Evaluation of Hematologic Endpoints Used To Assess Erythropoiesis-Stimulating Agents (ESAs): A Pooled Analysis of Data from over 10,000 Patients (pts) with Chemotherapy-Induced Anemia (CIA).

A set of commonly accepted hematologic endpoints is used to assess ESAs by regulatory bodies like the FDA, clinical guideline committees like NCCN, and independent health care assessment bodies like AHRQ. These include percentage of pts who received transfusions (TFNs) from week (wk) 1 to end of treatment period (EOTP), reached target Hb ≥11g/dL, and achieved a ≥3-point change in FACT-F score from baseline (BL). Other exploratory, unvalidated endpoints have been suggested (1 g/dL rise in Hb in 4 wks or change in area-under-the-Hb-concentration curve [ΔHbAUC]), but the AHRQ report (2006) characterized these as being insensitive to discriminate between ESAs. We examined the commonly accepted and the exploratory hematologic endpoints described above using pooled data from 10,328 pts with CIA enrolled in 20 clinical trials: 8079 darbepoetin alfa (DA) pts, 1677 epoetin alfa (EA) pts, and 572 placebo (PBO) pts. DA dosing frequencies were every 1 (QW), 2 (Q2W), or 3 (Q3W) wks; EA dosing frequencies (3 times a week and QW) were pooled. A mixed logit model was used to analyze the percentage of pts with TFNs and who achieved the target Hb, a ≥3-point change in FACT-F score, and a 1-g/dL Hb rise in 4 wks; a linear mixed model was used to analyze ΔHbAUC. Values were adjusted for BL Hb (<10 vs ≥10g/dL), dose adjustments, and whether pts received platinum chemotherapy (no vs yes). Methodology used to calculate ΔHbAUC was evaluated. For pts treated with an ESA, results (Table) appeared comparable regardless of agent or dosing frequency. For pts who received PBO, a higher percentage received TFNs, and a lower percentage achieved Hb ≥11g/dL and a ≥3-point change in FACT-F compared with pts receiving ESAs. We observed that a 1-g/dL rise in Hb in 4 wks was not reliable in predicting which pts would not receive a TFN, achieve target Hb, or achieve a ≥3-point change from baseline in FACT-F. ΔHbAUC was similar between ESA arms, but it was highly sensitive to pt characteristics and analytic method used (data not shown). In conclusion, these results indicate that the exploratory endpoints examined here lack clinical utility and validity, and provide no incremental benefit over commonly accepted endpoints in this setting.