Background: Whether abacavir (ABC) exposure contributes to myocardial infarction (MI) risk remains unclear. Large observational studies, including D:A:D, found ABC use associated with nearly 2-fold increased MI risk; other studies have found no association. We evaluated MI risk associated with recent ABC use among patients in the largest North American cohort study: NA-ACCORD.
Methods: Incident MIs from seven U.S. cohorts of HIV-infected persons in NA-ACCORD were centrally adjudicated using MESA criteria and classified per the Universal Definition of MI as either atherothrombotic (type 1) or demand ischemia (type 2). Adults who were ABC-naïve at entry were included and followed until MI, death, one year after last CD4 or HIV RNA measurement, or 12/31/2010. Recent ABC use was defined as prescription within the prior 6 months. We used pooled logistic regression models to estimate adjusted hazard ratios and 95% confidence intervals for MI risk associated with recent ABC use adjusting for demographics, cigarette smoking, diabetes, hypertension, renal impairment, high total cholesterol, high triglycerides, statin use, CD4, previous protease inhibitor use, calendar year, and cohort.
Results: 16,733 adults contributed 301 incident MIs and 64,607 person-years of follow up. Persons who initiated ABC were significantly more likely to have traditional MI risk factors (older age, smoking, hypertension, low HDL, high total cholesterol, and black race) and factors linked to inflammation (history of IDU, HCV infection, CD4 <200 cells/mm3, detectable HIV RNA, and history of AIDS). Without adjustment, the MI risk associated with recent ABC use was 1.88 (1.35, 2.60). In an adjusted model similar to that used in the D:A:D study, the MI risk associated with recent ABC use was 1.71 (1.11, 2.64). In a model that further adjusted for traditional MI and HIV-related risk factors measured prior to ABC use as confounders of the ABC/MI relationship, the MI risk associated with recent ABC use was 1.34 (0.96, 1.88); results stratified by MI type were similar.
Conclusions: We found an increased risk for MI associated with recent ABC use that diminished in magnitude and statistical significance after adjusting for traditional and HIV-associated MI risk factors, many of which were significantly more prevalent in ABC users. Further analyses are underway to account for potential time-dependent confounding of risks for MI.