Professional Documents
Culture Documents
Insulin resistance is prevalent in women with polycystic ovary syndrome (PCOS), Stéphanie B Mayer*,1, William
and plays a critical pathophysiologic role in both the metabolic and reproductive S Evans2 & John E Nestler1
complications of PCOS. This review focuses on the contribution of insulin resistance
1
Division of Endocrinology & Metabolism,
Virginia Commonwealth University,
to anovulation in PCOS and to the high risk for Type 2 diabetes, metabolic syndrome
Richmond, VA, USA
and early cardiovasular disease. Key points for clinicians emphasized by this review 2
Division of Endocrinology & Metabolism,
are the following: PCOS is a clinical diagnosis and alternative diagnoses must be University of Virginia, Charlottesville,
excluded; PCOS carries an inherent risk of insulin resistance and, hence, metabolic VA, USA
consequences for which women with PCOS should be screened regardless of BMI or *Author for correspondence:
Tel.: 804 828 9696
degree of obesity; and PCOS is associated with infertility and this should be discussed
Fax: 804 828 8389
early on in care of women diagnosed with PCOS, recognizing that there are several smayer@ mcvh-vcu.edu
possible strategies to address infertility in women with PCOS, each with its own risks
and benefits.
10.2217/WHE.14.73 © 2015 Future Medicine Ltd Womens Health (2015) 11(2), 137–149 ISSN 1745-5057 137
Review Mayer, Evans & Nestler
Insulin binds to
insulin receptors and
stimulates androgen
production
Liver
Figure 1. Insulin resistance in polycystic ovary syndrome, effects on ovarian androgen production.
FSH: Follicle-stimulating hormone; LH: Luteinizing hormone; SHBG: Sex hormone binding globulin.
unbound testosterone (also called free testosterone) women without PCOS found no effect of diazoxide
that acts on target tissues. Insulin is thought to increase on circulating serum testosterone concentrations [25] ,
circulating levels of free testosterone by two specific suggesting that women with PCOS have an inherent
mechanisms: first, by stimulating ovarian biosynthesis susceptibility to the hyperandrogenic effects of insulin.
and secretion of testosterone, and, second, by directly With regard to insulin’s suppression of hepatic pro-
suppressing hepatic production of SHBG. duction of SHBG, in vitro insulin has been shown to
Regarding insulin’s stimulation of ovarian testos- inhibit SHBG production by cultured HepG2 cells [21] ,
terone production, in vitro studies have demonstrated and, in vivo, administration of diazoxide to pharmaco-
greater stimulation of testosterone production by logically castrated women with PCOS resulted in an
human theca cells obtained from women with PCOS increase in levels of SHBG despite no change in sex
compared with theca cells obtained from normal steroids [22] .
women, and insulin has been shown to inhibit hepatic Finally, it is noteworthy that studies of women with
production of SHBG both in vitro [21] and in vivo [22] . both Type 2 diabetes (DM2) [26,27] and Type 1 diabe-
In vivo studies in which hyperinsulinemia was induced tes, including lean patients, [28–31] , reported a higher
resulted in increases in circulating androgens in women prevalence of PCOS in these women than in the gen-
with PCOS but not in normal women, both indepen- eral population. This suggests that insulin resistance,
dently of gonadotropin release [23] , and independently or more specifically hyperinsulinemia (whether endog-
of BMI [24] . Conversely, administration of diazoxide enous or exogenous), is responsible for the ovarian
to inhibit insulin release in obese women with PCOS hyperandrogenism in these diabetic patients.
resulted in a decrease in serum total and free testoster-
one levels with no change in gonadotropin level, sug- Prevalence of insulin resistance in women with
gesting a direct effect of insulin to stimulate ovarian PCOS
testosterone production [13] . A similarly designed study Insulin resistance is observed independently of obesity
of administration of diazoxide to normal non-obese in women with PCOS, and it is critical for physicians
treating women with PCOS to bear in mind that PCOS have been found to be at substantially increased
both obese and lean women with PCOS demonstrate risk of developing DM2 [35] .
hyperinsulinemic insulin resistance, independently of
their adiposity [32,33] . In 1989, Dunaif et al. [34] , using Detection of glucose intolerance in women
the hyperinsulinemic-euglycemic clamp technique to with PCOS
measure whole-body insulin sensitivity, demonstrated Neither insulin resistance nor biochemical evidence
that both obese and lean women with PCOS were of glucose intolerance is necessary for the diagnosis of
significantly insulin-resistant compared with their PCOS. Moreover, many women with PCOS will have
weight-matched normal controls, with lean women demonstrable insulin resistance with no clinical evi-
with PCOS being as insulin-resistant as obese nor- dence of glucose intolerance. Nonetheless, given the evi-
mal women, and obese normoglycemic women with dence above of a high prevalence of glucose intolerance
PCOS being as insulin-resistant as DM2 women. In and DM2 in PCOS, it behooves clinicians to screen for
this study, PCOS status and obesity synergistically these disorders at the time of PCOS diagnosis. Expert
reduced insulin sensitivity – that is, insulin sensitivity opinion would recommend screening all women with
was greatest in lean normal > lean PCOS = obese nor- PCOS for glucose intolerance at the time of diagnosis
mal > and worst of all in obese PCOS women, so that using a 2-h 75-g oral glucose tolerance test (OGTT)
having both PCOS and obesity worsened insulin resis- (regardless of BMI) and repeat screening at 2–5 year
tance in a complementary fashion [34] . A take home intervals [41,42] . An OGTT is the recommended screen-
point is that insulin resistance is a frequent feature of ing test for glucose intolerance, since, as noted below,
PCOS regardless of BMI. studies indicate that in women with PCOS the fasting
Methodologies to test accurately for insulin resistance serum glucose and HbA1c levels may be in the nor-
remain technically challenging and are not used in clini- mal range even when a woman with PCOS has IGT or
cal practice. However, there are several clinical param- DM2 when assessed by an OGTT [43–45] .
eters that reflect underlying insulin resistance, including In women with PCOS, the OGTT remains the gold
obesity, increased waist-to-hip ratio, presence of acan- standard test for screening for glucose intolerance.
