Shortly after the introduction of HAART, reports of impaired fasting glucose, glucose intolerance, diabetes, and hyperlipidemia appeared in the HIV literature. Given the close temporal relationship to the introduction of PIs, studies focused on the association with PIs. However, other factors such as restoration of health, immune reconstitution, and body composition changes (including lipoatrophy and visceral hypertrophy) may also contribute to the disturbances in metabolism. To clarify these issues, several groups have given HIV antiretroviral drugs to healthy, HIV-seronegative volunteers in order to define direct drug effects. In the sections later, the results in HIV-infected subjects will be compared to those in healthy volunteers to understand the effects of drug versus disease. Metabolic effects are specific to certain drugs and not an effect of the PI class as a whole.
Insulin Resistance, Glucose Intolerance, and Diabetes
With the reports of rapid onset of diabetes after introduction of HAART with a PI, researchers looked for effects of PI on insulin resistance. Subsequent studies looked at hepatic glucose production and insulin secretion.
Several factors contribute to the development of insulin resistance in the setting of HIV. Unlike other infectious states, in which insulin resistance is common, early in the epidemic, AIDS was found to be associated with an increase in insulin sensitivity (Table 25–4). Compared with 10 healthy controls, insulin sensitivity was higher in 10 stable patients with symptomatic HIV. However, with asymptomatic HIV infection, another early study found that there was no change in insulin sensitivity compared with healthy controls. Insulin resistance is common in healthy subjects. Thus, improvement of HIV infection alone may contribute to an observed decrease in insulin sensitivity. More recent studies have found insulin resistance in ARV-naïve, HIV-infected subjects.
Table 25–4 Effect of HIV and AIDS Status on Glucose and Lipid Metabolism Prior to the Introduction of HAART. ||Download (.pdf)
Table 25–4 Effect of HIV and AIDS Status on Glucose and Lipid Metabolism Prior to the Introduction of HAART.
|Insulin Resistance||Total Cholesterol||Triglyceride||VLDL||LDL||HDL|
|HIV||↔||↓||↑ (8%)||↑ (7%)||↓ (16%)||↓↓↓ (36%)|
|AIDS||↓||↓||↑↑↑↑ (99%)||↑↑↑↑ (98%)||↓↓↓ (31%)||↓↓↓ (37%)|
Body composition influences insulin sensitivity. Patients studied early in the epidemic, were thin, if not cachectic. In recent studies body weight was higher even in many ARV-naïve patients. Increased visceral fat in the abdomen has been linked to insulin resistance in subjects with or without HIV disease. Upper trunk fat is also independently, strongly associated with insulin resistance. Likewise, severe lipoatrophy has been linked to insulin resistance regardless of HIV status. The lesser levels of lipoatrophy seen in most HIV-infected patients may also contribute. It should be recognized that each of these factors may contribute in an additive way to insulin resistance.
Much attention has focused on the role of individual therapies in the induction of insulin resistance. Some PIs have been reported to decrease insulin-mediated glucose disposal per unit insulin level (M/I) during the hyperinsulinemic, euglycemic clamp, a technique during which insulin is infused at a steady rate and glucose infused to maintain euglycemia, which directly measures insulin action. In a double-blind, placebo-controlled study in healthy normal volunteers, a single dose of indinavir has been shown to decrease insulin-mediated glucose disposal by 34% (Table 25–5). Indinavir for 4 weeks has also been shown to cause a 17% decrease in insulin-mediated glucose disposal as well as deterioration in glucose tolerance. A single dose of ritonavir decreased insulin sensitivity by 15%. Lopinavir/ritonavir has less of an effect on insulin sensitivity, but the magnitude of this effect is not clear (see Table 25–5). In two studies, lopinavir/ritonavir given for 4 weeks caused no change in insulin sensitivity, whereas in shorter studies, lopinavir/ritonavir given for 1 to 5 days was associated with a 13% to 24% decrease in insulin sensitivity. The study reporting a 13% decrease compared rates in the same subjects on placebo and lopinavir/ritonavir; the study reporting the 24% decrease presented pooled data from some subjects who received both placebo and drug on different occasions, some drug only and some placebo only.
