance (Fst) and decay of heterozygosity between continents. The charts were obtained by merging the data
from HGDP, 1000 Genomes, and HapMap populations. A, Scatterplot of heterozygosity and geographic distance. The expected heterozygosity
calculated for PCOS susceptibility loci decreases together with geographic distance, evaluated by linear regression. B, Box and whiskers plot of Fst
distribution grouped for continents. The acronym above the bars indicates a significant difference vs Africa, Af; America, Am; Central Asia, CA;
Europe, Eu; East Asia, EA; Middle East/Mediterranean area, ME; and all other continents, ALL. Kruskal-Wallis and Dunn’s post-test (P⬍.0001). C,
Scatterplot of Fst and geographic distance calculated using the waypoints. Each point indicates the comparison between African vs other
populations, as specified in the figure legend. Fst calculated for PCOS susceptibility loci increases together with geographic distance, evaluatedby
linear regression.
E2416 Casarini and Brigante PCOS Genotypes and Phenotypes J Clin Endocrinol Metab, November 2014, 99(11):E2412–E2420
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 07 November 2014. at 07:29 For personal use only. No other uses without permission. . All rights reserved.
(P⬍.0001) calculated by linear regression (Figure 3C)
and indicating the strong contribution of genetic drift in
the establishment of PCOS markers distribution. The de-
cay of the expected heterozygosity (Figure 3A) strengthens
this observation. The result is corroborated by the increase
of Fst and decay of heterozygosity, calculated for a wider
range of genetic markers in a previous study, when plotted
against the distance from the putative starting point of
human migrations, in Ethiopia (36, 37). Differently from
what expected, the observed heterozygosity does not de-
cay with distance, suggesting a relatively recent genetic
admixture. Moreover, the Fst distributions in Africans vs
other populations (Figure 3A) is different among conti-
nents, indicating strongly diverse, nonhomogeneous allele
frequencies in the distribution of PCOS markers world-
wide (Figure 3).
Discussion
Previous stratification analyses using a wide number of
genetic markers showed that the modern human popula-
tion comprises six main genetic clusters depending on the
ethnic background, reflecting with surprising accuracy the
ethnicity and admixture degree of ancestry (45). Using the
PCOS markers human population was stratified into five
different genetic clusters falling within two main PCOS
phenotypic groups. Thus, PCOS results in a hyperandro-
genic and in a metabolic phenotype, reflecting the world
distribution of the degree of affinity to genetic clusters.
This analysis provides evidence that PCOS ethnic varia-
tions are strongly determined by the genetic background in
humans (27, 34, 46–48), as already demonstrated by a
comparative experiment between different PCOS mouse
strains (49). Genetic cluster analysis relies on the simul-
taneous combination of different SNPs, providing a higher
level of accuracy than case-control studies, which, in fact,
yielded conflicting results (50–53), not resulting in any
clear cause-effect indication related to ethnicity and pro-
viding a hundred putative markers not independently con-
firmed (19, 51). The study of ethnic variations of PCOS
genotype-phenotype link may be a useful approach for the
pharmacological treatment of the disease and during in-
fertility treatment.
PCOS phenotypes
The severity of the PCOS phenotypes may result from
different combinations of SNPs represented by the clusters
(Figure 1 and Figure 2), eg, Americans and Asians with a
prevalence of the metabolic phenotype, belonging to dif-
ferent prevalent clusters (3 and 2, respectively; Figures 1B
and 2). Conversely, European and Mediterranean/Middle
Eastern people share the hirsute-hyperandrogenic pheno-
type and a high affinity to cluster 2, but also the affinity to
clusters 1 and 4 is high among populations with a mid
hyperandrogenic phenotype, although characterized by
metabolic features rather than hirsutism. Thus, the ancient
humans’ migratory routes do not completely reflect the
current distribution of the PCOS phenotypes. Additional
considerations regarding the distribution of the genotypes
and phenotypes are available separately (Supplemental
Discussion).
