New Predictive Scoring System Better Identifies ESA Response in Patients With MDS

Author’s Perspective

Lead author Rena Buckstein, MD: “The novel ITACA score has three elements predictive of response to erythropoiesis-stimulating agents (ESAs): erythropoietin level <100 U/L, International Prognostic Scoring System low-risk disease, and transfusion independence (defined as <1 unit of red blood cells [RBCs] every 8 weeks for a 16-week period). With newer agents in development for this patient population, deploying this score may identify the anemic and/or RBC transfusion-dependent patients with low-risk myelodysplastic syndromes better suited for firstline agents other than ESAs.”

Erythropoiesis-stimulating agents (ESAs) are often a firstline treatment for patients with anemia and low-risk myelodysplastic syndromes (MDS), but response rates vary and can be difficult to predict.

To refine the predictive power of existing scoring systems and identify novel factors that predict response to ESAs, Rena Buckstein, MD, from the Odette Cancer Centre at Sunnybrook Health Sciences in Toronto, Ontario, Canada and researchers Enrico Balleari, MD, and Francesco Buccisano, MD, analyzed data from three large, prospective registries of patients with MDS in Italy (FISM and GROM) and Canada (MDS-CAN). They found that the ITACA score had the highest discriminating power to predict response, compared with the MDS-CAN, International Prognostic Scoring System–Revised [IPSS-R], and Nordic scores, according to results published in the American Journal of Hematology.

The study included 996 patients who had received an adequate trial of ESA therapy (defined as a minimum of erythropoietin [EPO] 40,000 IU/week or darbepoetin 300-500 μg every 2-3 weeks) for a minimum duration of 12 weeks. The median patient age at the time of ESA treatment was 76 years (range = 69-81 years), with a median time from diagnosis until ESA initiation of 1.6 months (range = 0.5-7.0 months).

The overall response rate (ORR) in 925 evaluable patients was 59 percent (range = 46%-62%). The researchers randomly divided the evaluable patient population into derivation (n=463) and validation (n=462) cohorts. Other than a longer time to ESA initiation in the validation set, the characteristics between the two cohorts were not statistically different (TABLE).

In multivariate analyses, the following factors predicted response:

  • red blood cell transfusion independence (<1 unit per 8 weeks in a 16-week period; odds ratio [OR] = 3.9; p<0.0001)
  • serum EPO concentration <100 U/L (OR=2.3; p=0.001)
  • IPSS low risk (OR=1.9; p=0.005)

The authors then assigned a score of 1 to each of those factors to develop the ITACA scoring system. Response rates for each of the risk categories (scores 0-3) were 23 percent, 43 percent, 67 percent, and 85 percent, respectively.

Unlike the Nordic, IPSS-R, and MDS-CAN scores, which “demonstrated stratified response rates that were statistically different, [the ITACA score] demonstrated the strongest agreement with ESA response in patients [with MDS],” the authors wrote.

The lack of central adjudication of erythroid response rates or analysis of response according to World Health Organization subtypes is a limitation of the study.

Dr. Buckstein added, “This predictive score does not yet identify the patients with very low response rates to ESAs and suggests that the Nordic score is still relevant for excluding these patients.” External validation of the ITACA score is still necessary, and incorporating biologic markers in future predictive scores will better identify non-responders, the authors concluded.

The authors reported funding from Celgene, Novartis, Alexion, Exjade, Amgen, Lundbeck, Sanofi, Gilead, and Roche.

Source: Buckstein R, Balleari E, Wells R, et al. ITACA: a new validated international erythropoietic stimulating agent–response score that further refines the predictive power of previous scoring systems. Am J Hematol. 2017 July 4. [Epub ahead of print]

TABLE. Response Rates to ITACA Score in Derivation and Validation Cohorts


Total p Value Validation


Total p Value
0 20 (77%) 6 (23%) 26 <0.0001 21 (78%) 6 (22%) 27 <0.0001
1 43 (57%) 33 (43%) 76 42 (57%) 31 (42%) 73
2 46 (33%) 92 (67%) 138 46 (36%) 83 (64%) 129
3 16 (15%) 89 (85%) 105 23 (21%) 84 (78%) 107
Total 125 220 345 132 204 336
ORR = overall response rate