The observation that TNF-α is elevated in individuals with advanced heart failure (HF) prompted several high-profile clinical trials investigating whether TNF inhibitors could be used to treat HF. The failure of these trials and reports of RA patients treated with TNF inhibitors developing new-onset or worsening HF has raised questions as to whether TNF inhibition is beneficial or detrimental to the myocardium. Here, Listing et al (Arthritis Rheum 2008; 58: 667) explore the association of TNF inhibitor use HF in a prospective cohort study of RA patients.
Methods
RA patients enrolled between May 1, 2001 and May 1, 2006 in the German RABBIT (Rheumatoid Arthritis-Observation of Biologic Therapy) cohort, a prospective cohort study following clinical outcomes in RA patients from the time of initiation of TNF inhibitors, were followed for reported incident HF confirmed by medical records review. The association of TNF inhibitor use with incident HF or worsening of prevalent HF over a maximum of 60 months was explored, accounting for RA disease and treatment characteristics. A confirmatory nested case-control study was performed comparing cases of incident HF to controls matched on age, gender, baseline CVD, BMI, and follow-up time.
Results
2,757 patients starting TNF inhibitors were compared to 1,491 staring a new non-biologic DMARD. The majority of patients were female (78%) with a mean age of 54 years. Cardiovascular disease of any type was present in 37% of patients and did not differ by TNF inhibitor treatment status. Patients treated with TNF inhibitors tended to be younger than those not treated and had higher disease activity scores and lower physical function status. Approximately 95% of patients completed follow-up or had the outcome of interest. Prevalent HF at baseline was noted in 98 patients, 8 of whom were classified as class III HF (7 of whom were treated with TNF inhibitors). Drop-outs did not differ from completers in demographic, RA, or baseline cardiovascular characteristics. Deaths during follow-up were seen in 101 patients, 14 secondary to HF.
Incident HF was observed in 25 patients and worsening of prevalent HF in 12 patients. Factors significantly associated with HF included age, male gender, cardiovascular disease at baseline, baseline functional capacity, and DAS28 at follow-up. For both incident HF and worsening of prevalent HF, the frequency of TNF inhibitor treatment did not significantly differ and the risk of incident HF was not significantly associated with TNF inhibitor use in models adjusted for age, gender, prevalent cardiovascular disease, BMI, baseline functional capacity, and DAS28 at follow-up (Hazard ratio 1.66 (95% CI 0.67-4.10)). A possible association with prednisolone dose and worsening of prevalent HF was observed.
Conclusions
In this large cohort of RA patients followed from the time of beginning TNF inhibitor therapy, the use of TNF inhibitors was not significantly associated with an increased risk of incident HF or worsening of prevalent HF.
Editorial Comment
Because TNF inhibitors are usually prescribed to patients with more active or severe RA, then outcomes associated with RA activity or severity may be incorrectly ascribed to the therapy rather than the disease. This concept, known as confounding by indication, can make it difficult to parse out helpful vs. harmful effects of therapeutics in non-randomized (i.e. cohort/case-control) studies. Here, the confounding effects of disease activity and functional status on HF were identified and adjustments made for their potential confounding effects. These results provide some reassurance that HF risk associated with TNF inhibitors is low and, by controlling inflammation and allowing patients to maintain physical functioning, may actually reduce the incidence of HF in the long run. The study is not perfect, and criticisms include small overall numbers of events, relatively short follow-up time, and a clinical outcome of HF that may permit misclassification. Nevertheless, these findings are reassuring. One striking observation is that the majority of subjects with class III HF were prescribed a TNF inhibitor, an action that would be contrary to FDA prescribing recommendations for U.S. physicians.