PCOS and genetic diversity among humans
Although other alternatives were proposed and dis-
cussed (54, 55), Fst remains a widely used measure of
genetic distance. The pairwise Fst calculated for the var-
ious populations vs Africans increases together with the
geographic distance from Africa. This result is consistent
with previous data showing that genetic diversity is de-
termined by a serial founder effect that occurred during the
ancient human migrations across the continents (36). The
increase of Fst with the geographic distance calculated for
PCOS markers in humans is the same of that previously
observed for a wider set of genetic markers from different
organisms (36, 56–59), suggesting that its distribution is
the result of the random genetic drift. Nevertheless, the
variations of PCOS phenotypes among continents is dis-
played through various characteristics and symptoms ap-
parently incoherent with the geographic distance (Figure
3), probably as a result of genetic admixture rather than
natural selection. In fact, each world population shares a
different degree of affinity with the five genetic clusters,
and therefore it may differently contribute to phenotype
determination. Limitations due to genetic admixture were
discussed separately (Supplemental Discussion).
All things considered, the overall continental variabil-
ity of the two main PCOS phenotypes is clearly linked to
a peculiar genetic background, resulting from genetic drift
and indicating that different genetic markers may reflect
convergent phenotypic features.
Natural selection or genetic drift?
These data suggest that PCOS genotype and phenotype
may be not strongly affected by natural selection during
human evolution. Nevertheless, the reason why the prev-
alence of PCOS is similar among the different world con-
tinents remains to be demonstrated. A large number of
evolution-based theories were produced and extensively
discussed (16), providing rational evaluations for the evo-
lution of different PCOS phenotypes in females, especially
the metabolic phenotype, but not for the overall constant
prevalence of the disease. Surprisingly, none of these the-
ories considers the male as the carrier of an hyperandro-
doi: 10.1210/jc.2014-2703 jcem.endojournals.org E2417
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 07 November 2014. at 07:29 For personal use only. No other uses without permission. . All rights reserved.
genic trait, which may reasonably provide an improve-
ment of his individual fitness, likely emphasizing the male
secondary sex characteristics. Indeed, both sexes share
most of their genomes and express the same traits, how-
ever resulting in antagonistic selection (60, 61). From this
point of view, the prevalence of PCOS among different
human populations may be the result of the balance be-
tween a positive selection in males against a negative se-
lection in females. Our results are strengthened by the
demonstration that natural selection favors in similar pro-
portions both protective and risk alleles for type 2 diabetes
(62), suggesting that the phenotype linked to the disease
results in opposite effects on the fitness of both sexes. Pre-
vious observations in other mammals support this hypoth-
esis, since the selection for the sex hormone testosterone
leads to antagonistic reproductive fitness between parents
and their opposite-sex progeny (63). On the other hand, it
is well known that evolution is often sex-dependent in
different species (64–67), given that alleles can have pos-
itive effects on fitness in one sex and negative in the other,
resulting in intralocus sexual conflict (61, 68). Different
genetic background for PCOS converging in two main
phenotypes together with overall constant prevalence of
the disease support the presence of intralocus sexual con-
flict, which may have affected the decay of observed
heterozygosity. Even if speculative, in humans this mech-
anism seems to be a “bug” inherited from admixture with
different hominids or from an ancient genome evolved in
environmental and social conditions, strongly different
from those in which Homo sapiens lived in the last
100 000 years (69, 70), but every hypothesis in this regard
must be demonstrated.
Summary
The phenotypic expression of PCOS varies among hu-
man populations, depending on ethnicity. The distribu-
tion of previously identified susceptibility disease markers
results in different, nonhomogeneous continental genetic
backgrounds by Bayesian clustering, reflecting the ethnic
distribution of the main PCOS phenotypes. Thus, a clear
indication for PCOS ethnicity is shown, taking into ac-
count a certain degree of genetic admixture between hu-
man populations. The genetic distance increases together
with the distance from Africa, suggesting that the modern
distribution of PCOS susceptibility markers is the result of
genetic drift likely due to a serial founder effect occurred
during the ancient human migrations as alternative to the
natural selection theory. Intralocus sexual conflict may
contribute to the maintenance of an overall constant prev-
alence of PCOS measured in females. The analysis of the
genetic background may lead to important implications
for the pharmacological approach to the disease.
