To assess circulating biochemical indices of endothelial function and nitro-oxidative stress in women with polycystic ovary syndrome (PCOS).
To assess circulating biochemical indices of endothelial function and nitro-oxidative stress in women with polycystic ovary syndrome (PCOS).
Seventeen women with PCOS and eighteen age- and body mass index-matched healthy volunteers.
Nitric oxide (NO) metabolite levels were assessed by chemiluminescence. Electron paramagnetic resonance spectroscopy with spin trapping was used to assess oxidative stress ex vivo and in vitro. Antioxidant capacity was measured using oxygen radical absorbance.
Biochemical indices of endothelial function, including NO metabolites, lipid-derived radicals and antioxidant capacity.
Plasma NO metabolites were similar in the two groups (nitrite: 257 ± 116 nmol/l [PCOS], 261 ± 135 nmol/l [controls] P = 0.93; nitrate: 27 ± 7 μmol/l [PCOS], 26 ± 6 μmol/l [controls] P = 0.89). Alkoxyl free radicals (lipid-derived) were detected as the dominant species, but levels were not different between women with PCOS and controls whether measured directly ex vivo (median 7.2 [range 0.17–16.73]e6 arbitrary units [a.u.] and 7.2 [1.7–11.9]e6 a.u., respectively, P = 0.57) or when stimulated in vitro to test radical generation capacity (1.23 [0.3–5.62]e7 a.u. and 1.1 [0.48–15.7]e7 a.u. respectively, P = 0.71). In regression analysis, visceral fat area was independently associated with in vitro oxidative potential (β = 0.6, P = 0.002). Total plasma antioxidant capacity (94 ± 30% [PCOS], 79 ± 24% [controls], P = 0.09) and plasma hydroperoxides (7.5 ± 4 μmol/l [PCOS], 6.7 ± 5 μmol/l [controls], P = 0.21) were not different between groups. However, lipophilic antioxidant capacity was lower in women with PCOS compared with controls (92 ± 32 and 125 ± 48%, respectively, P = 0.02).
Young overweight women with PCOS display a reduced lipophilic antioxidant capacity compared with healthy volunteers, but no change in circulating free radicals or nitro-oxidative stress.
Polycystic ovary syndrome (PCOS) is a common condition characterised by hyperandrogenism, oligo/anovulation and defects in insulin secretion and sensitivity, leading to an increased risk of type 2 diabetes. Patients also have an increased prevalence of hypertension, dyslipidaemia and premature atherosclerosis, although it is not yet clear whether this translates into an increased risk of cardiovascular mortality.
Endothelial dysfunction, an early marker of vascular disease, is a state linked to reduced nitric oxide (NO) bioavailability and increased oxidative stress. A recent meta-analysis of 21 studies comparing flow-mediated dilation, a noninvasive measure of endothelial function, in women with PCOS and control women, concluded that endothelial dysfunction was evident in women with PCOS even if they were young and non-obese. Few studies have examined NO bioavailability in women with PCOS but these have all used insensitive methodologies that fail to directly resolve total NO into its major component fractions, nitrite (NO2−) and nitrate (NO3−).[7-13] All failed to show any differences in total NO levels between women with PCOS and controls, although two studies did note an inverse relationship between total NO and fasting insulin,[7, 10] implying that insulin resistance may be mechanistically important in altering NO bioavailability in PCOS. Given that plasma nitrate is largely governed by dietary intake (approximately 75%) and plasma nitrite is reflective of endothelial NO bioavailability, analytical techniques based on colorimetric or fluorometric analysis do not provide sufficient sensitivity for plasma nitrite and may contribute to uncertainty regarding NO bioavailability in PCOS.
Oxidative stress, an imbalance arising from excess production of oxidants in the presence of reduced antioxidant capacity, can also reduce NO bioavailability and induce endothelial dysfunction. A recent meta-analysis found evidence of altered antioxidant capacity (increased superoxide dismutase and reduced glutathione levels) and indices of oxidative stress in women with PCOS compared with controls. However, the studies that underpinned the meta-analysis were generally limited to measurement of oxidant or antioxidant molecules in isolation and/or measurement of reaction end-products as surrogates of oxidative stress. Direct detection of reactive oxygen species (ROS) is challenging because of their potent reactivity but is essential if overestimation of oxidative burden is to be avoided. Hence, we sought to establish whether NO bioavailability and oxidant status are altered in a cohort of carefully characterised young women with PCOS who were free of overt cardiovascular disease, using sensitive, validated methodologies to directly assess plasma nitrite/nitrate, total antioxidant capacity and lipid-derived free radicals.
