Metabolic disturbances in

non-obese women with polycystic

ovary syndrome: a systematic
review and meta-analysis
Shiqin Zhu, M.D.,a,b,c Bingqian Zhang, M.D.,a,b,c Xiao Jiang, M.D.,a,b,c Zeyan Li, M.D.,d
Shigang Zhao, M.D., Ph.D.,a,b,c Linlin Cui, M.D., Ph.D.,a,b,c and Zi-Jiang Chen, M.D., Ph.D.a,b,c,e,f
a
Center for Reproductive Medicine, Shandong Provincial Hospital Affiliated to Shandong University, b National Research
Center for Assisted Reproductive Technology and Reproductive Genetics, c The Key Laboratory of Reproductive
Endocrinology (Shandong University), Ministry of Education, Jinan; and d Department of Urology, Qilu Hospital of
Shandong University, e Center for Reproductive Medicine, Ren Ji Hospital, School of Medicine, Shanghai Jiao Tong
University and f Shanghai Key Laboratory for Assisted Reproduction and Reproductive Genetics, Shanghai, People's
Republic of China

Objective: To explore metabolic disturbances in nonobese women with polycystic ovary syndrome (PCOS) compared with nonobese
healthy controls.
Design: Systematic review and meta-analysis.
Setting: Not applicable.
Patient(s): Nonobese women with PCOS and nonobese healthy controls.
Intervention(s): None.
Main Outcome Measure(s): Prevalence of metabolic disturbances including hyperinsulinemia, insulin resistance (IR), impaired fast-
ing glucose (IFG), impaired glucose intolerance (IGT), prediabetes, dyslipidemia, hypercholesterolemia, hypertriglyceridemia, and low
high-density lipoprotein (low-HDL), as well as other metabolic outcomes such as type 2 diabetes mellitus (T2DM), hypertension,
metabolic syndrome (Mets), myocardial infarction, stroke, cerebrovascular accident, arterial occlusive disease, and coronary heart
disease.
Result(s): Compared to nonobese controls, nonobese women with PCOS showed a higher prevalence of hyperinsulinemia (odds ratio
[OR], 36.27; 95% confidence interval [CI] 1.76–747.12), IR (OR, 5.70; 95% CI 1.46–22.32), IGT (OR, 3.42; 95% CI 1.56–7.52), T2DM (OR,
1.47; 95% CI 1.11–1.93), hypertriglyceridemia (OR, 10.46; 95% CI 1.39–78.56), low-HDL (OR, 4.03; 95% CI 1.26–12.95), and Mets (OR,
2.57; 95% CI 1.30–5.07). No significant difference was observed for IFG, pre-DM, dyslipidemia, hypercholesterolemia, and
hypertension. In subgroup analysis, Whites exhibited increased risks of IR, IGT, IFG, T2DM, hypertension, and Mets, whereas no
significant metabolic change was found in Asians. No study reported specifically an incidence of myocardial infarction, stroke,
cerebrovascular accident, arterial occlusive disease, and coronary heart disease in nonobese women with PCOS.
Conclusion(s): Nonobese women with PCOS also suffer from metabolic disturbances and the risk of long-term metabolic
complications. Further efforts should be made to elucidate underlying mechanisms and possible interventions in the early phase.
(Fertil SterilÒ 2019;111:168–77. Ó2018 by American Society for Reproductive Medicine.)
El resumen está disponible en Español al final del artículo.
Key Words: Polycystic ovary syndrome, lean, metabolic syndrome, diabetes mellitus, systematic review
Discuss: You can discuss this article with its authors and other readers at https://www.fertstertdialog.com/users/16110-fertility-
and-sterility/posts/39194-26341

Received May 19, 2018; revised and accepted September 25, 2018.
S.Z. has nothing to disclose. B.Z. reports grant from Shandong Provincial Hospital Affiliated to Shandong University. X.J. reports grant from Shandong Pro-
vincial Hospital Affiliated to Shandong University. Z.L. has nothing to disclose. S.Z. reports grant from Shandong Provincial Hospital Affiliated to Shan-
dong University. L.C. reports grant from Shandong Provincial Hospital Affiliated to Shandong University. Z.-J.C. reports grant from Shandong
Provincial Hospital Affiliated to Shandong University.
Funded by The National Key Research and Development Program of China (No. 2017YFC1001000), National Natural Science Foundation of China (No.
81501223), and Young Scholars Program of Shandong University.
Reprint requests: Linlin Cui, M.D., Ph.D., Center for Reproductive Medicine, Shandong Provincial Hospital Affiliated to Shandong University, Jingliu Road
157, 250001 Jinan, People's Republic of China (E-mail: [email protected]).

Fertility and Sterility® Vol. 111, No. 1, January 2019 0015-0282/$36.00

Copyright ©2018 American Society for Reproductive Medicine, Published by Elsevier Inc.
https://doi.org/10.1016/j.fertnstert.2018.09.013

