Authors: Alexander L. Wallace (aDepartment of Psychiatry, University of California, San Diego, USA), Marilyn A. Huestis (bInstitute of Emerging Health Professions, Thomas Jefferson University, Philadelphia, PA, USA), Ryan M. Sullivan (cUniversity of Wisconsin - Milwaukee, Milwaukee, WI, USA), Natasha E. Wade (aDepartment of Psychiatry, University of California, San Diego, USA)
Categories: Article, Cannabis, Amygdala, Depression, Adolescents, Hair toxicology
Source: Behavioural brain research
Authors: Alexander L. Wallace, Marilyn A. Huestis, Ryan M. Sullivan, Natasha E. Wade
Both cannabis use and depressive symptomology increase in prevalence throughout adolescence. Concurrently, the brain is undergoing neurodevelopment in important limbic regions, such as the amygdala. Prior research indicates the amygdala may also be related to cannabis use and depressive symptoms. We aimed to investigate the effects of adolescent cannabis use on amygdala volumes as well as the interaction of cannabis use and amygdala morphometry on depressive symptoms in youth.
Two-hundred-twenty-four participants (ages 12–15), balanced by sex assigned at birth, were selected from a sub-sample of the Adolescent Brain Cognitive Development (ABCD) Study based on hair toxicology and self-report measures of cannabis use. Participants positive for cannabinoids in hair and/or self-reported cannabis use were demographically matched to youth with no self-reported or confirmed cannabis use. The guardians of these youth reported depression symptoms on the Child Behavioral Checklist. Linear mixed effect models were run investigating cannabis use group on amygdala volumes bilaterally, controlling for whole brain volume and random effects of scanner type. Additional analyses examined cannabis group status and bilateral amygdala volume on depression symptoms.
Cannabis use was not significantly associated with amygdala volume but was associated with increased depressive symptoms (p<0.01). Cannabis group interacted with amygdala volume, such that individuals with smaller volumes had increased depressive symptoms within the cannabis group (p’s<0.01–0.02).
Aberrations in amygdala volume based on cannabis use were not found in early adolescence; however, more depressive symptoms were related to cannabis group. Youth who use cannabis and have smaller amygdala volumes were at increased risk for depressive symptomology, suggesting potential neurovulnerabilities to cannabis use.
The adolescent years reflect a key period of neurodevelopment [15, 53,54]. Concurrently, this time frame also marks the onset of cannabis use and/or psychopathology for some, with 18 % of 10th graders reporting cannabis use in the past year [43] and 19.5 % of 12–17 year-olds experiencing a major depressive episode [10]. Concerningly, youth with past-year depressive episodes also report higher rates of substance use than those without depressive episodes [26.1 % v. 11.5 %, respectively (Center for Behavioral Health Statistics and Quality, 2023)]. Thus, investigating the relationship between cannabis use and depression symptoms with underlying and developing neuroanatomy is important.
The amygdala is a subcortical temporal structure associated with socioemotional functioning that undergoes vast changes throughout development [54]. Amygdala volumes increase during childhood and adolescence, with notable differences in trajectory of volumetric growth, across time and by sex, as males exhibit accelerated growth patterns [24]. Importantly, the amygdala is rich in cannabinoid-receptor type-1 (CB1) [17], making it potentially vulnerable to the influence of exogenous cannabinoids.
Prior research indicates that either cannabis use or heightened depression symptomology during adolescence are linked to aberrant amygdala morphometry. For instance, a study of 16–18 year-olds found sex-based differences in amygdala volume in teens who use cannabis, such that female cannabis users had larger right amygdala volume, which correlated to increased depression symptoms [41]. Investigations in the broader cannabis research field, however, indicate that reduced amygdala volumes are more common [37,40,48]. Further, a narrative review summarized that cannabis use is linked with volumetric reductions and increased gray matter density, with particularly pronounced aberrations in individuals with earlier cannabis use onset [36]. On balance, two meta-analyses focused on adolescent and young adult neurodevelopment found minimal amygdala regional differences by cannabis use status [2,35]. Studies of amygdala volumetric differences between youth with depression and controls similarly find reduced volume [42,52], as well as sex-specific differences in growth over time [64]. Similar mixed findings are noted in the depression literature [46,49], suggesting increased need to study the relationship between amygdala volume and depression symptoms in youth. There is a paucity of research investigating adolescent cannabis use, amygdala volume, and their interaction on depression symptomology, which warrants the need for further investigation.
