A computer program designed by two scientists could one day be a weapon in the fight against antibiotic resistance in bacteria. Using complex algorithms, biologist Miriam Barlow of the University of California, Merced and mathematician Kristina Crona of American University in Washington, D.C. were able to prevent and actually reverse antibiotic resistance in a lab setting.
The discovery is exciting because antibiotic-resistant infections are some of the most pressing research concerns in modern medicine. These mutated bacteria, such as MRSA, C. diff and more, have the power to turn back the clock on all of the advances we've made since the advent of antibiotics.
To put the problem to scale, at least 2 million people are infected annually with antibiotic-resistant bacteriain the U.S., and 23,000 people die each year as a result, according to the Centers for Disease Control and Prevention. The global threat of antibacterial resistance is expected to get worse over time.
Currently, the way doctors fight drug-resistant infections is to start treating them with the most commonly used antibiotics and then cycle through increasingly rarely used antibiotics. Often, the first few drugs are ineffective and, in a worst-case scenario, doctors may end up using antibiotics that even increase an infection's resistance to medicine.
Here’s how the software test worked: Barlow focused on a single gene in E. coli bacteria that is responsible for making the organism resistant to our arsenal of antibiotics. Together with Crona, she devised a mathematical formula that matched known antibiotics to the bacteria at several stages of the mutation process. The matching process, which is faster and more exacting than a doctor’s intuition, spit out the optimal order of antibiotics to “cycle” through in order to prevent resistance. The matching process worked so well that the researchers were even able to drag E. coli back to its “wild type,” or its natural state before becoming resistant.
The name of their software, fittingly, is the “Time Machine."
"With our treatment plans, the protocol is to find an order of using antibiotics that will take these highly evolved genes conferring resistance to new antibiotics and to push them back to the wild type state, so they’re not conferring antibiotic resistance to our more modern drugs,” Barlow told the Huffington Post. "Without new antibiotics, resistance to the old antibiotics is increasing more and more all of the time, and we need to find ways of making those antibiotics last."
"In medical practice, many times doctors have great ideas because they have a feeling for the problem, based on their experience,” Crona explained. “But [the software] is beyond what a medical doctor can do in terms of his or her intuitions."
Barlow and Crona’s program takes the guesswork out of choosing which known antibiotic to use for E. coli bacteria at a given stage of mutation, bringing a sense of order and a higher chance of effectiveness to infection treatment. But their initial project, published Wednesday in the scientific journal PLOS ONE, is still just a “first step,” said Barlow.
More theoretical and laboratory work needs to be done before testing the protocol in a real-world setting with patients, although Barlow noted that a local hospital near UC Merced has expressed interest in allowing the researchers to analyze the hospital's strains of bacterial infections to see if they’re evolving in response to antibiotics routinely prescribed by doctors on staff.
"If we find strong evidence of that, then we can probably create a protocol similar to the one we’re publishing and present it to the hospital, and then they can choose if they want to incorporate it,” said Barlow.
Dr. Vincent Young, an infectious diseases professor at the University of Michigan Medical School, was not involved in the research, but regularly sees drug resistant infections in his clinical work. He agrees with Barlow and Crona that their research is an intriguing first step in defeating antibiotic resistance, as well as an exciting example of the promise that big data sets hold for medical treatment. However, he cautioned that there were some significant caveats to their work.
For one thing, explained Young, the strain of E. coli used to create their software doesn’t actually cause disease. The resistance gene they were able to influence was the same gene that they inserted into the lab-created strain, so the software was not tested on a bacteria that has a naturally-occurring resistance gene. There are also unanswered questions about whether it could work on other bacteria, and in real-world settings.
“It isn’t known if the same [results] would hold for some of these organisms carrying the resistance gene,” said Young. He also noted that the E. coli grew in isolation in the lab, shielded from the the competition of other bacteria and the pressures of a person’s immune system, so it’s unclear how the bacteria would respond to the antibiotic cycling if it were actually inside a person.
"All this being said, this work is a start in understanding how selection (in this case, the varying use of antibiotics) can influence the development and propagation of antibiotic resistance,” Young concluded. "You need to start with a relatively simple, defined system, study it to find trend and then continue to increase complexity and apply it to other systems to understand how generalizable the findings are."
In its current state, the free "Time Machine" software has the potential of putting the brakes on antibacterial resistance in up to 60 percent of cases, Barlow explained. But in time, with more research on different kinds of mutations in different types of bacteria, as well as tests in clinical trials, she hopes that her algorithms can serve as a template for other researchers to explore and manipulate other genes that cause many different kinds of bacteria to become resistant to antibiotics.
"We’re putting the program onto a user-friendly platform because this is a very mathematical format, and we’re trying to make it easier so that doctors and other researchers would be able to use it,” said Barlow. “Then hopefully other researchers will have an interest in doing this and be able to generate similar data for the resistance genes they research."