HLA Match and Mismatch Direction in Cord Blood (CB) Transplantation: Impact On Outcome and Implications for Cord Blood Unit Selection.

Abstract 2297

Poster Board II-274

Apparent immunological tolerance allows for transplantation of cord blood (CB) hematopoietic stem cells that are mismatched for human leukocyte antigens (HLA) with the recipient. HLA match, however, affects CB transplant outcome. To explore the possibility that the direction of HLA mismatch (MM) also has an impact on outcomes, we evaluated patients (pts) transplanted in the US with single CB units provided by the New York Blood Center, with 0-2 HLA MM, during the period 1993-2006. We hypothesized that CB grafts with MM in the rejection direction only (pt homozygous at the mismatched locus) would have higher rates of graft failure and, consequently, relatively poor post-transplant survival. Conversely, CB grafts with no rejection MM (i.e., GVHD only MM; CB unit homozygous at the mismatched locus) would have better engraftment and survival. Match level was evaluated at low-intermediate resolution for HLA-A and -B and at high resolution level for -DRB1. Of the 1,207 pts with outcome data (94% of eligible pts), 72% had hematological malignancies, 8% had bone marrow failure and the rest had various genetic diseases. Most pts (92%) had myeloablative conditioning, 26% were ≥16 years of age and 48% had non-Caucasian ancestry. CB grafts were HLA matched in 73 donor/patient pairs. Ninety-eight pairs (8% of total) had only unidirectional MM: 58 in the GVHD direction only (51 with 1 HLA MM; 7 with two) and 40 in the rejection direction only (30 with 1 HLA MM; 10 with two). There were 360 donor/patient pairs with 1 bidirectional MM, 524 with 2 bidirectional MM, 137 with 2 mismatches that were a mixture of bi- and uni-directional, and 5 with both types of unidirectional MM (different loci). In multivariate analyses, compared to CB grafts having 1 bidirectional MM, pts with matched grafts and those with GVHD only MM had faster ANC engraftment (RR=1.7, p=0.037 and RR=1.6, p=0.006, respectively) and improved TRM (RR=0.5, p=0.041 and RR=0.6, p=0.046, respectively), overall mortality (RR=0.6, p=0.041 and RR=0.6, p=0.016, respectively) and disease-free survival (RR=0.6, p=0.055 and RR=0.6, p=0.009, respectively). These grafts had no association with relapse risk in pts with hematological malignancies. In contrast, CB grafts with rejection only MM had a lower engraftment rate (RR=0.7, p=0.086) and a higher relapse rate. These associations were most apparent in patients with hematological malignancies (Table). These relationships were independent of other predictors in the multivariate models and of matching for the donor's non-inherited maternal antigens (NIMA), another factor we have recently found to affect outcome in MM CB transplants. Further, MM direction had no significant association with acute or chronic GVHD. No associations with outcome were found in grafts with a mix of both bi-directional and uni-directional MM or two unidirectional MM.

This study demonstrates that direction of HLA MM affects CB transplant outcome. CB units with GVHD only MM had outcomes similar to those of fully matched (6/6) grafts and, according to published reports in children with leukemia, possibly superior to 8/8 allele level bone marrow grafts from unrelated donors. CB units with rejection only MM were associated with poor engraftment and, in pts with hematological malignancies, with increased relapse risk. Assignment of HLA MM direction should be included in CB unit selection algorithms. For pts lacking fully matched grafts, selection of CB units with GVHD only MM will increase the probability of an optimal outcome.