Non-Permissive Hla-Dpb1 Mismatches are Associated with Increased 1-Year Mortality after Unrelated Donor Hematopoietic Cell Transplantation for Non-Malignant Disorders

Aim: The impact of T-cell epitope (TCE) permissive and non-permissive HLA-DPB1 mismatches on overall survival (OS) of patients with non-malignant disorders undergoing HLA 10/10 unrelated donor hematopoietic cell transplantation (URD-HCT) is currently unknown.

Methods: High-resolution HLA typing was performed using either Sanger or next-generation sequencing. The TCE3 v2.0 algorithm (https://www.ebi.ac.uk/ipd/imgt/hla/matching/dpb_v2) classified HLA-DPB1 mismatches into permissive (PM) and non-permissive (NPM). Cox proportional hazards regression was used for the multivariable analysis of OS.

Results: In total, 151 patients with non-malignant disorders underwent URD-HCT between 2007 and 2022. The median age was 9 years (range, 0-51), and the main indications for URD-HCT were inherited bone marrow failure syndromes (n=72; 47.7%) and acquired severe aplastic anemia (n=49; 32.4%). All patients received bone marrow as the graft source, and 145 patients (96%) received in vivo T-cell depletion with ATG. All pairs had HLA 10/10 high-resolution compatibility. Of these, 122 (80.8%) had HLA-DPB1 mismatches, 73 being PM and 49 NPM. In the Cox regression, adjusted for patient age and CMV status, a violation of the proportional hazards assumption (PHA) was observed (P=0.035). The Schoenfeld residuals indicated different risks before and after day +400 (Figure 1A). Thus, follow-up was split, and the PHA was held before and after day +400 (P = 0.59 and P = 0.17, respectively). Before day +400, the presence of NPM was strongly associated with increased mortality (HR: 10.8; 95% CI: 1.99-58.26; P=0.006). In contrast, NPM did not impact OS after day +400 (HR: 0.62; 95% CI: 0.08-4.61; P=0.64). Interestingly, PM and 12/12 matching did not show significant differences in OS before (HR: 2.73; 95% CI: 0.50-14.96; P=0.25) and after (HR: 1.21; 95% CI: 0.24-6.09; P=0.82) the day +400 (Figure 1B).

Conclusion: In conclusion, our study describes that NPMs are associated with poorer OS before day +400 in patients with non-malignant disorders receiving 10/10 URD-HCT. Therefore, such HLA-DPB1 incompatibilities should be avoided whenever possible. Additionally, in the absence of a donor with HLA 12/12 compatibility, a PM should be prioritized due to its better OS profile. These findings should be validated in independent cohorts.