Do Erythropoiesis-Stimulating Agents (ESAs) Affect Survival in Anemic Patients with Myelodysplastic Syndromes (MDS)?

Abstract 2794

ESAs play an important role in the amelioration of MDS-related anemia with Phase II trials demonstrating a 20–40% erythroid response rate. Randomized trials in breast, non-small cell lung, and head and neck cancers have suggested a deleterious effect of ESAs on progression-free and overall survival. In contrast, the single Phase III trial randomizing between ESA administration and supportive care in lower- risk MDS patients (E1996) demonstrated no differences in overall survival or in the incidence of acute myeloid leukemia (AML) transformation (Blood 2009). Retrospective analyses of MDS studies have also suggested no such negative impact of ESAs. To further examine the impact of ESA administration on progression to AML or death (ie, progression-free survival, PFS), we combined data on MDS patients from two sources: the International MDS Risk Assessment Workshop (IMRAW) pooled data on ESA untreated patients from 7 historical registries accrued when ESAs were not commercially available (Blood 1997) and Surveillance Epidemiology and End Results (SEER) registries from 2001–2005 linked to Medicare claims including ESA treated and untreated patients. Conceptually, comparison of these groups could function as an “historical experiment” comparing cohorts which differ primarily in the widespread availability of ESAs. Because IMRAW included baseline hemoglobin measurements but did not indicate transfusion use, while SEER-Medicare documents transfusion use but not hemoglobin, we identified MDS patients with a high likelihood of transfusion need as follows: IMRAW patients with hemoglobin <8.5, and SEER-Medicare with at least one transfusion within 3 months pre-or post-diagnosis. Patients were selected and grouped by risk (IMRAW: IPSS low and intermediate-1 versus intermediate-2 and high risk; SEER: WHO classification of refractory anemia, refractory anemia with ring sideroblasts, refractory cytopenia with multilineage dysplasia or del(5q) syndrome versus refractory anemia with excess blasts). Compared to IMRAW (N=199), SEER-Medicare (N=260) had a higher proportion of MDS lower-risk (74% versus 61%), and older (78% versus 37% aged > 75 years) patients. Cox proportional hazard (PH) estimates pooling ESA treated from SEER and (non-treated) IMRAW patients, controlling for age, sex, and risk group (Model 1) failed to find an effect of ESAs (HR 0.79, p=.147) on PFS. As expected, higher MDS risk (HR 3.05, p<.001) and increasing age were associated with increased hazard of AML or death. We further compared the IMRAW (all patients with Hgb < 8.5) to SEER-Medicare (transfusion-receiving) including both ESA-treated and ESA-naïve patients (Model 2) and re-estimated both Models 1 & 2 stratified by MDS prognostic risk group. Relative to IMRAW, SEER-Medicare patients (ESA users and non-users) demonstrated an increased risk of AML/death (HR 1.34, p=.020). ESA use was not associated with a significant difference in survival when comparison of ESA-treated SEER-Medicare to IMRAW (Model 1) was restricted to either MDS-high risk or MDS-lower risk patients. Subset analysis of Model 2 by risk group (ESA treated and naïve in SEER-Medicare compared with IMRAW) indicates a somewhat larger effect among higher-risk compared to lower-risk MDS (HR 1.51, p=0.037 versus HR 1.21, p=0.264). When ESA treated and untreated in SEER-Medicare alone were compared to each other, controlling for risk group, age, race, sex, comorbid conditions, and prior cancer, a strong protective effect of ESAs was found (PH model, HR 0.63, p<.001); however this difference was not sustained when analyses controlled for potential selection bias (HR: 1.54, p=0.14). Although the SEER-Medicare cohort demonstrated inferior progression-free survival compared to the IMRAW cohort, even when age and MDS risk group were controlled for, this analysis can identify no specific negative impact of the use of ESA in this population.