Age-Modulated Immuno-Metabolic Proteome Profiles of Deceased Donor Kidneys Predict 12-Month Posttransplant Outcome

Abstract

ABSTRACTBackgroundOrgan availability limits kidney transplantation, the best treatment for end-stage kidney disease. Deceased donor acceptance criteria have been relaxed to include older donors with higher risk of inferior posttransplant outcomes. Donor age, although significantly correlates with transplant outcomes, lacks granularity in predicting graft dysfunction. Better characterization of the biological mechanisms associated with deceased donor organ damage and specifically predictive of transplant outcome in recipients is key to developing new assessment criteria for donor kidneys and developing function-preserving interventions.Methods185 deceased donor pretransplant biopsies with clinical and demographic donor and recipient metadata were obtained from the Quality in Organ Donation biobank (QUOD), selected on the basis of 12-month paired posttransplant function and deep proteomic profiles acquired by mass spectrometry. Using a 2/3rd:1/3rdtraining:test data split, sampling equally across posttransplant function, we applied machine learning feature selection followed by protein-wise relaxed LASSO regression modeling, assessing the performance of the final set of protein models on the test data. Western blotting validated protein changes, and the biological relevance of the final set of protein models was externally validated by contextualization against a published dataset of human healthy and disease kidney transcriptomes.ResultsOur analysis revealed 144 proteins carrying outcome-predictive information, all of which showed donor-age modulated associations with posttransplant function, as opposed to age and protein/gene effects being independent terms. Observed associations with inflammatory, metabolic, protein processing and cell cycle pathways suggest biological targets for possible interventions pretransplant. Contextualization of our results against external spatial transcriptomic data suggest a sub-nephrotic spatial localization of the predictive signal.ConclusionsIntegrating kidney proteome information with clinical metadata enhances the resolution of donor kidney quality stratification, and the highlighted biological mechanisms open new research directions in developing predictive models and novel interventions during donor management or preservation to improve kidney transplant outcome.SIGNIFICANCE STATEMENTCurrently, organ quality assessment pretransplant relies on key factors such as donor age or clinical information, these lack granularity in depicting graft susceptibility and capacity to function posttransplant. A high-resolution proteomic profiling of 185 pretransplant biopsies of kidneys with known posttransplant function and complete metadata was performed. Integration of donor kidney proteomes with 56 clinical metadata variables using regularized regression modelling resulted in enhancing the resolution of donor kidney quality stratification. Immunometabolic and catabolic processes contributed to donor kidney susceptibility and worse transplant outcomes in an age modulated pattern, validated by western blotting. Comparison of kidney proteomes with a recent transcriptomics dataset of healthy and diseased kidneys provide an additional special single cell resolution to the findings of this study.