Researchers from MIT, Brigham and Women’s Hospital, and Duke University have developed a novel approach combining machine-learning algorithms and tissue models to identify the specific transporters used by drugs in the gastrointestinal tract. This breakthrough could lead to improvements in patient treatment and drug development.
Transporter proteins within the gastrointestinal system enable drug absorption. These proteins can impair drug efficacy if two medications that rely on the same transporter are administered together. Therefore, determining a drug’s specific transporters could prevent undesirable drug interactions.
This pioneering study led by Giovanni Traverso has disclosed that a frequently used antibiotic and blood thinner could interfere with each other, raising a possibility to make drugs safer and more effective by modeling these interactions. The insights could also inform the development of new drugs with improved absorption rates, achieved by using excipients that enhance their interactions with transporters.
The team used a tissue model, cultured from pig intestinal tissue, to assess drug absorbability. The model systematically exposes tissue to varying drug formulations, thereby measuring their absorption rates. In addition, the researchers manipulated the expression of multiple transporters within the tissue using small RNA strands, allowing intricate insights into how transporters interact with different drugs.
The researchers’ system tested 23 commonly used drugs, identifying the transporters utilized by each. A machine-learning model was then trained on this data and data from drug databases. This model was used to predict drug interactions based on the chemical structure of the drugs.
In an analysis of 28 frequently prescribed drugs and 1,595 experimental drugs, the model made nearly two million potential drug interaction predictions. One prediction of note was that the antibiotic doxycycline could interact with the blood thinner Warfarin, leading to increased Warfarin levels in the bloodstream.
This predictive model will aid in identifying potential interactions between drugs already in use and those in development. Drug developers can use this breakthrough method to modify the formulation of new drugs, preventing interactions or improving absorption rates.
This research was partly funded 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.