thosis nigricans, low HDL-c level, elevated triglyceride Legro et al. have demonstrated in both adult and ado-
level, and reduced serum SHBG. One clinical conse- lescent patients with PCOS that a fasting plasma glu-
quence of insulin resistance is impaired glucose toler- cose fails to identify a substantial portion of women
ance (IGT) and, more severely, overt DM2. In two pro- who have IGT or DM2 by OGTT testing [46] . Simi-
spective trials of women with PCOS conducted in the larly, recent evaluations of HbA1c for diagnosis of glu-
USA [35,36] , there was a 31–35% prevalence of IGT and a cose intolerance in women with PCOS have proved
7.5–10.0% prevalence of DM2 in young to middle-aged disappointing. A study of 252 Turkish women with
women with PCOS. In a prospective, controlled trial PCOS and 117 control women without PCOS [43]
over a 2–3-year follow-up, among women with PCOS compared HbA1c levels with OGTT results and
there was at baseline a 37% prevalence of IGT and 10% found that HgbA1c provided a less robust measure
prevalence of DM2, with a subsequent 16% conversion/ of testing with a 52.4% sensitivity, 74.4% specificity,
year from normal glucose tolerance (NGT) to IGT and 67.1% positive and 60.9% negative predictive values.
a 2% conversion/year from IGT to DM2. Consistent Again, women who had IGT or DM2 by OGTT were
with these observations, in the Nurses’ Health Study II mis-identified as normal by HbA1c testing. Similarly,
(NHSII) of 101,073 women followed for 8 years, the a retrospective study of 208 premenopausal women
conversion rate to DM2 was approximately twofold with PCOS [44] reported a poor sensitivity of 35% but
higher in oligomenorrheic women, independent of excellent specificity of 99% using HbA1c ≥ 6.5% for
weight, than those with normal menses [37] . the diagnosis of diabetes in women with PCOS, com-
Collectively, studies indicate that in the USA there pared with the diagnosis established by OGTT. The
is a 10-fold increased prevalence of DM2 in young authors concluded that the HbA1c is less reliable than
women with PCOS, and that 30–50% of obese women the OGTT to uncover glucose intolerance or diag-
with PCOS develop IGT or DM2 by the age of 30 years nose diabetes in PCOS. In another study of 111 adult
old (earlier than their peers without PCOS). Of note, women with PCOS who were prospectively screened
abnormalities of carbohydrate metabolism, resulting in for DM2, IGT (pre-DM) and insulin resistance using
IGT or DM2, tend to cluster in first degree relatives various diagnostic methods and OGTT as the gold
of women with PCOS [38,39] . Similar to patients with standard, screening with fasting plasma glucose failed
DM2 without PCOS, in patients with PCOS a positive to identify 41% of IGT and 20% of DM2 subjects.
first degree relative with DM2 correlates significantly Further, OGTT and HbA1c had only fair agreement
with the risk of having DM2 [40] . In short, women with (κ = 0.29) [45] .
Prevention of glucose intolerance in women levels of adiponectin and increased C-reactive protein
with PCOS levels. Anatomic and functional markers for CVD
Treatment options for reducing the rate of conversion are also increased in women with PCOS, including
to diabetes in women with PCOS focus on lifestyle increased carotid intima-medial thickness, coronary
modification, including the use of low glycemic index artery calcium (CAC) scores and abnormal flow medi-
and low carbohydrate diets, and use of insulin sensitiz- ated dilation, all suggesting an increased risk in women
ing agents, specifically metformin. Use of thiazolidin- with PCOS.
ediones had previously shown promise in this regard, In terms of lipids, in premenopausal women with
but given recent concerns for risk for bladder cancer PCOS, higher rates of dyslipidemia were found than
with prolonged pioglitazone use, caution with rosigli- in controls [49] . A case–control study of 195 non-His-
tazone and cardiac concerns, removal of troglitazone in panic white women with PCOS [50] found elevations
the face of idiosyncratic hepatitis, and concern to bone in LDL-c levels in women with PCOS, independent
health and increased fracture risk in postmenopausal of obesity [50] .
women, this class of agents has fallen out of favor. A prospective study followed 125 women with PCOS
A recent systematic review [47] of six articles from five and 142 controls over the course of 3 years with B-mode
studies, with 137 women included, found subtle differ- ultrasonography of the carotid arteries to measure carotid
ences between various dietary interventions in PCOS. intima-media wall thickness and plaque [51] . In those
Greater weight loss was seen in monounsaturated fat- women aged 45 years or older, there was significantly
enriched diets, and low-glycemic index diets seemed more carotid mean intimal medial thickness in PCOS
to improve menstrual irregularity. Greater reductions cases than in the control women, and this remained
in insulin resistance were seen with both low-carbo- significant after adjusting for age and BMI [51] . In a
hydrate and low-glycemic index diets. The authors study of 36 women with PCOS and 71 age- and BMI-
found an improved quality of life with the use of a low- matched control women aged 30–45 years (excluding
glycemic index diet and improved depression and self- those with diabetes and known coronary heart disease),
esteem with a high-protein diet. Conversely, an unde- evaluated by electron beam computed tomography
sirable increased in free androgen index resulted from to noninvasively measure CAC, 39% of women with
high-carbohydrate diets, perhaps due to comparatively PCOS compared with 21% of controls had CAC, and
enhanced insulin release suppressing SHBG. when compared with community dwelling women the
While several studies have demonstrated a reduc- prevalence was more startling still (9.9%; odds ratio,
tion in progression to DM2 in non-PCOS individuals 5.9) [52] . An association was found between extent of
treated with the drug metformin, the only such study coronary artery disease (CAD) on coronary angiogra-
specifically in PCOS is a retrospective chart review of 50 phy and presence of polycystic ovaries on ultrasound
women with PCOS, all of whom were started on met- in 143 women ≤60 years old referred for assessment of
formin therapy at diagnosis, had OGTT testing at base- chest pain or valvular disease in New Zealand in 1997,
line, did not have diabetes at baseline and had a mini- so that women with polycystic ovaries (42%) had more
mum of 1-year of follow-up with repeat OGTT testing. extensive coronary artery disease than women with nor-
The study found that, at baseline, 39 women (78%) had mal ovaries (number of segments with >50% stenosis,
NGT and 11 women (22%) had IGT. The women with 1.7 compared with 0.82). On logistic regression analysis
NGT were followed for an average of 43.3 months and of this data, the extent of coronary artery disease and
the women with IGT for 28.9 months. During these family history of heart disease predicted the presence of
periods, the annual conversion rate from NGT or IGT polycystic ovaries [53] .
