Cell-free DNA in ex-vivo lung perfusate is associated with low-quality lungs and lung transplant outcome

Abstract

Background: Cell-free DNA (cfDNA) in ex-vivo lung perfusion (EVLP) perfusate has been shown to potentially reflect lung injury; however, the relationship between cfDNA concentration with clinical EVLP lung outcomes has not been elucidated.

Methods: A discovery cohort of n = 100 clinical EVLP cases and a validation cohort (n = 50) were used in this single-center, retrospective cohort study. cfDNA was extracted and quantified from perfusate samples. The concentration of cfDNA at 1 hour and the change in cfDNA concentration per hour of EVLP in the transplanted and declined groups were compared by univariable and multivariable logistic regression. cfDNA was introduced as an additional factor in a machine-learning algorithm to predict lung utilization and postoperative outcome and the performance evaluated.

Results: Significantly higher cfDNA concentrations were observed in the declined group than in the transplanted group (1 hour: p < 0.001; delta/hour: p = 0.031). Multivariable analysis among the 1 hour factors showed that [cfDNA 1 hour] (OR 4.27, p = 0.010) was an independent prognostic factor. Increases in [cfDNA 1 hour], [cfDNA delta/hour], and both showed that both initial [cfDNA] and increases in [cfDNA] over time were independently correlated with the probability of a lung being declined. The validation analysis also confirmed higher [cfDNA 1 hour] in the declined group than in the transplanted group (p = 0.010). Addition of [cfDNA] features improved the performance of a machine-learning algorithm used to predict donor lung utilization.

Conclusions: The cfDNA concentration in EVLP perfusate correlates with the rate of decline of lungs for transplant from EVLP.

Keywords: cell-free DNA; donor lung evaluation; ex-vivo lung perfusion; lung transplantation; machine-learning.