168 VOL. 111 NO. 1 / JANUARY 2019

 Fertility and Sterility®

P
olycystic ovary syndrome (PCOS), as one of the most tional information if necessary. Abstracts containing neces-
common gynecological endocrine diseases, affects sary information were considered.
6%–8% of women of reproductive age (1, 2). In
addition to typical reproductive endocrinal changes, namely Inclusion and Exclusion Criteria
oligoovulation and anovulation, hyperandrogenism, and
polycystic ovaries (PCOs) morphology and metabolic Two independent reviewers (S.Z. and L.C.) identified and
disturbances are also identified in most patients (3, 4). selected the articles according to the inclusion and exclusion
Accumulated evidence has demonstrated increasing risks of criteria. The articles that compared the prevalence of meta-
metabolic complications including insulin resistance (IR), bolic abnormities between nonobese women with PCOS and
type 2 diabetes mellitus (T2DM), metabolic syndrome (Mets), BMI-matched controls were considered as eligible studies.
and cardiovascular disease (CVD) in this group of women The inclusion criteria were as follows: [1] observational
(5, 6). studies; [2] studies focused on evaluating prevalence of any
Obesity is extensively considered as the key basis of these metabolic abnormities between the nonobese premenopausal
metabolic complications, which can be observed in approxi- women with PCOS and nonobese premenopausal controls; [3]
mately half of women with PCOS. However, the metabolic dichotomous data were reported or sufficient data were avail-
features of patients with PCOS, but without obesity, are able to extrapolate the odds ratio (OR) and its 95% confidence
controversial. Some studies (3, 7, 8) demonstrated that interval (CI); [4] the most recent and complete study was
patients with PCOS are still at increased risk of metabolic selected when overlapped datasets appeared; and [5] studies
dysfunctions after the adjustment of body mass index reporting two independent cohorts were included separately.
(BMI). A previous meta-analysis (3) also indicated an elevated Polycystic ovary syndrome was defined in accordance
prevalence of impaired glucose tolerance (IGT) and Mets espe- with the National Institutes of Health (NIH) criteria (12), or
cially in lean women (BMI <25 kg/m2) with PCOS compared the European Society of Human Reproduction and Embry-
with BMI-matched controls. Whereas, Cheung et al. (9) and ology/American Society for Reproductive Medicine (ESHRE/
Ollila et al. (10) suggested equivalent metabolic risks between ASRM) criteria (13, 14), or the Androgen Excess and PCOS
nonobese women with PCOS and BMI-matched healthy Society (AES) criteria (15). For some studies before the 21st
controls. century, definitions of PCOS based on the original articles
Polycystic ovary syndrome is complex with heteroge- were also accepted. Nonobesity was defined as a BMI
neous manifestation and complicated etiology. Revealing <30 kg/m2 for Whites and BMI <25 kg/m2 for Asians,
the metabolic condition of nonobese women with PCOS can according to World Health Organization criteria (WHO
guide clinical screening and treatment, and indicate the 1998, WHO 2000) (16). Studies with BMI cutoff values <30
possible metabolic pathway in this patient group. The aim in Whites and 25 in Asians were included. Definitions of
of the present systematic review and meta-analysis is to esti- outcomes were based on the descriptions of original studies.
mate the differences in prevalence of glucose metabolic pro- Detailed definitions of outcomes in each study were listed in
files, cardiovascular risk factors, and Mets between nonobese Supplemental Table 2 (available online). The exclusion
women of reproductive age with and without PCOS and to ac- criteria were as follows: [1] animal studies, [2] participants
cess the effect of the confounding factors such as PCOS defi- were investigated after certain treatments, [3] lack of BMI
nition and ethnicity. classification, [4] invalid definition of PCOS, [5] <20
observations, and [6] studies involving data from adolescents.

MATERIALS AND METHODS Data Extraction

Search Strategy
Data were extracted from all eligible articles. These included
This systematic review and meta-analysis was reported sample size, age, race, study location, study design, baseline
based on the Preferred Reporting Item for Systematic Re- health state, definition of PCOS, BMI category, outcomes,
views and Meta-analyses statement (11). Electronic litera- definition of outcomes, calculated OR and its 95% CI, and
ture search was performed in MEDLINE, Web of Science, adjusted confounders.
Embase, and Cochrane library databases up to November
2017. The following terms were used in automatic search:
polycystic ovary syndrome, PCOS, Stein Leventhal Syn- Quality Assessment
drome, hyperlipidemia, hypertension, cardiovascular disease, Included studies were critically assessed according to the
CVD, Diabetes Mellitus, T2DM, Prediabetes, impaired Newcastle-Ottawa Quality Assessment Scale (NOS) (17), a
glucose tolerance, IGT, impaired fasting glucose, impaired valid tool recommended by the Cochrane Working Group to
fasting glucose (IFG), insulin resistance, IR, hyperinsuline- appraise the risk of bias of nonrandomized studies. As for
mia, Metabolic Syndrome, Mets, lean, non-obese, and cross-sectional studies, an adapted NOS scale was used. This
normal weight. Detailed search strategy was shown in modified scale, proposed by Barry et al. (18), adjusted the
Supplemental Table 1 (available online). Languages were item of ‘‘ascertainment of diagnosis’’ to ‘‘ascertainment of
limited to English and Chinese. exposure,’’ which made NOS applicable for cross-sectional
Reference lists and relevant reviews were also manually studies. Studies were graded into low quality (0–3 points), me-
searched to avoid omission of potential eligible studies. First dium quality (4–6 points), and high quality (7–9 points) ac-
authors and corresponding authors were contacted for addi- cording to NOS score. Two reviewers (S.Z. and L.C.)

