| |
Prevalence and Predictors of Metabolic Syndrome Among HIV-Infected and HIV-Uninfected Women in the Women's Interagency HIV Study: metabolic syndrome 50% in HIV+ women
|
| |
| |
[Clinical Science] JAIDS July 2008
"In this study of more than 2000 HIV-infected and HIV-uninfected women, 1 of 3 HIV-infected women met the criteria for diagnosis of the MetSynd, a prevalence 50% higher than in the HIV-uninfected women. The specific components of MetSynd most influenced by HIV serostatus were presence of increased triglycerides and decreased HDL....In our cohort, abnormal HDL and abnormal waist circumference were the most frequent components of the MetSynd, but the increased prevalence of MetSynd among HIV-infected women was primarily due to differences in abnormal lipid profiles....We also identify strong associations between the MetSynd and HIV viral load itself and between MetSynd and BMI, age, smoking status, and specific antiretroviral agents.....nevirapine use was protective....we noted a protective effect of African American ethnicity among HIV-infected women, driven entirely by their higher HDL and lower triglyceride levels compared with white or Hispanic women. In our cohort, HIV-infected white and Hispanic women had a higher prevalence of abnormal HDL and triglycerides than did African American women. Similar trends have been reported in HIV-seronegative women...
....A key question for patients and providers is the effect of antiretroviral therapy on MetSynd. In our cohort, stavudine was independently associated with the MetSynd. Our findings of an association of stavudine use with MetSynd extend those of Anastos et al44 who reported an association of stavudine use with higher triglyceride levels (166 vs 141 mg/dL among HIV-infected nonusers, P = 0.002) but not with any of the other measured lipids. We did not, however, find any associations between PI use and MetSynd. Furthermore, when we examined specific PIs, lopinavir/ritonavir, and atazanavir/ritonavir, we did not find any significant association with MetSynd. This is in contrast to several studies of mostly European men that have found an association of PIs with MetSynd,23,34,37,39 most likely due to its adverse effect on fasting glucose and triglycerides.....
.....We did note, however, a protective effect of nevirapine, which has been previously described in the Multicenter AIDS Cohort Study.23 We postulated that this may be due to a previously described salutary effect of nevirapine on HDL-C49,50; in our cohort, prevalence of abnormal HDL was lower in nevirapine users compared with women not taking nevirapine (49.8% vs 65.5%, P < 0.0001). Interestingly, we also found that nevirapine users had a lower prevalence of abnormal glucose compared with women not on nevirapine (12.8% vs 18.4%, P = 0.036). This finding should be investigated further in other cohorts. Antiretroviral-naive status and PI-based or non-PI-based HAART were not predictive of the MetSynd in our cohort.....
....In our population of HIV-infected women, plasma HIV-1 RNA >50, 000 was associated with the MetSynd, whereas CD4 cell counts were not.....HIV infection induces chronic immune activation52,53 and may thereby lead to a proinflammatory state....The link between inflammation and diabetes, atherosclerosis, and the MetSynd has been well characterized in HIV-negative individuals.7,8,14,58-64 Further study is needed to determine whether a similar relationship exists among HIV-infected individuals and whether inflammation confers an additional risk factor in the development of MetSynd...."
JAIDS Journal of Acquired Immune Deficiency Syndromes:Volume 48(3)1 July 2008pp 272-280
Sobieszczyk, Magdalena E MD, MPH*; Hoover, Donald R PhD; Anastos, Kathryn MD; Mulligan, Kathleen PhD; Tan, Tianren MS_; Shi, Qiuhu PhD; Gao, Wei MS, MPH#; Hyman, Charles MD**; Cohen, Mardge H MD; Cole, Stephen R PhD; Plankey, Michael W PhD; Levine, Alexandra M MD__; Justman, Jessica MD; and the Women's Interagency HIV Study
From the *Division of Infectious Diseases, Department of Medicine, Columbia University College of Physicians and Surgeons, New York, NY; Department of Statistics and Institute for Health, Rutgers University, Piscataway, NJ; Departments of Medicine, Epidemiology, and Population Health, Montefiore Medical Center and Lincoln Medical and Mental Health Center, Bronx, NY; Department of Medicine, University of California, San Francisco, CA; _Data Solutions, LLC, Bronx, NY; School of Public Health, New York Medical College, Valhalla, NY; #Montefiore Medical Center, Bronx, NY; **Department of Medicine, Kings County Hospital Center, Brooklyn, NY; CORE Center, Cook County Bureau of Health Services and Departments of Medicine, John H. Stroger Hospital and Rush Medical College, Chicago IL; Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD; Department of Medicine, Georgetown University School of Medicine, Washington, DC; __Department of Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA; and the Mailman School of Public Health, Columbia University, New York, NY.
