Craving is known to be a key factor in substance use disorders and can increase the likelihood of future drug use or relapse. Yet its neural basis—or, how the brain gives rise to craving—is not well understood.
In a new study, researchers from Yale, Dartmouth, and the French National Centre for Scientific Research (CNRS) have identified a stable brain pattern, or neuromarker, for drug and food craving. Their findings were published in Nature Neuroscience.
The discovery may be an important step toward understanding the brain basis of craving, addiction as a brain disorder, and how to better treat addiction in the future, researchers say. Importantly, this neuromarker may also be used to differentiate drug users from non-users, making it not only a neuromarker for craving, but also a potential neuromarker that may one day be used in diagnosis of substance use disorders.
For many diseases there are biological markers that doctors can use to diagnose and treat patients. To diagnose diabetes, for example, physicians test a blood marker called A1C.
“One benefit of having a stable biological indicator for a disease is that you can then give the test to any person and say that they do or do not have that disease,” said Hedy Kober, an associate professor of psychiatry at Yale School of Medicine and an author of the study. “And we don’t have that for psychopathology and certainly not for addiction.”
To determine if such a marker could be established for craving, Kober and her colleagues—Leonie Koban from CRNS and Tor Wager from Dartmouth College—used a machine learning algorithm. Their idea was that if many individuals experiencing similar levels of craving share a pattern of brain activity, then a machine learning algorithm might be able to detect that pattern and use it to predict craving levels based on brain images.
For the study, they used functional magnetic resonance imaging (fMRI) data—which offer insight into brain activity—and self-reported assessments of craving from 99 people to train and test the machine learning algorithm.