Relationship Between Donor Derived Cell-Free DNA and Tissue-Based Rejection-Related Transcripts In Heart Transplantation

Abstract

Introduction Endomyocardial biopsy (EMB)-based traditional microscopy remains the gold standard for the detection of cardiac allograft rejection, despite its limitation of inherent subjectivity leading to inter-reader variability. Alternative techniques now exist to surveil for allograft injury and classify rejection. Donor-derived cell-free DNA (dd-cfDNA) testing is now a validated blood-based assay used to surveil for allograft injury. The molecular microscope diagnostic system (MMDx) utilizes intragraft rejection-associated transcripts (RATs) to classify allograft rejection and identify injury. The use of dd-cfDNA and MMDx together provides objective molecular insight into allograft injury and rejection. The aim of this study was to measure the diagnostic agreement between dd-cfDNA and MMDx and assess the relationship between dd-cfDNA and MMDx-derived RATs which may provide further insight into the pathophysiology of allograft rejection and injury. Methods: This is a retrospective observational study of 186 endomyocardial biopsy (EMB) evaluated with traditional microscopy and MMDx. All samples were paired with dd-cfDNA from peripheral blood prior to EMB (up to 1 month). Diagnostic agreement between traditional microscopy, MMDx, and dd-cfDNA (threshold of 0.20%) were compared for assessment of allograft injury. In addition, the relationship between dd-cfDNA and individual RAT expression levels from MMDx was evaluated. Results MMDx characterized allograft tissue as no rejection (NR) (64.5%), antibody-mediated rejection (ABMR) (25.8%), T-cell-mediated rejection (TCMR) (4.8%), and mixed ABMR/ TCMR (4.8%). For the diagnosis of any type of rejection (TCMR, ABMR, and mixed rejection), there was substantial agreement between MMDx and dd-cfDNA (74.7% agreement). All transcript clusters (group of gene sets designated by MMDx) and individual transcripts considered abnormal from MMDx had significantly elevated dd-cfDNA. In addition, a positive correlation between dd-cfDNA levels and certain MMDx-derived RATs was observed. Tissue transcript clusters correlated with dd-cfDNA scores, including DSAST, GRIT, HT1, QCMAT and S4. For individual transcripts, tissue ROBO4 was significantly correlated with dd-cfDNA in both non-rejection and rejection as assessed by MMDx. Conclusion: Collectively, we have shown substantial diagnostic agreement between dd-cfDNA and MMDx. Furthermore, based on the findings presented, we postulate a common pathway between the release of dd-cfDNA and ROBO4 (a vascular endothelial-specific gene that stabilizes the vasculature) in the setting of AMR, which may provide a mechanistic rationale for observed elevations in dd-cfDNA in AMR, compared to ACR.