Does Antiretroviral Therapy Increase or Decrease the Risk of Cardiovascular Disease?

Abstract

Atherosclerosis is a complex inflammatory process that has been identified as an important problem in persons with HIV infection. Epidemiologic studies have linked certain antiretroviral medications (some nucleoside reverse transcriptase inhibitors and protease inhibitors) with a higher risk of coronary heart disease (CHD). Conversely, non-nucleoside reverse transcriptase inhibitors, entry inhibitors and integrase inhibitors appear neutral. HIV infection is a chronic inflammatory process associated with endothelial dysfunction, atherogenic dyslipidemia and a higher risk for CHD. Initiation of antiretroviral therapy in the short term appears to lower CHD risk, regardless of the specific agents used. However, adequately powered randomized trials of antiretroviral therapy with CHD as a primary endpoint are lacking. Thus, the evidence of whether antiretroviral therapy increases or decreases CHD risk in persons with HIV is perplexing. This article reviews the current controversy of the role of HIV and antiretroviral therapy in the development of cardiovascular disease.

Introduction

The seminal report by Henry documented the premature occurrence of coronary heart disease (CHD) in an HIV-infected man that fueled a controversy about the role of antiretroviral therapy in the development of CHD [1]. Despite careful epidemiologic analyses of CHD and numerous randomized trials we still do not have the ability to separate the effects of chronic HIV infection from those of antiretroviral therapy. The lack of an adequate animal model of HIV infection has made it difficult to answer the important question of whether antiretroviral therapy, HIV or both lower or raise the risk of developing CHD. Furthermore, there are no adequately power randomized trials with CHD as a primary endpoint in persons with HIV infection. It is critical to place this debate in the overall context that antiretroviral therapy has dramatically improved the morbidity and mortality from the acquired immunodeficiency syndrome (AIDS). However, recently published studies of mortality in HIV infection suggest that cardiovascular disease ranks as a leading cause behind AIDS-related opportunistic illnesses and liver disease [2–3]. In addition, the true burden of CHD in persons with HIV remains unclear because of advances in interventions that reduce CHD mortality and the inconsistency of prior reports to capture all events (e.g., coronary revascularization). The burden of CHD will clearly increase as patients with HIV infection grow older and the question will linger as to whether the risk is increased or merely reflects that of the general population. In the past few years, controversy has re-emerged surrounding the important question of whether HIV infection causes CHD stimulated by evidence from the SMART study [4]. On the other side, some studies indicate that in the short-term, antiretroviral therapy actually improves surrogate measures of CHD like endothelial reactivity [5]. The debate has become more heated with newer provocative data from D:A:D study suggesting that recent use of abacavir or didanosine is associated with a higher risk of CHD [6]. This review will outline our current understanding of the relationship between antiretroviral therapy, HIV and CHD and update the reader on recent data that address this important problem.

Coronary Heart Disease in HIV Infection

Since the initial report of premature CHD in HIV infection in 1998 a number of observational studies have suggested that antiretroviral therapy and HIV are both associated with elevated risk of myocardial infarctions (Table 1) [4, 6–11]. Unfortunately, there are no randomized clinical trials demonstrating a clear association with antiretroviral therapy and CHD endpoints like myocardial infarctions (MIs) or proven coronary artery disease. It may be that the risk associations merely reflect the occurrence of CHD in a population that is aging with other known risk factors. Indeed, epidemiologic evidence demonstrates high rates of known CHD risk factors like cigarette smoking, hypertension and diabetes mellitus in persons with HIV infection [12]. In one of the largest prospective observational studies, investigators from the Data Collection on Adverse Events of Anti-HIV Drugs (D:A:D) study demonstrated a relative risk of 1.26 per additional year of exposure to antiretroviral therapy [8]. They went on to identify protease inhibitors (PIs) as a class of antiretrovirals associated with a greater risk (RR=1.16 per year of additional use) [13]. Although the majority of studies suggest there is added risk with the use of some types of antiretroviral therapy and HIV, Bozzette and colleagues did not find added risk in their initial or follow up studies of veterans in the United States [7,14]. It is important to reiterate that most studies also document the importance of traditional risk factors for CHD including age, gender, cigarette smoking, hypertension, diabetes mellitus and family history of premature coronary artery disease.

