AIDS: Volume 19(7) 29 April 2005 Kamin, Daniel S; Grinspoon, Steven K

"……Taken together, these, and other data on hard outcomes suggest that a small but significant and increasing risk exists for cardiovascular disease related to HIV infection or ARV/PI use. Further studies are needed to clarify the contributions of infection, direct drug toxicity, and metabolic derangements to enhanced CVD risk in persons with HIV. The actual number of cardiovascular events remains small and must be placed in the overall perspective of the benefit of ARV therapies on immune function. However, the risk may rise with aging of the HIV population and longer duration of ARV exposure and metabolic abnormalities…..

….. Diabetes mellitus is a major risk factor for cardiovascular disease. Similarly, impaired glucose tolerance and insulin resistance are associated with increased risk of cardiovascular disease in the general population. Treatment of diabetes in the HIV-infected patients should also proceed along established lines, with the caveat that specific drug toxicities may push an individual with compensated insulin resistance into frank type II diabetes. In such a situation, modification of the ARV regimen may be reasonable in parallel with dietary modification and use of insulin-sensitizing agents……"

Use of highly active antiretroviral therapy (HAART) is associated with the development of traditional cardiovascular risk factors, including dyslipidemia and insulin resistance. Recent data also suggests that endothelial dysfunction, impaired fibrinolysis, and excess inflammation may contribute to increased cardiovascular risk in the HIV-positive population. Surrogate markers such as C-reactive protein (CRP), tissue plasminogen activator (tPA), and plasminogen activator inhibitor-1 (PAI-1) are increased in the HIV-positive population in association with metabolic abnormalities and altered fat distribution. Carotid intimal-medial thickness (IMT) and coronary calcification assessments suggest increased atherosclerotic disease among certain HIV-positive individuals. Finally, recent studies of large numbers of HIV-positive subjects have documented increased cardiovascular risk using hard endpoints such as myocardial infarction rates. This review will consider the pathogenesis, prevalence, and treatment of cardiovascular risk factors in the HIV-positive population. Pharmacologic strategies for dyslipidemia and abnormal glucose homeostasis will be reviewed in conjunction with non-drug strategies for disease prevention.

Altered lipid metabolism is known to occur in association with HIV disease itself. HDL cholesterol (HDL-C), LDL cholesterol (LDL-C), and total cholesterol (TC) levels are reduced, while triglyceride (TG) levels are increased in HIV-positive patients [1]. Early data suggest that circulating TG levels may be elevated from increased hepatic very low-density lipoprotein (VLDL) production and reduced clearance [1,2]. Cytokines, such as interferon alpha, may play a role in the abnormal lipid homeostasis seen in HIV-positive patients [3,4].

Lipid levels are also influenced by female sex, estrogen use, age, and heritable factors, including lipoprotein genotype in HIV-positive patients [5]. Various HIV-related factors such as viral load, CD4 count, proximity/severity of opportunistic infections, and mechanism of HIV acquisition appear to have modest or minimal direct affects on atherosclerotic lesions [6].

Protease inhibitor (PI) use has been linked to further abnormalities in the serum lipid profile in HIV-positive patients [5,7]. Increased TC, TG rich VLDL, and LDL-C are seen in PI-treated patients relative to PI-naive, HIV-positive individuals [5,8,9]. Data from prospective cohort studies report new-onset hypercholesterolemia and hypertriglyceridemia after 5 years of HAART therapy in 24 and 19% of subjects, respectively [10,11].

In contrast, recent longitudinal data from the Multicenter AIDS Cohort Study (MACS) reinforced that HIV infection itself is associated with reduced HDL-C, TC, and LDL-C. Cholesterol and LDL-C increase with use of HAART to pre-infection levels whereas HDL-C remains low [12].