VOL. 111 NO. 1 / JANUARY 2019 169

ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY

independently participated in quality assessment. Consensus disturbances including IR, HIN, IGT, IFG, pre-DM, and T2DM
was reached to resolve the disagreements. (7–10, 23–34); 7 reported CVD risk factors, including
dyslipidemia, hypertension, high-TC, high-TG, and low-HDL
(9, 25, 27, 30, 35–37) and 7 assessed Mets (7, 9, 27, 32, 38–
Outcomes of Interest
40). No study investigated the prevalence of myocardial
Some outcomes of interest were the prevalence of metabolic infarction, stroke, cerebrovascular accident, arterial
disturbances, which included hyperinsulinemia (HIN), IR, occlusive disease, and coronary heart disease in nonobese
IFG, IGT, pre-diabetes (pre-DM), dyslipidemia, hypercholes- women with PCOS. Studies were published from 1986 to
terolemia (high-TC), hypertriglyceridemia (high-TG), and 2017 and overall 9,967 participants were recruited. Multiple
low high-density lipoprotein (low-HDL). Other metabolic out- ethnic groups were included in the present review as
comes included T2DM, hypertension, Mets, myocardial follows: White (n ¼ 16) (7, 8, 10, 23, 27–31, 33–37, 39, 40),
infarction, stroke, cerebrovascular accident, arterial occlusive Mediterranean (n ¼ 2) (26, 32), East Asian (n ¼ 2) (9, 24),
disease, and coronary heart disease. and Indian (n ¼ 1) (38). Definitions of PCOS were mainly
according to the NIH criteria (n ¼ 6) (23, 25–27, 37, 40) and
Statistical Analysis the Rotterdam criteria (n ¼ 11) (7, 9, 24, 32–35, 37–40).
Four studies adopted original definitions (28–31) and 2
The OR and its 95% CI were used to estimate outcomes in non-
studies were based on structured interview and self-reports
obese women with PCOS versus nonobese healthy controls.
(10, 36).
Hazard ratio was considered equivalent to OR (19). Studies
with zero events in both groups were omitted automatically
from meta-analysis because of their limited contribution to Methodological Quality
the pooled estimation (20). Pooled ORs were calculated using
Assessment of risk of bias is illustrated in Supplemental
the fixed-effects model and heterogeneity between studies
Table 3 (available online). Overall, 9 studies were graded
was measured by the Q-test and I2 statistics. P< .1 and
into high quality. One study was recognized as low quality
I2>50% indicated significant heterogeneity, where the
because of limited data for NOS score. Seventeen studies re-
random-effects model was applied instead. Subgroup ana-
cruited subjects matching for confounders such as age, BMI,
lyses were stratified by ethnic, age, different definitions,
and waist and hip circumference. No history of medication
BMI category, study design, NOS quality, adjusted waist
was declared in 11 studies, whereas 2 studies adjusted the
circumference, and adjusted age. Sensitivity analyses were
condition of treatments. In 10 studies, early events were
performed to test the stability of overall analysis by omitting
excluded before cases recruitment. In 9 studies, other relevant
single research or changing effect model. Publication bias was
information such as smoking, alcohol use, and family history
assessed by Egger's regression test (21, 22). Nonparametric
of metabolic diseases, was also recorded (Supplemental
trim-and-fill method was conducted to test publication bias
Table 4, available online). Given the remarkable between-
when Egger's regression test was not applicable. Significant
studies heterogeneity of clinical characteristics, sensitivity
publication bias was defined as P< .1. Two-tail P value was
analyses and publication bias assessment were conducted
adopted in this review and P< .05 was considered significant.
for each analysis.
All statistical analyses were accomplished in STATA 12.0.

RESULTS Comparison in Glucose Metabolism Outcomes

Search Results According to meta-analysis, increased risks of IR (OR, 5.70;
Search results are illustrated in the flow chart (Fig. 1). A total 95% CI 1.46–22.32; P¼ .012; Ph ¼ .005; I2 ¼ 81.1%) and
of 7,432 articles were yielded in the electronic search and 27 IGT (OR, 3.42; 95% CI 1.56–7.52; P¼ .002; Ph ¼ .310; I2 ¼
articles were searched from reference list or reviews. After title 16.3%) were found in nonobese premenopausal women with
and abstract screening based on prior selection criteria, 118 PCOS. One study reported higher prevalence of HIN in nonob-
articles were identified for full text assessment. In further ex- ese women with PCOS (OR 36.27; 95% CI 1.76–747.12) (29). In
amination, 35 studies were excluded for irrelevant outcomes, addition, nonobese women with PCOS also showed increased
43 were excluded for undesirable participants, 9 were risk of T2DM (OR 1.47; 95% CI 1.11–1.93; P¼ .007; Ph ¼
excluded for foreign languages, 7 were excluded for involving .555; I2 ¼ 0). When the design of the study was restricted to
adolescents, and 2 were excluded for overlapped datasets. cohort, pooled OR remained significant (OR 1.48; 95% CI
Eventually, 22 full-text studies were included in the present 1.12–1.95; P¼ .007; Ph ¼ .245; I2 ¼ 29.0%). However, there
meta-analysis. was no remarkable difference of prevalence in IFG and pre-
DM (for IFG: OR 1.08; 95% CI 0.46–2.53; P¼ .864; Ph ¼ .109
I2 ¼ 50.4%; for pre-DM: OR 1.39; 95% CI 0.73–2.63;
Studies Characteristics P¼ .317; Ph ¼ .459; I2 ¼ 0) (Table 1, Fig. 2). Subgroups with
Characteristics of all 22 included studies are displayed in increased risks of metabolic disturbances were listed in
Supplemental Table 2. All eligible studies investigated the Supplemental Table 5 (available online). In subgroup analyses,
prevalence of metabolic abnormities in nonobese premeno- the risks of IR, IGT, IFG, and T2DM were elevated in nonobese
pausal women with PCOS and nonobese women without White patients (P< .05), whereas in other ethnic groups, only
PCOS. Among them, 16 studies focused on glucose metabolic elevated risk of IR was identified in Mediterraneans. Except

170 VOL. 111 NO. 1 / JANUARY 2019

 Fertility and Sterility®

FIGURE 1

Flow diagram for study selection.

Zhu. Metabolic disturbances in non-obese PCOS. Fertil Steril 2018.

for IR, which was pooled from studies using only Rotterdam 1.39–78.56; P¼ .022; Ph ¼ .554; I2 ¼ 0; for low-HDL: OR 4.03;
criteria, all other parameters of glucose metabolism were 95% CI 1.26–12.95; P¼ .019; Ph ¼ .626; I2 ¼ 0). But no signif-
analyzed further after dividing by PCOS definition. Significant icant difference was found in prevalence of high-TC, dyslipi-
increase was still found in IGT in cases based on the Rotterdam demia, and hypertension (for high-TC: OR 5.78; 95% CI
criteria (OR 2.74; 95% CI 1.19–6.31; P¼ .018), and T2DM in 0.31–107.92; for dyslipidemia: OR 1.87; 95% CI 0.85–4.13;
cases based on the NIH criteria (OR 2.76; 95% CI 1.12–6.88; P¼ .121; Ph ¼ .913; I2 ¼ 0; for hypertension: OR 2.44; 95%
P¼ .030) and on medical record (OR 1.37; 95% CI 1.03–1.84; CI 0.80–7.43; P¼ .117; Ph ¼ .117; I2 ¼ 53.3%) (Table 1,
P¼ .034). There is no difference of IFG based on the Rotterdam Fig. 3). In subgroup analyses, between-studies heterogeneity
criteria and pre-DM based on the NIH criteria. Impaired for hypertension could be explained by ethnicity. Whites dis-
glucose intolerance and IFG based on the NIH criteria as well played a fourfold increased risk of hypertension (OR 4.27;
as T2DM based on the Rotterdam criteria were not meta- 95% CI 2.04–8.96; P< .001; Ph ¼ .812; I2 ¼ 0). After being
analyzed because of the limited amount of studies included, divided by PCOS definition, elevated risk of hypertension re-
but no difference was reported in each of the above studies. mained in cases based on self-report (OR 4.37; 95% CI 2.04–
Detailed results of subgroup analyses are illustrated in 9.38; P< .001), whereas no difference was found in cases
Supplemental Tables 6–10 (available online). diagnosed by the Rotterdam criteria (OR 1.22; 95% CI 0.42–
3.60; P¼ .714). As for other parameters of CVD, both
high-TG and low-HDL were pooled from studies using the
Comparison in CVD Risk Factors Rotterdam criteria and only one study, which reported no dif-
Elevated prevalence of high-TG and low-HDL were showed in ference in high-TC, was adopted the Rotterdam, the NIH, and
nonobese patients with PCOS (for high-TG: OR 10.46; 95% CI the Androgen Excess and PCOS Society criteria. No difference

