(B-54) Identifying Liver Transplantation Rejection Mechanisms by Integration of Pre- and Post-Operative Donor and Recipient Immunologic Features

Introduction:: Liver ischemia-reperfusion injuries (LIRI) arise during liver transplantation and contribute to the high rate of chronic liver transplant rejection. LIRI occur across the liver transplant process, from pre-operation ischemic injuries during the liver transport to post-operation reperfusion injuries as the recipient’s immune system responds to the donor’s liver. Current metrics of LIRI severity—such as liver biopsy pathology scores—are insufficiently predictive of transplant outcome. While research has identified many types and mechanisms of LIRI, the relationship between LIRI and transplant outcome (donor liver acceptance or rejection) is incompletely understood, precluding development of preventative therapies. This gap in understanding is often attributed to the multitude of potential LIRI sources as the drivers of LIRI vary across the transplant process and manifest across both recipient and donor tissues. Thus, we hypothesize that recognizing mechanisms relevant to transplant rejection likely requires integration of longitudinal donor and recipient measurements reflecting LIRI across the transplant process.


Materials and Methods:: Integrating longitudinal donor and recipient measurements inherently leads to high-dimensional datasets that are difficult to analyze via traditional, matrix-based bioinformatic methods like partial least squares regression (PLS). Matrix-based methods require ‘flattening’ measurements, often by concatenating multiple time-points into a single matrix and obfuscating temporal patterns. Tensors—organized, high-dimensional arrays of data, akin to higher-dimensional versions of matrices—better represent such high-dimensional longitudinal measurements by representing each dimension in the dataset as different dimensions within a tensor (Figure 1C). Here, we apply a form of PLS generalized to tensor-formatted data—tensor partial least squares (tPLS)—to regress transplant outcomes against longitudinal measurements derived from both transplant donor and recipient sources. Longitudinal measurements included peripheral blood cytokine and liver function test (LFT) measurements alongside cytokine measurements derived from the donor portal vein shortly before and after liver transplant surgery (Figure 1A-B).


Results, Conclusions, and Discussions:: Results and Discussion: tPLS successfully integrated pre- and post-transplant measurements collected from donor and recipient. We found that tPLS reduced our dataset to two time-varied components that associate with liver transplant rejection (Figure 2A) and elevated liver damage in the first week post-operation (Figure 2B). These two components were highly predictive of transplant outcomes, predicting transplant rejection with an accuracy of 69% and outperforming models trained on each measurement type alone. Biological interpretation finds that component 1 associates with post-operative Th1 polarization while component 2 associates with pre-operative T_reg suppression (Figures 2C-D). Together, these components highlight that Th1 polarization combined with T_reg suppression led to increased risk of liver transplant rejection through unsuppressed Th1 activation that induces IFN-γ driven liver damage (Figure 2E).

Conclusions: Overall, these results demonstrate that outcome-informed tensor decomposition methods can capture biological mechanisms relevant to liver transplant rejection across host and recipient measurements collected across the transplant process. Interpretation of our results highlights Th1 polarization and immunoregulatory suppression as key determinants in liver transplant rejection. Ultimately, these efforts demonstrate the power of outcome-informed tensor decomposition techniques in capturing immunological signatures and highlight potential therapeutic targets for improving liver transplant outcomes.


Acknowledgements (Optional): : This work was funded by NIH P01AI120944 from the National Institute of Allergy and Infectious Diseases (NIAID).



References (Optional): :