If my brain was working i’d be able to explain in great detail the findings of this paper but unfortunately it has been irrevocably destroyed by years of cannabis use , oops i should have said abuse.
So here’s the abstract
You can download the paper and see the results at https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2829657
Key Points
Question Are recent cannabis use and lifetime cannabis use associated with differences in brain function during cognitive tasks?
Findings In this cross-sectional study of 1003 young adults, heavy lifetime cannabis use was associated with lower brain activation during a working memory task; this association remained after removing individuals with recent cannabis use. These results were not explained by differences in demographic variables, age at first cannabis use, alcohol use, or nicotine use.
Meaning These findings suggest that cannabis use is associated with short- and long-term brain function outcomes, especially during working memory tasks.
Importance Cannabis use has increased globally, but its effects on brain function are not fully known, highlighting the need to better determine recent and long-term brain activation outcomes of cannabis use.
Objective To examine the association of lifetime history of heavy cannabis use and recent cannabis use with brain activation across a range of brain functions in a large sample of young adults in the US.
Design, Setting, and Participants This cross-sectional study used data (2017 release) from the Human Connectome Project (collected between August 2012 and 2015). Young adults (aged 22-36 years) with magnetic resonance imaging (MRI), urine toxicology, and cannabis use data were included in the analysis. Data were analyzed from January 31 to July 30, 2024.
Exposures History of heavy cannabis use was assessed using the Semi-Structured Assessment for the Genetics of Alcoholism, with variables for lifetime history and diagnosis of cannabis dependence. Individuals were grouped as heavy lifetime cannabis users if they had greater than 1000 uses, as moderate users if they had 10 to 999 uses, and as nonusers if they had fewer than 10 uses. Participants provided urine samples on the day of scanning to assess recent use. Diagnosis of cannabis dependence (per Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria) was also included.
Main Outcomes and Measures Brain activation was assessed during each of the 7 tasks administered during the functional MRI session (working memory, reward, emotion, language, motor, relational assessment, and theory of mind). Mean activation from regions associated with the primary contrast for each task was used. The primary analysis was a linear mixed-effects regression model (one model per task) examining the association of lifetime cannabis and recent cannabis use on the mean brain activation value.
Results The sample comprised 1003 adults (mean [SD] age, 28.7 [3.7] years; 470 men [46.9%] and 533 women [53.1%]). A total of 63 participants were Asian (6.3%), 137 were Black (13.7%), and 762 were White (76.0%). For lifetime history criteria, 88 participants (8.8%) were classified as heavy cannabis users, 179 (17.8%) as moderate users, and 736 (73.4%) as nonusers. Heavy lifetime use (Cohen d = −0.28 [95% CI, −0.50 to −0.06]; false discovery rate corrected P = .02) was associated with lower activation on the working memory task. Regions associated with a history of heavy use included the anterior insula, medial prefrontal cortex, and dorsolateral prefrontal cortex. Recent cannabis use was associated with poorer performance and lower brain activation in the working memory and motor tasks, but the associations between recent use and brain activation did not survive false discovery rate correction. No other tasks were associated with lifetime history of heavy use, recent use, or dependence diagnosis.
Conclusions and Relevance In this study of young adults, lifetime history of heavy cannabis use was associated with lower brain activation during a working memory task. These findings identify negative outcomes associated with heavy lifetime cannabis use and working memory in healthy young adults that may be long lasting.