Table 1 –

Selected Epidemiologic Studies of the Association of HIV and Antiretroviral Therapy and Coronary Heart Disease

Study	Population	Study design	Events	Antiretroviral therapy or HIV effect	Other significant risk factors
VA7	36,766	Retrospective	1,207 CHD	No effect with ART or PI	Not evaluated
D:A:D8	23,490	Prospective	345 MI	↑ risk with ART (RR=1.26, 95% CI-1.12–1.41)	Smoking, Age, Gender, HTN, DM, Lipids
D:A:D6	23,437	Prospective	345 MI	↑ risk with PIs (RR=1.16, 95% CI-1.10–1.23)	Smoking, Age, Gender, HTN, DM, Lipids
Kaiser9	4,408	Retrospective	86 MI	↑ risk in HIV+ vs. HIV- (6.5 vs. 3.8 CHD hospitalization events per year, P=0.003) No change in risk with PI use	Not evaluated
French10	34,976	Retrospective	60 MIs	↑ risk with PIs (RR=2.56, 95% CI-1.03–6.34)	Age
Insight11	7,542	Retrospective	127 MIs	↑ risk with PI use (HR=1.71, 95% CI-1.07–2.74)	Smoking, Age, HTN, DM

MI-Myocardial infarction; CHD-Coronary heart disease; HTN-Hypertension; DM-Diabetes mellitus; HR-Hazard ratio; RR-Relative risk; PIs-Protease inhibitors; ART-Antiretroviral therapy;

Several new surprising findings were reported over the past couple of years further complicating our understanding of the relationship between CHD, HIV infection and antiretroviral therapy. The D:A:D investigators reported that recent use of abacavir or didanosine-containing regimens have a greater risk for the development of CHD [6]. The adjusted relative risk of CHD for ongoing use of abacavir and didanosine was 1.70 and 1.41, respectively, compared to those never using these agents or those who had stopped them for 6 months or greater [6]. Interestingly, while there was no cumulative affect initially reported with abacavir, a more recent analysis suggests there may be a cumulative effect (RR, 1.07, 95% CI, 1.00–1.14) [15]. Conversely, investigators from the SMART trial demonstrated a higher risk of CHD with discontinuation of antiretroviral therapy (relative hazard=1.6, 95% CI-1.00–2.5, P=0.05) [4]. Thus, the picture has become increasingly complicated with both untreated HIV infection being implicated in the occurrence of CHD endpoints and specific antiretrovirals also contributing to risk.

The Role of HIV Infection in the Development of Coronary Heart Disease

Several lines of evidence suggest a role of HIV infection in the development of CHD. These include HIV associated atherogenic dyslipidemia, higher CHD event rates in persons discontinuing antiretroviral therapy and endothelial dysfunction in untreated HIV that improves with treatment. Atherosclerosis is an inflammatory process with many known risk factors including cigarette smoking, hypertension, diabetes mellitus, a family history of premature CHD and dyslipidemia [16]. Important factors involved in the pathogenesis of atherosclerosis include the formation and migration of lipid-laden macrophages containing oxidized particles of low density lipoprotein (LDL), smooth muscle proliferation of blood vessels, endothelial and platelet dysfunction, plaque formation, plaque destabilization and coronary artery clot formation. Understanding the role of HIV in this process is challenging because many patients in the pre-HAART era died from AIDS-related illnesses prior to the overt development of CHD and survivors in the post-HAART era typically receive antiretroviral therapy that may have independent effects on atherogenesis. HIV infection has long been associated with alterations in lipid metabolism and the presence of an atherogenic lipid profile [17–19]. HIV infection and inflammation induces changes in circulating lipids and apolipoproteins with more advanced immunosuppression resulting in hypertriglyceridemia, and lower levels of total cholesterol (TC), LDL-C, and high density lipoprotein cholesterol (HDL-C) [17–19]. In an elegant study from the Multi-center AIDS Cohort Study Group, Riddler reported average decreases in TC of −30 mg/dL, LDL-C of −22 mg/dL and HDL-C of −12 mg/dL after HIV seroconversion [19]. Suppression of HIV replication reversed many of these changes with the exception of HDL-C that remained below pre-seroconversion levels by an average of −9 mg/dL [19]. In general, there are a myriad of changes that are proatherogenic in untreated HIV infection including higher concentrations of small dense LDL-C particles, higher levels of triglyceride rich particles, and increased platelet-activating factor acetylhydrolase activity [20].

