Introduction: Human leukocyte antigen (HLA) compatibility between patient and donor is a mainstay in allogeneic hematopoietic cell transplantation (HCT) to reduce the risk of potentially fatal graft-versus-host disease (GvHD). Traditionally, mismatches at HLA class I (HLA-A, HLA-B, HLA-C) and class II (HLA-DRB1) have been associated with significantly reduced overall survival (OS) and increased risks of non-relapse mortality (NRM) and acute (a)GvHD compared to fully matched donors (Lee et al. Blood 2007). This has resulted in a preference for 8/8 (10/10 in Europe) HLA-matched donors in unrelated HCT (Dehn et al. Blood 2019). More recently, interest in expanding the donor pool to patients lacking a fully matched donor (Shaw et al. JCO 2021 & TCT 2023) and advances in the management and prevention of GvHD, in particular with post-transplant cyclophosphamide (PTCy) (Bolaños-Meade et al. NEJM 2023), are increasing the use of mismatched (i.e. <10/10) unrelated donors. However, the role of HLA mismatches in current HCT practice, including new strategies for GvHD prophylaxis, remains unclear. Here, we evaluated the effects of mismatching at the major HLA loci in a large contemporary cohort of patients transplanted for hematological malignancies. Methods: We studied outcomes in 17,276 adult patients transplanted from unrelated donors between 2005-2020 reported to the EBMT Registry. 67.5% of the patients were treated for acute myeloid or lymphoblastic leukemia, or myelodysplastic or myeloproliferative neoplasms. 23.5% of the transplants were performed across one (9/10; n=3,561) or two (8/10; n=499) high-resolution HLA mismatches. Reduced-intensity conditioning was used in 56% of the patients, with 76% having in vivo T-cell depletion, and 91% receiving peripheral blood stem cells. Standard GvHD prophylaxis based on calcineurin inhibitors alone or in combination with other drugs was used in all patients. PTCy was used in 7% of the 10/10 (n=924) vs 15% (n=599) of the <10/10 transplants. Univariable and multivariable models were constructed to analyze the effect of the number, class, and locus of HLA mismatches compared to 10/10 transplants in the presence or absence of PTCy. Results: In the whole cohort, the presence of one or two mismatches at the five main HLA loci significantly reduced OS (45% [95% CI 40-50%] and 47% [95% CI 45-49%] in 8/10 and 9/10, respectively, vs 52% [95% CI 51-53%] in 10/10 at 60 months; p<0.001; Figure 1A). In multivariable analysis adjusted for the most relevant clinical and transplant-related variables including the use of PTCy, the presence of one (HR 1.24 [99% CI 1.15-1.34]; p<0.001) or two (HR 1.29 [99% CI 1.09-1.54]; p<0.001) HLA mismatches was confirmed to be associated with a higher risk of mortality when compared to 10/10 transplants. When the type of mismatch was taken into account, results showed that only HLA class I (HR 1.31 [99% CI 1.20-1.42]; p<0.001) and not HLA class II (HR 1.07 [99% CI 0.93-1.22]; p=0.23) single mismatches were significantly associated with worse survival. For HLA class I, these associations were stronger for HLA-A (HR 1.37 [99% CI 1.21-1.54]; p<0.001) and HLA-B (HR 1.44 [99% CI 1.23-1.69]; p<0.001) than for HLA-C (HR 1.16 [99% CI 1.01-1.33]; p=0.005). Similar associations were observed for other endpoints, including GvHD-free, relapse-free (GRFS), and relapse-free (RFS) survival, NRM, and aGVHD grades II-IV and III-IV. In contrast, no significant differences were observed for the risks of relapse or chronic (c)GvHD. The use of PTCy significantly reduced the risks of aGvHD and mortality compared to transplants performed with standard prophylaxis. However, the effects of HLA mismatches were similar regardless of the administration of PTCy (non-significant interaction; p=0.22), with a single mismatch conferring significantly increased risks of mortality both in the presence (HR 1.38 [99% CI 1.14-1.68]; p<0.001) and in the absence (HR 1.23 [99% CI 1.13-1.33]; p<0.001) of PTCy ( Figure 1B). Conclusion: In contemporary HCT, HLA disparity is associated with increased risks of mortality, mainly driven by HLA class I mismatches. These associations are present even under GvHD prophylaxis with PTCy, challenging the notion that GvHD prophylaxis can completely overcome the effects of histocompatibility, and highlighting the need to continue defining better tolerated mismatches to provide the best possible outcome for all patients.
Read full abstract