Comprehensive evaluation of time-to-response parameter as a predictor of treatment failure following imatinib therapy in chronic phase chronic myeloid leukemia: which parameter at which time-point does matter?

Publication Type:

Journal Article


American journal of hematology, Volume 85, Issue 11, p.856-62 (2010)


2010, Center-Authored Paper, Clinical Research Division, Leukemia, Myeloid, Chronic-Phase, Piperazines, Predictive Value of Tests, Proportional Hazards Models, Pyrimidines, Remission Induction, ROC Curve, Time Factors, Treatment Failure


Early recognition of high-risk patient is important to improve long-term outcomes following imatinib therapy for chronic myeloid leukemia (CML). Some controversy surrounds the question, which of short-term response parameters at which time-point, including complete cytogenetic response (CCyR) or major molecular response (MMR) at 6 or 12 months, is the best predictor for treatment outcomes. In this comprehensive analysis, we adopted landmark analysis method, time-dependent Cox's proportional hazard model, and receiver-operating characteristics (ROC) method to analyze time-to-response parameter as predictor of long-term outcomes in 187 chronic phase (CP) CML patients. Regardless of the methods of analysis, earlier achievement of short-term response such as CCyR or MMR could predict the higher probability of achieving better interim outcome (such as treatment failure or loss of response [LOR]). Similar to the findings from other studies, our ROC analysis provided cutoff time points for MMR (18-36 months) and CCyR (6-12 months) that were the best predictors for LOR or treatment failure, which can be an indirect evidence supporting the ELN recommendation. The patient who achieves short-term response rapidly will have a lower risk of losing response or failing after imatinib therapy in CML patients.