Dr Grant D Brinkworth, CSIRO—Human Nutrition, PO Box 10041, BC Adelaide, South Australia 5000, Australia. Email firstname.lastname@example.org
Objective There remains a large degree of disagreement about the association of polycystic ovary syndrome (PCOS) with impaired endothelial dysfunction and cardiovascular disease (CVD) risk. The purpose of this study was to determine whether overweight and obese women with PCOS have impaired endothelial function compared with weight-matched controls without PCOS and whether endothelial function is associated with cardiovascular risk markers and hormonal parameters.
Design Cross-sectional analysis.
Setting An outpatient trial at the Commonwealth Scientific Industrial Research Organisation Clinical Research Unit.
Population Overweight and obese women with PCOS (n= 12) and weight-matched controls without PCOS (n= 10).
Methods Endothelial function, cardiovascular risk markers and hormonal parameters were assessed in the patients.
Main outcome measures Endothelial function was assessed by flow-mediated dilatation (FMD) of the brachial artery using high-resolution ultrasound. Lipid profile, fasting insulin level, glucose level, insulin resistance, C-reactive protein level, folate level, Vitamin B12 level and hormonal parameters.
Results Women with PCOS had significantly higher testosterone levels (P < 0.001) and free androgen index (P= 0.006) compared with the controls without PCOS. Both groups were normoinsulinaemic, and there were no significant differences in any of the markers of CVD between women with and without PCOS. Furthermore, FMD was similar in both groups (PCOS 6.1 ± 1.2% versus control 5.6 ± 1.0%, P= 0.77).
Conclusions Compared with a group of weight-matched women with similar metabolic profiles, normoinsulinemic, overweight and obese women with PCOS did not show any greater impairment in endothelial function assessed by FMD. A normoinsulinemic phenotype of PCOS with low metabolic risk factors may reduce the risk of endothelial dysfunction in overweight and obese women with this syndrome. Further studies are required that directly compare FMD in normoinsulinemic and hyperinsulinaemic women with PCOS.
Polycystic ovary syndrome (PCOS), characterised by menstrual dysfunction, anovulation and hyperandrogenism, is a common reproductive endocrine disorder,1 affecting 5–10% of women of reproductive age.2,3 Most women with PCOS also show an adverse cardiovascular risk profile, with an increased incidence of components of the metabolic syndrome, including insulin resistance, hypertension, dyslipidaemia, impaired glucose tolerance and type II diabetes.4–8 This women group may therefore have a higher risk of coronary artery disease.9–12 However, long-term follow-up data examining the effects of PCOS on cardiovascular morbidity and mortality are conflicting.13–15
Endothelial dysfunction plays a key role in the development and progression of atherosclerosis and has shown to be an early risk marker of atherosclerosis and an independent predictor of future cardiac events.16 Measurement of postischaemic flow-mediated dilatation (FMD) of the brachial artery is an established method to assess endothelial function of conduit arteries.17 Endothelial dysfunction, demonstrated as reduced FMD, is present in a number of states including hypercholesterolaemia,18 diabetes,19 obesity and insulin resistance.20,21 Despite this, studies examining endothelial function in women with PCOS have yielded inconsistent results. Mather et al.22 showed that women with PCOS had normal endothelial function despite marked insulin resistance and hyperandrogenism. Similarly, Bickerton et al.23 using a venous occlusion plethysmography technique to assess differences in reactive hyperaemia of the forearm microcirculation found no evidence of endothelial dysfunction in healthy women with PCOS compared with age- and weight-matched controls without PCOS. In contrast, Paradisi et al.24 showed that impaired endothelium-dependent vasodilatation related to both androgen levels and insulin resistance in obese women with PCOS compared with an age- and weight-matched control group of women without PCOS. Kravariti et al.25 also showed that asymptomatic women with PCOS have impaired FMD at an early age that was associated with insulin resistance, total testosterone level and total cholesterol level. In support, three additional studies conducted in young women with normal weight showed impaired endothelial structure and function in women with PCOS that was correlated with insulin resistance.26–28 The exact reason for the differences between these previous studies is not entirely clear. Hence, some doubt remains about the effects of PCOS as a risk factor for endothelial dysfunction and cardiovascular disease (CVD).
