| |
Response (safety, efficacy) to Newly Prescribed Lipid-Lowering Therapy in Patients With and Without HIV Infection
|
| |
| |
Michael J. Silverberg, PhD, MPH; Wendy Leyden, MPH; Leo Hurley, MPH; Alan S. Go, MD; Charles P. Quesenberry, Jr., PhD; Daniel Klein, MD; and Michael A. Horberg, MD, MAS
Annals of Internal Medicine 3 March 2009 | Volume 150 Issue 5 | Pages 301-313
In a large population-based cohort study, we found that patients with dyslipidemia and HIV infection who were prescribed lipid-lowering therapy had smaller decreases in LDL cholesterol and triglyceride levels than patients with dyslipidemia and no HIV infection, although the differences were greater for triglyceride levels. Because LDL cholesterol level is the primary therapeutic target for dyslipidemia management, it is encouraging that differences in response according to HIV infection status were modest (2.7% smaller decreases in patients with HIV infection). We also determined that patients with HIV infection tolerated lipid-lowering therapy well but had higher rates of rhabdomyolysis and certain liver and muscle laboratory adverse events than patients without HIV infection.
Our findings suggest that the reduced triglyceride level responses in patients with HIV infection may be related to concomitant use of antiretrovirals that aggravate hyperlipidemia, particularly PIs.
Despite these higher adverse event rates, patients with HIV infection had similar adherence to statins and fibrates compared with patients without HIV infection and had lower discontinuation rates. Overall, discontinuation rates for lipid-lowering therapy were lower than rates reported in the literature (38).
Although the absolute risk for adverse laboratory events of grade 3 or 4 was relatively low (<5%), patients with HIV infection had up to nearly 6-fold greater odds of targeted laboratory abnormalities than patients without HIV infection beginning any lipid-lowering therapy (Table 5).
"Compared with patients without HIV infection, patients with HIV infection were less likely to achieve "optimal" LDL cholesterol levels, defined as less than 2.6 mmol/L (100 mg/dL) (23.1% vs. 30.2%; P < 0.001). Similarly, patients with HIV infection were less likely to achieve "normal" triglyceride levels, defined as less than 1.7 mmol/L (150 mg/mL) (5.2% vs. 12.2%; P = 0.003), or triglyceride levels below "very high," defined as 5.7 mmol/L (500 mg/dL) (60.1% vs. 79.2%; P < 0..001)."
"Patients with HIV infection beginning statin treatment and receiving PI-based, NNRTI-based, or mixed ART had similar decreases in LDL cholesterol levels that were all reduced compared with responses in patients without HIV infection, although the result was not significant for mixed ART (Table 3). However, the ART class effect was large for triglyceride level responses. Compared with patients without HIV infection, patients with HIV infection who were receiving gemfibrozil along with PI-based ART or mixed ART had a smaller reduction in triglyceride levels. However, patients without HIV infection had responses similar to those in patients with HIV infection who were receiving NNRTIs only (Table 3). Inferences were similar for the effect of ART class on absolute changes in lipid levels (Table 2)."
"For patients with elevated LDL cholesterol levels, mean adherence to statin therapy was similar in those with HIV infection (83.8%) and those without HIV infection (83.3%; P = 0.64), although patients with HIV infection were somewhat more likely to have 96% or greater adherence (57.4% vs. 52.9% [P = 0.037]). For patients with elevated triglyceride levels, mean adherence to fibrate therapy was also similar in patients with HIV infection (75.2%) and those without HIV infection (72.9%; P = 0.37); again, a higher percentage of patients with HIV infection had 96% or greater adherence (42.8% vs. 34.7% [P = 0.063])."
"Sensitivity analyses restricted to patients with 96% or greater adherence resulted in adjusted percentages of decline in LDL cholesterol level of 27.6 and 30.4 (P = 0.003) for highly adherent patients with and without HIV infection receiving statins, respectively. Further restriction to highly adherent statin use for more than 9 months resulted in adjusted percentages of decline in LDL cholesterol level of 23.8% for patients with HIV infection and 28.2% (P = 0..002) for patients without HIV infection. Similarly, adjusted percentages of decline in triglyceride levels for highly adherent patients receiving gemfibrozil were 41.4% for patients with HIV infection and 61.2% (P < 0.001) for patients without HIV infection. Further restriction to highly adherent use of fibrates for more than 9 months resulted in adjusted percentages of decline in triglyceride levels of 67.3% and 83.9% (P = 0.010) for patients with and without HIV infection, respectively."