Women with PCOS (n = 17; age 16–45 years) were recruited from the endocrine clinic at the University Hospital of Wales (UHW). A diagnosis of PCOS was based on the Rotterdam criteria. Congenital adrenal hyperplasia, Cushing syndrome, hyperprolactinaemia, androgen-secreting tumours and thyroid disease were excluded by biochemical testing. Women were excluded from participation if they were pregnant, breastfeeding, had a history of hypertension, hyperlipidaemia or diabetes, or had a history of current or recent (within 3 months) use of anti-diabetics, lipid-lowering agents, antioxidant medication, anti-hypertensives and/or anti-androgens. Healthy volunteers (HVs, n = 18; age 16–45 years) were recruited from among medical students and staff within our institution. Volunteers had regular menstrual cycles (every 27–32 days). Their healthy state was established by history, physical examination and hormonal evaluation (thyroid function, prolactin, testosterone and 17-hydroxyprogesterone); those with features of hirsutism or a family history of PCOS were excluded.
Women attended our Clinical Research Facility at 08.00 hours after an overnight fast. Height, weight, hip and waist circumference were measured as per our published protocols. Abdominal subcutaneous and visceral fat areas were measured by computed tomography (CT; Hawkeye; GE Medical Systems, Hatfield, UK) on one cross-sectional scan obtained at the level of L4–L5. Scans were performed with women in the supine position using standard acquisition parameters (140 kV, 2.5 mA, 10-mm slice width, 13.6-s rotation time, 256-pixel matrix). The CT images were segmented into fat and non-fat areas according to our previously published protocols.
Fasting blood samples were drawn from an antecubital vein. Serum total cholesterol, high-density lipoprotein cholesterol and triglycerides were assayed using an Aeroset automated analyser (Abbott Diagnostics, Maidenhead, UK); low-density lipoprotein cholesterol was calculated using Friedewald's formula. Insulin was measured using an immunometric assay specific for human insulin (Invitron, Monmouth, UK) and glucose was measured using the Aeroset chemistry system (Abbott Diagnostics). High-sensitivity C-reactive protein was assayed by nephelometry (BN™ II system; Dade Behring, Milton Keynes, UK) and total testosterone was measured by liquid chromatography-tandem mass spectrometry (Quattro™ Premier XE triple quadrupole tandem mass spectrometer; Waters Ltd, Watford, UK). The intra-assay and inter-assay coefficients of variation were all <9%. After basal sampling, women underwent a standard 75-g oral glucose tolerance test. Glucose and insulin were measured at 0, 30, 60, 90 and 120 minutes. The area under the curve (AUC) for insulin and glucose was calculated using the trapezoid method. For the NO and oxidative stress measures, samples were promptly centrifuged (1024 g for 10 minutes at 4°C) to yield platelet-poor plasma. Ex vivo oxidative stress measurements (lipid-derived radicals) were taken immediately while the remaining samples were stored at −80°C until analysis.
The NO metabolites were assessed using ozone-based chemiluminescence, as previously described. Briefly, for plasma nitrite analysis, 5 ml tri-iodide reagent (I3−) was placed in a glass purge vessel and heated at 50°C via a water bath thermostatically controlled hotplate. The carrier gas (O2-free nitrogen) purging I3− was linked to a sodium hydroxide trap (1 mol/l), connected to an NO analyser (Sievers NOA 280i; Analytix, Boldon, UK). Samples (200 μl) were injected directly into the purge vessel through a rubber septum injection inlet. For nitrate analysis, plasma (15 μl) was injected into vanadium chloride heated at 90°C before detection via an NO analyser (Sievers NOA 280; Analytix). Results were compared with a sodium nitrite or sodium nitrate standard curve performed daily to account for temperature variation. Room temperature was 20 ± 2°C. In our hands the limit of sensitivity of these assays for plasma nitrite is >10 nmol/l and for plasma nitrate is >500 nmol/l, and the intra-assay coefficients of variation were <5 and <8%, respectively.
