How to select optimal cord blood (CB) remains an important clinical question. We developed and validated an index of CB engraftment, the cord blood index (CBI), which uses three weighted variables representing cell doses and HLA mismatches. We modeled the neutrophil engraftment time with competing events by random survival forests for competing risks as a function of the predictors: total nucleated cells, CD34, colony-forming units for granulocytes/macrophages, and the number of HLA mismatches at the antigen and allele levels. The CBI defined three groups that had different neutrophil engraftment rates at day 30 (High, 83.7% [95% CI, 79.2-88.1%]; Intermediate, 77.0% [95% CI, 73.7-80.2%]; Low, 68.4% [95% CI, 63.6-73.2%]), platelet engraftment rates at day 60 (High, 70.4% [95% CI, 64.9-75.9%]; Intermediate, 62.3% [95% CI, 58.5-66.0%]; Low, 49.3% [95% CI, 44.2-54.5%]), and non-relapse mortality at day 100 (High, 14.1% [95% CI, 9.9-18.3%]; Intermediate, 16.4% [95% CI, 13.5-19.3%]; Low, 21.3% [95% CI, 17.1-25.5%]). This novel approach is clinically beneficial and can be adopted immediately because it uses easily obtained pre-freeze data of CB.
© 2021. The Author(s), under exclusive licence to Springer Nature Limited.