ABSTRACT
Background: Antiretroviral agents, particularly protease inhibitors (PIs), may adversely affect lipid levels in patients with HIV infection. However, it is not known whether HIV-associated dyslipidemia is more difficult to treat.
Objective: To compare the effectiveness and safety of lipid-lowering therapy in patients with and without HIV infection.
Design: Retrospective cohort study.
Setting: Integrated health care delivery system from 1996 to 2005.
Patients: 829 patients with HIV infection and 6941 patients without HIV infection beginning lipid-lowering therapy for elevated low-density lipoprotein cholesterol or triglyceride levels.
Measurements: Percentage change in lipids within 12 months and adverse liver- and muscle-related clinical and laboratory events.
Results: Compared with patients without HIV infection, patients with HIV infection beginning statin therapy had smaller reductions in low-density lipoprotein cholesterol levels (25.6% vs. 28.3%; P = 0.001), which did not vary by antiretroviral therapy class. __Patients with HIV infection beginning gemfibrozil therapy had substantially smaller reductions in triglyceride levels than patients without HIV infection (44.2% vs. 59.3%; P < 0.001), and reductions with gemfibrozil varied by antiretroviral therapy class (44.0% [P = 0.001] in patients receiving PIs only, 26.4% [P < 0.001] in patients receiving PIs and nonnucleoside reverse transcriptase inhibitors [NNRTIs], and 60.3% [P = 0.94] in patients receiving NNRTIs only). Rhabdomyolysis was diagnosed in 3 patients with HIV infection and 1 patient without HIV infection. No clinically recognized cases of myositis or myopathy were observed. The risk for laboratory adverse events was low (<5%), although it was increased in patients with HIV infection.
In addition, patients with HIV infection who used pravastatin had a reduced LDL cholesterol level response compared with patients without HIV infection receiving any statin, whereas patients with HIV infection receiving simvastatin, lovastatin, or atorvastatin had no significant differences. Finally, patients with HIV infection who were prescribed gemfibrozil also had substantially reduced percentages of decline in triglyceride levels compared with patients without HIV infection. Similar inferences were observed for adjusted absolute changes in LDL cholesterol and triglyceride levels
Limitations: Laboratory measurements were not uniformly performed according to HIV status, and adequate fasting before lipoprotein testing could not be verified. Results may not be completely generalizable to uninsured persons, women, or certain racial or ethnic minorities.
Conclusion: Dyslipidemia, particularly hypertriglyceridemia, is more difficult to treat in patients with HIV infection than in the general population. However, patients with HIV infection receiving NNRTI-based antiretroviral therapy and gemfibrozil had triglyceride responses similar to those in patients without HIV infection.
Funding: GlaxoSmithKline.
Combination antiretroviral therapy (ART) has led to dramatic reductions in HIV-associated morbidity and mortality (1), resulting in more older persons living with HIV/AIDS (2). However, an ongoing concern for patients with HIV infection is the higher prevalence of age-related comorbid conditions, such as coronary disease, compared with general population rates (3, 4). Research indicates a higher risk for coronary events for patients with HIV infection who are prescribed protease inhibitors (PIs) (5) and, more recently, certain nucleoside reverse transcriptase inhibitors (NRTIs) (6). Factors contributing to the excess coronary disease risk in this population include dyslipidemia and other metabolic side effects associated with ART (7) and higher rates of tobacco use (8). Some data indicate possible direct atherogenic effects of the HIV virus itself (9) and ART (10).
