Endocrine, vol. 30, no. 1, 19–26, August 2006 0969–711X/06/30:19–26/$30.

00
ENDO (Online) ISSN 1559-0100 © 2006 by Humana Press Inc. All rights of any nature whatsoever reserved.

The Role of Genes and Environment in the Etiology of PCOS

Evanthia Diamanti-Kandarakis,1 Helen Kandarakis,1 and Richard S. Legro2
1
Endocrine Section, First Department of Medicine, University of Athens Medical School, Athens, Greece;
and 2Department of Obstetrics and Gynecology, Pennsylvania State University College of Medicine, Hershey, PA, USA

Both genes and the environment contribute to PCOS. ment in PCOS, and unfortunately there has been little work
Obesity, exacerbated by poor dietary choices and phys- in gene–environment interactions in this area. The overall
ical inactivity, worsens PCOS in susceptible individu- hypothesis is that a genetic predisposition combined with
als. The role of other environmental modifiers such as a favorable environment leads to PCOS.
infectious agents or toxins are speculative. Phenotype
confusion has characterized genetic studies of PCOS. The Diagnostic Dilemna of PCOS
Although several loci have been proposed as PCOS
The diagnosis of PCOS has traditionally been based on a
genes including CYP11A, the insulin gene, the follista-
history of oligomenorrhea and/or hyperandrogenism, either
tin gene, and a region near the insulin receptor, the
clinical, i.e., most commonly hirsutism, or biochemical, i.e.,
evidence supporting linkage is not overwhelming. The
elevated circulating total or bioavailable androgens; and/
strongest case can be made for the region near the insu-
or polycystic ovaries. The criteria that emerged from the
lin receptor gene (but not involving this gene), as it has
1990 NIH–NICHD conference identified PCOS as unex-
been identified in two separate studies, and perhaps most
plained hyperandrogenic chronic anovulation, making it
importantly has not yet been refuted by larger studies.
in essence a diagnosis of exclusion (1). The “consensus”
However, the responsible gene at chromosome 19p13.3
definition did not include the polycystic ovary morphol-
remains to be identified. To date, no gene has been
ogy, most commonly found today on ultrasound consisting
identified that causes or contributes substantially to the
of multiple 2–8 mm subcapsular preantral follicles and
development of a PCOS phenotype.
increased ovarian volume (2); however, these ultrasound
criteria for polycystic ovaries are also a shifting target (3).
Key Words: Insulin resistance; familial studies; associ-
Ultrasound criteria were recently incorporated in the revised
ation studies; linkage; obesity.
2003 Rotterdam diagnostic criteria, which requires two out
of the three outlined cardinal stigmata for PCOS: oligome-
norrhea, hyperandrogenism, and/or polycystic ovaries (4).
Introduction
Polycystic ovary syndrome (PCOS) is the most common, The Difficult Genetics of Complex Diseases
but least understood, endocrinopathy. Although a major
None of these expert generated definitions include insu-
genetic contribution is suspected, there have been no clear
lin resistance, a common but not inevitable finding in PCOS.
genes or family of genes identified that cause or contribute
This diagnostic dilemma has hampered clinical and genetic
to PCOS. There are a number of environmental factors that
studies of PCOS. The larger the number of distinct pheno-
clearly also contribute to PCOS, most notably obesity. The
types within the affected category, the more complex the
rapid emergence of PCOS as a major endocrinopathy has
genetic analysis and the greater the likelihood that inves-
gone hand in hand with the epidemics of obesity and type
tigators using different diagnostic criteria will arrive at dif-
2 diabetes sweeping developed countries. While these dis-
ferent conclusions. Other factors that contribute to difficulty
orders also have a genetic component, clearly the environ-
in performing genetic studies in PCOS include the associ-
ment has changed radically in the last century to further
ated infertility and low fecundity (selection bias against the
these epidemics and it is reasonable to assume that similar
transmission of PCOS genes). Thus, it is rare to find large
changes have also factored in the PCOS epidemic. This
pedigrees with multiple affected women with which to per-
article will explore the contributions of genes and environ-
form linkage analysis.
Another difficulty is assigning phenotypes to premen-
Received December 11, 2005; Accepted December 11, 2005. archal girls and postmenopausal women (who lack menses,
Author to whom all correspondence and reprint requests should be addressed: and have indeterminate ovarian morphology, and no clear
Richard S. Legro, MD, Department of Ob/Gyn, PO Box 850, 500 University
Drive, M.S. Hershey Medical Center, Hershey PA, 17033. E-mail: rsli@psu. cutoffs for hyperandrogenemia), a problem that also limits
edu the utilization of pedigrees. Although a male phenotype has

