Researchers from MIT, Brigham and Women’s Hospital, and Duke University have developed a system using tissue models and machine-learning algorithms to identify how different drugs navigate through the lining of the digestive tract, which could have significant implications for the world of medicine.
Orally-taken drugs often rely on transporter proteins within the digestive tract’s cells to pass through, but it’s not clear which transporters different drugs utilize. If two drugs use the same transporter, they may interfere with each other and thus should not be co-administered.
The research team created an experimental setup using pig intestinal tissue cultivated in a lab, which they used to expose the tissue to varying drug formulations and measure how effectively they were absorbed. Using siRNA, a type of RNA molecule, they reduced the expression of each transporter in different sections of tissue, which enabled them to determine how each transporter interacts with a range of drugs.
The team tested 23 commonly used drugs using this system, allowing them to identify the specific transporters each drug used. Machine-learning was then used to train a model on this data, as well as data from several drug databases, and the model was able to predict which drugs would interact with which transporters based on similarities between chemical structures.
In testing the model, the researchers discovered that doxycycline (an antibiotic) could interact with warfarin (a commonly used blood thinner). Subsequent patient data from Massachusetts General Hospital and Brigham and Women’s Hospital showed that when doxycycline was given to patients already taking warfarin, the level of warfarin in the patients’ bloodstream increased, but then decreased once they finished the doxycycline course.
The team’s system could also identify potential interactions between drugs currently in development. Drug developers could use this knowledge to modify new drug formulations to avoid interactions with other drugs or make them more absorbable.
The findings, funded in part by the U.S. National Institutes of Health, the Department of Mechanical Engineering at MIT, and the Division of Gastroenterology at Brigham and Women’s Hospital, are a sign-post for the medical community. Identifying and predicting drug interactions using this model could significantly improve patient safety and drug effectiveness and decrease the risk of unexpected toxicities.
