Prognosis in chronic myeloid leukaemia: biology of the disease vs. treatment.

CML is best regarded as a benign neoplasm, which regularly transforms into a rapidly fatal malignant one. Survival is determined by the timing of disease transformation. The rate of transformation and death is low initially and increases gradually to a stable annual rate, reached during the third year after diagnosis. This averages 25% per year in large populations of Ph positive patients. Transformation appears to occur randomly; a patient's risk of transformation can be defined, but the time when it will occur cannot be predicted. There is no evidence that conventional antileukaemic therapy changes the risk of transformation; thus, survival is determined principally by the intrinsic biology of the disease. A number of features recorded at the time of diagnosis correlate significantly with survival and can serve as prognostic parameters. Multivariate regression and Cox model analyses can generate hazard ratio formulae which provide quantitative estimates of patients' risk. Several Cox models have been described by different groups, and it is possible that two or more models may describe survival equally well. A four-variable model described by the International CGL Prognosis Study Group in 1984 has recently been tested in a prospective study by the Italian Cooperative Study Group on Chronic Myeloid Leukaemia, and successfully classified patients into three groups with significantly different outcomes. This model produces excellent results in analyses of large patient populations, but is less satisfactory when applied to smaller series. It is likely that current prognostic models can be improved substantially to provide more accurate definition of patient risk over a broader range of hazard ratio values. Such models should prove useful in evaluating therapeutic schedules, selecting patients for investigational or hazardous treatment and making decisions regarding bone marrow transplantation.