19
20 Role of Genes and Environment in the Etiology of PCOS / Diamanti-Kandarakis, Kandarakis, and Legro Endocrine

been postulated, there are no rigorously established clinical in most studies as the criteria for diagnosing PCOS in pro-
or biochemical features that can be used to identify “PCOS bands. Earlier studies tended to characterize relatives on
males.” This makes formal segregation analysis as well as the basis of questionnaires, whereas later studies focused
genetic linkage studies more difficult. Finally, the lack of on more intensive phenotyping. More recent studies have
animals that spontaneously develop a PCOS-like pheno- focused on hyperandrogenemia, which appears to be one
type, especially mice, precludes the use of usually powerful of the strongest genetically inherited characteristics in fam-
tools of genetic mapping. The animal model (rhesus mon- ilial cases (6,7).
key) that best reflects the complex metabolic and reproduc- These studies strongly support the clustering of reproduc-
tive abnormalities found in women with PCOS has resulted tive abnormalities such as clinical and biochemical hyper-
from prenatal iatrogenic androgen exposure in utero, at least androgenism, polycystic ovaries, and, to a lesser degree,
providing support for the idea of some potential endocrine oligomenorrhea in first degree relatives. Similarly, there
disruptor contributing to PCOS in humans, but most im- appears to be an associated increased prevalence of biochem-
portantly the role of the environment in shaping PCOS (5). ical insulin resistance, hyperinsulinemia in first degree rela-
The idea that PCOS is a heritable disease is supported by tives. In one study 50% of sisters of women diagnosed with
studies documenting ethnic predisposition, familial aggre- PCOS had elevated total or bioavailable testosterone levels,
gation, concordant twin studies, and association with other suggesting that hyperandrogenemia is a common trait, and
Mendelian disorders. This chapter will review the evidence has a bimodal distribution suggesting a dominant form of
that PCOS is a genetic disease and will briefly overview the inheritance (6).
largely negative genetic marker data that have emerged in
the last decade. Twin Studies

Familial Aggregation There have been a paucity of studies examining twins

affected with PCOS. Twin studies offer a method for exam-
The foundation of genetic studies is the evidence that ining the relative contributions of genes and environment
disease clusters in families. Although familial clustering by comparing the prevalence and penetrance of traits in
may also occasionally have environmental causes (selec- monozygotic (MZ) to that in dizygotic (DZ) twins. A higher
tive exposure to the putative agent differentially affects prevalence of the trait and more complete penetrance in
family members), but there has been no environmental MZ compared to DZ twins is strongly supportive of a trait
endocrine disruptor or infectious agent identified to date under genetic influence. A twin study from Australia (7)
that causes PCOS. None of the existing family studies of reported on both MZ and DZ twins, who were studied with
PCOS convincingly establishes a mode of inheritance either ultrasound as well as clinical and biochemical parameters.
because the number of families studied was too small; the This study noted a high degree of discordance among the
parental phenotypes could not be firmly established; and/ twins for polycystic ovaries on ultrasound; however, there
or the male phenotype is uncertain. Despite the heteroge- was a significant genetic component to androgen and insu-
neity in study design and the inability to obtain comprehen- lin levels (7,8). A recent large twin series applied Rotterdam
sive phenotype information to permit a formal segregation PCOS diagnostic criteria to a large existing sample (with
analysis, collectively the existing literature strongly sug- the exception of no ultrasound data). Clinical signs and
gests the clustering of PCOS in families with a mode of symptoms from 1332 monozygotic twins (genetically iden-
inheritance suggestive of an autosomal dominant pattern tical) and 1873 dizygotic twins/singleton sisters of twins
(Table 1). This may equally suggest a publication bias as registered with The Netherlands Twin Register pointed to
subsequent analyses of genetic markers utilizing larger sam- a strong contribution of genetic factors to PCOS. The resem-
ple size often utilize a non-Mendelian model mode of gene- blance in MZ twin sisters (tetrachoric correlation = 0.71,
tic analysis (see below). This may also reflect that fact that these correlations are an index of twin similarity with higher
retrospective series tend to identify large families through rates associated with higher heritability) for PCOS was
selection bias, whereas prospective series tend to identify about twice as large as in DZ twin and other sisters (tetra-
smaller families, perhaps better indicative of low fecundity choric correlation = 0.38) (9).
in these families.
These studies need to be critically judged based on their
Modes of Genetic Analysis
strengths and weakness. Diagnostic criteria used to assign
Utilized in PCOS Studies
affected status differed among the studies as did the meth-
ods with which the status of first- and second-degree rela- A variety of approaches have been used to find genes
tives were ascertained (Table 1). By and large, ovarian mor- that contribute to PCOS including both genetic association
phology determined from tissue biopsy, direct visualization, studies in unrelated probands and familial linkage studies.
or diagnostic imaging, in association with menstrual distur- Both approaches to date have been characterized by rely-
bances and evidence for hyperandrogenism, have been used ing on the candidate gene approach, which is again ham-
Vol. 30, No. 1 Role of Genes and Environment in the Etiology of PCOS / Diamanti-Kandarakis, Kandarakis, and Legro 21