Strategies to manage the increased coronary disease risk in this population include lifestyle modifications and possible adjustment to more "lipid-friendly" ART regimens (11). However, pharmacologic lipid-lowering therapy may be needed to achieve target lipoprotein goals. Recent guidelines recommend that patients with HIV infection be managed according to National Cholesterol Education Panel Adult Treatment Panel III guidelines but stress that therapeutic interventions should consider known adverse drug-drug interactions between PIs and lovastatin and simvastatin (12). Therefore, in patients with HIV infection, pravastatin and atorvastatin are recommended first-line therapy for elevated low-density lipoprotein (LDL) cholesterol levels, and gemfibrozil and fenofibrate are recommended agents for hypertriglyceridemia.
However, whether HIV-associated dyslipidemia responds to pharmacologic therapies similarly to dyslipidemia in the general population is not established. The few studies evaluating responses in patients with HIV infection involved relatively small samples (13-28), only 1 of which included a comparison with patients without HIV infection (13). Data are also lacking on the effect of different ART regimens on lipid-lowering therapy responses (13) and whether HIV status is associated with tolerability of therapy. We address these knowledge gaps by examining a large, diverse cohort of persons with and without HIV infection receiving care in an integrated health care delivery system. We hypothesized that responses to lipid-lowering therapy may be less robust in patients with HIV infection, possibly in part because of concomitant ART.
Results
Persons who started lipid-lowering therapy included 616 patients with elevated LDL cholesterol levels and HIV infection, 213 patients with elevated triglyceride levels and HIV infection, 5451 patients with elevated LDL cholesterol levels and no HIV infection, and 1490 patients with elevated triglyceride levels and no HIV infection. Compared with patients without HIV infection, patients with HIV infection had higher baseline lipid levels, a higher prevalence of coronary disease risk factors and hepatitis B or C infection, and a lower prevalence of known coronary disease or diabetes. The groups did not differ according to age or sex (Table 1). Compared with patients without HIV infection, patients with HIV infection had longer follow-up (7.8 months [SD, 3.3] vs 7.0 months [SD, 3.5] [P < 0.001] for patients with elevated LDL cholesterol levels and 8.2 months [SD, 3.2] vs 6.9 months [SD, 3.4] [P < 0.001] for patients with elevated triglyceride levels). Patients with HIV infection also had more posttreatment lipid tests than patients without HIV infection (mean, 2.5 tests [SD, 1.6] vs 2.0 tests [SD, 1.2] [P < 0.001] for patients with elevated LDL cholesterol levels and 3.1 tests [SD, 2.1] vs 2.0 tests [SD, 1.2] [P < 0.001] for patients with elevated triglyceride levels)._
Effectiveness of Lipid-Lowering Therapy
Figure 2 shows the observed changes in LDL cholesterol levels and triglyceride levels for 50 randomly selected patients with and without HIV infection. In adjusted linear regression models, patients with HIV infection receiving any lipid-lowering therapy had similar percentages of decline in LDL cholesterol level and significantly smaller percentages of decline in triglyceride levels compared with patients without HIV infection (Figure 3). However, among statin users, patients with HIV infection had smaller adjusted percentages of decline in LDL cholesterol levels than patients without HIV infection (P = 0.001). In addition, patients with HIV infection who used pravastatin had a reduced LDL cholesterol level response compared with patients without HIV infection receiving any statin, whereas patients with HIV infection receiving simvastatin, lovastatin, or atorvastatin had no significant differences. Finally, patients with HIV infection who were prescribed gemfibrozil also had substantially reduced percentages of decline in triglyceride levels compared with patients without HIV infection. Similar inferences were observed for adjusted absolute changes in LDL cholesterol and triglyceride levels (Table 2.)
Figure 3. Adjusted percentage changes in LDL cholesterol and triglyceride levels within 12 months of lipid-lowering therapy.
Results are based on linear regression with adjustment for age, sex, year, lipid-lowering therapy class, months of follow-up, baseline LDL cholesterol and triglyceride levels, number of coronary disease risk factors, past coronary disease or diabetes diagnoses, and hepatitis B or C infection. Model for any statin use is also adjusted for dose-equivalents of different individual statins and concomitant use of other lipid-lowering therapy classes. Model for gemfibrozil is also adjusted for medication dose and concomitant use of other lipid-lowering therapy classes. Ator = atorvastatin; HIV+ = HIV-infected; HIV- = HIV-uninfected; LDL = low-density lipoprotein; Lova = lovastatin; Prav = pravastatin; Simv = simvastatin.