Cannabis use and exposure is documented through multiple means, including self-report and toxicological assessment. Both methods have their own strengths, such as detecting low-level use by self-report and providing objective verification of substance consumption via toxicological testing. Hair analysis is a robust means of determining use and provides more detailed information on cannabinoid content than self-report of cannabis use (e.g., tetrahydrocannabinol (THC) and cannabidiol (CBD) concentration). Prior work found relationships between cannabinoid concentrations and symptoms of psychosis and psychological well-being through using hair samples in cannabis-using adults [11,45,47]. Hippocampal volume was linked to hair cannabinoid concentrations [12,66]. Recently, our group found that combining hair with self-report in a large, matched sample of adolescents revealed more robust relationships between cognition and cannabis exposure than reliance on self-report alone [62]. This may be particularly important given other findings which indicate the utility of hair to identify underreporting of substance use in youth [60].
Given the potentially unique vulnerability of early adolescent neurodevelopment, we aim to investigate neuroanatomical and mood correlates of cannabis exposure at 13–14 years-old. Using the large, diverse Adolescent Brain Cognitive Development Study (ABCD), we combine self-reported cannabis use with objective hair analysis to identify youth exposed to cannabis, and match them to non-exposed controls. We expected that youth with cannabis use (CU) would demonstrate smaller amygdala volumes. Further, we hypothesized there will be a group by amygdala interaction, such that youth with cannabis use and smaller amygdala would demonstrate increased depressive symptoms.
Participants were enrolled in the ABCD Study at baseline between 2016 and 2019 through stratified probability sampling of schools within catchment areas to mirror census data [14], with data from the Year 4 follow-up visit included here from ABCD Data Release 5.1 (https://doi.org/10.15154/z563-zd24). Data were collected between September 2020 and January 2022, and all data are available via the National Institute of Mental Health (NIMH) Data Archive (https://nda.nih.gov/study.html?id=2313).
Baseline exclusion criteria included major neurological disorders, history of a traumatic brain injury, moderate to severe intellectual disability or current substance use disorder, gestational age younger than 28 weeks or birth weight less than 1.2 kg, birth complications requiring hospitalization greater than one month, or standard magnetic resonance imaging (MRI) contraindications. In addition, exclusion criteria in the present analyses included lack of MRI scan or hair sample data at Year 4 Follow-Up.
Through combining self-reported cannabis use with hair toxicological results, 114 participants were identified as youth with cannabis use (CU) and were sociodemographically-matched by age, sex, race/ethnicity, parental education, and household income to non-using controls (youth who denied cannabis use and had no positive toxicological hair samples for any substance). However, four participants (two from each group) had outlier brain volumes and were excluded (see outlier analysis). Therefore, the final study sample was 224 total participants (112 controls and 112 who use cannabis; 49 with positive hair cannabinoids, 31 with positive self-report, and 32 with concordant hair and self-report). Although the ABCD Study is longitudinal, participants with multiple completed hair analyses are limited within ABCD Data Release 5.1, restricting use of this grouping method to only Year 4 Follow-Up. Thus, previously collected data from prior yearly follow-ups (i.e., Baseline to Year 3) are provided as descriptive information but were not included in the main analyses.
Participants and their guardians attended an annual study session at their local data collection site and completed written informed assent and consent, respectively. The study protocol included parent- and youth-reported mental health questionnaires [3], substance use assessment [33], biosamples [56], and, on alternating years after Baseline (including Year 4 Follow-Up), neuroimaging [9], in addition to other measures. All aspects of the protocol were approved by a central UC San Diego IRB and follow the Declaration of Helsinki.
As mentioned, participants were identified as those with cannabis exposure through hair analysis and/or self-report. For some participants, there was insufficient quantity of hair to test all hair samples for both parent cannabinoids (THC, CBD) and tetrahydrocannabinol carboxylic (THCCOOH), the primary THC metabolite that definitively confirms cannabis ingestion. If testing was only completed for either parent cannabinoids or THCCOOH, it was included and categorized as confirmed positive or negative, even if either THCCOOH or parent cannabinoids were not assayed. Although positive hair toxicology for any other substance was exclusionary for the non-cannabis use group, participants within the cannabis use group could have positive toxicology for another substance other than cannabis use.
At baseline, participants’ guardians reported participant sex assigned at birth (referred to herein as sex), race/ethnicity, highest level of parental education attained, and household annual income [3].
Internalizing symptoms were measured with the Child Behavior Checklist (CBCL). The CBCL is a parent report questionnaire rating whether statements are consistent with their child’s behaviors [1]. Rating are compiled to create summary scores as well as Diagnostic and Statistical Manual of Mental Disorders Version 5 (DSM-5) syndrome rating scores [3]. T-scores from parent reported DSM-5 depression subscale were utilized in the present analyses.