Discontinuation of antiretroviral therapy has been associated with a higher short-term risk of CHD events like MI suggesting that the final common pathway of events in atherosclerosis like plaque destabilization and/or clot formation may be more likely to occur in untreated HIV infection. In the SMART study, 5,472 participants were randomized to continue or interrupt antiretroviral therapy. In the drug conservation arm where treatment was interrupted there were 48 cardiovascular events compared to 31 in the virus suppression arm that remained on antiretroviral therapy (relative hazard of 1.60, 1.0–2.5, P=0.05) [4]. The precise role of untreated HIV infection in these events remains unclear.

Surrogate markers of CHD like endothelial dysfunction and intima media thickness (IMT) have been abnormal in studies of brachial artery reactivity and carotid IMT in persons with HIV infection. In a sub-study of a large randomized trial of persons initiating antiretroviral therapy, 82 persons had endothelial function measured by flow-mediated dilatation (FMD) of the brachial artery [5]. FMD was abnormal at baseline (3.68%) and significantly improved with treatment at 24 weeks by an average of 1.48% but did not return to normal [5]. There was no difference in whether persons were receiving a non-nucleoside reverse transcriptase inhibitor (NNRTI) based regimen, protease inhibitor (PI) based regimen or nucleoside/tide reverse transcriptase (NRTI) sparing regimen. This study suggests that HIV infection induces endothelial dysfunction and that lowering of HIV replication partially reverses these changes. Indeed there was a modest inverse correlation with HIV RNA levels and the improvement in endothelial dysfunction (r_s=−0.30, P=0.017). Hsue and colleagues reported HIV infection was an independent predictor of thicker carotid IMT compared to uninfected controls (mean IMT, 0.91 mm vs. 0.74 mm, P<0.001) [21]. Importantly, several risk factors were also associated with thicker IMT including age (P<0.001), LDL-C (P<0.001), cigarette pack-years (P<0.001), and Latino race (P<0.024) on multivariate analysis. Many of these patients with HIV infection were taking antiretroviral therapy and the smaller sample size may have limited the ability of the investigators to determine whether HIV and/or antiretroviral therapy were truly associated with thicker IMT.

Thus, HIV infection appears to be associated with an atherogenic dyslipidemia, endothelial dysfunction and the relative occurrence of higher event rates of CHD in persons that discontinue antiretroviral therapy. It remains unclear whether HIV infection is associated with more rapid progression of atherosclerosis.

Antiretroviral Therapy and Coronary Heart Disease

There are currently five FDA-approved classes of antiretrovirals. Two classes, entry inhibitors (e.g., enfuvirtide and maraviroc) and integrase inhibitors (raltegravir), have not been linked with CHD, dyslipidemia, insulin resistance or adipose tissue changes. Within the remaining classes, PIs and NRTIs have been most closely associated with CHD. Although there may be class affects with many antiretrovirals there appears to be some differences in the risk of CHD and their specific abilities to induce metabolic disturbances. For example, most PIs, with the exception of atazanavir, are associated with elevations in TC, triglycerides and LDL-C [22]. NNRTIs produce increases in TC, HDL-C, LDL-C, and triglycerides; however, epidemiologic studies suggest they are not associated with higher risk of CHD [13,23]. Among the NRTIs, while the thymidine analogs induce greater increases in TC, LDL-C, and triglycerides they have not been specifically linked with a higher risk of CHD events [6,24]. Thus, it is important to examine the evidence within each class of antiretrovirals that links specific agents with the development of CHD.

