Higher High-Density Lipoprotein Cholesterol in African-American Women with Polycystic Ovary Syndrome Compared with Caucasian Counterparts

Abstract

Context: Studies have demonstrated lipid differences among African-Americans and Caucasians and between women with polycystic ovary syndrome (PCOS) and normally ovulating women. However, few studies have examined racial differences in lipoprotein levels in women with PCOS.

Objective: This study compared lipoprotein levels in African-American and Caucasian women with PCOS.

Design and Setting: We performed a retrospective chart review of 398 subjects seen as new patients for PCOS at the Duke University Medical Center Endocrinology Clinic in Durham, NC.

Patients: We identified 126 charts appropriate for review, based on a diagnosis of PCOS (using the 1990 National Institutes of Health criteria), a self-reported race of either Caucasian or African-American, and a body mass index (BMI) higher than 25. We excluded patients taking glucophage, oral contraceptives, or lipid-lowering medications.

Main Outcome Measure: Age, BMI, total cholesterol, high-density lipoprotein (HDL) cholesterol, non-HDL cholesterol, random triglycerides (TG), and oral glucose tolerance test measurements were collected and included in the analysis.

Results: African-American women with PCOS had higher HDL cholesterol levels (52.6 vs. 47.5 mg/dl, P = 0.019), lower non-HDL cholesterol (134.1 vs. 154.6 mg/dl, P = 0.046), and lower TG levels (97.5 vs. 168.2 mg/dl, P < 0.001) than Caucasian women. These differences could not be attributed to age, BMI, or differences in insulin resistance as determined by homeostasis model assessment of insulin resistance.

Conclusion: African-American women with PCOS appear to have a more favorable lipid profile than Caucasian women with PCOS having higher HDL cholesterol, lower non-HDL cholesterol, and lower TG when BMI and insulin resistance are equal.

African American women with PCOS have higher HDL, lower non-HDL, and lower triglycerides than Caucasian PCOS women.

Many studies have demonstrated that African-Americans have a different cardiovascular risk profile when compared with Caucasians. When comparing the two races, African-Americans are more insulin resistant, have a higher prevalence of diabetes, have higher systolic and diastolic blood pressure, and are more overweight (1). Studies have also demonstrated higher cardiovascular mortality in African-Americans (1).

Despite the higher cardiovascular risk, it appears that African-Americans have a more favorable lipoprotein profile compared with Caucasians. African-Americans tend to have lower triglycerides (TG) (2), higher high-density lipoprotein (HDL) cholesterol (3), and lower non-HDL cholesterol (4), although not all studies have found these differences (5).

Although lipid differences between African-Americans and Caucasians have been established, it has not been determined whether these racial differences hold true in women with polycystic ovary syndrome (PCOS).

Compared with non-PCOS women, women with PCOS tend to have lower HDL cholesterol (6,7). This lower HDL cholesterol, combined with increased low-density lipoprotein cholesterol and TG and a higher prevalence of obesity (7), has led to the speculation that PCOS women have a particularly high cardiovascular risk.

It is important to compare racial differences in women with PCOS to better understand the cardiovascular risk that affects this population. We hypothesized that HDL cholesterol levels would be higher in African-American women with PCOS compared with Caucasian counterparts. This study examines the difference in lipoprotein values between overweight and obese African-American and Caucasian women diagnosed with PCOS.

Subjects and Methods

This was a retrospective chart review, conducted in an academic health center setting, designed to evaluate the differences in lipid values between African-American and Caucasian women with clearly diagnosed PCOS. The Institutional Review Board of Duke University Medical Center approved this study.

Subjects

The study population included 398 women aged 18–50 yr seen by one of our investigators (A.J.B.) in a new-patient, adult medical endocrinology PCOS clinic at Duke University Medical Center. Women were referred to this clinic from a variety of sources, including obstetric-gynecology, primary care, and subspecialty clinics and self-referral. All 398 subjects had PCOS based on the 1990 National Institutes of Health expert panel definition (8). All women had oligo/anovulation and either clinical or biochemical hyperandrogenism. Hyperprolactinemia, uncontrolled thyroid disease, and pregnancy were ruled out routinely. Testing for nonclassical congenital adrenal hyperplasia was done in patients with Ashkenazi Jewish ethnicity. Testing for premature ovarian failure, Cushing’s syndrome, or an androgen-secreting tumor was performed when these diagnoses were suggested by history or exam. Subjects who previously participated in an endocrine research study were excluded.

For the final analysis, subjects with body mass index (BMI) below 25 kg/m² were excluded because there were very few African-American PCOS women with BMI under 25 kg/m² compared with Caucasian PCOS women. Subjects with missing variables (BMI, age, race, lipids, or oral glucose tolerance test values) and those taking lipid-altering medications, glucophage, or hormonal therapy (such as oral contraceptives) were also excluded from analysis.

