Multidimensional biomarkers for a deep-characterizing of long COVID in the brain

• Grant number: 24/09944-6
• Support Opportunities: Regular Research Grants
• Field of knowledge: Biological Sciences - Biochemistry - Metabolism and Bioenergetics
• Principal Investigator: Artur Martins Novaes Coutinho
• Grantee: Artur Martins Novaes Coutinho
• Host Institution: Faculdade de Medicina (FM). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Associated researchers: Carlos Alberto Buchpiguel ; Daniele de Paula Faria ; Débora Guerini de Souza

Abstract

The coronavirus disease 2019 (COVID-19) is a health condition with consequences that are not fully understood. It can trigger chronic inflammation and affect organs such as the brain. It has been reported that many individuals affected by the disease have persistent neurological manifestations, symptoms that are part of a condition called "long COVID." Long COVID can include fatigue, inattention, lack of concentration, memory complaints, lack of motivation, anxiety, and depressive symptoms. However, the neurobiological basis of these sequelae has yet to be elucidated. To fill this gap in current knowledge about COVID, we will evaluate biological data obtained by multimodal medical examinations in individuals with moderate or severe post-COVID-19 neurological manifestations. This project will include participants from the "NeuroCovid" Brazil cohort, established in 2023 by the proponents, including individuals of both sexes with or without neurological manifestations (Long COVID). Individuals will be assessed using clinical and cognitive evaluations, structural neuroimaging (magnetic resonance imaging - MRI), and molecular imaging/PET with the radiopharmaceutical [11C]acetate, an index of brain inflammation. Participants will also have blood taken to assess plasma proteins associated with inflammation and brain damage. This data will be used for deep-phenotyping. We will use voxel-level neuroimaging data and concentrations of 370 plasma proteins to train an artificial intelligence model capable of predicting and identifying neurological manifestations of long COVID. To do this, we will use a machine learning platform recently developed by the group, which combines multiple dimensionality reduction techniques and models to predict and identify neurological damage associated with long COVID using data from PET neuroimaging and proteomics. We aim to advance our understanding of the pathophysiological basis of the brain associated with long COVID and thus open up new perspectives for treating and managing such prevalent neurological manifestations. (AU)