Class I Graft-Versus-Host Mismatch Based on Predicted Indirectly Recognizable HLA Epitopes (PIRCHES) Is Associated with Worse Post-Transplant Outcomes in Patients Receiving Peripheral Blood, T-Cell Replete Haploidentical Hematopoietic Cell Transplantation with Post-Transplant Cyclophosphamide

Abstract

Introduction High dose post-transplant cyclophosphamide has improved the outcomes and expanded the use of haploidentical hematopoietic cell transplantation (haplo-HCT). The impact of HLA disparity in this setting, however, is unclear and is traditionally measured at the antigen or allele level. The Predicted Indirectly Recognizable HLA Epitopes (PIRCHES) model provides a novel tool to calculate the number of mismatched HLA peptides between donor and host that can be processed and presented by shared HLA antigens for T cell recognition. No data exist on the relationship between PIRCHES mismatch and clinical outcome in haplo-HCT with post-transplant cyclophosphamide. Methods 148 patients who received a peripheral blood, T-cell replete haplo-HCT with post-transplant cyclophosphamide at a single center between 2009 and 2016 were retrospectively analyzed. The PIRCHES mismatch load was calculated using the PIRCHES matching tool online and was categorized by class and vector. The primary outcome was incidence of relapse. The association between PIRCHES mismatch and outcome was analyzed using the Cox proportional hazard model or Gray's sub-distribution method for competing risk as appropriate. This study was approved by the Institutional Review Board. Results The median follow-up for survivors was 34.0 months (range, 14.5-78.3 months). One-hundred (67.6%) patients were deceased and 61 (41.2%) patients relapsed during follow-up. Class I graft-versus-host (GvH) PIRCHES mismatch had a median of 11 with a range of 0 to 56. Class II GvH PIRCHES mismatch had a median of 29 with a range of 0 to 122. Class I GvH PIRCHES mismatch was associated with worse acute graft-versus-host disease (aGvHD) (grades II-IV) (HR 1.021 per unit PIRCHES mismatch increase; 95% CI 1.001-1.040; p=0.036). Class I GvH PIRCHES mismatch was not associated with treatment-related mortality (TRM), but was associated with a trend towards worsened overall survival (OS) (HR 1.015 per unit PIRCHES mismatch increase; 95% CI 0.999-1.033; p=0.073). Class I host-versus-graft (HvG) PIRCHES mismatch was associated with worse relapse-free survival (RFS) (HR 1.018; 95% CI 1.001-1.035; p=0.036) and worse OS (HR 1.018; 95% CI 1.001-1.035; p=0.036) but not relapse. Class II PIRCHES mismatch was not associated with relapse, OS, RFS, TRM, or aGvHD. Both class I and II PIRCHES mismatch were not associated with chronic GvHD or engraftment. Discussion Class I GvH PIRCHES mismatch was associated with greater aGvHD and a trend towards worsened OS, while class I HvG PIRCHES mismatch was associated with worsened RFS and OS. Class II PIRCHES mismatch was not associated with outcomes and appears to be tolerated by this patient cohort. PIRCHES mismatch represents a novel strategy to predict clinical outcome in haplo-HCT. Further studies using registry data and prospective studies are needed to validate these findings.