Polycystic ovary syndrome (PCOS) is a heterogeneous clinical syndrome among women of reproductive age. National Institutes of Health consensus criteria for the diagnosis of PCOS requires menstrual irregularities attributable to anovulation, either biochemical or clinical evidence of hyperandrogenism, and the exclusion of other diagnoses.1 Although PCOS is associated with an adverse cardiovascular risk profile, such as obesity, insulin resistance, type 2 diabetes mellitus, dyslipidemia, and hypertension in cross-sectional studies2–5 and natural history studies,6,7 questions remain about the independent association of PCOS with incident cardiovascular disease in longitudinal studies. Establishing this relationship is challenging because of the spontaneous resolution of PCOS symptoms over time8,9 and the higher body mass index (BMI, calculated as weight (kg)/[height (m)]2) associated with PCOS4 that may mediate cardiovascular disease risk. The determination of PCOS as an independent risk factor for cardiovascular disease may lead to clearer guidelines for cardiovascular disease prevention in these women.
We used data from the Coronary Artery Risk Development in Young Adults study, which is a large, established cohort of young African-American and white adults observed for 20 years, to investigate the association of PCOS and the subsequent development of cardiovascular risk factors. We hypothesized that women who fulfilled National Institutes of Health criteria for PCOS between ages 20 and 32 years would be at higher risk for subsequent development of incident diabetes, dyslipidemia, and hypertension. Additionally, we hypothesized that normal-weight women with PCOS would have the same degree of cardiovascular risk as overweight women with PCOS, and that the persistence of PCOS symptomatology over time would be associated with increased cardiovascular risk.
MATERIALS AND METHODS
The Coronary Artery Risk Development in Young Adults study is a prospective investigation of cardiovascular risk factors in a U.S. population of African-American and white young adults.10 The study enrolled 5,115 men and women aged 18–30 years at baseline in 1985–1986, who were recruited from four cities (Birmingham, Alabama; Chicago, Illinois; Minneapolis, Minnesota; and Oakland, California). The Institutional Review Board at each of the study sites approved the study protocols, and written informed consent was obtained from all participants. The sampling strategy yielded a cohort balanced by age, gender (54% women), education (40% with 12 years or less of education), and race (52% African American). Participants underwent a baseline examination and follow-up examinations at years 2, 5, 7, 10, 15, and 20, with retention rates of 91%, 88%, 81%, 79%, 74%, and 72%, respectively. The Coronary Artery Risk Development in Young Adults Women's Study is an ancillary study of women who attended an additional examination at year 16; which was designed to examine the role of androgens and polycystic ovaries in the development of cardiovascular disease. To be eligible for the Coronary Artery Risk Development in Young Adults Women's Study, women had to have attended the year 15 examination, have at least one ovary, and not be pregnant. Approximately 86% of eligible Coronary Artery Risk Development in Young Adults women were successfully recruited and examined for Coronary Artery Risk Development in Young Adults Women's Study during year 16 (n=1,163).
Our study sample included 1,127 women present at both the year 2 (ages 20–32) and year 16 (ages 34–46) examinations. The year 2 examination was considered to be baseline unless otherwise noted.
