Introduction: The challenges of finding HLA-matched donors for hematopoietic stem cell transplantation (HSCT) has led to the increasing use of haploidentical donors (haplo-HSCT), with post-transplant cyclophosphamide (PTCy) reducing the immunological impact of HLA mismatches. However, the criteria for optimal donor selection from an immunological standpoint remain controversial.

The Predicted Indirectly Recognizable HLA Epitopes (PIRCHE) algorithm is a computational tool that estimates the immunogenic potential of HLA mismatches by predicting HLA-derived peptides presented by HLA molecules, modeling indirect allorecognition by CD4⁺ and CD8+T cells. The HLA-B leader polymorphism refers to a dimorphic variant in exon 1 of the HLA-B gene encoding the signal peptide, involving either methionine (M) or threonine (T) at position -21. This variation influences HLA-E stabilization and immune modulation, potentially affecting alloreactivity and transplant outcomes.

Aim: To evaluate the immunogenicity of HLA mismatches using the PIRCHE algorithm and HLA-B leader dimorphism in haplo-HSCT with PTCy, and their association with post-transplant outcomes, including acute and chronic graft-versus-host disease (aGVHD, cGVHD), relapse-free survival (RFS), GVHD-free relapse-free survival (GRFS) and overall survival (OS).

Materials and Methods: A retrospective analysis was conducted on 193 patients who underwent haplo-HSCT at the University Hospital of Salamanca, Spain. HLA typing (HLA-A, -B, -C, -DRB1, -DQB1, -DPB1) was performed using PCR-SSOP and/or next-generation sequencing, when appropiate. Molecular HLA compatibility was assessed with the PIRCHE algorithm, generating scores including PS-A, PS-B, PS-C, PS-I, PS-II, and combined PS-I+II. HLA-B leader dimorphism and mismatches were determined using BLEAT and the IPD-IMGT/HLA database. Clinical data were extracted from medical records. Categorical variables were reported as frequencies and percentages; continuous variables as medians and ranges. ROC curves were used to determine optimal cut-off values for PIRCHE scores in predicting outcomes. Survival analyses were performed using Kaplan-Meier curves and log-rank tests. Multivariate Cox regression models included variables with p < 0.1 in univariate analyses.

Results: The median age of the patients was 57 years, with a male predominance (56%). Acute myeloid leukemia and myelodysplastic syndromes were the main indications for haplo-HSCT (52%). Median follow-up was 26.9 months (range, 5.5–103.1). Median OS was 18.7 months (range, 0.1–103.1). Median RFS was 12.8 months (range, 0.1–99.7). The cumulative incidence of relapse during follow-up was 18.7%, and overall mortality was 36.3%. The cumulative incidence of grade III–IV aGVHD was 15.7%, and cGVHD occurred in 24.4% of patients. Non-relapse mortality (NRM) was 9.9% at day 100 and 22.8% at 1 year.

Patients with PSI+II >49 (as defined by ROC analysis) had significantly reduced GRFS (HR: 1.93, p=0.001). Similar associations were observed for HCT-CI ≥3 (HR: 1.85, p=0.004) and donor-recipient HLA-B leader mismatches (HR: 2.07, p=0.039). In multivariate analysis, both PS and HLA-B leader mismatch (MTM) were independent predictors of reduced GRFS.

Higher relapse risk was independently associated with high/very high DRI (HR: 2.59, p=0.006), HCT-CI ≥3 (HR: 2.43, p=0.014), and PS-II >28 (HR: 2.19, p=0.037).

No significant association was found between the PIRCHE score or HLA-B leader dimorphism and the risk of aGVHD.

Regarding cGVHD, patients with PS-B >1 had increased risk (HR: 1.82, p=0.042), as did those with HLA-B leader mismatches (HR: 1.69, p=0.065) and recipients carrying the threonine/methionine (TM) dimorphism (HR: 2.21, p=0.021). Both PS-B and B leader dimorphism remained independent predictors of cGVHD in multivariate analysis.

Conclusions: Molecular HLA compatibility assessment using the PIRCHE algorithm and HLA-B leader typing offers additional predictive value for post-transplant outcomes in haplo-HSCT with PTCy and may support more refined donor selection strategies in this setting.

This content is only available as a PDF.
2025
Sign in via your Institution