Although not absolute, to date, most evidence in PCOS has reported a close association between endothelial dysfunction with insulin resistance and hyperandrogenaemia to a less consistent extent. Collectively, this suggests that insulin resistance, rather than PCOS per se, is the major casual factor of endothelial dysfunction in women with PCOS. Due to great diversity of symptoms, signs, metabolic aberrations and hormonal profiles among women with PCOS, the presence of insulin resistance and degree of hyperandrogenism may have impact on the presence of endothelial dysfunction. Therefore, women with PCOS with normoinsulinaemia or low levels of insulin resistance could have normal endothelial function compared with women without PCOS. However, as a greater body of research becomes available and the effects of PCOS and its related metabolic consequences on vascular functioning become better characterised, this will assist to unravel the associated effects of this disorder on the risk for CVD.
To our knowledge, there is a lack of data comparing endothelial function in normoinsulinaemic women with and without PCOS. Therefore, the purpose of this study was to compare endothelial function assessed by brachial artery ultrasound in a group of overweight and obese women with and without PCOS who were largely normoinsulinaemic and to evaluate the relationship between endothelial function, hyperandrogenism, insulin resistance and other cardiovascular risk markers.
Materials and methods
Twelve women with PCOS and ten normal controls, matched for body weight were recruited for the study (Table 1). All subjects were overweight and obese [body mass index (BMI), 27–44 kg/m2], premenopausal women, aged 23–43 years and had been weight stable (<2.0 kg weight change) for at least 3 months prior to enrolment. Presence of PCOS was diagnosed according to the Rotterdam consensus group29 when at least two of the following three criteria were present: oligo or amenorrhoea; hyperandrogenism including elevated serum concentrations of testosterone (testosterone >2.8 nmol/l) and/or a hirsutism score of more than 8 (according to Ferriman–Gallwey)30 or positive ultrasound presentation of polycystic ovaries by transvaginal scan defined as the presence of 12 or more follicles in each ovary measuring 2–9 mm in diameter and/or increased ovarian volume of >10 ml. Exclusion criteria included cancer, liver disease, renal disease, haematological disease, CVD, diabetes, Cushing syndrome, androgen-secreting tumours, late-onset 21-hydroxylase deficiency, thyroid dysfunction, hyperprolactinaemia or pregnancy. All women provided written informed consent, and the study protocols and procedures were approved by the Human Ethics Committee of the Commonwealth Scientific Industrial Research Organisation (CSIRO).
Table 1. Baseline characteristics and hormone profiles of women with PCOS and controls without PCOS
PCOS (n= 12)
Control (n= 10)
FAI, Free androgen index.
Values are represented as means ± SE.
31.9 ± 1.8
37.2 ± 1.7
Body weight (kg)
95.6 ± 4.5
92.5 ± 4.9
36.2 ± 1.7
34.4 ± 1.5
Fat-free mass (kg)
60.8 ± 2.6
59.6 ± 3.0
Fat mass (kg)
34.6 ± 2.1
34.0 ± 2.5
Percent body fat
36.1 ± 2.3
36.1 ± 2.6
Waist circumference (cm)
113.6 ± 3.6
107.6 ± 2.9
3.4 ± 0.2
2.1 ± 0.2
Sex-hormone-binding globulin (nmol/l)
25.1 ± 3.4
25.6 ± 1.9
14.5 ± 2.6
8.2 ± 1.2
Vitamin B12 (pmol/l)
280.3 ± 32.7
271.8 ± 39.1
16.7 ± 2.4
17.6 ± 2.5
After an overnight fast, subjects attended the CSIRO Clinical Research Unit during which height, body weight, body composition and waist circumference were measured prior to a venous blood sample being drawn for determination of blood lipids, glucose, insulin, C-reactive protein (CRP), folate and vitamin B12. Endothelial function was then assessed in each subject by FMD of the brachial artery. All subjects were asked to refrain from consuming alcohol and participating in vigorous physical activity for at least 24 hours before assessment.
Anthropometric measures and body composition
Height was measured to the nearest 0.1 cm using a stadiometer (SECA, Hamburg, Germany) with subjects in the free-standing position. Body weight was measured to the nearest 0.05 kg, with subjects wearing light clothing and no shoes, using calibrated electronic digital scales (A&D Mercury, Model AMZ 14, Tokyo, Japan). Waist circumference (minimal) was measured in duplicate. Fat mass, lean mass and percent body fat were measured by bioelectrical impedance analysis (ImpDF50; Impedimed Pty Ltd, Eight Miles Plains, Australia).
