Introduction: Timely detection and management of graft injury is essential for favorable long-term outcomes in pediatric liver transplant (LT). New non-invasive biomarkers may offer potential for enhanced clinical surveillance. We assessed donor-derived cell-free DNA (ddcfDNA), a promising biomarker for diagnosis of acute rejection, in combination with liver function tests (LFTs) on biopsy-proven acute rejection (BPAR) cases.
Methods: A cross-sectional study including pediatric LT patients undergoing surveillance or for-cause biopsy was performed. Biopsy-paired blood samples were collected and stored until ddcfDNA quantification (AlloSeq cfDNA, CareDx). Demographic, peritransplant, and biochemical data were collected. The histopathological outcomes were independently classified by two pathologists according to the Banff criteria. Patients underwent surveillance biopsy if their LFTs<2 the upper limit normal (ULN) while for-cause biopsy was performed if their LFTs≥2ULN. Patients in the surveillance biopsy group included BPAR-free and subclinical-BPAR (subBPAR) cases according to histopathology assessment. ROC curves were used to evaluate ddcfDNA’s and LFTs' performance for diagnosing BPAR and to establish a ddcfDNA threshold that could be used with LFT for identifying subBPAR from other clinically stable patients.
Results: 139 biopsy-paired ddcfDNA samples from 55 BPAR (32 subBPAR and 23 clinically overt BPAR) and 84 BPAR-free patients (40 indeterminate for T cell-mediated rejection and 44 normal histology) were analyzed. DdcfDNA levels were significantly higher in BPAR cases compared to those BPAR-free (p<0.001). ROC threshold (AUC) for identifying BPAR was 10.4% (74.2%) for ddcfDNA, 48UI/L (71.9%) for ALT, and 36UI/L (74.7%) for AST. Moreover, ddcfDNA ROC threshold (AUC) for identifying subBPAR cases was 1.9% (63.8%). To distinguish between BPAR-free and patients with BPAR, including clinically overt BPAR and subBPAR patients, we used a combined approach. Patients with LFTs≥2ULN (n=30) and those with LFTs<2ULN and ddcfDNA≥1.9% (n=80) were considered at risk for rejection while those with LFTs<2ULN and ddcfDNA<1.9% (n=29) as clinically stable. Using this approach we correctly identified 30 of 32 subBPAR cases. The sensitivity (96.4%) and negative predictive value (93.1%) were notorious with only 2 false negatives and 27 true negatives that could have avoided surveillance biopsies.
Conclusions: This study suggests the use of ddcfDNA as a biomarker to guide surveillance biopsy and immunosuppression therapy in children with LT. Further evaluations, including peritransplant and pharmacological variables, and longitudinal ddcfDNA assessment are being performed to identify multiple risk factors and to evaluate the use of ddcfDNA for individualized therapy strategies.