A total of 398 charts were reviewed. Of these, 218 were excluded because they were taking metformin or hormonal therapy or due to inability to confirm PCOS diagnosis from available data. Twelve subjects were excluded because they were taking lipid-altering medications. Forty-two women did not have all data variables present and were excluded from the analysis. Final analysis included 126 women with PCOS (94 Caucasian subjects, 32 African-American subjects).

Measured outcomes

Laboratory tests and measurements from the physical exam were recorded from the subject’s initial visit to the PCOS clinic or from the date closest to initial visit. Recorded outcomes included age, self-reported race, blood pressure, height, weight, calculated BMI, lipoprotein profiles (including total cholesterol, TG, HDL cholesterol, and non-HDL), and oral glucose tolerance (including fasting and 2-h glucose and insulin administered with a 75-g oral glucose challenge). Homeostasis model assessment of insulin resistance (HOMA-IR) was calculated using the following equation: fasting serum insulin (milli-international units per milliliter) × fasting plasma glucose (millimoles per liter)/22.5. Fasting lipids were used in this analysis if available, and nonfasting values were used if not. Fasting status of lipid measurements was not always known.

Upon scheduling their initial appointment, subjects were asked demographic information including race and were scored as African-American, Alaskan Native, American Indian, Asian, Hawaii-Pacific, White, multiracial, or other. Due to the small sample sizes of the other racial groups, only the African-American and Caucasian groups are compared in this paper.

Statistical analysis

Continuous variables are presented as mean (sd). A general linear model was constructed for each lipid to evaluate its association with race after accounting for the age, BMI, and insulin sensitivity of the subject. A log transformation was applied to account for evidence of nonnormality. The linearity of each covariate was evaluated, and appropriate transformations were applied when necessary. All analyses were performed in SAS version 8.2 (Cary, NC). A P value of 0.05 was considered statistically significant.

Results

Participants were young women (mean age 30.5 yr) with a mean BMI of 38 kg/m² (Table 1). They were normotensive and normoglycemic on glucose tolerance testing. Average HOMA-IR at 4.2 indicated insulin resistance as a group. Total testosterone was available for 82 participants (63 Caucasian, 19 African-American). Mean values were comparable between the races (mean ± sd = 63.4 ± 33.3 for Caucasians vs. 69.9 ± 32.4 ng/dl for African-Americans).

Table 1.

Study population characteristics

Characteristic	Entire study population	Caucasian	African-American
n	126	94	32
Age (yr)	30.5 (7.0)	30.5 (6.8)	30.6 (7.6)
BMI (kg/m2)	38.0 (8.0)	37.0 (7.1)	41.0 (9.6)
Systolic blood pressure	125.5 (13.6)	124.9 (11.6)	127.3 (18.5)
Diastolic blood pressure	77.7 (11.2)	76.9 (9.8)	79.8 (14.4)
Fasting glucose (mmol/liter)	5.1 (1.0)	5.1 (1.0)	5.1 (0.8)
Fasting insulin (μU/ml)	18.0 (22.2)	15.6 (12.7)	24.9 (37.9)
2-h glucose (mmol/liter)	6.4 (2.3)	6.3 (2.3)	6.6 (2.2)
2-h insulin (μU/ml)	100.0 (111.7)	88.6 (98.8)	136.2 (141.6)
HOMA-IR	4.2 (6.0)	3.6 (3.0)	6.1 (10.4)

All characteristics are presented as mean (sd). 

Our primary objective was to compare HDL cholesterol between African-American and Caucasian PCOS women. After adjusting for age, BMI, and HOMA-IR. African-American women had significantly higher HDL cholesterol than Caucasian women (52.6 vs. 47.5 mg/dl, P = 0.019) (Table 2). African-American women also had significantly lower non-HDL cholesterol (134.1 vs. 154.6 mg/dl, P = 0.046) and lower TG (97.5 vs. 168.2 mg/dl, P < 0.001). There were no significant differences found in total cholesterol between the races (P = 0.090).

Table 2.

Comparison of lipid profiles between African-American and Caucasian PCOS women

Characteristic	Entire study population	Caucasian	African-American	P value
HDL cholesterol	48.8 (11.3)	47.5 (10.6)	52.6 (12.5)	0.019
Non-HDL cholesterol	149.4 (47.5)	154.6 (47.8)	134.1 (43.5)	0.046
Total cholesterol	198.2 (45.9)	202.1 (47.0)	186.7 (41.1)	0.090
TG	150.2 (103.1)	168.2 (110.4)	97.5 (49.8)	<0.001

All characteristics are presented as mean (sd). All units are milligrams per deciliter. P values comparing lipids across race are adjusted for age, BMI, and HOMA-IR. 

Discussion

Our study found that African-American women with PCOS had higher HDL cholesterol levels, lower non-HDL cholesterol, and lower TG when compared with their Caucasian counterparts. These differences could not be attributed to differences in insulin resistance (HOMA-IR), age, or BMI between the races.