Table 25–5. Effect of Protease Inhibitors and Nonnucleoside Reverse Transcriptase Inhibitors on Glucose and Lipid Metabolism.a ||Download (.pdf)
Table 25–5. Effect of Protease Inhibitors and Nonnucleoside Reverse Transcriptase Inhibitors on Glucose and Lipid Metabolism.a
|Drug||Fasting Glucose||Insulinb Resistance||EGP||Total Cholesterol||TG||LDL||HDL|
|Indinavir||↑ (HIV negative)||↑↑↑ (−34% M/I) (HIV negative)||↑↑ (HIV negative)||↔||↔ (HIV negative)||↔ (HIV negative) ↑ (HIV infected)c||↔↑ (0%-20%)|
|Ritonavir||↔||↑↑ (−18% M/I) (HIV negative)||↑ (HIV negative)||↑↑||↑↑↑ (150% HIV negative)||↔(HIV negative) ↑(HIV infected)c||↓ (0%-5%)|
|Lopinavir/ritonavir||↔ (HIV negative)||↔/↑d (−13%-24% M/Id) (HIV negative)||NA||↔||↑↑ (80% HIV negative)||↔ (HIV negative) ↑ (HIV infected)c||↔|
|Tipranavir/ritonavir||↔||NA||NA||NA||↑↑ (80% HIV infected)||NA||NA|
|Atazanavir/ritonavir||↔ (HIV negative)||↔ (HIV negative)||NA||NA||↑||↔↑||↔↑ (0%-20%)|
|Amprenavir||↔||↔ (HIV negative)||↔ (HIV negative)||↔↑||↔↑||↑ (HIV infected)||↑ (0%-20%)|
|Atazanavir||↔||↔ (HIV negative)||NA||↔/↑||↔(HIV negative)||↔(HIV negative) ↔(HIV infected)||↔/↑ (0%-20%)|
Of note, not all PIs decrease insulin-mediated glucose disposal. In double-blind, placebo-controlled studies, atazanavir and amprenavir had no effect on insulin sensitivity in healthy normal volunteers (see Table 25–5). NNRTIs have not been associated with insulin resistance. In studies where PIs are replaced with NNRTIs, insulin resistance improved.
The mechanism of insulin resistance with PIs includes the acute blockade of the peripheral insulin-regulated glucose transporter (GLUT)4. In vitro studies have shown that PIs (indinavir, ritonavir, and amprenavir) selectively inhibit 2-deoxyglucose transport into 3T3-L1 adipocytes without affecting early insulin-signaling events or the translocation of intracellular GLUT4 to the surface. Indinavir has also been shown to block partially GLUT2, the glucose transporter postulated to be involved in glucose sensing in the pancreas and regulation of insulin secretion. Recently, an analog of the peptidomimetic phenylalanine moiety found in all PIs has been shown to inhibit GLUT4-induced glucose transport in vitro. Because serum levels of PIs vary, some PIs, such as amprenavir, may block GLUT4 in vitro but have no effect in patients.
Increased insulin resistance has also been found in HIV-infected subjects on NRTI therapy. However, it is unclear if the effects of NRTIs are a direct effect of the drug, reactivation of the immune system, restoration of health, or changes in body composition. When stavudine was given to healthy volunteers, there was no decrease in M/I.
In addition to peripheral insulin resistance, impairment in insulin secretion was reported in HIV patients on PI therapy. In HIV-infected subjects treated with either nelfinavir, indinavir, lopinavir/ritonavir, or saquinavir, beta cell function assessed by first-phase insulin secretion showed a 25% decrease. However, insulin secretion in the HIV-infected patients was higher than controls before PI therapy and was reduced only to that of controls after PI therapy. The HIV-infected patients had suppression of HIV RNA levels and increases in CD4 counts. More recently, healthy normal volunteers were given lopinavir/ritonavir for 4 weeks, and no effect was seen on first-phase insulin secretion. Thus, it is not clear whether currently used PIs alter insulin secretion.
Hepatic glucose production is also increased in some patients on PIs. Endogenous glucose production, comprised mostly of hepatic gluconeogenesis and glycogenolysis, is the largest determinant of fasting glucose and can be measured by tracer stable isotope technology. In studies of healthy normal volunteers, indinavir increased endogenous glucose production in the fasting state. During the hyperinsulinemic, euglycemic clamp, indinavir blunted the ability of insulin to suppress endogenous glucose production. Endogenous glucose production was not increased in rats treated with intravenous indinavir during a hyperinsulinemic, euglycemic clamp study. However, the method of stable isotope analysis was different in the rat and human studies, which may explain the discrepancy in results. In humans, full-dose ritonavir has a small detrimental effect on endogenous glucose production, while amprenavir had no effect. The effects of other PIs remain to be determined.