Assessment of endothelial function by brachial ultrasound
Endothelial function was assessed by FMD (vascular reactivity) using brachial artery ultrasound, with subjects lying supine in a quiet, temperature-controlled room (22–25°C). A 7.5-MHz linear array ultrasound transducer of an Acuson Aspen ultrasound (Siemens, Mountain View, CA, USA) was positioned perpendicular to the long axis of the artery of the dominant arm just above the antecubital fossa. The diameter of the artery was measured from two-dimensional B-mode ultrasound images of the centre of the brachial artery, identified when the clearest picture of the anterior and posterior intimal layers was obtained, with the focus set to the depth of the near wall. The depth, gain and persistence settings were set to optimise images of the lumen–arterial wall interface and magnified using a resolution box function. After a baseline image was obtained, for the production of reactive hyperaemia, a sphygmomanometer cuff was placed around the mid-point of the dominant forearm (i.e. distal to the scanned part of the artery) and inflated to a pressure of 250 mmHg for 5 minutes. The cuff was then rapidly deflated and the brachial arterial diameter scanned every 15 seconds for 3 minutes.
All images were recorded for subsequent off-line analysis. FMD of the brachial artery was expressed as the maximal percent change in the arterial diameter from baseline to 3 minutes after cuff deflation. Heart rate was recorded continuously using an electrocardiogram, and the scans were read at the end of diastole, coincidentally with the R-wave. All assessments were performed by the same operator, who was blinded to hormonal status of the women. Based on assessments performed on 2 separate days in nine subjects by this operator, the within-subject coefficient of variation of the endothelium-dependent response was 8.4%, which is similar to that reported in other laboratories.25,28,31
Sex-hormone-binding globulin and total testosterone levels were measured as described previously.32 Serum lipids, glucose and CRP concentrations were measured in one run on a Roche Hitachi analyser using standard Roche enzymatic kits (Roche Diagnostics Co., Indianapolis, IN, USA). A modified Friedewald equation was used to calculate low-density lipoprotein cholesterol.33 Insulin was determined in duplicate using a commercially available enzyme immunoassay kit (Mercodia, Uppsala, Sweden). The homeostatic model assessment of insulin resistance (HOMA-R) was used as a surrogate measure of insulin sensitivity based on fasting glucose and insulin concentrations, which is calculated as [fasting serum insulin level (milliunits/l) × fasting glucose level (mmol/l)]/22.5.34 Normoinsulinaemia was defined as a HOMA score of <3.9 based on a large population of women with PCOS.35 Serum concentrations of Vitamin B12 and folate were measured in a certified commercial laboratory (Institute of Medical and Veterinary Science, Adelaide, South Australia, Australia).
Statistical analyses were performed by SPSS for windows 11.5.0 (SPSS, Chicago, IL, USA). Normally distributed data are presented as means ± standard error (SE). For non-normally distributed variables (e.g. glucose, insulin and HOMA score), the logarithm of each was used for parametric statistical analyses and geometric means ± SE presented. Group differences were determined using an unpaired Student’s t tests and analysis of covariance to control for the effects of age. Pearson’s correlation coefficients were calculated to determine relationships between the variables examined. The level of statistical significance was set at P≤ 0.05.
By design, the groups were well matched for body weight, and in addition to BMI, waist circumference and body composition were similar in both groups (Table 1). As expected, the levels of testosterone and free androgen index (FAI) were significantly higher in the women with PCOS than in the controls without PCOS. All women with PCOS had clear diagnosis confirmation through the presence of menstrual irregularity and hyperandrogenism.
Cardiovascular risk markers and endothelial function
Table 2 shows the results of the biochemical markers of CVD risk. There were no significant differences between the two groups for lipid fractions, fasting glucose, insulin, CRP or insulin resistance as determined by HOMA. Brachial artery diameter between the groups was not significantly different (PCOS 0.42 ± 0.02 mm versus control 0.39 ± 0.02 mm, P= 0.15). FMD in response to reactive hyperaemia was also similar between the groups (PCOS 6.1 ± 1.2% versus control 5.6 ± 1.0%, P= 0.77; Figure 1).
Table 2. Metabolic markers of cardiovascular risk in women with PCOS and controls without PCOS
PCOS (n= 12)
Control (n= 10)
HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; HOMA-R, homeostatic model assessment of insulin resistance, CRP, C-reactive protein.
Values are represented as means ± SE.
