Abstract:The utility of the MLC assay as a test of HLA‐D region matching and predictor of graft‐versus‐host disease (GvHD) was evaluated in 435 patients receiving marrow grafts from unrelated donors. Donors and recipients were phenotyped for HLA‐A, B and DR antigens by serology, tested in MLC, and retrospectively genotyped for DRB1, B3, B4, B5, DQB1 and DPB1 alleles by PCR/SSOP. Of the 244 HLA‐A, B, DR‐identical donor‐recipient pairs with evaluable MLC and DRB1 typing results available, 208 were matched for HLA‐A, B and DRB1, while 36 were matched for HLA‐A and B and mismatched for a DRB1 allele. Donor anti‐recipient relative responses (RR) in MLC, corresponding to the GvHD vector in marrow transplantation, ranged from 7.2 to 100%, with a median of 4.0%. A comparison of reactivity in MLC between pairs matched versus mismatched for DRB1 alleles showed a significant overlap in the distribution of RRs. Using optimally‐defined RR cutoffs of 4 and 16%, no correlation between MLC results and risk of developing clinically significant grades III‐IV GvHD (p=0.6 and 0.5, respectively) was found when the contribution of DRB1 mismatch was accounted for. Matching for DRB1 alleles, in contrast, was a better predictor of clinically significant GvHD, with DRB1‐matched transplant recipients less likely to develop grades III‐IV GvHD than DRB1‐mismatched recipients (p=0.14). Among the 208 patients and donors matched for DRB1 alleles, the MLC, although reactive (RR>4.0%) in 45% of cases, did not predict GvHD. Overall, these results underscore the limitations in using the MLC to predict DRB1 matching or risk of clinically significant GvHD among patients receiving unrelated marrow grafts. The availability of DRB1 allele matching by sequence‐specific oligonucleotide probes (SSOP) or by direct sequencing provides a method for donor matching that is rapid, precise and superior to the MLC for predict