U.S. researchers have used machine learning to predict the effectiveness of multi-drug treatment combinations for tuberculosis (TB), which could help in the design of new therapy regimens.
By examining study data from TB drug pairs in vitro, they were able to predict how three or four drugs could affect treatment in vivo and work out rules governing drug choices among these pairwise building blocks that would create effective multi-drug therapies.
“Using the design rules we’ve established and tested, we can substitute one drug pair for another drug pair and know with a high degree of confidence that the drug pair should work in concert with the other drug pair to kill the TB bacteria in the rodent model,” explained researcher Bree Aldridge, associate professor of molecular biology and microbiology at Tufts University School of Medicine in Boston, Massachusetts.
“The selection process we developed is both more streamlined and more accurate in predicting success than prior processes, which necessarily considered fewer combinations.”
The research, published in the journal Cell Reports Medicine, follows an earlier study released this week showing that a deep learning program can be as effective as radiologists in identifying tuberculosis on chest X-rays.
Aldridge and co-workers conducted their current research after noting that the interaction of drug pairs predicts the impact of high-order drug combinations in Mycobacterium tuberculosis, Escherichia coli, and cancer cells.
A challenge in the design of tuberculosis treatments is the necessity to combine three or more antibiotics.
While there are nearly 6000 possible combinations involving three and four separate drugs from a collection of 20 overall, this equates to just 190 drug pairs, meaning that pairwise assessment would improve efficiency 30 times over.
The team examined whether the behavior of high-order drug combinations of three or more drugs could be predicted from the underlying low-order combinations by examining study data involving two-drug combinations of 12 drugs aimed against tuberculosis.
Specifically, the team analyzed pairwise drug combination response data from a large-scale study that contained in vitro measurement of two- and three-drug combinations among 10 commonly used anti-TB drugs, which they then expanded with two more drugs.
The 10-drug set included bedaquiline, clofazimine, ethambutol, isoniazid, linezolid, moxifloxacin, pretomanid, pyrazinamide, rifapentine, rifampicin. This was expanded to include SQ109 and sutezolid.
The investigators trained machine learning models to predict the impact of three or four drugs in preclinical models and revealed the principles that drug pairs needed to satisfy in order to be effective within a higher order multi-drug regimen.
From this, they created simple rulesets for guiding the construction of combination therapies, which the researchers describe as “predictive, intuitive, and practical.”
They conclude: “We envision the experimental and computational framework devised in this study, supported by other evaluation methods, can be used early in regimen development to winnow down the thousands of potential combinations to a priority list for further evaluation using [pharmacokinetic/pharmacodynamic] studies and animal testing.”