Tolerability and Safety of HIV Protease Inhibitors in Adults: heart, bones, lipids, body changes, diabetes
JAIDS Journal of Acquired Immune Deficiency Syndromes: Volume 37(1) 1 September 2004 pp 1111-1124
Sax, Paul E MD*; Kumar, Princy MD†
*Brigham and Women's Hospital, Boston, MA, and †Georgetown University Hospital, Washington, DC.
Summary: Antiretroviral drugs are associated with both short-term and long-term adverse events. Like other HIV drugs, protease inhibitors (PIs) may affect metabolic processes influencing body shape and body tissue composition, appearance, bone integrity, and cardiovascular status.
However, numerous confounding variables including age, cigarette smoking, body mass index (BMI), duration of HIV infection, degree of immunodeficiency, concomitant antiretroviral agents, extent of previous treatment, and duration of treatment all blur the relationship between PI use and adverse events.
Recent data suggest that the early PIs appear to have greater effects on such surrogate markers of disease risk as insulin resistance and cholesterol and triglyceride levels than the recently developed PIs. These data also suggest that evaluation of PIs as a class should be reconsidered and that it is probably not appropriate to extrapolate safety data obtained from individuals treated with first-generation agents in the era of potent combination antiretroviral therapy to those treated with recently developed PIs.
Because PIs remain a critical component of successful antiretroviral therapy, evaluation of potential long-term complications with prolonged PI use is essential, as is delineation of the significant differences in safety profiles among individual PIs.
Antiretroviral drug therapy is associated with both short- and long-term adverse effects. The magnitude of these events is often influenced by such variables as age, sex, body mass index (BMI), disease status, degree of immunodeficiency, and life style. From a clinical point of view, the acute adverse effects of rash, central nervous system (CNS) disturbance, nausea, vomiting, and diarrhea are true inconveniences to the patient at best and, at worst, mandate cessation of treatment. Such adverse effects may have a psychosocial, stigmatizing impact conducive to low self-esteem, anxiety, and depression if these events are perceived to be severe and frequent. Long-term therapy may cause potentially permanent and irreversible physiological damage leading to chronic or progressive pathologic conditions. Additionally, as a significant reduction in HIV-related mortality has occurred since the introduction of combination regimens in 1996, metabolic complications associated with combination antiretroviral therapy (ART) have emerged as a prominent cause of morbidity and poor quality of life among surviving patients. Antiretroviral treatment may also interfere with other medication(s), resulting in altered drug disposition and unwanted or even life-threatening events. Whether acute or chronic, severe adverse effects are the main reason for nonadherence to a given therapeutic regimen.
SHORT-TERM ADVERSE EFFECTS ASSOCIATED WITH HIV PROTEASE INHIBITORS
Short-term adverse effects associated with therapy occur within hours or days of beginning treatment. When the clinical manifestations of these symptoms are mild, palliative therapy can often be provided on an as-needed basis, such as the use of loperamide to reduce diarrhea. Short-term adverse effects may resolve spontaneously or subside. In some cases, complete resolution only occurs if therapy is withdrawn. The severity of the short-term side effects associated with PI treatment varies with the agent used and some adverse effects occur with only a few agents in the class (Table 1).
| ||908 ||908/r ||APV/r |
| ||2 Caps bid ||4 Pills qd ||8 Caps/2 Caps bid |
| ||1400 mg ||1400mg/200mg ||1200mg/200mg |
|Anemia ||- ||- ||- |
|Fatigue ||2% ||1% ||5% |
|Fever ||- ||- ||- |
|GI symptoms || || || |
|-abdominal pain ||1% ||1% ||13% |
|-diarrhea ||5% ||13% ||25% |
|-nausea ||5% ||3% ||23% |
|-vomiting ||2% ||3% ||10% |
|Headache ||2% ||4% ||15% |
|Jaundice ||- ||- ||- |
|Nethrolithiasis ||- ||- ||- |
|Paresthesias ||- ||- ||- |
|Rash ||8% ||3% ||3% |
|Dyspnea ||- ||- ||- |
| ||ATV ||IDV ||LPV/r |
| ||2 Caps qd ||2 Caps tid ||3 Caps bid |
| ||400mg ||800mg ||400mg |
|Anemia ||- ||1% ||- |
|Fatigue ||2% ||- ||- |
|Fever ||4% ||2% ||- |
|GI symptoms || || || |
|-abdominal pain ||6% ||17% ||4% |
|-diarrhea ||6% ||3% ||16% |
|-nausea ||16% ||12% ||7% |
|-vomiting ||6% ||8% ||2% |
|Headache ||14% ||5% ||2% |
|Jaundice ||8% ||2% ||- |
|Nethrolithiasis ||- ||9% ||- |
|Paresthesias ||1% ||- ||- |
|Rash ||9% ||1, ||1% |
|Dyspnea ||- ||- ||- |
| ||NFV ||RTV ||SQV |
| ||5 Tabs bid ||6 Caps bid ||6 Caps bid |
| ||1250mg ||600mg ||600mg |
|Anemia ||- ||- ||- |
|Fatigue ||- ||10% ||- |
|Fever ||- ||1% ||- |
|GI symptoms || || || |
|-abdominal pain ||- ||- ||2% |
|-diarrhea ||20% ||15% ||20% |
|-nausea ||3% ||26% ||11% |
|-vomiting ||- ||14% ||3% |
|Headache ||- ||6% ||5% |
|Jaundice ||- ||- ||- |
|Nethrolithiasis ||- ||- ||- |
|Paresthesias ||- ||6% ||- |
|Rash ||2% ||- ||- |
|Dyspnea ||- ||- ||- |
