Clinical Relevance of Donor-Derived Cell-Free DNA in Monitoring Kidney Transplant Rejection.

KEY POINTS: Donor-derived cell-free DNA showed strong discriminative ability for rejection after kidney transplantation. Clinical interpretability was improved by calculating donor-derived cell-free DNA result-specific likelihood ratio and corresponding post-test rejection risk. We showed that donor-derived cell-free DNA testing could help reduce unnecessary kidney biopsies, delineating a key clinical context in a low rejection prevalence cohort.

BACKGROUND: Over the past 25 years, donor-derived cell-free DNA (dd-cfDNA) has emerged as a noninvasive biomarker for detecting allograft rejection in kidney transplant recipients, yet its clinical interpretability and context(s) of use in post-transplant monitoring remain largely unexplored.

METHODS: In this observational cohort study, we analyzed 472 retrospective and 450 prospective plasma samples as the derivation and validation cohorts, respectively. Samples were collected at the time of either an indication or a protocol kidney biopsy. We evaluated three logistic regression models for rejection discrimination: a clinical model, the dd-cfDNA alone, and their combination. Result-specific likelihood ratios were derived from the dd-cfDNA-based model probabilities and applied to the clinical model predictions to obtain post-test probabilities of rejection. By comparing pre-test and post-test probabilities, we assessed the impact of dd-cfDNA on rejection risk stratification.

RESULTS: dd-cfDNA alone demonstrated strong discriminative ability, with area under the receiver operating characteristic curves of 0.79 (95% confidence interval, 0.73 to 0.84) and 0.80 (95% confidence interval, 0.72 to 0.87) in the derivation and validation cohorts, respectively. Combining dd-cfDNA with clinical parameters improved the above area under the receiver operating characteristic curves to 0.82 and 0.87, respectively. We compared pre-test and post-test probabilities to determine how dd-cfDNA testing influenced rejection risk estimation. Incorporating dd-cfDNA testing results clarified risk stratification by reducing the number of samples categorized as intermediate (10%-30%) pre-test risk. Overall, 253/345 pre-test intermediate risk samples were reclassified, with 117/162 (72%) and 33/91 (36%) correctly assigned to low (≤10%) and high (≥30%) post-test risk, respectively. With 10% probability threshold for warranting a kidney biopsy, 482 biopsies would have been safely avoided using post-test probabilities, with 437 (91%) of these being protocol biopsies.

CONCLUSIONS: Our study confirmed the strong association between dd-cfDNA% and rejection after kidney transplantation. Using test result-specific likelihood ratio, we evaluated the impact of integrating dd-cfDNA% for estimating time point-specific rejection risk, demonstrating its potential to reduce the number of unnecessary biopsies in a cohort with a relatively low rejection prevalence.