The CD3- 16+ 56+ NK cell count independently predicts autologous blood stem cell mobilization.

Abstract

Better predictive factors for autologous blood stem cell mobilization (BSCM) are needed. The purpose of this study was to determine if an independent association exists between lymphocyte or NK cell counts and BSCM. Data were analyzed on 141 consecutive patients aged 19-69 years (median 45) who received combined chemotherapy plus G-CSF for BSCM, and who had measurements of immune cells prior to BSCM. Of the 141 patients, 41% had breast cancer, 14% Hodgkin's disease, 34% non-Hodgkin's lymphoma, and 11% other diagnoses. BSCM involved dose-intensive cyclophosphamide, etoposide, cisplatin (DICEP) plus G-CSF 300 microg (<70 kg) or 480 microg (>70 kg) for 45% of patients, while the remaining 55% received other chemotherapy plus similar doses of G-CSF. Only a single apheresis was performed for 94% of patients. The following factors were analyzed for predictors of BSCM: age, gender, prior chemotherapy, prior radiotherapy, diagnosis, disease status, marrow involvement, mobilization regimen, Hb, WBC, platelet count, B cell, T cell, and NK cell counts. The peripheral blood CD34+ counts on the first day of apheresis (PBCD34) were 6-1783 x 10(6)/l (median 150). The PBCD34 count correlated strongly with the number of CD34+ cells collected/l blood apheresed and with the number of CD34+ cells collected/kg. By multivariate analysis using continuous variables, relapsed status (P = 0.0003), not using DICEP mobilization (P = 0.0001), female gender (P = 0.0057), low platelet count (P = 0.051), and low CD3- 16+ 56+ count (P = 0.0158) were associated with low PBCD34 counts. Using categorical variables, the only factors that independently predicted a PBCD34 count <150 x 10(6)/l were: >1 prior chemotherapy regimen (odds ratio = 5.12, P = 0.0003), not using DICEP mobilization (odds ratio = 4.94, P = 0.0001), and CD3- 16+ 56+ count <125 x 10(6)/l (odds ratio= 2.58, P = 0.0157). In conclusion, the CD3- 16+ 56+ count may be a useful additional predictor of BSCM and warrants further study.