We used National Institutes of Health consensus criteria for the diagnosis of PCOS, which includes menstrual irregularities attributable to anovulation and either biochemical or clinical evidence of hyperandrogenism.1 Polycystic ovary syndrome was determined by self-report of clinical symptoms, including oligomenorrhea and hirsutism, and serum androgen measures. At the year 16 examination, women were queried about symptoms during two timeframes, past (ages 20–30) and current (ages 34–46). Women were asked about the length and regularity of menstrual cycles. Those who indicated either regular or irregular menstrual cycles 34 days or more were considered as fulfilling criteria for oligomenorrhea. Women who reported unwanted hair growth, excluding the lower leg and underarm, were considered to fulfill criteria for the clinical symptom of hirsutism. Androgen measures were obtained from year 2 stored sera (ages 20–32) and newly collected year 16 sera (ages 34–46). Androgens were assayed by the Obstetrics/Gynecology Research and Diagnostic Laboratory at the University of Alabama, Birmingham. Total testosterone was measured using a competitive immunoassay (Beckman Coulter, Fullerton, CA) using direct chemiluminescent technology on the Beckman Access Automated System. Sex hormone-binding globulin was determined using equilibrium analysis, and free testosterone was calculated on the basis of measured total testosterone and sex hormone-binding globulin.11 Given that testosterone levels are known to decrease as women with PCOS age,8 we used age-specific cut-offs to define biochemical hyperandrogenism. Biochemical hyperandrogenism at year 2 was defined as 76 ng/dL or more of total testosterone or 0.69 ng/dL or more of free testosterone based on the 95th percentile for the nonoligomenorrheic, nonhirsute women at year 2. Biochemical hyperandrogenism at year 16 was defined as 55 ng/dL of total testosterone or 0.44 ng/dL of free testosterone based on the same criteria at year 16.
For our main analysis, participants were classified as having PCOS at ages 20–32 if they reported oligomenorrhea between 20 and 30 years of age and either reported hirsutism between 20 and 30 years of age or fulfilled criteria for biochemical hyperandrogenism at year 2.
Participants were classified as having PCOS at year 16 if they reported current oligomenorrhea and either reported current hirsutism or fulfilled criteria for biochemical hyperandrogenism at year 16. We classified women with persistent PCOS based on the two timeframes, year 2 (ages 20–32) and year 16 (ages 34–46): 1) “never PCOS” includes women who did not fulfill criteria for PCOS at either timeframe; 2) “early PCOS” includes women who fulfilled criteria at year 2 only; and 3) “persistent PCOS” includes women who fulfilled criteria for PCOS at both timeframes. To reduce the possibility of postmenopausal symptoms limiting true assessment of PCOS, we excluded women based on follicle-stimulating hormone more than 40 microinternational units/mL (n=77) and self-report of no menses within the previous 12 months (n=116) at the year 16 examination. Women using hormonal contraception (n=148) were also excluded. Additionally, the 40 women who fulfilled criteria for PCOS at year 16 only were excluded to prevent possible inclusion of women with irregular menstrual cycles attributable to perimenopause.
Self-reported sociodemographic data (age, race, maximally obtained education, and parity) and lifestyle information (physical activity, alcohol abuse, and cigarette status) were collected using self- and interviewer-administered questionnaires. For parity, participants were categorized as either nulliparous (0 births) or parous (one or more births). Physical activity was reported as total exercise units based on the Coronary Artery Risk Development in Young Adults Physical Activity History Questionnaire, which assessed intensity and frequency of participation in a variety of leisure and occupational activities.12 For alcohol abuse, participants were dichotomized based on the at-risk consensus threshold of the National Institute on Alcohol Abuse and Alcoholism, which were met if a woman consumed more than seven drinks per week. For cigarette status, participants were categorized as current users if they smoked more than five cigarettes per week. Family history of diabetes was defined as one or more first-degree relatives with diabetes.
Measurements of weight, height, and waist circumference were obtained by trained and certified clinical staff according to standardized protocols previously described.13 Body weight was measured with participants wearing light clothing using a calibrated balance beam scale to the nearest 0.2 kg. Height, without shoes, was measured with a vertical ruler to the nearest 0.5 cm. Waist circumference was measured at the minimum abdominal girth to the nearest 0.5 cm.
