Determinants of Endothelial Function in Human Immunodeficiency Virus Infection: A Complex Interplay Among Therapy, Disease, and Host Factors


Kristin E. Mondy, MD, Infectious Diseases Division, Department of Internal Medicine, Washington University School of Medicine, 660 South Euclid Avenue, Campus Box 8051, St Louis, MO 63110


In the era of highly active antiretroviral therapy (HAART), human immunodeficiency virus (HIV) infection has become a chronic disease in which patients may develop significant metabolic complications and risk factors for cardiovascular disease (CVD), including insulin resistance, visceral fat deposition, and increases in atherogenic cholesterol and triglyceride levels. Epidemiologic studies have found that persons infected with HIV are likely to be at higher risk for premature CVD compared with the general population, and clinical studies examining endothelial function in HIV-infected cohorts have supported such conclusions. The mechanisms underlying the regulation of endothelial function in HIV-infected persons appear to be multifactorial, including direct effects of HIV on the endothelium, indirect effects of HIV on lipids and inflammatory cytokines, HAART-related effects, and traditional/host factors. Better understanding of these processes can lead to improved strategies for the long-term management of HIV infection.

The use of combination highly active antiretroviral therapy (HAART) to treat human immunodeficiency virus (HIV) infection since the mid-1990s has been associated with metabolic complications that may increase patients' cardiovascular disease (CVD) risk in the long term.1,2 Specific antiretroviral therapies may lead to insulin resistance, visceral fat deposition, and increases in atherogenic cholesterol and triglyceride levels,3,4 all components of the metabolic syndrome. In addition, cumulative exposure time to protease inhibitors (PIs) was recently found to be an independent risk factor for myocardial infarction (MI) in a large observational study examining MI incidence as the main clinical outcome.2 Other recent data have implicated HIV infection itself as a potential contributor to elevated cardiovascular risk.5 Currently, over 1 million persons in the United States have been diagnosed with HIV infection/AIDS and can be expected to receive antiretroviral therapy at some point during the course of their disease. The prevalence of HIV infection continues to rise as mortality declines and transmission rates remain relatively constant, thus creating a greater public health concern for the long-term management of HIV infection and complications of therapy, such as premature CVD. Finally, the metabolic syndrome associated with HIV infection also shares similarities with the metabolic syndrome, which is estimated to be prevalent in as many as 47 million Americans, regardless of HIV status.6

Given that many of the metabolic complications observed in HIV-infected persons are likely risk factors for CVD, the magnitude of overall risk may be large, but at this time it remains unknown. Cardiovascular morbidity may take years to manifest clinically, and HAART has been in widespread use for little more than 10 years. Methods to identify those at greatest risk before the development of such complications are not well defined. Thus, there is great interest in assessing long-term cardiovascular risk in patients receiving HAART before development of clinically overt heart disease and determining nontraditional risk factors (eg, PI use) that may accelerate the atherosclerotic process. In HIV-negative persons, recent research efforts have focused on identifying those at high risk for atherosclerotic heart disease before the development of atherosclerotic lesions. Endothelial function appears to be important in the early development of atherosclerosis because endothelial cells provide regulation of vascular tone, thrombogenesis, lipid breakdown, inflammation, and vessel growth.7 Derangements in these regulatory functions are likely to predispose vessels to atherosclerosis, thus providing the basis for studying endothelial dysfunction as a surrogate for CVD risk. Endothelial dysfunction occurs early in the process of atherosclerosis and can be predictive of the later development of CVD.8,9 In addition, endothelial dysfunction may be present independently of other cardiac risk factors and hence serve as an important modifier of the more traditional cardiac risk profile of an individual.8–10