Individual PIs likely have substantially different affects on the lipid profile. For example, data from the Swiss Cohort study suggest that ritonavir, but not indinavir or nelfinavir, is associated with increased TG levels [10]. Purnell et al. [13] demonstrated significant effects of ritonavir on TG levels after 2 weeks in HIV-negative patients. Similarly, low dose ritonavir in combination with lopinavir over 4 weeks also increased TG levels (but not LDL-C) in HIV-negative men [14]. The newer PI atazanavir appears to have a significantly less pronounced effect on serum lipid levels [15,16]. The mechanisms by which PIs (notably ritonavir) influence serum TGs in HIV-infected patients have not been clearly defined. Preliminary data from rat models [17,18] suggest that PIs may prevent proteosomal degradation of nascent apolipoprotein B, a principle protein component of circulating TGs, leading to increased production of VLDL particles. Furthermore, alterations in apolipoprotein B metabolism may be explained in part by ritonavir-induced intra-hepatocyte accumulation of nuclear transcription factors such as sterol regulatory binding proteins [19], resulting in up-regulation of metabolic pathways and excessive production of VLDL.

Other antiretroviral (ARV) medications may affect serum lipids. Kumar et al. [20] reported on a 48-week randomized trial of three different ARV regimens in 258 treatment-naive HIV-positive subjects. PI sparing regimens (zidovudine/lamivudine + abacavir) raised fasting TC and TG least in comparison with regimens containing a PI (zidovudine/lamivudine + nelfinavir) or stavudine and a PI. In a 48-week open-label randomized trial of the non-nucleoside reverse transcriptase inhibitors (NNRTIs) nevirapine or efavirenz in 1216 treatment-naive HIV-positive individuals, van Leth et al. [21] found that nevirapine versus efavirenz-containing regimens were associated with significantly higher HDL-C and lower TC to HDL-C ratios. An earlier study by van der Valk et al. [22] demonstrated similar beneficial effects on HDL-C for subjects receiving the nevirapine versus PI-containing regimens. More recently, cross-sectional data from the Data Collection on Adverse Events of Anti-HIV Drugs (DAD) study (n = 7483 patients) [23] reinforce the notion that exposure to NNRTIs is associated with modest yet significantly increased TG (> 200 mg/dl) levels (odds ratio, 1.90; 95% confidence interval, 1.06-3.39), but not with low HDL-C (< 40 mg/dl) or increased LDL-C (> 160 mg/dl). Taken together, the data indicate that PI-sparing regimens with nevirapine and without stavudine may have the least adverse influence on serum lipids in HIV-positive individuals.

Abnormalities in glucose homeostasis are common among HIV-positive individuals treated with HAART. HIV-positive individuals may be diabetic, or more commonly, demonstrate impaired glucose tolerance and insulin resistance. Use of the WHO definitions (fasting glucose > 126 mg/dl defines diabetes, and fasting glucose > 110 mg/dl impaired fasting glycemia) is recommended to characterize glucose abnormalities in the HIV-positive population [24]. The glucose response to a standard 75 g glucose challenge may also be useful to identify individuals with impaired glucose tolerance. Insulin resistance is common during HAART therapy, and may manifest as fasting hyperinsulinemia [25] or reduced glucose disposal after an oral or IV glucose challenge [26]; fasting glucose need not be elevated.

The prevalence of diabetes is increased among HIV-infected patients treated with HAART. Recent data [27] from the MACS cohort using the WHO criteria found diabetes in 14% of men, with an odds ratio of 4.4 after adjustment for age and body mass index (BMI). The hazard ratio for developing diabetes over 3 years of follow-up was 3.1 versus the control HIV-negative population. Exposure to a PI-containing HAART regimen, stavudine or efavirenz, were each independently associated with the development of diabetes. The prevalence of insulin resistance among those treated with HAART is not known, but data suggest that up to 60% of those treated may be affected [5,28], depending on the criteria and techniques used.

Mechanisms of insulin resistance in the HIV-positive population are not known, but may relate to altered nutrient metabolism, changes in body composition, and/or direct effects of antiviral agents. Preliminary data include the demonstration of altered lipolysis and increased serum free fatty acids in HIV-positive individuals [29,30]. Excess free fatty acids in the circulation may reduce insulin sensitivity through inappropriate lipid storage in muscle [31] and liver [32], resulting in impaired glucose utilization [33] and insulin-mediated inhibition of glycogenolysis and gluconeogenesis [34].