Trained research assistants queried participants on lifetime (at Baseline) and past-year substance use (at follow-up visits; [33]), reminding them of study confidentiality and ensuring no one could overhear the interview. Though drug classes were individually reported, the present analyses focused on cannabinoid products. All cannabis products were combined into a binary variable indicating whether any type of cannabis product was used since their last study visit.
All participants were asked to allow a trained research assistant to collect a small sample (~100 mg hair collected near the root) from 3 to 4 regions around the crown of the head. Participants with hair styles that would be disrupted or where the collection would be clearly visible were not asked for a sample. Samples were then packaged and stored until sample analysis. Only a subsample of collected hair underwent toxicological analysis due to financial limitations, with the vast majority being youth who report substance use or who are at elevated risk for substance initiation; full details on sample selection are available elsewhere [60,61].
Hair samples selected for toxicological analysis were shipped to Psychemedics (Culver City, CA) and trimmed to 3.9 cm to provide a three-month window of substance use detection. In order to mitigate against external contamination (e.g., environmental contamination), samples were washed for 15-minutes with 2 mL isopropanol per 12 mg hair, three 30-min phosphate buffer washes, and two 60-min phosphate buffer washes [26]. Samples were screened by immunoassay for THCCOOH (LOD=5 pg/10 mg) then tested by liquid chromatography-mass spectrometry (LC-MS/MS), while parent cannabinoids (THC, CBD) were tested directly by GC-MS/MS. Sensitivity was maximized by testing to the limit of detection (THCCOOH, LOD/LOQ=0.02 pg/mg; THC and CBD =5 pg/mg).
MRI scans were completed locally at all 21 sites. MRI data were harmonized following techniques utilized by other multi-site MRI studies [9]. MRI data acquisition had slight variations depending on scanner type (GE, Siemens, or Philips). Whole brain T1 scan voxel size was 1 mm×1 mm × 1 mm, 256 ×256 matrix, 176–225 slices, FOV of 256 ×240–256, FOV phase of 93.75–100 %, TR of 6.31–2500 ms, TE of 2–2.9 ms, flip angle of 8 degrees, and image acquisition time of 38–7:12 minutes. ABCD staff members reviewed images for quality including FreeSurfer reconstructions. Structural MRI scans were processed in a uniform pipeline yielding cortical volumes for Desikan-Killany regions (Desikan et al., 2006); full processing pipeline is outlined in [22]. Left and right amygdala volumes as well as whole brain estimates were utilized for the present analyses.
Analyses were conducted using R version 4.1.2 [50] via RStudio [51].
The R package ‘MatchIt’ [28] was used to match the cannabis use group to controls by age, sex, race/ethnicity, parental education, and household income.
Neuroanatomical volume data were checked for outliers (defined as >3SDs ± the mean within the full subsample). There was one outlier in the non-cannabis use group for whole brain volume; this participant was excluded from all analyses as whole brain volume was included as a covariate in all analyses. Additionally, three participants (two cannabis use and one non-cannabis use) had outliers in both the left and right amygdala volume and so were excluded from the models, bringing the final sample for analysis to 224.
Sociodemographic differences were tested using analysis of variance (ANOVA) and chi-square tests. The R package ‘lme4’ [6] was used to run nested linear models, with fixed effects of cannabis group (CU or control) and whole brain volume and random MRI scanner type effects (GE, Siemens, or Phillips) predicting amygdala volumes, separately for left and right. Follow-up models tested whether parent-reported depression symptoms were predicted by bilateral amygdala volume, cannabis group status, or the interaction of the two, controlling for whole brain volume and random effect of MRI scanner type. Maximally complex models (nested linear mixed effect models (LMM)) were run [4]. Models that were overfitted due to the complexity of the model were rerun as linear regressions without random effects of MRI scanner type.
As outlined in the Participants’ section, only participants with processed hair samples were included in the study. Listwise-deletion was utilized for participants with missing demographic and MRI data at Year 4. Missing self-reported substance use data were allowed if there was confirmation of cannabis use via hair toxicology.
Participants were 12–15 years-old (mean=14.30 years, SD=0.72) and balanced by sex (49.5 % Female; N=111). Participants predominately had a parent with an Associate’s or Bachelor’s Degree (40.2 %; N=86), had a household income of <$50,000 (66.5 %; N=149), and identified as White (60.3 %; N=135) (see Table 1). As groups were matched based on sociodemographic information, there were no significant differences by cannabis use group.
