Adverse Effects of Cannabis on Adolescent Brain Development: A Longitudinal Study

Abstract

Cannabis is widely perceived as a safe recreational drug and its use is increasing in youth. It is important to understand the implications of cannabis use during childhood and adolescence on brain development. This is the first longitudinal study that compared resting functional connectivity of frontally mediated networks between 43 healthy controls (HCs; 20 females; age M = 16.5 ± 2.7) and 22 treatment-seeking adolescents with cannabis use disorder (CUD; 8 females; age M = 17.6 ± 2.4). Increases in resting functional connectivity between caudal anterior cingulate cortex (ACC) and superior frontal gyrus across time were found in HC, but not in CUD. CUD showed a decrease in functional connectivity between caudal ACC and dorsolateral and orbitofrontal cortices across time. Lower functional connectivity between caudal ACC cortex and orbitofrontal cortex at baseline predicted higher amounts of cannabis use during the following 18 months. Finally, high amounts of cannabis use during the 18-month interval predicted lower intelligence quotient and slower cognitive function measured at follow-up. These data provide compelling longitudinal evidence suggesting that repeated exposure to cannabis during adolescence may have detrimental effects on brain resting functional connectivity, intelligence, and cognitive function.

Keywords: cannabis; cognition; development; functional connectivity; longitudinal.

Figure 1.

(A) Three-dimensional MNI brain with slices cut at x = −12 y = 9, z = 43 showing regions with lower functional connectivity in adolescents with CUD than healthy controls HC: (1) caudal ACC (red cluster) and (2) left DLPFC (BA 9; green cluster). (B) Line graph showing trajectory of functional connectivity from Time 1 scan to Time 2 scan for the CUD (purple line) and HC (green line) groups. CUD had significant decrease in functional connectivity across time (P = 0.040), HC did not. CUD showed significantly lower functional connectivity than HC at Time 2 (P = 0.016).

Figure 2.

(A) Three-dimensional MNI brain with slices cut at x = −12 y = 9, z = 11 showing regions with lower functional connectivity in adolescents with CUD than HC: (1) caudal ACC (red cluster) and (2) left SFG (BA 10; green cluster). (B) Line graph showing trajectory of functional connectivity from Time 1 scan to Time 2 scan for the CUD (purple line) and HC (green line) groups. HC had significant increase in functional connectivity across time (P = 0.003), CUD did not. CUD showed significantly lower functional connectivity than HC at Time 2 (P = 0.006).

Figure 3.

(A) Three-dimensional MNI brain with slices cut at x = −12 y = 9, z = −16 showing regions with lower functional connectivity in adolescents with cannabis use disorder (CUD) than HC: (1) caudal ACC (red cluster) and (2) left OFC (BA 11; green cluster). (B) Line graph showing trajectory of functional connectivity from Time 1 scan to Time 2 scan for the CUD (purple line) and HC (green line) groups. CUD showed significantly lower functional connectivity than HC at Time 2 (P = 0.006).

Figure 4.

Partial regression plot. Values in x and y axes are residuals that illustrate the relationship between Caudal ACC–OFC functional connectivity at Time 1 and cannabis exposure during the scanning interval (18 months) after removing the linear effects of other independent variables in the hierarchical linear regression model: lifetime substance use (cannabis and alcohol) up to Time 1 (step 1), age at Time 1 (step 2), and functional connectivity between Caudal ACC and DLPFC and SFG (step 3). ACC, anterior cingulate cortex; functional connectivity, functional connectivity; DLPFC, dorsolateral prefrontal cortex; SFG, superior frontal gyrus.

Figure 5.

Partial regression plots. Values in x and y axes are residuals that illustrate the relationship between cannabis exposure during the scanning interval (18 months) and (A) full-scale IQ and (B) reaction time (RT) in the ANT at Time 2 after removing the linear effects of other independent variables in the hierarchical linear regression model: corresponding Time 1 measure of IQ or executive functioning (step 1) and age at Time 1 (step 2). IQ, intelligence quotient; ANT, attention network task.