The Role of Nucleoside/Nucleotide Reverse Transcriptase Inhibitors

Current treatment guidelines recommend a backbone of NRTIs combined with PIs, NNRTIs or Integrase inhibitors for treatment-naive persons with HIV infection [25]. The D:A:D study recently presented somewhat surprising data on five NRTIs and their risk association with CHD (Table 1) [6]. In this study of 33,347 subjects, there were 517 persons with a MI. They hypothesized that zidovudine and stavudine were more likely to be associated with the development of an MI. However, recent use (within 6 months of the MI) but not cumulative or prior exposure to Abacavir was associated with relative rate of MI of 1.94 (95% CI, 1.48–2.55, P<0.001). Similarly, the recent use of didanosine was associated with a relative rate of MI of 1.53 (95% CI, 1.10–2.13, P=0.01). Interestingly, recent use of lamivudine was also associated with a higher relative rate of 1.69 (95% CI, 1.02–2.80, P=0.04). These analyses were adjusted for demographic characteristics and other risk factors for CHD. Importantly, the risk disappeared 6 months after stopping the abacavir. In a report update from the same group, cumulative exposure to abacavir was now also associated with a higher risk of CHD (RR-1.07, 95% CI, 1.00–1.14) [15]. Data from the SMART study seemed to confirm the D:A:D findings with patients who were on the viral suppression arm receiving abacavir having a higher risk of MI events with recent use (RR-4.3, 95% CI, 1.4–13.0, P<0.01) (Table 1) [26]. However, two other reports of subjects enrolled in randomized trials of abacavir did not confirm this association (Table 1) [27–28]. These latter reports were limited by a smaller number of CHD endpoints. In the D:A:D study, recent or cumulative exposure to tenofovir was not associated with increased risk [15]. Additional data on the use of tenofovir suggests in may lower certain lipid fractions like triglyceride and cholesterol levels relative to other NRTIs [29].

Are there any mechanisms that might account for the association of abacavir and possibly didanosine with a higher risk of CHD events? Some investigators have postulated that abacavir increases platelet adhesiveness and reduces endothelial reactivity [30]. Higher levels of pro-inflammatory cytokines like IL-6 and markers of inflammation and cardiovascular risk like high-sensitivity C-reactive protein have been documented in subjects taking these drugs (Table 2) [26]. Thus, in the setting of atherosclerosis, one might hypothesize that the use of abacavir may lead to plaque destabilization and clot formation with the resultant acute CHD event. There is limited mechanistic information on the association of didanosine. The basis for these associations is largely determined by retrospective, non-randomized observational studies. Some have also argued that the association abacavir and didanosine reflects a channeling bias of persons at higher risk of CHD. Thus, clinicians may have been more likely to switch those that they perceived at higher risk of CHD to these regimens because they thought that thymidine analogs were more likely to generate worsening atherosclerosis. The D:A:D investigators have been careful to assess for channeling bias by comparing baseline CHD risk for example, and have not found obvious differences. Nevertheless, until randomized trial data confirms this observation questions about its validity will remain.

Table 2 –

Summary of Recent Important Studies of HIV and Antiretroviral Therapy and the Risk of Coronary Heart Disease

Study	Study Type	Population	Major New Findings	Summary of Results
SMART10	RCT	5,472 80 CVD events	Interruption of treatment associated with higher risk of cardiovascular disease.	Fatal or non-fatal cardiovascular disease in treatment interruption group had a higher hazard ratio=1.6 (95% CI, 1.0–2.5)
SMART26	RCT, nested case-control	2,752 70 CVD events	Use of Abacavir associated with higher risk of Major CVD events including MI, revascularization, stroke and cardiac death.	Use of abacavir but not didanosine associated with a higher risk of major CVD events with a hazard ratio=1.80, (95% CI, 1.04–3.1) IL-6 levels were 16.2% higher (P=0.02) and hs-CRP levels were 27.1% higher (P=0.02) in Abacavir compared to other NRTIs
D:A:D9	Prospective Observational	33,347 517 MIs	Abacavir and Didanosine associated with higher risk of myocardial infarction.	Recent use & risk of MI -Abacavir=1.89 (CI, 1.47–2.45, P=0.0001) Recent use & risk of MI-Didanosine=1.49 (CI, 1.14–1.95, P=0.004)
GSK27	54 RCTs	14,683 45 CHD 18 MIs	No association of abacavir with CHD or MI	Rate of CHD with abacavir vs. no abacavir (3.45 vs. 5.82 per 1000 PYRs, RR=0.59 (95% CI, 0.35–1.01, P=0.06) Rate of MI with abacavir vs. no abacavir (2.04 vs. 2.36 per 1000 PYRs, RR=0.86 (95% CI, 0.40–1.86, P=0.71)
ACTG 500128	Prospective Observational Study of RCTs	3,205 27 MIs 63 Severe CVD	No association with recent use of abacavir and development of an MI or severe CVD	For risk of MI, recent use of Abacavir had Adjusted RR=1.2 (95% CI, 0.5–3.1, P=0.82). For risk of severe CVD, recent use of Abacavir had Adjusted RR=0.8 (95% CI, 0.4–1.5, P=0.50)
ACTG 5152s5	RCT	82	Antiretroviral therapy resulted in improved endothelial function after 24 weeks of treatment	Baseline FMD was 3.68% (IQR, 1.98–5.51). FMD increased by 0.74% (IQR, −0.62% to +2.74%, P=0.003) at 4 weeks and 1.48% (IQR, −0.20% to +4.30%, P=0.001) at 24 weeks.

