Donor-derived cell-free DNA with clinical and echocardiographic data as independent predictors of Coronary Allograft Vasculopathy in heart-transplant recipients

Background Coronary allograft vasculopathy (CAV) remains a leading cause of late graft failure after heart transplantation (HTx). Because CAV often develops silently, identifying circulating biomarkers and clinical parameters able to detect early vascular injury is of major clinical interest. Methods We analyzed 117 HTx recipients evaluated a median of 85.7 months after transplantation. Patients were stratified according to the presence of CAV. Clinical characteristics, echocardiographic parameters, and circulating biomarkers reflecting myocardial injury, heart failure, and graft damage—high-sensitivity troponin I (hs-TnI), N-terminal pro-brain natriuretic peptide (NT-proBNP), mid-regional pro-adrenomedullin (MR-proADM), and donor-derived cell-free DNA (dd-cfDNA)—were assessed. Univariate logistic regression identified variables associated with CAV. Multivariable logistic regression models were constructed, and model performance was evaluated using ROC analysis and confusion matrices. Results Fifty-seven patients developed CAV while 60 did not. Patients with CAV were older and had a significantly longer time since HTx. Mycophenolate therapy was more frequently used in patients without CAV. Echocardiographic indices of diastolic dysfunction were significantly altered in CAV (Table 1). Among circulating biomarkers, MR-proADM and dd-cfDNA were significantly elevated in patients with CAV (Table 1). Univariate analysis identified twelve variables associated with CAV, including time since HTx, age, BMI, diastolic function parameters, MR-proADM, and dd-cfDNA, whereas mycophenolate therapy was associated with reduced risk (Figure 1A). ROC analysis demonstrated the best discrimination for dd-cfDNA (AUC 0.766) and MR-proADM (AUC 0.736) (Figure 1B). Because several predictors reflected overlapping clinical or physiological domains, correlation analysis was performed to identify collinear variables before multivariable modeling (Figure 1C). Representative variables from each correlated cluster were then selected to avoid redundancy and improve model stability. In multivariable analysis, time since HTx and dd-cfDNA remained independent predictors of CAV, while mycophenolate therapy showed a protective association (Figure 1D). A parsimonious stepwise model including time since HTx, dd-cfDNA, E/e′ ratio, BMI, and mycophenolate demonstrated strong discrimination (AUC 0.856), comparable to the full model (Figure 1E,F). Conclusions Circulating biomarkers reflecting graft injury and vascular stress are strongly associated with coronary allograft vasculopathy after heart transplantation. dd-cfDNA emerged as an independent predictor of CAV, while mycophenolate therapy was associated with reduced risk. Integration of biomarkers with echocardiographic and clinical parameters enabled accurate discrimination of CAV, highlighting the potential of multimodal biomarker-based approaches for improving CAV risk stratification in transplant recipients.Table 1Figure1