Table 1
Summary of Diagnostic Criteria for the Proband in Familial Studies of PCOS and Proposed Mode on Inheritance
Author Diagnostic criteria for PCOS Number studied Mode of inheritance/familial cluster
Cooper et al. Oligomenorrhea, hirsutism, polycystic 18 PCOS women and their Autosomal dominant with reduced
(41) ovaries (by culdoscopy, gynecography, first degree relatives and a penetrance
or wedge resection) control group
Givens et al. Oligomenorrhea, hirsutism, and 3 multigeneration kindreds (?X-linked) dominant
(42) polycystic ovaries (exam and surgery)
Ferriman and Hirsutism and/or oligomenorrhea, 381 PCOS women and relatives Modified dominant
Purdie (43) 60% with polycystic ovaries and a control group
(by air contrast gynecography)
Hague et al. Clinical symptoms (menstrual 50 PCOS women and Segregation ratios exceeded
(44) dysfunction, hyperandrogenism, 17 women with CAH and a autosomal dominant pattern
obesity, and infertility) and polycystic control group
ovaries by transabdominal ultrasound
Lunde et al. Clinical symptoms (menstrual 132 PCOS women and first Unclear, most consistent with
(45) irregularities, hirsutism, infertility, and second degree relatives autosomal dominant
and obesity) and multicystic ovaries and a control group
on wedge resection
Carey et al. Polycystic ovaries (by transabdominal 10 kindreds and 62 relatives Autosomal dominant with 90%
(15) ultrasound) penetrance
Norman et al. Elevated androgens, decreased SHBG, 5 families with 24 females Not stated, metabolic abnormalities
(46) and polycystic ovaries on ultrasound and 8 males associated with insulin resistance
in families
Legro et al. Elevated testosterone levels combined 80 PCOS probands and Hyperandrogenemia consistent with
(6) with oligomenorrhea (≤ 6 menses/yr) 115 sisters an autosomal dominant trait
Govind et al. Polycystic ovary morphology on 29 families with 53 sisters Autosomal dominant
(47) ultrasonograpy and 18 brothers
Kahsar-Millar Oligomenorrhea and either hirsutism 90 PCOS probands, 50 sisters, Increased prevalence of symptoms in
et al. (48) or elevated testosterone levels 78 mothers in first degree relatives suggesting
genetic trait
Legro et al. Elevated testosterone levels combined 346 PCOS probands, Increased prevalence of hyper-
(25) with oligomenorrhea (≤ 6 menses/yr) 307 sisters insulinemia in sisters segregating
with hyperandrogenemia
Legro et al. Elevated testosterone levels combined 87 PCOS probands, Increased DHEAS levels in
(49) with oligomenorrhea (≤ 6 menses/yr) 119 brothers brothers
Sir-Petermann Hyperandrogenism (elevated free 106 PCOS Probands, Increased diabetes (approx twofold)
et al. (50) androgen index or hirsutism) AND 200 parents and insulin resistance among
oligomenorrhea PCOS parents
Yildiz et al. Elevated testosterone levels AND 52 PCOS Probands, Elevated testosterone levels in
(51) oligomenorrhea 102 first degree relatives sisters and mothers, increased
rates of glucose intolerance in
first degree family members
Kaushal et al. Polycystic ovary morphology on 20 PCO probands Increased prevalence of insulin
(52) ultrasonograpy and 20 brothers resistance and endothelial
dysfunction in brothers.