PI therapy is associated with higher rates of diabetes mellitus, impaired glucose tolerance, and hyperinsulinemia among HIV-positive individuals [35]. This effect is not class specific, and newer PI drugs such as atazanavir appear less likely to affect glucose homoestastis [36]. A direct link between ARV medication and abnormal glucose homeostasis is substantiated by physiologically rigorous evaluation of indinavir [37], and more recently, lopinavir/ritonavir [14] in HIV-negative individuals.

The mechanisms behind PI-induced insulin resistance are complex and multi-factorial. Initial studies suggested an effect on Glut-4-mediated glucose transport [38,39]. Islet cell dysfunction [40] and dysregulated hepatic glucose production [41] may complicate glucose homeostasis further in HIV-positive individuals receiving HAART. Preliminary data suggest that PIs may inhibit the processing of insulin from pro-insulin [5]. Mitochondrial toxicity could contribute to the detrimental effect of NRTIs on tissue insulin sensitivity either through impaired oxidative phosphorylation and excess lipid accumulation in liver or muscle or via a reduction in the absolute or relative amounts of subcutaneous fat.

NRTIs use can be linked to abnormal glucose homeostasis as well. These effects may be indirect through chronic changes in fat distribution [25,42]. Furthermore, stavudine administration may lead to changes in lipolysis, resulting in increased serum free fatty acids and decreased insulin sensitivity [30]. A similar independent effect of stavudine on insulin sensitivity was also noted in the MACS cohort [27].

Abnormal body fat partitioning among HIV-positive individuals strongly predicts insulin resistance and/or glucose intolerance. In a study comparing serum glucose and insulin responses to a standard oral glucose challenge in HIV-positive subjects (n = 71) with evidence of fat redistribution by clinical exam [26], the prevalence of DM was 7% compared to 0.5% (P = 0.01 for difference) in age and BMI-matched Framingham subjects (n = 213). The prevalence of impaired glucose tolerance (2 h serum glucose > 140 mg/dl) was markedly higher at 35% in the lipodystrophic group compared to 5% (P = 0.001) in the matched Framingham subjects [26]. As in the general population, excess trunk or visceral fat is a risk factor for insulin resistance among those with HIV infection, independent of PI use [26,43]. In addition, insulin resistance is itself independently associated with fat loss in HIV-positive individuals [44,45].

Additional factors can influence the development of insulin resistance in HIV-infected patients, including increasing age and hepatitis C co-infection [28,35,46]. In sum, the particular cause(s) of insulin resistance in any one patient will vary depending on many of the factors noted. Assessment of glucose homeostasis in HIV-infected patients should include measurement of periodic fasting glucose in all subjects, particularly in those with risk factors for the development of diabetes or in whom ARV regimens have changed or are about to change. Performance of glucose tolerance testing may also be useful in patients with significant risk factors for diabetes mellitus.

HIV-positive individuals receiving HAART commonly manifest evidence of fat redistribution, characterized by loss of subcutaneous extremity fat, relative preservation of fat in the trunk, and an increased waist-to-hip ratio (WHR) [28,43,47]. It is well established that increased truncal adiposity confers heightened risk of cardiovascular complications in the general population [48]. Intra-abdominal fat delivers excess free fatty acids directly into the portal blood system [49] and secretes cytokines and other factors that contribute to insulin resistance, impaired fibrinolysis [50,51], and endothelial dysfunction [52].

Excess truncal adiposity is an established risk factor for insulin resistance in HIV-negative individuals. In the HIV-positive population, increased WHR is often a function of increased waist and reduced hip circumference, and is associated with increased metabolic risk indices including hyperlipidemia [53]. In the general population, advancing age [54], male gender [54], and ethnicity [55,56] also predict excess abdominal adiposity, and are likely to contribute to body composition changes in HIV-positive patients.

While it is unclear if changes in trunk and subcutaneous fat are linked as part of a single syndrome [57], evidence from prospective studies in treatment-naive subjects suggest such changes may occur simultaneously. Mallon et al. [58] prospectively studied 40 HIV-positive, therapy naïve men for 96 weeks and demonstrated decreased extremity fat and increased relative central fat accumulation. Similarly, Dube et al. [59] demonstrated increased trunk fat and reduced extremity fat in response to NNRTI and NRTI regimens, respectively, with reduced extremity fat most clearly linked to the PI nelfinavir and an NRTI regimen containing stavudine. The relative risk of depleted extremity subcutaneous fat with respect to cardiovascular disease (CVD) is not known, but Hadigan et al. [53] calculated a higher predicted risk of myocardial infarction in subjects with predominant lipoatrophy.

