Parameters of Predictive Value in Chronic Myeloid Leukemia-The Prognostic Impact of Histopathological Variables in a Multivariate Regression Analysis.

The prognostic significance of disease features at the time of diagnosis was examined in 113 patients (75 males/38 females, median age 49 years) with chronic myeloid leukemia (CML) using multivariate regression analysis. In the course of this study we tested the validity of the prognostic model described by Sokal and coworkers(3), the predictive value of histopathological variables, and determined the disease-specific loss in life expectancy. The median survival time was 35 months in our patients. By using Cox's proportional hazards model for covariate analysis of censored survival data(26)'(32) two prognostic models were derived. The first consisted of the following parameters-age, presence of pseudo-Gaucher cells in the marrow, spleen size and the product of myeloblasts and normoblasts in the peripheral blood. The second model included age, the presence of myelofibrosis, the number of megakaryocytes in the bone marrow and the liver size. Both models provided a risk status that showed an excellent predictive relationship to relative survival rates and life expectancy. Application of the Sokal model(3) to our subpopulation of CML patients failed to reveal a significant segregation of the intermediate group from the other risk groups, even when assessing the disease-specific loss in life expectancy. Our results suggest that an amendment of the generally accepted model for prognostic evaluation of survival in CML should be made by inclusion of histological variables and the determination of relative survival rates.