Brain activity classifies adolescents with and without a familial history of substance use disorders

Front Hum Neurosci. 2015 Apr 22;9:219. doi: 10.3389/fnhum.2015.00219. eCollection 2015.


We aimed to uncover differences in brain circuits of adolescents with parental positive or negative histories of substance use disorders (SUD), when performing a task that elicits emotional conflict, testing whether the brain circuits could serve as endophenotype markers to distinguish these adolescents. We acquired functional magnetic resonance imaging data from 11 adolescents with a positive familial history of SUD (FH+ group) and seven adolescents with a negative familial history of SUD (FH- group) when performing an emotional stroop task. We extracted brain features from the conflict-related contrast images in group level analyses and granger causality indices (GCIs) that measure the causal interactions among regions. Support vector machine (SVM) was applied to classify the FH+ and FH- adolescents. Adolescents with FH+ showed greater activity and weaker connectivity related to emotional conflict, decision making and reward system including anterior cingulate cortex (ACC), prefrontal cortex (PFC), and ventral tegmental area (VTA). High classification accuracies were achieved with leave-one-out cross validation (89.75% for the maximum conflict, 96.71% when combining maximum conflict and general conflict contrast, 97.28% when combining activity of the two contrasts and GCIs). Individual contributions of the brain features to the classification were further investigated, indicating that activation in PFC, ACC, VTA and effective connectivity from PFC to ACC play the most important roles. We concluded that fundamental differences of neural substrates underlying cognitive behaviors of adolescents with parental positive or negative histories of SUD provide new insight into potential neurobiological mechanisms contributing to the elevated risk of FH+ individuals for developing SUD.

Keywords: brain connectivity; emotional conflict; fMRI; family history; machine learning; risk; substance use disorders.