VOL. 111 NO. 1 / JANUARY 2019 171

ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY

TABLE 1

Meta-analysis results for glucose metabolic disturbances and cardiovascular disease risk factors.
Heterogeneity
Outcome No. of studies Effects model OR (95%CI) P value Ph value I2 (%)
Comparison in glucose metabolism disturbances
HIN 1 36.27 (1.76, 747.12)
IR 3 Random 5.70 (1.46, 22.32) .012 0.005 81.1
IFG 4 Random 1.08 (0.46, 2.53) .864 0.109 50.4
IGT 4 Fixed 3.42 (1.56, 7.52) .002 0.310 16.3
Pre-DM 3 Fixed 1.39 (0.73, 2.63) .317 0.459 0
T2DM 5 Fixed 1.47 (1.11, 1.93) .007 0.555 0
T2DM cohort 3 Fixed 1.48 (1.12, 1.95) .007 0.245 29
Comparison in CVD risk factors
Dyslipidemia 2 Fixed 1.87 (0.85, 4.13) .121 0.913 0
high-TC 1 5.78 (0.31, 107.92)
high-TG 2 Fixed 10.46 (1.39, 78.56) .022 0.554 0
low-HDL 2 Fixed 4.03 (1.26, 12.95) .019 0.626 0
Hypertension 3 Random 2.44 (0.80, 7.43) .117 0.117 53.3
Note: CI ¼ confidence interval; CVD ¼ cardiovascular disease; high-TC ¼ hypercholesterolemia; high-TG ¼ hypertriglyceridemia; HIN ¼ hyperinsulinemia; IFG ¼ impaired fasting glucose; IGT ¼
impaired glucose intolerance; IR ¼ insulin resistance; low-HDL ¼ low high-density lipoprotein; Pre-DM ¼ IGT plus IFG; OR ¼ odds ratio; T2DM ¼ type 2 diabetes mellitus.
Zhu. Metabolic disturbances in non-obese PCOS. Fertil Steril 2018.

FIGURE 2

Meta-analysis on insulin resistance (IR), impaired glucose tolerance (IGT), impaired fasting glucose (IFG), and type 2 diabetes mellitus (T2DM):
nonobese women with PCOS versus nonobese healthy controls. Forest plot displayed odds of IR (A), IGT (B), IFG (C), and T2DM (D) in
nonobese women with PCOS versus nonobese healthy controls.
Zhu. Metabolic disturbances in non-obese PCOS. Fertil Steril 2018.

172 VOL. 111 NO. 1 / JANUARY 2019

 Fertility and Sterility®

FIGURE 3

Forrest plot displayed odds of hypertension and metabolic syndrome (Mets) in nonobese women with PCOS versus nonobese healthy controls. (A)
Meta-analysis on hypertension; (B) meta-analysis on Mets.
Zhu. Metabolic disturbances in non-obese PCOS. Fertil Steril 2018.

was found in dyslipidemia on either the NIH or the Rotterdam glucose metabolic disturbances including HIN, IR, IGT, and
criteria. Detailed results are shown in Supplemental Table 5 T2DM compared with BMI-matched controls. In addition, fre-
and Supplemental Tables 11–14, available online. quencies of high-TG, low-HDL, and Mets were increased in
nonobese women with PCOS, which indicated more risk of
adverse metabolic outcomes such as CVD.
Comparison in Mets Risk
Insulin resistance was supposed to be strongly correlated
Using the fixed-effects model, a 2.6-fold increased risk of with obesity, but in our meta-analysis, it was also more
Mets was found in nonobese women with PCOS compared frequently encountered in nonobese patients with PCOS
with nonobese controls (OR 2.57; 95% CI 1.30–5.07; than in nonobese controls. This suggested that the pathophys-
P¼ .007; Ph ¼ .338; I2 ¼ 12.1%) (Fig. 3). In subgroup analyses, iological change may not only be a result consequent to
consistent results were observed in the White group (OR 2.40; obesity but also an intrinsic etiologic basis for PCOS regard-
95% CI 1.05–5.49; P¼ .039) and the group in which Mets was less of obesity. Relative genes variations, defect of the certain
defined by the National Cholesterol Education Program signaling pathway, and the vicious circle between the two
(NCEP) Adult Treatment Panel III criteria (OR 3.77; 95% CI with hyperandrogenism are proposed as possible underlying
1.62–8.77; P¼ .002) (Supplemental Table 5). However, no dif- mechanisms. Previous genome-wide association studies (41,
ference was found in the non-White group and the group in 42) revealed that INSR and DENND1A conferred to a PCOS
which IDF was used as the definition for Mets. All eligible risk. In subsequent study (42) these two genes were
studies on Mets were based on the Rotterdam criteria demonstrated to be associated with IR in patients with
(Supplemental Table 15, available online). PCOS, which suggested a genetic basis. In addition, tissue-
specific defects in the insulin signal transduction were also
Sensitivity Analyses and Publication Bias involved in IR in PCOS. Except for well-accepted abnormality
in the adipocyte of subjects with PCOS, myotubes also dis-
Sensitivity analyses verified the robustness of the results of all
played impaired insulin responsiveness for glucose disposal.
outcomes except for T2DM. Significant variations were intro-
It was correlated with diminished insulin-stimulated Akt
duced in pooled OR of T2DM after omitting one study (OR
phosphorylation, a key step of insulin signaling pathway, in
1.96; 95% CI 0.95–4.02). No obvious publication bias was de-
skeletal muscle of patients with PCOS (43, 44). It could in
tected according to Egger's tests and trim and filled analyses
part explain IR in nonobese patient with PCOS. In addition,
(Supplemental Table 16, available online). A sensitivity anal-
animal studies (45, 46) indicated that hyperandrogenism
ysis graph of Mets is shown in Supplemental Figure 1, avail-
could contribute to IR by increasing serine phosphorylation
able online.
on myotubes and facilitating visceral fat accumulation. The
paralleled change of serum-free T and reduced insulin-
DISCUSSION stimulated glucose disposal were also confirmed in women
The present meta-analysis showed that even without obesity, with PCOS (44). Hyperinsulinemia was also observed in non-
PCOS patients still presented an increased prevalence of obese women with PCOS according to the present meta-

