RESUMO
BACKGROUND: Our knowledge of de novo anti-HLA donor-specific antibodies (dnDSA) in liver transplantation continues to be defined. We hypothesized that differences of HLA-DR/DQ mismatches can improve precision in alloimmune risk categorization and be applied to tailor immunosuppression. METHODS: A retrospective chart review of 244 pediatric patients consecutively transplanted at our center between 2003 and 2019 was performed to identify patients tested for dnDSA. Records were queried for: demographics, pre-transplant diagnosis, biopsy-proven T-cell-mediated rejection (TCMR), radiology proven biliary complications, tacrolimus trough levels, dnDSA characteristics, and HLA typing. The eplet mismatch analyses were performed using HLAMatchmaker™ 3.1. All statistical analyses were conducted using R software version 3.40. RESULTS: There were 99 dnDSA-negative patients and 73 dnDSA-positive patients (n = 70 against class II and n = 3 against class I and II). ROC analysis identified optimal cutoff of eplet mismatch load for dnDSA and defined risk groups for an alloimmune outcome. Kaplan-Meier curves and log-rank tests showed high eplet mismatch load was associated with shorter dnDSA-free survival (log-rank p = .001). Multivariable Cox regression models showed that tacrolimus coefficient of variation and tacrolimus mean levels were significantly associated with dnDSA-free survival (p < .001 and p = .036). Fisher's exact test showed that dnDSA was associated with an increased likelihood of TCMR (OR 14.94; 95% CI 3.65 - 61.19; p < .001). Patients without TCMR were more likely to have dnDSA to HLA-DQ7 and less likely to have dnDSA to HLA-DQ2 (p = .03, p = .080). CONCLUSIONS: Mismatched epitope load predicts dnDSA-free survival in pediatric liver transplant, while dnDSA specificity may determine alloimmune outcome.