Haplo-PtCy: adjusting the HLA barrier

In this issue of Blood, Fuchs et al¹ unraveled the role of patient-donor HLA matching in the outcome of haploidentical hematopoietic cell transplantation (HCT) as treatment for acute leukemia (AL) or myelodysplastic syndrome (MDS).

Since its description in the 1990s, successful HCT from related donors mismatched for an entire HLA haplotype (haplo-HCT) has challenged the dogma of complete patient-donor HLA matching as a prerequisite for engraftment and prevention of lethal graft-versus-host disease (GvHD).² The introduction of posttransplant cyclophosphamide (PtCy)-based GvHD prophylaxis inititated the era of T-cell replete haplo-HCT (haplo-PtCy). This resulted in the continuously increasing clinical use this treatment modality has witnessed over the last decade, and further fueled discussions about the relevance of the HLA barrier in HCT. This question is of considerable practical importance because many patients have several haploidentical family donors. Based on currently available evidence, donor-selection guidelines for haplo-HCT consider HLA only by recommending avoidance of alloantibodies in the patient against donor-specific HLA allotypes.³ 

By definition, patients share 1 entire HLA haplotype (ie, 1 allele each of the 6 loci HLA-A, HLA-B, HLA-C, HLA-DRB1, HLA-DQB1, HLA-DPB1) with their haploidentical family donors, whereas the other haplotype is unshared (see figure). However, the unshared haplotypes of patient and donor can, by chance, carry the same allele at ≥1 of the 6 HLA loci, resulting in a minimum of 0 and a maximum of 6 mismatched HLA alleles. It is unknown whether the number and/or the locus of these accidentally shared HLA alleles has any impact on HCT outcome. Moreover, qualitative differences in HLA mismatches, such as the B leader⁴ or DPB1 T-cell epitope (TCE) nonpermissiveness,⁵ relevant in unrelated HCT, have not been comprehensively studied in haplo-HCT.

Fuchs et al¹ systematically addressed these questions in a cohort of 1434 patients, who were treated for AL or MDS by haplo-PtCy, reported to the Center for International Blood and Marrow Transplant Research. They found that the number of HLA mismatches on the unshared haplotype did not have any significant impact on outcome. In contrast, HLA-DRB1 graft-versus-host (GvH) mismatches were associated with lower relapse rates and, when combined with a match at HLA-DQB1, better disease-free survival (DFS). Moreover, B leader matches and DPB1 TCE nonpermissive GvH (DPB1-NP) mismatches were associated with improved DFS and overall survival (see figure). This reflected lower transplant-related mortality (TRM) for the B leader. For DPB1-NP, statistical power was more limited than for the other models, because DPB1 typing was available for less than half of the cohort. It is tempting to speculate that the survival advantage for DPB1-NP might result from lower relapse and lower TRM, as evidenced by the lower hazard ratios for these 2 endpoints, although the differences were not statistically significant.

The findings from Fuchs et al¹ elegantly demonstrate how the HLA barrier has not been erased but needs to be adjusted in the setting of haplo-PtCy. Locus-specific effects for HLA class II and biological matching models, rather than the number of HLA mismatches, turned out to be the key to pinpointing the impact of patient-donor HLA matching status on outcome. Interestingly, HLA class II DRB1 (and possibly DPB1-NP) mismatches were associated with lower relapse, but not higher acute GvHD, in this setting. Targeted genomic and/or global transcriptional downregulation of HLA class II is a mechanism of immune evasion in leukemia relapse after HCT, hinting at its importance in graft-versus-leukemia (GvL).^6-8 The observation that the benefits of defined HLA class II mismatches on relapse were not negated by increased GvHD might be explained by the immunological effects of PtCy, which selectively eliminates activated (ie, in the early transplant phase alloreactive) effector T cells while sparing regulatory T cells.² This mode of action might classify PtCy as a game changer for T-cell alloreactivity, which raises the bar of HLA disparity needed for severe toxic GvHD but not for GvL. Of note, permissive, rather than nonpermissive, DPB1 TCE mismatches are associated with improved survival in unrelated HCT with calcineurin inhibitor-based immune prophylaxis.⁵ In the context of PtCy, stronger T-cell alloreactivity against DPB1 TCE nonpermissive mismatches, mechanistically based on the degree of immunopeptidome divergence between HLA-DP allotypes determining the number and diversity of alloreactive T cells,⁹ might be required for efficient GvL without excessive GvHD. Further research is needed to understand whether these effects of PtCy also apply to its use in unrelated HCT, in which relevant clinical trials are under way.

Interestingly, the survival advantage for B leader matches was associated with lower TRM but not with reduced relapse. In B leader matched haplo-HCT, the HLA-B alleles on the unshared haplotype of patient and donor carry the same amino acid at position −21 (ie, threonine [T] or methionine [M]) (see figure). M leaders, but not T leaders, are able to supply peptides binding to HLA-E, thereby impacting the function of CD94/NKG2A-bearing natural killer cells¹⁰ and, possibly, HLA-E–restricted T cells. It remains to be elucidated if and how these immune cells are involved in the beneficial effects of B-leader matching.

Based on their findings, Fuchs et al¹ developed a webtool to predict DFS for a given patient with defined relevant clinical characteristics, after HCT from haploidentical donors, according to their HLA–B leader, DRB1/DQB1, and DPB1 TCE matching status (http://haplodonorselector.b12x.org/v1.0/). This freely accessible tool will prove useful in guiding clinicians in the frequent need to select the optimal donor for haplo-HCT. Of note, the tool does not take into account the patient’s alloantibody status, which was not investigated here because of the unavailability of the relevant data but has been identified as critical in numerous other studies³; hence, it should be integrated into the decision making. It will also be important to update the webtool once evidence on additional relevant variables, such as killer immunoglobulin-like receptors, emerges.

In summary, Fuchs et al provide long-awaited comprehensive evidence on the role of HLA matching in haplo-HCT. Apart from their practical relevance, these data remind us that qualitative, rather than quantitative, effects are the key to understanding the role of patient-donor HLA disparity in HCT.