As a direct measure of ex vivo oxidative stress, we used electron paramagnetic resonance (EPR) spectroscopy coupled with spin trapping, using α-phenyl-N-tert-butyl nitrone (PBN) to detect free radicals in the circulation. Briefly, the nitrone-based spin trap on interaction with a free radical forms a stable spin adduct, detectable via EPR spectroscopy and exhibiting unique spectral characteristics. Blood samples were collected directly into a 6-ml EDTA vacutainer containing 2 ml PBN (200 mmol/l; Sigma-Aldrich, Poole, UK). Lipid-radical/PBN adducts underwent repeated toluene (Sigma-Aldrich) extraction followed by reconstitution in 100 μl chloroform (Sigma-Aldrich) before EPR analysis (e-scan; Bruker, Coventry, UK). The PBN/radical adducts were soluble in non-aqueous solvents, suggesting that the radical species were lipid-derived. To test in vitro lipid-radical formation potential ferrous sulphate (100 mmol/l, Sigma-Aldrich) was added to plasma and PBN (125 mmol/l, 1:1:1, volume:volume) before lipid extraction and EPR analysis as described above. Typical measurement conditions were: modulation amplitude 1.43 G, power 48.1 mW, time constant 40.96 seconds, sweep time 41.94 seconds. The EPR signals generated are proportional to radical amount, so peak height was used to reflect relative spin adduct concentration. As an additional biomarker of oxidative stress, total plasma hydroperoxide concentrations were determined using the ferrous iron/xylenol orange assay. Total plasma antioxidant capacity was assessed by oxygen radical absorbance capacity (ORACfl), whereas lipophilic antioxidant capacity was assessed using a modified extraction process as previously described. Briefly, fluorescein (10 nmol/l, Sigma-Aldrich) was used as a target for a free radical attack from 2,2′-azobis(2,4-amidinopropane)dihydrochloride (AAPH, 240 mmol/l), a thermally activated peroxyl-radical. ORAC measured the ability of plasma to buffer the AAPH insult. For lipophilic antioxidant capacity, 100 μl plasma was transferred to a glass tube followed by 200 μl ethanol (Fisher Scientific, Loughborough, UK), 100 μl water and 400 μl toluene, followed by mixing. The mixture was left to settle for 1–2 minutes until two phases appeared, followed by centrifugation (3300 g for 5 minutes at room temperature). The toluene phase was subsequently removed, and added to a separate amber tube. This step was repeated. The toluene extract was dried under nitrogen flow in preparation for lipophilic ORACfl analysis. The ORAC assay was carried out on a FLUOstar OPTIMA (BMG Labtech, Aylesbury, UK, ex: 485 nm, em: 527 nm, 37°C), as previously described. Samples were run in triplicate and results were expressed in arbitrary units compared with a standard antioxidant, tempo 100 μmol/l (Sigma-Aldrich) equivalent.
Data were analysed using SPSS version 18.0 (SPSS Inc., Chicago, IL, USA). The Kolmogorov–Smirnov test was used to check the data for normality. Analysis between groups was performed using the independent t-test or Mann–Whitney U-test for normally or non-normally distributed data, respectively. Age-adjusted correlation coefficients were used to explore the strength of the relationships between in vitro/ex vivo oxidation and testosterone, regional fat area and insulin sensitivity. Multiple regression analysis was performed to explore the dependence of in vitro and ex vivo oxidation on age, regional fat area, testosterone and insulin AUC. Results are expressed as mean ± SD or median (range). A P-value <0.05 was regarded as statistically significant. We based our sample size calculations on our previous data, which demonstrated a 0.28-fold shift in SD in lipid-derived radicals in women with type 2 diabetes compared with controls. To detect a similar shift in SD with >90% power at the 5% α level, we sought to recruit at least 14 women in each group. We considered a shift in SD of this magnitude as clinically significant, because this is associated with endothelial dysfunction.