pered by the same drawbacks as our diagnostic criteria, lar gene or region, and it has not so far helped us to get in
inadequate understanding of the fundamental pathophysi- any way closer to the exit of this genomic labyrinth. There
ology of PCOS. Molecular defects in such candidate genes has been no genome-wide scan performed in either asso-
as gonadotropins and their receptors, in enzymes involved ciation or linkage studies, and one problem has been the
in steroidogenesis, as well as those underlying insulin ac- failure to accumulate a sufficient number of individuals or
tion and secretion pathways, have been under continuous families to perform adequately powered studies using these
and intense investigation with variable results. broader tools. Thus, findings of positive and negative genes
Although this scientific race is quite stimulatory, it has in PCOS must be corrected for the relatively small sample
not been possible to establish firmly the role of any particu- size and the limitations of our vision.
22 Role of Genes and Environment in the Etiology of PCOS / Diamanti-Kandarakis, Kandarakis, and Legro Endocrine

Table 2 notypes. Finally, an association may exist, but not because

Common Flaws in Genetic Association Studies the disease allele is the chosen candidate gene, but because
Small sample size the allele is in linkage disequilibrium with a nearby gene.
Subgroup analysis and multiple comparisons Thus, the association is real, only the gene of interest is
Poorly matched control group wrong.
Failure to attempt study replication
Family-Based Association Studies
Failure to detect linkage disequilibrium with adjacent loci
To control for genetic mismatching in case-control stud-
Adapted from ref. 11. ies, Spielman and Ewens have suggested further assessment
of positive associations with a transmission disequilibrium
test (TDT) (17). This requires the genotyping of both par-
ents for, preferably, heterozygous alleles as well as having
Association Studies affected and unaffected offspring. The parent who is heter-
Association studies are case control studies that test ozygous for this allele should more often transmit the dis-
whether a particular allele occurs at a higher frequency ease allele to the affected offspring than the other allele(s)
among affected other than unaffected individuals. They (17). Chi square testing can document increased transmis-
involve correlation within a population, and not the inher- sion of putative disease markers. Family-based tests, such
itance of alleles within a family (10), and therefore family as this one, provide for an appropriate “internal control”
members are not required to participate. Association stud- with the use of unaffected siblings, whose environment and
ies are most commonly performed with alleles of a gene overall genetic background (i.e., same ethnic and family
thought to have biological significance to the etiology of group) are as closely similar to the proband as possible (18).
the complex trait or where familial-based linkage studies
Linkage Analysis
have shown linkage disequilibrium. This does not preclude
their use in studying random alleles of seemingly unrelated Finally, there are modes of linkage analysis, which use
genes. DNA sequence polymorphisms (normal variants) that are
Association studies have been criticized by a number of near or within a gene of interest to track within a family the
researchers for their tendency to highlight positive results inheritance of a disease-causing mutation in that gene.
leading to a publication bias, and the tendency to over- These have classically been considered the most powerful
interpret these results as identifying etiologic genes (10, means of identifying disease genes, and have the most com-
11). This has certainly been the case with PCOS where in- plex forms of analyses. Parametric methods assume a spe-
numerable negative and positive associations have been cific model of inheritance (i.e., autosomal recessive, etc.),
reported. Association studies have provided a number of whereas nonparametric methods do not. The nonmodel
potential loci with genetic variants that may create or add to based Affected Relative Pair analyses (sib pairs, cousin pairs,
a PCOS phenotype (many of these borrowed from similar grandparent/child, etc.), have the advantage of not specify-
studies in families with type 2 diabetes), including, to men- ing a genetic model and of eliminating the problem of mis-
tion a few such genes, Calpain 10 (12), IRS-1 and IRS-2 classifying young or “nonpenetrant” relatives, who are clin-
(13), and SHBG (14). Positive associations between a candi- ically unaffected but may carry the trait allele and have been
date allele and a complex disease are subject to many poten- used for PCOS studies. Linkage with PCOS was observed
tial errors (Table 2). One of the most common is that smaller over a broad region of chromosome 19p13.2 in a recent
sample sizes are more likely to yield positive relationships large study of 367 PCOS families. The strongest evi-dence
that do not hold up when the sample size increases, as we for association was observed with D19S884 (chi square =
have seen in PCOS with a CYP17 (which encodes a key 11.85; nominal p < 0.0006; permutation p = 0.034) (Fig. 1)
enzyme involved in androgen biosynthesis) polymorphism (18). The present analysis suggests that a PCOS suscepti-
(15). These were not replicated even by the same group in bility locus maps very close to D19S884.
independent data sets (16).
Population stratification between cases and controls is Pitfalls of PCOS Genetic Studies: Examples
another problem. Failure to match cases with controls on Some of the pitfalls with specific genes will be illustrated
the basis of racial background can lead to spurious results by the following examples of two candidate genes follista-
based on racial differences in allele frequencies unrelated tin and side chain cleavage that have been implicated in the
to disease status. Results should be adjusted for multiple etiology of PCOS.
comparisons, as frequently both multiple genes and multi-
ple alleles of these genes (at different sites) are tested. Fur- Follistatin
thermore, a priori phenotypes should be used to avoid post Follistatin, an activin binding protein, is widely expressed,
hoc phenotype drift, which can occur if the pirmary analy- as is activin and could play an important role in PCOS.
ses do not reach significance with the initial proposed phe- Activin stimulates pituitary FSH secretion and acts directly
Vol. 30, No. 1 Role of Genes and Environment in the Etiology of PCOS / Diamanti-Kandarakis, Kandarakis, and Legro 23

