Heart allograft rejection: molecular diagnosis using intra-graft targeted gene expression profiling.

Summary

In an international cohort of 671 FFPE endomyocardial biopsies, targeted gene expression classifiers for AMR and ACR achieved AUCs ~0.81–0.85 in internal and external validation, aligning with pathologic severity. The workflow produces automated clinician-friendly reports, enabling immediate translational use alongside histopathology.

Key Findings

• Molecular classifiers for AMR and ACR achieved AUC 0.812/0.849 (internal) and 0.822/0.815 (external).
• AMR signatures mapped to IFN-γ signaling, endothelial activation, and monocyte–macrophage recruitment.
• ACR signatures mapped to TCR/CD3/CD28 signaling; model scores correlated with pathologic severity.
• Automated reporting developed for clinical implementation.

Clinical Implications

Centers can add FFPE-based molecular classifiers to EMB workflows to improve diagnostic confidence for AMR/ACR, especially in borderline or ambiguous cases, and to standardize reporting across sites.

Limitations

• Clinical utility metrics (e.g., change in management/outcomes) were not tested.
• Generalizability to all platforms/laboratories requires further inter-lab concordance studies.

Future Directions

Prospective trials should test whether molecular classifier–guided management improves rejection detection, reduces unnecessary therapy, and impacts graft outcomes.