Poster #110
Whole Blood Transcriptomic Signatures as Predictors of Organ Structural Integrity
Mentors: Sanju Sinha, MD and Anamika Yadav
Predicting organ health from blood is an impactful open problem with little progress over the last decades. Two key challenges hinder progress: the lack of paired blood and organ data (since organ sampling is invasive) and the absence of clear definitions for healthy organ states. Using a large post-mortem cohort (GTex) of 970 individuals covering 40 tissues with available histopathology images, Sinha Lab previously developed the Structural Age Gap (SAG)—which quantifies the deviation between an individual’s observed tissue architecture and the expected structure for their chronological age. In this study, using whole blood transcriptomic profiles from the same individuals, we tested the feasibility of predicting Structural Age Gap directly from blood samples. We first computed correlations between the expression of each gene in blood and SAG score across 40 tissues. This analysis identified 12 tissue types with at least 100 genes significantly correlated with SAG scores, exceeding thresholds established by shuffled-label controls. To characterize these genes, we performed two tests: 1) whether these genes are specifically expressed in the tissues where they predict SAG score, and 2) whether specific pathways are enriched in top-ranking genes. We here demonstrate that whole blood expression comprises information for at least 12 tissues structural integrity, setting the stage for machine learning model development.