Grade 3/4 recurrent or chronic diarrhea is the adverse effect most often associated with PIs, but it rarely occurs with such PIs as indinavir, atazanavir, or fosamprenavir (GW433908) (Table 1). In contrast, both nelfinavir and lopinavir/ritonavir (LPV/RTV) have been associated with grade 3/4 diarrhea in a significant number of patients. Caution must be exercised in comparing such adverse event data, however, because the clinical trials that produced the data may have involved different accompanying nucleoside/nucleotide reverse transcriptase inhibitor (NRTI) drugs and patient populations. In one randomized, double-blind phase III study, 316 patients were randomly assigned to 1 of 2 nelfinavir doses (500 mg or 750 mg 3 times daily [tid]) or placebo; all received zidovudine/lamivudine (ZDV/3TC) as the NRTI backbone. Moderate or severe diarrhea occurred in 15 and 20% of the patients receiving nelfinavir 500 mg and 750 mg, respectively. Additionally, 2 (10%) of the 20 patients who experienced grade 3/4 diarrhea withdrew from the study. In a randomized, double-blind, multicenter, multinational study, 686 patients with >400 copies of HIV RNA per mL were randomly assigned to LPV/RTV (400 mg/100 mg bid) or nelfinavir 750 mg tid, both in combination with stavudine (d4T)/3TC. Moderate to severe diarrhea of possible, probable, or unknown relation to PIs was the most common adverse event in both treatment groups, occurring in 51/328 patients (15.6%) in the LPV/RTV arm and 56/327 patients (17.1%) in the nelfinavir arm. As shown in Table 1, diarrhea of this severity was rarely seen in treatment-naive patients treated with either fosamprenavir or atazanavir, occurring in 5 and 6% of patients, respectively.
Nephrolithiasis is a unique adverse effect of indinavir, occurring in as many as 12.4% of patients; the risk appears to increase over time, perhaps due to lower compliance with required fluid intake. In addition, indinavir can induce a retinoidlike effect characterized by dry skin, changes in hair, and fingernail and toe nail disorders. Both nephrolithiasis and retinoid toxicity appear to be increased with greater drug exposure, with a higher incidence of these complications when the agent is given boosted with RTV. Additionally, jaundice associated with indirect hyperbilirubinemia may occur with both indinavir and atazanavir (Table 2). In a dose-escalating phase II study of 420 antiretroviral-naive patients randomly allocated to atazanavir or nelfinavir, grade 3/4 elevations in total bilirubin (predominantly unconjugated) occurred asymptomatically in 20, 41, and 49% of subjects treated with atazanavir 200 mg, 400 mg, or 500 mg, respectively, compared with 1% in the nelfinavir group (P < 0.0001). Jaundice occurred only in atazanavir-treated subjects (6, 6, and 12% in the atazanavir 200 mg, 400 mg, and 500 mg groups, respectively). Bilirubin elevations were not associated with hepatotoxicity. In the extended phase of this study, when nelfinavir-treated patients were allowed to switch to the atazanavir arm, 76% of those who switched developed elevated bilirubin levels, including 13% with grade 3/4. Among those who continued on atazanavir 400 mg, 83% had elevated bilirubin levels; including 26% with grade 3/4. Jaundice appeared in 6% of patients who switched from nelfinavir to atazanavir. A number of laboratory abnormalities were also observed with PI treatment, most of which occurred in <10% of patients (Table 2).
GRADE 3/4 COMMON LAB ABNORMALITIES
| ||908 ||908/r ||ATV |
| ||2 Caps bid ||4 Pills qd ||2 Caps qd |
| ||1400mg ||1400mg/200mg ||400mg |
|Hematology || || || |
|-hemoglobin ||- ||- ||- |
|-thrombocytopenia ||- ||- ||- |
|-neutropenia ||3% ||- ||5% |
|Chemistry || || || |
|-ALT ||6% ||4% ||6% |
|-AST ||6% ||4% ||3% |
|-Lipase ||8% ||5% ||4% |
|-Bilirubin ||- ||- ||22% |
| ||ATV/r ||IDV ||LPV/r |
| ||2 Caps/1-qd ||2 Caps tid ||3 Caps bid |
| ||300mg/100mg ||800mg ||400mg |
|-hemoglobin ||- ||1% ||- |
|-thrombocytopenia ||- ||1% ||4% |
|-neutropenia ||4% ||2% ||2% |
|Chemistry || || || |
|-ALT ||3% ||4% ||7% |
|-AST ||- ||3% ||8% |
|-Lipase ||4% ||- ||- |
|-Bilirubin ||40% ||12% ||3% |
| ||NFV ||RTV ||SQV |
| ||5 tabs bid ||6 Caps bid ||6 Caps bid |
| ||1250mg ||600mg ||600mg |
|-hemoglobin ||2% ||- ||1% |
|-thrombocytopenia ||- ||- ||- |
|-neutropenia ||5% ||- ||3% |
|Chemistry || || || |
|-ALT ||1% ||8% ||2% |
|-AST ||1% ||10% ||4% |
|-Lipase ||- ||- ||- |
|-Bilirubin ||- ||- ||2% |
Relationship of Tolerability to Adherence
Short-term adverse effects can be challenging, both for asymptomatic, previously untreated patients beginning their first regimen and for patients with more advanced HIV disease, in whom medication adverse effects may augment HIV-related symptoms. Frequently occurring or severe adverse effects (e.g., chronic diarrhea) can interfere with daily activities, including employment, and can be as stigmatizing as appearance-related adverse effects (e.g., jaundice or rash) or HIV disease itself. A variety of studies have reported a correlation between nonadherence to treatment and adverse effects. In the cross-sectional multicenter Adherence Italian Cohort Naive Anti-retrovirals (ICoNA) substudy enrolling 358 patients treated with combination ART, nonadherent participants had a higher mean overall symptom score (P < 0.001) and mean medication adverse effect score (P < 0.001) than compliant patients. In this study, nausea, anorexia, anxiety, confusion, insomnia, vision problems, taste perversion, and abnormal fat accumulation were significantly correlated with reduced adherence to treatment. In another multicenter study of 336 patients treated with 2 NRTIs and 1 PI, fatigue and diarrhea were the most often reported adverse effects of therapy (81.3 and 75,% respectively). Follow-up for 4 months demonstrated that patients with a high incidence of adverse effects at the onset of treatment were more likely to become nonadherent to treatment.2 Recently, in a retrospective study of 345 randomly selected antiretroviral-naive patients who initiated combination ART on 6 different regimens, the discontinuation rate was 61,% with 24% citing adverse events as the primary reason. Nausea (27%), diarrhea (18%), vomiting (16%), and gastrointestinal (GI) disturbances (12%) were the main reasons for discontinuation. CNS abnormalities (10%) were also cited as a reason for discontinuation. Demonstrating that appearance (as a marker of disease) can undermine the patient's confidence and commitment to ART, the ICoNA study showed that adipose tissue alteration is also associated with nonadherence to treatment.
