Association of Cannabis Use During Adolescence With Neurodevelopment

Key Points Question To what extent is cannabis use associated with magnetic resonance imaging–measured cerebral cortical thickness development during adolescence? Findings In this cohort study, linear mixed-effects model analysis using 1598 magnetic resonance images from 799 participants revealed that cannabis use was associated with accelerated age-related cortical thinning from 14 to 19 years of age in predominantly prefrontal regions. The spatial pattern of cannabis-related cortical thinning was significantly associated with a positron emission tomography–assessed map of cannabinoid 1 receptor availability. Meaning Results suggest that cannabis use during middle to late adolescence may be associated with altered cerebral cortical development, particularly in regions rich in cannabinoid 1 receptors.


eAppendix 1. MRI Acquisition and Image Processing
MRI scanning was conducted at the eight IMAGEN assessment sites using 3T whole body MRI systems. 1 Image-acquisition utilized parameters that were compatible with all scanners in order to ensure comparability of data across the different scanners. Details surrounding image acquisition protocols and quality checks have been described elsewhere, including extensive standardization across MRI scanner. 1 The CIVET pipeline was used for extraction of cortical surfaces and estimation of local cortical thickness. The following steps were performed as part of this processing pipeline. 2 Native MR images were linearly registered to a standardized MNI-Talairach space based on the ICBM152 dataset. [3][4][5][6] Intensity non-uniformity artifacts were corrected for using the N3 algorithm. 7 Brain extraction was implemented using FSL's Brain Extraction Tool (BET). 8 Classification of white matter (WM), gray matter (GM), and cerebrospinal fluid (CSF) was carried out using the INSECT algorithm. 9,10 The CIVET pipeline includes the CLASP algorithm used to generate high-resolution hemispheric surfaces with 40,962 vertices per hemisphere. [11][12][13][14] Hemispheric surfaces were generated for both the WM/GM interface, as well as the GM/CSF interface. In order to establish correspondence of vertices between subjects, the surfaces for each hemisphere were non-linearly registered to an average surface created from the ICBM152 dataset. 12,15 A reverse linear transformation was performed on each subject's images, allowing for cortical thickness estimations to be made at each cortical point in the MR image's native space. 16 To increase the signal-to-noise ratio, each subject's cortical thickness map was blurred using a 20-millimeter full width at half maximum surface-based diffusion smoothing kernel. 15 This kernel size closely approximates previously recommended values, affording optimal sensitivity for cortical thickness analysis. 17 Following from previous work by members of our group, quality control of CIVET output was performed with regard to (a) registration, (b) surface extraction, and (c) gray-white surface-surface intersections. 18 Absolute cortical thickness measures were also reviewed as extreme values can provide cues of poor surface recognition.

eAppendix 4. Between-Group Analyses
It bears noting that nearly identical results were obtained when primary analyses in present study were rerun using a between-group design (i.e., participants reporting moderate-to-heavy lifetime cannabis use at 5-year follow-up versus those who remained cannabis-naïve). In particular, moderate-to-heavy lifetime cannabis users were defined as participants endorsing 10-19 uses or more (i.e., ≥ 4) on the lifetime cannabis use item at 5-year-follow-up. At 5-year follow-up, participants reporting moderate-to-heavy lifetime cannabis use (n = 161) exhibited reduced cortical thickness in a number of prefrontal areas relative to those who remained cannabis-naïve (n = 430). No significant between-group differences were observed with regard to baseline cortical thickness. Linear mixed-effects model analysis (591 subjects; 1182 MRIs) revealed accelerated age-related cortical thinning among participants who transitioned to moderate-to-heavy lifetime cannabis use relative to those who remained cannabis-naïve. See eFigures 4 and 5, below.

eAppendix 5. Cannabis-Related Thinning and Impulsiveness
Of the 799 participants in the present study, 697 (87.2%) had available Barratt Impulsivity Scale (BIS) data at 5-year follow-up (this measure was not administered at baseline).
Average thickness (at baseline and follow-up) was calculated for each significant cluster in the LMM analysis examining the influence of cannabis use on age-related cortical thinning. For all participants, symmetrized percent change (SPC) (i.e., change in cortical thickness, in millimeters per year, with respect to the mean cortical thickness across both time points) was calculated for each significant cluster (i.e., right dorsomedial prefrontal, left dorsal prefrontal, and left inferior parietal clusters). Given that three BIS scales were tested in each of the three cortical regions, a corrected p value of 0.006 was adopted (0.05/9). Critically, cannabis-related cortical thinning in the right dorsal prefrontal cortex accounted for unique variance in attentional impulsiveness at 5-year follow-up while controlling for sex, site, baseline age, baseline brain volume, baseline pubertal development, IQPR, and IQVC (b = -.119, p = .003).
We reran our analysis examining the extent to which symmetrized percent change (SPC) in the right dorsomedial prefrontal cluster was uniquely associated with attentional impulsivity at follow-up while controlling for both parent and youth reports of ADHD symptomatology at baseline. Notably, the association was still significant. Results from the GLM are displayed below.  We also tested for associations between cortical thickness change in prefrontal regions exhibiting cannabis-related thinning (i.e., significant clusters from the LMM analysis) and change in psychopathology scores (from baseline and follow-up) (assessed using the Strength and Difficulties Questionnaire, or SDQ). Sex, age at baseline, site, baseline total brain volume, baseline pubertal stage, baseline SES, and baseline IQ were accounted for across analyses. Across all tests, no significant associations were found.

eFigure 2. Baseline Local Cortical Thickness and Lifetime Cannabis Use at Follow-Up
Brain areas where baseline local cortical thickness is negatively associated with dimensional measure of lifetime cannabis use at 5-year follow-up (n = 799). Figure is shown at p ≤ 0.005, uncorrected. Controlled for age, total brain volume, sex, handedness, AUDIT Alcohol Consumption score, and site. No regions passed threshold for positive associations.