Acknowledgments
We thank Professor Manuela Simoni for her commitment, sup-
port, and guidance in the field of endocrinology.
Address all correspondence and requests for reprints to: Livio
Casarini, PhD, Unit of Endocrinology, Nuovo Ospedale Civile
Sant’Agostino Estense (NOCSAE), Via P. Giardini 1355, 41126
Modena, Italy. E-mail: livio.casarini@unimore.it.
This work was supported by a grant of the Italian Ministry of
Education, University and Research, No. PRIN 2010C8ERKX.
Disclosure Summary: The authors have nothing to disclose.
References
1. Domecq JP, Prutsky G, Mullan RJ, et al. Lifestyle modification pro-
grams in polycystic ovary syndrome: Systematic review and meta-
analysis. J Clin Endocrinol Metab. 2013;98:4655–4663.
2. Legro RS, Arslanian SA, Ehrmann DA, et al. Diagnosis and treat-
ment of polycystic ovary syndrome: An Endocrine Society Clinical
Practice Guideline. J Clin Endocrinol Metab. 2013;98:4565–4592.
3. Moran LJ, Ranasinha S, Zoungas S, McNaughton SA, Brown WJ,
Teede HJ. The contribution of diet, physical activity and sedentary
behaviour to body mass index in women with and without polycystic
ovary syndrome. Hum Reprod. 2013;28:2276–2283.
4. Rotterdam ESHRE/ASRM-Sponsored PCOS consensus workshop
group. Revised 2003 consensus on diagnostic criteria and long-term
health risks related to polycystic ovary syndrome (PCOS). Hum
Reprod. 2004;19:41–47.
5. Kauffman RP, Baker TE, Baker VM, DiMarino P, Castracane VD.
Endocrine and metabolic differences among phenotypic expressions
of polycystic ovary syndrome according to the 2003 Rotterdam con-
sensus criteria. Am J Obstet Gynecol. 2008;198:670.e1–7.
6. Mojiminiyi OA, Safar FH, Al Rumaih H, Diejomaoh M. Variations
in alanine aminotransferase levels within the normal range predict
metabolic and androgenic phenotypes in women of reproductive
age. Scand J Clin Lab Invest. 2010;70:554–560.
7. Wijeyaratne CN, Seneviratne Rde A, Dahanayake S, et al. Phenotype
and metabolic profile of South Asian women with polycystic ovary
syndrome (PCOS): Results of a large database from a specialist En-
docrine Clinic. Hum Reprod. 2011;26:202–213.
8. Dunaif A, Fauser BC. Renaming PCOS—a two-state solution. J Clin
Endocrinol Metab. 2013;98:4325–4328.
9. Zhang HY, Guo CX, Zhu FF, Qu PP, Lin WJ, Xiong J. Clinical
characteristics, metabolic features, and phenotype of Chinese
women with polycystic ovary syndrome: A large-scale case-control
study. Arch Gynecol Obstet. 2013;287:525–531.
10. Zhao Y, Qiao J. Ethnic differences in the phenotypic expression of
polycystic ovary syndrome. Steroids. 2013;78:755–760.
11. Diamanti-Kandarakis E, Kouli CR, Bergiele AT, et al. A survey of
the polycystic ovary syndrome in the Greek island of Lesbos: Hor-
monal and metabolic profile. J Clin Endocrinol Metab. 1999;84:
4006–4011.
12. Asunción M, Calvo RM, San Millán JL, Sancho J, Avila S, Escobar-
Morreale HF. A prospective study of the prevalence of the polycystic
ovary syndrome in unselected Caucasian women from Spain. J Clin
Endocrinol Metab. 2000;85:2434–2438.
13. Kumarapeli V, Seneviratne Rde A, Wijeyaratne CN, Yapa RM,
Dodampahala SH. A simple screening approach for assessing com-
munity prevalence and phenotype of polycystic ovary syndrome in
a semi-urban population in Sri Lanka. Am J Epidemiol. 2008;168:
321–328.