to DM2 was 0% (none had developed diabetes), and Most notably, in a study, 30 young (mean age
1.4% from NGT to IGT [48] . This rate of conversion 22 years old) and lean (mean BMI 22 kg/m2) women
from NGT to IGT was ∼10-fold lower than the con- with PCOS, without any prior known cardiovascular
version to IGT among women with PCOS with NGT disease (CVD), were compared with 30 healthy age-
not treated with metformin reported in the prospective and BMI-matched control women with flow-mediated
studies mentioned above (16–19% NGT to IGT) [35,36] . dilation of the brachial artery and by Doppler ultra-
sound of the carotid arteries for intima-media thick-
Cardiovascular disease in PCOS ness [54] . Compared with the normal control women,
Women with PCOS are often found to have an accumu- these young and lean women with PCOS were found
lation of risk factors for cardiovascular disease (CVD) to have impaired flow-mediated dilation, accompanied
including obesity, hypertension, low levels of HDL- by an increased serum endothelin-1 level, and increased
c, elevated triglyceride levels, increased plasminogen intima-media thickness of the carotid arteries. Hence,
activator inhibitor (PAI-1), increased endothelin-1, low at an early age, these lean women with PCOS were
already exhibiting anatomic and functional cardiac disease was 1.4 for PCOS, this did not meet statistical
abnormalities placing them at risk for premature CVD. significance [62] .
Given the high prevalence of cardiovascular risk Conversely and consistent with the results of the
factors, it would not be surprising that women with NHS and WISE studies, in a 20-year retrospective
PCOS would also have a higher prevalence of the study of 2301 women with PCOS totalling >12,000
metabolic syndrome (MBS). As background, the MBS person-years, Mani et al. found that the prevalence of
in women is defined as having three out of five of the MI in the age group 45–54 was 1.9%, age 55–64 was
following criteria: waist circumference >35 inches 6.0% and >65 years 27.3%. Prevalence of angina age
(88 cm); triglyceride ≥150 mg/dl; HDL-c <50 mg/dl; 45–54 was 2.6%, age 55–64 was 6.0% and > 65 years
blood pressure ≥130/85 mmHg; fasting plasma glu- was 27.3%. Age-group-specific odds ratios for the
cose ≥110 mg/dl. While MBS confers a two- to three- prevalence of MI and angina compared with the local
fold higher risk for a cardiovascular event, the risk in female population ranged between 2.6 and 12.9. The
women with MBS may be greater than that of men highest odds ratio was for MI in the group of women
with MBS. Seven studies within a larger meta-analysis with PCOS >65 years old [63] .
found that the risk of CV events and death is 30% When considered collectively, the majority of data
higher in women than men meeting NCEP 2001 support the idea that women with PCOS, indepen-
ATP III criteria for metabolic syndrome [55] . dently of adiposity, are at increased risk for atheroscle-
In a retrospective chart review of 106 women rosis, and should be screened and targeted for primary
with PCOS, the prevalence of metabolic syndrome prevention of CVD.
was 43%, nearly two- to threefold higher than that This is in line with guideline recommendations of
reported for age-matched women in the general popu- the Androgen Excess and PCOS society [64] .
lation from NHANES III data, even when matched
by BMI [56] . In this study, >90% of PCOS women had Reproductive consequences in PCOS
at least one of the five cardiovascular risk factors that Prevalence
comprise the MBS, and the most prevalent risk factor PCOS affects 5–10% of women of childbearing age
among the women was a low HDL-c. The higher prev- in the USA (3–5 million women), and is currently the
alence of the MBS in women with PCOS is not con- leading cause of anovulatory infertility [65,66] . Anovu-
fined to the USA, but has been confirmed in multiple lation presents clinically as irregular menstrual cycles
studies conducted worldwide [57–59] . (either amenorrhea or oligomenorrhea) and infertility.
Studies of cardiac outcomes in PCOS are limited Seventy-five percent of women with PCOS have infer-
in number, but overall support an increased risk of tility due to anovulation. In addition, early (first tri-
cardiovascular events. In the Nurses Health Study mester) miscarriages are threefold the rates of women
(NHS), which followed 82,439 women for 14 years, without PCOS [67] .
in women with very irregular periods the risk ratio for Weight loss of 5–10% can restore ovarian function
coronary heart disease was 1.5 and the risk ratio for and ovulation in approximately 50% of patients with
fatal myocardial infarction (MI) was increased two- PCOS and is thus an important pillar of fertility care
fold [60] . Further epidemiological evidence of increased in women with this disorder. It seems likely that this
cardiovascular events in postmenopausal women with phenomenon is mediated via improved insulin sensitiv-
a history of PCOS (defined by irregular menses and ity. However, sustained weight loss can be difficult to
elevated androgen measurements) was found in a sub- achieve which in turn results in the need for additional
set of the NIH’s National Heart, Lung, and Blood therapeutic strategies in many cases.
Institute (NHLBI) sponsored prospective Women’s
Ischemia Syndrome Evaluation (WISE) study. The First-line medical strategies
cumulative 5-year cardiovascular event rate was 21.1% The two most commonly used medical therapies to
for women with clinical features of PCOS (n = 104) treat infertility in women with PCOS are clomiphene
versus 11.3% for women without clinical features of citrate and metformin. Clomiphene citrate acts as an
PCOS (n = 286) [61] . anti-estrogen which displaces endogenous estrogen
Counter to these data is a retrospective review of from the hypothalamic estrogen receptor and thus
death certificates in the UK from women with PCOS removes negative feedback by endogenous estro-
not showing a statistically significant increase in MI gens. Clomiphene acts as an indirect stimulator of
compared with the general population. However, this FSH secretion. It restores ovulation in nearly 80% of
study was powered to detect a threefold increase in risk women and results in an approximate 50% conception
and only 786/1028 charts were analyzed, and while rate usually within three cycles of use. The primary
the standardized mortality ratio for ischemic heart adverse effect of clomiphene citrate is the increased rate
of multiple gestations. Metformin is thought to treat ing metformin with clomiphene citrate in women with
infertility by improving insulin sensitivity. In a trial PCOS and BMI <32 kg/m2 found no statistically sig-
of 16 obese women with PCOS treated with metfor- nificant difference for any outcome (ovulation, preg-
min for 6 months and evaluated by the euglycemic- nancy, live birth, miscarriage and multiple pregnancy
hyperinsulinemic clamp technique, glucose utilization rates) between these two treatment modalities [72] . The
and insulin action improved and was accompanied authors did caution that there were conflicting findings
by significant increases in the levels of sex hormone- and heterogeneity across the RCTs.