Among patients receiving any lipid-lowering therapy, we also determined the percentage of patients achieving lipid levels below key threshold levels specified in the Adult Treatment Panel III guidelines (30). Compared with patients without HIV infection, patients with HIV infection were less likely to achieve "optimal" LDL cholesterol levels, defined as less than 2.6 mmol/L (100 mg/dL) (23.1% vs. 30.2%; P < 0.001). Similarly, patients with HIV infection were less likely to achieve "normal" triglyceride levels, defined as less than 1.7 mmol/L (150 mg/mL) (5.2% vs. 12.2%; P = 0.003), or triglyceride levels below "very high," defined as 5.7 mmol/L (500 mg/dL) (60.1% vs. 79.2%; P < 0.001).
Patients with HIV infection beginning statin treatment and receiving PI-based, NNRTI-based, or mixed ART had similar decreases in LDL cholesterol levels that were all reduced compared with responses in patients without HIV infection, although the result was not significant for mixed ART (Table 3). However, the ART class effect was large for triglyceride level responses. Compared with patients without HIV infection, patients with HIV infection who were receiving gemfibrozil along with PI-based ART or mixed ART had a smaller reduction in triglyceride levels. However, patients without HIV infection had responses similar to those in patients with HIV infection who were receiving NNRTIs only (Table 3). Inferences were similar for the effect of ART class on absolute changes in lipid levels (Table 2).
We further determined whether changes in therapy during follow-up influenced results. Overall, differences in dyslipidemia treatment changes by HIV infection status were few, although patients with HIV infection were less likely to discontinue lipid-lowering therapy and more likely to switch therapeutic classes or individual drugs within a therapeutic class (Table 4). Changes in ART during follow-up probably had little effect on lipid level changes, because nearly 80% of patients with HIV infection maintained the same ART class over 12 months, and few (<5%) discontinued ART. We also computed medication possession ratios to approximate adherence to statin and fibrate therapy. For patients with elevated LDL cholesterol levels, mean adherence to statin therapy was similar in those with HIV infection (83.8%) and those without HIV infection (83.3%; P = 0.64), although patients with HIV infection were somewhat more likely to have 96% or greater adherence (57.4% vs. 52.9% [P = 0.037]). For patients with elevated triglyceride levels, mean adherence to fibrate therapy was also similar in patients with HIV infection (75.2%) and those without HIV infection (72.9%; P = 0.37); again, a higher percentage of patients with HIV infection had 96% or greater adherence (42.8% vs. 34.7% [P = 0.063]).
Sensitivity analyses restricted to patients with 96% or greater adherence resulted in adjusted percentages of decline in LDL cholesterol level of 27.6 and 30.4 (P = 0.003) for highly adherent patients with and without HIV infection receiving statins, respectively. Further restriction to highly adherent statin use for more than 9 months resulted in adjusted percentages of decline in LDL cholesterol level of 23.8% for patients with HIV infection and 28.2% (P = 0.002) for patients without HIV infection. Similarly, adjusted percentages of decline in triglyceride levels for highly adherent patients receiving gemfibrozil were 41.4% for patients with HIV infection and 61.2% (P < 0.001) for patients without HIV infection. Further restriction to highly adherent use of fibrates for more than 9 months resulted in adjusted percentages of decline in triglyceride levels of 67.3% and 83.9% (P = 0.010) for patients with and without HIV infection, respectively.
Adverse Events
Within 1 year after starting lipid-lowering therapy, 3 patients with HIV infection and 1 patient without HIV infection were hospitalized with rhabdomyolysis, corresponding to a higher crude incidence rate for patients with HIV infection with elevated LDL cholesterol levels (5.0 vs. 0.3 cases per 1000 person-years; P = 0.036) and elevated triglyceride levels (6.9 vs. 0.0 cases per 1000 person-years; P = 0.145). No cases of myositis or myopathy were observed during follow-up; 1 patient without HIV infection had liver failure. Although the absolute risk for adverse laboratory events of grade 3 or 4 was relatively low (<5%), patients with HIV infection had up to nearly 6-fold greater odds of targeted laboratory abnormalities than patients without HIV infection beginning any lipid-lowering therapy (Table 5).
