Can prognostic scoring systems for chronic myeloid leukemia as established in adults be applied to pediatric patients?

In contrast to adult medicine, specific scoring systems predicting the treatment response for an individual pediatric patient (pt) with chronic myeloid leukemia (CML) have not yet been defined. We evaluated to what extend prognostic scores as described for adults (e.g., Sokal, Hasford, EUTOS score) resulted in comparable risk group categorizations in a pediatric cohort. Parameters for score calculation were extracted from a data set of 90 patients enrolled into trial CML-PAED-II and treated by a standard dose of imatinib. At month 3 and at month 6, treatment response was analyzed based on the transcript ratio BCR-ABL1/ABL1. By the EUTOS, Hasford, and Sokal scores 81, 59, and 62 % of the patients were categorized as low risk, respectively; 19, 14, and 16 % of the patients as high risk, respectively; and by Hasford and Sokal scores 27 and 22 % of the patients, respectively, as intermediate risk. Twenty-seven out of 72 patients analyzable (38 %) exhibited a transcript ratio >10 % at month 3. We show that only the EUTOS score, but not the Sokal and Hasford score, correlates with this early outcome (p = 0.008). Analyzing the EUTOS score separately, we can demonstrate that lowering the cutoff from 87 to 48 points for categorization in low- and high-risk individuals increases the odds ratio from 2.4 (95 % CI 0.6 to 10.4) to 3.6 (95 % CI 1.3 to 10.9). Data are provided on the distribution of risk categories and resulting discrepancies when adult scores are applied on children and adolescents with CML at diagnosis. A larger number of patients and longer follow-up are still needed to develop a prognostic score specifically adapted to the pediatric and adolescent age cohorts.