Cardiovascular risk factors were assessed at each Coronary Artery Risk Development in Young Adults examination. Diabetes was defined as a fasting plasma glucose 126 mg/dL or more or use of diabetic medications. Fasting serum total cholesterol, high-density lipoprotein (HDL) cholesterol, and triglyceride levels were measured, and low-density lipoprotein (LDL) cholesterol was calculated using the Friedewald equation.14 Dyslipidemia was defined as LDL more than 160 mg/dL, HDL less than 40 mg/dL, triglycerides more than 200 mg/dL, or use of cholesterol-lowering medications. Blood pressure was measured by trained and certified clinical staff. After a 5-minute rest, blood pressure was measured three times at 1-minute intervals in the right arm of the seated participant using a Hawksley random zero sphygmomanometer and an appropriate-size cuff.13 The mean of the last two blood pressure values was used in the analyses. Hypertension was defined as blood pressure 140/90 mm Hg or more or the use of antihypertensive medications.
For each outcome, cumulative incident diabetes, dyslipidemia, or hypertension was defined as the development of these conditions at any examination visit through year 20, excluding those cases present at year 2.
Descriptive statistics included means (±standard deviation) and median (interquartile range) for continuous variables and proportions for categorical variables. Differences between means or proportions were compared using t tests or the Wilcoxon rank sum test for continuous variables and χ2 tests for categorical variables.
For the main analysis, multivariable logistic regression was used to assess the odds of cumulative incident diabetes, dyslipidemia, and hypertension based on PCOS classification at 20–32 years of age. Prevalent cases at baseline were specifically excluded for each outcome. The first model (model 1) adjusted for age, race, BMI, education, parity, and family history of diabetes obtained at baseline. The second model (model 2) adjusted for the same covariates as model 1 plus year 20 BMI. We also assessed the association between PCOS and diabetes based on model 1 by adjusting for fasting insulin levels at baseline. Likewise, we further explored the association between PCOS and dyslipidemia based on model 1 by adjusting for fasting LDL cholesterol, HDL cholesterol, and triglyceride levels at baseline.
Next, we investigated the effect of PCOS and BMI as a combined risk factor for cumulative incident diabetes and dyslipidemia. We used a categorical variable with four groups as the predictor: 1) women without PCOS at ages 20–32 and baseline BMI less than 25 (reference group); 2) women without PCOS at ages 20–32 and baseline BMI more than 25; 3) women with PCOS at ages 20–32 and baseline BMI less than 25; and 4) women with PCOS at ages 20–32 and baseline BMI more than 25. All the other elements of the analysis were identical to those of the main analysis.
As an additional analysis, we used the PCOS classification of never PCOS, early PCOS, and persistent PCOS as previously described to assess the odds of diabetes and dyslipidemia developing between the year 15 examination and the year 20 examination. The reference group was considered to be the never PCOS group of women for this analysis. Prevalent cases at year 15 were specifically excluded for each outcome. The logistic regression model was adjusted with covariates obtained at the year 15 examination. Stata 10 was used for all statistical analyses.
Of the 1,127 women included in the analyses, 53 (4.7%) fulfilled criteria for PCOS at ages 20–32 (Table 1). White race, nulliparity, and a higher mean fasting insulin level were more common among women with PCOS. The two groups did not differ by maximally obtained education, BMI, waist circumference, physical activity, alcohol use, and current tobacco use. Women with PCOS also did not differ significantly from those without PCOS for the presence of hypertension, diabetes, or dyslipidemia at baseline.
Women with PCOS at ages 20–32 were more likely to have incident diabetes develop by the time they reached 38–50 years of age (Table 2). Polycystic ovary syndrome was associated with twofold higher odds of incident diabetes, which persisted after adjusting for potential confounders, including baseline BMI and year 20 BMI. When baseline BMI was excluded from the multivariable model, the association of PCOS with diabetes remained (adjusted odds ratio 2.6, confidence interval 1.3–5.2). Additional adjustment for fasting insulin levels did not change this association (adjusted odds ratio 2.1, confidence interval 1.0–4.4). In the adjusted logistic regression model with or without year 20 BMI, women with PCOS also had a twofold higher odds of incident dyslipidemia over the course of 18 years of follow-up (Table 2). After additional adjustment for baseline LDL cholesterol, HDL cholesterol, and triglyceride levels, the magnitude of the association remained unchanged but was no longer statistically significant (adjusted odds ratio 1.9, confidence interval 0.9–3.8). Given the race differences suggested in Table 1, we examined potential effect modification by race and found that the association between PCOS and either incident diabetes or incident dyslipidemia did not differ between the two race groups (P for interaction=0.4 and 0.2, respectively). The association between PCOS and incident hypertension was not statistically significant (Table 2).