The probability of hepatitis C seropositivity was higher among the seropositive than seronegative women (30% vs 18%, P < 0.0001). Overall, 3% of both HIV-infected and HIV-uninfected women were using an antidiabetic medication at the index visit.....Among the HIV-infected women, the mean CD4 cell count was 474 cells/μL and mean HIV-1 RNA plasma level was 3.1 log10 copies/mL....The prevalence of MetSynd was significantly higher among the HIV-infected than HIV-uninfected women (33% vs 22%, P < 0.0001) (Table 2). HIV-infected women had higher prevalence of elevated triglycerides (35% vs 16%, P < 0.0001), low HDL-C (63% vs 41%, P < 0.0001), and elevated blood pressure (32% vs 26%, P = 0.003). In HIV-infected compared with HIV-uninfected women, mean triglyceride level was higher (154 vs 101 mg/dL, P < 0.0001) and mean HDL-C was lower (46 vs 55 mg/dL, P < 0.0001). By contrast, the prevalence of abnormal waist circumference and elevated glucose was similar in the 2 groups, and there were no differences in mean fasting glucose or blood pressure measurements between HIV-infected and HIV-uninfected individuals....MetSynd was less frequent among African American women (31%) than among white (42%) or Hispanic women (34%, P = 0.030). Of note, the use of particular antiretroviral agents did not differ significantly between HIV-seropositive African American compared with white and Hispanic women.....Although African American women had a lower prevalence of hypertriglyceridemia (27% vs 52% for white, P < 0.0001) and low HDL (58% vs 71%, for white P < 0.0001), they had higher prevalence of elevated blood pressure (38% vs 26% and 24% for white and Hispanic, respectively, P < 0.0001). There were no differences in prevalence of abnormal waist circumference or abnormal glucose by ethnic or racial category.....Nevirapine use (OR = 0.75, P = 0.016) was protective.....There was a higher prevalence of MetSynd in women with HIV-1 RNA >50,000 copies/mL (OR = 1.36, P = 0.019). The use of ritonavir-boosted PIs was significantly associated with MetSynd in univariate (OR = 1.26, P = 0.002) but not multivariate analysis (OR = 1.15, P = 0.134). HAART use, with or without a PI, was not associated with MetSynd.....Among the HIV-seronegative women, older age, higher BMI, current smoking, and higher parity were significantly associated with MetSynd in multivariate models. There was no significant association of MetSynd with race/ethnicity among HIV-seronegative women.....
....HIV serostatus was independently and significantly associated with MetSynd in univariate (OR = 1.69, P < 0.0001) and multivariate analyses (OR = 1.79, P < 0.0001) with HIV-1-seropositive and HIV-1-seronegative women combined (Table 4). Other factors significantly associated with MetSynd in the multivariate analysis among all women in the study group included white and Hispanic race/ethnicity compared with African American (OR = 1.97, P < 0.0001 for white and OR = 1.42, P = 0.0002 for Hispanic); older age (OR = 1.38 per 5 year increase, P < 0.0001); and higher BMI (OR = 2.20 for BMI 26-30 and OR = 6.54 for BMI >30 kg/m2 vs ≦21-25, P < 0.0001). Hepatitis C infection was not associated with MetSynd in either serogroup in multivariate analysis.
Abstract
Objectives: To assess the prevalence of metabolic syndrome (MetSynd) among participants of the Women's Interagency HIV Study and to describe the association of MetSynd with HIV infection, antiretroviral therapies, and sociodemographic factors.
Methods: Prevalence of MetSynd, defined by updated Adult Treatment Panel III guidelines, was assessed among 2393 (1725 seropositive and 668 seronegative) participants from the Women's Interagency HIV Study seen between October 2000 and October 2004.
