The Cardiovascular Risk of Young Women with Polycystic Ovary Syndrome: An Observational, Analytical, Prospective Case-Control Study

To evaluate the cardiovascular risk of polycystic ovary syndrome (PCOS), we investigated lipid profile, metabolic pattern, and echocardiography in 30 young women with PCOS and 30 healthy age- and body mass index (BMI)-matched women. PCOS women had higher fasting glucose and insulin levels, homeostasis model assessment score of insulin sensitivity, total cholesterol (TC) and low density lipoprotein cholesterol (LDL-C) levels, and TC/high density lipoprotein cholesterol (HDL-C) ratio and lower HDL-C levels than controls. Additionally, PCOS women had higher left atrium size (32.0 ± 4.9 vs. 27.4 ± 2.1 mm; P < 0.0001) and left ventricular mass index (80.5 ± 18.1 vs. 56.1 ± 5.4 g/m²; P < 0.0001) and lower left ventricular ejection fraction (64.4 ± 4.1 vs. 67.1 ± 2.6%; P = 0.003) and early to late mitral flow velocity ratio (1.6 ± 0.4 vs. 2.1 ± 0.2; P < 0.0001) than controls. When patients and controls were grouped according to BMI [normal weight (BMI, >18 and <25 kg/m²), overweight (BMI, 25.1–30 kg/m²), and obese (BMI, >30 kg/m²)], the differences between PCOS women and controls were maintained in overweight and obese women. In normal weight PCOS women, a significant increase in left ventricular mass index and a decrease in diastolic filling were observed, notwithstanding no change in TC, LDL-C, HDL-C, TC/HDL-C ratio, and TG compared with controls. In conclusion, our data show the detrimental effect of PCOS on the cardiovascular system even in young women asymptomatic for cardiac disease.

THE POLYCYSTIC OVARY syndrome (PCOS) is one of the most common endocrine diseases in women, affecting up to 10% of women in reproductive age (1, 2). PCOS is characterized by chronic anovulatory cycles, oligo- or amenorrhea, hirsutism, and insulin resistance; obesity is also common (3–5). PCOS not only has a negative effect on fertility, but it is also considered a clear-cut plurimetabolic syndrome (2, 4, 6), being associated with type 2 diabetes mellitus, hypertension, and dyslipidemia (7–9). Insulin resistance is probably the major risk factor for the occurrence of cardiovascular (CV) disease (CVD) in PCOS (3).

The risk of coronary artery disease (10–12) and myocardial infarction (13) has been reported to be increased in patients with PCOS compared with regularly cycling women even if mortality because of circulatory disease does not seem to be increased (14–16). In PCOS women, endothelial and diastolic dysfunction have been shown and associated with both elevated androgen levels and insulin resistance (17, 18). Recently, together with classical CV risk factors, such as total (TC) and high density lipoprotein cholesterol (HDL-C) levels, obesity, homocysteine, and left ventricular (LV) hypertrophy (LVH) (19) have been shown to be independently associated with an increased CV risk. LVH is an important predictor of CV morbidity and mortality (20, 21), but determinants of LV mass (LVM) in nonhypertensive subjects are still incompletely understood (22–24).

The present study aims at further investigating the CV risk of young women with PCOS. We 1) studied the prevalence of LVH and altered diastolic filling and systolic performance by echocardiography, and 2) analyzed any potential relationship between cardiac and metabolic parameters, such as insulin secretion and lipid profile, in a selected cohort of 30 young women with PCOS.

Subjects and Methods

Subjects

Thirty young (<35 yr) women with PCOS and 30 age- and body mass index (BMI)-matched controls were enrolled in this pilot study. The controls were defined as age- and BMI-matched with cases when the number of years ± age of cases and the BMI (kilograms per meter squared) of cases were less than to 2 yr and less than to 1 kg/m², respectively.

PCOS was defined according to clinical [Ferriman-Gallwey score, >8) (25) oligomenorrhea or amenorrhea > 6 months; biological LH/FSH ratio, >2; hyperandrogenism] and ultrasonographic (26, 27) findings. In Table 1 are shown the clinical and biochemical diagnostic features of the PCOS women.

