Researchers from MIT, Brigham and Women’s Hospital, and Duke University have developed a machine-learning, multipronged strategy to identify which transporter proteins a drug uses to navigate through a patient’s digestive tract. Knowledge in this area is key to improving drug efficacy and patient safety as drugs using the same transporter proteins can interfere with each other. The research has already revealed an interaction between a common antibiotic and a blood-thinner.
The team developed a tissue model to calculate a drug’s absorbability and utilized short strands of RNA to suppress each transporter’s expression. They examined 23 medicines and gathered extensive data to train a machine-learning model to make predictions about which drugs interact with which transporters. The researchers then assessed 28 widely used drugs and 1,595 experimental drugs, providing nearly two million predictions of potential drug interactions. This process identified that doxycycline, an antibiotic, could interact with warfarin, a commonly used blood-thinner. This interaction was confirmed using data from a patient database, which detected elevated warfarin levels in the bloodstream while taking doxycycline, which normalized after ceasing the antibiotic. The model also predicted interactions with other drugs such as digoxin, levetiracetam, and tacrolimus.
The study, while identifying interactions between current drugs, could also be applied during the development phase of new drugs. Drug companies could modify drug formulation to prevent interactions or enhance absorption. Biotech company Vivtex is exploring these possibilities. The research was funded by the National Institutes of Health, MIT’s Department of Mechanical Engineering, and the Division of Gastroenterology at Brigham and Women’s Hospital among other contributors.