Fig. 1. Map of D19S884 region of chromosome 19p13.2. Locations of the short tandem repeat polymorphisms used in the analysis of
367 PCOS families, including those for (A) the 6.6 MB region of strongest evidence for linkage with PCOS in the INSR region and (B)
the narrow region around D19S884. D19S884 is indicated by an arrow. Adapted from ref. 18.

Fig. 2. Influence of body weight in women with PCOS (proband) and their sisters who have varying phenotypes, those with fully penetrant
PCOS, those with regular periods and hyperandrogenemia alone, and those who are completely unaffected with regular menses and
normal serum androgens. Mean BMI steps down as the phenotype improves. Adapted from ref. 6.

on the ovary to promote follicular maturation. Neutraliza- association study conducted on 97 PCOS women and
tion of activin would be expected to cause reductions in FSH matched controls revealed significant association of a pen-
levels and arrested follicular maturation as occurs in trans- tanucleotide repeat (tttta)n polymorphism at position −528
genic mice overexpressing follistatin. In addition, activin from the ATG initiation codon in the 5'-region of the CYP11A
inhibits thecal androgen biosynthesis and its removal by gene, which encodes the cholesterol side-chain cleavage
binding to follistatin might lead to unrestrained theca andro- enzyme and total serum testosterone levels (22). Although
gen synthesis driven by LH. The follistatin locus emerged no regulatory role has been assigned to this polymorphism
as a promising candidate from the study of 39 affected sib- in terms of CYP11A gene transcription, the investigators
pairs (19), although the strength of the link faded in a larger suggested that allelic variants of the CYP11A gene have a
follow-up series that included detailed sequence analysis role in the hyperandrogenemia of PCOS. This conclusion
of the follistatin gene (Fig. 2) (20). The follistatin gene has was supported by another study that found an association
also not been found to be linked to PCOS in studies by other between at least four repeats and PCOS, and within the
investigators. PCOS group an association of this allele group with a lower
testosterone level (23). In a study with a larger sample size,
CYP11A Urbanek et al. found modestly increased sharing of alleles
The rate-limiting step in sex steroid biosynthesis is deter- at the CYP11A locus among affected sib-pairs (19), but the
mined by the activity of side-chain cleavage encoded by sharing was not statistically significant. Most recently, the
P450scc encoded by the CYP11A gene. Gharani et al. exam- original group in a larger replication study, using both case-
ined 20 families with multiple affected women and found control and family-based association methods, report no
evidence for weak linkage to the CYP11A locus (21). An association with the CYP11A gene (24).
24 Role of Genes and Environment in the Etiology of PCOS / Diamanti-Kandarakis, Kandarakis, and Legro Endocrine

