Identification of Tendency to Alcohol Misuse From the Structural Brain Networks

Abstract

The propensity to engage in risky behaviors including excessive alcohol consumption may impose increased medical, emotional, and psychosocial burdens. Personality and behavioral traits of individuals may contribute in part to the involvement in risky behaviors, and therefore the classification of one’s traits may help identify those who are at risk for future onset of the addictive disorder and related behavioral issues such as alcohol misuse. Personality and behavioral characteristics including impulsivity, anger, reward sensitivity, and avoidance were assessed in a large sample of healthy young adults (n = 475). Participants also underwent diffusion tensor imaging for the analysis of structural brain networks. A data-driven clustering using personality and behavioral traits of the participants identified four subtypes. As compared with individuals clustered into the neutral type, individuals with a high level of impulsivity (A subtype) and those with high levels of reward sensitivity, impulsivity, anger, and avoidance (B subtype) showed significant associations with problem drinking. In contrast, individuals with high levels of impulsivity, anger, and avoidance but not reward sensitivity (C subtype) showed a pattern of social drinking that was similar to those of the neutral subtype. Furthermore, logistic regression analysis with ridge estimators was applied to demonstrate the neurobiological relevance for the identified subtypes according to distinct patterns of structural brain connectivity within the addiction circuitry [neutral vs. A subtype, the area under the receiver operator characteristic curve (AUC) = 0.74, 95% CI = 0.67–0.81; neutral vs. B subtype, AUC = 0.74, 95% CI = 0.66–0.82; neutral vs. C subtype, AUC = 0.77, 95% CI = 0.70–0.84]. The current findings enable the characterization of individuals according to subtypes based on personality and behavioral traits that are also corroborated by neuroimaging data and may provide a platform to better predict individual risks for addictive disorders.