Clinical impact of predicted indirectly recognizable HLA epitopes (PIRCHE) in single unit cord blood transplantation using tacrolimus and methotrexate for graft-versus-host disease prophylaxis

Abstract Background: The Predicted Indirectly Recognizable HLA Epitopes (PIRCHE) algorithm estimates the potential for indirect recognition of mismatched HLA derived peptides by donor T cells inducing alloimmune responses. A previous study reported that, in cord blood transplantation (CBT) predominantly using tacrolimus and mycophenolate mofetil for graft-versus-host disease (GVHD) prophylaxis, the presentation of class I epitopes derived from mismatched recipient HLA-B, as predicted by the PIRCHE algorithm, was associated with severe pre-engraftment immune reactions (PIR) and increased non-relapse mortality (NRM). In contrast, GVHD prophylaxis with tacrolimus and methotrexate (Tac+MTX) in CBT has been reported to be associated with a lower incidence of PIR. Therefore, the impact of PIRCHE under this distinct immunosuppressive context in CBT remains unclear. Aim: To evaluate the clinical impact of PIRCHE in single-unit CBT using Tac+MTX for GVHD prophylaxis. Methods: We retrospectively analyzed 97 patients with acute leukemia who underwent their first allogeneic stem cell transplantation using single-unit CBT with Tac+MTX between January 2011 and January 2024 at our center. High-resolution HLA typing (HLA-A, -B, -C, -DRB1) was performed for all patients and donors. PIRCHE scores were calculated based on mismatched recipient peptides predicted to be presented by donor HLA class Ⅰ and Ⅱ. Patients were categorized into three groups based on thresholds that closely reflected the tertile distribution of PIRCHE scores: P1-L (PIRCHE Ⅰ low, ≤5), P1-Int (intermediate, 6–10), and P1-H (high, >10); and P2-L (PIRCHE Ⅱ low, ≤5), P2-Int (intermediate, 6–15), and P2-H (high, >16). Group comparisons of patient characteristics and outcomes, including overall survival (OS), relapse, NRM, incidence of GVHD were conducted. Results: The median age at transplant was 44 years (range, 17–68 years), and 44 patients were male. The patients' diseases were acute myeloid leukemia (n = 67) and acute lymphoblastic leukemia (n = 30). Disease status at transplant was CR1 (n = 49), CR2 (n = 28), and non-CR (n = 20). Number of HLA allele mismatch were 0–2 (n = 17), 3–5 (n = 69) and 6–8 (n = 12). Most of patients (n = 87) received the myeloablative conditioning. The number of patients in each group was as follows: P1-L (n = 31), P1-Int (n = 35), P1-H (n = 31); and P2-L (n = 28), P2-Int (n = 27), P2-H (n = 42). There were no significant differences in characteristics among the groups. The cumulative incidence of grade Ⅱ–Ⅳ acute GVHD at day 100 showed no significant difference: P1-L 19% (95% CI 8%–35%), P1-Int 17% (7–31), P1-H 7% (1–19), p = 0.30; P2-L 14% (4–30), P2-Int 22% (9–40), P2-H 10% (3–21), p = 0.36. At 1 year, the incidence of any grade chronic GVHD also showed no significant differences: P1-L 11% (3–26), P1-Int 6% (1–18), P1-H 0% (0–0), p = 0.23; P2-L 0% (0–0), P2-Int 12% (3–29), P2-H 5% (1–16), p = 0.19. In PIRCHE Ⅰ groups, no significant differences were observed in 2-year OS, NRM, and cumulative incidence of relapse (CIR). OS: P1-L 75% (54–87), P1-Int 79% (61–90), P1-H 56% (34–73), p = 0.57. NRM: P1-L 12% (3–27), P1-Int 4% (0–16), P1-H 15% (6–28), p = 0.36. CIR: P1-L 25% (11–42), P1-Int 22% (9–38), P1-H 33% (16–51), p = 0.53. In contrast, PIRCHE Ⅱ groups showed a non-linear pattern, with significantly better outcomes in P2-Int. OS: P2-L 64% (41–79), P2-Int 91% (69–98), P2-H 63% (45–76), p = 0.042. NRM was similar among groups: 11% in all three (p = 0.82). CIR: P2-L 23% (9–41), P2-Int 10% (2–27), P2-H 39% (23–54), p = 0.016. Multivariate analyses for CIR and OS were performed, including age ≥50 years, conditioning intensity, disease status, number of HLA allele mismatches, and PIRCHE score categories as covariates. For CIR, CR2 (HR 8.37; 95% CI, 2.35–29.88; p = 0.0011), non-CR (HR 15.15; 3.06–75.03; p < 0.001), and P2-Int (HR 0.19; 0.045–0.80; p = 0.023) were identified as independent predictors. For OS, both non-CR status (HR 3.76; 1.35–10.45; p = 0.011) and P2-Int (HR 0.22; 0.061–0.77; p = 0.018) were also identified as independent predictors. Conclusion: PIRCHE Ⅱ scores estimate CD4+ T-cell–mediated indirect immunogenicity. An intermediate PIRCHE Ⅱ score was independently associated with improved OS in CBT using Tac+MTX, likely due to reduced relapse. This non-linear association may be explained by a balance between CD4+ T-cell–mediated alloreactive antitumor effects and immune exhaustion, although further immunological validation is needed.