1 Integrated Program in Neuroscience, McGill University, Montreal, Quebec, Canada.
2 Department of Psychology, McGill University, Montreal, Quebec, Canada.
3 Department of Psychiatry, McGill University, Montreal, Quebec, Canada.
4 Department of Cellular and Molecular Medicine, University of Ottawa, Ottawa, Ontario, Canada.
5 University of Ottawa Institute of Mental Health Research, Ottawa, Ontario, Canada.
6 School of Psychoeducation, Université de Montréal, Montreal, Quebec, Canada.
7 Azrieli Research Center, Centre Hospitalier Universitaire Sainte-Justine, Montreal, Quebec, Canada.
8 Department of Psychology, Université de Québec à Montréal, Montreal, Quebec, Canada.
9 Department of Psychology, Université Laval, Quebec City, Quebec, Canada.
10 Institute of Genetic, Neurobiological and Social Foundations of Child Development, Tomsk State University, Tomsk, Russia.
11 Department of Social and Preventative Medicine, Université de Montréal, Montreal, Quebec, Canada.
12 Departments of Pediatrics and Psychology, Université de Montréal, Montreal, Quebec, Canada.
13 School of Public Health and Sports Science, University College Dublin, Dublin, Ireland.
14 Institut National de la Santé et de la Recherche Médicale, U669, Paris, France.
15 Department of Psychiatry and Addictology, Université de Montréal, Montreal, Quebec, Canada.
16 Department of Neurology and Neurosurgery, Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada.
17 Center for Studies in Behavioral Neurobiology, Concordia University, Montreal, Quebec, Canada.

Background: We recently reported evidence that a biopsychosocial model predicts lifetime histories of commonly comorbid early-onset psychiatric disorders. This transdiagnostic model was produced when incorporating either positron emission tomography-measured midbrain dopamine autoreceptors or functional magnetic resonance imaging (fMRI)-measured mesocorticolimbic responses during a monetary incentive delay task. Here, we tested whether a variant of this model incorporating brain responses to alcohol cues specifically predicts future alcohol problems.

Methods: Forty-four youths were assessed longitudinally for externalizing (EXT) behaviors between ages 10 and 16 years, completed fMRI scans and the Childhood Trauma Questionnaire (CTQ) at age 18, and completed the Alcohol Use Disorders Identification Test (AUDIT) at age 25. Binomial logistic regressions were then run with each region of interest.

Results: As hypothesized, the combination of high mesocorticolimbic responses to alcohol cues, high adolescent EXT behaviors, and high CTQ scores predicted higher AUDIT scores 7 years later. Model performance was improved by adding sex as a fourth factor. Significant mesocorticolimbic contributors to the models included the ventral (p = .028) and associative striatum (p = .017), anterior (p = .029) and posterior cingulate (p = .005), and ventromedial prefrontal cortex (p = .030). All models exhibited good-to-excellent classification performance, with area under the curve values ranging from 0.83 to 0.87, predictive accuracy from 80% to 84%, sensitivity from 74% to 79%, and specificity from 84% to 92%.

Conclusions: Together, this work delineates a novel hierarchical model with precision psychiatry factors superimposed on transdiagnostic liability. High mesocorticolimbic reactivity to specific stimuli might shape the expression of psychopathology.