Targeted GVL through HLA mismatch in double cord blood transplant Open Access

In this issue of Blood Advances, Wynn et al¹ demonstrate that, in the setting of double cord blood transplant (DCBT), shared human leukocyte antigen (HLA) mismatches between the patient and the losing cord blood unit (LU) when mismatched with the winning unit (WU), may prime a targeted immune response from the WU that significantly decreases relapse risk without increasing graft-versus-host disease (GVHD), or treatment related mortality (TRM). Furthermore, they identify class 1 loci, and specifically HLA-A, as key mediators of this effect, whereas class 2 loci were more closely associated with GVHD and TRM risk.

This study builds on earlier observations that DCBT may be associated with lower relapse risk compared to single-unit cord blood transplant (CBT), potentially due to immune priming through graft-vs-graft interactions. Verneris et al² reported a significantly lower incidence of relapse in recipients of 2 partially HLA-matched umbilical cord blood (UCB) units vs a single unit, especially among patients who underwent transplantation in the first or second complete remission (16% vs 31%, P = .03). Brunstein et al³ later found that a higher degree of allele-level mismatch, particularly when present in the predominant engrafting unit, was associated with reduced relapse without worsening GVHD or nonrelapse mortality. Wynn et al further advance our understanding of these immune interactions by demonstrating that it is not simply the degree of mismatch, but rather the immunologic configuration of shared mismatches between the patient and LU that primes a beneficial response from the WU. These findings suggest that the immunologic rejection of the LU is not merely a collateral event, but may represent an opportunity to harness targeted graft-versus-leukemia (GVL) effects without increasing GVHD risk, particularly when mismatch effects can be targeted to specific HLA loci.

Wynn et al observed that HLA-A mismatches shared between the patient and LU, but not class 2 mismatches, were significantly associated with reduced relapse risk, reinforcing the fundamental role of class 1–restricted, CD8⁺ T-cell–mediated GVL. Gutman et al⁴ previously demonstrated a graft-vs-graft CD8⁺ T-cell–mediated immune response was present in DCBT transplant recipients and responsible for rejection of the LU by the WU. The findings of Wynn et al suggest that this response may be extended to recipient leukemia cells when they share class 1 molecules with the LU. Although Wynn et al did not observe a similar correlation between relapse and class 2 mismatches, others, including Lamers et al⁵ have demonstrated an HLA class 2–specific CD4⁺ T-cell–mediated response in rejection of the LU by the WU. Moreover, these HLA class 2 specific T cells recognized primary leukemic cells when the class 2 mismatch was shared between the LU and the patient. These findings would suggest both class 1 restricted CD8⁺ T cells and class 2 restricted CD4⁺ T cells are playing a role in GVL. This is supported by the finding that loss or downregulation of HLA expression is a major mechanism of relapse, particularly in the haploidentical transplant setting. Vago et al⁶ report that downregulation of HLA class 2 molecules at relapse occurs across transplant platforms and impairs CD4⁺-mediated GVL. The absence of a class 2 effect in the Wynn et al analysis, therefore, may reflect cohort limitations, more than biological absence underscoring the need for further investigation of how different HLA loci contribute to GVL.

Wynn et al provide compelling evidence that immunologic interactions between HLA-A mismatched WU and shared LU/patient may be responsible for reduced relapse incidence. Although the clinical application of these findings remains limited by our current inability to predict which unit will engraft in DCBTs, the concept of targeted class 1 or class 2 HLA-mismatch supports a role for development of rational immune engineering of the cord blood transplant platform. Recent data from Borrill et al⁷ demonstrate that nonengrafting, highly mismatched pooled granulocyte products, presumably with significant HLA diversity, when infused along with an unmanipulated cord blood unit, lead to robust CD8⁺ T-cell expansion, cytokine activation, and durable remissions in children with refractory leukemia. Similarly, in our ongoing trial,⁸ a T-cell–depleted, ex vivo expanded, pooled cord product given after an unmanipulated unit has been associated with early expansion of engrafting donor T cells, which we hypothesize may similarly reflect a primed immune response and may contribute to improved leukemia-free survival. These examples support further development of methodologies in which intentional mismatch may prime an immune response to augment GVL without additional toxicity.

Together, these findings highlight the potential to harness immunologic mismatch as a therapeutic tool. Wynn et al provide compelling evidence that targeted mismatch at specific class 1 loci can stimulate a GVL response without excess toxicity, reinforcing the need for continued investigation into strategies that intentionally direct immune responses to optimize transplant outcomes. The findings of Wynn et al reinforce the importance of immunologic interactions shaping transplant outcomes, and support further development of methodologies to precisely stimulate these responses. Studies on the immune microenvironment, such as innate immune cells, antigen-presenting cell dynamics, and the cytokine milieu are needed to better understand how these HLA-driven effects can affect long-term leukemia control.

Conflict-of-interest disclosure: The authors declare no competing financial interests.