Search

The association between neurodegenerative diseases (NDs) and obesity has been well studied in recent years. Obesity is a syndrome of multifactorial etiology characterized by an excessive accumulation and release of fatty acids (FA) in adipose and non-adipose tissue. An excess of FA generates a metabolic condition known as lipotoxicity, which triggers pathological cellular and molecular responses, causing dysregulation of homeostasis and a decrease in cell viability. This condition is a hallmark of NDs, and astrocytes are particularly sensitive to it, given their crucial role in energy production and oxidative stress management in the brain. However, analyzing cellular mechanisms associated with these conditions represents a challenge. In this regard, metabolomics is an approach that allows biochemical analysis from the comprehensive perspective of cell physiology. This technique allows cellular metabolic profiles to be determined in different biological contexts, such as those of NDs and specific metabolic insults, including lipotoxicity. Since data provided by metabolomics can be complex and difficult to interpret, alternative data analysis techniques such as machine learning (ML) have grown exponentially in areas related to omics data. Here, we developed an ML model yielding a 93% area under the receiving operating characteristic (ROC) curve, with sensibility and specificity values of 80% and 93%, respectively. This study aimed to analyze the metabolomic profiles of human astrocytes under lipotoxic conditions to provide powerful insights, such as potential biomarkers for scenarios of lipotoxicity induced by palmitic acid (PA). In this work, we propose that dysregulation in seleno-amino acid metabolism, urea cycle, and glutamate metabolism pathways are major triggers in astrocyte lipotoxic scenarios, while increased metabolites such as alanine, adenosine, and glutamate are suggested as potential biomarkers, which, to our knowledge, have not been identified in human astrocytes and are proposed as candidates for further research and validation.