Aspartame and thyroid cancer: Unraveling the connection through network toxicology, machine learning, and molecular docking

Abstract Aspartame, a prevalent artificial sweetener, has been implicated as a potential risk factor due to its controversial carcinogenicity. This study sought to explore the potential computationally predicted between aspartame and thyroid cancer and uncover the underlying molecular mechanisms. By integrating network toxicology, machine learning, and molecular docking approaches, we identified aspartame-related targets from multiple databases and thyroid cancer-associated targets from GEO datasets. After intersecting these datasets, 52 overlapping genes were pinpointed as potential mediators computationally linked to aspartame exposure to thyroid cancer. Gene Ontology and KEGG enrichment analyses revealed these genes’ involvement in critical biological processes and pathways, such as protein metabolism, extracellular matrix remodeling, lysosomal activity, and renin-angiotensin signaling. Machine learning models, particularly the Lasso+plsRglm algorithm, were constructed to predict key genes, with SHAP analysis highlighting CTSC, ECE1, F10, HMGCR, and APLNR as pivotal predictive features. Molecular docking further confirmed stable interactions between aspartame and these five proteins, supporting the plausibility of direct compound-target interactions. This research provides novel computational evidence suggesting that aspartame may be associated with thyroid cancer initiation and progression through the regulation of multiple genes and signaling pathways, offering insights into its potential role in thyroid cancer pathogenesis and implications for safety assessment as a food additive.