Short-term adverse effects can be an impediment to continued therapy. They can include intolerable GI disturbances (most older PIs), renal toxicity (indinavir), jaundice (indinavir, atazanavir), and to a lesser extent CNS-associated events. These effects can often be managed with prophylactic or adjunctive therapy, and it is essential to educate patients on how to anticipate and manage potential short-term adverse effects to ensure the success of antiretroviral treatment. The management of short-term adverse effects has been addressed elsewhere.
POTENTIAL LONG-TERM COMPLICATIONS OF HIV PROTEASE INHIBITOR-BASED THERAPY
HIV Infection vs. Protease Inhibitors: Relative Contribution to Adverse Events
The early stages of HIV infection induce metabolic and endocrine effects that are generally asymptomatic. As infection progresses and CD4+ cells are depleted, metabolic and endocrine adverse effects become evident. Endocrine alterations have been documented in the thyroid, adrenals, and gonads in some patients.25 In addition, HIV appears to affect pancreatic and liver function, impairing insulin clearance or glucose homeostasis or lipid metabolism.26-29 The introduction of PI-based combination therapy appears to have compounded this problem. In 1997 several reports described alterations in glucose metabolism in patients treated with the then recently approved PIs. Glucose intolerance was attributed to insulin resistance possibly associated with triglyceride storage defects, and lipodystrophy in PI-treated individuals was ascribed to lipid redistribution. Accumulating data on PIs suggested a possible causal relationship between PI use and abnormal lipid levels, hyperglycemia, insulin resistance, new-onset diabetes mellitus, and altered bone metabolism. The causal relationship between altered glucose metabolism and PI use was confirmed by the resolution of hyperglycemia or diabetes following PI discontinuation.
25. Sellmeyer DE, Grunfeld C. Endocrine and metabolic disturbances in human immunodeficiency virus infection and the acquired immune deficiency syndrome. Endocr Rev. 1996;17:518-532.
26. Hommes MJ, Romijn JA, Endert E, et al. Resting energy expenditure and substrate oxidation in human immunodeficiency virus (HIV)-infected asymptomatic men: HIV affects host metabolism in the early asymptomatic stage. Am J Clin Nutr. 1991;54:311-315.
27. Hommes MJ, Romijn JA, Endert E, et al. Basal fuel homoeostasis in symptomatic human immunodeficiency virus infection. Clin Sci (Lond). 1991;80:359-365.
28. Hommes MJ, Romijn JA, Endert E, et al. Insulin sensitivity and insulin clearance in human immunodeficiency virus-infected men. Metabolism. 1991;40:651-656.
29. Heyligenberg R, Romijn JA, Hommes MJ, et al. Non-insulin-mediated glucose uptake in human immunodeficiency virus-infected men. Clin Sci (Lond). 1993;84:209-216.
The interrelatedness between insulin resistance, impaired glucose tolerance, dyslipidemia, and body habitus changes has been termed lipodystrophy syndrome and is currently the object of intense investigation. Lipodystrophy has been reported in the range of 5-75%; the most likely explanations for this wide range are variations in case definition, different methods of assessment, and different length of follow-up among study subjects.
The changes in body fat distribution observed in patients with the lipodystrophy syndrome during combination ART primarily involve subcutaneous lipoatrophy and central (visceral) adiposity. The former is most clinically evident in the extremities, face, and buttocks, whereas localized fat gain is most prominent in the midsection as well as the breast (gynecomastia in men) or supraclavicular or intrascapular (buffalo hump) areas. Because antiretroviral drugs are used in combination, identifying the specific treatment cause of lipodystrophy has been difficult. Early studies suggested that PIs were at the origin of the fat redistribution, but recent work suggests a significant role for NRTIs in the establishment of the syndrome, particularly lipoatrophy. Prospective data also demonstrate that the individual choice of the accompanying NRTI combination is important in determining the rate of peripheral fat loss. The adipose tissue loss associated with lipodystrophy is not to be mistaken for the increasing loss in lean body mass that occurs in many patients with progressing HIV disease. This condition, termed wasting syndrome when it exceeds a 10% decrease in body weight compared with baseline, appears unrelated to lipodystrophy and is strongly associated with viral load.