1.8 ± 0.5
1.4 ± 0.2
Total cholesterol (mmol/l)
4.9 ± 0.3
5.0 ± 0.2
1.4 ± 0.1
1.4 ± 0.1
2.5 ± 0.3
3.0 ± 0.2
Fasting glucose (mmol/l)
5.3 ± 0.3
5.2 ± 0.2
Fasting insulin (milliunits/l)
12.6 ± 4.0
9.2 ± 2.1
2.9 ± 1.0
1.7 ± 0.5
5.9 ± 1.0
4.3 ± 2.6
The results of the analyses examining correlations of various anthropometric, metabolic and hormonal parameters with FMD for all subjects are shown in Table 3. A significant correlation between FMD and age was evident (r=−0.52, P= 0.01) but not between any other factor including BMI, waist circumference, percent body fat, fasting glucose, insulin resistance, blood lipid subfractions, CRP or any hormonal parameters. After controlling for age, FMD was still not different between the women with PCOS and the controls without PCOS (P= 0.37).
Table 3. Correlational analyses of various anthropometric, metabolic and hormonal parameters with brachial FMD response in all subjects
In this study, we showed no effect of PCOS on endothelial function measured by brachial artery FMD. We also showed no differences in a number of established CVD risk markers in overweight and obese women with and without PCOS or correlations between these variables and the endothelial function.
Our principal result is similar to that reported in two previous studies.22,23 Mather et al.22 using the same brachial artery FMD measure observed no differences in endothelial function in obese women with PCOS compared with age-matched but not weight-matched controls without PCOS. Similarly, Bickerton et al.23 who assessed endothelial function using a standard venous occlusion plethysmography technique to measure reactive hyperaemic forearm blood flow showed similar responses in age- and weight-matched women with and without PCOS. In contrast, Paradisi et al.24 reported diminished leg blood flow responses to the vasodilator methacholine chloride in obese women with PCOS compared with age- and weight-matched controls without PCOS, which was related to androgen levels. Meyer et al.26,36 showed an increased endothelial dysfunction (assessed by brachial artery FMD) in overweight and obese women with PCOS, which was related to insulin resistance. Other independent studies have also reported impaired brachial artery FMD in both normal-weight and overweight young women with PCOS, which was strongly associated with insulin resistance and hyperandrogenism to a lesser extent.25,27,28,37 A limitation of our study is the relatively small sample size examined. Nevertheless, there was still sufficient power (85%, P < 0.05) to detect a minimum difference in FMD of 4% between the PCOS and the control group, which is the approximate effect size repeatedly observed in previous studies.25,26,37 This suggests that because there was sufficient power to detect a differential effect size that could have been expected with PCOS, that other reasons maybe responsible for the discrepant findings between our’s and previous studies. However, these remain unclear at this time.
Unlike previous studies, the two groups examined in the present investigation were not precisely matched for age. An age-related decline in brachial artery FMD, including women with PCOS has been reported,26,38,39 which was confirmed by our data, i.e. the higher the age, the more impaired was FMD. It might be argued that this may have resulted in a greater impairment of FMD due to age-related factors in the control group. However, after adjustment of age, there was still no significant difference in FMD observed between subjects with and without PCOS, suggesting that our results were only weakly affected by this variable. Furthermore, based on data obtained from a large observational study,40 in our subjects, on average, the 5-year age advancement in the control group was most likely to have only caused a greater reduction in FMD of ∼1%. Therefore, although a confounding effect of age bias between the two groups cannot be entirely dismissed, based on these data, it would be considered too small to have fully masked any observable effect that could have been expected with PCOS if adjustments for age were made. Nevertheless, compared with other studies that report endothelial dysfunction with PCOS,25–28 the mean age of subjects evaluated in our experiment and previous investigations that have not detected any effect of PCOS on endothelial function22,23 were older (aged 30 versus 20 years). It is therefore still possible that the, overall, older age of subjects studied in the latter investigations may have mitigated any effects between subjects with and without PCOS that could explain some of the variation between the studies. Such an effect would suggest the possibility that PCOS-related changes in endothelial function maybe more apparent in younger women.