Age and smoking appear to behave similarly as risk factors for atherosclerosis when comparing HIV-negative and HAART-treated HIV-positive individuals [6]. Some studies report that smoking rates among HIV-positive individuals are as high as 60%, making smoking a substantial modifiable cardiovascular risk factor in this population [60-62].

Significant evidence suggests that markers of impaired fibrinolysis are increased in HIV in association with insulin resistance and various altered metabolic pathways. The anti-fibrinolytic factor PAI-1 is increased in association with insulin-resistance [63] and correlates with risk of myocardial infarction [64]. Similarly, increased tPA is associated with myocardial infarction and stroke in HIV-negative individuals [65,66]. Increased homocysteine concentrations also correlate with excess cardiovascular risk [67]. Among HIV-positive subjects treated with HAART, homocysteine [68], tPA [69]and PAI-1 [70] are increased. Metformin treatment decreased serum tPA and PAI-1 activity in patients with HIV lipodystrophy, in association with improvement in insulin sensitivity [69].

The pathogenesis of atherosclerosis now includes chronic systemic inflammatory activity. As such, C-reactive protein (CRP) has become a clinical tool used to predict complications from atherosclerotic disease [71]. Only limited studies have assessed CRP in HIV-infected patients. In a recent study, increased CRP predicted cardiovascular mortality in HIV-positive women enrolled in the Women's Interagency HIV Study [72]. Dolan et al. demonstrated increased CRP as a function of increased central fat accumulation in HIV-positive women, controlling for ARV use and immune function [73].

Adiponectin, an abundant gene product in adipose tissue [74,75], has anti-diabetic properties and may affect the insulin signaling protein insulin receptor substrate-1 [76,77], up-regulate muscle β-oxidation [76], and suppress hepatic glucose production [78]. Reduced adiponectin levels have been identified in many insulin-resistant states such as obesity [79], diabetes [79], and inherited lipodystrophy syndromes [80]. Adiponectin also has anti-inflammatory properties. It suppresses inflammatory cell infiltration of the vascular intimal space in animal and cell culture models of atherogenesis [81-83], and deficiency of adiponectin up-regulates endothelial adhesion molecules [82].

Preliminary data suggest that reduced adiponectin concentrations may increase the risk of myocardial infarction, but this relationship has not been studied in HIV [84]. Cross-sectional studies have identified significant independent correlations between low adiponectin and traditional and non-traditional CVD risk such as insulin resistance [85-89], excess visceral adiposity [89], depleted peripheral subcutaneous fat [90], disadvantageous lipid profiles [89], higher liver fat content [86], and higher circulating levels of CRP and TNF-α [87,88]. Future therapeutic strategies may include treatment for low adiponectin if reduced concentrations consistently predict excess cardiovascular disease. The thiazolidinedione-class of insulin-sensitizing agents increase levels of adiponectin expression in HIV-infected patients with lipodystrophy [91-93] and may be useful in this regard (see treatment below).

Studies investigating cardiovascular disease in the HIV-positive population generally fall into three categories: (1) risk modeling to predict outcome in HIV-infected individuals; (2) analysis of surrogate markers known to be associated with increased cardiovascular disease in the general population (e.g. increased carotid IMT or markers of endothelial dysfunction); and (3) evaluation of hard cardiovascular endpoints including stroke and myocardial infarction rates.

The Framingham score is a calculated measure of cardiovascular disease risk validated in the general population [94]. The Framingham score takes into account six parameters: age, sex, TC, HDL-C, systemic blood pressure, diabetes, and smoking, and has been shown to accurately predict myocardial infarction rates in longitudinal assessment of Framingham Study participants [94]. Egger et al. [95] used the Framingham score to determine risk of cardiovascular disease among individuals receiving HAART. While age, sex, and smoking were the primary determinants of risk, estimated treatment-induced complications added substantial risk. For instance, a 50-year-old male non-smoker receiving HAART with and without metabolic complications had a calculated 10-year CVD risk rate of 26.5 and 13%, respectively. Comparable values for women were 18.1 and 5.5%.

