A predictive model of response to erythropoietin stimulating agents in myelodysplastic syndrome: from the Canadian MDS patient registry.

ABSTRACT

Prediction of response to erythropoietin stimulating agents (ESAs) in anemic MDS patients is often based on the Nordic score. We wished to validate the Nordic score (IWG 2006 response criteria) in a larger cohort and determine if other variables such as IPSS/IPSS-R, ferritin, LDH, and a novel European ESA response score (Santini 2013) were of prognostic importance. We analyzed 208 ESA-treated MDS patients (WHO 2008 criteria) from a prospective registry. Ninety-four and 93% had lower risk scores by IPSS (low/int - 1) and IPSS-R (low/very low), respectively. Erythroid response was achieved in 94 patients (47%); responses were similar with erythropoietin (50%) and darbepoetin (39%; p = 0.2). The Nordic and European scores were both validated on univariate analysis. Variables independently predictive of response in multivariate analysis were low-risk IPSS score (OR 0.1, p = 0.0016) and serum EPO level < 100 mIU/mL (OR 8.7, p < 0.0001). We propose a new ESA response score, consisting of (a) IPSS low score (1 point) and (b) serum EPO levels < 100 mIU/ml (2 points), yielding scores ranging from 0 to 3, with response rates varying from 17 to 81%. The Nordic score has validity but we observed lower than the expected response rates in the best risk group. Our proposed scoring system appears more discriminating but needs validation.