Effects of ABO-Incompatible Blood Transfusion on Immune Response and Rejection After Organ Transplantation

Peri-operative blood transfusion is common during solid-organ transplantation, yet the impact of ABO-incompatible (ABOi) blood exposure on post-transplant immunity and acute rejection (AR) remains uncertain.

We assembled a multicenter cohort of 468 adult kidney, liver and heart recipients (2018-2024). Detailed, time-stamped transfusion records were linked to serial immune-phenotyping (30-plex cytokines, flow-cytometric cell subsets) collected from pre-operative baseline to 12 months. The causal effect of ABOi transfusion on biopsy-proven AR was estimated with marginal structural models that incorporated daily inverse-probability-of-treatment weighting and Fine-Gray competing-risk adjustment. We further developed static (XGBoost-Cox) and dynamic attention-LSTM models to predict 30-, 90- and 365-day AR; model interpretability employed SHAP values and integrated gradients.

Fifty-six recipients (12%) received ≥ 1 unit of ABOi blood. After adjustment for baseline characteristics, surgical bleeding, tacrolimus exposure and other time-varying confounders, ABOi transfusion independently increased 1-year AR risk (HR 1.65; 95% CI 1.12-2.44). A clear dose-response was observed: transfusion of > 3 incompatible red-cell units doubled AR incidence. ABOi exposure produced an early IL-6 surge (80 pg mL^{- 1} vs 60 pg mL^{- 1} in controls, p < 0.001) and sustained elevation of CD8⁺/Treg ratios. Unsupervised clustering of 10 immune markers identified a hyper-inflammatory phenotype (Cluster A) with a two-fold higher AR hazard, strongly correlated with ABOi load. The dynamic model achieved AUROCs of 0.83, 0.79 and 0.75 for 30-, 90- and 365-day prediction, respectively, and provided patient-level risk passports 24 h before clinical rejection.

Peri-operative ABOi transfusion is a modifiable, dose-dependent driver of AR across solid-organ transplants. Early IL-6 release and cytotoxic T-cell expansion constitute plausible biological mediators. Minimizing incompatible antigen load and leveraging dynamic risk analytics may improve graft surveillance and inform targeted immunomodulation.