Chronic myeloid leukemia (CML) has a reliable prognostic scoring system during the chronic phase, but blast phase (BP) lacks comprehensive analysis of prognostic factors or standardized scoring systems. Despite the era of tyrosine kinase inhibitors (TKIs), outcomes for BP patients remain poor. From June 13 to 16, 2024, the 29th European Hematology Association (EHA) Annual Congress was held in Madrid, Spain. At the congress, based on data from CML-BP patients diagnosed across multiple countries, the European Leukemia Network established a prognostic scoring system for BP-CML, clearly stratifying CML-BP patients into three distinct survival groups. Hematology Frontier is honored to feature commentary from Professor Bingcheng Liu, Director of the Leukemia Diagnosis and Treatment Center at the Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences, on this study.Research Methodology
Using a Cox proportional hazards model for overall survival (OS), the study incorporated various factors including platelets, hemoglobin, white blood cells, basophils, eosinophils, peripheral blood blasts, bone marrow blasts, age, immune phenotype, transcript type (typical vs. atypical), high-risk additional chromosomal abnormalities, BCR::ABL1 kinase domain mutations, central nervous system involvement, extramedullary disease (EM), time from CML diagnosis to BP occurrence, initial CML phase at diagnosis, and gender. Treatment details unknown at diagnosis were excluded. Model selection utilized likelihood ratio tests, and score assignment used the minimum p-value method. Missing data were handled using multiple imputation with fully conditional specification, and model validation employed five-fold cross-validation.
Research Findings
275 eligible patients were included for evaluative analysis, with a median follow-up of 45 months post-BP diagnosis. The median OS was 18.9 months (95% CI: 16.8 to 26.1 months), with 170 deaths recorded. On average, each variable had 10% missing data. Significant variables in the final model included peripheral blood blasts (HR: 1.06 per 10%, 95% CI: 1.02 to 1.05, P = 0.049), platelets (HR: 0.92 per 100 Gpt/L, 95% CI: 0.86 to 0.99, P = 0.03), (BP diagnosis age / 10)² (HR: 1.04 per unit, 95% CI: 1.02 to 1.05, P < 0.001), initial CML phase (BP vs. CP, HR: 0.67, 95% CI: 0.48 to 0.94, P = 0.02), immune phenotype (lymphoid vs. other phenotypes, HR: 0.42, 95% CI: 0.29 to 0.61, P < 0.001), and extramedullary disease (presence of EM vs. no EM, HR: 1.72, 95% CI: 1.13 to 2.60, P = 0.01).
Using this model, a prognostic scoring system stratified patients into three groups: low-risk (14% of patients, median OS 97 months, 95% CI: 67 months to indeterminate), intermediate-risk (59%, median survival 22 months, 95% CI: 18 to 35), and high-risk (27%, median survival 9 months, 95% CI: 6 to 13). Cross-validation demonstrated average Harrell’s C statistic of 0.69 for the full model and 0.66 for the scoring system, indicating good performance.
Research Conclusion
This multicenter study identified six key prognostic factors for CML-BP and proposed a novel scoring system. While internal validation showed good performance, external validation with independent datasets is strongly recommended. Despite inherent limitations in registry data, such as occasional data loss and heterogeneity among participating centers, this study provides significant insights into prognostic factors for CML-BP, with implications for patient treatment decisions and prognosis improvement.
Expert Commentary
Professor Bingcheng Liu: Even in the TKI era, long-term outcomes for CML-BP remain challenging, especially for patients unable to undergo curative allogeneic stem cell transplantation due to donor availability, physical condition, or economic factors. This study, utilizing multinational CML-BP registry data and incorporating multiple clinical indicators, established and validated a prognostic assessment system. The system clearly demonstrates heterogeneity among BP patients and significant survival differences between groups. This suggests that stratifying treatment based on prognosis, donor availability, and patient conditions, including transplantation, could improve outcomes. The study did not include treatment details before and after BP onset, limiting assessment of treatment factors on the stratification system.