In HIV-infected patients treated with PI-containing regimens, body habitus changes can be associated with hypercholesterolemia, hypertriglyceridemia, and insulin resistance. Recent reports also suggest that NRTIs alone or in combination with PIs are associated with lipodystrophy. Although such NRTIs as d4T and ZDV have been associated more specifically with lipoatrophy, it remains unclear whether the mechanism of NRTI-associated fat loss is related to that associated with the PI-induced fat accumulation; indeed, it has been suggested that NRTI-induced lipoatrophy may be related to mitochondrial toxicity.
Because insulin plays a key role in the differentiation process of adipocytes, insulin resistance may lead to alterations of lipid metabolism, as suggested by observed increases in triglyceride and cholesterol plasma concentrations. Additionally, fat redistribution associated with lipodystrophy may also interfere with insulin signaling, compounding the problem. Although the exact mechanism that leads to the lipodystrophy syndrome is unclear, various studies suggest that individual PIs target specific intracellular sites that may preferentially favor the development of either dyslipidemia or impaired glucose tolerance. Presumably, development of the lipodystrophy syndrome leads to an increased risk of both diabetes and cardiovascular disease. A recent study suggests that the paradoxical combination of fat loss and fat accumulation in different body compartments (visceral vs. subcutaneous, peripheral vs. truncal) is the result of a mobilization process, although other studies suggest otherwise. It has been proposed that the redistribution is related to the ubiquitous use of multidrug therapy containing at least an NRTI and a PI rather than to a PI alone. NRTIs have been shown to cause mitochondrial toxicity and lipoatrophy primarily via inhibition of mitochondrial DNA polymerase-g. Because white subcutaneous fat, located in the extremities and face, contains far fewer mitochondria than brown visceral fat, located predominantly in the abdomen and upper back, mitochondrial toxicity drives apoptosis disproportionately in white fat, resulting in white fat wasting, and brown fat remains the only available repository for lipids, resulting in redistribution of low-density lipoprotein (LDL) to brown fat.
Clinical Effects on Plasma Lipid Profile
The variable effects of PIs on mature adipocytes and on the adipocyte differentiation process are reflected in differences in lipid serum profiles of patients with HIV infections treated with various antiviral combinations. Early studies reported PI-induced dyslipidemia as a class effect. However, recent work demonstrates that the effect of PIs on serum lipids is variable.
Studies with earlier-generation PIs show that total cholesterol levels may increase by as much as 30% after 2-4 years of use and triglyceride levels by as much as 80%. In a recent study of 212 patients receiving PI-based ART, RTV- and LPV/RTV-containing regimens caused the highest incidence of hypercholesterolemia and hypertriglyceridemia after 12 months of therapy. These data were confirmed in a study of 353 previously treated patients who received LPV/RTV treatment. In this cohort, previous PI use was a predictor of worse hypertriglyceridemia and hypercholesterolemia, with as many as 25% of patients with triglycerides levels >400 mg/dL and 30% of patients with total cholesterol levels >240 mg/dL after 3 months of treatment. Similar effects, particularly with respect to high levels of RTV-induced lipid increases, have been observed in other studies in both adults and children.
In a comparative study of 93 HIV-infected adults treated with PIs, RTV also produced higher levels of cholesterol, LDL cholesterol (LDL-C), and triglycerides than nelfinavir or indinavir (Table 3). Saquinavir has been shown to cause a slightly higher increase in lipids than indinavir or nelfinavir and, like nelfinavir, is more likely to cause severe hypertriglyceridemia.
Change From baseline
| ||PI-Naïve ||RTV || ||IDV || ||NFV || |
| ||N=28 ||n=46 ||P ||n=26 ||P ||n=21 ||P |
|Chol ||2% ||43% ||.001 ||16% ||.05 ||29% ||.01 |
|HDL-C ||-8% ||0% ||- ||11% ||- ||33% ||- |
|LDL-C ||15% ||50% ||.001 ||26% ||- ||44% ||.05 |
|Triglyc ||6% ||101% ||.01 ||-7% ||- ||6% ||- |
In a study of 187 HIV-positive patients, higher fasting serum cholesterol and lipid levels were associated with ART, particularly among patients treated with PI-containing regimens. Multivariate analysis revealed RTV, indinavir, saquinavir, and the NNRTI efavirenz to be associated with a significant increase in total cholesterol or LDL-C.
The 2 most recently developed PIs appear to have a more favorable lipid profile than previously developed agents. In an ongoing prospective phase II extension trial of atazanavir 400 mg with d4T and 3TC, median cholesterol, fasting LDL-C, and fasting triglycerides remained unchanged at 48 weeks (158 vs. 169 mg/dL, 101 vs. 101 mg/dL, and 115 vs. 126 mg/dL, respectively). In this trial, switching from nelfinavir to atazanavir resulted in a decrease in lipid levels by week 24.19 Similarly, in an open randomized study of fosamprenavir (GW433908) vs. nelfinavir administered with abacavir and 3TC, at 48 weeks, no significant change in mean triglyceride serum levels had occurred (151 mg/dL at baseline vs. 152 mg/dL at week 48), compared with a 30% increase in the nelfinavir group (from 154 mg/dL at baseline to 200 mg/dL at week 48). Mean total cholesterol (TC) and LDL-C values rose by approximately 25% in both groups: from 152 mg/dL to 197 mg/dL and from 86 mg/dL to 119 mg/dL, respectively, in the GW433908 group, and from 153 mg/dL to 202 mg/dL and from 89 mg/dL to 122 mg/dL, respectively, in the nelfinavir group. The incidence of grade 3-4 abnormalities in TC and triglycerides was <1%. High-density lipoprotein cholesterol (HDL-C) levels also increased in both arms, from 37 mg/dL in the fosamprenavir arm and 36 mg/dL in the nelfinavir arm to optimal values of 49 mg/dL and 44 mg/dL, respectively.
