A two-threshold donor-derived cell-free DNA algorithm improves detection of molecular rejection in heart transplantation

BACKGROUND: Donor-derived cell-free DNA (dd-cfDNA), a noninvasive biomarker for heart allograft rejection, was initially developed based on the percent of total cfDNA. A two-threshold algorithm (2TA) was recently developed, assigning high risk of rejection if dd-cfDNA was ≥0.26% or donor-quantity score (DQS) was ≥18 copies/mL based on histologic biopsy assessment. The present study evaluated 2TA for its performance in detecting molecular rejection using the Molecular Microscope Diagnostic System (MMDx). METHODS: We assessed endomyocardial biopsies with matched dd-cfDNA assessments from the multicenter, prospective Trifecta-Heart study (ClinicalTrials.gov, NCT04707872). RESULTS: Among 366 biopsy-matched samples from 236 patients, MMDx classified 108 biopsies as rejection and 258 as no rejection. Median dd-cfDNA% and DQS were significantly higher in rejection vs non-rejection (0.67% vs 0.07%; 44 vs 1 copies/mL; p<0.001 for both) and were higher in antibody-mediated, T cell-mediated, and mixed rejection. Compared to the previous 0.15% dd-cfDNA cutoff, 2TA provided higher specificity (83.7% vs 76.7%) but similar sensitivity (83.3% vs 82.4%), reducing false positives by 30% with AUC of 0.87, a net reclassification index of 7.23%, and detecting molecular rejection better than histologic rejection. Each 0.05% increase in dd-cfDNA% and 5-copy/ml increase in DQS raised the odds of MMDx rejection by 14% and 20%, respectively. CONCLUSIONS: Using dd-cfDNA% with DQS for heart transplant rejection improves specificity and AUC compared to the previous cutoff, predicting molecular rejection better than histologic rejection.