Discussion
In a large population-based cohort study, we found that patients with dyslipidemia and HIV infection who were prescribed lipid-lowering therapy had smaller decreases in LDL cholesterol and triglyceride levels than patients with dyslipidemia and no HIV infection, although the differences were greater for triglyceride levels. Because LDL cholesterol level is the primary therapeutic target for dyslipidemia management, it is encouraging that differences in response according to HIV infection status were modest (2.7% smaller decreases in patients with HIV infection). We also determined that patients with HIV infection tolerated lipid-lowering therapy well but had higher rates of rhabdomyolysis and certain liver and muscle laboratory adverse events than patients without HIV infection.
A major strength of our study was the large, population-based cohort sampled from typical clinical care settings. Our results are also generalizable to persons with access to health care. Kaiser Permanente of Northern California provides care to approximately 30% of all Northern Californians in its most populated areas, and data indicate that members are similar to the local surrounding and statewide population with respect to age, sex, and race or ethnicity, with only slight underrepresentation of those in lower and higher income and education categories (37). Other strengths include the examination of therapeutic responses to lipid-lowering therapies by using a new user design (29) and the systematic comparison of treated HIV-infected patients with matched HIV-uninfected patients. Furthermore, access to comprehensive clinical and administrative health plan databases facilitated high-quality case ascertainment of HIV status; dyslipidemia and therapy; adverse events; and potential important confounders, including coronary disease risk factors and pretreatment lipid levels.
Our findings suggest that the reduced triglyceride level responses in patients with HIV infection may be related to concomitant use of antiretrovirals that aggravate hyperlipidemia, particularly PIs. Patients with HIV infection receiving PI-containing ART had the smallest decreases in triglyceride levels, whereas patients with HIV infection receiving NNRTI-based ART had decreases similar to those of patients without HIV infection even after adjustment for confounders. Responses of LDL cholesterol levels did not seem to be influenced by ART regimen class. Other potential explanations for the reduced lipid responses in patients with HIV infection are lower adherence rates to lipid-lowering therapies and use of less effective statins because of drug toxicity concerns. However, our study results do not support these mechanisms because patients with HIV infection had better adherence to both statins and fibrates, despite higher rates of adverse events. Furthermore, we accounted for dose equivalence of different statins in the analysis.
Only 1 published study of 106 patients included a comparison of responses to lipid-lowering therapies according to HIV infection status, and patients with HIV infection were less likely to achieve target goals for total cholesterol and triglyceride levels and had smaller decreases in LDL cholesterol levels than patients without HIV infection (13). However, that study did not account for potential confounders, which is particularly notable because patients with HIV infection in the study had higher lipid levels at the start of lipid-lowering therapy and a higher prevalence of coronary disease or coronary disease risk equivalents than patients without HIV infection.
Several previous studies evaluated the efficacy of lipid-lowering therapy in patients with HIV infection (14-28). In a trial of 106 PI-treated patients, fibrates were associated with a 41% decrease in triglyceride levels and a 22% decrease in total cholesterol levels, whereas statins were associated with a 35% decrease in triglyceride levels and a 25% decrease in total cholesterol levels (20). Other smaller trials focused on individual lipid-lowering agents and observed modest decreases in lipid levels for pravastatin (14, 16, 23, 27), atorvastatin (22), rosuvastatin (19, 24), gemfibrozil (22, 25), fenofibrate (14, 15), niacin (21), and ezetimibe (17, 26). The largest observational study to date involved 245 patients with HIV infection receiving both ART and lipid-lowering therapy followed for up to 12 months (18). Patients treated with statins had a 13% decline in total cholesterol levels, and those treated with fibrates had a 29% decline in triglyceride levels; no differences were observed in lipoprotein responses when we compared patients receiving PI-based and NNRTI-based ART regimens (18). Another study of 103 patients with HIV infection receiving ART also indicated modest decreases in total cholesterol (19%) and triglyceride levels (26%) after the start of lipid-lowering therapy (28). The investigators reported that discontinuing PIs was the only predictor of improved lipoprotein responses (28).