The Environment and PCOS higher saturated fat content in diet of U.S. women. There-
fore, it was hypothesized that diet alone does not explain
An overemphasis on genes is a simplistic explanation of
differences in body mass, because their food differed only
the PCOS epidemic. The rise in PCOS in populations where
in the quality of consumed fats and not in quantity. From
the gene pool has been relatively constant confirms that
these data it was concluded that genetic and lifestyle factors
environmental factors are assuming an ever more impor-
contribute to body weight differences, and there may be
tant role. The development of obesity is linked to the devel-
substantial geographic differences in the activities of daily
opment of PCOS in susceptible individuals. The modern
life including exercise- and nonexercise-related activities.
living environment in developed countries is characterized
Randomized trials of varying hypocaloric diets in women
by low daily energy expenditure and an abundant and inex-
with PCOS have resulted in appropriate short-term weight
pensive food supply, making positive energy balance com-
loss after 4 wk (with a 1000 kcal deficit/d designed to lose
mon, but we propose this too is simplistic and it is likely
1 kg/wk, mean weight loss at 4 wk was 4 kg) (31) and after
that other factors in the environment perhaps unrecognized
12 wk (on an approx 1400 kcal/d diet designed to lose about
or little recognized may also play a role, for instance expo-
0.45 kg/wk, subjects lost approx 8 kg) (32). There was also
sure to infectious agents, or environmental toxins.
an improvement in body composition, insulin levels, andro-
Obesity and Diet gen levels, and menstrual frequency with dietary-induced
weight loss (31,32). Neither of these studies found a bene-
Obesity is critical in the development of the PCOS phe-
fit to a particular dietary composition (high protein versus
notype and weight exacerbates many (most notably insulin
low protein) although their sample size was small and both
resistance), if not all, of the PCOS symptoms. Family stud-
had high dropout rates (31,32).
ies provide an interesting perspective on the role of weight
in the PCOS phenotype. There are interesting findings among Environmental Toxins
PCOS sisters revealing the role of obesity. Body weight dif- Organochlorine pesticides were extensively used through-
fers between the phenotypes in affected PCOS sisters (6). out the world until they were banned in the 1970s (33).
Sisters with irregular cycles and hyperandrogenemia are Owing to their resistance to enzymatic degradation and to
heavier than sisters with regular cycles and hyperandrogen- their concentration in fat, these products are particularly
emia, while unaffected sisters have lower body weights than persistent in living organisms and accumulate in their fat
the affected ones (Fig. 2) (25). Retrospective studies suggest over years (33). Since then, several organochlorine com-
that a different intrauterine environment is linked to dif- pounds have been characterized as estrogenic and as hor-
ferent PCOS phenotypes like PCOS women with increased mone-disrupters (34). Furthermore, organochlorines were
birth weight born from overweight mothers (26), or girls found in greater concentrations in the adipose tissue of
with premature pubarche and lower birth weights who de- women affected by breast cancer than in healthy controls
velop a more severe PCOS phenotype at menarche (27). (35). However, their role in the development of a PCOS
For obese women with PCOS, dietary therapy is often phenotype are unknown.
the first line recommended therapy and marked improve-
ment including restoration of ovulation has been noted with Medications
minimal weight loss approaching 5% of the baseline weight A better example of an “environmental” substance impli-
(28,29). The mechanisms for this are complex for weight cated in the development of a PCOS phenotype is valproic
loss is associated with simultaneous improvements in cir- acid, which is a short-chained fatty acid that is widely used
culating androgens, gonadotropins, and insulin levels. It is to treat epilepsy and bipolar disorders as well as migraines
difficult to tease out the benefits of exercise per se on improv- and generalized mood disorders. There are studies to sug-
ing PCOS, but intervention programs that have combined gest that women with these disorders and treated with val-
both dietary and exercise interventions have had excellent proic acid may develop stigmata of PCOS, including poly-
outcomes in terms of improved ovulation and/or pregnancy cystic ovaries, hyperandrogenism, obesity, and anovula-
rates. tion (36), and that these stigmata may be reversible with
Poor dietary choices including relying on energy dense discontinuation of the medication (37). Although this is a
foods instead of unprocessed grains, fruits, and vegetables, highly contentious area and there may be clear ethnic dif-
and larger portion sizes have been implicated as contributors ferences in susceptibility, recent studies suggest that weight
to the obesity epidemic. Interestingly not only the quantity gain on medication is essential to developing the full PCOS
of food but the quality and the type of nutrition as well may phenotype.
alter PCOS phenotype, possibly interacting with different
genetic patterns. One study demonstrated that women with Infectious Causes
PCOS from Sicily are less obese than women from Pennsyl- The possibility that one or more viruses could have con-
vania (30). However, total calorie intake and dietary con- tributed to the epidemic of PCOS has received little consid-
stituents were similar between countries, except there was eration. Most people view the likelihood that an infection
Vol. 30, No. 1 Role of Genes and Environment in the Etiology of PCOS / Diamanti-Kandarakis, Kandarakis, and Legro 25