binding globulin and decreases in free testosterone A meta-analysis of 13 RCTs including 428 women
and androstenedione levels [68] . Among these women, with PCOS found a 46% rate of ovulation in those
44% resumed normal cycles and two cases of sponta- PCOS women assigned to metformin monotherapy
neous pregnancy occurred during treatment. Current versus 24% in those receiving placebo [73] . Another
expert theory proposes that reducing levels of circulat- such meta-analysis found that metformin was 50%
ing insulin will decrease intra-ovarian concentrations better than placebo for ovulation induction in infer-
of androgens and normalize gonadotropin secretory tile PCOS patients, but was not of confirmed ben-
dynamics. efit versus placebo for achievement of pregnancy [74] .
One study followed 61 obese women with PCOS, A subsequent meta-analysis in 2008 of 17 studies in
randomized to either metformin or placebo, for spon- 1639 women compared metformin versus placebo, and
taneous ovulation over 35 days, and then administer- metformin + clomiphene versus clomiphene alone [75] .
ing clomiphene for 5 days to those women who did In the 10 trials comparing metformin monotherapy
not ovulate (21 in metformin arm and 25 in placebo to placebo, metformin improved the odds of ovula-
arm), while continuing either metformin or placebo, tion. There was a 1.56-fold increase in clinical preg-
and then monitoring serum progesterone levels. The nancy rate, but this was not statistically significant [75] .
authors found that the women randomized to metfor- Metformin combined with clomiphene increased the
min receiving clomiphene had a significant reduction likelihood of ovulation (and pregnancy), compared
in mean AUC of insulin after oral glucose administra- with clomiphene therapy alone [75] . A Cochrane meta-
tion, whereas those on placebo did not, and 19 of 21 analysis of metformin monotherapy in PCOS found
women on metformin + clomiphene ovulated (90%) an improved ovulation rate in 16 RCTs with 1208
versus 2 of 25 on placebo + clomiphene ovulated subjects. There was also an improved rate for clinical
(8%) [69] . In another study of non-obese women with pregnancy with metformin use in eight RCTs in 707
PCOS with normal OGTT, the rate of ovulation after subjects. In combination with clomiphene, metformin
initiation of metformin therapy demonstrated a graded improved clinical pregnancy rates but this combina-
increase over a 6 month time course with peak ovula- tion did not lead to improved live-birth rates. Met-
tion rate at months 5 and 6 of therapy [70] . In a head formin was also associated with a significantly higher
to head randomized controlled trial (RCT) of clomi- incidence of gastrointestinal disturbances than placebo
phene citrate to metformin in non-obese anovulatory but no serious adverse effects were reported [76] .
women with PCOS [71] , 6 months of administration In 2007 the Pregnancy in Polycystic Ovary Syn-
of metformin was more effective than six cycles of clo- drome study (PPCOS) randomized 626 infertile
miphene citrate based on a non-statistically different women with PCOS to clomiphene citrate plus pla-
ovulation rate between the two treatment groups (62.9 cebo, extended-release metformin plus placebo, or a
vs 67.0%), but with a significantly higher pregnancy combination of metformin and clomiphene for up to
rate in the metformin group (15.1 vs 7.2%), a lower 6 months. Adding metformin to clomiphene citrate
rate of abortion with metformin therapy over 6 months increased the ovulation rate by 20%. In this study, the
(9.7 vs 37.5%), as well as a positive trend for live-birth live-birth rate was 22.5% in the clomiphene group,
rate favoring metformin therapy (83.9 vs 56.3%). The 7.2% in the metformin group and 26.8% in the com-
cumulative pregnancy rate was significantly higher bination-therapy group, but the difference in live-birth
in the metformin group than the clomiphene citrate rates between the clomiphene group and combination-
group (68.9 vs 34.0%). Ovulation rates were higher therapy group did not attain statistical significance.
initially with clomiphene citrate use in the first cycle Notably, multiparity was 6.0% in the clomiphene
but steadily rose with longer duration of metformin group, 0% in the metformin group and 3.1% in the
use over the course of 6 months to surpass rates with combination-therapy group [77] .
clomiphene citrate by month 6 (month 1: metformin Because of the lack of a statistically significant dif-
42% vs clomiphene 83%; in month 6: metformin ference in live-birth rates between clomiphene alone
86% versus clomiphene 48%) [71] . In 2012, a system- and clomiphene in combination with metformin in
atic review and meta-analysis of four RCTs compar- this study, the use of metformin for fertility induction
has somewhat fallen out of favor and does not figure difference in pregnancy complications between arms
in the Endocrine Society’s guidelines on PCOS [42] . in preeclampsia prevalence, GDM, or preterm deliv-
However, it should be noted that in the PPCOS study ery, or in a composite of those outcomes. Women in
patients were not allowed a run-in time of metformin the metformin group gained less weight during preg-
therapy which may well have affected results. Since nancy than those on placebo. There was no difference
its publication – and not included in the Endocrine in fetal weight [83] .
Society guideline – a 2012 study of 320 anovulatory A prospective pharmacokinetic study of metformin
women with PCOS randomized to metformin versus concentrations in breast milk samples in six women
placebo found a 45% increase in live-birth rate after found that the relative infant dose was <0.5% of the
metformin use compared with placebo. Metformin mother’s weight adjusted dose, and the authors con-
significantly improved both pregnancy and live-birth cluded that infant exposure to metformin through
rates over placebo (pregnancy rate: 53.6 vs 40.4%; live- breast milk is low [84] . Similarly, another study in seven
birth rate: 41.9 vs 28.8%) [78] . Perhaps this discrepancy breastfeeding women taking metformin, concentra-
is due to methodological differences in study protocols tions of metformin were measured by high performance
by which the PPCOS study allowed insufficient time liquid chromatography and found that the mean rela-
for sufficient improvement in insulin sensitivity with tive infant dose was 0.28% (0.16–0.4%). Metformin
metformin therapy. was found in undetectable to very low plasma concen-
There are data suggesting that metformin use in trations in the infants. No health concerns were noted
PCOS may reduce the rate of early pregnancy loss. in the infants evaluated [85] .
