Mode of death in patients with heart failure: how trends of cardiovascular mortality and death due to cancer changed in the last 25 years

Background Mutual interactions between heart failure (HF) and cancer are increasing, as many patients with a diagnosis of HF tend to die due to malignancy rather than to cardiovascular diseases. Randomized controlled trials (RCTs) span decades of HF treatment and may help to clarify the burden of cancer in the HF cohorts, as well as the temporal trends of the modes of death in this population. Aims We sought to perform a meta-analysis of RCTs to assess the temporal trends of cardiovascular mortality and death due to cancer, within the last twenty-five years. Methods We evaluated cardiovascular death and mortality due to cancer in phase 3 RCTs which involved patients with a diagnosis of chronic HF, irrespective of left ventricular ejection fraction phenotypes. Only studies which presented data on both cardiovascular and cancer death were included. RCTs published before January 1st, 2001, which recruited less than 1,000 patients and/or whose length in follow-up was less than 12 months were excluded. The pooled risk ratios (RRs) of cardiovascular and cancer death of both the control and treatment arms were calculated using a random-effects model. Temporal trends and the impact of patient and RCT characteristics on mortality outcomes were evaluated by meta-regression analysis. Results Among the 47 RCTs finally screened, 43 (91%) presented data on cardiovascular death, instead cancer mortality was assessed in only 17 (36%) studies. Of the overall 72,488 enrollees (aged 67 years; 69% men), 13,446 died, of whom 12,204 (91%) for cardiovascular diseases and 1,242 (9%) due to cancer, which was the most frequently reported mode of death among the non-cardiovascular. Cardiovascular mortality did not show a significant decline (p=0.294), instead cancer death decreased over time (p=0.002). Cancer mortality was not influenced by treatment (RR 1.12; 95% CI, 0.99–1.27), unlike cardiovascular death which presented a significant reduction in favor of treatments, compared to the control arms (RR 0.92; 95% CI, 0.88–0.96). Conclusions Among the RCTs on HF published in the last 25 years, only 36% assessed cancer as a mode of death, although malignancy was the most common cause of mortality, among the non-cardiovascular. The paucity of RCTs reporting data on cancer death may bias temporal trends. Cancer mortality should be recognized as a major outcome of the RCTs recruiting individuals with a diagnosis of chronic HF.