causes PCOS or obesity with the same skepticism that the 6. Legro, R. S., Driscoll, D., Strauss, J. F., et al. (1998). Proc.
scientific community once viewed the causes of peptic ulcer Natl. Acad. Sci. USA 95, 14956–14960.
7. Jahanfar, S., Eden, J. A., Warren, P., et al. (1995). Fertil. Steril.
disease (interestingly also as a fault of the individual—too 63, 478–486.
much stress in their life!). There is a significant literature 8. Jahanfar, S., Maleki, H., Mosavi, A. R., and Jahanfar, M. (2004).
that demonstrates that viruses may cause obesity in animals Gynecol. Endocrinol. 18, 327–334.
(38). A human virus, adenovirus-36 (Ad-36), has produced 9. Vink, J. M., Sadrzadeh, S., Lambalk, C. B., and Boomsma,
D. I. (2006). J. Clin. Endocrinol. Metab. 91, 2100–2104.
obesity in chickens, mice, and monkeys. About 70–100% 10. Lander, E. S. and Schork, N. J. (1994). Science 265, 2037–2048
of animals inoculated with Ad-36 became obese (100% of (erratum, 266, 353).
monkeys) (39). A study in humans demonstrated that both 11. Cardon, L. R. and Bell, J. I. (2001). Nat. Rev. Genet. 2, 91–99.
obese and nonobese subjects positive for Ad-36 antibodies 12. Ehrmann, D. A., Schwarz, P. E., Hara, M., et al. (2002). J. Clin.
Endocrinol. Metab. 87, 1669–1673.
had significantly higher BMIs compared to AB− subjects, 13. El Mkadem, S. A., Lautier, C., Macari, F., et al. (2001). Diabetes
thus demonstrating a syndrome similar to animals infected 50, 2164–2168.
with Ad-36 in the laboratory (40). These data show an asso- 14. Cousin, P., Calemard-Michel, L., Lejeune, H., et al. (2004). J.
ciation, but causality for human obesity has not been proven. Clin. Endocrinol. Metab. 89, 917–924.
15. Carey, A. H., Chan, K. L., Short, F., et al. (1993). Clin. Endo-
There are no comparable studies of infectious agents in crinol. 38, 653–658.
PCOS, and these data are cited for edification. 16. Gharani, N., Waterworth, D. M., Williamson, R., and Franks,
S. (1996). J. Clin. Endocrinol. Metab. 81, 4174.
17. Spielman, R. S. and Ewens, W. J. (1996). Am. J. Hum. Genet.
Summary 59, 983–989.
18. Urbanek, M., Woodroffe, A., Ewens, K. G., et al. (2005). J. Clin.
There is evidence that both genes and the environment Endocrinol. Metab. 90, 6623–6629.
contribute to PCOS. Recognizing environmental modifi- 19. Urbanek, M., Legro, R. S., Driscoll, D. A., et al. (1999). Proc.
ers of PCOS in addition to poor diet and physical inactivity Natl. Acad. Sci. USA 96, 8573–8578.
should be a priority for future researchers. There are likely 20. Urbanek, M., Wu, X., Vickery, K. R., et al. (2000). J. Clin.
Endocrinol. Metab. 85, 4455–4461.
man-made common substances that affect predisposition 21. Gharani, N., Waterworth, D. M., Batty, S., et al. (1997). Hum.
to PCOS, and one may also even widely speculate that infec- Mol. Genet. 6, 397–402.
tious agents contribute to the risk. 22. Diamanti-Kandarakis, E., Bartzis, M. I., Bergiele, A. T., et al.
Thus, there have been many tantalizing starts, but no (2000). Fertil. Steril. 73, 735–741.
23. San Millan, J. L., Sancho, J., Calvo, R. M., and Escobar-
finishes identifying genes associated with PCOS. This may Morreale, H. F. (2001). Fertil. Steril. 75, 797–802.
be partially due to the fact that the majority of genes studied 24. Gaasenbeek, M., Powell, B. L., Sovio, U., et al. (2004). J. Clin.
have been candidate genes based on current understanding Endocrinol. Metab. 89, 2408–2413.
of the pathophysiology of PCOS. Given that the pathophys- 25. Legro, R. S., Bentley-Lewis, R., Driscoll, D., et al. (2002). J.
Clin. Endocrinol. Metab. 87, 2128–2133.
iology is poorly understood, it is not surprising that these 26. Cresswell, J. L., Barker, D. J., Osmond, C., et al. (1997). Lancet
candidates have yielded little. No one to date has performed 350, 1131–1135.
a genome-wide scan on an adequate sample size to discover 27. Ibanez, L., Potau, N., Francois, I., and de Zegher, F. (1998).
new genes and regions unrelated to the expert’s best guesses, J. Clin. Endocrinol. Metab. 83, 3558–3562.
28. Kim, L. H., Taylor, A. E., and Barbieri, R. L. (2000). Fertil.
although these studies are now in progress. The PCOS gene Steril. 73, 1097–1098.
or genes are awaiting clearer unfolding 29. Kiddy, D. S., Hamilton-Fairley, D., Bush, A., et al. (1992).
Clin. Endocrinol. 36, 105–111.
30. Carmina, E., Legro, R. S., Stamets, K., et al. (2003). Hum.
Acknowledgments Reprod. 18, 2289–2293.
31. Stamets, K., Taylor, D. S., Kunselman, A., et al. (2004). Fertil.
This work was supported by the National Cooperative Steril. 81, 630–637.
Program in Infertility Research (NCPIR) U54 HD34449, a 32. Moran, L. J., Noakes, M., Clifton, P. M., et al. (2003). J. Clin.
GCRC grant MO1 RR 10732 to Pennsylvania State Univer- Endocrinol. Metab. 88, 812–819.
sity and K24 HD01476 (to RSL). 33. Pelletier, C., Imbeault, P., and Tremblay, A. (2003). Obes. Rev.
4, 17–24.
34. Steinmetz, R., Young, P. C., Caperell-Grant, A., et al. (1996).
References Cancer Res. 56, 5403–5409.
35. Dewailly, E., Dodin, S., Verreault, R., et al. (1994). J. Natl.
1. Zawadski, J. K. and Dunaif, A. (1992). In: Polycystic ovary Cancer Inst. 86, 232–234.
syndrome. Blackwell Scientific: Boston. 36. Isojarvi, J. I., Laatikainen, T. J., Pakarinen, A. J., et al. (1993).
2. Balen, A. H., Laven, J. S., Tan, S. L., and Dewailly, D. (2003). N. Engl. J. Med. 329, 1383–1388.
Hum. Reprod. Update 9, 505–514. 37. Isojarvi, J. I., Rattya, J., Myllyla, V. V., et al. (1998). Ann.
3. Jonard, S., Robert, Y., and Dewailly, D. (2005). Hum. Reprod. Neurol. 43, 446–451.
20, 2893–2898. 38. Dhurandhar, N. V. (2004). Drug News Perspect. 17, 307–313.
4. Rotterdam ESHRE/ASRM sponsored PCOS consensus work- 39. Dhurandhar, N. V., Whigham, L. D., Abbott, D. H., et al.
shop group. (2004). Hum. Reprod. 19, 41–47. (2002). J. Nutr. 132, 3155–3160.
5. Abbott, D. H., Dumesic, D. A., and Franks, S. (2002). J. Endo- 40. Dhurandhar, N. V., Kulkarni, P. R., Ajinkya, S. M., et al. (1997).
crinol. 174, 1–5. Obes. Res. 5, 464–469.
26 Role of Genes and Environment in the Etiology of PCOS / Diamanti-Kandarakis, Kandarakis, and Legro Endocrine