VOL. 111 NO. 1 / JANUARY 2019 173

ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY

analysis. Except for a compensatory response to IR, reduced Mets. In our study, the prevalence of metabolic
hepatic clearances of insulin provided correlative evidence disturbances, including IR, IGT, high-TG, low-HDL, and
as well (47, 48). The intrinsic abnormalities in IR Mets, were increased in nonobese women with PCOS, as diag-
contributed to elevated risk for adverse metabolic outcomes nosed by the Rotterdam consensus. It was partly consistent
including IGT, T2DM, and Mets in nonobese women with with the study of Yildiz et al. (65), which reported a twofold
PCOS, which was demonstrated by the present study. increased risk of Mets in patients with PCOS regardless of
According to the present study, nonobese patients with diagnostic criteria adopted.
PCOS showed higher prevalence of high-TG and low-HDL The results of the present study provided relative solid ev-
compared with their counterparts. Dyslipidemia was one of idence for metabolic deterioration in nonobese women with
the most frequent cardiovascular metabolic disturbances in PCOS based on systemic review and standard meta-analysis
women with PCOS, which could exaggerate the risk for of all published studies. However, several limitations still ex-
CVD (35). The reason for abnormal lipid metabolism is multi- isted. First, the heterogeneity caused by different definitions
factorial, and involves abnormal hormone changes such as of nonobesity and metabolic disturbances in the studies,
insulin, androgen, and estrogen (E), as well as environmental and lack of baseline information, including medical history,
factors (49). Enhanced ApoCI-mediated block of several en- family history, smoking, and alcohol conditions, would still
zymes and receptors within pathway of lipoprotein transfer bring biases. As a remedy, subgroup analyses and sensitivity
and metabolism was supposed as the possible mechanism in analyses were conducted in the present meta-analysis. Sec-
normal-weight women with PCOS (50). The guidelines from ond, in most of the studies, the cutoff values of BMI were
the Androgen Excess and PCOS Society recommend that lipid below the WHO criteria, which might shrink the case group
pattern screening be performed in all women with PCOS (51). and cover up the difference of outcomes to some extent. Pro-
However, no difference was found in hypertension, another spective cohort studies based on homogeneous populations of
CVD risk factor, in nonobese women with and without large sample size with strict definition of nonobesity are still
PCOS. The relative young age of subjects enrolled might be needed to verify the conclusion. Third, because the number of
an important confounding factor. Cohort studies with a eligible studies in each subgroup was limited, the results of
longer duration of follow-up are needed. subgroup analyses should be explained with more caution.
Our study identified a significant increase in Mets inci- In addition, metabolic dysfunction is a life-long process
dence in nonobese patients with PCOS. Mets is a cluster of that would exaggerate risk of adverse metabolic outcomes
dysfunction in glucose and lipid metabolism including central such as CVD with aging. The relative young age of the
obesity, glucose intolerance, dyslipidemia, and hypertension enrolled subjects in the study may conceal some outcomes
(52, 53). It is a well-documented risk factor for T2DM and of interest. Thus the conclusions should be confirmed in a
CVD. Insulin resistance and central obesity are considered well-designed longitudinal cohort study with confounding
to play a common etiologic role in Mets and PCOS (54, 55). factors precisely controlled.
Impaired insulin sensitivity and responsiveness would result In conclusion, our study demonstrated that nonobese
in hyperglycemia and hyperlipidemia by reducing adipocyte women with PCOS also suffered from glucose and lipid meta-
glucose uptake and stimulating adipocyte lipolysis (56, 57). bolic disturbances, as well as metabolic complications
Central obesity with excess visceral fat could exacerbate IR including T2DM and Mets. It suggested that early screening
in normal weight subjects (58, 59). After adjustment for and intervention should be extended to this subset of patients.
waist circumference, no difference was found for Mets Future study should focus on elucidating the underlying
incidence according to the results of subgroup analysis in mechanism and finding the optimal time point for screening
the present study. It suggested that accumulation of visceral and intervention.
fat might be a key reason for the development of Mets in
nonobese women with PCOS. Notably, various diagnostic Acknowledgments: The authors thank Professor Helena J.
criteria of Mets were adopted in the included literature, Teede and Professor Onno E. Janssen for providing the pre-
which was a great confounding factor. On subgroup liminary information.
analyses, increased frequency of Mets as defined only in
Adult Treatment Panel III was maintained in nonobese
women with PCOS. REFERENCES
Differences of ethnicity and PCOS diagnostic criteria were
1. Diamanti-Kandarakis E, Kouli CR, Bergiele AT, Filandra FA, Tsianateli TC,
noted to result in significant clinical heterogeneity in the pre- Spina GG, et al. A survey of the polycystic ovary syndrome in the Greek island
sent study. Therefore, subgroup analyses based on these two of Lesbos: hormonal and metabolic profile. J Clin Endocrinol Metab 1999;
variables were conducted. Differences were maintained for 84:4006–11.
IR, IGT, IFG, T2DM, hypertension, and Mets in nonobese 2. Norman RJ, Dewailly D, Legro RS, Hickey TE. Polycystic ovary syndrome. Lan-
White, and IR in Mediterranean, but none in Asian. Specific cet 2007;370:685–97.
3. Moran LJ, Misso ML, Wild RA, Norman RJ. Impaired glucose tolerance, type
genetic background (42, 60) and environmental exposures,
2 diabetes and metabolic syndrome in polycystic ovary syndrome: a system-
such as high-fat diet and smoking (61, 62), were the most
atic review and meta-analysis. Hum Reprod Update 2010;16:347–63.
possible reasons. With regard to PCOS diagnostic criteria, 4. Cooney LG, Lee I, Sammel MD, Dokras A. High prevalence of moderate and
some previous studies (63, 64) showed that women severe depressive and anxiety symptoms in polycystic ovary syndrome: a sys-
diagnosed with the NIH criteria had a higher prevalence of tematic review and meta-analysis. Hum Reprod 2017;32:1075–91.