Results:
HIV-1 infection was independently associated with MetSynd [33% vs 22%, P < 0.0001 in HIV-seropositive compared with HIV-seronegative women; adjusted odds ratio (OR) 1.79 (95% confidence interval 1.48, 2.16)]. HIV-infected women had higher mean triglyceride (154 vs 101 mg/dL, P < 0.0001) and lower mean high-density lipoprotein cholesterol levels (46 vs 55 mg/dL, P < 0.0001). Most notable factors associated with higher prevalence of MetSynd among HIV-infected women included older age (OR = 1.38 per 5 year increase, P < 0.0001); higher body mass index; current smoking; HIV-1 RNA (OR = 1.36, P = 0.019, for >50,000 vs <80 copies/mL); and use of stavudine (OR = 1.28, P = 0.009). Nevirapine use was protective (OR = 0.75, P = 0.016). There was no significant association of MetSynd with ritonavir-boosted protease inhibitors (OR = 1.15, P = 0.134).
Conclusions: MetSynd is more prevalent in HIV-seropositive than HIV-seronegative women. This increased prevalence was due to dyslipidemias rather than higher blood pressure, glucose, or waist circumference.
INTRODUCTION
Although potent antiretroviral therapy has reduced HIV related morbidity and mortality, concerns about treatment-related metabolic complications persist.1-4 These complications resemble metabolic and body composition abnormalities of the metabolic syndrome (MetSynd) described with increasing frequency in the general adult population.5,6 The MetSynd is clinically defined as having at least 3 of the following 5 conditions: impaired fasting glucose, increased waist circumference, elevated triglycerides, low high-density lipoprotein cholesterol (HDL-C), and hypertension.7 This constellation of metabolic and physical abnormalities is frequently associated with increased risk of insulin resistance and cardiovascular morbidity and mortality in the general population.7-11
There is considerable controversy about the completeness and precision of the current MetSynd definition and whether certain risk factors associated with proinflammatory states should be included in the definition.12,13 Despite this controversy, most agree that this constellation of risk factors predicts increased future risk of diabetes, cardiovascular events, and other complications.12-16 Current MetSynd criteria may therefore identify individuals who could benefit from lifestyle modification and in some cases pharmacologic interventions to delay the progression to diabetes and cardiovascular disease.
According to recent data from National Health and Nutrition Examination Survey III (NHANES III), the overall age-adjusted prevalence of MetSynd in adults increased from 29% to 32% over the last decade; women in the 20-39 age group experienced the most dramatic change, with MetSynd almost doubling in prevalence from 11% to 19%.17 In addition, data from NHANES III also indicate ethnic differences in the distribution of the syndrome; the highest prevalence was noted among Mexican American (36%) compared with white women (23%).18 This increase in MetSynd prevalence was largely ascribed to an increase in the prevalence of hypertriglyceridemia and high blood pressure. Other population-based studies of US adults have reported an association of high baseline body mass index (BMI), older age, and lack of alcohol intake with an increased risk for the MetSynd.19-21
There are limited data about the prevalence of the MetSynd among HIV-infected persons, but it seems to be higher than in the general population.22,23 Recent data from the Data Collection on Adverse Events of Anti-HIV Drugs (DAD) study, which does not include HIV negative participants, found that men were more likely than women to have MetSynd and that therapy with nonnucleoside reverse transcriptase inhibitors (NNRTIs) may be protective.24 Few studies have evaluated the risks and prevalence of MetSynd in the growing population of HIV-infected women. We, therefore, examined the prevalence of MetSynd among participants of the Women's Interagency HIV Study (WIHS) and assessed the association of MetSynd with HIV status, use of particular antiretroviral medications, and sociodemographic factors.
