Despite more than 11 million typed volunteers in hematopoietic donor registries worldwide (www.bmdw.org), many patients do not find a fully HLA matched donor. The questions become therefore which donor is the closest match, how much would the detected disparity impact patient outcomes and would a transplant from the best partially matched donor still be preferable compared to non-transplant therapies.
In this month’s case, respondents favored a donor mismatched for one HLA-DQ ‘antigen’ (i.e.: DQB1*0101 vs. 0301) to donors mismatched at one HLA-A ‘allele’ (i.e. A*0101 vs. *0102) or one C ‘allele’ (i.e., C*0701 vs. *0702), and the top choice coincides with my preference as well. Reports of unrelated donor marrow transplantation from the U.S. and the Japan Marrow Donor Program including more than 5,000 patients failed to detect a significant association of an isolated HLA-DQB1 mismatch with any patient outcomes1,2. Thus, this donor would provide similar survival to a fully HLA-A, B, C, DRB1, DQB1 matched unrelated donor.
Mismatching for HLA-A was associated with increased mortality in both the US and Japanese registry studies, while one isolated mismatch for HLA-B or C less significant in the US study1, and HLA-C was not significant in the Japanese study2, thus I would have ranked Donor 2 better than Donor 1. The fact that the mismatch at HLA-A is for an allele and not an antigen does not mitigate the risk, as in the US analysis an A allele mismatch was associated with a mortality relative risk of 1.5 (95% C.I. 1.2-1.9)1. Since Donor 2 is mismatched for one HLA-C allele, and the relative risk of death for such a mismatch was 1.0 (95% C.I. 0.8-1.3), I would have expected this donor to have garnered as many votes as Donor 3 that was mismatched at HLA-DQ1. Obviously not all disparities are equally functional, but there are only early reports about potential criteria for identifying permissible mismatches by polymorphisms away from the sites of peptide anchor residues3. The KIR-binding HLA-C allotypes are usually identical within allelic disparities, but if the mismatch were for an HLA-C antigen and for a KIR-ligand allotype, consideration should have been given to enrolling the patient on a trial to test the potential benefits of donor NK alloreactivity with a T-depleted graft, based on data from mismatched related transplants4.
The literature does not provide definitive evidence from prospective controlled studies that a closely matched unrelated donor transplant improves survival of a patient with AML in second remission when compared to an autologous transplant or non-transplant therapies, but a case series and consensus guidelines of an expert panel support this recommendation especially with high-risk cytogenetics, based on retrospective studies5,6. I will not address the issue of selecting cord blood units or a related donor mismatched for a whole HLA haplotype, since a suitably matched unrelated donor is already in hand.
Donor age is often ignored in the selection process, but data from large transplant data sets found a significant albeit small (~3% per decade of age) association of younger donor age with less mortality7, another factor that did not weigh negatively in the selection of Donor 1 as much as I would have expected.
We often try to select a CMV-seronegative donor for a seronegative patient, and this might be one reason why Donor 1 scored relatively high in the poll, but such a donor does not improve survival compared to a CMV-seropositive donor7. We are similarly concerned about selecting female donors for male patients, because it increases the incidence of acute and chronic GVHD, the duration of chronic GVHD, and the non-relapse mortality, but the use of a female donor is also associated with less post-transplant relapse, and therefore survival of male patients with malignancy is in general not affected by the donor gender7. Apparently, these morbidity concerns did not influence respondents' votes since Donor 1, a female, was given as much preference as Donor 2, a male. Delayed red cell recovery occurs after donor major ABO disparity but has not been reported with Rh major disparities. Delayed transfusion reactions with intravascular hemolysis can occur with transplants from donor type O into recipient A or B, but are rare and occur almost exclusively in patients treated with a calcineurin inhibitor alone. Therefore, donor ABO would not be a helpful criterion for donor selection in this case, if the planned GVHD prophylaxis is a drug combination.
Should we choose mobilized blood or marrow stem cells from Donor 3? The literature has provided relatively good guidance with an individual patient meta-analysis of 9 randomized trials in sibling transplantation, but retrospective series show little difference in relapse and survival between patients treated with one of the two sources from an unrelated donor8. The increased risk of acute and chronic GVHD should deter the use of mobilized blood, but the high risk of relapse in a patient with AML with high risk cytogenetics in second remission should favor mobilized blood. To help answer this question, patients should be enrolled on the ongoing BMT-CTN 0201 trial that randomizes 1:1 between the two sources (https://web.emmes.com/study/bmt).
Address correspondence to claudio.anasetti@moffitt.org.