BACKGROUND: Cord blood is an important hematopoietic stem cell source for treating over 80 FDA-approved diseases, driving demand for HSC-containing-umbilical cord blood (UCB) unit cryopreservation for future use. Total nucleated cell (TNC) and CD34+ cell amounts are key determinants of transplant success. Predicting processing outcomes is clinically valuable, particularly for public stem cell banks where storing high-quality UCB units is a priority. This study aimed to establish selection models for UCB units with high isolation yields of these critical cellular determinants. METHODS: We first performed univariate analysis and multiple linear regression on 3338 UCB units processed at MekoStem Stem Cell Bank (Vietnam) from 2019 to 2023 to explore correlations between potential variables (maternal age, mode of delivery, baby blood type, birth weight, gestational age, infant sex, blood volume, procurement time, and time to processing after collection) and TNC and CD34+ cell counts. Subsequently, Bayesian Model Averaging and exhaustive search were applied to develop predictive models. The developed models were then validated using a dataset of 660 UCB units processed in 2024. RESULTS: Results identified an optimal selection model incorporating birth weight, delivery mode, blood volume, and gestational age for predicting high isolation yields of both TNC and CD34+ cells. Reference values for these variables were birth weight >3200 g, vaginal delivery mode, blood volume >79.22 mL, and gestational age >39 weeks or ≤39 weeks for TNC or CD34+ cell models, respectively. CONCLUSIONS: The models proposed in this study demonstrated robust predictive performance in external validation and can be applied to any type of stem cell bank.
Pham et al. (Fri,) studied this question.