14. Azziz R, Woods KS, Reyna R, Key TJ, Knochenhauer ES, Yildiz BO.
The prevalence and features of the polycystic ovary syndrome in an
E2418 Casarini and Brigante PCOS Genotypes and Phenotypes J Clin Endocrinol Metab, November 2014, 99(11):E2412–E2420
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 07 November 2014. at 07:29 For personal use only. No other uses without permission. . All rights reserved.
unselected population. J Clin Endocrinol Metab. 2004;89:2745–
2749.
15. Wang S, Alvero R. Racial and ethnic differences in physiology and
clinical symptoms of polycystic ovary syndrome. Semin Reprod
Med. 2013;31:365–369.
16. Corbett S, Morin-Papunen L. The polycystic ovary syndrome and
recent human evolution. Mol Cell Endocrinol. 2013;373:39–50.
17. Azziz R, Dumesic DA, Goodarzi MO. Polycystic ovary syndrome:
An ancient disorder? Fertil Steril. 2011;95:1544–1548.
18. Holte J, Bergh T, Berne C, Berglund L, Lithell H. Enhanced early
insulin response to glucose in relation to insulin resistance in women
with polycystic ovary syndrome and normal glucose tolerance. J Clin
Endocrinol Metab. 1994;79:1052–1058.
19. Louwers YV, Stolk L, Uitterlinden AG, Laven JS. Cross-ethnic meta-
analysis of genetic variants for polycystic ovary syndrome. J Clin
Endocrinol Metab. 2013;98:E2006–12.
20. Chen ZJ, Zhao H, He L, et al. Genome-wide association study iden-
tifies susceptibility loci for polycystic ovary syndrome on chromo-
some 2p16.3, 2p21 and 9q33.3. Nat Genet. 2011;43:55–59.
21. Shi Y, Zhao H, Shi Y, et al. Genome-wide association study identifies
eight new risk loci for polycystic ovary syndrome. Nat Genet. 2012;
44:1020–1025.
22. Simoni M, Tempfer CB, Destenaves B, Fauser BC. Functional ge-
netic polymorphisms and female reproductive disorders: Part I:
Polycystic ovary syndrome and ovarian response. Hum Reprod Up-
date. 2008;14:459–484.
23. Unsal T, Konac E, Yesilkaya E, et al. Genetic polymorphisms of
FSHR,CYP17,CYP1A1,CAPN10,INSR,SERPINE1 genes in ad-
olescent girls with polycystic ovary syndrome. J Assist Reprod
Genet. 2009;26:205–216.
24. Gu BH, Park JM, Baek KH. Genetic variations of follicle stimulating
hormone receptor are associated with polycystic ovary syndrome.
Int J Mol Med. 2010;26:107–112.
25. Eriksen MB, Brusgaard K, Andersen M, et al. Association of poly-
cystic ovary syndrome susceptibility single nucleotide polymor-
phism rs2479106 and PCOS in Caucasian patients with PCOS or
hirsutism as referral diagnosis. Eur J Obstet Gynecol Reprod Biol.
2012;163:39–42.
26. Goodarzi MO, Jones MR, Li X, et al. Replication of association of
DENND1A and THADA variants with polycystic ovary syndrome
in European cohorts. J Med Genet. 2012;49:90–95.
27. Li T, Zhao H, Zhao X, et al. Identification of YAP1 as a novel
susceptibility gene for polycystic ovary syndrome. J Med Genet.
2012;49:254–257.
28. Welt CK, Styrkarsdottir U, Ehrmann DA, et al. Variants in
DENND1A are associated with polycystic ovary syndrome in
women of European ancestry. J Clin Endocrinol Metab. 2012;97:
E1342–1347.
29. Mutharasan P, Galdones E, Peñalver Bernabé B, et al. Evidence for
chromosome 2p16.3 polycystic ovary syndrome susceptibility locus
in affected women of European ancestry. J Clin Endocrinol Metab.
2013;98:E185–190.
30. Coop G, Pickrell JK, Novembre J, et al. The role of geography in
human adaptation. PLoS Genet. 2009; 5:e1000500.
31. Pickrell JK, Coop G, Novembre J, et al. Signals of recent positive
selection in a worldwide sample of human populations. Genome
Research. 2009;19:826–837.