The control group consisted of 30 healthy volunteer females with regular menstrual cycles. Their healthy state was determined by medical history, physical and pelvic examination, and complete blood chemistry. The normal menstrual cycle was diagnosed after a 3-month prestudy period. During this period, all healthy women recorded in a daily diary the characteristics of their menses. A normal menstrual cycle was defined as cyclic uterine bleedings with a duration of 4–5 d and a frequency of 26–32 d/month. The quantity of blood flow was defined as normal subjectively by each woman and objectively using a serum hemoglobin assay. At the entry, the normal ovulatory state was confirmed by transvaginal ultrasonography and plasma progesterone levels during the luteal phase of the cycle.

Exclusion criteria for all subjects included age less than 18 or more than 30 yr, pregnancy, hypothyroidism, hyperprolactinemia, Cushing’s syndrome, nonclassical congenital adrenal hyperplasia, and current or previous (within the last 6 months) use of oral contraceptives, glucocorticoids, antiandrogens, ovulation induction agents, antidiabetic and antiobesity drugs, or other hormonal drugs. The presence of hyperprolactinemia was excluded with a single assay of plasma prolactin (PRL) levels (normal, <25 ng/ml) (28). In women with a serum PRL levels greater than 25 ng/ml, hyperprolactinemia was excluded considering the average value of serum PRL assayed at 0800 h for three times every 15 min. Nonclassical congenital adrenal hyperplasia was excluded with a single measure of serum 17-hydroxyprogesterone levels (normal, <6.0 nmol/liter) (29).

Women with clinical and/or biochemical hyperandrogenism alone were excluded from the control group. None of the patients was affected by any neoplastic or cardiovascular disorder, and none of the women (PCOS and controls) had hypertension or used any drug during the 3 months preceding the study. None of the women had a BMI less than 18 kg/m².

The study was performed according to the guidelines of the Helsinki Declaration on human experimentation and was approved by the local ethics committee of University Federico II of Naples. All patients gave their written informed consent before the study.

Study protocol

According to a previous study protocol (30) all patients and controls were submitted to a three-step study.

Clinical study.

Height, weight, BMI, waist to hip ratio, and measurements of heart rate, systolic (SBP) and diastolic (DBP) blood pressure were evaluated by standard methods. BMI was measured as the ratio between the weight and the square of the height. A BMI between 25 and 30 kg/m² was considered an index of overweight whereas a BMI greater than 30 kg/m² was considered an index of obesity (31). The waist to hip ratio was measured as the ratio between the waist, considered to be the smallest circumference of torso between the 12th rib and the iliac crest, and the circumference of the hip, considered as the maximal extension of the buttocks. All measurements were performed when the patients were in a standing position with relaxed abdomen, arms at their sides, and joined feet (32). Blood pressure was measured in the right arm, with the subjects in a relaxed sitting position. The average of six measurements (three taken by each of two examiners) with a mercury sphygmomanometer was used.

Biochemical study.

Fasting glucose, triglycerides (TG), TC, low density lipoprotein cholesterol (LDL-C) and HDL-C levels were measured by standard procedures in the morning between 0800 and 0900 h after an overnight fast and resting in bed during early follicular phase (d 2–5) of the spontaneous or progesterone-induced withdrawal bleeding. Samples were performed 10 min after needle insertion in duplicate and immediately centrifuged, and the serum was stored at −80 C until analysis. The TC/HDL-C ratio was also calculated for each woman (33). Hypertriglyceridemia was diagnosed when triglycerides levels were greater than 250 mg/dl (34), whereas hypercholesterolemia was diagnosed when total cholesterol levels were greater than 240 mg/dl (35). Impaired glucose tolerance was diagnosed after an oral glucose tolerance test (75 g glucose diluted in 250 ml saline solution, measuring blood glucose every 30 min for 2 h). In line with World Health Organization criteria, diabetes mellitus was diagnosed when fasting blood glucose levels were greater than 126 mg/dl in two consecutive determinations or 200 mg/dl or more 2 h after oral glucose; reduced glucose tolerance was diagnosed when blood glucose levels were 126–200 mg/dl 2 h after oral glucose (36). Fasting and glucose load-stimulated insulin levels were measured in patients and controls to estimate the insulin sensitivity; the homeostasis model assessment (HOMA) index, which is considered an index of insulin resistance (37), was also calculated.

