Understanding the Complications of Antiretroviral Drugs

After the use of combination antiretroviral (ARV) therapy led to improvements in survival among patients with HIV infection and AIDS [1], a wide variety of body composition and metabolic changes appeared that initially were blamed on HIV protease inhibitors (PIs) as a class [2]. However, the global changes attributed to PIs seemed unlikely, for several reasons [3]. First, an understanding of the literature published before the PI era indicates that some changes represented reversal of the effects of HIV infection—essentially, restoration to health. Second, studies always involved HIV-infected patients receiving therapy with 3 drugs, so attribution to a PI was not definitive. Third, the diverse structures of PIs, which were directed against a viral protease with little homology to mammalian proteases, made it unlikely that all PIs would have the same off-target complications at their respective therapeutic levels. Furthermore, facile theories often linked insulin resistance and hypertriglyceridemia.

As HIV-infected patients have lived longer, there has been increasing concern over ARV drug toxicities, such as insulin resistance, hypertriglyceridemia with increases in atherogenic non–high-density lipoprotein cholesterol, and coronary artery disease. It therefore became important to define the direct complications of specific ARV drugs. As a consequence, many of us have tested the effects of ARV drugs, especially PIs, on healthy, HIV-seronegative volunteers, to separate the direct toxicities of the specific ARV drug from the effects of controlling HIV infection (reviewed in detail elsewhere [4]). Comparison of the effects seen in HIV-seronegative volunteers with the changes seen in HIV-infected patients receiving combination therapy or undergoing switch therapy in which 1 drug was changed has painted a clearer picture of ARV drug toxicity.

We have learned that the increases in low-density lipoprotein cholesterol seen with combination therapy are mostly reversal of the decrease in low-density lipoprotein seen in the host response to HIV [4]. The similar decrease in high-density lipoprotein that is attributable to HIV infection is poorly responsive to PI-based therapy but improves significantly with nonnucleoside reverse-transcriptase inhibitor therapy, although not always back to normal. The increases seen in triglycerides are drug specific, because they are dominantly attributable to ritonavir-containing regimens and are not a general property of PI drugs. Indeed, there is evidence that efavirenz increases triglycerides. Indinavir is the most potent inducer of insulin resistance but has little effect on triglycerides. We have learned that time of exposure matters. Acute or short-term administration leads to more insulin resistance, whereas insulin resistance decreases with 4 weeks of exposure. Increases in triglycerides are seen within 2–4 weeks after initiation of ritonavir treatment. After longer periods, both PI and nonnucleoside reverse-transcriptase inhibitor regimens for HIV-infected individuals will lead to insulin resistance that likely represents restoration to health, including body composition changes [5]. The 4-week period is not enough to see any change in body composition; therefore, it makes a good time for study, because one can see the effects of most metabolic drugs and diets and there is enough time for the body to respond to acute metabolic changes [4].

The article by Dubé et al. [6] in this issue of Clinical Infectious Diseases extends these observations to another area, endothelial dysfunction, a vascular abnormality that predicts cardiovascular disease [7, 8]. Endothelial dysfunction has the advantage of being modulated within weeks, so it represents a more dynamic measure of cardiovascular risk that complements slowly changing direct measures of atherosclerosis, such as carotid intimamedial thickness. Endothelial dysfunction may also play a direct role in increasing atherosclerosis. Research groups from Indiana University had shown elsewhere that indinavir was a potent inducer of endothelial dysfunction in healthy, HIV-seronegative volunteers, but that effect was unrelated to the induction of insulin resistance [9–11]. Dubé et al. [6] have published an informative study with negative results. Neither atazanavir nor lopinavir-ritonavir induced endothelial dysfunction in a similar 4-week exposure among HIV-seronegative volunteers, despite the fact that lopinavir-ritonavir induced its usual hypertriglyceridemia.

Thus, Dubé et al. [6] have given us several important findings. First, as with other complications of ARV therapy, induction of endothelial dysfunction is drug specific, not class specific. Second, induction of endothelial dysfunction may be a direct effect, because it has now been shown that it is not merely mediated by hypertriglyceridemia or insulin resistance, although these metabolic changes may be associated with endothelial dysfunction under other circumstances [12]. Finally, Dubé et al. [6] point out that, in looking for an explanation of how PIs as a class might induce myocardial infarction above that expected from the changes in lipids [13], induction of endothelial dysfunction is no longer a likely candidate. Further research in this area is needed, but the recent finding that PI therapy as a class is associated with higher fibrinogen levels [14] suggests one possible mechanism. In the interim, studies such as those by Dubé et al. [6] and others [4] allow us to more intelligently advise our patients about the possible complications of very effective drugs.