Unbiased and Machine Learning Approaches to the Identification of Baseline Predictors of PASC

Introduction: The goal of this study is to use Systems Immunology and AI to define the interplay between baseline metabolite and microbiome signatures and immune cells to mediate susceptibility to the development of post-acute sequelae of SARS-CoV-2 (PASC) or Long COVID. We will define how transcriptional and epigenetic changes, at the single cell level, mediated by these metabolites and cytokines mechanistically underpin the biology and immunology of Long COVID.

Objective: To define the baseline clinical and immunological features that mechanistically mediate the development of Long COVID.

Methods: In Aim 1, we will use stool, urine and plasma samples to perform shotgun metagenomics and metabolomics to identify changes in microbial community structure and/or function, their associated changes in circulating metabolites (plasma and excreted into urine), and how these regulate immune cell function as measured by flow cytometry and cytokines in plasma. In Aim 2, we will employ state-of-the-art single cell RNA and single cell ATAC sequencing to define the transcriptional and epigenetic changes at baseline which persist and predict development of Long COVID. Throughout, we will utilize next generation machine learning and AI algorithms to model Omic features as drivers of Long COVID and LLMs for causal inference and identification of therapeutic strategies that can target these baseline signatures predicting the development of Long COVID.

Results: In preliminary findings subject to change, we have discovered distinct post-infection cytokine signatures mediating Long COVID for each of the pre-existing co-morbidities of obesity, HIV, and lung disease. A common feature underlying potential susceptibility to Long COVID across all of these co-morbidities is the presence of heightened interferon and inflammatory (IL-1b, IL-6, TNFa) cytokines. Importantly, our longitudinal analysis of samples up to 3 years after initial infection show that these signatures of elevated cytokines persist in individuals, suggesting they may result from baseline changes in immune function which we seek to confirm and validate in the ongoing studies in this proposal.

Conclusion/Discussion: We have generated strong evidence that pre-existing co-morbidities shape the cytokine profile in Long COVID which can contribute mechanistically to symptom development. We will now expand on these post-infection findings by looking for baseline multi-Omic signatures that predict susceptibility to developing Long COVID.

Key Topics:

• Assay and in vitro studies to gain mechanistic insights
• Chronic immune dysfunction
• Collaborative and systems biology approaches
• Long COVID and other chronic conditions
• Therapeutic targets and drug development