Noninvasive Measures of Endothelial Function

Recently developed noninvasive tests of endothelial function such as brachial artery reactivity/flow-mediated dilation (brachial FMD) are well correlated with coronary endothelial function, atherosclerotic disease progression, and cardiovascular events in HIV-negative persons.8–10 Accordingly, this type of research tool has been increasingly utilized in HIV research as well. Most commonly, noninvasive measurement of endothelial function involves interrupting brachial artery blood flow to the arm with a blood pressure cuff inflated to suprasystolic pressure that, when released, creates reactive hyperemia caused by dilation of distal vessels. Additional shear stress and pressure changes from this hyperemia result in the local release of nitric oxide and further vessel dilation.8,9 Changes in vessel diameter are then measured with high-resolution ultrasonography before, during, and for several minutes after blood flow interruption. This test has shown a high degree of accuracy and reproducibility in HIV-negative persons.9,11 Changes in brachial artery diameter are often subsequently measured in response to sublingual nitroglycerin, a vasodilator that acts independently of endothelial function. This additional test provides an internal control to evaluate endothelium-independent artery dilation. Other more invasive tests for endothelial function may include the use of intravascular administration of nitric oxide agonists such as acetylcholine or methacholine. Although reproducibility of endothelial function tests may be quite high when performed at an experienced research facility, an important caveat of these tests is that endothelial function may be affected in the short term by a multitude of factors, including use of specific medications (ie, anti-inflammatory agents or antioxidants), caffeine intake, smoking, other diseases causing systemic inflammation, and hormonal changes.8,9,11,12 For these reasons, testing must be performed in a carefully controlled environment and cross-sectional studies may not be as useful as longitudinal studies in estimating cardiovascular risk.

Clinical Studies of Endothelial Function in HIV Infection

To date, most studies examining endothelial function in HIV-infected persons have generally been cross-sectional or limited to very short-term follow-up in small cohorts (Table). Stein and colleagues,13 in one of the earliest clinical studies of endothelial function in HIV infection, found that HIV-infected patients receiving PIs had significantly reduced endothelial function (by brachial FMD) compared with patients not receiving PIs. Further analysis revealed that the majority of the PI effect was related to elevated lipid and glucose levels. These findings were also supported by a very small study of HIV-negative participants in which endothelium-dependent vasodilation by methacholine was markedly impaired after 4 weeks of indinavir monotherapy.14 The changes seemed related in part to insulin resistance that developed during therapy. Indinavir, a first-generation PI, has been previously associated with multiple adverse metabolic complications including insulin resistance, hyperlipidemia, and visceral fat accumulation,15,16 so these results are not surprising. In contrast, a similar, more recent study using the newer PI lopinavir/ritonavir actually showed improvements in endothelial function over a 1-month period.17 The opposing results of these studies high-light that different PIs are known to have varying metabolic effects, and thus extrapolating findings of cardiovascular and endothelial function studies to an entire antiretroviral drug class is likely to lead to incorrect conclusions.

Table Table.  Clinical Studies of Endothelial Function in HIV
StudyParticipantsDesignOutcomeIndependent Predictors of Endothelial Dysfunction in HIV
Stein et al, 20011337 HIV+ patients taking HAARTCross-sectionalImpaired brachial FMD in HIV+ patients taking PI vs no PIPI use (due mainly to effects on atherogenic lipoproteins and glucose), brachial artery diameter, blood pressure, heart rate
Nolan et al, 20032124 HIV+ patients taking HAART; 24 controls (matched by age, sex, and BMI)Cross-sectionalNo significant differences in brachial FMD between HIV+ and control participantsPercentage of naive CD4 T cells, age, pulse pressure, brachial artery diameter
Van Wijk et al, 20062037 HIV+ patients taking HAART; 27 controlsCross-sectionalImpaired brachial FMD in HIV+ vs control participants, worse in HIV+ patients with MSApolipoprotein B
Solages et al, 20061975 HIV+ patients taking HAART (63% Hep C+); 223 HIV-controlsCross-sectionalImpaired brachial FMD in HIV+ patientsCurrent intravenous drug use, α-high-density lipoprotein triglyceride level
Torriani et al, 20071882 HIV+ patients starting HAARTLongitudinalImprovement in FMD at 24 weeks in all participantsStudy ongoing, but no significant association with lipids. HIV RNA decline is the sole predictor of improvement in FMD
Abbreviations: BMI, body mass index; FMD, flow-mediated dilatation; HAART, highly active antiretroviral therapy; Hep C, hepatitis C; HIV, human immunodeficiency virus; MS, metabolic syndrome; PI, protease inhibitor.

