Purpose: There is no consensus on the best choice of an alternative donor (umbilical cord blood [UCB], haploidentical, one-antigen mismatched [7/8]-bone marrow [BM], or 7/8-peripheral blood [PB]) for hematopoietic cell transplantation (HCT) for patients lacking an HLA-matched related or unrelated donor.
Methods: We report composite end points of graft-versus-host disease (GVHD)-free relapse-free survival (GRFS) and chronic GVHD (cGVHD)-free relapse-free survival (CRFS) in 2,198 patients who underwent UCB (n = 838), haploidentical (n = 159), 7/8-BM (n = 241), or 7/8-PB (n = 960) HCT. All groups were divided by myeloablative conditioning (MAC) intensity or reduced intensity conditioning (RIC), except haploidentical group in which most received RIC. To account for multiple testing, P < .0071 in multivariable analysis and P < .00025 in direct pairwise comparisons were considered statistically significant.
Results: In multivariable analysis, haploidentical group had the best GRFS, CRFS, and overall survival (OS). In the direct pairwise comparison of other groups, among those who received MAC, there was no difference in GRFS or CRFS among UCB, 7/8-BM, and 7/8-PB with serotherapy (alemtuzumab or antithymocyte globulin) groups. In contrast, the 7/8-PB without serotherapy group had significantly inferior GRFS, higher cGVHD, and a trend toward worse CRFS (hazard ratio [HR], 1.38; 95% CI, 1.13 to 1.69; P = .002) than the 7/8-BM group and higher cGVHD and trend toward inferior CRFS (HR, 1.36; 95% CI, 1.14 to 1.63; P = .0006) than the UCB group. Among patients with RIC, all groups had significantly inferior GRFS and CRFS compared with the haploidentical group.
Conclusion: Recognizing the limitations of a registry retrospective analysis and the possibility of center selection bias in choosing donors, our data support the use of UCB, 7/8-BM, or 7/8-PB (with serotherapy) grafts for patients undergoing MAC HCT and haploidentical grafts for patients undergoing RIC HCT. The haploidentical group had the best GRFS, CRFS, and OS of all groups.