Because of the lack of a significant association between PCOS and incident hypertension, the subsequent analyses focused on incident diabetes and dyslipidemia. Using PCOS classification and baseline BMI as a combined predictor, PCOS was associated with a threefold higher odds of diabetes developing among normal-weight women with PCOS compared with normal-weight women without PCOS over the course of 18 years of follow-up (Table 3). Although these results are limited by wide confidence intervals, the magnitude of this risk appeared to be greater than that of overweight women without PCOS, but less than that of overweight women with PCOS. After further adjustment with year 20 BMI, the independent association for PCOS and diabetes regardless of baseline BMI category remained. The same pattern of association was observed for dyslipidemia, although some associations did not reach the level of statistical significance in this analysis (Table 3).
Of the 746 women available for the secondary analysis of persistent PCOS, 2.0% (15/746) met criteria for persistent PCOS and 3.5% (26/746) were classified as having early PCOS. Thus, spontaneous resolution of PCOS symptomatology over the course of 14 years occurred in 63% of women classified with PCOS at ages 20–32 (26/41). Women with persistent PCOS had a sevenfold higher odds of diabetes developing over the subsequent 5 years compared with women without PCOS, although wide confidence intervals limit the precision of the point estimate. Those with early PCOS did not demonstrate a statistically significant increased odds of diabetes (Table 4). In contrast, women with early PCOS, but not persistent PCOS, had a significantly increased odds of incident dyslipidemia compared with those without PCOS.
In a large, biracial cohort of young U.S. women, we found that PCOS among women in their 20s was associated with an increased odds of subsequent incident diabetes and dyslipidemia, but not hypertension, by the fifth decade of life. We observed that even normal-weight women with PCOS were at an increased risk for diabetes compared with normal-weight women without PCOS. In addition, women with persistent PCOS symptomatology were at highest risk for diabetes. The results of these longitudinal analyses support the hypothesis that PCOS places women at an increased risk for cardiovascular disease and have important implications for screening, surveillance, and risk factor modification in these young women.
Natural history studies have observed small cohorts of women with PCOS identified by ovarian wedge resection and noted an increased risk for diabetes,6,7 hypertriglyceridemia,6 and hypercholesterolemia6 compared with rates in the general population. The Nurses' Health Study II, a prospective cohort study, assessed menstrual cycle history between 18 and 22 years based on recall in 101,073 women and found that women with long or irregular cycles had a twofold increased risk of incident diabetes after adjusting for self-reported BMI at age 18 years and other confounders.15
Our study supports the existing natural history studies and extends the Nurses' Health Study in several ways. We studied a large and well-characterized cohort of white and African-American women who have been observed for more than 18 years. We used a standardized definition of PCOS supported by measures of serum androgen levels, as well as participant self-report. We were also able to assess PCOS at two time points, allowing for the assessment of persistent symptomatology. Furthermore, we examined the incidence of three important cardiovascular disease risk factors over the course of this 18-year period, diabetes, dyslipidemia, and hypertension.
