Long-acting anemia drugs show higher risk of mortality

Jay B. Wish, MD

This is a retrospective study of Japanese dialysis patients receiving ESAs. It compares mortality risk between patients receiving short-acting ESAs (epoetin) and long-acting ESAs (darbepoetin and methoxy polyethylene glycol-epoetin beta [CERA]). 

As a retrospective study, it is subject to residual confounding and is hypothesis generating; it cannot conclusively establish cause and effect. The authors used sophisticated statistical modeling to minimize the chance of residual confounding, including three levels of adjustment for patient and facility characteristics, propensity score matching and instrumental variable analysis. 

The usage in the study of short-acting vs. long-acting ESAs – 36.3% and 63.7% of patients, respectively – was similar to U.S. prescribing patterns. Irrespective of the adjustment model, patients receiving long-acting ESAs had a higher rate of death from all causes, cardiovascular disease, cardiac disease, stroke, non-cardiovascular disease, stroke and malignancy. The hazard ratio for long-acting ESAs and all-cause mortality ranged from 1.11 to 1.20 depending upon the model used but was always statistically significant.  The statistically significant hazard ratio persisted for all patient subgroups except those with the lowest ESA dose and lowest erythropoietin resistance index, suggesting that the toxicity of long-acting ESAs is dose related. 

Interestingly, the toxicity of long-acting ESAs was not proportionate to their duration of action as patients receiving darbepoetin had a higher all-cause mortality rate than those receiving CERA. Since CERA stands for continuous erythropoietin receptor activator, one could speculate that the toxicity of longer-acting ESAs might be related to the duration of receptor activation leading to off-target effects on non-erythropoietic tissues. However, since darbepoetin was associated with higher all-cause mortality than CERA, despite the shorter duration of action of the former, it is possible that the modification of the recombinant epoetin molecule or the darbepoetin vehicle contributes to its toxicity. 

It should be noted that a previous study (Winkelmayer W, et al. Am J Kidney Dis. 2015; doi:10.1053/j.ajkd.2015.02.339) comparing long-term outcomes among dialysis facilities using primarily darbepoetin vs. epoetin found no differences between the groups. Limitations of the Sakaguchi and colleagues study include missing data, lack of information regarding iron supplementation (which tends to be conservative in Japanese patients), potential crossover of ESA use and the single ethnicity of the patients, which may limit extrapolation to other populations. 

Only a long-term randomized controlled trial comparing short-acting vs. long-acting ESAs can conclusively prove a cause and effect relationship between ESA duration of action and mortality risk. Such a study is unlikely to occur as newer agents for the treatment of anemia, such as hypoxia inducible factor stabilizers, are on the horizon.