DISCUSSION
In this study of more than 2000 HIV-infected and HIV-uninfected women, 1 of 3 HIV-infected women met the criteria for diagnosis of the MetSynd, a prevalence 50% higher than in the HIV-uninfected women. The specific components of MetSynd most influenced by HIV serostatus were presence of increased triglycerides and decreased HDL. Among the HIV-infected women, elevated HIV-1 RNA levels, stavudine use, white race, and Hispanic ethnicity were independently associated with the MetSynd. In addition, nevirapine and light to moderate alcohol use were protective. Although heavy alcohol use was not statistically protective, this could reflect the small number of persons in this group. In our cohort, older age was associated with the MetSynd, perhaps explained by increased prevalence of cholesterol and glucose abnormalities with age.28,29 We note that there were no significant differences in the proportion of HIV-infected and HIV-uninfected women with 1 or 2 components of the MetSynd. The prevalence of abnormal glucose (including diabetes) was very similar in the 2 serogroups. In addition, the prevalence of abnormal waist circumference in the HIV-seropositive women (42%) was nearly identical to that in the HIV-seronegative women (43%).30 This is similar to findings reported by others who noted that rates of central hypertrophy were comparable among HIV-seropositive and HIV-seronegative women. Our finding of similar prevalence of these 2 components in both serogroups suggests that they are important in the development of MetSynd and may be related to one another. In fact, a correlation between abdominal obesity and insulin resistance in HIV-seropositive and HIV-seronegative individuals has been well described by others31-33; although we did not assess insulin resistance, impaired fasting plasma glucose and diabetes are part of the same spectrum.8 Of note, and in contrast to recent studies, in the multivariate models, HAART or ritonavir-boosted PI use was not significantly associated with the MetSynd.34
Previous literature on MetSynd among HIV-1-seropositive persons reports an increased prevalence compared with seronegative controls. Most of the data, however, pertain to predominantly male cohorts. Estimates range from 17% to 45% depending on the population studied and definitions used.22,23,28,32,34-38 Palella et al,23 for example, reported a prevalence of 33% (OR = 1.43, P = 0.002) among HIV-infected men in the Multicenter AIDS Cohort Study compared with 27% among the HIV-negative controls, with increased waist circumference the most frequently occurring component in both HIV-infected and HIV-uninfected men. Among HIV-infected men, factors associated with MetSynd included older age and HAART use. In a cross-sectional study of 477 HIV-infected men and women from the Nutrition for Healthy Living (NFHL) study,39 the overall prevalence of MetSynd was 24%, lower than what was noted in our cohort. A recently published study by Mondy et al40 found a similar prevalence of the MetSynd among HIV-1-infected men and women from an urban Midwestern outpatient clinic and matched control subjects from the NHANES cohort. Potential explanations for the different findings may include differences in racial/ethnic characteristics or the higher rate of obesity (BMI > 30 kg/m2) among women in our study group or maybe a gender difference in the effect of HIV on MetSynd. In the NFHL study, women accounted for only 25% of the cohort. The most common metabolic abnormalities in the NFHL study were low HDL (prevalence 54%) and high triglycerides (prevalence 47%).
In our cohort, abnormal HDL and abnormal waist circumference were the most frequent components of the MetSynd, but the increased prevalence of MetSynd among HIV-infected women was primarily due to differences in abnormal lipid profiles. Whereas ethnicity and race were not predictive for MetSynd among HIV-negative women, we noted a protective effect of African American ethnicity among HIV-infected women, driven entirely by their higher HDL and lower triglyceride levels compared with white or Hispanic women. In our cohort, HIV-infected white and Hispanic women had a higher prevalence of abnormal HDL and triglycerides than did African American women. Similar trends have been reported in HIV-seronegative women: The NHANES III data, for example, report that African American women have higher HDL-C and lower triglyceride levels than white and Hispanic women.41 Of note, the use of particular antiretroviral agents did not differ significantly between African American compared with white or Hispanic women in our cohort. The observed ethnic/racial differences in MetSynd prevalence are therefore not explained by race-based differences in antiretroviral utilization patterns at the index visit.
Of note, diabetes was the least frequently occurring component of MetSynd in our HIV-infected women. It is, however, important to note that we did not assess for insulin resistance in our cohort and that diabetes occurs later in the spectrum of glucose intolerance and insulin resistance. The use of fasting glucose levels to diagnose glucose intolerance may underestimate the prevalence of insulin resistance. In our cohort, hepatitis C, previously associated with insulin resistance and/or diabetes,42 was not associated with MetSynd in either serogroup. As MetSynd itself predicts the development of diabetes in the general population,43 further study is needed to see whether this is among the HIV-seropositive populations, including those with HIV and hepatitis C virus coinfection.
A key question for patients and providers is the effect of antiretroviral therapy on MetSynd. In our cohort, stavudine was independently associated with the MetSynd. Our findings of an association of stavudine use with MetSynd extend those of Anastos et al44 who reported an association of stavudine use with higher triglyceride levels (166 vs 141 mg/dL among HIV-infected nonusers, P = 0.002) but not with any of the other measured lipids. We did not, however, find any associations between PI use and MetSynd. Furthermore, when we examined specific PIs, lopinavir/ritonavir, and atazanavir/ritonavir, we did not find any significant association with MetSynd. This is in contrast to several studies of mostly European men that have found an association of PIs with MetSynd,23,34,37,39 most likely due to its adverse effect on fasting glucose and triglycerides. Jacobson et al39 showed an increased risk of MetSynd among lopinavir/ritonavir users in the NFHL cohort. In a cross-sectional study of 710 HIV-infected patients from Spain, past (OR = 2.96, 95% confidence interval of 1.03 to 3.55) and current PI use (OR = 4.18, 95% confidence interval of 1.4 to 12.5) and stavudine use were associated with presence of MetSynd.34 Mondy et al,40 on the other hand, did not find differences between individuals with or without MetSynd with respect to type of antiretrovirals used, and protease use was associated only with higher triglyceride levels. It has been shown that PI has variable effect on the metabolic profile, such as, for example, an increase in HDL-C with nelfinavir45,46 and higher triglyceride levels with inclusion of ritonavir in the regimen,44 whereas stavudine has been shown to be associated with elevated triglycerides.44,47,48 These variable effects may be responsible for the lack of association between certain agents and the MetSynd, or there may be differences by sex or race in metabolic responses.