174 VOL. 111 NO. 1 / JANUARY 2019

 Fertility and Sterility®

5. Moran LJ, Norman RJ, Teede HJ. Metabolic risk in PCOS: phenotype and tients with polycystic ovary syndrome than in women who do not have
adiposity impact. Trends Endocrinol Metab 2015;26:136–43. hyperandrogenism. Fertil Steril 2010;94:2214–21.
6. Ali AT. Polycystic ovary syndrome and metabolic syndrome. Ceska Gynekol 27. Faloia E, Canibus P, Gatti C, Frezza F, Santangelo M, Garrapa G, et al.
2015;80:279–89. Body composition, fat distribution and metabolic characteristics in lean
7. Attaoua R, El Mkadem SA, Radian S, Fica S, Hanzu F, Albu A, et al. FTO gene and obese women with polycystic ovary syndrome. J Endocrinol Invest
associates to metabolic syndrome in women with polycystic ovary syndrome. 2004;27:424–9.
Biochem Biophys Res Commun 2008;373:230–4. 28. Morin-Papunen LC, Vauhkonen I, Koivunen RM, Ruokonen A,
8. Morgan CL, Jenkins-Jones S, Currie CJ, Rees DA. Evaluation of adverse Tapanainen JS. Insulin sensitivity, insulin secretion, and metabolic and hor-
outcome in young women with polycystic ovary syndrome versus matched, monal parameters in healthy women and women with polycystic ovarian
reference controls: a retrospective, observational study. J Clin Endocrinol syndrome. Hum Reprod 2000;15:1266–74.
Metab 2012;97:3251–60. 29. Dos Reis R, Foss M, de Moura MD, Ferriani R, Silva de Sa M. Insulin secretion
9. Cheung L, Ma R, Lam P, Lok I, Haines C, So W, et al. Cardiovascular risks and in obese and non-obese women with polycystic ovary syndrome and its rela-
metabolic syndrome in Hong Kong Chinese women with polycystic ovary tionship with hyperandrogenism. Gynecol Endocrinol 1995;9:45–50.
syndrome. Hum Reprod 2008;23:1431–8. 30. Conway GS, Agrawal R, Betteridge D, Jacobs H. Risk factors for coronary ar-
10. Ollila M-M, West S, Kein€anen-Kiukaanniemi S, Jokelainen J, Auvinen J, tery disease in lean and obese women with the polycystic ovary syndrome.
Puukka K, et al. Overweight and obese but not normal weight women Clin Endocrinol (Oxf) 1992;37:119–25.
with PCOS are at increased risk of Type 2 diabetes mellitus—a prospective, 31. Dunaif A, Segal KR, Futterweit W, Dobrjansky A. Profound peripheral insulin
population-based cohort study. Hum Reprod 2017;32:423–31. resistance, independent of obesity, in polycystic ovary syndrome. Diabetes
11. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for sys- 1989;38:1165–74.
tematic reviews and meta-analyses: the PRISMA statement. J Clin Epidemiol 32. Çaglar GS, Kiseli M, Seker R, Ozdemir ED, Karadag D, Demirtas S. Athero-
2009;62:1006–12. genic dyslipidemia, subclinical atherosclerosis, non-alcoholic fatty liver dis-
12. Zawadski J. Diagnostic criteria for polycystic ovary syndrome, towards a ease and insulin resistance in polycystic ovarian syndrome/[Polikistik over
rational approach. Polycystic ovary syndrome, 1992:39–50. sendromunda ins€ ulin direnci, aterojenik dislipidemi, subklinik aterosklerozis
13. Rotterdam ESHRE/ASRM-Sponsored PCOS consensus workshop group. ve non-alkolik yag er hastalıg
lı karacig ı]. Turkish J Biochem 2015;40:24–30.
et al. Revised 2003 consensus on diagnostic criteria and long-term health 33. Alebic MS, Bulum T, Stojanovic N, Duvnjak L. Definition of insulin resistance
risks related to polycystic ovary syndrome (PCOS). Hum Reprod 2004;19: using the homeostasis model assessment (HOMA-IR) in IVF patients diag-
417. nosed with polycystic ovary syndrome (PCOS) according to the Rotterdam
14. Fauser BC, Tarlatzis BC, Rebar RW, Legro RS, Balen AH, Lobo R, et al. criteria. Endocrine 2014;47:625–30.
Consensus on women's health aspects of polycystic ovary syndrome 34. Pavicic Baldani D, Skrgatic L, Sprem Goldstajn M, Zlopasa G, Kralik Oguic S,
(PCOS): the Amsterdam ESHRE/ASRM-Sponsored 3rd PCOS Consensus Canic T, et al. Clinical and biochemical characteristics of polycystic ovary syn-
Workshop Group. Fertil Steril 2012;97:28–38.e25. drome in Croatian population. Coll Antropol 2012;36:1413–8.
15. Azziz R, Carmina E, Dewailly D, Diamanti-Kandarakis E, Escobar- 35. Blagojevic IP, Eror T, Pelivanovic J, Jelic S, Kotur-Stevuljevic J, Ignjatovic S.