Hadigan et al. also used the Framingham Equation to predict CVD risk in patients selected for evidence of fat redistribution [53]. Thirty percent of HIV-positive subjects with fat redistribution versus 13% of an age- and BMI-matched HIV-negative control group (P = 0.001 for difference) demonstrated a significant 10-year risk of myocardial infarction [53]. When HIV- positive subjects were matched to HIV-negative controls for WHR, the 10-year risks were no different, suggesting that increased CVD risk was associated with fat redistribution in this population. Sub-analysis revealed an increased 10-year risk of CVD in subjects with lipoatrophy in comparison to subjects with central lipohypertrophy or mixed lipodystrophy [53].

Vascular imaging studies represent a robust means of evaluating cardiovascular risk in HIV-positive individuals. B-mode high-resolution ultrasound imaging of peripheral arteries is used to quantify vessel thickness and presence of plaques within the carotid and femoral arteries. Significant work has validated these methods in the general population [96,97]. Flow-mediated dilation (FMD) of the brachial or carotid artery, assessed by measuring flow with high-resolution ultrasound after variable lengths of arterial occlusion, is also considered an early surrogate marker for endothelial dysfunction and subsequent plaque formation [98-100], as is response to vasodilators [101].

The full clinical implications of vascular imaging findings in HIV-positive subjects is not yet clear. Depairon et al. [6] studied 168 HIV-infected men and women, 136 of whom had received PIs. In comparison with HIV-negative individuals, the HIV-positive group manifested more plaque lesions in the carotid or femoral arteries (55 versus 38%; P = 0.03). Protease inhibitor therapy did not independently predict lesion number, whereas traditional factors such as age, male gender, LDL-C, and smoking did. Hsue et al. [102] used carotid IMT to corroborate these findings, while also reporting a significantly higher rate of IMT thickness progression over 1 year in the 121 HIV-positive subjects versus 27 age and sex matched HIV-negative controls (average progression 0.074 mm ± 0.13 mm versus -0.006 mm ± 0.05 mm, respectively; P = 0.002 for difference). Significant predictors of IMT progression in the HIV-positive subjects included age and Hispanic ethnicity. In contrast, Maggi et al. [103] evaluated 104 HIV-infected individuals receiving (n = 55) and naive (n = 47) to PIs, and compared them with age and BMI-matched HIV-negative individuals. More than half of those receiving PIs demonstrated abnormal carotid artery IMT in comparison with 14.9 and 6.7% among PI-naive and healthy control subjects, respectively. Seminari et al. [104] reported similar results, with increased carotid disease in HIV-infected subjects receiving PI therapy.

Stein et al. [105] demonstrated impaired FMD in HIV-positive subjects receiving PIs in comparison with PI-naive subjects. The difference in FMD was explained by abnormal lipoprotein concentrations. Using FMD, Nolan et al. [106] found that amplitude of immune reconstitution (assessed via percentage of T cells with the 'naive' marker CD44RA) correlated negatively with FMD, whereas traditional risk factors had no affect. Different study populations, techniques, statistical power, and complex interactions between multiple risk factors may complicate comparison between studies. Taken together, molecular and clinical studies performed to date provide initial evidence for vascular disease and endothelial dysfunction among HIV-positive patients. Further studies are needed to assess abnormal endothelial function and altered vascular biology in HIV-positive individuals, focusing on the relative contributions of HIV itself, metabolic factors, and pharmacologic toxicity.

Initial case reports and series [107-109] suggested that increased cardiovascular disease could be caused by HAART. Data from large-scale retrospective studies are mixed in terms of whether use of ARV medications contributes to increased myocardial infarction rates in the HIV population.

Klein et al. [110] used retrospective data from the Kaiser Permanente Medical Care Program of Northern California database to show that, with 3.6 mean years and 14 823 person-years of follow-up, the myocardial infarction rate was higher among HIV-positive adult men versus HIV-negative matched controls (6.5 events per 1000 person-years versus 3.8, P = 0.003 for difference). Importantly, among male health plan members with HIV, PI and ARV medication use did not influence myocardial infarction rates.