Echocardiographic study.

M-Mode, two-dimensional, and pulsed Doppler echocardiography studies were performed by one operator (L.S.), who was blind with respect to the presence of metabolic abnormalities or arterial hypertension, using ultrasound systems (Apogee CX, Interspec, Inc., Ambler, PA) with a 3.5-mHz transducer during at least three consecutive cardiac cycles. All patients were studied in the left lateral position after a 10-min resting period according to the recommendations of the American Society of Echocardiography (38). The following measurements were recorded on M-mode tracing: interventricular septum and LV posterior wall thickness. The frequency-normalized mean velocity of circumferential fiber shortening end-diastolic and end-systolic volumes (EDV and ESV) and ejection fraction [EF = EDV − ESV/EDV (%)] were estimated according to the Quinones method (39). The LV ejection fraction (LVEF) is normal when it is above 50%. The LVM was calculated by Devereux’s formula according to Penn’s convention with the following regression-corrected cube formula: LVM = 1.04[(ISV + LVID + PWT)³ − (LVID)³] − 14 g. LVH was diagnosed when LVM values, corrected for body surface area [LVM index (LVMi)], were 135 g/m² or greater in males and 110 g/m² or greater in females (40). Doppler studies provided indexes of ventricular filling that were derived from the mitral flow velocities curves, i.e. maximal early diastolic flow velocity (E; centimeters per second), maximal late diastolic flow velocity (A; centimeters per second), and the ratio between E and A curves (E/A; normal, >1).

Assays

Plasma LH, FSH, PRL, estradiol, progesterone, testosterone, Δ⁴-androstenedione, and dehydroepiandrosterone sulfate were measured by specific RIA as previously described (41, 42). Serum 17-hydroxyprogesterone levels were determined by RIA (Diagnostic Systems Laboratories, Webster, TX) with a sensitivity of 0.01 ng/ml and intra- and interassay coefficients of variation of 8.9% and 9.0%, respectively. SHBG levels were measured using an immunoradiometric assay, as previously described (28, 43). In each woman the free androgen index (FAI) was calculated using the following formula: testosterone (nmol/liter)/SHBG × 100 (44).

Serum insulin was measured by a solid phase chemiluminescent enzyme immunoassay using commercially available kits (Immunolite, Diagnostic Products Corp., Los Angeles, CA). Blood glucose levels were determined by the glucose oxidase method. Serum TC, HDL-C, LDL-C, and TG levels were measured with an autoanalyzer (Monarch 1000, Instrumentation Laboratory, Milan, Italy) using commercially available kits (IL TEST, Instrumentation Laboratory) as previously reported (45, 46). Serum LDL-C was evaluated using Friedewald’s formula: TC − HDL-C − 1/5 TG (47).

Statistical analysis

Statistical analysis was performed using the SPSS 11.5.2.1 package (SPSS, Inc., Chicago, IL). Data were expressed as the mean ± sd. Stepwise multiple linear regression analysis was performed to evaluate the relative importance of LVMi, as a dependent variable, against predictor variables: age, BMI, HOMA, FAI, lipid profile, SBP, DBP, and number of cigarettes smoked daily. Differences in physical activity and smoking habitus (percentage of women smoking) were evaluated using Mann-Whitney U test and χ² test, respectively. The other comparisons between patients and controls were performed using a t test for unpaired data. Significance was set at 5%.

Results

Clinical study (Table 2)

As expected, PCOS patients had significantly (P < 0.05) higher LH, testosterone, Δ⁴-androstenedione, and SHBG levels than controls. DBP and mean blood pressure were also significantly (P < 0.05) higher in the patients than in the controls, whereas SBP and heart rate were similar in the two groups. The endocrine parameters were in line with the diagnosis of PCOS in the patients group. In particular, the FAI was significantly (P < 0.05) higher in patients than controls.