Table 1 shows the baseline clinical and metabolic characteristics of the women with PCOS and healthy volunteers. There were no significant differences observed between the groups in age, body mass index (BMI), waist/hip circumference, subcutaneous/visceral/total fat area, lipids, high-sensitivity C-reactive protein or glucose AUC. As anticipated, insulin AUC and total testosterone were higher in women with PCOS.
|PCOS||Healthy volunteers||P value|
|Age (years)||31 ± 6||31 ± 7||0.9|
|Weight (kg)||78 ± 21||76 ± 15||0.68|
|BMI (kg/m2)||30 ± 6||29 ± 6||0.61|
|Waist circumference (cm)||91 ± 15||86 ± 13||0.31|
|Hip circumference (cm)||111 ± 16||106 ± 12||0.24|
|Visceral fat area (cm2)||31 ± 23||26 ± 14||0.46|
|Subcutaneous fat area (cm2)||287 ± 119||298 ± 114||0.78|
|Total fat area (cm2)||318 ± 133||324 ± 124||0.89|
|Testosterone (nmol/l)||1.4 ± 0.6||0.9 ± 0.6||0.02|
|hsCRP (mg/l)||1.25 (0.24–21.8)||0.9 (0.17–16.73)||0.73|
|Total cholesterol (mmol/l)||4.6 ± 1.3||4.8 ± 1.1||0.67|
|Triglycerides (mmol/l)||1.2 ± 1.4||1.0 ± 0.5||0.52|
|LDL cholesterol (mmol/l)||2.4 ± 1.4||2.5 ± 1.3||0.79|
|HDL cholesterol (mmol/l)||1.2 ± 0.5||1.3 ± 0.6||0.65|
|Insulin AUC||80 503 ± 46 700||52 681 ± 29 614||0.04|
|Glucose AUC||764 ± 217||692 ± 133||0.24|
No significant differences were found between women with PCOS and healthy volunteers in plasma nitrite (257 ± 116 nmol/l and 261 ± 135 nmol/l, respectively, P = 0.93) (Figure 2F) or nitrate (27 ± 7 and 26 ± 6 μmol/l, respectively, P = 0.89) (Figure 2G) concentration. These values are within the normal range that we have detected using similar methodologies in healthy women and several disease states.
Qualitatively, the spectral characteristics (hyperfine coupling constants) were used to specifically identify radical species against reference/standard values. The radicals trapped were identified as lipid-alkoxyl radicals (LO−, coupling constants αN = 14 Gauss (G), αH = 2.3 G) as previously described (Figure 1). The EPR spectra also revealed the presence of a PBN radical artefact (αN = 15 G; αH = 5 G). To control for possible experimental initiation of radical/PBN adducts, blood was collected directly into EDTA vacutainers containing 200 mmol/l PBN covered in foil, to account for transient extracellular leakage of iron and photolytic degradation of PBN. No species were detected from PBN alone (Figure S1). In addition, no species were detected when blood was taken into a syringe and transferred to an EDTA vacutainer containing PBN (Figure S1).
Quantitatively, circulating lipid-derived radicals were similar between women with PCOS (median 7.2 [range 0.17–16.73]e6 arbitrary units [a.u.]) and healthy volunteers (7.2 [1.7–11.9]e6 a.u, P = 0.57) (Figure 2A). Ferrous sulphate-oxidised plasma was used as a measure of in vitro lipid radical formation potential. No difference in susceptibility to form lipid-derived radicals was observed between women with PCOS and healthy volunteers (1.23 [0.3–5.62]e7 a.u. and 1.1 [0.48–15.7]e7 a.u., respectively, P = 0.71) (Figure 2B). Plasma hydroperoxide levels were also similar between women with PCOS and healthy volunteers (7.5 ± 4 and 6.7 ± 5 μmol/l, respectively, P = 0.21, Figure 2C).
Results reflect the ability of plasma (or selectively the lipophilic components of plasma) to buffer against AAPH, a thermally activated peroxyl-radical insult. Total plasma antioxidant capacity was similar between the PCOS group and healthy volunteers (94 ± 30 and 79 ± 24%, respectively, P = 0.09) (Figure 2D). However, women with PCOS displayed a reduced lipophilic antioxidant capacity in comparison to healthy volunteers (92 ± 32 and 125 ± 48%, respectively, P = 0.02) (Figure 2E).