We found that lipid-lowering therapy was well tolerated in patients with HIV infection, and few adverse events related to drug therapy were identified. This finding is consistent with other studies in patients with HIV infection, which indicated no documented adverse events (17, 19, 20, 22, 23, 25-27) or a few patients with myalgia (14, 24), high creatine kinase levels (14, 28), elevated liver enzymes (24), mild gastrointestinal symptoms (20), or rhabdomyolysis (28). However, previous studies included only relatively small samples and did not compare the risk for adverse events between patients with and without HIV infection. We determined that compared with patients without HIV infection, patients with HIV infection had a higher risk for rhabdomyolysis and up to a 6-fold higher risk for certain laboratory abnormalities. Despite these higher adverse event rates, patients with HIV infection had similar adherence to statins and fibrates compared with patients without HIV infection and had lower discontinuation rates. Overall, discontinuation rates for lipid-lowering therapy were lower than rates reported in the literature (38).
Our study has limitations. First, follow-up measurement of lipids was not uniform. Patients with HIV infection had more frequent follow-up measurement, resulting in fewer exclusions due to lack of follow-up data. Nevertheless, the resultant bias is expected to be minimal because most (>75%) patients, both those with and those without HIV infection, had at least 1 follow-up measurement. The more-frequent measurements for patients with HIV infection also resulted in longer follow-up and thus increased the opportunity to observe improvements in lipid levels. However, analyses adjusted for differences in follow-up by HIV infection status, and sensitivity analyses restricted to patients with more than 9 months of follow-up showed results similar to those of the primary study. An additional related limitation was that patients with HIV infection were more likely to be tested for potential laboratory abnormalities. However, the magnitude of the increased risk for patients with HIV infection was similar for tests performed more frequently (for example, measurement of alanine aminotransferase) compared with tests performed less frequently (for example, measurement of creatine kinase). Nevertheless, the possibility of ascertainment bias for adverse events cannot be ruled out.
An additional limitation is that we performed lipoprotein measurements in the context of usual clinical care, so despite routine instructions given to patients, we could not verify adequate fasting before lipoprotein testing. However, standard laboratory practices required a direct LDL cholesterol level measurement if it was known that the patient had not been fasting or if the triglyceride level was greater than 4.5 mmol/L (400 mg/dL). Furthermore, it is unlikely that adherence to the fasting instructions differed between patients with and without HIV infection.
We also cannot exclude the potential effect of treatment selection bias, despite accounting for a wide variety of relevant confounders. In fact, patients with HIV infection were not prescribed the same statin medications as patients without HIV infection because of the risk for toxicity. However, analyses adjusted for equivalent doses between statins with respect to potency. Furthermore, our analysis was restricted to patients starting lipid-lowering therapy, which minimized the effect of confounding by indication. Finally, our results may not be generalizable to patients without health insurance or to study populations with more women or certain racial or ethnic groups. However, our analysis of insured patients reduces confounding due to unequal access to care, and our population is highly representative of persons in California.
In summary, dyslipidemia in patients with HIV infection, and in particular hypertriglyceridemia, seems to be more difficult to treat than dyslipidemia in the general population. The reduced triglyceride responses for patients with HIV infection beginning gemfibrozil therapy may be related, at least in part, to concomitant PI-based or mixed ART. Patients with HIV infection receiving NNRTI-based ART and gemfibrozil had triglyceride responses similar to those of patients without HIV infection. Future studies should further evaluate lipid responses in patients using newer-generation antiretrovirals. Overall, lipid-lowering therapy was well tolerated, but patients with HIV infection had a higher risk for rhabdomyolysis and certain liver- and muscle-related laboratory adverse events. Given the challenges in treating dyslipidemia in patients with HIV infection and considering the recommendation for more aggressive goals of target lipid levels in patients with moderate-to-high risk for coronary disease (39), optimizing management of non-lipoprotein-related coronary disease risk, such as lifestyle factors, obesity, and hypertension, may be even more critical for patients with HIV infection.
|
|
| |
| |
|
|
|