Currently, the National Institutes of Health-sponsored AIDS Clinical Trials Group has an ongoing substudy (5152s) to examine endothelial function over time as part of a larger therapy-naive trial (ACTG 5142). In this substudy, patients blinded to 1 of 3 treatment arms had brachial FMD measured before and after starting therapy.18 Preliminary data showed that at baseline, all participants appeared to have markedly impaired brachial FMD and that all treatment arms experienced improvement after 24 weeks of therapy, even the groups receiving PIs (Figure 1). Nonetheless, mean brachial FMD was still considered abnormal at short-term follow-up. Improvement in FMD was significantly associated only with 24-week reductions in HIV RNA (r2=−0.30; P=.02). There were no significant FMD differences between groups after adjusting for baseline brachial artery diameter or any relation between baseline FMD and pre-HAART HIV RNA level. Additional longitudinal study of this cohort is under way, but these findings suggest that HIV infection itself, particularly uncontrolled HIV infection, may have a greater role in causing early endothelial dysfunction than HAART. Recent results of a very large HAART interruption study also found higher than expected rates of adverse cardiovascular outcomes,5 further supporting an underlying role for HIV infection in the regulation of endothelial function. Consequently, the use of HAART treatment interruption as part of clinical practice or to avoid metabolic complications is strongly discouraged at this time.

Figure 1.

Twenty-four-week results from ACTG 5152s, an ongoing study in treatment-naive participants starting their first highly active antiretroviral therapy regimen. Patients are on either efavirenz + nucleosides (A), lopinavir/ritonavir + nucleosides (B), or a nucleoside-sparing arm of lopinavir/ritonavir + efavirenz (C). Flow-mediated dilatation (FMD) appears to be improving in all groups regardless of treatment. P values correspond to between-arm differences. Adapted from HLR Communications.40

Other cross-sectional clinical studies in HIV-positive persons have found an association between endothelial dysfunction and traditional CVD risk factors as well as nontraditional factors. In one of the larger studies that examined brachial FMD in HIV-infected vs noninfected persons, Solages and colleagues19 found that although HIV-infected persons had significantly reduced brachial FMD compared with noninfected persons, the decline was not significantly associated with use of PIs or other antiretroviral therapies. Rather, only current injection drug use and levels of α-high-density lipoprotein triglyceride levels were significantly related to endothelial function in a multivariate model. Of note, a significant number of participants had hepatitis C coinfection, so other potentially confounding factors existed, which could have affected underlying inflammation and vascular function. Another smaller study examined several surrogates for CVD risk (brachial FMD, carotid intima-media thickness, and aortic pulse wave velocity) in HIV-infected patients receiving HAART in relation to the traditional risk factor of the metabolic syndrome.20 Patients with HIV and no metabolic syndrome exhibited endothelial dysfunction comparable to that of age-matched persons with type 2 diabetes, whereas HIV-infected patients with the metabolic syndrome had even further impairment in endothelial function. Finally, another study comparing HIV-infected men receiving PI therapy with controls matched by age and body mass index found no significant difference in endothelial function between groups.21 Among patients with HIV infection, however, higher brachial FMD was significantly associated with a lower percentage of naive CD4 T cells, suggesting a role for immune activation in regulating endothelial function.

Potential Mechanisms of Endothelial Dysfunction

HIV infection and its treatment may have direct or indirect effects on endothelial function, including direct infection or activation of endothelial cells by HIV, vascular injury as a result of chronic inflammation and immune activation, dysregulation of the nitric oxide synthase system, or metabolic effects that serve to amplify traditional risk factors for endothelial dysfunction (eg, dyslipidemia and insulin resistance [Figure 2]). Unfortunately, studies to further elucidate these pathogenic mechanisms have primarily been conducted in vitro. Recently, the HIV viral protein Tat was shown to cause a significant decrease in endothelial nitric oxide synthase (eNOS) mRNA and related protein expression in the endothelial cells of porcine coronary arteries.22 In vitro, certain PIs were found to reduce eNOS levels, resulting in an increase in reactive oxygen species leading to vascular damage.23 Finally, both the Tat and gp120 proteins of the HIV virion may activate endothelial cells, resulting in increased expression of specific adhesion molecules (eg, ICAM-1, E-selectin) that may contribute to endothelial dysfunction and a prothrombotic state.24,25

Figure 2.

Interrelationship of factors contributing to endothelial function in human immunodeficiency virus (HIV). HAART indicates highly active antiretroviral therapy.