Groups did not significantly differ by baseline parent-reported depression symptoms (p=.74), left amygdala volume (p=.20), or right amygdala volume (p=.37). Baseline characteristics by group are reported in Table 1.
Cannabis group status was not significantly associated with left or right amygdala volume after controlling for whole brain volume and random effects of MRI scanner type.
Due to overfitting, both left and right amygdala secondary analysis models were run as linear regressions per the analytic plan.
Cannabis use was associated with increased depression symptoms (b=3.17, p<0.01, R^2^=0.05). Additionally, there was a significant interaction between cannabis group and left amygdala volume (b=2.75, p<0.01, R^2^=0.05) where smaller amygdala volume in youth who use cannabis was related to higher depression symptoms, but larger volume sizes in the cannabis use group were related to decreased depression symptoms (see Fig. 1). Inversely, in the non-cannabis use group, larger amygdala volumes were associated with increased depression symptoms compared to smaller amygdala volumes.
Similar to the left amygdala models, in the right amygdala models, cannabis use was significantly associated with increased parent reported depression symptoms (b=3.04, p<0.01, R^2^=0.05). There was a significant cannabis use group by right amygdala volume interaction on depression symptoms (b=3.61, p=0.02, R^2^=0.05). Amygdala volumes significantly moderated depression symptoms by cannabis use group, where smaller volumes in the cannabis use group was associated with higher depression symptoms compared to cannabis using individuals with larger right amygdala volumes (see Fig. 2). Further, in the control group, larger amygdala volumes were associated with increased depression symptoms compared to smaller amygdala volumes.
The goal of this study was to examine 1) the relationship between cannabis use status on adolescent bilateral amygdala volume and 2) the potential interaction between cannabis use and amygdala volume on parent-reported depressive symptoms. While we did not find a direct relationship between cannabis use status and amygdala volume, there was an association between cannabis use and depressive symptoms, such that cannabis use status was linked with increased parent-reported depression symptomology. Further, there were significant interactions in both right and left amygdala volume models and cannabis use status on depression symptomology. These findings link cannabis use with adolescent depression symptoms based on parent report, as well as differences in depressive symptomology based on variations in subcortical volume.
This work builds on previous research that investigated the effects of cannabis use on amygdala volume [7]. While most studies note decreased amygdala volume with cannabis use [37,40,48], we did not find a main effect of cannabis use with amygdala volume. This may be due to the relative age of our sample, as most studies investigating the relationships between cannabis use and amygdala volume utilize samples of older adolescents and young adults who had much greater frequency and amount of cannabis exposure. Indeed, previous work suggests that only with long-term cannabis exposure we begin to detect differences in volume alterations [35]. Our findings lend credence to this notion given the age of our sample ranges from 12 to 15-years-old. Still, increased longitudinal monitoring of cannabis use through self-report interviews and testing of hair samples for cannabinoids would allow for closer examination of how cannabis use onset and repeated cannabis use exposure may impact regional volume differences overtime.
We found a main effect between adolescent cannabis use status and increased depressive symptoms. This aligns with previous literature linking cannabis use and depression symptomology [13,31,59,65]. Adolescent cannabis use, in particular, can lead to increased risk of depression, noting this period as a vulnerable developmental window [18]. While our findings support this body of research, we further noted a significant interaction between cannabis use and amygdala volume on depressive symptoms. Specifically, smaller amygdala volume in individuals who used cannabis were associated with increased depressive symptoms. Previous research examining amygdala volume and depression suggested that smaller amygdala volume in pediatric populations are associated with increased depression [42,52]; however, these findings tend to “flip” in adulthood, where larger amygdala volume is associated with increased depression [23]. While we did not find a main effect of amygdala volume on depression, individuals who used cannabis did show a similar pattern of smaller amygdala volume and increased depressive symptomology.
Of note, relationships between amygdala volume and depression differed in control participants where larger amygdala volumes were associated with increased depressive symptoms. As stated above, previous literature shows the opposite relationship in pediatric populations [52]. While transdiagnostic models of adolescent psychiatric symptoms demonstrate grey matter volume loss in the amygdala [32], it is possible that in our sample of early adolescents we are capturing amygdala volumes prior to these developmental changes. In this way, our sample would be undergoing early stages of pruning [19] and potentially showcasing early altered pruning processes due to depressive symptoms [38]. However, as our findings are cross-sectional in nature, additional work leveraging longitudinal data are needed to better understand this process overtime. Further, past work demonstrates that other factors such as early life stress and socioeconomic status may attenuate amygdala volume in youth [42,63]. While we matched our samples based on household income—among other sociodemographic information—we did not specifically investigate these factors as they are beyond the scope of the current study. Future investigations are warranted to tease apart the relationships between amygdala volume and depression in individuals who do not use cannabis.