Adipocytokine levels have been explored in HIV infection, and they may explain some of the results in glucose metabolism. Adiponectin, a hormone secreted by adipocytes, has been shown to increase peripheral and hepatic insulin sensitivity. Adiponectin levels inversely correlate with the amount of visceral fat in HIV-negative subjects. In patients with HIV-associated peripheral lipoatrophy, adiponectin levels are reduced. Low adiponectin levels have been proposed to mediate some of the insulin resistance observed in peripheral lipoatrophy and visceral lipohypertrophy. The mechanism which reduces adiponectin levels is unknown. Some have attributed the reduction to PI therapy. In vitro studies of cultured fat cells have suggested that PI treatment suppresses adiponectin mRNA and protein expression. However, two studies in healthy normal volunteers found that in fact adiponectin levels were increased during treatment with the PIs indinavir or lopinavir/ritonavir. The discrepancy in the effect of PIs on adiponectin levels in vitro and in humans is not well understood.
Levels of leptin, another hormone secreted by adipocytes, correlate with insulin resistance. However, some diseases with very low leptin levels also have insulin resistance. Leptin levels have been shown to be decreased in HIV patients with peripheral lipoatrophy.
Current guidelines set forth by the International AIDS Society-USA (IAS-USA) suggest measuring fasting glucose levels before and during PI therapy. In patients with risk factors for diabetes or with peripheral lipoatrophy or visceral lipohypertrophy, oral glucose tolerance testing may be warranted. Patients with preexisting abnormalities in glucose metabolism or with first-degree relatives with diabetes mellitus should consider avoiding the PIs most associated with insulin resistance; atazanavir and newer PIs may be better choices.
Treatment of PI-induced diabetes should ideally be tailored to the specific alteration in glucose metabolism. Although there are no studies of treatment of PI-induced diabetes per se, there are studies of therapy of patients who have HIV-associated lipoatrophy and lipohypertrophy who also are taking PIs. Thiazolidinediones can ameliorate peripheral insulin resistance induced by PIs. Thiazolidinediones improve insulin resistance in patients with HIV-associated lipoatrophy and lipohypertrophy. Proliferation of lipomas has been reported in one patient with HIV-associated lipoatrophy. Given the recent findings that thiazolidinediones decrease BMD and may increase fracture risk, caution is warranted in HIV-infected patients as they may be at higher risk for bone loss and fracture. Metformin decreases hepatic glucose production and peripheral insulin resistance. Metformin should be used with caution in combination with NNRTI's stavudine and didanosine, which, like metformin, are associated with increased lactic acidosis. Sulfonylureas may prove to be useful in improving insulin secretion. Due to the risk of hypoglycemia, sulfonylureas should be used with caution in treating mild PI-induced diabetes.
Other medications used to treat opportunistic infections are associated with hyperglycemia and hypoglycemia. Pentamidine causes pancreatic beta cell toxicity, acutely leading to hypoglycemia, and over the long-term, it causes diabetes mellitus. Hypoglycemia during pentamidine treatment is associated with increased length of treatment, higher cumulative doses, and renal insufficiency. Patients who develop hypoglycemia in association with pentamidine therapy are at increased long-term risk of developing diabetes mellitus. HIV-infected patients who develop diabetes mellitus following pentamidine therapy have low C peptide levels, suggesting beta cell destruction. Aerosolized pentamidine therapy has also been reported to cause hypoglycemia and diabetes mellitus. Pentamidine, trimethoprim-sulfamethoxazole, and the nucleoside analogs didanosine and zalcitabine have been associated with acute pancreatitis. Megestrol acetate, which has intrinsic glucocorticoid activity, may be associated with diabetes mellitus in HIV-infected patients, perhaps through its glucocorticoid activity, although the rate of hyperglycemia in controlled clinical trials appears to be low. GH can also cause insulin resistance, leading to hyperglycemia and diabetes. Medications used in HIV-infected patients that can affect the endocrine system are listed in Table 25–2.
Alterations in lipid and lipoprotein profiles are common in patients infected with HIV. The observed changes can be due to a number of interrelated issues, including HIV infection itself, antiviral medications, body composition changes, and immune reconstitution. A rational approach to disturbances in lipid metabolism is to assess the status of the HIV infection, medications used to treat HIV, and presence of body composition changes (including peripheral lipoatrophy and central lipohypertrophy). The following section reviews the lipid and lipoprotein profiles individually, with an emphasis on studies prospectively measuring fasting lipid levels.