HIV and HAART may also adversely affect endothelial function through indirect mechanisms by altering lipid, glucose, and fat metabolism as suggested by the clinical studies described above. Metabolic derangements may lead to some or all of the components of the metabolic syndrome, an atherogenic, proinflammatory condition that has been shown to contribute to endothelial dysfunction in studies of persons without HIV infection.26 HIV-positive persons receiving first-generation PIs, for example, may develop hepatic and peripheral tissue insulin resistance, together with increased lipolysis, visceral fat accumulation (lipohypertrophy), and hypertriglyceridemia that contribute to the metabolic syndrome.15,16,27,28 Acutely in vitro, these drugs may cause moderate suppression of GLUT4 receptor transport of glucose into adipocytes.29 Downstream, impaired activation of sterol regulatory element binding protein 1 (SREBP-1) in adipose and hepatic cells may occur, resulting in dysregulation of adipocyte differentiation and glucose and lipid metabolism.30,31 This finding has been further supported by reduced expression of PPARγ, which is activated by SREBP-1, and decreased expression of PPARγ-dependent adipocytokines and insulin-signaling molecules in fatty tissues of HIV-infected patients taking PIs compared with seronegative controls.32 Treatment with a PPARγ agonist (rosiglitazone) in HIV-negative participants with the metabolic syndrome has been shown to improve endothelial function and reduce serum markers of inflammation.33 Among HIV-infected patients, a study comparing rosiglitazone vs metformin for lipodystrophy resulted in significant improvement in brachial FMD in the metformin arm only.34 The absolute improvement in FMD among patients taking metformin was very small (1.5%) and of limited clinical significance, however. Of note, rosiglitazone has been associated with exacerbations in dyslipidemia among HIV-infected persons35; thus, any benefit of improved insulin sensitivity on endothelial function from this drug may have been negated by its untoward lipid effects. In addition, HIV-positive patients receiving metformin had a significant decrease in visceral fat, suggesting that changes in body fat may also influence endothelial function.34 Currently, it remains unclear whether HIV-associated lipodystrophy and insulin resistance are significant contributors toward endothelial function in this population; therefore, further longitudinal studies that incorporate sensitive assessments of body fat changes and insulin sensitivity will be important.

As demonstrated in the Table, cross-sectional clinical studies of brachial FMD in HIV appear to support a role for dyslipidemia in endothelial dysfunction. Specific drugs from the 3 most widely used antiretroviral classes (nucleoside reverse transcriptase inhibitors, PIs, and non-nucleoside reverse transcriptase inhibitors) may all have substantial effects on lipids.4,28 HIV infection itself is also known to cause unfavorable lipid changes, including a reduction in high-density lipoprotein cholesterol and elevation in triglycerides.28,36 These lipid effects could therefore be a potential explanation for part of the negative effect of HIV infection on endothelial function. Although treatment may help normalize lipid values to some extent, certain antiretroviral agents (particularly ritonavir) may still worsen HIV's atherogenic profile by contributing to further increases in triglyceride and low-density lipoprotein cholesterol levels, while having very little effect on high-density lipoprotein cholesterol.36 Two studies to date examining statin therapy among persons infected with HIV who were receiving HAART have demonstrated a trend toward improved endothelial function (absolute increase of 0.7%–1.2%) by brachial FMD,37,38 further supporting a role for lipids as mediators of endothelial dysfunction. Data in HIV-negative persons support additional anti-inflammatory properties of statins, however; thus, the effects of these drugs on endothelial function in HIV may be multifactorial and merit further study.39 For now, HIV guidelines specifically address dyslipidemia in this population and recommend the routine use of appropriate statins to achieve target lipid levels.28


With the development of potent antiretroviral therapies, HIV infection has become a chronic disease in which management of long-term metabolic complications is a significant component of routine care. Large epidemiologic studies have found that persons infected with HIV are likely to be at higher risk for premature CVD compared with the general population, and several clinical studies examining endothelial function in HIV cohorts have supported such conclusions. The mechanisms underlying the regulation of endothelial function in HIV-infected persons are not well elucidated but are likely to be multifactorial, including direct effects of HIV on endothelial cells, indirect effects of HIV on lipids and proinflammatory processes, HAART-related metabolic derangements, and traditional/host factors. Longitudinal studies in HIV-infected persons naive to HAART and those receiving HAART (with and without controlled HIV viremia) are needed to better characterize the contribution of HIV infection and its therapy to endothelial function. Ideally, in the long term, changes in endothelial function can potentially be correlated with actual cardiovascular outcomes. In the meantime, current data support the overall mortality benefits of HAART and the importance of treating traditional cardiovascular risk factors such as dyslipidemia, insulin resistance, and obesity and modifying behavioral factors such as tobacco and illegal drug use.


This work was supported in part by grants from the National Institutes of Health (K23 AI065336-01) and the Centers for Disease Control and Prevention (200-2002-00611).