In women with PCOS, after adjustment for age, in vitro oxidative capacity correlated moderately with testosterone (r = 0.64, P = 0.07) and insulin AUC (r = 0.51, P = 0.04), and strongly with visceral fat area (r = 0.76, P = 001). No significant relationships were noted with ex vivo radical generation in women with PCOS. In control women, in vitro oxidative capacity correlated moderately with subcutaneous fat area (r = 0.53, P = 0.03) and visceral fat area (r = 0.58, P = 0.01), and negatively with testosterone (r = –0.55, P = 0.02). Ex vivo radical generation correlated moderately with visceral fat area (r = 0.49, P = 0.046). When women with PCOS and control women were analysed together, in vitro oxidative capacity correlated moderately with insulin AUC (r = 0.42, P = 0.01) and strongly with visceral fat area (r = 0.72, P < 0.001) but in multiple linear regression analysis, only visceral fat area remained significant in the model (β = 0.6, P = 0.002).
Previous studies have measured indices of global NO metabolism and oxidant status in women with PCOS with mixed results. To our knowledge this is the first study to specifically assess plasma nitrite (reflecting endothelial NO bioavailability), and directly measure free radical formation in blood in women with PCOS using a series of sensitive, reference standard methodologies. We were unable to find any evidence for altered NO bioavailability or metabolism in our patients, other than that women with PCOS did show reduced lipophilic antioxidant capacity compared with healthy volunteers.
Our study is the first to employ a series of sensitive methodologies to specifically assess plasma NO metabolites and directly detect free radical formation in the circulation of women with PCOS. We also undertook a detailed metabolic and anthropometric characterisation so as to understand the relationships of NO/oxidative status with insulin sensitivity and body fat distribution. However, although we used the well-established Rotterdam criteria to classify our PCOS population, this may be associated with a less severe metabolic phenotype than other definitions of the syndrome. Hence it is possible that women defined by National Institutes of Health or Androgen Excess Society criteria might have more significant disturbances in cardiovascular homeostasis and oxidative burden. Our cohort was also relatively young, and it is tempting to speculate that disturbances in vascular NO metabolism and oxidative stress may not emerge until later in the disease course, in line with findings from studies of carotid intima media thickness where meaningful differences in atheroma burden were not apparent until middle age. Long-term dietary influences on antioxidant status are also difficult to adjust for, but we sought to minimise these effects by asking women to abstain from antioxidant medication before participation.
Plasma nitrite and nitrate levels were unaltered in women with PCOS compared with age- and BMI-matched controls. These findings are consistent with previous reports[7-13] but contrast with observations from many studies of endothelial function in women with PCOS, which have shown reduced flow-mediated dilation compared with healthy volunteers. These discrepancies may relate to the difficulties in measuring local, endothelial-derived nitric oxide, which has an extremely short circulatory half-life—estimated at <1 second. Researchers are therefore reliant on measurement of nitrite and nitrate, the major metabolites of NO generated by stepwise oxidation, which are widely used as an index of endothelial NO synthase activity.[22, 23] However, plasma nitrite and particularly nitrate reflect not only endogenous NO production but also dietary nitrate ingestion. We were careful to minimise the influence of dietary variation by fasting women overnight before measurement; however, we cannot entirely rule out the possibility that dietary factors might have impacted upon our findings. Furthermore, in some instances plasma NO status may not reflect tissue NO status. Notwithstanding these limitations, our observations do not support a major alteration in NO bioavailability in women with PCOS, with values measured in the control and PCOS group similar to those measured across a broad range of healthy women.