HIV infection is associated with a mild increase in triglyceride and very low density (VLDL) cholesterol levels (see Table 25–4). Triglyceride and VLDL cholesterol levels rise in association with advancing HIV disease and correlate with HIV RNA levels. The increased triglycerides are thought to be due to interferon-alpha induced decreased clearance of triglycerides and, to a lesser extent, increased VLDL production.
Several antiviral medications can increase triglyceride levels. Full-dose ritonavir can cause a two- to three-fold increase in triglyceride levels, probably by increased VLDL production (see Table 25–5). Ritonavir has been shown in vitro to inhibit the degradation of apolipoprotein B, a protein involved in the formation of VLDL particles. Other studies have suggested that activated sterol regulatory-binding proteins (SREBPs) in the liver are involved in the increase in VLDL production. Because of its ability to inhibit the hepatic enzyme CYP3A4, ritonavir is used to increase the pharmacologic doses of other PIs metabolized through the same cytochrome system. Boosting doses of ritonavir (100 mg twice daily) have also been shown to increase triglycerides, albeit to a lesser extent. The combination lopinavir/ritonavir given to healthy normal volunteers increased triglycerides and VLDL cholesterol levels by 83% and 33%, respectively. Tipranavir/ritonavir produces similar increases to those of lopinavir/ritonavir in HIV-infected individuals. Not all PIs alter triglyceride levels; in healthy normal volunteers, administration of indinavir and atazanavir resulted in no change in triglyceride levels (see Table 25–5). The data are mixed on amprenavir and nelfinavir, with some studies showing an increase in triglycerides and others finding no effect.
The NNRTI efavirenz is associated with increased triglycerides. Again, this is not a class effect as another NNRTI, nevirapine, has not been associated with alterations in triglyceride metabolism.
The effects of NRTIs on lipid metabolism have not been well studied. In some, but not all studies, stavudine use was associated with increased triglyceride levels. The most informative trial compared stavudine with tenofovir and found an increase in triglycerides in the stavudine arm, but not in the tenofovir arm. Given that all subjects got efavirenz, a likely interpretation is that there was a lipid-lowering effect of tenofovir; other data support this interpretation, as adding tenofovir to an effective ARV regimen lowers lipids.
The original clinical syndrome of HIV lipodystrophy has also been reported to be associated with increased triglyceride levels. One study found that 57% of patients with both peripheral lipoatrophy and central lipohypertrophy had triglyceride levels above 300 mg/dL. It has long been recognized that visceral obesity in the general population is associated with high triglycerides. However, recent data suggest that lower body fat is protective, as increased levels of lower body fat are associated with lower triglycerides in both HIV-infected patients and controls. Therefore, the loss of lower body fat in HIV-associated lipoatrophy is another reason why triglycerides are high in HIV infection. Hypertriglyceridemia is well known to be multifactorial; genes, diet, alcohol, and physical activity also play a role. In HIV infection, one can add the synergistic effects of the host response to HIV itself, treatment with ritonavir and efavirenz, and HIV-lipoatrophy to the pathogenesis.
After the introduction of HAART, a reported increase in low-density lipoprotein (LDL) cholesterol levels was largely attributed to PI therapy. Recently, it has become increasingly clear that factors other than PIs contribute to this rise in LDL cholesterol levels. In the early stages of HIV infection, LDL cholesterol levels fall (see Table 25–4). With effective therapy, LDL cholesterol levels rise in response to suppression of the infection, independent of the type of therapy. Most, but not all PIs have been associated with increases in LDL cholesterol levels, but average levels are not high. Studies in healthy normal volunteers have provided insight into the direct effects of PIs on cholesterol metabolism apart from those associated with HIV infection. Indinavir, ritonavir, lopinavir/ritonavir, and atazanavir all have no effects on LDL cholesterol levels in healthy normal volunteers (see Table 25–5). In patients with HIV infection, treatment with the NNRTI nevirapine raises LDL cholesterol levels. Studies that involve switching patients from PIs to the NNRTIs nevirapine and efavirenz found that LDL cholesterol levels do not change. Hence, the increase in LDL cholesterol seen in HIV-infected patients is not solely an effect of PI therapy, but likely represents suppression of HIV and restoration to health. Tenofovir may lower LDL levels.