PI Substitution and Dyslipidemia
Substitution of PI therapy with nevirapine, efavirenz, or abacavir has been studied as a strategy to improve both metabolic and morphologic abnormalities attributed to PIs. A relatively brief interruption in therapy (about 7 weeks) has resulted in a significant improvement in TC, LDL-C, and triglyceride levels, and most clinical studies have demonstrated improvements when substituting an NNRTI or an NRTI for a PI. In some studies, however, the changes were mixed or marginal, or lipid profiles did not improve. For instance, in one study, of 17 patients with lipodystrophy who switched from a PI to nevirapine, 7 of 17 (41%) had an objective improvement, as assessed by anthropomorphic measures, but none experienced a complete reversal in symptoms. In these patients, only triglyceride levels decreased. In another study, of 23 patients, most of whom (73%) were initially receiving indinavir, cholesterol decreased 22% and triglycerides 57% at 6 months following the switch to nevirapine. In 2 other such studies, however, improvements were marginal or did not occur.
Recently, an open-label randomized study evaluated the effect of switching from a PI or an NNRTI to the NRTI abacavir in 20 patients who experienced therapy-induced lipodystrophy. After 48 weeks of treatment with abacavir, cholesterol levels decreased significantly but remained at or above 5.2 mM (200 mg/dL). A nonsignificant decrease in triglycerides was also observed. This trend was also observed in a recent randomized trial enrolling 460 patients treated with PIs and switching to nevirapine, efavirenz, or abacavir. The switch caused no significant changes in triglyceride levels with any of these agents, and decreases in TC in the abacavir group only after 12 months of treatment. In an intent-to-treat analysis at 12 months, 10, 6, and 13% of patients in the nevirapine, efavirenz, and abacavir groups, respectively, reached a protocol-defined endpoint-a nonsignificant difference. In an as-treated analysis, significantly more patients reached a virologic failure endpoint in the abacavir group (14%) than in the nevirapine (7%) or efavirenz (5%) groups (P = 0.03). However, significantly more patients discontinued treatment in the nevirapine and efavirenz groups (17% in both cases) than in the abacavir group (6%; P = 0.01) due to adverse events.
These results suggest that PI-induced dyslipidemia is highly variable between individuals, depends on the PI used, whether it is boosted with RTV, and is more likely to occur with older agents. Substitution of the older PIs with nevirapine, efavirenz, or abacavir may lead to some improvement in lipids, although significant discrepancies exist between studies. Additionally, switching from a dual-NRTI plus PI combination to a triple-NRTI combination that would include abacavir is not recommended in those with documented or presumed NRTI resistance, due to the increased risk of virologic failure. With all of these PI substitution strategies, improvement in body habitus abnormalities has not been consistently demonstrated. It is hoped that with recently developed PIs such as fosamprenavir or atazanavir dyslipidemia will be reduced.
Insulin Resistance and Glucose Homeostasis
Insulin resistance, a key component of the lipodystrophy syndrome, is estimated to occur in 24% of adults over age 20 in the general population in the United States. Incidence and prevalence data for insulin resistance, disorders of glucose metabolism, and diabetes in HIV disease are derived from a limited number of studies, some with small data sets and some using nonuniform criteria for the determination of diabetes or insulin resistance. As observed in the general population, insulin resistance and impaired glucose tolerance are more common than diabetes mellitus and hyperglycemia in HIV-infected individuals treated with PI-containing regimens.
Mechanism of Insulin Resistance
Dyslipidemia usually appears after 3-6 months of PI therapy, whereas studies in both HIV-seronegative volunteers and therapy-naive patients with HIV97 have shown that insulin resistance may develop within 4 weeks of starting indinavir. In the latter studies that evaluated lipids at 4 and 8 weeks, respectively, lipid levels remained unchanged at the time of evaluation. However, other studies suggest that indinavir, but not RTV or nelfinavir, causes insulin or steady-state plasma glucose levels to rise within 16-24 months of therapy. A study on healthy volunteers demonstrated that even a single dose of indinavir is capable of decreasing both total and nonoxidative insulin-stimulated glucose turnover.99 By contrast, in a study of 14 patients with stable HIV infection treated with amprenavir, increases in lipid levels and body fat occurred early, whereas insulin resistance and changes in fasting serum glucose had not happened by week 48. These studies suggest that although the development of insulin resistance may precede dyslipidemia with most PIs, exceptions, such as that seen with amprenavir, do exist. Similarly, a recent randomized study enrolled 30 healthy volunteers to evaluate the effects of 5 days of treatment with atazanavir, LPV/RTV, or placebo on glucose metabolism estimated with the hyperinsulinemic euglycemic clamp technique. Whereas LPV/RTV significantly decreased glucose disposal per unit of insulin and glucose storage rate compared with placebo (7.54 mg/kg• min/u/mL vs 9.87 mg/ kg•min/u/mL, p = 0.009; and 2.61 mg/kg•min vs. 4.01 mg/kg•min, P = 0.003, respectively), atazanavir had no effect on either of these parameters.
Currently, the site(s) of insulin resistance is (are) not well documented. Recent studies show that insulin resistance is not confined to adipose tissue but also arises in skeletal muscle. Furthermore, pancreatic b-cells appear unable to release more insulin to compensate for the decrease in insulin sensitivity.