Morreale HF, Futterweit W, et al. Positions statement: criteria for defining Women with polycystic ovary syndrome and risk of cardiovascular disease.
polycystic ovary syndrome as a predominantly hyperandrogenic syndrome: J Med Biochem 2017;36:259–69.
an Androgen Excess Society guideline. J Clin Endocrinol Metab 2006;91: 36. Joham AE, Boyle JA, Zoungas S, Teede HJ. Hypertension in reproductive-
4237–45. aged women with polycystic ovary syndrome and association with obesity.
16. World Health Organization. The Asia-Pacific perspective: redefining obesity Am J Hypertens 2014;28:847–51.
and its treatment. Sydney: Health Communications Australia; 2000 Feb. 37. Rocha MP, Marcondes JA, Barcellos CR, Hayashida SA, Curi DD, da
17. Wells G, Shea B, O'Connell D, Peterson J, Welch V, Losos M, et al. The Fonseca AM,^ et al. Dyslipidemia in women with polycystic ovary syndrome:
Newcastle-Ottawa Scale (NOS) for assessing the quality if nonrandomized incidence, pattern and predictors. Gynecol Endocrinol 2011;27:814–9.
studies in meta-analyses. 2009. Epub Available at: http://www.ohri.ca/pro 38. Sharma S, Majumdar A. Prevalence of metabolic syndrome in relation to
grams/clinical_epidemiology/oxford.htm [cited 2009 Oct 19] 2013. Ac- body mass index and polycystic ovarian syndrome in Indian women.
cessed January 23, 2018. J Hum Reprod Sci 2015;8:202.
18. Barry JA, Kuczmierczyk AR, Hardiman PJ. Anxiety and depression in polycy- 39. Panidis D, Tziomalos K, Macut D, Kandaraki EA, Tsourdi EA, Papadakis E,
stic ovary syndrome: a systematic review and meta-analysis. Hum Reprod et al. Age-and body mass index-related differences in the prevalence of
2011;26:2442–51. metabolic syndrome in women with polycystic ovary syndrome. Gynecol En-
19. Pan A, Sun Q, Okereke OI, Rexrode KM, Hu FB. Depression and risk of stroke docrinol 2013;29:926–30.
morbidity and mortality: a meta-analysis and systematic review. JAMA 2011; 40. Cussons AJ, Watts GF, Burke V, Shaw JE, Zimmet PZ, Stuckey BG. Cardiome-
306:1241–9. tabolic risk in polycystic ovary syndrome: a comparison of different ap-
20. Higgins JP, Green S. Cochrane handbook for systematic reviews of interven- proaches to defining the metabolic syndrome. Hum Reprod 2008;23:2352–8.
tions. John Wiley & Sons; 2011. 41. Chen ZJ, Zhao H, He L, Shi Y, Qin Y, Shi Y, et al. Genome-wide association
21. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test study identifies susceptibility loci for polycystic ovary syndrome on chromo-
for publication bias. Biometrics 1994;50:1088–101. some 2p16.3, 2p21 and 9q33.3. Nat Genet 2011;43:55–9.
22. Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected 42. Shi Y, Zhao H, Shi Y, Cao Y, Yang D, Li Z, et al. Genome-wide association
by a simple, graphical test. BMJ 1997;315:629–34. study identifies eight new risk loci for polycystic ovary syndrome. Nat Genet
23. Dahan M, Morris D. The impact of body mass index (BMI) on the incidence of 2012;44:1020–5.
type 2 diabetes (DMII) in women with and without polycystic ovary syn- 43. Ciaraldi TP, Aroda V, Mudaliar S, Chang RJ, Henry RR. Polycystic ovary syn-
drome (PCOS). Fertil Steril 2013;100:S348–9. drome is associated with tissue-specific differences in insulin resistance. J
24. Kim JJ, Choi YM, Cho YM, Jung HS, Chae SJ, Hwang KR, et al. Prevalence of Clin Endocrinol Metab 2009;94:157–63.
elevated glycated hemoglobin in women with polycystic ovary syndrome. 44. Hojlund K, Glintborg D, Andersen NR, Birk JB, Treebak JT, Frosig C, et al.
Hum Reprod 2012;27:1439–44. Impaired insulin-stimulated phosphorylation of Akt and AS160 in skeletal
25. Wang ET, Calderon-Margalit R, Cedars MI, Daviglus ML, Merkin SS, muscle of women with polycystic ovary syndrome is reversed by pioglitazone
Schreiner PJ, et al. Polycystic ovary syndrome and risk for long-term diabetes treatment. Diabetes 2008;57:357–66.
and dyslipidemia. Obstet Gynecol 2011;117:6. 45. Manneras L, Cajander S, Holmang A, Seleskovic Z, Lystig T, Lonn M, et al. A
26. Luque-Ramírez M, Alpan ~es M, Escobar-Morreale HF. The determinants of in- new rat model exhibiting both ovarian and metabolic characteristics of poly-
sulin sensitivity, b-cell function, and glucose tolerance are different in pa- cystic ovary syndrome. Endocrinology 2007;148:3781–91.