RCT-Randomized clinical trial; CVD-Cardiovascular disease; MI-Myocardial infarction; IL-6-Interleukin 6; hs-CRP-High sensitivity C-Reactive Protein; RR-Relative risk; NRTIs-Nucleoside/tide reverse transcriptase inhibitors; PYRs-Person-Years; IQR-Interquartile range; FMD-Flow-mediated dilatation; ACTG-AIDS Clinical Trials Group; GSK-GlaxoSmithKline.

The Role of Protease Inhibitors

Several epidemiologic studies have directly implicated PIs being associated with a higher risk of CHD events (Table 1). Mary-Krause reported a higher risk of MIs in the French Hospital National Database Study with the use of PIs. Those exposed to ≥30 months of PIs were more likely than those exposed to <18 months of PIs to have an MI (SMR=3.6, 95% CI, 1.8–6.2). Similarly, those exposed to ≥30 months of PIs were more likely than the general HIV uninfected population to have an MI (SMR=2.9, 95% CI, 1.5–5.0). In that study, none of the models were adjusted for lipid levels, diabetes mellitus, cigarette smoking or hypertension because the data was not available. In the D:A:D study, the relative risk of MI was increased 1.16 per additional year of exposure to PIs and remained increased even when adjusted for many other known traditional risk factors (RR=1.10, 95% CI-1.02–1.18, P=0.002) [6].

Are there specific agents within the class of PIs that are more likely to be associated with the risk of CHD events? To date, only one study has examined this with sufficient power to determine any associations. Worm and colleagues recently reported data from the D:A:D study on the relationship of four PIs, nelfinavir, indinavir, lopinavir-ritonavir and saquinavir, and the risk of MI [15]. The relative risk for MI was increased with the use of indinavir (RR-1.12, 95% CI-1.07–1.18) and lopinavir-ritonavir (RR-1.13, 95% CI, 1.05–1.21). No association was observed with saquinavir or nelfinavir. There was insufficient data available to analyze the effects of amprenavir, atazanavir, darunavir or tipranavir.

Are there mechanisms that might explain how some PIs increase the risk of CHD? Indinavir leads to insulin resistance by inhibiting the GLUT-4 receptor and reduces endothelial reactivity by a nitric oxide mediated mechanism [16, 31]. Lopinavir-ritonavir induces significant dyslipidemia in healthy volunteers and HIV infected persons [32]. And Noor reported that lopinavir-ritonavir impaired insulin sensitivity and glucose uptake more so than atazanavir-ritonavir [33]. Thus, it is possible that some PIs may lead to worsen atherosclerosis through several different mechanisms.