In a retrospective study of PCOS women treated for In follow-up questionnaires regarding breast size
infertility in one clinic, rates of early pregnancy loss increment and breastfeeding patterns, 1-year post-
were lower (8.8%; 6 of 68 pregnancies) in the group partum, from participants of a completed RCT [83]
treated with meformin, as compared with 41.9% (13 of in which 240 participants with PCOS were random-
31 pregnancies) in the control group [79] . Subsequently ized to treatment with metformin or placebo from
in a prospective study of 208 pregnant women (98 the first trimester to delivery, no difference was found
women with PCOS treated with metformin through- between the two groups. It was concluded that met-
out pregnancy and 110 normal pregnant controls), there formin had no impact on breastfeeding based on the
was a reduction in miscarriage rate (9.1 vs 20%), ges- 186 responders [86] .
tational diabetes (0 vs 13%) and gestational hyperten- In conclusion, it is likely that metformin use in
sion (0 vs 11%) and a non-significant decrease in pre- pregnancy and lactation is safe.
eclampsia (0 vs 3%), in women with PCOS treated with
metformin preconception to 37 weeks gestation [80] . Individualized care
On an individual basis, the clinician must ultimately
Safety of metformin in pregnancy & lactation tailor fertility treatment to the patient’s particular
Metformin is considered a Class B drug in pregnancy, timeline. The onset of action of metformin will be
meaning that metformin is not teratogenic when used slower than clomiphene citrate. It is critical to discuss
in animal models of pregnancy. When compared with patients that clomiphene is a fertility drug that
with insulin for treatment of gestational diabetes acts directly to induce ovulation with a rapid onset of
(GDM), in 751 women with GDM randomized at action and that it is associated with significant multi-
20–33 weeks gestation to either open-label metfor- parity [87] . In contrast, metformin affects metabolism
min (supplemental insulin if required) or insulin, to indirectly induce ovulation with a slow onset of
metformin was found to be safe. No serious adverse action. It results in a more physiologic ovulation and
events were associated with metformin use [81] . In a release of a single egg at a time. An individual patient’s
prospective study of 126 live births (122 pregnancies) favored timeline will help to direct therapy.
to 109 women with PCOS who conceived while on
metformin and continued metformin through preg- Alternative strategies
nancy, it was determined that metformin use reduced Letrozole is an aromatase inhibitor which induces
the prevalence of gestational diabetes, was not found ovulation without anti-estrogenic effects that can
to be teratogenic, and did not adversely affect birth be seen with clomiphene. In clomiphene-resistant
length and weight, growth or motor-social devel- women with PCOS unable to achieve fertility, alter-
opment in the first 18 months of these babies’ lives native strategies of adding metformin to clomiphene,
compared with controls [82] . An RCT of 257 pregnant versus using letrozole with and without additional
women with PCOS randomized to metformin ver- metformin, versus bilateral ovarian drilling can be
sus placebo from first trimester to delivery found no considered. In one RCT of 146 women with PCOS
who did not conceive with clomiphene, there was no PCOS. Insulin resistance unifies the defects seen in
significant difference in ovulation (p = 0.24), preg- PCOS of both increased risk for metabolic syndrome
nancy rate (p = 0.32) or abortion rate (p = 0.51) com- and anovulation and reproductive concerns. Future
paring those randomized to letrozole plus metformin directions of research to tackle unresolved issues in
versus bilateral ovarian drilling [88] . An RCT of letro- PCOS care and understanding of underlying patho-
zole compared with combined metformin and clomi- physiologic mechanism may include pharmacoge-
phene citrate, in 250 women with PCOS who did not netic evaluation of responders versus non-responders
conceive on clomiphene alone, found no statistically to metformin, the role of free fatty acid stimula-
significant difference either in ovulation frequency or tion of androgens and the role that insulin plays,
in pregnancy rates (14.7 vs 14.4%) between these two and whether there are ethnic/genetic differences in
arms [89] . Finally, a head to head combined metformin- PCOS presentations. The molecular defect funda-
letrozole versus metformin-clomiphene citrate trial in mentally underlying insulin resistance in PCOS has
59 women with PCOS who were clomiphene-resistant been hinted at by findings of post-receptor defects
found no statistical difference in pregnancy rate in leading to abnormal patterns of phosphorylation of
the metformin-letrozole group (10 patients, 34.5%) specific residues of the insulin receptor. The latter
versus the metformin-clomiphene group (5 patients, remains to be fully elucidated.
16.67%); however, there were more full term preg-
nancies with metformin-letrozole (10 patients, 34.5% Financial & competing interests disclosure
vs 3 patients, 10%) [90] . The authors note that in this The authors have no relevant affiliations or financial in-
last study metformin was allowed a 6–8 week run-in volvement with any organization or entity with a financial
period prior to additional agent use. As noted pre- interest in or financial conflict with the subject matter or
viously the pathophysiology of metformin affecting materials discussed in the manuscript. This includes employ-
insulin resistance can take time. ment, consultancies, honoraria, stock ownership or options,
expert testimony, grants or patents received or pending, or
Conclusion & future perspective royalties.
In summary, insulin resistance plays a critical role in No writing assistance was utilized in the production of this
both metabolism and reproduction of women with manuscript.
Executive summary
Evolution of a definition
• Polycystic ovary syndrome (PCOS) is a clinical diagnosis.
• PCOS can be present at any BMI, that is, in both thin and obese women.
• When considering the diagnosis of PCOS, obtain tests to assess androgen status and exclude the following
conditions: hyperprolactinemia, thyroid dysfunction, non-classical congenital adrenal hyperplasia and
Cushing’s syndrome if clinically indicated.