Studies aiming to elucidate the mechanism of insulin resistance and glucose intolerance have used differentiated 3T3-L1 adipocytes in culture. However, as with lipid metabolism, different PIs appear to produce different mechanisms of resistance. Whereas a direct inhibition of the glucose transporter protein GLUT4 has been demonstrated with indinavir, RTV, amprenavir, and saquinavir, nelfinavir appears to inhibit glucose uptake by interfering with insulin-dependent GLUT4 translocation or activation at the level of protein kinase B within the signaling cascade that leads to GLUT4 activation. Concurrent increase in lipolysis even in the presence of insulin was also noted, suggesting a decrease in the antilipolytic activity of insulin. Furthermore, nelfinavir appears to increase the concentration of the GLUT1 glucose transporter and stimulate basal glucose transport independently of insulin action in 3T3-L1 preadipocytes and L6 myotube muscle cell precursors. Other studies demonstrate that insulin resistance is not confined solely to adipocyte-related mechanisms. Indinavir has been shown to reduce insulin-stimulated glycogen synthesis in HepG2 hepatoma cells. Since GLUT4 is not present in HepG2 cells, the inhibitory mechanism is not likely to include inhibition of this transporter.
Hyperglycemia and Diabetes Mellitus
The prevalence of diabetes mellitus in the general adult population was estimated at 5.1% between 1988 and 1994.107 Among patients with HIV infection who developed the lipodystrophy syndrome often associated with PI treatment, 1-6% of patients will eventually develop new-onset diabetes (clinically similar to type II diabetes). Additionally, up to 60-85% of PI-treated patients will have evidence of some level of insulin resistance when intensively studied. The relationship between the introduction of PI therapy and the incidence of diabetes was demonstrated in an epidemiologic study carried out between 1994 and 2000 among HIV-positive members of the Northern California Kaiser Permanente Health Plan. The age-adjusted incidence of diabetes mellitus peaked in 1997 following the introduction of PIs in 1996 (from 97.1 cases per 10,000 in 1994 to 188.2, compared with 63.1 cases per 10,000 in 1994 to 82.3 in 1998 among HIV-seronegative individuals). Among members who were treated with early PIs, the incidence reached 319 cases per 10,000 in 1996 but decreased in 2000 to 100.8 cases per 10,000 (presumably due to the replacement of PIs with NNRTIs), compared with 83.4 and 88.4 cases per 10,000 in the seronegative and HIV population not treated with a PI, respectively. Multivariate Poisson regression modeling, adjusted for demographic factors, revealed a significant association between PI use and diabetes only when the models were stratified by year: the relative risk of diabetes for PI-treated members vs. no PI therapy was 2.49 (95% CI = 1.04-5.96). Although the time to development of diabetes or hyperglycemia is highly variable, studies have shown that alterations of glucose metabolism generally occur within 3 months after initiation of PI therapy, supporting the rapid rise in diabetes cases observed in the Northern California Kaiser Permanente Health Plan cohort.
Factors in the Development of Hyperglycemia and Diabetes
Studies with PIs in healthy volunteers indicate that individuals who develop glucose intolerance or diabetes had first-degree relatives with diabetes or a family history of the disease, suggesting that the propensity of certain HIV-infected patients treated with PIs to develop diabetes may involve a genetic component. In a retrospective case-control study of 49 patients with HIV who developed diabetes, traditional risk factors such as high mean BMI, high prevalence of fat accumulation, as well as a family history were associated with the development of the disease. Furthermore, patients with HIV had higher levels of serum alanine aminotransferase (ALT) than matched controls (66.3 vs. 43.7 U/L, respectively; P = 0.013). These data suggest that in addition to traditional risk factors, ART-induced metabolic changes and liver injury evidenced by increased serum ALT may play a role in the pathogenesis of diabetes in HIV-infected individuals. Recently, a prospective multicenter cohort study identified 69 cases of diabetes in 1785 women, the incidence of diabetes among PI users was 2.8 cases per 100 patient-years (PY) compared with 1.2 cases per 100 PY among both reverse transcriptase inhibitor (RTI) users or ART-naive patients (P = 0.01), and 1.4 cases per 100 PY among HIV-negative individuals (P = 0.06). PI use, age, and BMI were identified by multivariate analysis as independent risk factors for the development of diabetes mellitus. However, because of the small number of cases that occurred compared with the benefit of PI therapy, monitoring for glucose intolerance among older, heavier patients should be considered rather than categorically avoiding PI treatment.
Changes in bone mineral metabolism and bone histomorphometry were noted in HIV-infected individuals before the widespread use of potent ART. A recent prospective study has also reported asymptomatic osteonecrotic lesions of the hip using MRI in 15 of 339 HIV-infected patients (4.4%). This condition occurred predominantly in patients with hyperlipidemia, and in those using corticosteroids and alcohol, and appeared to be unrelated to combination ART. In another study evaluating 112 men including 60 patients treated with PIs, 35 on a non-PI regimen, and 17 untreated uninfected controls, 50% of those treated with PIs had osteopenia or osteoporosis according to the World Health Organization (WHO) classification, compared with 23% of those not treated with a PI and 29% of the control subjects (P = 0.02). However, when other risk factors such as low BMI, weight loss, steroid use, and smoking were taken into consideration, the association with PIs disappeared. A longitudinal study of 54 patients on a stable combination ART regimen for over a year suggests that loss in bone mineral density is related to initial BMI rather than to PI use; in fact, the data from this study suggest that indinavir, but not nelfinavir, is associated with increasing bone mineral density over time.119 The data suggest a shift in mesenchymal stem cell differentiation away from adipose to osteoblastic, an effect that may be related to tumor necrosis factor (TNF-a) activity.
