Evaluating Mortality with Erythropoietin Stimulating Agents (ESAs) in the Oncology Setting: The Importance of Selecting Trials That Include Survival or Loco-Regional Tumor Control as a Primary Endpoint

Abstract

In February 2008, we reported increased risks of mortality (hazard ratio (HR) 1.10, 95% confidence interval (CI) 1.01–1.20) and venous thromboembolism (VTE) (relative risk (RR) 1.57, 95% CI 1.31–1.87) among anemic cancer patients. Fifty-one trials including 13,611 patients and 38 trials including 8,172 patients were included in the mortality and VTE analyses, respectively. At ASH 2007, we reported that trials within the nephrology setting that measure survival as a primary or secondary safety endpoint should be used over trials that measure efficacy as a primary or secondary endpoint to detect significant safety signals. Herein, we compare the percentage of trials and patients included in various published meta-analyses within the cancer setting that prospectively evaluated overall survival, disease-free survival, progression-free survival and/or loco-regional tumor control as a primary outcome, to determine whether these studies can better identify safety signals than studies that do not include these primary endpoints. Randomized trials of ESA administration to anemic cancer patients that evaluated survival or loco-regional control as primary endpoints were selected for review. Overviews published since 2006 were identified in Medline databases. HRs for mortality and 95% CIs were evaluated for the survival-focused meta-analyses and published overviews. A subset analysis of our 2008 meta-analysis including studies that prospectively evaluated overall survival, disease-free survival, progression-free survival and/or loco-regional tumor control as a primary outcome resulted in an increased risk of mortality (hazard ratio (HR) 1.18, 95% confidence interval (CI) 1.04, 1.35). Among recently reported meta-analyses, safety signals are most apparent when the percentage of patients and trials that evaluated survival or loco-regional tumor control as a primary outcome is greatest, emphasizing the importance of trial selection when conducting safety-focused meta-analyses.Wilson et al. 2007Bohlius et al. 2006Seidenfeld et al. 2006Bennett et al. 2008Bennett et al. subset analysis 2008# of trials (# of patients)28 (5,308)42 (8,167)35 (6,918)51 (13,611)14 (5,785)Median/Mean # of patients146/216147/201147/201223/271370/413% of trials included that prospectively evaluated overall survival, disease –free survival, progression-free survival, or loco-regional tumor control as a primary outcome (# included, total #)7.1% (2/28)16.7% (7/42)17.1% (6/35)27.5% (14/51)100% (14/14)% of patients included that prospectively evaluated overall survival, disease –free survival, progression-free survival, or loco-regional tumor control as a primary outcome (# included, total #)24.3% (1,290/5,308)26.7% (2,181/8,167)29..9% (2,068/6,918)39.1% (5,324/13,611)100% (5,787/5,787)HR for mortality(95% CI)1.03 (0.92–1.16)1.08 (0.99–1.18)1.11 (0.99–1.23)1.10 (1.01–1.20)1.18 (1.04, 1.35)