Cardiovascular disease in polycystic ovary syndrome
• It is reasonable to consider all women with PCOS at risk of insulin resistance.
• Clinical markers of insulin resistance in PCOS include obesity, increased waist-to-hip ratio, presence of
acanthosis nigricans, low HDL-C, high triglycerides and low sex hormone-binding globulin.
• Screening for women with PCOS should include a measure of adiposity (BMI, waist circumference), serial
blood pressure measurements, serum lipid panel and evaluation for glucose intolerance using an oral glucose
tolerance test. A fasting plasma glucose and HbA1c are not satisfactory screening tests for glucose intolerance
in PCOS.
• Consideration of dietary and lifestyle intervention and/or initiation of an insulin sensitizer (metformin) at the
time of diagnosis of PCOS for primary prevention of DM2 and cardiovascular disease are recommended.
Reproductive consequence in polycystic ovary syndrome
• The clinician must:
–– Assess the individual woman’s sense of readiness or imperative for pregnancy at the time of presentation
and diagnosis, or shortly thereafter.
–– Recognize that:
• Metformin may take a longer time to increase ovulation rates but may do so in a more physiologic
fashion with less multiparity and less hyperstimulation syndrome
• Clomiphene citrate may achieve swifter onset of increased ovulation in a pharmacologic fashion which
may be more commonly associated with multiparity.
• Alternative strategies for infertility are actively being explored.
27 Peppard HR, Marfori J, Iuorno MJ, Nestler JE. Prevalence 41 Salley KE, Wickham EP, Cheang KI, Essah PA, Karjane NW,
of polycystic ovary syndrome among premenopausal women Nestler JE. Glucose intolerance in polycystic ovary syndrome
with Type 2 diabetes. Diabetes Care 24(6), 1050–1052 – a position statement of the Androgen Excess Society. J. Clin.
(2001). Endocrinol. Metab. 92(12), 4546–4556 (2007).
28 Escobar-Morreale HF, Roldan B, Barrio R et al. High 42 Legro RS, Arslanian SA, Ehrmann DA et al. Diagnosis
prevalence of the polycystic ovary syndrome and hirsutism and treatment of polycystic ovary syndrome: an endocrine
in women with Type 1 diabetes mellitus. J. Clin. Endocrinol. society clinical practice guideline. J. Clin. Endocrinol. Metab.
Metab. 85(11), 4182–4187 (2000). 98(12), 4565–4592 (2013).
29 Zachurzok A, Deja G, Gawlik A, Drosdzol-Cop A, • Most recent formal guidelines for PCOS care endorsed by
Malecka-Tendera E. Hyperandrogenism in adolescent girls the Endocrine Society.
with Type 1 diabetes mellitus treated with intensive and
43 Celik C, Abali R, Bastu E, Tasdemir N, Tasdemir UG, Gul
continuous subcutaneous insulin therapy. Endokrynol. Pol.
A. Assessment of impaired glucose tolerance prevalence
64(2), 121–128 (2013).
with hemoglobin A(1)c and oral glucose tolerance test
30 Miyoshi A, Nagai S, Takeda M et al. Ovarian morphology in 252 Turkish women with polycystic ovary syndrome:
and prevalence of polycystic ovary syndrome in Japanese a prospective, controlled study. Hum. Reprod. 28(4),
women with Type 1 diabetes mellitus. J. Diabetes Invest. 4(3), 1062–1068 (2013).
326–329 (2013).
44 Velling Magnussen L, Mumm H, Andersen M, Glintborg
31 Codner E, Cassorla F. Puberty and ovarian function in girls D. Hemoglobin A1c as a tool for the diagnosis of Type 2
with Type 1 diabetes mellitus. Horm. Res. 71(1), 12–21 diabetes in 208 premenopausal women with polycystic ovary
(2009). syndrome. Fertil. Steril. 96(5), 1275–1280 (2011).
32 Chang RJ, Nakamura RM, Judd HL, Kaplan SA. Insulin 45 Hurd WW, Abdel-Rahman MY, Ismail SA, Abdellah MA,
resistance in nonobese patients with polycystic ovarian Schmotzer CL, Sood A. Comparison of diabetes mellitus
disease. J. Clin. Endocrinol. Metab. 57(2), 356–359 (1983). and insulin resistance screening methods for women with
33 Toprak S, Yonem A, Cakir B et al. Insulin resistance in polycystic ovary syndrome. Fertil. Steril. 96(4), 1043–1047
nonobese patients with polycystic ovary syndrome. Horm. (2011).
Res. 55(2), 65–70 (2001). 46 Legro RS, Finegood D, Dunaif A. A fasting glucose to
34 Dunaif A, Segal KR, Futterweit W, Dobrjansky A. Profound insulin ratio is a useful measure of insulin sensitivity in
peripheral insulin resistance, independent of obesity, in women with polycystic ovary syndrome. J. Clin. Endocrinol.
polycystic ovary syndrome. Diabetes 38(9), 1165–1174 Metab. 83(8), 2694–2698 (1998).
(1989). 47 Moran LJ, Ko H, Misso M et al. Dietary composition in the
•• Data supporting the premise that PCOS confers an treatment of polycystic ovary syndrome: a systematic review
increased risk of insulin resistance that is independent of to inform evidence-based guidelines. J. Acad. Nutr. Diet.
BMI. 113(4), 520–545 (2013).
35 Legro RS, Kunselman AR, Dodson WC, Dunaif A. • Systematic review of dietary interventions in women with
Prevalence and predictors of risk for Type 2 diabetes mellitus PCOS suggesting that low glycemic index and higher
and impaired glucose tolerance in polycystic ovary syndrome: fat/lower carbohydrate diets may have some benefit over
a prospective, controlled study in 254 affected women. traditional low fat low calorie diets in women with PCOS.
J. Clin. Endocrinol. Metab. 84(1), 165–169 (1999). 48 Sharma ST, Wickham EP 3rd, Nestler JE. Changes in
36 Ehrmann DA, Barnes RB, Rosenfield RL, Cavaghan MK, glucose tolerance with metformin treatment in polycystic
Imperial J. Prevalence of impaired glucose tolerance and ovary syndrome: a retrospective analysis. Endocr. Pract.
diabetes in women with polycystic ovary syndrome. Diabetes 13(4), 373–379 (2007).
