Chronic myelogenous leukemia: a multivariate analysis of the associations of patient characteristics and therapy with survival

Abstract

Abstract The prognostic importance of patient pretreatment clinical and laboratory features was investigated in a group of 303 patients with Philadelphia chromosome-positive benign-phase chronic myelogenous leukemia. Intensive chemotherapy was given to 97 patients, and 78 underwent an early elective splenectomy. The overall median survival time, dated from hospital admission, was 39 months. Patient characteristics associated with shortened survival were age 60 years or older, black race, the presence of hepatomegaly, splenomegaly, symptoms, weight loss, and poor performance status. Adverse blood and bone marrow parameters were anemia, thrombocytosis or thrombocytopenia, a high proportion of peripheral blasts plus promyelocytes or of basophils, a high proportion of marrow blasts or basophils, decreased marrow megakaryocytes, and cytogenetic abnormalities in addition to the Philadelphia chromosome. Several of these factors were interrelated. A multivariate regression analysis demonstrated that the combination blood basophilia, race, additional cytogenetic abnormalities, age and marrow basophilia had the strongest predictive relationship to survival time. This resulted in a model segregating patients into low-, intermediate-, and high-risk groups, with median survivals of 53, 39, and 25 months, respectively. Another model was derived that did not include the marrow features and identified splenomegaly and platelet counts as adding to the prognosis prediction by blood basophilia, race, and age. Evaluation of the effect of therapy, after adjusting for differences in prognostic characteristics, showed that intensive chemotherapy was associated with survival prolongation among patients at intermediate and high risk of death. We conclude that a combination of pretreatment factors identifies different risk subcategories in patients with chronic myelogenous leukemia and is helpful in assessing overall prognosis and treatment effect.