For the roughly half of blood cancer patients who relapse after stem cell transplantation, understanding why the cancer escapes immune surveillance is critical to planning second-line treatment. A large multicenter analysis now clarifies that the answer often lies in the geometry of immune recognition itself — and that donor choice at the outset substantially determines whether this escape route is available.
Across 533 post-transplant relapses drawn from 27 centers worldwide, genomic loss of mismatched human leukocyte antigen (HLA) alleles was detected in 15.6% of cases overall — but the rate varied dramatically by donor type: 28.7% in haploidentical family donor transplants, 7.2% in matched unrelated adult donors, and just 2.7% in cord blood recipients. Crucially, a newly developed next-generation sequencing pipeline and a web-based phasing tool built on HLA data from approximately five million individuals revealed that the spatial arrangement of HLA mismatches across a patient's two haplotypes — not merely their number — drives risk. When all mismatches clustered on a single haplotype, HLA loss occurred in 27.6% of relapses; when mismatches were distributed across both haplotypes, the rate dropped to 5.4%. Clinically, HLA loss rendered donor lymphocyte infusions from the original donor essentially ineffective, while switching to a second transplant from a different donor conferred a meaningful survival advantage.
This work reframes HLA incompatibility from a binary transplant risk factor into a directional, mappable vulnerability that oncologists could assess before the first graft is chosen. The findings are particularly significant given the rapid expansion of haploidentical transplantation over the past decade. While the study is observational and cannot fully account for disease heterogeneity across centers, its scale and genomic rigor are notable. The practical implication — that haplotype phasing could inform both donor selection and post-relapse treatment sequencing — represents a potentially paradigm-shifting integration of population genomics into transplant decision-making, warranting prospective validation.