41. Cooper, H. E., Spellacy, W. N., Prem, K. A., and Cohen, W. D. 47. Govind, A., Obhrai, M. S., and Clayton, R. N. (1999). J. Clin.
(1968). Am. J. Obstet. Gynecol. 100, 371–387. Endocrinol. Metab. 84, 38–43.
42. Wilroy, R. S. Jr., Givens, J. R., Wiser, W. L., et al. (1975). Birth 48. Kahsar-Miller, M. D., Nixon, C., Boots, L. R., et al. (2001).
Defects Orig. Artic. Ser. 11, 81–85. Fertil. Steril. 75, 53–58.
43. Ferriman, D. and Purdie, A. W. (1979). Clin. Endocrinol. 11, 49. Legro, R. S., Kunselman, A. R., Demers, L., et al. (2002). J. Clin.
291–300. Endocrinol. Metab. 87, 2134–2138.
44. Hague, W. M., Adams, J., Reeders, S. T., et al. (1988). Clin. 50. Sir-Petermann, T., Angel, B., Maliqueo, M., et al. (2002). Dia-
Endocrinol. 29, 593–605. betologia 45, 959–964.
45. Lunde, O., Magnus, P., Sandvik, L., and Hoglo, S. (1989). 51. Yildiz, B. O., Yarali, H., Oguz, H., and Bayraktar, M. (2003).
Gynecol. Obstet. Invest. 28, 23–30. J. Clin. Endocrinol. Metab. 88, 2031–2036.
46. Norman, R. J., Masters, S., and Hague, W. (1996). Fertil. Steril. 52. Kaushal, R., Parchure, N., Bano, G., et al. (2004). Clin. Endo-
66, 942–947. crinol. (Oxf.) 60, 322–328.