Mary Krause et al. [111] analyzed myocardial infarction rates among 34 976 HIV-positive men from the French Hospital Database on HIV who were followed for a median of 33 months (88 029 person-years). This retrospective study found that the myocardial infarction rates among men increased in relationship to duration of PI use (< 18 months PI use: 10.8 events per 10 000 person-years; 18-29 months PI use: 15.9 events; > 30 months PI use: 33.8 events; expected: 1.1 events). Mary-Krause et al. did not report data on myocardial infarction rates in HIV-positive, therapy-naive individuals. Retrospective analysis from the HIV Outpatient Study cohort [62] demonstrated similar findings, although there was no relationship between length of exposure to PI and risk of myocardial infarction.

Currier et al. [112] investigated coronary heart disease rates based on administrative claims data for the greater California Medicare (Medi-Cal) population. The incidence of CVD among young men (up to age 34) and women (up to age 44) was significantly higher in HIV-positive compared with HIV-negative patients. The covariate-adjusted relative risk for the development of CVD in individuals receiving ARV medications compared with those not receiving medications was 2.06 (P < 0.001) in HIV-positive individuals aged 18-33 years. Notably, increased risk of CVD was not found for older HIV-positive patients.

In contrast, a large retrospective study using a Veteran's Affairs database for the years 1993 and 2001 (36 766 patients, 122 000 patient-years) found that PI use was not associated with increased risk of myocardial infarction [113]. However, the median amount of time study subjects had received a PI was 16 months. Although the study did not compare event rates to age-matched HIV-negative controls, admission rates and deaths from vascular events decreased over time between 1995 and 2001 among the HIV-positive cohort.

Prospective studies have recently been published which were specifically designed to detect adverse CVD outcomes. The DAD investigators prospectively examined patient data collected at multiple sites throughout the world between the years 1999 and 2002 [61]. A total of 36 199 person-years were considered, and 126 individuals had myocardial infarction (MI). The incidence of MI increased in association with increasing exposure to combination ARV therapy, with a 26% relative increase in MI rate per additional year of study, adjusting for age, sex, body mass index, race, family history, smoking, and prior cardiovascular disease. Age, male sex, previous cardiovascular disease and smoking also independently predicted myocardial infarction in the DAD population. Current or past smoking (56% of the study subjects) was associated with an increased myocardial infarction rate of 2.17 [95% confidence interval (CI), 1.30-3.62], substantially higher than the medication effect. Controlling for TC and TG reduced the ARV-associated relative rate of MI down to 1.16 (95% CI, 1.02-1.33), whereas the presence of subjective lipodystrophy, diabetes, BMI, and HIV disease parameters did not affect the MI rate. The annual rate of myocardial infarction was low, 0.6%, and many more subjects died of traditional HIV-related complications than cardiovascular diseases. Furthermore, missing data and unmeasured potential co-variates, such as WHR and measures of insulin resistance, may have contributed to the observed MI rates.

Taken together, these, and other data [114] on hard outcomes suggest that a small but significant and increasing risk exists for cardiovascular disease related to HIV infection or ARV/PI use. Further studies are needed to clarify the contributions of infection, direct drug toxicity, and metabolic derangements to enhanced CVD risk in persons with HIV. The actual number of cardiovascular events remains small and must be placed in the overall perspective of the benefit of ARV therapies on immune function. However, the risk may rise with aging of the HIV population and longer duration of ARV exposure and metabolic abnormalities.

Treatment of HIV-positive individuals with established cardiovascular disease should proceed according to standard guidelines. To the extent that specific ARV medications are thought to contribute to increased lipid or glucose levels, consideration can be given to changing ARV medications and/or using traditional lipid- or glucose-lowering strategies, depending on the circumstances. Until definitive data are available on the efficacy of lipid-lowering and insulin-sensitizing pharmacologic strategies, the primary focus of treatment should be on lifestyle modification, including diet, exercise, and smoking cessation. Changing HIV medications can be difficult and consideration should also be given to prevention of metabolic abnormalities with the choice of initial ARV regimen. The HIV-positive population is relatively young in comparison with patients targeted for CVD risk modification in the general population, but prolonged exposure to potentially atherogenic ARV medications and associated metabolic/inflammatory abnormalities further increase risk in this population.

