Analysis of 608 Umbilical Cord Blood (UCB) Transplants: HLA-Match Is a Critical Determinant of Transplant-Related Mortality (TRM) in the Post-Engraftment Period Even in the Absence of Acute Graft-vs-Host Disease (aGVHD).

With the recognition of the critical importance of UCB cell dose upon hematopoietic recovery, neutrophil recovery after UCB transplantation (UCBT) has improved. However, TRM in the early post-engraftment period (days 30–180) remains a barrier to transplant success. Therefore, how graft characteristics impact upon the risk for TRM during this period is important but has not been well characterized. Therefore, we analyzed the impact of cell dose and HLA-match upon TRM between days 30–180 in recipients of single unit 3–6/6 HLA-A,B antigen and DRB1-allele matched UCBT who were transplanted for leukemia or myelodysplasia with units from the NYBC’s National Cord Blood Program and achieved sustained donor engraftment. Patients dying from days 0 to 29 or those with primary or secondary graft failure were excluded, and patients were censored at the time of relapse. 608 patients fulfilled study criteria and had a median age of 9.0 years (range 0.4–58 years). The cumulative incidence of TRM was 36% (95%CI: 32–39). Using 2.5–4.9 x 10⁷ TNC/kg as the reference (n = 229), the impact of cell dose was analyzed. While patients (n = 141) receiving a lower dose of 0.7–2.4 had a relative risk (RR) of TRM of 1.5 (p = 0.009) during this period, TRM in patients receiving higher doses were similar to the reference [5.0–9.9 (n = 160) RR 0.9, p = 0.4; and ≥ 10.0 (n = 78) RR 0.8, p = 0.4)]. However, HLA-match had an impact upon TRM at all levels of match [reference: 5/6 recipients (n = 194); 6/6 (n = 32) RR 0.3, p = 0.063; 4/6 (n = 342) RR 1.7, p = 0.001; and 3/6 (n = 40) RR 2.2, p = 0.003]. To exclude the possibility that disease stage accounted for this finding, a multivariate Cox regression analysis was performed including TNC dose, HLA-match and disease stage, and showed that low dose and HLA-mismatch remained the significant predictors of TRM. Further, when patients were divided into those without significant aGVHD (only grade 0 or 1) (n = 292) vs patients with grade 2–4 aGVHD (n = 301), the effect of HLA-match on TRM was most pronounced in those without significant aGVHD [reference: 5/6 recipients (n = 102); 6/6 (n = 25) RR 0.0, p = NS; 4/6 (n = 151) RR 2.4, p = 0.004; 3/6 (n = 14) RR 2.8, p = 0.050]. Of the transplant-related deaths (n = 60) from days 30–180 in this “without significant aGVHD” group, infection was reported to be a major cause of death in 36 (60%). In summary, above the threshold of 2.5 x 10⁷ TNC/kg, cell dose has no demonstrable effect on the risk for day 180 TRM in the post-engraftment period, whereas the adverse impact of HLA-mismatch is both striking and retained even in the absence of aGVHD. We postulate that the adverse effect of HLA-mismatch is mediated not only by induction of aGVHD, but also by an effect on immune reconstitution as manifested by death from infection. This data has significant implications for graft selection implying that both dose and HLA-match should be considered, and investigation of how to “trade-off” each of these factors should be a priority. Further, it argues that recipients of HLA-mismatched grafts will require aggressive supportive care and strategies to augment immune reconstitution. Finally, improved outcome in UCBT may be dependent upon the ability to obtain units that are both of sufficient size and match which will require an increase in the global UCB inventory.