Equally unresolved is the putative association between adipose tissue redistribution and bone mineral density that would find its basis in differentiation reprogramming. Early findings suggested that osteopenia was independent from adipose tissue distribution. However, a multivariate regression analysis of fat distribution and bone density in a study enrolling HIV-infected men with lipodystrophy, 20 HIV-infected men without lipodystrophy, and 18 age- and BMI-matched healthy controls showed an association between visceral fat accumulation and bone mineral density (P = 0.007). Further investigation suggested that indinavir and nelfinavir have opposite effects on intervertebral bone marrow fat and lumbar spine density, supporting comparable differences inferred in other studies of PI-based mechanisms governing fat distribution and bone turnover.
Non-PI antiretroviral agents in the treatment regimen may confound the effects of PIs on bone. For instance, NRTI-induced lactic acidemia has been associated with reduced spinal bone mineral density; and, in a retrospective study of patients on dual NRTI-only therapy, whole-body dual-energy x-ray absorptiometry revealed a significant decrease, although within normal limits, in bone mineral density over a 2-year assessment period.
In summary, the data linking PI-based therapy to osteopenia or osteoporosis are conflicting and methodologically problematic, particularly when other risk factors for bone disease are considered. Non-PI antiretroviral agents, HIV itself, and traditional conditions associated with diminished bone mineral density all may play a role in the relatively high rate of osteopenia seen in HIV-infected patients in several reports. Better-designed and well-controlled studies are needed to conclusively link or dissociate PI use and abnormalities bone metabolism.
ADVERSE OUTCOMES POTENTIALLY RELATED TO COMBINATION ART-INDUCED ADVERSE EVENTS
Dyslipidemia is a major contributor to the development of coronary heart disease (CHD), and strict guidelines have been proposed to prevent the development of CHD in the general population. Hyperglycemia and diabetes cause physiological changes in the vascular system that increase the risk of atherosclerosis and may lead to coronary artery disease (CAD), myocardial infarction (MI), and stroke. Because PIs are associated to various extents with dyslipidemia, hyperglycemia, or both, there is concern regarding increased cardiovascular events in PI-treated HIV patients, with a number of anecdotal cases having been reported.
Retrospective studies investigating the influence of ART on cardiovascular disease have yielded mixed results. In a study of 951 patients with HIV, 16 of whom had CHD (including 8 MIs), the incidence of CHD was associated with traditional risk factors, including cigarette smoking, hypertension, family history, and cholesterol levels, but not with PI use. In this study, however, longer exposure to NRTIs (3.7 vs. 2.5 years; P = 0.02) and low CD4+ cell counts also increased the risk of CHD. In a larger, short-term retrospective study using an anonymous database including 36,766 patients who had been treated at Veterans Administration hospitals, 41.6% of whom had been treated with PIs, there was no relationship between the incidence of cardiovascular or cerebrovascular events and combination ART over a median 17-month period of use. In an ongoing study including 4159 individuals with active HIV infections and 39,877 controls over the age of 35 enrolled in the Kaiser Permanente Medical Care Program (KPMCP) of Northern California, the prevalence of CHD events was analyzed over a 5.5-year period. During that time, 72 patients were hospitalized for CHD events, 47 of which were MIs. HIV-positive cases were at a significantly higher risk for CHD than HIV-negative cases (6.5/1000 PY vs. 3.8/1000 PY; P = 0.003).136 The rates of hospitalization for CHD were similar in PI-treated and non-PI-treated patients, however (6.7/1000 PY vs. 6.2/1000 PY; P = NS), suggesting no effect of PIs on CHD rate in this study as well. Additionally, because patients with no prior ART also had an increased risk of CHD (5.7/1000 PY), the data suggested that HIV or HIV comorbidities may contribute significantly to cardiovascular risk.
A number of studies, however, have shown a relationship between CHD and PI or ART use. A retrospective evaluation of 73,336 seropositive subjects treated with combination ART between 1996 and 1999 suggested that PI-containing regimens were associated with an increased incidence of MI with increasing length of exposure. The study analyzed an epidemiologic database including 68 French university hospitals belonging to 29 HIV treatment centers. Data collected included CD4 cell counts, HIV RNA levels, clinical events, treatment schedules, clinical trials (if applicable), mortality, and cause of death. Compared with the age-adjusted general population expectation of a rate of 10.8 MIs per 10,000 PY, patients who were exposed to PIs exhibited rates of 8.2, 15.9, and 33.8 MIs per 10,000 PY with PI exposures of <18 months, 18-29 months, and >=30 months, respectively. Similarly, in a retrospective analysis of 5672 patients from the HIV Outpatient Study over 9 years, the rate of MIs was greater among patients treated with PIs (19 MIs in 3247 patients) than among those treated with non-PI regimens (2 MIs in 2425 patients). Additionally, in a cohort of 2671 patients with HIV treated with various regimens for 12 years, the incidence of CHD and cerebrovascular disease (ischemic stroke or transient ischemic attack) was evaluated at 5.9 events/1000 PY (43 cases) and 5.0 events/1000 PY (37 cases), respectively. This represents a 2- to 3-fold greater incidence than in the general population. Cases were more likely than controls to be associated with PI use (as a class, but not individually) and the NRTI d4T.