32. Pritchard JK, Pickrell JK, Coop G. The genetics of human adapta-
tion: Hard sweeps, soft sweeps, and polygenic adaptation. Curr Biol.
2010;20:R208–215.
33. Cann HM, de Toma C, Cazes L, et al. A human genome diversity cell
line panel. Science. 2002;296:261–262.
34. Cavalli-Sforza LL. The human genome diversity project: Past, pres-
ent and future. Nat Rev Genet. 2005;6:333–340.
35. Cui L, Zhao H, Zhang B, et al. Genotype-phenotype correlations of
PCOS susceptibility SNPs identified by GWAS in a large cohort of
Han Chinese women. Hum Reprod. 2013;28:538–544.
36. Ramachandran S, Deshpande O, Roseman CC, Rosenberg NA,
Feldman MW, Cavalli-Sforza LL. Support from the relationship of
genetic and geographic distance in human populations for a serial
founder effect originating in Africa. Proc Natl Acad SciUSA. 2005;
102:15942–15947.
37. Li JZ, Absher DM, Tang H, et al. Worldwide human relationships
inferred from genome-wide patterns of variation. Science. 2008;
319:1100–1104.
38. Abecasis GR, Altshuler D, Auton A, et al. A map of human genome
variation from population-scale sequencing. Nature. 2010;467:
1061–1073.
39. The International HapMap Consortium. The International Hap-
Map Project. Nature. 2003;426:789–796.
40. Pritchard JK, Stephens M, Donnelly P. Inference of population struc-
ture using multilocus genotype data. Genetics. 2000;155:945–959.
41. Evanno G, Regnaut S, Goudet J. Detecting the number of clusters of
individuals using the software STRUCTURE: a simulation study.
Mol Ecol. 2005;14:2611–2620.
42. Sharov AA, Dudekula DB, Ko MS. A web-based tool for principal
component and significance analysis of microarray data. Bioinfor-
matics. 2005;21:2548–2549.
43. Wright S. Isolation by distance. Genetics. 1943;28:114–138.
44. Henn BM, Cavalli-Sforza LL, Feldman MW. The great human ex-
pansion. Proc Natl Acad SciUSA. 2012;109:17758–17764.
45. Rosenberg NA, Pritchard JK, Weber JL, et al. Genetic structure of
human populations. Science. 2002;298:2381–2385.
46. Valkenburg O, Uitterlinden AG, Piersma D, et al. Genetic polymor-
phisms of GnRH and gonadotrophic hormone receptors affect the
phenotype of polycystic ovary syndrome. Hum Reprod. 2009;24:
2014–2022.
47. Hwang JY, Lee EJ, Jin Go M, et al. Genome-wide association study
identifies GYS2 as a novel genetic factor for polycystic ovary syn-
drome through obesity-related condition. J Hum Genet. 2012;57:
660–664.
48. Zadeh-Vakili A, Ramezani Tehrani F, Daneshpour MS, Zarkesh M,
Saadat N, Azizi F. Genetic polymorphism of vitamin D receptor gene
affects the phenotype of PCOS. Gene. 2013;515:193–196.
49. Dowling AR, Nedorezov LB, Qiu X, Marino JS, Hill JW. Genetic
factors modulate the impact of pubertal androgen excess on insulin
sensitivity and fertility. PLoS One. 2013;8:e79849.
50. Fu L, Zhang Z, Zhang A, et al. Association study between FSHR
Ala307Thr and Ser680Asn variants and polycystic ovary syndrome
(PCOS) in Northern Chinese Han women. J Assist Reprod Genet.
2013;30:717–721.
51. Pau C, Saxena R, Welt CK. Evaluating reported candidate gene
associations with polycystic ovary syndrome. Fertil Steril. 2013;99:
1774–1778.
52. Skrgati L, Baldani DP, Gersak K, Cerne JZ, Ferk P, Cori M. Genetic
polymorphisms of INS, INSR and IRS-1 genes are not associated
with polycystic ovary syndrome in Croatian women. Coll Antropol.
2013;37:141–146.
53. Xu P, Shen SM, Zhang XL, et al. Haplotype analysis of single nu-
cleotide polymorphisms in anti-Müllerian hormone gene in Chinese
PCOS women. Arch Gynecol Obstet. 2013;288:125–130.
