A new prognostic system for multiple myeloma based on easily available parameters.

The prognostic significance of different presenting features in 180 patients with multiple myeloma (MM) from a single institution was analysed. Out of eight variables isolated from the univariate analysis only two (blood urea and serum albumin), were significant in the multivariate model. Derived from these two simple variables, the relative risk of each patient was calculated, and subsequently two subpopulations of patients could be recognized. The first group included patients with a very active myeloma and a high risk of death soon after diagnosis, their median survival being of only 11.6 months, and the second one comprised patients with low risk of death during the first year and a median survival of 28 months. A hazard function derived from two-thirds of the patient population (training group) was successfully validated in the remaining subset of patients (test group). Finally, the three major available myeloma staging systems (Durie & Salmon's, Merlini et al's, and the one proposed by the British Medical Research Council) were tested in the present series, and only the latter one showed prognostic validity.