However, in the absence of well-controlled studies, it is not possible to distinguish the relative contribution of PI-induced metabolic derangements, the degree of immunodeficiency, the state of chronic inflammation due to HIV infection, and traditional cardiac risk factors to the cardiac event rate. In an effort to address this important issue, the Data Collection on Adverse Events of Anti-HIV Drugs (DAD) study was designed to control for these variables. The study enrolled 23,468 patients from 11 previously established cohorts from 21 countries in Europe, the United States, and Australia between December 1999 and April 2001. One hundred new cases of MI were required to give the study sufficient power to detect a 2-fold difference in the incidence of MI between 2 comparable groups according to their exposure to combination ART. During the course of the study, patients were prospectively followed at each clinic visit. Standardized data collection forms were filled at enrollment and every 8 months thereafter. New cases of MI were reported to the study coordinator and coded according to the WHO Multinational Monitoring of Trends and Determinants in Cardiovascular Disease project. Twenty-four percent of patients were women, and the median age of the entire population was 39 years; 75% of patients had been previously exposed to combination ART, and 67.1% had been previously exposed to PIs. The incidence of MI increased with the duration of exposure to combination ART: from 0.24 relative rate for untreated patients to 1.34 for those exposed to combination ART for <1 year, 1.73 for those exposed for 2-3 years, 1.98 for those exposed for 3-4 years, and 2.55 for those exposed for >4 years of combination ART (P for trend <0.001). Older age (P < 0.001), smoking habit (P = 0.007), previous CVD (P < 0.001), and male gender (P = 0.04) were also associated with an increased incidence of MI, as were metabolic disorders such as higher cholesterol and triglyceride levels, hypertension, diabetes, and lipodystrophy. Although combination ART independently increased the relative rate of MI per year of exposure by 26,% it is important to emphasize that the overall risk of MI was still markedly lower than the risk of AIDS-defining illnesses seen before the advent of potent combination ART.
Surrogate Markers for CHD Risk
As in studies directly evaluating the effects of combination ART on CHD events, studies that have attempted to evaluate the effect of combination ART on surrogate markers for the risk of CHD have yielded mixed results. Because subclinical atherosclerosis is a surrogate marker of MI and stroke risk, studies have explored the potential of a relationship between carotid or femoral intima-media thickness (IMT) and PI use. In an early study of 102 patients, PI use was associated with an increased risk of atherosclerotic plaque; however, in this study, the comparative group comprised both patients treated with ART and therapy-naive individuals. Additionally, the baseline characteristics were not well balanced between groups; in particular, patients in the PI group had more advanced disease and lower CD4+ cell counts than controls. In another study evaluating HIV-infected patients treated with PIs, HIV-infected individuals who were treatment naive, and HIV-negative controls, carotid IMT was greater in the first 2 groups than in controls. Additionally, in a recent cross-sectional study in which baseline IMT values were established in 106 patients with HIV who had been treated with PIs for a median duration of 4 years, and progression was followed for an additional year, baseline IMT was greater in these patients than in matched, uninfected historical controls. Furthermore, the rate of progression of IMT at 1 year was greater in 21 of the patients than in historical HIV-negative individuals; progression was associated with age and duration of PI use.
In a study of 168 HIV-infected patients that included 136 previously treated with PIs for a mean of 2 years and 68 HIV-negative individuals, however, the presence of plaque was exclusively associated with age, male gender, high HDL-C levels, smoking, and HIV infection, but not with PI use. Furthermore, a prospective, longitudinal, matched cohort study evaluating baseline characteristics of 134 individuals distributed evenly into 3 groups of uninfected subjects, HIV-positive PI-naive patients, and HIV-positive patients with a median PI use of at least 2 years, multivariate analysis identified increased age, higher BMI, TC, LDL-C, and triglycerides as risk factors for increased carotid IMTs. Carotid IMTs were similar for all 3 groups, suggesting that prior PI use did not increase the risk of atherosclerosis. Follow-up in this study will determine whether the rate of progression of carotid IMT is influenced by PI treatment.
An increased incidence of CHD risks implies potential changes in cardiac function. However, a recent study found cardiac function (left ventricular systolic function and cardiac valve regurgitation) in patients treated with early PIs similar to that in patients treated with PI-free regimens. Interventricular septum thickness and left ventricle posterior wall thickness were somewhat greater in PI-treated patients, however, than in those treated with non-PI-containing regimens (P = 0.049 and 0.047, respectively).
In summary, current data on cardiovascular risks associated with PIs in particular and combination ART in general are inconsistent. The preponderance of the data suggest that ART does increase the risk of CHD somewhat but that the absolute risk remains low compared with the risk of AIDS-related complications in the absence of treatment. As a result, this additional risk of cardiac disease attributed to ART should not be a deterrent to the use of ART if indicated. Additional follow-up of prospectively designed longitudinal studies (such as DAD) are needed to confirm these early findings, to place them into their appropriate clinical context, and to explore further treatment-specific effects.
PI-based ART has dramatically improved the prognosis for HIV-infected patients. Problems with tolerability, however, have limited their use to some extent. Most of the older agents in this class are associated with significant GI symptoms. The newer PIs, fosamprenavir and atazanavir, however, exhibit a more favorable GI adverse event profile. Although atazanavir may cause significant hyperbilirubinemia in many patients, with jaundice in a proportion of them, this hyperbilirubinemia is infrequently a cause of treatment cessation. Rash may occur somewhat more commonly with fosampranavir than with other PIs, although again, this rarely requires stopping the drug. Educating patients and providing prophylactic management of short-term adverse effects as required are essential to prevent patients from skipping doses or discontinuing treatment. Most older PIs have demonstrated adverse effects on both lipid and glucose metabolism. Although most PIs significantly worsen lipid profiles, leading to both hypercholesterolemia and hypertriglyceridemia, atazanavir has a relatively neutral effect on plasma lipids and fosamprenavir appears to alter triglyceride levels to a minimal degree, if at all. These 2 recently approved PIs demonstrate improved subjective and metabolic tolerability profiles, which should lead to increased choice of PI-based therapy as a component of antiretroviral treatment.