Abstract

Erythropoiesis-stimulating agents (ESAs) reduce transfusion use in patients with anemia of CKD (A-CKD) but ESA use is associated with worse cardiovascular outcomes in all trials to date. Several studies indicate that patients with A-CKD who have initial poor responses to ESAs have even worse cardiovascular outcomes than others. No specific baseline characteristics identify poor ESA responders. If these patients could be identified early in their therapy, their anemia management might be altered to reduce their known greater risks of major adverse cardiovascular events (MACE). A standardized and early definition of poor ESA responders is desirable but has not been established.

We analyzed three groups of patients with anemia of CKD not on dialysis (A-CKD-NOD), for their initial response to ESA therapy, ESA dosing over time, and subsequent MACE outcomes. One group consisted of all patients in the TREAT trial receiving darbepoetin alfa (n=1981; New Engl J Med 2009 and 2010). The other two groups consisted of all patients in the phase 3 trials of the ESA peginesatide for A-CKD-NOD (New Engl J Med 2013), based on the ESA to which they were randomized. Patients receiving peginesatide (n=600) comprised one group, and the control group receiving darbepoetin alfa therapy (n=304) comprised the third group (for this group, events are too few for further analysis). In all patients, peginesatide was given once monthly and darbepoetin was given every two weeks, initially at a prespecified dose and then titrated to achieve and maintain target hemoglobin (Hgb) levels of 11-12 g/dL. All analyses used the patient populations with both baseline and week 4-5 Hgb values. Peginesatide was non-inferior to darbepoetin in raising Hgb in the trials. Our analyses compared early Hgb responses at week 4-5 for each patient with later Hgb levels, cumulative ESA dosage, and with primary outcomes of MACE (time to first event of stroke, myocardial infarction, or death) for each group. Patient level data was available for all patients, and all data has been reported publically. All outcomes were assessed by blinded independent review. The relationships between MACE outcomes and the hemoglobin response were assessed using three methods: Kaplan-Meier analysis with log-rank test, adjusted Cox model with model selection, and bootstrap simulations.

Results: In all groups, patients’ baseline Hgb was 10-10.5 g/dL. No individual or composite baseline characteristics were predictive for defining a poor responder group. “Poor responders” for each group, identified by week 4-5 Hgb values, remained poor responders at week 12. The average Hgb level after 12 weeks remained marginally lower in the poor responder group. Over time, the poor responder group received more ESA dose yet achieved lower Hgb increments and sustained more MACE events.

Table
Study groups
CKD not on dialysis 
Poor responder: Hgb increment from baseline n (%) Relative Risk for MACE for poor vs better responder 
Peginesatide 600 0.3 g/dL or less 112 (19) 2.56 
Darbepoetin (TREAT trial) 1981 0.1 g/dL or less 451 (23) 1.22 
Study groups
CKD not on dialysis 
Poor responder: Hgb increment from baseline n (%) Relative Risk for MACE for poor vs better responder 
Peginesatide 600 0.3 g/dL or less 112 (19) 2.56 
Darbepoetin (TREAT trial) 1981 0.1 g/dL or less 451 (23) 1.22 

In exploratory analyses, patients with little or no increment in Hgb at 4 weeks on ESA therapy, defined by having 0.3 g/dL Hgb increment with peginesatide or 0.1 g/dL or less Hgb increment with darbepoetin, have the greatest risk for MACE outcomes and should be studied prospectively for alternatives to continuing or escalating usual ESA therapy, to determine if the poor responder findings are prognostic or predictive (or both). The risk of a MACE outcome is less common but not eliminated among better responders. In sensitivity analyses, by using additional adverse outcomes (such as onset of heart failure or of hospitalization) in composite models, more events are identified and allow for greater precision for risk estimates but may not be as reliable clinically as are MACE events.

This work is independent and does not represent views of any government agency.

Disclosures

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.