The risk of cannabis use disorder is mediated by altered brain connectivity: A chronnectome study

Giovanni Fazio, Daniele Olivo, Nadine D. Wolf, Dušan Hirjak, Mike M. Schmitgen, Florian Werler, Miriam Witteman, Katharina M. Kubera, Vince D. Calhoun, Wolfgang Reith, Robert Christian Wolf, Fabio SambataroAddiction 2024 · DOI

TL;DR

This study reveals that individuals at risk for Cannabis Use Disorder (CUD) exhibit altered dynamic functional connectivity (dFNC) in their brains, characterized by increased fluidity and a tendency to remain in specific network states for longer periods. These findings suggest that dFNC could serve as an early biomarker for CUD risk, potentially paving the way for targeted interventions.

What Problem Does This Paper Address?

Cannabis Use Disorder (CUD) is a growing public health concern, yet the underlying brain mechanisms that predispose individuals to developing this disorder remain poorly understood. While previous research has identified some static differences in brain connectivity in individuals with CUD, these studies often overlook the dynamic, time-varying nature of brain activity. The brain is not a static entity; its functional connections constantly shift and reconfigure to process information. Understanding these dynamic changes is crucial for a more complete picture of how CUD develops. This paper aims to fill this gap by investigating dynamic functional connectivity (dFNC) in individuals at risk for CUD, exploring how their brain networks interact and change over time.

Methods

The researchers employed resting-state functional Magnetic Resonance Imaging (rs-fMRI) to capture brain activity while participants were at rest. To analyze the dynamic nature of brain connectivity, they utilized a novel approach called dynamic functional connectivity (dFNC). This involved:

  • Spatial Group Independent Component Analysis (ICA): This technique was used to decompose the rs-fMRI data into 53 intrinsic connectivity networks (ICNs) that represent distinct functional brain systems.
  • Sliding Window Analysis: The time series of these ICNs were then analyzed using a sliding window approach. This allowed the researchers to capture how functional connectivity (FC) between these networks changed over short periods.
  • Meta-state Analysis: This method modeled the dFNC as a weighted sum of "prototype connectivity patterns" (CPs). By discretizing the time courses of these CPs, they could identify different "meta-states" of brain connectivity. Key metrics derived from this analysis included the number of meta-states, the frequency of transitions between them, the "state span" (the range of connectivity states visited), and the "total distance" traveled within the state space.
  • Cluster States Analysis: Complementary to the meta-state analysis, this approach used k-means clustering to identify recurring patterns of functional connectivity (FC) across the sliding windows. This allowed them to define distinct "cluster states" of brain network organization and measure how long participants spent in each state (mean dwell time).

Datasets

The study utilized rs-fMRI data from two groups:

  • At-risk individuals for CUD (n=39): Identified using the Cannabis Use Disorder Identification Test (CUDIT-R) with a score of 8 or higher.
  • Control group (n=55): Individuals without a high risk for CUD.

All participants were screened for current or past mental disorders and substance use (other than cannabis and nicotine) to ensure a relatively clean sample.

Key Insights and Ideas

  • Increased Brain Network Fluidity in At-Risk Individuals: The most striking finding is that individuals at risk for CUD exhibit a significantly more dynamic brain connectivity. They transition between different network states more frequently and explore a wider range of connectivity patterns.
  • Prolonged Engagement in Specific Network States: While their overall connectivity is more fluid, at-risk individuals also tend to spend longer periods in specific "cluster states." This suggests a potential difficulty in shifting away from certain network configurations.
  • Subcortical and Cognitive Network Involvement: The prolonged dwell time in a particular cluster state was associated with higher within-network coupling in key areas like the subcortical, sensorimotor, visual, cognitive-control, and default-mode networks. This points to alterations in the interplay between regions crucial for reward processing, executive functions, and emotional regulation.
  • Potential for Early Biomarkers: The observed dFNC alterations appear early in the risk phase of CUD, suggesting that these dynamic connectivity patterns could serve as valuable biomarkers for identifying individuals at high risk before they develop full-blown disorder.

Key Results

  • Meta-state Analysis: Individuals at risk for CUD showed significantly increased "number of states," "change between states," "state span," and "total distance" compared to controls. These increases were robust even after accounting for psychopathological differences (e.g., depression, anxiety, ADHD).
  • Cluster State Analysis: At-risk individuals spent significantly longer in "State 3" compared to controls. This state was characterized by high within-network coupling in several key brain domains and reduced between-network interactions.
  • Correlations: Notably, the meta-state dynamic indices were not significantly correlated with psychometric scores or cannabis use patterns, suggesting that these dFNC changes are distinct from these factors.

Limitations and Future Work

The authors acknowledge several limitations:

  • Small Sample Size: The relatively small sample size might limit the generalizability of the findings and the detection of more subtle associations. Larger, more diverse samples are needed for replication.
  • Cross-Sectional Design: The study's cross-sectional nature prevents the evaluation of risk trajectories. Longitudinal studies are crucial to establish whether these dFNC alterations precede and predict the development of CUD.
  • Potential Confounding Factors: While efforts were made to control for anxiety and ADHD, the authors note that higher scores in these areas in the at-risk group could have partially influenced the results.

Future work should focus on longitudinal studies to confirm the predictive value of dFNC for CUD development. Additionally, exploring how dFNC can be used to stratify individuals at risk and as biomarkers for preclinical pharmacological trials to test treatment efficacy is a promising avenue.

Summary updated

Abstract

The brain mechanisms underlying the risk of cannabis use disorder (CUD) are poorly understood. Several studies have reported changes in functional connectivity (FC) in CUD, although none have focused on the study of time-varying patterns of FC. To fill this important gap of knowledge, 39 individuals at risk for CUD and 55 controls, stratified by their score on a self-screening questionnaire for cannabis-related problems (CUDIT-R), underwent resting-state functional magnetic resonance imaging. Dynamic functional connectivity (dFNC) was estimated using independent component analysis, sliding-time window correlations, cluster states and meta-state indices of global dynamics and were compared among groups. At-risk individuals stayed longer in a cluster state with higher within and reduced between network dFNC for the subcortical, sensory-motor, visual, cognitive-control and default-mode networks, relative to controls. More globally, at-risk individuals had a greater number of meta-states and transitions between them and a longer state span and total distance between meta-states in the state space. Our findings suggest that the risk of CUD is associated with an increased dynamic fluidity and dynamic range of FC. This may result in altered stability and engagement of the brain networks, which can ultimately translate into altered cortical and subcortical function conveying CUD risk. Identifying these changes in brain function can pave the way for early pharmacological and neurostimulation treatment of CUD, as much as they could facilitate the stratification of high-risk individuals.
BibTeX
@inproceedings{fazio2024risk,
  author = {Giovanni Fazio and Daniele Olivo and Nadine D. Wolf and Dušan Hirjak and Mike M. Schmitgen and Florian Werler and Miriam Witteman and Katharina M. Kubera and Vince D. Calhoun and Wolfgang Reith and Robert Christian Wolf and Fabio Sambataro},
  title = {{The risk of cannabis use disorder is mediated by altered brain connectivity: A chronnectome study}},
  booktitle = {Addiction},
  year = {2024},
  doi = {10.1111/adb.13395},
}
Figure