Molecular Diagnosis of Heart Allograft Rejection Using Intra-Graft Targeted Gene Expression Profiling: the heart HistoMX system

Background and aimsTissular gene expression profiling has the potential to refine the diagnosis of cardiac allograft rejection. Contrary to whole-transcriptome approaches, targeted molecular profiling applicable to formalin-fixed paraffin-embedded (FFPE) endomyocardial biopsies (EMB) can be easily implemented in clinical practice. We aimed to develop and validate the first rejection molecular diagnostic system dedicated to heart transplantation (HTx). MethodsAn international multicenter study was designed, building a deep phenotyped cohort of HTx recipients recruited between 2011 and 2021 at 4 referral centers. Detailed donors, recipients, clinical, immunological, biological, and histological parameters were collected. EMBs were graded according to international working formulations. Tissue gene expression was analyzed on FFPE-EMB using the consensus Banff Human Organ Transplant gene set. Molecular classifiers of antibody-mediated (AMR) and acute cellular rejection (ACR) were built. Discrimination and calibration were assessed in the development and validation sets (NCT06436027). ResultsA total of 591 biopsies were included: 188 AMR (pAMR1(I+): n=51; pAMR1(H+): n=58; pAMR2-3: n=79), 289 ACR (1R n=174; 2-3R n=115) and 114 matched non-rejection cases. Biopsies were split in a derivation (n=476) and a validation set (n=115). AMR top significant transcripts were related to the IFN-gamma inducible pathway, endothelial activation, and monocyte-macrophage recruitment. ACR was characterized by transcripts related to T-cell receptor, CD3 receptor activation, and CD28 signaling. ACR and AMR molecular rejection models were strongly associated with the pathology severity of rejection and accurately identified rejection in the derivation (ROC-AUC: AMR=0.831, ACR:=0.837) and validation sets (ROC-AUC: AMR=0.812; ACR=0.849). Calibration was adequate. The robustness of the molecular classifiers were reinforced by various sensitivity analyzes. An automated report was developed to enhance the reproducibility and clinical applicability of the molecular analysis. ConclusionsIn this study, the first tissue-based rejection molecular diagnostic system applicable to FFPE- EMB and dedicated to heart transplantation rejection was developed and internally validated. This tool has the potential to refine the diagnosis of rejection.