Biochemical study (Table 3)

PCOS patients had significantly (P < 0.05) higher fasting glucose and insulin levels, HOMA, TC, LDL-C, and TC/HDL-C ratio, whereas HDL-C levels were significantly (P < 0.05) lower than control values. TG levels were similar in patients and controls. Neither the patients nor the controls had hypercholesterolemia or hypertriglyceridemia.

Echocardiography study (Table 4)

Young women with PCOS had a cardiac size significantly (P < 0.05) increased compared with controls. They had also interventricular septum, LV posterior wall thickness, ESV, and LVMi significantly (P < 0.05) higher than controls. Two of 30 (6.7%) had LVH (>110 g/m²). Additionally, PCOS patients had significantly (P < 0.05) lower LVEF and E/A than controls, although all patients had normal LVEF, and two patients had abnormal E/A (6.7%).

When patients and controls were grouped according to BMI into three groups, namely subjects of normal weight (BMI, >18 < 25 kg/m²), overweight (BMI, 25.1–30 kg/m²), and obese (BMI, >30 kg/m²), it was observed that the majority of the differences in the metabolic profile and cardiac findings persisted (Table 5). As expected, there was a progressive impairment of metabolic profile and echocardiographic pattern from patients with normal weight to those with obesity (Table 5). Interestingly, normal weight PCOS women had a significant (P < 0.05) increase in LVMi and a decrease in diastolic filling, but no difference in TC, LDL-C, HDL-C, TC/HDL-C ratio, or TG, compared with controls. The other two groups had the same pattern described in the series of patients as a whole.

The final model of stepwise multiple linear regression analysis in PCOS patients showed that LVMi is linearly related only to HOMA (coefficient = 6.764; β =0.874; P < 0.0001; constant = 58.647).

Discussion

The results of this prospective controlled study in selected young women with PCOS showed a significant impairment of glucose and lipid profile associated with an increased LVM and decreased LV performance and diastolic filling. Most of these abnormalities persisted even in young patients with normal weight, suggesting that the pathogenesis of cardiac abnormalities in PCOS is not only dependent on BMI. In fact, only in the patients was LVMi significantly correlated with both BMI and HOMA index.

PCOS women represent an intriguing biological model illustrating hormonal effects on cardiovascular risk. In fact, several findings indicate a relationship between heart disease and PCOS, i.e. dyslipidemia (9), insulin resistance (3, 7), increased LVM, and diastolic dysfunction (48, 49). PCOS as a putative cause of CVD is, however, still questioned, because definitive data to demonstrate a direct influence of PCOS on heart disease are still lacking (50–52). On the basis of the multiple findings linking CVD with PCOS, we designed this prospective controlled study to investigate different variables affecting heart structure and function. We selected a group of young patients to exclude any role of long-standing undiagnosed PCOS and investigated the early phases of the disease. To better understand the role of increased BMI frequent in PCOS, we included a group of lean patients and controls. This would allow us to detect cardiac abnormalities not related to overweight and obesity.

PCOS women were reported to have higher TG levels and lower HDL-C values (53). Because insulin is a major positive regulator of lipoprotein lipase that is involved in the pathway of HDL-C production, dyslipidemia is probably secondary to insulin resistance, although hyperandrogenism may affect lipoproteins and lipids independently of insulin levels and body weight (9). In our series we confirmed an unfavorable lipid profile; however, we did not find any increase in TG levels. Our young patients with PCOS had an increase in TC and a decrease in HDL-C levels as major findings associated with increased glucose and insulin levels and HOMA index. We also confirmed increased LVM and diastolic dysfunction in young PCOS women. Interestingly, these abnormal echocardiographic aspects were present even in the 12 normal weight women who did not show alterations in lipid profile. Furthermore, these young PCOS women showed greater LVMi than their age- and BMI-matched controls. Although it is well known that ventricular mass indexed for height is significantly higher in overweight than in normal weight subjects (54), our PCOS women showed an increased LVMi that was not weight dependent. Therefore, PCOS could be considered an aggravating factor causing LVH and could play a role in the early cardiovascular disease in PCOS. Among the various echocardiographic parameters that were different between PCOS and controls, the increased LVMi in young PCOS patients has clinical importance, because it represents the main cause of both LVH and diastolic dysfunction in PCOS women, worsening from early to elderly age.

