A multi-institutional team of researchers from MIT, Brigham and Women’s Hospital, and Duke University have developed a strategy that uses machine learning and tissue models to identify transport proteins that different drugs interact with in the gastrointestinal (GI) tract.
When a drug is taken orally, it must pass through the lining of the digestive tract. The specific transport proteins which help the drug exit the digestive system have remained unknown for many medications. Understanding the specific proteins utilized by particular drugs can improve patient treatment, as drugs reliant on the same transporter can interfere with each other and therefore should not be prescribed simultaneously.
An experimental method to gauge a drug’s absorbability was adapted by the researchers for the study based on pig intestinal tissue. Short strands of RNA, known as siRNA, were used to suppress the expression of various transporters. The experimental setup facilitated the exposure of different drug formulations to the tissue to measure absorption.
The researchers examined 23 commonly used drugs and used the generated data to train a machine-learning model. The software was also given data from different drug databases to deduce patterns between the chemical structure of a drug and its interaction with specific transporters.
Using the trained model, the researchers conducted an analysis on an additional 28 widely used drugs as well as 1,595 experimental drugs. The study led to nearly 2 million predictions regarding potential drug interactions. Notably, the researchers discovered potential reactions between commonly used drugs such as the antibiotic doxycycline with the blood thinner warfarin, the heart failure treatment digoxin, antiseizure medication levetiracetam, and immunosuppressant tacrolimus.
To verify their predictions, patient data was analyzed for cases where the patients had been taking the predicted drugs concurrently with doxycycline. The gathered data confirmed that when doxycycline is used in combination with warfarin, the blood concentration of warfarin increased, then decreased once the doxycycline dosage was halted.
The study’s approach can also be applied to drugs currently in development, allowing pharmaceutical developers to adjust the composition of new drug molecules to prevent potential interactions. A biotech company co-founded by members of the research team, Vivtex, is exploring such applications.
The researchers are optimistic that their innovative approach will aid in predicting and preventing harmful drug interactions, effectively improving the safety and effectiveness of many medications. This research was partially supported 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. The findings have been published in Nature Biomedical Engineering.
