Connectome-based predictive modelling of smoking severity in smokers.

Abstract

The functional connectivity within and between networks could provide a framework to characterize the neurobiological mechanism of nicotine addiction. This study examined the brain regions that were functionally connected in response to smoking cues and established the brain-behaviour relationships in smokers. Sixty-seven male smokers were enrolled and scanned while performing the cue-reactivity and Stroop task. A whole-brain analysis approach, connectome-based predictive modelling (CPM), was conducted on the data from the cue-reactivity task to identify the networks that could predict the smoking severity with the Shen atlas as templates. Then, the brain-behaviour relationships were verified in a different brain state (Stroop task). CPM identified the smoking severity-related network, as indicated by a significant correlation between predicted and actual smoking severity scores (r = 0.31, p = 0.02). Identified networks mainly involved the canonical networks implicated in the reward process (motor/sensory network and salience network) and executive control (frontoparietal network). Network strength in the Stroop task marginally significantly predicted smoking severity scores (r= 0.23, p= 0.06), partially replicating the brain-behaviour relationship. The CPM results identified the whole-brain neural network related to smoking severity, which was cross-validated by the AAL and Shen atlas. These findings contribute to more profound insights into neural substrates underlying the smoking severity.