Smoking has long been considered a major contributor to overall cardiovascular morbidity and mortality in the general population, and CVD risk drops sharply after smoking cessation. Smoking is particularly common in the HIV population; the DAD study reported 56% of their cohort had been or were current smokers [61]. Prevention and smoking treatment programs may offer substantial benefit to HIV-positive patients. Moderate exercise (such as brisk walking) is known to reduce risk of myocardial infarction significantly in the general population. Recent work [115-117] found that routine aerobic activity and muscle conditioning improves trunk adiposity and lipid parameters for HIV-positive individuals. A regular exercise program might therefore be beneficial to modify cardiovascular disease risk in HIV-positive individuals.

In the general population, considerable debate centers on the appropriate lipoprotein values to initiate lipid-lowering therapy [118,119]. Traditionally, the National Cholesterol Education Program (NCEP) guidelines have based treatment on the LDL-C concentration and consideration of well-recognized CVD risk factors [120]. With regard to TGs, the major established adverse effect of severe hypertriglyceridemia is pancreatitis, most often seen with triglyceride > 1000 mg/dl. High TG levels may also increase the risk of developing CVD [121,122]. Current NCEP guidelines suggest that high (200-499 mg/dl) triglycerides be treated with medication when two or more risk factors are present; those with very high (³ 500 mg/dl) TGs should be medically treated regardless, especially if there is a history of pancreatitis or a drug regimen containing a substance known to cause pancreatitis (e.g. didanosine) [123].

Treatment of hyperlipidemia includes replacing ARV medications known to cause hyperlipidemia and/or using anti-lipid medications and lifestyle change to directly address high LDL-C and/or high TGs. Multiple 'switch' studies have found that substitution of NRTI or NNRTI medications for PIs can be beneficial without compromising anti-viral efficacy[124-133]. However, favorable effects have not been uniformly reported [132,134,135]. Drug treatment to reduce LDL-C and TGs needs to be balanced against the potentially significant drug-drug interactions (i.e. between ritonavir and specific HMG-CoA reductase inhibitors [136]) and lack of trials comparing the safety and efficacy of ARV switching and anti-lipid medications with regard to improving serum TGs and LDL-C. Nevertheless, in the HIV-positive population, specific drugs in the statin and fibrate class of medications improve LDL-C and TGs, respectively [137,138]. Detailed guidelines for the management of lipid abnormalities in HIV-infected patients have recently been published [123,139-141].

Diabetes mellitus is a major risk factor for cardiovascular disease. Similarly, impaired glucose tolerance and insulin resistance are associated with increased risk of cardiovascular disease in the general population [142,143]. Treatment of diabetes in the HIV-infected patients should also proceed along established lines, with the caveat that specific drug toxicities may push an individual with compensated insulin resistance into frank type II diabetes. In such a situation, modification of the ARV regimen may be reasonable in parallel with dietary modification and use of insulin-sensitizing agents.

A number of studies have now investigated the use of insulin-sensitizing agents in patients with HIV disease. Saint-Marc and Touraine evaluated metformin, a biguanide insulin-sensitizing agent, in a placebo-controlled study of 27 HIV-positive individuals with central adiposity after starting PI-containing HAART [144]. Insulin AUC (following a 75 g oral glucose tolerance test), visceral fat, triglycerides and BMI decreased significantly in the metformin-treated group compared with placebo after 8 weeks. Hadigan et al. [69,145,146] used a randomized, double-blind approach in HIV-positive men with lipodystrophy and also found improved insulin sensitivity and decreased trunk adiposity. Lactate levels were stable throughout the studies. Hadigan et al. also found that metformin treatment improved abnormal levels of serum markers (tPA and PAI-1) associated with increased cardiovascular disease risk. Taken together, these data suggest that metformin may improve insulin sensitivity, central adiposity, and other CVD risk factors in HIV-infected patients with truncal adiposity and insulin resistance. Individuals with azotemia, liver disease, or lactic acidosis associated with NRTI therapy should not receive metformin. Furthermore, metformin use is associated with weight loss and may be inappropriate for subjects with significant lipoatrophy. Driscoll et al. recently showed that the combination of progressive resistance/aerobic training and metformin was more efficacious than metformin alone in reducing WHR and insulin resistance in HIV-positive patients with fat redistribution [147].