Many previous studies have documented that PCOS is associated with an increased risk of dyslipidaemia and insulin resistance,4–8 which are metabolic disturbances integrally associated with endothelial dysfunction.17,41,42 Inverse correlations between FMD and fasting insulin levels and surrogate parameters of insulin resistance have been reported in PCOS25,27,28,37 and other women groups including type II diabetes and the metabolic syndrome.19,43 Clinical trials have further shown that impairments of endothelial dysfunction in PCOS are reversed by metformin therapy, and troglitazone treatment associated with a decrease in insulin resistance.37,44 Collectively, this highlights the importance of insulin resistance in endothelial dysfunction in PCOS and suggests that insulin resistance is the major causal factor in the pathophysiology of metabolic risk factors rather than the presence of PCOS per se. In this study, there was an absence of difference between the groups for any of the surrogate CVD risk markers assessed, including insulin resistance. In fact, subjects in our study on a whole showed metabolic profiles that were relatively normal45–47 and normoinsulinaemic according to HOMA (insulin resistance surrogate) based on previously defined criteria in PCOS.35 Moreover, we did not find any correlations between FMD and either insulin resistance or fasting insulin levels; but this could have been caused by the small study group or that the relatively low range of FMD and insulin values observed were not sensitive enough to detect correlations. Although insulin resistance and metabolic aberrations are common in overweight women with PCOS, they are not universal,35,48 suggesting that PCOS is a heterogeneous disorder with different phenotypes that may represent groups with varying risk of CVD risk. In the present investigation, the study of a population that consisted predominately of women with a phenotype of low insulin resistance and CVD risk markers could therefore be a likely explanation for the similar metabolic profile between the study groups, which translated to equivocal FMD response. Similarly, in the study by Mather et al.22 who also found no evidence of impaired FMD in women with PCOS, despite marked insulin resistance, showed that plasma high-density lipoprotein levels in women with PCOS did not differ from those in controls without PCOS, indicating that the level of insulin resistance may have been somewhat less than that in subjects reported in previous studies. Based on this evidence, the divergence of the findings in others and our study are most likely related to the characteristics of the population studied, such that endothelial dysfunction maybe more prevalent in PCOS phenotypes that show a high degree of insulin resistance and an adverse metabolic profile. However, the role of insulin resistance on vascular health in PCOS to some extent remains unclear and is an area requiring further clarification. Future studies should specifically compare endothelial function in hyperinsulinaemic and normoinsulinaemic women with PCOS. Furthermore to this point, recent investigations have determined metabolic disparity in women with PCOS between different ethnic groups.48,49 For instance, Wijeyaratne et al.49 showed that South Asian women with PCOS show more severe metabolic disturbance and insulin resistance than Caucasian women of European descent. This suggests an additional need to study differences in endothelial function and CVD risk in women with PCOS based on ethnicity.
A separate line of evidence suggests that sex hormones may also have important effects on CVD and endothelial function. In previous studies, increase in CVD in postmenopausal women with elevated testosterone levels was reported,50,51 although there is conflicting evidence suggesting an opposing effect.52 In PCOS, an inverse relationship between FMD and endothelial function and testosterone levels has been reported,24,25,37 suggesting an impact of androgens on the endothelium and atherogenesis. However, this relationship does not seem universal, with a number of other studies,22,26 including ours showing no association between these variables. In fact, testosterone therapy has been shown to improve endothelial function in postmenopausal women.53 A recently published study in a large group of women with PCOS showed a positive association between FAI and increased FMD,26 suggesting that hyperandrogenism may have a protective effect against metabolic disturbances that promote endothelial dysfunction. Further research is still required to establish the relationship between hormones and their effects on endothelial function and cardiovascular risk in PCOS.
A vascular protective role of folic acid and B-vitamins has been reported previously.54 In this study, mean serum concentrations of vitamin B12 and folate were comparable between the PCOS and the control groups. In agreement with us, another study reported similar values and no differences for these variables between women with PCOS and weight-matched controls without PCOS.55
CRP is an inflammatory molecule that has been directly implicated in the atherosclerotic process and identified as an independent biomarker of CVD risk and future cardiac events.56 No differences in CRP concentrations were observed between the women with PCOS and controls without PCOS. Our values are consistent with those previously reported in women of similar age and weight, which also showed no effect of PCOS on this measure.23,26 In contrast, a large study has shown CRP to be higher among both lean and obese women with PCOS than among controls.57 However, the results of this study was limited by the lack of insulin sensitivity measurement, given that another report has shown that CRP is not different between a PCOS and a control group after adjustment for insulin sensitivity.58 This suggests that PCOS per se is not directly associated with an inflammatory state.
In conclusion, we found no evidence of endothelial dysfunction or increased cardiovascular risk markers in normoinsulinaemic, overweight and obese women with PCOS compared with a weight-matched control group that show a similar metabolic profile. At this time, the limited studies available have shown no consistent demonstration of an increased incidence of CVD in PCOS.14,15 However, based on the current level of evidence, and the existence of metabolically diverse phenotypes, unequivocal statements regarding the clinical implications of PCOS on endothelial function and its subsequent effects on CVD risk cannot be made. Further, long-term follow-up studies are necessary to establish the CVD risks associated with PCOS and metabolic phenotypes that exist within this syndrome.
We like to thank the research support staff of the Clinical Research Unit at CSIRO—Human Nutrition for assistance in organising and performing various tests for this trial.