Oxidative stress was also unaltered in the women with PCOS compared with healthy volunteers. This is in contrast to most previous studies in women with PCOS, which have shown an increased oxidative burden but have relied upon measurement of reaction end-products as biomarkers of damage to lipids and proteins. These measures are associated with a complex biochemistry where the different thermodynamic and kinetic properties may contribute to overestimation of oxidative burden and inconsistent results. To our knowledge, our study is the only one to directly assess oxidative burden by EPR spectroscopy, the only analytical technique capable of direct detection of free radical species. Analysis of EPR spectra identified alkoxyl free radicals (LO−) as the dominant species present in the circulation of both groups of women. Previous studies have suggested that alkoxyl radicals evolve during a reaction catalysed by Fe2+ reductive decomposition of extracellular lipid hydroperoxides formed subsequent to primary radical-mediated damage to membrane phospholipids. Lipid-derived radical concentrations, whether measured directly ex vivo or in vitro, and plasma hydroperoxide levels were not different in women with PCOS compared with controls. However, the strongest relationships of oxidative burden were with visceral fat area and insulin AUC, implying that oxidative stress is linked predominantly with central adiposity and insulin resistance rather than PCOS per se. This may reflect accelerated adipocyte lipolysis in the pro-inflammatory obese state, leading to increased nonesterified fatty acids and subsequently increased ROS generation in mononuclear cells.
We found that lipophilic but not total antioxidant capacity was reduced in women with PCOS compared with healthy volunteers. Previous studies have noted reduced total lipophilic antioxidant capacity in women with PCOS, independent of BMI or insulin resistance.[27, 28] In contrast, a meta-analysis of six studies of total antioxidant capacity, including 470 women, found no significant difference between women with PCOS and controls. This is in agreement with our findings, although we noted a trend towards an increased total antioxidant capacity in women with PCOS, which did not quite achieve statistical significance, suggesting that antioxidant activity in the hydrophilic compartment may have undergone a compensatory increase to maintain homeostasis. The antioxidant capacity of the aqueous compartment is accounted for by proteins such as albumin and ascorbic acid whereas fat-soluble antioxidants such as carotenoids and α-tocopherol are located in the lipoprotein core. Individual carotenoids may reflect dietary intake of fruits and vegetables, whereas plasma tocopherol concentrations correlate with vitamin E intake. Circulating vitamin E concentrations may be lower in women with PCOS compared with controls but vitamin A and β-carotene levels appear to be unchanged.
A depletion of antioxidant defences, accompanied by increased ROS production, is a hallmark of other diseases characterised by insulin resistance, notably type 2 diabetes. This redox imbalance leads to increased production of free radicals such as superoxide, which promote vascular smooth muscle contraction via inhibition of endothelial-dependent relaxation. Lipophilic antioxidant capacity may be especially important in regulating oxidation of low-density lipoprotein cholesterol. Oxidation of low-density lipoprotein induces an inflammatory cascade and binding to scavenger receptors on the surface of macrophages leading to foam cell generation and atherosclerotic plaque formation. Although we did not measure oxidised low-density lipoprotein, others have found that circulating concentrations are increased in women with PCOS[30, 31] although this is not a consistent finding.
Further studies are needed to establish the causes and consequences of altered antioxidant capacity in women with PCOS but in the meantime our study confirms that there is little evidence of abnormal NO/oxidative metabolism in young, overweight women with PCOS. This is in agreement with our recent findings from a large population-based study in which we found no evidence for an increased incidence of cardiovascular events in young women with PCOS. Although these data may reassure clinicians treating young women with PCOS that cardiovascular risk is not increased at this age, young women with PCOS are at increased risk of type 2 diabetes, which is worsened by weight gain. As we demonstrated a strong association of visceral fat with oxidative capacity, we speculate that weight loss may be the most important measure in reducing oxidative stress and cardiometabolic risk in women with PCOS, although further trials are still needed to confirm if this is the case.
GRW, MU, WDE, HLB, PEJ and DAR have no conflicts of interest to declare.
GRW contributed to study design, analysis, data collection, writing of manuscript and final approval. PEJ and DAR contributed to study design, data collection, writing of manuscript and final approval. MU, WDE and HLB contributed to data collection, and final approval.
The study was approved by Cardiff University (study sponsors), Cardiff & Vale University Health Board and the South East Wales research ethics committee. All women gave written informed consent before study commencement (date awarded: 26 September 2008; ethical approval: ref. 08/WSE04/53).
Our research was funded by grants from The Wales Heart Research Institute and Mrs John Nixon Scholarship.