Deciphering the Molecular Mechanism of Post-Acute Sequelae of COVID-19 through Comorbidity Network Analysis

Abstract

Introduction: The post-acute sequelae of COVID-19 presents a significant health challenge in the post-pandemic world. Our study aims to analyze longitudinal electronic health records to determine the impact of COVID-19 on disease progression, provide molecular insights into these mechanisms, and identify associated biomarkers. Method: We included 58,710 patients with COVID-19 records from 01/01/2020 to 31/08/2022 and at least one hospital admission before and after the acute phase of COVID-19 (28 days) as the treatment group. A healthy control group of 174,071 individuals was established for comparison using propensity score matching based on pre-existing diseases (before COVID-19). We built a comorbidity network using Pearson correlation coefficient differences between pairs of pre-existing disease and post-infection disease in both groups. Disease-protein mapping and protein-protein interaction network analysis revealed the impact of COVID-19 on disease trajectories through protein interactions in the human body. Results: The disparity in the weight of prevalent disease comorbidity patterns between the treatment and control groups highlights the impact of COVID-19. Certain specific comorbidity patterns show a more pronounced influence by COVID-19. For each comorbidity pattern, overlapping proteins directly associated with pre-existing diseases, post-infection diseases, and COVID-19 help to elucidate the biological mechanism of COVID-19's impact on each comorbidity pattern. Proteins essential for explaining the biological mechanism can be identified based on their weights. Conclusion: Disease comorbidity associations influenced by COVID-19, as identified through longitudinal electronic health records and disease-protein mapping, can help elucidate the biological mechanisms of COVID-19, discover intervention methods, and decode the molecular basis of comorbidity associations. This analysis can also yield potential biomarkers and corresponding treatments for specific disease patterns.