TNF Antagonists and Cardiovascular Disease Risk

Chicago—Data from a large epidemiologic study presented at the ACR meeting suggest a trend toward reduced cardiovascular disease (CVD) risk among patients with rheumatoid arthritis (RA) during the first 6 months of initiating treatment with tumor necrosis factor (TNF) antagonists. If confirmed in future clinical trials, this result suggests that potent immunosuppression with TNF antagonists may reduce CVD risk compared with nonbiologic disease-modifying antirheumatic drugs (DMARDs), according to Daniel H. Solomon, MD, the study’s lead author who works in the division of rheumatology at Brigham & Women’s Hospital in Boston. Dr. Solomon discussed the results in an oral abstract session. Prior evidence had suggested that TNF inhibitors may reduce the risk of CVD events by a number of potential mechanisms, including improving insulin resistance, improving endothelial dysfunction, slowing plaque formation, inhibiting plaque rupture, and reducing ischemic damage. However, data are lacking on who may be at most risk of CVD, the timing and pattern of CVD in RA patients, and the specific types of cardiovascular events. To examine some of these issues, Dr. Solomon and colleagues conducted a comparative effectiveness study of CVD risk associated with new use of TNF inhibitors versus new use of nonbiologic DMARDs among patients with RA. Using data collected from a large US health maintenance organization, Medicare, and Medicaid, they identified patients with a diagnosis of ≥2 RA diseases between 1998 and 2007 previously treated with methotrexate who received subsequent treatment with a TNF inhibitor or nonbiologic DMARDs. A comparison of the effect of RA on CVD risk was done from the time patients started receiving either the addition or substitution of a TNF inhibitor or DMARD to methotrexate. Investigators assessed the CVD risk by relevant subgroups, which included age, sex, prior CVD or diabetes mellitus, and use of statins or nonsteroidal anti-inflammatory drug. Patients were matched based on propensity score (PS), which controls for variables that predict interventions and reduces potential confounders to a single variable. Among the potentially eligible patients with RA using methotrexate identified in the databases, 7780 new users of TNF inhibitors were PS-matched to 9034 new users of DMARDs. Using Cox proportional hazards regression models fit to the composite CVD end point, the study conducted 2 analyses: an intent-to-treat analysis that carried the first exposure forward and an as-treated analysis that censored patients 30 days after exposure ended. Based on evaluation of the primary end point (a composite of myocardial infarction, stroke, and revascularization), the study found the incidence rates per 100 person-years for the composite cardiovascular end point were 2.26 (95% confidence interval [CI], 1.78-2.87) for TNF inhibitors and 2.99 (95% CI, 2.41-3.71) for DMARDs. Analysis of outcomes at 6 months showed that the glucocorticoid adjusted PS-matched hazard ratio (HR) for the intent-to-treat analysis was 0.79 (95% CI, 0.60-1.04) and 0.72 (95% CI, 0.62-0.99) for the as-treated analysis. For the as-treated analysis, the HR at 12 months was 0.86 (95% CI, 0.66-1.13). The only variable that appeared to influence outcomes was age. Among persons ≥65 years of age, a potential cardiovascular benefit was seen for those treated with TNF inhibitors. The HR for these patients was 0.51 (95% CI, 0.33-0.78) compared with 1.20 (95% CI, 0.72-2.01) for patients <65 years of age. According to Dr. Solomon, the age effect was a novel finding without a biologic explanation.