¹McGill University, Integrated Program in Neuroscience, Montreal, Canada, ²McGill University, Psychology, Montreal, Canada, ³McGill University, Psychiatry, Montreal, Canada, ⁴Université de Montréal, Départment de Psychiatrie et d’Addictologie, Montreal, Canada, ⁵McGill University, McConnell Brain Imaging Centre, Montreal, Canada, ⁶Université de Montréal, School of Psychoeducation, Montreal, Canada, ⁷Université de Québec à Montréal, Psychology, Montreal, Canada, ⁸Université de Laval, Psychology, Quebec City, Canada, ⁹Université de Montréal, Department of Social and Preventative Medicine, Montreal, Canada

Background: One of the best-replicated findings in biological psychiatry is that people with substance use disorders (SUDs) have low dopamine D2/3 receptor availability. It remains controversial, however, whether this is a pre-existing vulnerability trait or an effect of heavy drug use.

Aims & Objectives: To address the above question, we conducted repeat positron emission tomography (PET) scans in young adults before and after they developed an SUD.

Method: PET scans with the high-affinity D2/3 receptor ligand, [¹⁸F]fallypride, were conducted in longitudinally followed youth at ages 18 (Wave-1, n=58) and 25 (Wave-2, n=47). Binding potential (BPND) values were derived for 12 striatal and extrastriatal regions of interest (ROIs). Linear mixed-effects models examined effects of Group (Emerging SUD vs Control), Time, and ROI, including all two- and three-way interactions. Separate mixed-effects models decomposed substance use into within-person change (Wave-2 minus Wave-1) and between-person averages across waves to predict changes in BPND. Post hoc comparisons and correlations were Bonferroni- and Benjamini–Hochberg-corrected, respectively. Since the PET scans were 90 minutes, we confirmed that derived striatal BPND values corrected highly with those from 180-min scans (r ≥ 0.86, p < 0.0001).

Results: As reflected by a Group x Wave x ROI interaction (p < 0.001), Wave-1 (age 18) BPND values were higher in the Emerging SUD than Control group in the ventral (VS; p = 0.02), associative (AS; p = 0.050), and sensorimotor striatum (SMS; p < .001). At Wave-2 (age 25), BPND values were lower in the Emerging SUD than Control group in the AS (p < 0.001) and SMS (p < 0.001). During the intervening seven years, both groups exhibited decreases in BPND values, but the magnitude of these decreases was greater in those who developed an SUD in the AS (p < 0.001), SMS (p < 0.001), and VS (p < 0.001). Similar group differences were observed for participants who developed any DSM-5 disorder, with significantly greater Wave-1 to Wave-2 BPND declines in the AS (p = 0.003) and SMS (p < 0.001). Alcohol was the most commonly used substance, alcohol use disorder was the most common SUD, and individual differences in lifetime alcohol use occasions correlated negatively with Wave-2 BPND values (AS: r = –0.363, p = 0.017; SMS: r = –0.332, p = 0.030). Greater increases in alcohol use from Wave-1 to Wave-2 predicted larger decreases in striatal BPND (VS: β = −0.004, p < 0.001; AS: β = −0.003, p < 0.001; SMS: β = −0.004, p < 0.001), and heavier drinkers on average had lower BPND (AS: β = −0.001, p = 0.033; SMS: β = −0.003, p < 0.001). None of these effects were influenced by sex or driven by stressful life events (all p > 0.221).

Discussion & Conclusions: This study provides the first longitudinal evidence in humans that striatal D2/3 receptor availability is elevated in high-risk participants prior to the onset of an SUD (and other DSM-5 disorders) and then declines in those who develop a disorder, plausibly reflecting compensatory responses to repeated drug-induced dopamine surges.