Determination of the characteristics of accelerated disease in chronic myelogenous leukemia (CML) helps in individual prognostication, and in the introduction and analysis of investigative approaches based on risk-benefit ratios. The outcome of 357 patients with Philadelphia chromosome-positive CML was analysed from the time of development of suspected features of accelerated disease. Median survivals shorter than 18 months were associated with the appearance of any of the following: cytogenetic clonal evolution; extramedullary disease; peripheral blasts greater than or equal to 15%, peripheral blasts and promyelocytes greater than or equal to 30%, or peripheral basophils greater than or equal to 20%; platelet count less than 1.0 X 10(5)/microliters; marrow blasts greater than or equal to 15%, marrow blasts and promyelocytes greater than or equal to 30%, or marrow basophils greater than or equal to 20%. Relative hazard ratios, or risk of death per unit time, were calculated based on the relative survivals of patients who did or did not develop the particular feature of accelerated disease, after accounting for the time to development of the characteristic. Further analysis identified five features that have additive independent prognostic importance: cytogenetic clonal evolution; peripheral blasts greater than or equal to 15%; peripheral basophils greater than or equal to 20%; peripheral blasts and promyelocytes greater than or equal to 30%; and thrombocytopenia. By providing an objective estimate of prognosis in accelerated disease, the model identifies patients in need of different therapeutic interventions before the development of blastic crisis.

To study the influence of chronologic age on treatment outcome in patients with advanced, diffuse large-cell (histiocytic) lymphoma (DHL), we reviewed the results of two recent Southwest Oncology Group (SWOG) clinical trials. From 1974 to 1982, members entered 307 eligible patients treated with cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) with or without bleomycin, and CHOP with or without immunotherapy using BCG, levamisole, or both. Complete response (CR) rates declined progressively with advancing age: 65% in those under 40, 60% in the 40 to 54 age group, 55% in the 55 to 64 age group, and 37% in those 65 and older (P = .001). Likewise, survival decreased significantly in older patients: medians were 101 +, 52, 34, and 16 months, respectively (P less than .001). Treatment guidelines included an initial dose reduction of 50% for patients aged 65 or older and for younger patients with bone marrow compromise. Despite protocol specifications, 23 of 81 patients aged 65 or older received initial full-dose therapy. When these patients were compared with younger patients on whom full-dose chemotherapy was started, survival curves, but not CR rates, were still significantly different. There were no significant differences in duration of CR or frequency of treatment complications. These data suggest that older age is associated with a worse prognosis in advanced DHL. Moreover, the initial dose reduction for patients aged 65 or older may have contributed to their inferior outcomes.

No abstract

As a means of assessing the importance of variation in treatment effect among patient subsets, we derived posterior distributions for subset-specific treatment effects. The effects are represented by combinations of terms for treatment and treatment-by-covariate interaction effects in familiar regression models. Exchange-ability among the interactions is a key assumption; thus, the results are of interest primarily in the context of examining a collection of subsets with no definite a priori distinction relative to treatment effect. Exchangeability leads to a shrinking of the posterior distributions of the interaction terms toward the natural origin of 0, offsetting the tendency of the estimated effects to disperse. The method is applied to parameter estimates from a proportional hazards regression analysis of survival data from a clinical trial, invoking the approximate multivariate normal distribution of the estimates. No subjective prior distributions are required. Vague priors are used for all of the regression coefficients except the treatment-by-covariate interactions, which are assumed to follow a normal distribution.