A critical comparison of Sokal, Euro, and Eutos risk scores in chronic myeloid leukemia patients treated with tyrosine kinase inhibitors.

Background: In the era of molecular hematology, the prognosis of BCR- ABL+chronic myeloid leukemia (CML) is still based on three prognostic systems including few clinical and hematological variables: age, platelet count, blast cells, eosinophils, basophils (percentage in peripheral blood) and spleen size (assessed by manual palpation; maximum distance in cm from costal margin). The three prognostic systems are the Sokal score (Sokal et al, Blood 1984;63:789-799), the EURO score (Hasford et al, J Natl Cancer Inst 1998;90:850-858), and the EUTOS score (Hasford et al, Blood 2011;118:686- 692), based on the analysis of international, multicentric series of newly diag- nosed, chronic phase (CP) CML patients, treated with conventional chemother- apy (Sokal), interferon-alfa (IFNa)-based regimes (EURO), or imatinib-based regimes (EUTOS), respectively. Aims: In all the main studies exploring the efficacy of tyrosine-kinase inhibitors (TKIs), at least one of these scores has been used: all three scores were able to predict the response and/or the outcome, but the three scores are different, and there is no agreement on which score may be more useful and should be adopted. The aim of our analysis was to identify which score has been more frequently used and which one is able to give more consistent results. Methods: We reviewed the most relevant reports of the 13 important clinical studies, published in peer-reviewed journals between 2003 and 2014. More- over, we analyzed 559 patients enrolled within 3 multicentric prospective studies conducted by the GIMEMA CML WP (NCT00514488, NCT00510926, obser- vational trial CML023). Results: The Sokal score was used in 12 studies, the EURO score in 2 studies, the EUTOS score in 4 studies (Table 1). The results are difficult to interpret, because sometimes only the p-values for differences (not the response rates), and sometimes only the response rates (not the p-value) were reported. The relationship between the risk and survival was analyzed in few studies: the Sokal score was able to predict the survival in three studies (Hammersmith, Czeck registry and Swedish registry), the EURO in only one study (Swedish registry), and the EUTOS in no studies. Therefore, we tested all three scores in 559 newly diagnosed CP-CML patients who were enrolled in GIMEMA studies with imatinib front-line. No score was able to predict the early molecular response. The complete cytogenetic response (CCyR) and the major molecular response (MMR) rates by 1 year were better predicted by Sokal (p=.006 and. 001) and EURO (p=.002 and .002) scores than by EUTOS score (p=.009 and .010). The rate of deep molecular response (MR4.0) by 6 years was better pre- dicted by Sokal score (p<.001) than by EURO (p=.019) or EUTOS (p=.031) score. The 6-year overall survival was better predicted by Sokal (p=.002) and EURO (p=.003) score than by EUTOS score (p=.160). Summary and Conclusions: There are no studies supporting the superiority of any risk score over the others. The EUTOS score could be the first choice because based on TKIs-treated patients. However, the Sokal score may still be considered the reference risk score, because it has been validated in many more independent studies, with more consistent results. All the three scores could be refined introducing other biological variables (i.e. clonal chromosome abnormalities in Ph+cells). However, more progress in CML biology is needed to move from the era of clinical/hematologic prognosis to an era of molecular- based prognosis.