Share This Page
More than 100 million people in the US contend with chronic pain, according to the Institute of Medicine. Many will seek relief by trying different drugs, but finding the right treatment can often be a challenge. Over the past decade, researchers have used a variety of neuroimaging techniques to try to better understand the brain’s role in pain. A new study from University of Oxford suggests that neuroimaging could also be used to determine whether a particular analgesic, or pain relief, treatment will be effective for a particular person.
A guessing game
It can be difficult for clinicians to know what drugs or other treatment to prescribe when a patient comes into the office complaining of chronic pain, says Tor Wager, a neuroscientist who studies pain at the University of Colorado at Boulder.
“We know a fair amount clinically about how opiates work. But you don’t know which drug is going to work for which person,” he says. “You try it—sometimes it works and sometimes it doesn’t. And why it works, why it doesn’t, is kind of hidden.”
Neuroimaging techniques might uncover some of that hidden information to help better direct prescriptions. David Borsook, a professor of Anesthesiology at Harvard Medical School, has been calling for expanded use of functional magnetic resonance imaging (fMRI) in clinical trials of pain for more than a decade.
“We use medications for chronic pain and, for the most part, we have no idea how they modify brain systems that are involved in the pain phenotype in a good or bad way,” he says. “Imaging techniques allow us to measure the invisible, so to speak—so we can understand the correlation of function, structure, and feeling. And as we learn more about how pain and pain drugs act on certain circuits, we have a chance to predict how effective a drug may be.”
Wager agrees. “Pain is a complex issue. But it is a sensation, in part, that is created in your brain. It’s an interpretation of signals coming up from your body—and a very dynamic process.” That’s why, he says, neuroimaging techniques can not only assist in helping us to understand the neural signature of pain but also of its relief.
Decoding the patterns
Newly developed drugs have a high failure rate, says Eugene Duff, a researcher from the University of Oxford in the United Kingdom.
“This has been a major problem, particular when it comes to pain drugs. Well over fifty percent of drugs fail in Stage 2 clinical trials,” he says. “Human clinical trials are very expensive. And because of that high failure rate, clinical trials can become a big bottleneck in the drug development process. So we thought that there might be a way to use fMRI to give us a bit more confidence about which new drugs will be successful.”
Single neuroimaging studies, with their limited samples, may not offer the best strategy to give us that confidence, though, he says. Data-mining, or the practice of looking for insights across a large number of studies, however, just might.
Duff and his colleagues used a machine-learning technique (training a computer program to better classify results by adding more and more data), to analyze the fMRI data from eight separate clinical trials testing the efficacy of different analgesic medications. The program looked for patterns of activity underlying the associated pain relief. They found distinct changes to the anterior cingulate cortices, the pain-sensitive regions of the brain, in patients who received a drug versus a placebo.
“The idea of the study was not to target a particular brain region in isolation. The idea was to find an overall pattern that was as optimally predictive of the presence of an analgesic versus placebo as possible,” he says. “We were able to generate representations of this pattern which correspond very well to how the brain changes with modulation of pain.”
In the future, scientists could use fMRI to look for this pain relief pattern when testing newly developed analgesic compounds before sending them to clinical trial. If present, it might suggest that a drug will be effective before considerable time and money is spent, reducing that overall drug failure rate.
Pain and beyond
Duff says that this technique is a good proof of concept—and this kind of data mining also might be used in the future to see whether future drugs are working effectively.
“This reveals a new direction, where we can combine and integrate a series of studies, rather than just look to one study, and then look for distinct patterns of activity,” he says. “By looking at consistencies across a large class of drugs we can get more confident that the signatures we’re identifying are predictive of pain relief for new compounds.”
Borsook agrees: He says it shows the utility of imaging to the drug development process—and not just for pain drug development. It might also work for other treatments for other brain-related diseases and disorders. To that end, Duff’s group plans to expand the range of conditions they are studying.
“We’ve become more interested in looking at the resting state networks—and how they may vary in different conditions as well as how they may be modulated by different drugs,” he says. “By combining many studies that aren’t too different, we may be able to probe and look for general signatures that tell us whether a drug is or isn’t working. And that might help us develop new treatments, and get them to the people who need them, that much faster.”