Researchers have identified a robust functional connectivity brain signature that can accurately characterize neural circuit abnormalities in cocaine use disorder (CUD). The signature can, in turn, be used to predict the response of CUD patients to rTMS (repetitive transcranial magnetic therapy), a form of non-invasive brain stimulation therapy.

CUD, characterized by compulsive cocaine use, affects over one million Americans. While talk-based therapies for the disorder do exist, CUD patients have a high rate of relapse. This has motivated many researchers to search for improved treatment approaches.

Different kinds of brain scans have been used in past studies to discover how the brain differs in people who do and do not have cocaine use disorder. Structural MRI has found reduced grey-matter volume in several parts of the cerebral cortex. (The brain’s grey matter consists mainly of neuronal cell bodies, while white matter refers to axons that connect them.) Another kind of scan, measuring cumulative neural activity throughout the brain, has revealed certain patterns in people who are craving cocaine, relative to healthy controls: stronger activation in a brain region called the left DLPFC (dorsolateral prefrontal cortex) and bilateral occipital cortex.

The new study, led by Drs. Yu Zhang and Kanhao Zhao of Lehigh University, and reported in Nature Mental Health, used a third kind of brain imaging that measures functional connectivity (FC)—the way neurons connect with one another across brain regions under specific conditions. FC imaging can be a more sensitive indicator of brain dysfunction compared with scans that reveal brain structure and cumulative neuronal activity.

To date, functional MRI scans have succeeded in establishing biomarkers for neuropsychiatric disorders in several instances—signatures that characterize a particular illness or subset of patients, and/or their response to specific treatments. One example noted by Drs. Zhang, Zhao and colleagues is the signature of elevated activity in presynaptic dopamine neurons in the brain’s striatum, which has been shown to predict antipsychotic treatment response in schizophrenia. Another important example is the observation that increased functional connectivity within the brain’s default mode network (DMN)—a network active when the brain is not focused on a particular task—is an indicator of depression. rTMS brain stimulation treatment has been observed to result in a lowering of connectivity in the DMN and a corresponding reduction in depression symptoms in many depressed patients.

Could the team find a similar signal for cocaine use disorder?  The team, which included three recipients of BBRF grants—2019 BBRF Young Investigator Gregory A. Fonzo, Ph.D., 2016 BBRF Young Investigator Desmond J. Oathes, Ph.D., and 2012 BBRF Young Investigator Amit Etkin, M.D., Ph.D.—used machine learning to “train” a computer on functional connectivity data from CUD patients and controls. FC features were extracted from resting-state fMRI scans of a “discovery” group comprising 71 people with CUD and 58 healthy controls. Once they identified a signature, the team retested it within the participant groups and then tried to validate it in an independent cohort consisting of 81 CUD patients and 82 healthy controls. Finally, in a group of 45 CUD patients who were randomly assigned to receive either active rTMS treatment or a “sham” (i.e., placebo) version of it, the researchers sought to determine whether the FC signature for CUD that they had identified was useful in predicting treatment response.

The team succeeded in identifying a functional connectivity signature that was accurate, sensitive, and specific for discriminating people with CUD from healthy controls. This means that the signature generated very few false positives or negatives, a strong indication of its potential utility. They also succeeded in replicating the result in the independent cohort of CUD patients, “making this study a significant milestone,” the team said, in finding a CUD biomarker based on neural circuit abnormalities.

Specifically, the signature of CUD that emerged provided evidence of network-level abnormalities involving the frontoparietal control network (FPC), the default mode network ( DMN), the dorsal attention network (DAN), and the ventral attention network (VAN). The result was consistent with past studies suggesting hyperconnectivity in CUD patients in brain regions involved in response inhibition, as well as with the hypothesis that chronic cocaine use is associated with a decline in inhibitory control within the cortex.  Among other things, the signature involved reduced connectivity between the DMN and the limbic network (LIM), which in turn appeared to be related to cognitive complaints and anxiety in CUD patients. The team said the DMN and LIM are therefore worthy of investigation as potential treatment targets in future research. The limbic system’s hyperactivity in CUD might be “counteracted by enhancing FPC activity, which tends to be pathologically suppressed in individuals with CUD,” the team said. Based on other evidence, they said the results also supported exploration of the medial prefrontal cortex (mPFC) as a treatment target for CUD.

The functional connectivity signature linked with the presence of CUD “was significantly and specifically associated with the response to active—but not sham—rTMS treatment,” the team reported. rTMS has been given experimentally for treatment of cocaine abuse and addiction, but it is not approved by the FDA for that use. The finding that the CUD signature was associated with rTMS treatment response “highlights the role of the prefrontal cortex across the LIM, DMN, and FPC” in reducing craving in CUD patients, the team said.

The researchers looked forward to future studies which might, they said, include attempts to directly manipulate dysfunctional circuits to address cocaine-craving symptoms and to explore whether the circuit dysfunctions captured in the newly reported signature have a causal connection with the disorder.