RESUMEN
Successful allogeneic hematopoietic stem cell transplantation (alloHSCT) relies significantly on adequate allograft cell composition to achieve sustained engraftment, and a minimum of 2 × 108 total nucleated cells (TNCs) per kilogram of recipient body weight has been identified as the prerequisite cell dose for successful engraftment of marrow-derived products. To meet this minimum requirement, marrow harvest volumes are estimated based on anticipated TNC concentrations of 18.3 × 106/mL. However, there is considerable variability in marrow TNC concentrations. Thus, an algorithm that incorporates baseline donor characteristics to predict TNC concentrations could optimize outcomes for both donors and recipients. For this study, donor baseline characteristics and corresponding unstimulated marrow products harvested between 2004 and 2017 at a single large-volume donor center were collected. Multivariable analysis was used to identify significant predictors of TNC concentration. Two models-ordinary least squares (OLS) and least absolute shrinkage and selection operator (LASSO) regression-were compared for their fitness to the data and their utility in predicting TNCs. Donors with higher body mass index, younger age, male sex, white race/ethnicity, smaller harvest volumes, lower preharvest hematocrit, higher preharvest platelet count, and higher preharvest WBC count predicted significantly higher TNC concentrations in marrow products. When comparing predictive models that incorporate these characteristics, the cross-validated LASSO and bootstrapped OLS provided the best fit. We now supply these formulas to be validated in other datasets before clinical use. TNC concentration in marrow products can be predicted using donor characteristics, most of which are readily available during the donor clinical assessment. The ability to predict marrow allograft TNC concentrations can optimize collection volumes during a harvest.