Normothermic machine perfusion (NMP) is a new organ preservation method that maintains organs at physiological conditions through an artificial circulatory system prior to transplantation. By allowing functional assessment of livers NMP has enabled the transplant of organs that would normally be discarded, helping alleviate the problem of donor liver shortage. Currently, changes in metabolite levels (e.g. lactate) in the circulating blood (perfusate) are used to determine liver functionality and viability during NMP. However, there is a lack of consensus on which measures best predict transplant outcomes. A greater understanding of the molecular changes in the organ during NMP could help inform such criteria. In this study, we investigate transcriptomic changes observed livers that have been stored using NMP.
Biopsies were collected for 14 livers that underwent NMP at the start and end of cold ischemic time (storage on ice) as well as after 3 and 6 hours of perfusion. Following perfusion, livers were labelled by surgeons as viable or non-viable based on currently accepted viability criteria. RNAseq data was generated in these samples. Differential gene expression genes analysis was then performed to compare each viability group to cold ischemic time.
An active transcriptome was identified within viable livers, while non-viable livers displayed a comparatively inert transcriptome. Clustering of samples using these DEGs identified one liver that was considered clinically viable, but whose transcriptome was much more similar to that of non-viable livers, highlighting that gene expression may capture molecular changes related to viability that are not capture by clinical criteria. Finally, our gene expression signature predicted with high accuracy (AUROC ) viable NMP livers in an out-of-sample cohort.
This study highlights that gene expression can distinguish functional from non-functional livers following NMP and that investigation of the molecular changes could lead to identification of novel biomarkers for liver viability.