Use of neural network models to predict liver transplantation waitlist mortality.

Current liver transplantation (LT) organ allocation relies on Model for End-Stage Liver Disease-sodium scores to predict mortality in patients awaiting LT. This study aims to develop neural network (NN) models that more accurately predict LT waitlist mortality. The study evaluates patients listed for LT between February 27, 2002, and June 30, 2021, using the Organ Procurement and Transplantation Network/United Network for Organ Sharing registry. We excluded patients listed with Model for End-Stage Liver Disease (MELD) exception scores and those listed for multiorgan transplant, except for liver-kidney transplant. A subset of data from the waiting list was used to create a mortality prediction model at 90 days after listing with 105,140 patients. A total of 28 variables were selected for model creation. The data were split using random sampling into training, validation, and test data sets in a 60:20:20 ratio. The performance of the model was assessed using area under the receiver operating curve (AUC-ROC) and area under the precision-recall curve (AUC-PR). AUC-ROC for 90-day mortality was 0.936 (95% confidence interval [CI], 0.934-0.937), and AUC-PR was 0.758 (95% CI, 0.754-0.762). The NN 90-day mortality model outperformed MELD-based models for both AUC-ROC and AUC-PR. The 90-day mortality model specifically identified more waitlist deaths with a higher recall (sensitivity) of 0.807 (95% CI, 0.803-0.811) versus 0.413 (95% CI, 0.409-0.418; p < 0.001). The performance metrics were compared by breaking the test data set into multiple patient subsets by ethnicity, gender, region, age, diagnosis group, and year of listing. The NN 90-day mortality model outperformed MELD-based models across all subsets in predicting mortality. In conclusion, organ allocation based on NN modeling has the potential to decrease waitlist mortality and lead to more equitable allocation systems in LT.

The molecular phenotypes of injury, steatohepatitis, and fibrosis in liver transplant biopsies in the INTERLIVER study.

To extend previous molecular analyses of rejection in liver transplant biopsies in the INTERLIVER study (ClinicalTrials.gov #NCT03193151), the present study aimed to define the gene expression selective for parenchymal injury, fibrosis, and steatohepatitis. We analyzed genome-wide microarray measurements from 337 liver transplant biopsies from 13 centers. We examined expression of genes previously annotated as increased in injury and fibrosis using principal component analysis (PCA). PC1 reflected parenchymal injury and related inflammation in the early posttransplant period, slowly regressing over many months. PC2 separated early injury from late fibrosis. Positive PC3 identified a distinct mildly inflamed state correlating with histologic steatohepatitis. Injury PCs correlated with liver function and histologic abnormalities. A classifier trained on histologic steatohepatitis predicted histologic steatohepatitis with cross-validated AUC = 0.83, and was associated with pathways reflecting metabolic abnormalities distinct from fibrosis. PC2 predicted histologic fibrosis (AUC = 0.80), as did a molecular fibrosis classifier (AUC = 0.74). The fibrosis classifier correlated with matrix remodeling pathways with minimal overlap with those selective for steatohepatitis, although some biopsies had both. Genome-wide assessment of liver transplant biopsies can not only detect molecular changes induced by rejection but also those correlating with parenchymal injury, steatohepatitis, and fibrosis, offering potential insights into disease mechanisms for primary diseases.

The molecular diagnosis of rejection in liver transplant biopsies: First results of the INTERLIVER study.

Molecular diagnosis of rejection is emerging in kidney, heart, and lung transplant biopsies and could offer insights for liver transplant biopsies. We measured gene expression by microarrays in 235 liver transplant biopsies from 10 centers. Unsupervised archetypal analysis based on expression of previously annotated rejection-related transcripts identified 4 groups: normal "R1normal " (N = 129), T cell-mediated rejection (TCMR) "R2TCMR " (N = 37), early injury "R3injury " (N = 61), and fibrosis "R4late " (N = 8). Groups differed in median time posttransplant, for example, R3injury 99 days vs R4late 3117 days. R2TCMR biopsies expressed typical TCMR-related transcripts, for example, intense IFNG-induced effects. R3injury displayed increased expression of parenchymal injury transcripts (eg, hypoxia-inducible factor EGLN1). R4late biopsies showed immunoglobulin transcripts and injury-related transcripts. R2TCMR correlated with histologic rejection although with many discrepancies, and R4late with fibrosis. R2TCMR , R3injury , and R4late correlated with liver function abnormalities. Supervised classifiers trained on histologic rejection showed less agreement with histology than unsupervised R2TCMR scores. No confirmed cases of clinical antibody-mediated rejection (ABMR) were present in the population, and strategies that previously revealed ABMR in kidney and heart transplants failed to reveal a liver ABMR phenotype. In conclusion, molecular analysis of liver transplant biopsies detects rejection, has the potential to resolve ambiguities, and could assist with immunosuppressive management.