Can the revised IPSS predict response to erythropoietic-stimulating agents in patients with classical IPSS low or intermediate-1 MDS?

To the editor:

The “classical” International Prognostic Scoring System (IPSS), based on cytogenetics, marrow blast percentage, and number of cytopenias, has played a major role in prognosis assessment in myelodysplastic syndromes (MDS).¹  The recently published revised IPSS (IPSS-R), using the same parameters, but with 5 rather than 3 cytogenetic subgroups and new cutoff values for cytopenias and marrow blast percentages, refines the original IPSS prognostic value.^2,3  However, its prognostic value for response to erythropoiesis-stimulating agents (ESA) has not been assessed. We analyzed it retrospectively in 456 IPSS low/intermediate (Int)-1–risk MDS patients treated with ESA in France, Germany, and Italy.

Those 456 patients had serum erythropoietin (EPO) <500 mU/mL and hemoglobin (Hb) ≤10 g/dL and had received ESA (EPO alfa or β 40 000-60 000 IU/week, or darbepoetin 150-300 µg/week) for at least 12 weeks. In addition to IPSS-R parameters, age, sex, serum EPO level, serum ferritin (SF), red blood cell (RBC) transfusion requirement before ESA onset were assessed for response to ESA (based on International Working Group 2006 criteria), and overall survival (OS) from ESA onset. Characteristics of the 456 patients at ESA onset are listed in Table 1. Seventy-one percent of the patients had never received RBC transfusions, and their median Hb level was 9.3g/dL (range 7.0-10); 29% of patients had received at least 4 RBC concentrates/8 weeks before ESA onset (with a maximum of 12 concentrates). Median SF was 357 ng/mL and serum EPO was 60 mU/mL (range 6-483). IPSS was low in 55% and Int-1 in 45% of patients. IPSS-R was very low in 15%, low in 61%, intermediate in 19%, and high in 4% of the patients.

A total of 303 (61%) patients had an erythroid response, including 72% and 52% of low and Int-1 risk patients, respectively (P = .001). Using IPSS-R, 85%, 68%, 48%, and 31% of patients had erythroid response in the very low, low, intermediate, and high-risk groups, respectively (P < .0001).

Other prognostic factors of erythroid response, in univariate analysis, included individual IPSS-R parameters analyzed according to IPSS-R thresholds (Hb level, platelet count, absolute neutrophil count, marrow blast %, cytogenetics), serum EPO level, SF (variables tested as continuous variables), and previous RBC transfusions. In multivariate analysis, IPSS-R, serum EPO, and SF remained significantly associated with erythroid response (P < .0001, P < .0001, and P = .002, respectively).

Applying 1 point to each of the following unfavorable variables of response to ESA, serum EPO >200 mU/mL (=1), SF >350 ng/mL (=1), and IPSS-R (very low = 0, low = 1, intermediate = 2, and high = 3) yielded a score ranging from 0 to 4, with response rates of 85%, 80%, 64%, 40%, and 20%, respectively. As expected, IPSS-R also had strong prognostic value for OS (not shown).

Thus, in this patient cohort with overall favorable prognostic factors of response to ESA according to the Nordic score⁴  (ie, serum EPO <500 mU/mL and no or limited transfusion dependence), IPSS-R alone, and even better, a score ≥3 (using IPSS-R, serum EPO, and SF) proved useful to identifying patients with low response to ESA who also have worse OS⁵  and may require alternative treatments.