LVH is one of the several metabolic and cardiovascular risk factors associated with insulin resistance (55) and/or visceral obesity (56). In our series, there was a linear relation between LVMi and HOMA. Hyperinsulinemia is a predictor of coronary artery disease (57–60), and insulin resistance has been proposed as the key factor linking hypertension, glucose intolerance, obesity, lipid abnormalities, and coronary heart disease in an association called metabolic syndrome X (61). In fact, the increased LVM in PCOS women could be caused by an increase in blood pressure, even in the absence of hypertension (62). Indeed, in our series we found slightly increased mean blood pressure and DBP in PCOS women compared with controls.

In accord with previous studies (48, 49), we also report diastolic dysfunction in PCOS. Specifically, Tiras et al. (49) suggested that patients with PCOS have diastolic dysfunction; furthermore, no difference in E/A was observed between PCOS and control women. On the contrary, we showed both reduced ejection fraction of LV and E/A, suggesting impaired LV diastolic filling. However, whether the E/A ratio is an early predictor of diastolic dysfunction in PCOS remains to be determined. In addition, Prelevic et al. (63) reported that systolic flow velocity is lower in PCOS than in age-matched control women, and there is an inverse relationship between serum fasting insulin and LV systolic outflow parameters; furthermore, increased insulin levels in PCOS are associated with decreased cardiac flow (63).

Although we did not find any association between LVM or diastolic dysfunction and hyperandrogenism, cross-sectional data consistently showed a strong obesity-independent association of androgen excess in women with a cluster of CVD risk factors, including insulin resistance, dyslipidemia, and impaired fibrinolysis. It was therefore suggested that the chronically abnormal hormonal and metabolic milieu found in women with PCOS, starting from adolescence, may predispose these women to premature atherosclerosis. Furthermore, based on calculated risk profiles, PCOS women were predicted to have a 7-fold increased relative risk for myocardial infarction; in fact, Birdsall et al. (11) showed significant associations between the presence of polycystic ovaries and the presence and severity of cardiovascular artery disease and a family history of myocardial infarction as well as with elevated levels of insulin and TG and lower levels of HDL-C (11).

Some findings of clinical studies showed that women with CVD were affected more frequently than controls by clinical symptoms of androgen excess, such as hirsutism and polycystic ovaries (3, 4, 50, 64). Because many women with PCOS are overweight, and most, if not all, are insulin resistant, it is a matter of debate whether these symptoms are secondary to obesity and insulin resistance or whether hyperandrogenism itself contributes to obesity, insulin resistance, and hyperinsulinemia.

Our findings suggest that young PCOS women have increased LVM and diastolic dysfunction, neither of which is weight dependent, demonstrating that PCOS women are candidates for early CVD. Additional studies are needed to better clarify the exact role of insulin resistance and/or hyperandrogenism in the CVD, particularly in the complex etiopathogenetic scenario of the increased LVM in young PCOS patients.

Acknowledgements

We are grateful to Dr. Francesco Manguso (Department of Clinical and Experimental Medicine, Gastroenterology Unit, “Federico II” University, Naples, Italy) for his invaluable assistance in statistical analysis.

Abbreviations:

 
• BMI,

  Body mass index;

 
• CV,

  cardiovascular;

 
• CVD,

  CV disease;

 
• DBP,

  diastolic blood pressure;

 
• E/A,

  maximal early diastolic flow velocity/maximal late diastolic flow velocity ratio;

 
• EDV,

  end-diastolic volume;

 
• ESV,

  end-systolic volume;

 
• FAI,

  free androgen index;

 
• HDL-C,

  high density lipoprotein cholesterol;

 
• HOMA,

  homeostasis model assessment;

 
• LDL-C,

  low density lipoprotein cholesterol;

 
• LV,

  left ventricular;

 
• LVEF,

  LV ejection fraction;

 
• LVH,

  LV hypertrophy;

 
• LVM,

  LV mass;

 
• LVMi,

  LV mass index;

 
• PCOS,

  polycystic ovary syndrome;

 
• PRL,

  prolactin;

 
• SBP,

  systolic blood pressure;

 
• TC,

  total cholesterol;

 
• TG,

  triglycerides.