Interactions between cannabis use and amygdala volumes on depressive symptoms may be due to several underlying mechanisms. Differences in the relationship between amygdala volume and depression symptoms in cannabis using individuals may suggest a premorbid susceptibility to cannabis use. While some studies demonstrate premorbid differences in brain volume predicting future cannabis use [30], we did not see any significant differences between cannabis and non-cannabis use groups in amygdala volume from baseline scans. Rather, some individuals may possess genetic vulnerabilities to cannabis use [27,58], with expression of these vulnerabilities contributing to variable amygdala development and related depressive symptomology. Additionally, these findings may be due to changes in neurodevelopmental trajectories based on various factors of cannabis use. Specifically, THC potency [8], cannabidiol concentrations [5], lifetime duration of cannabis use [16,20], and route of cannabis consumption (e. g., smoking vs edibles; [44]) all may differentially impact adolescent neurodevelopment trajectories and depressive symptomology. Indeed, our study investigated whether participants were positive for cannabis use via hair toxicology and self-report, broadly. Additionally, some participants within the cannabis group had positive hair toxicology for other substance use. While these factors were not examined fully here, it is possible that other substance use exposure may have impacted neurodevelopmental trajectories and depression symptomatology [21].
Further, sex differences in the amygdala have been noted in response to adolescent cannabis exposure [41] as well as development of depression symptomatology [64], which suggest that examining these trajectories stratified by sex may produce slightly different results. Though the current study was underpowered to study these unique effects, sex differences are still one of many important factors to consider when investigating altered neurodevelopmental trajectories. Additional factors such as parental history of depression [57] and parental substance use [39] are extensively documented to have effects on adolescent psychopathology, substance use, and brain development. Both parental history of substance use and depression may speak to a genetic susceptibility, which could be demonstrated by differences in amygdala volume, their relationships to varying depression symptoms, or even parental perceptions of adolescent depression. Future investigations should more specifically target these qualitative factors to disentangle the complex nature of cannabis and brain development.
As our findings lend support that early cannabis use exposure may signal increased risk of depressive symptoms, this evidence suggests that clinicians should be screening for cannabis use even in early adolescence. While most reports begin tracking cannabis use onset starting at age 12 [25], the ABCD study demonstrates that a small number of youth initiate cannabis use as early as age 9 [34,55]. Although hair toxicology data suggests that self-reported numbers are underestimates [60], cannabis use is rare at this age with only 1.2 % of youth self-reporting cannabis use in these early years [55]. However, despite the low prevalence, our study suggests that youth who are using cannabis are at increased risk for depression symptoms. As a cross-sectional study, we cannot disentangle the causality between cannabis use and depressive symptoms, but cannabis use should be screened for in early adolescence as a potential marker for increased risk of mental health difficulties.
Like most studies, there are multiple limitations to highlight. Given the limited processing of hair samples for the same participants across all time points and the grouping methods employed here, we were only able to conduct cross-sectional analyses. As the ABCD Study processes more hair samples, longitudinal analyses can be conducted to more specifically study how positive cannabinoid hair results impact brain development and psychiatric symptoms. As mentioned above, we examined whether participants used cannabis broadly; however, future examinations should look more granularly at differences between THC and CBD to consider the different impact they may have in neurodevelopment. While our study incorporated parent reports of adolescent depression symptoms, as youth progress into adolescence, they start to express additional symptomology beyond parent-report [29]. Due to concerns for power given the missingness of concordant parent and self-report in this sub-sample, we did not investigate across both modalities of reporting. However, future utilization of multi-informant reports of depressive symptoms as more hair samples are analyzed may provide differing results.
Our study provides evidence that cannabis use is associated with increased depressive symptomology in early adolescence. While we did not find a direct relationship between adolescent cannabis use and amygdala volumes, we did find that cannabis use uniquely impacts depressive symptoms depending on amygdala morphometry, implying downstream behavioral changes for some youth depending on subcortical neuro-correlates. These differences may be due to underlying vulnerabilities of some youth or varying neurotoxicity of cannabis products. Future work should continue to investigate more specific cannabis use markers combined with longitudinal analyses to further elucidate the impact of adolescent cannabis use and neuroanatomical characteristics on psychiatric symptoms in the developing brain.