HDL cholesterol

When comparing levels of HDL cholesterol between races in overweight and obese women diagnosed with PCOS, our study found that African-American women had a higher level of HDL cholesterol by 5.1 mg/dl when compared with Caucasians (Table 2). The clinical significance of point differences in HDL cholesterol levels and its translation into cardiovascular outcomes has not been determined. However, based on the most recent National Cholesterol Education Program Adult Treatment Panel III guidelines for women (9), which define low HDL cholesterol as less than 50 mg/dl, our Caucasian group had a low HDL cholesterol value at 47.5 mg/dl, whereas the African-American group did not, with HDL cholesterol levels at 52.6 mg/dl. Furthermore, low HDL cholesterol is one of five components in diagnosing metabolic syndrome, a clinical marker of high cardiovascular risk. The higher level of HDL cholesterol demonstrated in overweight and obese African-American women may diminish the likelihood that they will be categorized as having metabolic syndrome. This presents an apparent paradox because other data indicate that African-American women have higher cardiovascular mortality than Caucasians, a fact that should be reflected in any widely used clinical categorization scheme (10). This raises the question about whether current HDL cholesterol cutoff points may inappropriately exclude some high-risk individuals from a diagnosis of metabolic syndrome even though their actual cardiovascular risk warrants it. Because risk for cardiovascular mortality among women with PCOS and the role of metabolic syndrome in that risk are not well characterized, the clinical significance of this paradox is not clear. However, our data suggest that there may be racial differences in HDL cholesterol that could affect cardiovascular risk classification.

Multiple studies have shown that women with PCOS have depressed levels of HDL cholesterol when compared with control women (6,7). To our knowledge only one previous study has compared HDL cholesterol values between African-American and Caucasian PCOS women (11). Welt et al. (11) compared 172 Caucasians, 44 African-Americans, 25 Hispanics, and 21 Asians and found no racial differences in HDL cholesterol levels. However, our participants had higher BMI (participants with BMI <25 were excluded), and BMI in our groups were similar between the races (41 kg/m² for African-Americans vs. 37 kg/m² for Caucasians), whereas they differed in Welt’s study (36 kg/m² for African-Americans vs. 31 kg/m² for Caucasians). The BMI difference in our racial groups did not meet statistical significance, and our analyses controlled even for the nonsignificant BMI difference.

Non-HDL cholesterol

African-American PCOS women in our study had an average non-HDL cholesterol value 20 mg/dl lower than Caucasian participants (Table 2). This lipid fraction, made up of atherogenic lipid components, has been shown to predict cardiovascular mortality in a racially diverse cohort (12) and cardiovascular events in women (13) and to correlate with intimal medial thickness in young adults (6). It is thus an important indicator of cardiovascular risk, and the lower (more favorable) value found in African-American PCOS women appears to present the same paradox described for HDL cholesterol above.

TG

African-American PCOS women had lower random TG than Caucasian PCOS women in our study (Table 2). This result is consistent with multiple studies in non-PCOS, insulin-resistant populations that compare TG levels between racial groups (14,15). Our results are consistent with the findings in these studies and reinforce the critical point made by others (10,14) that lower TG (and higher HDL cholesterol) seen in African-American women may result in a missed diagnosis of metabolic syndrome and an underestimation of cardiovascular risk in this group.

Limitations

Although our study classified race by self-report, any racial category is limited because it does not reveal population level behavioral differences or account for individual differences within these racial classifications. Future studies should capture information that would allow a more detailed exploration of these issues, such as education, parity, physical activity, social class, alcohol intake, and smoking history. Cholesterol readings were not all fasting measurements, which could affect measurement for TG. However, recent literature suggests that non-fasting TG levels are associated with cardiovascular events, and lipid assessments may move toward nonfasting values (16). To have values present for subjects for each variable, we included 126 of the eligible 168 subjects. Although this smaller group represents more accurate data, the numbers of subjects in each racial group were uneven. Finally, due to the study population of an academic center, an unidentifiable selection bias may exist.

Conclusion

The comparison of lipid values between African-American and Caucasian PCOS women while controlling for BMI, age, and insulin resistance has not been previously examined. Our study found that at equal levels of insulin resistance (HOMA-IR) and BMI, African-American PCOS women have higher levels of HDL cholesterol, lower non-HDL cholesterol, and lower random TG. This pattern suggests a more favorable cardiovascular risk in African-American PCOS women, which contrasts with the higher observed cardiovascular mortality seen in African-American women overall, when compared with Caucasian women. This apparent discrepancy between risk factors and actual risk has implications for clinical care, particularly in the identification of individuals who should receive intensive preventive health care. The difference in lipid values when insulin resistance is held constant also highlights the existence of important modulators of HDL and non-HDL cholesterol other than insulin resistance. These data contribute to the discussion in the literature regarding whether the link between insulin resistance and metabolic syndrome is fundamental or may be different for African-American and Caucasian individuals.