RESUMO
It is unknown whether clinical characteristics can successfully predict which multiple myeloma (MM) patients would be poor mobilizers with growth factor (GF) alone so they can be assigned to mobilization with chemotherapy + GF or GF + plerixafor. MM patients (N = 477) who underwent autologous mobilization with GF were retrospectively reviewed and assigned into training and validation cohorts. In multiple regression analysis, age, platelet count at time of mobilization, type of GF utilized, and extent of exposure to lenalidomide independently correlated with peripheral blood (PB)-CD34+ and were integrated in a predicting score (PS) for poor mobilizers, defined as PB-CD34+ < 20/mm(3) 4 days after initiation of GF. There was no correlation between institution, gender, time between diagnosis, and mobilization or plasma cells in the bone marrow at time of mobilization and PBCD34+. The PS cut-off found in the training cohort to have 90% sensitivity for prediction of poor mobilizers performed with 89.7% sensitivity but only 34.8% specificity in the validation cohort. Conversely, the PS cut-off developed to have 90% specificity performed with 86.9% specificity but only 37% sensitivity. We conclude that clinical characteristics identifiable before initiation of mobilization should not be used to stratify MM patients for different mobilization strategies.