54. Nei M. Analysis of gene diversity in subdivided populations. Proc
Natl Acad SciUSA. 1973;70:3321–3323.
55. Excoffier L, Smouse PE, Quattro JM. Analysis of molecular variance
inferred from metric distances among DNA haplotypes: Application
to human mitochondrial DNA restriction data. Genetics. 1992;131:
479–491.
56. Linz B, Balloux F, Moodley Y, et al. An African origin for the in-
timate association between humans and Helicobacter pylori. Na-
ture. 2007;445:915–918.
57. Stewart JF, Liu Y, Tauer CG, Nelson CD. Microsatellite versus
AFLP analyses of pre-management introgression levels in loblolly
pine (Pinus taeda L.) and shortleaf pine. (P. echinata Mill.). Tree
Genetics, Genomes. 2010;6:853–862.
58. Legrand D, Vautrin D, Lachaise D, Cariou ML. Microsatellite vari-
ation suggests a recent fine-scale population structure of Drosophila
doi: 10.1210/jc.2014-2703 jcem.endojournals.org E2419
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 07 November 2014. at 07:29 For personal use only. No other uses without permission. . All rights reserved.
sechellia, a species endemic of the Seychelles archipelago. Genetica.
2011;139:909–919.
59. Taylor SM, Antonia AL, Parobek CM, et al. Plasmodium falciparum
sulfadoxine resistance is geographically and genetically clustered
within the DR Congo. Sci Rep. 2013;3:1165.
60. Cox RM, Calsbeek R. Sexually antagonistic selection, sexual di-
morphism, and the resolution of intralocus sexual conflict. Am Nat.
2009;173:176–187.
61. Pennell TM, Morrow EH. Two sexes, one genome: The evolution-
ary dynamics of intralocus sexual conflict. Ecol Evol. 2013;3:1819 –
1834.
62. Ayub Q, Moutsianas L, Chen Y, et al. Revisiting the thrifty gene
hypothesis via 65 loci associated with susceptibility to type 2 dia-
betes. Am J Hum Genet. 2014;94:176–185.
63. Mills SC, Koskela E, Mappes T. Intralocus sexual conflict for fitness:
Sexually antagonistic alleles for testosterone. Proc Biol Sci. 2012;
279:1889–1895.
64. Mainguy J, Côté SD, Festa-Bianchet M, Coltman DW. Father-off-
spring phenotypic correlations suggest intralocus sexual conflict for
a fitness-linked trait in a wild sexually dimorphic mammal. Proc Biol
Sci. 2009;276:4067–4075.
65. Delph LF, Andicoechea J, Steven JC, Herlihy CR, Scarpino SV, Bell
DL. Environment-dependent intralocus sexual conflict in a dioe-
cious plant. New Phytol. 2011;192:542–552.
66. Lewis Z, Wedell N, Hunt J. Evidence for strong intralocus sexual
conflict in the Indian meal moth, Plodia interpunctella. Evolution.
2011;65:2085–2097.
67. Postma E, Spyrou N, Rollins LA, Brooks RC. Sex-dependent selec-
tion differentially shapes genetic variation on and off the guppy Y
chromosome. Evolution. 2011;65:2145–2156.
68. Stearns SC, Govindaraju DR, Ewbank D, Byars SG. Constraints on
the coevolution of contemporary human males and females. Proc
Biol Sci. 2012;279:4836–4844.
69. Vasseur E, Quintana-Murci L. The impact of natural selection on
health and disease: Uses of the population genetics approach in hu-
mans. Evol Appl. 2013;6:596–607.
70. Wang S, Lachance J, Tishkoff SA, Hey J, Xing J. Apparent Variation
in Neanderthal Admixture among African Populations is Consistent
with Gene Flow from non-African Populations. Genome Biol Evol.
2013;5:2075–2081.
E2420 Casarini and Brigante PCOS Genotypes and Phenotypes J Clin Endocrinol Metab, November 2014, 99(11):E2412–E2420
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 07 November 2014. at 07:29 For personal use only. No other uses without permission. . All rights reserved.