VOL. 111 NO. 1 / JANUARY 2019 175

ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY

46. Allemand MC, Irving BA, Asmann YW, Klaus KA, Tatpati L, Coddington CC, 57. Kumar A, Lawrence JC, Jung DY, Ko HJ, Keller SR, Kim JK, et al. Fat
et al. Effect of testosterone on insulin stimulated IRS1 Ser phosphorylation in cell-specific ablation of rictor in mice impairs insulin-regulated fat cell
primary rat myotubes—a potential model for PCOS-related insulin resis- and whole-body glucose and lipid metabolism. Diabetes 2010;59:
tance. PLoS One 2009;4:e4274. 1397–406.
47. Ciampelli M, Fulghesu AM, Cucinelli F, Pavone V, Caruso A, Mancuso S, 58. Weisberg SP, McCann D, Desai M, Rosenbaum M, Leibel RL, Ferrante AW.
et al. Heterogeneity in beta cell activity, hepatic insulin clearance and periph- Obesity is associated with macrophage accumulation in adipose tissue. J
eral insulin sensitivity in women with polycystic ovary syndrome. Hum Re- Clin Invest 2003;112:1796–808.
prod 1997;12:1897–901. 59. Yudkin JS, Stehouwer C, Emeis J, Coppack S. C-reactive protein in healthy
48. Amato MC, Vesco R, Vigneri E, Ciresi A, Giordano C. Hyperinsulinism and subjects: associations with obesity, insulin resistance, and endothelial
polycystic ovary syndrome (PCOS): role of insulin clearance. J Endocrinol dysfunction: a potential role for cytokines originating from adipose tissue?
Invest 2015;38:1319–26. Arterioscler Thromb Vasc Biol 1999;19:972–8.
49. Wild RA. Dyslipidemia in PCOS. Steroids 2012;77:295–9. 60. Hayes MG, Urbanek M, Ehrmann DA, Armstrong LL, Lee JY, Sisk R, et al.
50. Huang S, Qiao J, Li R, Wang L, Li M. Can serum apolipoprotein C-I demon- Genome-wide association of polycystic ovary syndrome implicates alter-
strate metabolic abnormality early in women with polycystic ovary syn- ations in gonadotropin secretion in European ancestry populations. Nat
drome? Fertil Steril 2010;94:205–10. Commun 2015;6:7502.
51. Wild RA, Carmina E, Diamanti-Kandarakis E, Dokras A, Escobar- 61. Garruti G, de Palo R, de Angelisc M. Weighing the impact of diet and lifestyle
Morreale HF, Futterweit W, et al. Assessment of cardiovascular risk and on female reproductive function. Curr Med Chem 2017;24:1. https://
prevention of cardiovascular disease in women with the polycystic ovary syn- doi.org/10.2174/0929867324666170518101008.
drome: a consensus statement by the Androgen Excess and Polycystic Ovary 62. Glintborg D, Mumm H, Hougaard D, Ravn P, Andersen M. Ethnic differences
Syndrome (AE-PCOS) Society. J Clin Endocr Metab 2010;95:2038–49. in Rotterdam criteria and metabolic risk factors in a multiethnic group of
52. Reaven GM. Role of insulin resistance in human disease. Diabetes 1988;37: women with PCOS studied in Denmark. Clin Endocrinol (Oxf) 2010;73:
1595–607. 732–8.
53. Samson SL, Garber AJ. Metabolic syndrome. Endocrinol Metab Clin North 63. Moran L, Teede H. Metabolic features of the reproductive pheno-
Am 2014;43:1–23. types of polycystic ovary syndrome. Hum Reprod Update 2009;15:
54. Alberti KGM, Zimmet P, Shaw J. The metabolic syndrome—a new world- 477–88.
wide definition. Lancet 2005;366:1059–62. 64. Anaforoglu I, Algun E, Incecayir O, Ersoy K. Higher metabolic risk with Na-
55. Grundy SM, Cleeman JI, Daniels SR, Donato KA, Eckel RH, Franklin BA, et al. tional Institutes of Health versus Rotterdam diagnostic criteria for polycystic
Diagnosis and management of the metabolic syndrome. Circulation 2005; ovarian syndrome in Turkish women. Metab Syndr Relat Disord 2011;9:
112:2735–52. 375–80.
56. Boucher J, Kahn CR. Differential roles of insulin and IGF-1 receptor in brown 65. Yildiz BO, Bozdag G, Yapici Z, Esinler I, Yarali H. Prevalence, phenotype and
and white adipose tissue and development of lipoatrophic diabetes. Dia- cardiometabolic risk of polycystic ovary syndrome under different diagnostic
betes 2013;62:A37. criteria. Hum Reprod 2012;27:3067–73.

176 VOL. 111 NO. 1 / JANUARY 2019

 Fertility and Sterility®

Alteraciones metabolicas en mujeres no obesas con síndrome de ovario poliquístico: Una revisi
on sistem
atica y meta-an
alisis
Objetivo: Comparar alteraciones metab
olicas entre mujeres no obesas con síndrome de ovario poliquístico (PCOS) y controles sanos.
~o: Revisi
Disen on sistematica y meta-analisis.
Entorno: No aplicable.
Paciente(s): mujeres no obesas con PCOS y controles sanos no obesas.
Intervencion(es): Ninguna.
Principales medidas de resultado: Prevalencia de alteraciones metab olicas incluyendo hiperinsulinemia, resistencia a la insulina (IR),
alteraci
on de la glucosa en ayunas (IFG), alteraci
on de la intolerancia a la glucosa (IGT), prediabetes, dislipidemia, hipercolesteloremia,
hipertrigliceridemia, niveles bajos de la lipoproteína de alta densidad (low-HDL), así como, otros resultados metab olicos como la dia-
betes mellitus tipo II (T2DM), hipertensi
on, síndrome metab olico (Mets), infarto de miocardio, apoplejía, accidente cerebrovascular, en-
fermedad arterial oclusiva, y enfermedad coronaria.
Resultados: Comparado con controles no obesas, las mujeres no obesas con PCOS mostraron una prevalencia mas alta de hiperinsu-
linemia (Odds Ratio [OR] 36.27; 95% intervalo de confianza[CI] 1.76–747.12), IR (OR, 5.70; 95% CI 1.46–22.32), IGT (OR, 3.42; 95% CI
1.56–7.52), T2DM (OR, 1.47; 95% CI 1.11–1.93), hipertrigliceridemia (OR, 10.46; 95% CI 1.39–78.56), low-HDL (OR, 4.03; 95% CI 1.26–
12.95), y Mets (OR, 2.57; 95% CI 1.30–5.07). No se observaron diferencias significativas para IFG, pre-DM, dislipidemia, hipercolester-
olemia ni hipertension. En el analisis de subgrupos, las mujeres blancas mostraron un riesgo incrementado de IR, IGT, IFG, T2DM,
hipertensi
on y Mets, mientras que en las mujeres asiaticas no se encontraron cambios metab olicos. Ning
un estudio ha informado es-
pecíficamente de la incidencia de infarto de miocardio, apoplejía, accidente cerebrovascular, enfermedad arterial oclusiva ni enferme-
dad coronaria en mujeres no obesas con PCOS.
Conclusion: Las mujeres no obesas con PCOS tambien sufren alteraciones metabolicas y tienen riesgo de complicaciones metabolicas a
largo plazo. Se deben realizar mas esfuerzos para aclarar los mecanismos subyacentes y las posibles intervenciones en la fase inicial.

VOL. 111 NO. 1 / JANUARY 2019 177