Care 22(1), 141–146 (1999). 49 Talbott E, Clerici A, Berga SL et al. Adverse lipid and
37 Solomon CG, Hu FB, Dunaif A et al. Long or highly coronary heart disease risk profiles in young women with
irregular menstrual cycles as a marker for risk of Type 2 polycystic ovary syndrome: results of a case–control study.
diabetes mellitus. JAMA 286(19), 2421–2426 (2001). J. Clin. Epidemiol. 51(5), 415–422 (1998).
38 Yildiz BO, Yarali H, Oguz H, Bayraktar M. Glucose 50 Legro RS, Kunselman AR, Dunaif A. Prevalence and
intolerance, insulin resistance, and hyperandrogenemia predictors of dyslipidemia in women with polycystic ovary
in first degree relatives of women with polycystic ovary syndrome. Am. J. Med. 111(8), 607–613 (2001).
syndrome. J. Clin. Endocrinol. Metab. 88(5), 2031–2036 51 Talbott EO, Guzick DS, Sutton-Tyrrell K et al. Evidence
(2003). for association between polycystic ovary syndrome and
39 Norman RJ, Masters S, Hague W. Hyperinsulinemia is premature carotid atherosclerosis in middle-aged women.
common in family members of women with polycystic ovary Arterioscler. Thromb. Vasc. Biol. 20(11), 2414–2421 (2000).
syndrome. Fertil. Steril. 66(6), 942–947 (1996). 52 Christian RC, Dumesic DA, Behrenbeck T, Oberg AL,
40 Ehrmann DA, Kasza K, Azziz R, Legro RS, Ghazzi MN. Sheedy PF 2nd, Fitzpatrick LA. Prevalence and predictors of
Effects of race and family history of Type 2 diabetes on coronary artery calcification in women with polycystic ovary
metabolic status of women with polycystic ovary syndrome. syndrome. J. Clin. Endocrinol. Metab. 88(6), 2562–2568
J. Clin. Endocrinol. Metab. 90(1), 66–71 (2005). (2003).
53 Birdsall MA, Farquhar CM, White HD. Association between 66 Asuncion M, Calvo RM, San Millan JL, Sancho J, Avila S,
polycystic ovaries and extent of coronary artery disease in Escobar-Morreale HF. A prospective study of the prevalence
women having cardiac catheterization. Ann. Intern. Med. of the polycystic ovary syndrome in unselected Caucasian
126(1), 32–35 (1997). women from Spain. J. Clin. Endocrinol. Metab. 85(7),
54 Orio F, Jr., Palomba S, Cascella T et al. Early impairment of 2434–2438 (2000).
endothelial structure and function in young normal-weight 67 Homburg R. The management of infertility associated with
women with polycystic ovary syndrome. J. Clin. Endocrinol. polycystic ovary syndrome. Reprod. Biol. Endocrinol. 1, 109
Metab. 89(9), 4588–4593 (2004). (2003).
55 Gami AS, Witt BJ, Howard DE et al. Metabolic syndrome and 68 Diamanti-Kandarakis E, Kouli C, Tsianateli T, Bergiele A.
risk of incident cardiovascular events and death: a systematic Therapeutic effects of metformin on insulin resistance and
review and meta-analysis of longitudinal studies. J. Am. Coll. hyperandrogenism in polycystic ovary syndrome. Eur. J.
Cardiol. 49(4), 403–414 (2007). Endocrinol. 138(3), 269–274 (1998).
56 Apridonidze T, Essah PA, Iuorno MJ, Nestler JE. Prevalence 69 Nestler JE, Jakubowicz DJ, Evans WS, Pasquali R. Effects
and characteristics of the metabolic syndrome in women with of metformin on spontaneous and clomiphene-induced
polycystic ovary syndrome. J. Clin. Endocrinol. Metab. 90(4), ovulation in the polycystic ovary syndrome. N. Engl. J. Med.
1929–1935 (2005). 338(26), 1876–1880 (1998).
57 Thathapudi S, Kodati V, Erukkambattu J, Katragadda A, 70 Baillargeon JP, Jakubowicz DJ, Iuorno MJ, Jakubowicz S,
Addepally U, Hasan Q. Anthropometric and biochemical Nestler JE. Effects of metformin and rosiglitazone, alone
characteristics of polycystic ovarian syndrome in South Indian and in combination, in nonobese women with polycystic
women using AES-2006 criteria. Int. J. Endocrinol. Metabol. ovary syndrome and normal indices of insulin sensitivity.
12(1), e12470 (2014). Fertil. Steril. 82(4), 893–902 (2004).
58 Orio F, Vuolo L, Palomba S, Lombardi G, Colao A. Metabolic 71 Palomba S, Orio F Jr, Falbo A et al. Prospective parallel
and cardiovascular consequences of polycystic ovary randomized, double-blind, double-dummy controlled
syndrome. Minerva Ginecol. 60(1), 39–51 (2008). clinical trial comparing clomiphene citrate and metformin
59 Chekir C, Nakatsuka M, Kamada Y, Noguchi S, Sasaki A, as the first-line treatment for ovulation induction in
Hiramatsu Y. Impaired uterine perfusion associated with nonobese anovulatory women with polycystic ovary
metabolic disorders in women with polycystic ovary syndrome. syndrome. J. Clin. Endocrinol. Metab. 90(7), 4068–4074
Acta Obstet. Gynecol. Scand. 84(2), 189–195 (2005). (2005).
60 Solomon CG, Hu FB, Dunaif A et al. Menstrual cycle 72 Misso ML, Costello MF, Garrubba M et al. Metformin
irregularity and risk for future cardiovascular disease. J. Clin. versus clomiphene citrate for infertility in non-obese women
Endocrinol. Metab. 87(5), 2013–2017 (2002). with polycystic ovary syndrome: a systematic review and
meta-analysis. Hum. Reprod. Update 19(1), 2–11 (2013).
61 Shaw LJ, Bairey Merz CN, Azziz R et al. Postmenopausal
women with a history of irregular menses and elevated 73 Lord JM, Flight IH, Norman RJ. Metformin in polycystic
androgen measurements at high risk for worsening ovary syndrome: systematic review and meta-analysis. BMJ
cardiovascular event-free survival: results from the National 327(7421), 951–953 (2003).