HIV infection and the NNRTIs have significant effects on high-density lipoprotein (HDL) metabolism. Early in the course of HIV infection, prior to the appearance of clinically evident disease, HDL cholesterol levels decline to levels around 25 to 35 mg/dL (see Table 25–4). With advancing HIV disease, HDL cholesterol levels continue to decline to less than 50% of baseline values. The pathogenesis of these changes is not well understood. Shortly after the introduction of PIs, one cross-sectional study reported decreased HDL cholesterol levels in HIV-infected patients. However, subsequent studies have failed to show a decrease in HDL cholesterol levels in either HIV-infected patients or healthy normal volunteers. Indeed, some studies have found modest increases (13%-21%) in HDL cholesterol levels during treatment with indinavir, nelfinavir, amprenavir, and atazanavir therapy in prospective studies of HIV-infected patients. More impressive is the nearly 50% increase in HDL cholesterol seen during treatment with the NNRTI nevirapine. Efavirenz has also been shown to increase HDL cholesterol levels by 15% to 23%. When patients were switched from PIs to the NNRTIs efavirenz and nevirapine, somewhat smaller increases were seen again supporting the concept that HAART with PI induces small increases in HDL. Hence, the increase in HDL cholesterol levels is likely directly due to NNRTI therapy. In some studies, HDL levels have been reported to be lower in patients with HIV-associated lipodystrophy, but the extent to which that decrease is due to HIV is not clear. Visceral obesity and upper trunk fat are associated with lower HDL levels.
A panel of the IAS-USA has suggested guidelines for the diagnosis and treatment of dyslipidemia in HIV-infected subjects. The National Cholesterol Education Program guidelines are recommended for initiating therapy in patients infected with HIV. To monitor for dyslipidemia, a fasting lipid panel should be obtained before initiating or switching therapy. Lipid panels should be repeated 3 to 6 months after starting ARV therapy. In those patients with preexisting cardiovascular disease (CVD) or uncontrollable hyperlipidemia, switching HIV therapy should be considered if the regimen is one linked to the observed dyslipidemia and it is refractory to conventional treatment.
Hypolipidemic therapy should be tailored to the type of dyslipidemia. For patients with triglyceride levels greater than 500 mg/dL, fibrate therapy is recommended. Studies have shown that gemfibrozil and fenofibrate are effective in reducing triglyceride levels in patients infected with HIV, but those who start at high triglyceride levels do not reach normal levels. Fish oil lowers triglycerides, but raises LDL, leading to no net decrease in CVD risk. Niacin is effective and has the advantage of raising HDL levels as well as lowering triglycerides, but niacin-induced insulin resistance may be problematic in HIV-infected patients. HMG-CoA reductase inhibitors may be better second-line agents for lowering triglycerides. As with HIV-negative patients, combination pharmacologic therapy may be required to correct extremely elevated triglyceride levels.
HMG-CoA reductase inhibitors are effective, first-line agents for the treatment of hypercholesterolemia. However, there are multiple drug-drug interactions that are important in HIV-infected patients. Some statins are metabolized by Cyp3A4 which is induced or inhibited by some ARV drugs. In particular, the combination of PI and simvastatin or lovastatin should not be used. PI, especially ritonavir, which is used to boost levels of many HIV ARV drugs to therapeutic levels, inhibit Cyp3A4 which is the major metabolic pathway for simvastatin and lovastatin. Ritonavir-based regimens increase simvastatin levels by 5- to 32-fold and multiple cases of rhabdomyolysis have been reported on such combinations. Atovastatin activity increases twofold, so 80 mg atorvastatin should be avoided. Lopinavir/ritonavir increases rosuvastatin levels two- to fivefold, and tipranvir/ritonavir increases rosuvastatin levels twofold and atorvastatin levels eightfold by unknown mechanisms. If those PI combinations are used, only low doses of rosuvastatin (5 mg) and atorvastatin (10 mg) should be tried. Thus, pravastatin and fluvastatin XL are recommended as first-line statins for patients on PI- or ritonavir-based HAART.
The effects of HIV protease inhibitors on glucose and lipid metabolism are shown in Table 25–5.
HIV, Antiretroviral Therapy, and Risk of Atherosclerosis
The changes in lipid and glucose metabolism seen in HIV raise the question about whether atherosclerosis increases due to HIV or its therapies. Retrospective studies report an increased prevalence of CVD in HIV-infected patients. Some studies also found an association with ARV therapy, particularly PI therapy, in addition to traditional risk factors such as age, gender, smoking, and LDL and HDL cholesterol levels. Smoking is more prevalent in those with HIV infection versus noninfected controls. Cross-sectional studies of the intima media thickness (IMT) of the carotid and femoral arteries by ultrasound have also shown that traditional CVD risk factors provide the dominant contribution to increased plaque. However, after adjusting for traditional CVD risk factors, HIV infection is an independent risk factor for increased IMT, similar in magnitude to male sex, diabetes, and smoking.