Polycystic ovary syndrome as an independent risk factor for diabetes can be explained by several mechanisms. Studies have shown a unique postbinding defect in insulin signal transduction in skeletal muscle of women with PCOS, resulting in increased serine phosphorylation of the insulin receptor and insulin receptor substrate-1.16 We did not see a change in the association between PCOS and diabetes after adjusting for fasting insulin levels; however, this does not account for the significant insulin resistance observed in women with PCOS compared with controls, as demonstrated in euglycemic clamp studies. Hyperandrogenism also may play a role in insulin resistance because insulin and androgen levels are known to be positively correlated in women with PCOS.17 Androgen administration to healthy women has been shown to be associated with the development of insulin resistance; conversely, the blockade of androgen action led to improved insulin sensitivity.18
The association between PCOS and hypertension has been less clear in the literature. Some studies show that PCOS is associated with hypertension based on ambulatory 24-hour monitoring or outpatient diagnoses,4,6,19 whereas other studies have not reached the same conclusion.7 The use of different diagnostic criteria for PCOS may account for these inconsistent findings. Our results suggest that PCOS is not associated with incident hypertension; however, this does not exclude the possibility of an association between a history of PCOS and future hypertension as this cohort continues to age and transition through menopause. However, we do note that hypertension is not uncommon by the year 20 examination, affecting 28.4% of women in our cohort.
Our study indicates that PCOS is an independent risk factor for subsequent disease, most notably diabetes. The results of this study support the recommendation by a consensus panel representing the European Society of Human Reproduction and Embryology and the American Society of Reproductive Medicine,20 as well as the Androgen Excess Society,21 to screen all women with PCOS for diabetes. Although retrospective data suggest that metformin therapy may protect against impaired glucose tolerance and diabetes in women with PCOS,22 clinical trials are needed before metformin can be routinely recommended for women with PCOS with normal glucose tolerance.
Certain limitations should be noted when interpreting our study results. Although the use of questionnaires for the diagnosis of PCOS according to symptomatology has been validated,23 misclassification of cases may occur. Additionally, the retrospective diagnosis of PCOS for women at ages 20–32 is subject to recall bias. However, the ascertainment of cases in these women is complemented by serum androgen levels and the prevalence of PCOS noted in this study is consistent with the literature. Furthermore, misclassification is likely to have resulted in a bias to the null. The considerable strengths of our study include the standardized definition of PCOS, the large cohort of African-American and white women, the long follow-up period, and the adequacy of our outcome assessment. The 18-year follow-up period is further strengthened by an overall retention rate of 72% within the Coronary Artery Risk Development in Young Adults cohort.
In conclusion, in this large, biracial cohort with extensive follow-up, we find that PCOS in early adulthood is associated with an increased long-term risk of diabetes and dyslipidemia, independent of BMI.
1. Zawadski J, Dunaif A. Diagnostic criteria for polycystic ovary syndrome: Towards a rational approach. In: Dunaif A, Givens J, Haseltine F, Merriam G, editors. Polycystic ovary syndrome. Boston (MA): Blackwell Scientific Publications; 1992. p. 377–84.
2. Legro RS, Kunselman AR, Dodson WC, Dunaif A. Prevalence and predictors of risk for type 2 diabetes mellitus and impaired glucose tolerance in polycystic ovary syndrome: a prospective, controlled study in 254 affected women. J Clin Endocrinol Metab 1999;84:165–9.
3. Legro RS, Kunselman AR, Dunaif A. Prevalence and predictors of dyslipidemia in women with polycystic ovary syndrome. Am J Med 2001;111:607–13.
4. Lo JC, Feigenbaum SL, Yang J, Pressman AR, Selby JV, Go AS. Epidemiology and adverse cardiovascular risk profile of diagnosed polycystic ovary syndrome. J Clin Endocrinol Metab 2006;91:1357–63.
5. Taponen S, Martikainen H, Jarvelin MR, Sovio U, Laitinen J, Pouta A, et al. Metabolic cardiovascular disease risk factors in women with self-reported symptoms of oligomenorrhea and/or hirsutism: Northern Finland Birth Cohort 1966 Study. J Clin Endocrinol Metab 2004;89:2114–8.