Institutes of Health – National Heart, Lung, and Blood 74 Kashyap S, Wells GA, Rosenwaks Z. Insulin-sensitizing
Institute sponsored Women’s Ischemia Syndrome Evaluation. agents as primary therapy for patients with polycystic
J. Clin. Endocrinol. Metab. 93(4), 1276–1284 (2008). ovarian syndrome. Hum. Reprod. 19(11), 2474–2483
62 Pierpoint T, Mckeigue PM, Isaacs AJ, Wild SH, Jacobs HS. (2004).
Mortality of women with polycystic ovary syndrome at long- 75 Creanga AA, Bradley HM, Mccormick C, Witkop CT. Use
term follow-up. J. Clin. Epidemiol. 51(7), 581–586 (1998). of metformin in polycystic ovary syndrome: a meta-analysis.
63 Mani H, Levy MJ, Davies MJ et al. Diabetes and Obstet. Gynecol. 111(4), 959–968 (2008).
cardiovascular events in women with polycystic ovary 76 Tang T, Lord JM, Norman RJ, Yasmin E, Balen AH.
syndrome: a 20-year retrospective cohort study. Clin. Insulin-sensitising drugs (metformin, rosiglitazone,
Endocrinol. (Oxf.) 78(6), 926–934 (2013). pioglitazone, D-chiro-inositol) for women with polycystic
64 Wild RA, Carmina E, Diamanti-Kandarakis E et al. ovary syndrome, oligo amenorrhoea and subfertility.
Assessment of cardiovascular risk and prevention of Cochrane Database Syst. Rev. 5, Cd003053 (2012).
cardiovascular disease in women with the polycystic ovary 77 Legro RS, Barnhart HX, Schlaff WD et al. Clomiphene,
syndrome: a consensus statement by the Androgen Excess metformin, or both for infertility in the polycystic ovary
and Polycystic Ovary Syndrome (AE-PCOS) Society. J. Clin. syndrome. N. Engl. J. Med. 356(6), 551–566 (2007).
Endocrinol. Metab. 95(5), 2038–2049 (2010). • Pregnancy in PCOS study comparing clomiphene
• Guideline by the Androgen Excess and PCOS Society versus metformin or combination therapy for infertility.
regarding cardiovascular disease and PCOS. There is some controversy regarding the interpretation
65 Knochenhauer ES, Key TJ, Kahsar-Miller M, Waggoner of these findings and the methodology of metformin
W, Boots LR, Azziz R. Prevalence of the polycystic ovary administration to consider.
syndrome in unselected black and white women of the 78 Morin-Papunen L, Rantala AS, Unkila-Kallio L et al.
southeastern United States: a prospective study. J. Clin. Metformin improves pregnancy and live-birth rates
Endocrinol. Metab. 83(9), 3078–3082 (1998).
in women with polycystic ovary syndrome (PCOS): a 85 Hale TW, Kristensen JH, Hackett LP, Kohan R, Ilett KF.
multicenter, double-blind, placebo-controlled randomized Transfer of metformin into human milk. Diabetologia 45(11),
trial. J. Clin. Endocrinol. Metab. 97(5), 1492–1500 (2012). 1509–1514 (2002).
•• Study results published after the Endocrine Society clinical 86 Vanky E, Nordskar JJ, Leithe H, Hjorth-Hansen AK,
practice guidelines on PCOS, therefore not included. Martinussen M, Carlsen SM. Breast size increment during
pregnancy and breastfeeding in mothers with polycystic
79 Jakubowicz DJ, Iuorno MJ, Jakubowicz S, Roberts KA,
ovary syndrome: a follow-up study of a randomised
Nestler JE. Effects of metformin on early pregnancy loss in
controlled trial on metformin versus placebo. BJOG 119(11),
the polycystic ovary syndrome. J. Clin. Endocrinol. Metab.
1403–1409 (2012).
87(2), 524–529 (2002).
87 Kulkarni AD, Jamieson DJ, Jones HW, Jr et al. Fertility
80 De Leo V, Musacchio MC, Piomboni P, Di Sabatino A,
treatments and multiple births in the United States. N. Engl.
Morgante G. The administration of metformin during
J. Med. 369(23), 2218–2225 (2013).
pregnancy reduces polycystic ovary syndrome related
gestational complications. Eur. J. Obstet. Gynecol. Reprod. 88 Abd Elgafor I. Efficacy of combined metformin-letrozole in
Biol. 157(1), 63–66 (2011). comparison with bilateral ovarian drilling in clomiphene-
resistant infertile women with polycystic ovarian syndrome.
81 Rowan JA, Hague WM, Gao W, Battin MR, Moore MP.
Arch. Gynecol. Obstet. 288(1), 119–123 (2013).
Metformin versus insulin for the treatment of gestational
diabetes. N. Engl. J. Med. 358(19), 2003–2015 (2008). 89 Abu Hashim H, Shokeir T, Badawy A. Letrozole versus
combined metformin and clomiphene citrate for ovulation
82 Glueck CJ, Goldenberg N, Pranikoff J, Loftspring M, Sieve
induction in clomiphene-resistant women with polycystic
L, Wang P. Height, weight, and motor-social development
ovary syndrome: a randomized controlled trial. Fertil. Steril.
during the first 18 months of life in 126 infants born to 109
94(4), 1405–1409 (2010).
mothers with polycystic ovary syndrome who conceived on
and continued metformin through pregnancy. Hum. Reprod. 90 Sohrabvand F, Ansari S, Bagheri M. Efficacy of combined
19(6), 1323–1330 (2004). metformin-letrozole in comparison with metformin-
clomiphene citrate in clomiphene-resistant infertile women
83 Vanky E, Stridsklev S, Heimstad R et al. Metformin versus
with polycystic ovarian disease. Hum. Reprod. 21(6),
placebo from first trimester to delivery in polycystic ovary
1432–1435 (2006).
syndrome: a randomized, controlled multicenter study.
J. Clin. Endocrinol. Metab. 95(12), E448–E455 (2010).
84 Eyal S, Easterling TR, Carr D et al. Pharmacokinetics of
metformin during pregnancy. Drug Metab. Dispos. 38(5),
833–840 (2010).