PROGNOSTIC FACTORS IN MULTIPLE-MYELOMA - SELECTION USING COXS PROPORTIONAL HAZARD MODEL

The pretreatment characteristics of 210 patients with multiple myeloma , observed between 1980 and 1994, were evaluated as potential prognost ic factors for survival. Multivariate analysis according to Cox's prop ortional hazard model identified in the 160 dead patients with myeloma , among 26 different single prognostic variables, the following factor s in order of importance: beta(2)-microglobulin; bone marrow plasma ce ll percentage, hemoglobinemia, degree of lyric bone lesions, serum cre atinine, and serum albumin. By analysis of these variables a prognosti c index (PI), that considers the regression coefficients derived by Co x's model of all significant factors, was obtained. Using this it was possible to separate the whole patient group into three stages: stage I (PI < 1.485,67 patients), stage II (PI: 1.485-2.090, 76 patients), a nd stage III (PI > 2.090, 67 patients), with a median survivals of 68, 36 and 13 months (P < 0.0001), respectively. Also the responses to th erapy (P < 0.0001) and the survival curves (P < 0.00001) presented sig nificant differences among the three subgroups. Knowledge of these fac tors could be of value in predicting prognosis and in planning therapy in patients with multiple myeloma.