We did note, however, a protective effect of nevirapine, which has been previously described in the Multicenter AIDS Cohort Study.23 We postulated that this may be due to a previously described salutary effect of nevirapine on HDL-C49,50; in our cohort, prevalence of abnormal HDL was lower in nevirapine users compared with women not taking nevirapine (49.8% vs 65.5%, P < 0.0001). Interestingly, we also found that nevirapine users had a lower prevalence of abnormal glucose compared with women not on nevirapine (12.8% vs 18.4%, P = 0.036). This finding should be investigated further in other cohorts. Antiretroviral-naive status and PI-based or non-PI-based HAART were not predictive of the MetSynd in our cohort.
There are few data about the association of HIV-specific factors with the MetSynd, and the reports have been conflicting. Palella et al23 noted that men with a CD4 cell count <200 cells/μL were more likely to have the MetSynd, whereas Mondy et al36 reported that in a cohort of 472 predominantly male individuals, higher CD4 cell count was an independent predictor of MetSynd. Recent data from the Data Collection on Adverse Events of Anti-HIV Drugs (DAD) study cohort of over 10,000 HIV-infected individuals demonstrated that both CD4 cell count (OR 1.02 per 100 cells/μL increment) and HIV RNA (OR 1.04 per log10 increase) were associated with the MetSynd24; an association between higher risk of MetSynd and increasing viral loads was also reported by Jacobson et al.39 In our population of HIV-infected women, plasma HIV-1 RNA >50, 000 was associated with the MetSynd, whereas CD4 cell counts were not. It is possible that the chronic immune activation seen with untreated22,23 and treated24,25 HIV infection may explain the association of higher plasma HIV-1 RNA with the MetSynd,51 although data remain conflicting.39 HIV infection induces chronic immune activation52,53 and may thereby lead to a proinflammatory state, illustrated by increased levels of C-reactive protein, a marker of inflammation, in HIV-infected individuals.54,55 There are some data that this marker remains elevated despite potent and suppressive antiretroviral therapy56,57 and may be due to viral factors related to inflammation. The link between inflammation and diabetes, atherosclerosis, and the MetSynd has been well characterized in HIV-negative individuals.7,8,14,58-64 Further study is needed to determine whether a similar relationship exists among HIV-infected individuals and whether inflammation confers an additional risk factor in the development of MetSynd.
The strengths of our study rest in the fact that we report data for both HIV-infected and HIV-uninfected women from a large database of prospectively recorded variables required to define MetSynd. Moreover, the demographic profile of the WIHS cohort reflects the characteristics of HIV-infected women living in the United States, and these results are thus likely to be applicable to the general population of HIV-infected women in the United States. To our knowledge, this is the largest study evaluating the prevalence and risk factors of this syndrome among HIV-1-infected women, with the additional advantage of an HIV-negative comparison group.
Limitations of the study include the cross-sectional design that precludes inferences about causality. Longitudinal follow-up data can further elucidate associations of MetSynd with particular antiretroviral regimens and characteristics of HIV disease.
In conclusion, our study demonstrates that 33% of HIV-infected women meet criteria for diagnosis of MetSynd, a prevalence which is 50% higher than that seen in HIV-negative comparators, mainly due to an increased prevalence of high triglyceride and low HDL levels among the HIV infected. We also identify strong associations between the MetSynd and HIV viral load itself and between MetSynd and BMI, age, smoking status, and specific antiretroviral agents. These findings have implications for the care of HIV-infected women in terms of decisions regarding antiretroviral therapy and the importance of early interventions to modify lifestyle factors that may lead to development of abnormal lipid profiles. Further studies are needed to examine the incidence of the MetSynd over time and to elucidate the role of inflammation and its association with the MetSynd in this population.
|
|
| |
| |
|
|
|
