The treatment policy of chronic myeloid leukemia (CML), particularly with tyrosine kinase inhibitors, has been influenced by several recent studies that were well designed and rapidly performed, but their interpretation is of some concern because different end points and methodologies were used. To understand and compare the results of the previous and future studies and to translate their conclusion into clinical practice, there is a need for common definitions and methods for analyses of CML studies. A panel of experts was appointed by the European LeukemiaNet with the aim of developing a set of definitions and recommendations to be used in design, analyses, and reporting of phase 3 clinical trials in this disease.
IntroductionApplied statistics are important tools in medical evaluations. The relevance of statistical designs and statistical results in trials is based on concise definitions regarding diagnosis, management, and treatment strategies. The choice of adequate statistical tests depends on the parameters to be analyzed and on specific end points.After the initial descriptions of chronic myeloid leukemia (CML) more than 160 years ago, little progress was made in its treatment for more than a century. 1 Survival prolongation was first achieved with drugs, such as hydroxyurea. 1 Then, major improvements were obtained with allogeneic hematopoietic stem cell transplantation. First priority at that time was to analyze survival.The understanding of the pathogenesis of CML began with the discovery of the Philadelphia (Ph) chromosome followed later on by the recognition of its molecular counterpart, the BCR-ABL fusion gene. In CML, clinical trials performed during the past 2 to 3 decades have profoundly improved the outcome of patients with CML. 2 In Europe, the coordinators of national CML Study Groups (European Investigators on CML) joined forces to perform European trials and long-term observations, 3 to conduct common meta-analyses, 4,5 and to elaborate new prognostic scores. 6,7 With the advents of IFN-␣ and of tyrosine kinase inhibitors (TKIs), analyses of treatment response moved more into focus.Thus, the relationship between responses and survival became of particular interest. The cytogenetic outcome is usually considered as a reliable surrogate marker for survival in CML, 7,8 whereas the relationship between molecular response and survival is still under investigation. 9 Molecular response is probably more relevant to the issue of treatment discontinuation without relapse and cure. 10 On the other hand, because of the recent improvement of survival offered by the use of TKIs, other diseases and deaths from causes other than CML progression have become more frequent in CML patients. Analysis of survival may include these new parameters. Quality of life and compliance have also to be considered in the context of this chronic disease with an expected long duration of treatment.Consequently, statistical methods in addition to Kaplan-Meier (KM) 11 analyses and initial Cox regression models 12 should be a...