6. Wild S, Pierpoint T, McKeigue P, Jacobs H. Cardiovascular disease in women with polycystic ovary syndrome at long-term follow-up: a retrospective cohort study. Clin Endocrinol (Oxf) 2000;52:595–600.
7. Lunde O, Tanbo T. Polycystic ovary syndrome: a follow-up study on diabetes mellitus, cardiovascular disease and malignancy 15–25 years after ovarian wedge resection. Gynecol Endocrinol 2007;23:704–9.
8. Winters SJ, Talbott E, Guzick DS, Zborowski J, McHugh KP. Serum testosterone levels decrease in middle age in women with the polycystic ovary syndrome. Fertil Steril 2000;73:724–9.
9. Elting MW, Korsen TJ, Rekers-Mombarg LT, Schoemaker J. Women with polycystic ovary syndrome gain regular menstrual cycles when ageing. Hum Reprod 2000;15:24–8.
10. Friedman GD, Cutter GR, Donahue RP, Hughes GH, Hulley SB, Jacobs DR Jr, et al. CARDIA: study design, recruitment, and some characteristics of the examined subjects. J Clin Epidemiol 1988;41:1105–16.
11. Pearlman WH, Crepy O, Murphy M. Testosterone-binding levels in the serum of women during the normal menstrual cycle, pregnancy, and the post-partum period. J Clin Endocrinol Metab 1967;27:1012–8.
12. Sidney S, Jacobs DR Jr, Haskell WL, Armstrong MA, Dimicco A, Oberman A, et al. Comparison of two methods of assessing physical activity in the Coronary Artery Risk Development in Young Adults (CARDIA) Study. Am J Epidemiol 1991;133:1231–45.
13. Folsom AR, Burke GL, Ballew C, Jacobs DR Jr, Haskell WL, Donahue RP, et al. Relation of body fatness and its distribution to cardiovascular risk factors in young blacks and whites. The role of insulin. Am J Epidemiol 1989;130:911–24.
14. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972;18:499–502.
15. Solomon CG, Hu FB, Dunaif A, Rich-Edwards J, Willett WC, Hunter DJ, et al. Long or highly irregular menstrual cycles as a marker for risk of type 2 diabetes mellitus. JAMA 2001;286:2421–6.
16. Corbould A, Kim YB, Youngren JF, Pender C, Kahn BB, Lee A, et al. Insulin resistance in the skeletal muscle of women with PCOS involves intrinsic and acquired defects in insulin signaling. Am J Physiol Endocrinol Metab 2005;288:E1047–54.
17. Burghen GA, Givens JR, Kitabchi AE. Correlation of hyperandrogenism with hyperinsulinism in polycystic ovarian disease. J Clin Endocrinol Metab 1980;50:113–6.
18. Corbould A. Effects of androgens on insulin action in women: is androgen excess a component of female metabolic syndrome? Diabetes Metab Res Rev 2008;24:520–32.
19. Holte J, Gennarelli G, Berne C, Bergh T, Lithell H. Elevated ambulatory day-time blood pressure in women with polycystic ovary syndrome: a sign of a pre-hypertensive state? Hum Reprod 1996;11:23–8.
20. Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Hum Reprod 2004;19:41.
21. Salley KE, Wickham EP, Cheang KI, Essah PA, Karjane NW, Nestler JE. Glucose intolerance in polycystic ovary syndrome–a position statement of the Androgen Excess Society. J Clin Endocrinol Metab 2007;92:4546–56.
22. Sharma ST, Wickham EP III, Nestler JE. Changes in glucose tolerance with metformin treatment in polycystic ovary syndrome: a retrospective analysis. Endocr Pract 2007;13:373–9.
23. Pedersen SD, Brar S, Faris P, Corenblum B. Polycystic ovary syndrome: validated questionnaire for use in diagnosis. Can Fam Physician 2007;53:1042–7, 1041.
Figure. No caption available.