The Role of Non-nucleoside Reverse Transcriptase Inhibitors

The most important data on risk associated with NNRTI use comes from the D:A:D study [13]. Overall, the risk of CHD did not change significantly with cumulative use of NNRTI based therapy. When adjusted for other traditional CHD risk factors, NNRTI use was associated with a stable risk (RR-1.00, 95% CI=0.93–1.09, P=0.92). One potential mechanism to explain a lesser risk of CHD compared to PI usage is the difference in the profile of lipids with NNRTI use. For example, in the 2NN study, 1,267 subjects were randomized to efavirenz (EFV) or nevirapine (NVP) [23]. HDL-C increased 42.5% and 32.7% in the NVP and EFV groups, respectively. It has been postulated that the relatively higher increase in HDL-C may be associated with a counter balance to the other lipid changes and lower overall risk of CHD events. While these changes in lipid may be considered favorable, it is important to note that total cholesterol and LDL-C also increased 26.9% and 35.4% and 31.1% and 40.1% in the NVP and EFV groups, respectively, in the 2NN study. Thus, while favorable changes in HDL-C are reported with NNRTI based regimens, it is not clear this effect explains the observations of lower CHD event rates in the D:A:D cohort.

What research is needed?

A critically important question is quantifying the role of chronic HIV infection in the development and progression of atherosclerosis. This must be distinguished from the effects of specific antiretroviral therapies. Identification of relevant predictive biomarkers and surrogate markers for CHD in persons with HIV infection is needed. We need to understand if the composition of atherosclerotic plaques are similar in HIV infected and uninfected persons. Are the recent observations of higher risk of CHD with specific NRTIs or PIs real or confounded by other factors obscured by non-randomized study designs? Most importantly, we need to determine what interventions will lower the risk of CHD events. For additional information on critical research questions about the relationship between HIV, CHD and antiretroviral therapy readers should refer to a recent State of the Research Executive Summary and accompanying articles published in Circulation [34].

In the meantime, what should clinicians do?

All patients with HIV infection should be evaluated regularly for CHD risk. Appropriate use of CHD risk predictor scores like those developed by the Framingham study are encouraged. Treatment of HIV infection earlier in the course of disease may reduce inflammation and perhaps alter the progression of atherosclerosis. Indeed some experts have recommended treating HIV earlier based upon emerging data from cohort studies [35]. The most recent Department of Health and Human Services Guidelines for the use of antiretroviral therapy in HIV-infected Adults and Adolescents indicate there is support for earlier initiation of antiretroviral therapy [25]. When initiating antiretroviral therapy, avoidance of specific antiretroviral agents currently linked to a higher risk of CHD should be advocated whenever possible particularly in persons with other risk factors for CHD. Aggressive risk reduction should be pursued by treating known risk factors like cigarette smoking, hypertension, diabetes mellitus and dyslipidemia. Clinicians should also be familiar with the drug-drug interactions between antiretroviral agents and medications used to treat these common co-morbidities.

Summary and Conclusions

Atherosclerosis and CHD is a complex inflammatory process that may be altered by the presence of chronic HIV infection. It is further complicated by the presence of many common persistent infections in the same host like cytomegalovirus, herpes viruses, hepatitis C and hepatitis B. Many of these other infections may also be contributing to the evolution of atherosclerotic disease. Antiretroviral therapy improves inflammation and endothelial reactivity in the short term. However, selected antiretroviral agents have been associated with specific disturbances known to be associated with a higher risk of CHD including dyslipidemia, insulin resistance and endothelial dysfunction. Indeed some PIs and NRTIs have been linked to a higher risk of CHD in epidemiologic studies. Despite this association, it is critical to recognize that it is currently not possible to differentiate the effects of chronic HIV infection and those of antiretroviral therapy on the development of atherosclerosis and CHD. Furthermore, the effects on increasing CHD risk appear to be modest relative to other known risk factors like cigarette smoking, hypertension and diabetes mellitus. Recent evidence suggests that treatment of HIV infection earlier in the course of disease lowers overall mortality. Part of this reduction in mortality may be, in theory, related to an overall reduction in inflammation. Whether earlier initiation of antiretroviral therapy reduces the rate of progression of atherosclerosis is unknown. As persons with HIV infection continue to age it will become increasingly important for HIV clinicians to proactively assess and intervene to reduce the risk of the development of CHD and aggressively manage those with known CHD. Thus, if you ask whether antiretrovirals increase or decrease the risk of CHD, the answer for now is probably both.