Elevated fractional donor-derived cell-free DNA during subclinical graft injury after liver transplantation.

Personalized immunosuppression (IS) promises to improve the balance of necessary control of alloreactivity and dose-dependent adverse effects of long-term IS such as kidney insufficiency, infections, and malignancies. The majority of liver transplantation (LT) recipients exhibit graft injuries (graft inflammation and/or fibrosis) that are not eligible for an IS reduction according to current Banff criteria, even when liver enzymes are normal or only marginally elevated. This cross-sectional study evaluated the noninvasive prediction of such subclinical graft injuries in surveillance liver biopsies via donor-derived cell-free DNA (dd-cfDNA). Absolute and fractional dd-cfDNA increased stepwise from patients without histological signs of rejection (n = 26) over subclinical graft injury (n = 61), including subclinical T cell-mediated rejection to clinical overt T cell-mediated rejection (n = 21). Thus, fractional plasma dd-cfDNA was significantly elevated paired to surveillance biopsies with relevant subclinical graft injury according to 2016 Banff criteria compared with those with minimal or absent histological graft injury. In contrast, the presence of donor-specific anti-human leukocyte antigen antibodies was not associated with the amount of dd-cfDNA. The sensitivity and specificity of fractional dd-cfDNA to noninvasively predict relevant subclinical graft injury was rather limited with 73% and 52% at the cutoff value of 2.1% fractional dd-cfDNA. The positive predictive value of fractional dd-cfDNA above 2.1% was 76% to noninvasively predict subclinical graft injury, calculated on the prevalence of graft injury in our prospective surveillance biopsy program, whereas the negative predictive values was not predictive (47%). In conclusion, dd-cfDNA has a rather limited diagnostic fidelity in addition to other noninvasive markers for the assessment of subclinical graft injury in personalized IS approaches after LT in a cross-sectional setting.