Peroxisome proliferator-activated receptor agonists (the thiazolidinedione class of medications) have become important agents used to treat type II diabetes in non-HIV-infected individuals. Physiologic properties include stimulation of adipocyte differentiation and function as well as improvement (albeit indirectly) in skeletal muscle glucose uptake [148]. Decreased PPAR- expression is seen in subcutaneous fat from PI-treated human subjects [149]. Five studies have been completed to date using PPAR- agonists [91-93,150,151]. Gelato et al. [92] showed a significant 20% increase in abdominal subcutaneous fat in association with improved insulin sensitivity in a group of highly insulin-resistant lipoatrophic patients. Similarly, Hadigan et al. demonstrated a 23% increase in leg subcutaneous fat [assessed by computerized tomography (CT)] in lipoatrophic HIV-infected patients with insulin resistance in a 3-month randomized, placebo-controlled study [93]. In both studies, insulin resistance improved significantly as shown by euglycemic, hyperinsulinemic clamp. In addition, modest but significant negative effects on TC and TG were seen, consistent with the pattern seen in non-HIV-infected patients. Adiponectin rose consistently in response to rosiglitazone, which (see above) may also confer a metabolic advantage.

In contrast, Sutinen et al. [151] and Carr et al. [91] found no changes in abdominal subcutaneous fat or extremity fat in randomized, placebo-controlled studies. Neither study chose patients based on insulin resistance. Of note all studies demonstrated improvement in insulin sensitivity, which may by itself afford improved CVD risk. Sutinen et al. [151] also measured liver fat with quantitative magnetic resonance, and found decreased hepatic fat content with therapy in conjunction with decreased serum transaminases. Further studies are needed to define how the use of PPAR- agonists can best improve metabolic parameters among selected HIV-positive individuals.

Women with HIV infection represent a growing proportion of the total HIV population. However, major prospective and retrospective studies evaluating cardiovascular disease risk and HIV therapy have included relatively few women. Pre-menopausal women are relatively protected from cardiovascular disease, in part by the influence of estrogens on lipoprotein sub-fractions. Indeed, data from the DAD study [61] indicate that female sex is protective against the risk of myocardial infarction. However, women taking combination antiretroviral therapy develop metabolic and body composition changes consistent with the male-type pattern classically associated with increased CVD risk. Women with HIV are now living longer and HIV-positive women may experience increased CVD risk associated with aging and menopause. Pernerstofer-Schoen et al. [152] prospectively evaluated male and female HIV-positive individuals starting a PI-based combination ARV drug regimen, comparing them with matched HIV-negative individuals. The authors observed that the 24-week HDL-C to LDL-C ratio fell more in HIV-infected women versus men, and that circulating levels of E-selectin, an endothelium-associated marker of inflammation and atheroclerotic risk, remained elevated in women versus men, each in comparison with the pre-treatment values. Further studies are needed to define gender-specific CVD risk factors in HIV-positive individuals.

Cardiovascular risk factors, including hyperlipidemia, insulin resistance and changes in body composition have increased in association with the use of potent ARV medications. Prediction modeling, surrogate markers and hard cardiovascular endpoints suggest increased CVD in persons with HIV. The relative increase in CVD risk is still small in an absolute sense, and the overwhelming effect of ARV is positive in terms of improvement in immune function and related morbidity and mortality. Nonetheless, as HIV-positive individuals live longer on ARV medications, cardiovascular disease could become increasingly prevalent. CVD risk assessment should occur regularly, especially after initiation and change in ARV medication use. Life-style modification strategies that have been well studied in the general population (e.g. smoking cessation, diet, and increased physical activity) should be rigorously implemented. Preliminary recommendations regarding drug treatment of HIV-associated hyperlipidemia and insulin resistance can be made from a limited number of studies. Future research will develop more specific risk stratification paradigms and treatment regimes for prevention of CVD in persons with HIV infection.

Supported in part by NIH DK 59535 and the Mary Fisher Clinical AIDS Research and Education Fund.