As more states and countries have legalized the production and sale of cannabis for recreational and medical use,1 there has been an associated increase in the potency of cannabis products,2 cannabis use rates,3,4 and prevalence of cannabis use disorder.5 Greater accessibility of cannabis has also been associated with higher rates of cannabis-related motor vehicle crashes,6,7 and frequent cannabis use is associated with increased risk for hyperemesis syndrome8 and cardiovascular disease.9,10 Despite these negative effects, there is an increasing perception that cannabis is harmless.11 Thus, better understanding of recent and long-term effects of cannabis is critical for informing public health policies. Meta-analytic evidence indicates that short-term effects of cannabis include decreases in cognitive performance (eg, episodic verbal memory), but these reductions may not persist after 72 hours of abstinence.12 Given the cognitive effects of cannabis and the disruption of the endogenous cannabinoid system by tetrahydrocannabinol (THC),13,14 it may be that brain regions with high cannabinoid 1 (CB1) receptor density15 might be altered by cannabis. For example, there is evidence that cannabis use among adolescents is negatively associated with the thickness of the left prefrontal cortex (PFC) and right PFC and that the spatial pattern of cannabis-related cortical thinning is related to CB1 receptor density.16
Numerous brain imaging studies have examined the effects of cannabis on brain function. For example, relative to nonusers, frequent cannabis users showed a greater response to cannabis cues in the striatum and medial PFC, and activation of these regions correlated with cannabis craving.17 There may also be developmental interaction effects.18 For example, individuals with cannabis dependence, relative to matched control participants, showed greater functional connectivity density (ie, hyperconnectivity with surrounding regions) in the ventral striatum, and effects were more pronounced in individuals who began cannabis use earlier in life.19 Evidence has indicated that cannabis use reduces neural activation related to memory,20 executive function,21,22 emotion,23,24 reward processing,25 and social processing,26 but most of these previous studies had fewer than 30 participants with cannabis use history.20 Furthermore, whereas several efforts have successfully meta-analyzed the cognitive effects of cannabis across multiple domains,12,27 few have addressed the effects of cannabis use on brain function across multiple domains. It is also challenging to account for effects on multiple brain regions with an interpretable and clinically meaningful outcome, even though activation patterns of brain regions during tasks are not independent and, instead, are often highly correlated across regions. Evidence from a 2024 study suggests that brain analysis should consider features such as function, architectonics, connectivity, and topography.28 Such approaches, however, have seldom been applied to analysis of the effects of cannabis on brain function to help advance knowledge of the influence of history of use or recent use. Such work stands to improve understanding of how cannabis affects neural processing relevant to social, cognitive, and emotional function.
To address these knowledge gaps, we used data from the Human Connectome Project (HCP) for this study. The HCP has data across 7 tasks covering a range of brain functions. It also assesses lifetime cannabis use, cannabis dependence diagnosis (per Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition [DSM-IV] criteria), and age at first use and uses a urine toxicology screen at the time of scanning to assess for recent cannabis use. These data allowed us to disentangle outcomes associated with a lifetime history of cannabis use from those associated with recent use. The HCP dataset also allowed us to adjust for group differences between individuals with heavy, moderate, and no cannabis use, given that demographic and socioeconomic factors can influence brain function.29 We were also able to control for comorbid substance (eg, alcohol or nicotine) use, which is necessary to reduce the likelihood that any observed outcomes of cannabis use are actually attributable to use of other substances. Given that the largest effects of cannabis use are on learning, working memory,30 and verbal episodic memory,12 we hypothesized that cannabis would be associated with activation during working memory and language tasks, and that this association would be present for recent use and lifetime history of use.
Procedures for this cross-sectional study were approved by the Washington University Institutional Review Board. Participants provided written consent. We preregistered our analysis on the Open Science Framework repository.31 The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
The HCP study consisted of 1206 adults aged 22 to 37 years,32 of whom 1005 had functional magnetic resonance imaging (MRI) data. The goal of the study was to “chart the neural pathways that underlie brain function and behavior.”32 Participants were recruited to complete MRI scans at Washington University in Saint Louis, Missouri. The study sampled twins and their nontwin siblings, along with singleton community members.32 Sibling status, representing a biological relationship, was based on self-report and genotyping from blood or saliva samples. The study was launched in 2010, and scanning began in August 2012 and concluded in 2015.33 For this analysis, we accessed data from the 2017 preprocessed data release. Although we initially planned to examine the language task only, we decided to analyze all 7 functional MRI tasks to broaden our understanding of brain function outcomes of cannabis use.
To assess recent use, participants provided urine samples on the day of scanning that were tested for the presence of cannabis metabolites. For analysis, we categorized individuals as having recent use if they had a positive THC result on the Accutest multidrug screen (Jant). A positive result is indicated if urine concentrations of the THC metabolite 11-Nor-Δ9-tetrahydrocannabinol-9-carboxylic acid (THC-COOH) exceed 50 ng/mL. This result typically indicates use in the past 10 days; however, very frequent users may have THC-COOH levels greater than 50 ng/mL for 1 month, or on some occasions several months, post use.34
To assess lifetime use, participants completed the Semi-Structured Assessment for the Genetics of Alcoholism (SSAGA).35 The interview assessed total lifetime number of cannabis uses on a Likert-rated scale (with response levels of never, 1-5, 6-10, 11-100, 101-999, and >1000). We categorized individuals as nonusers if they had used 10 or fewer times, as moderate users if they had used 11 to 999 times, and as heavy users if they had used 1000 or more times. The lifetime use variable was treated as an ordered factor (nonusers, 0; moderate users, 1; and heavy users, 2).
The SSAGA also assessed a diagnosis of cannabis dependence (lifetime) (per DSM-IV criteria) and age at first cannabis use (never or <14, 15-17, 18-20, or >21 years). We treated these as covariates.
Participants completed demographic screening to indicate age, sex, race, income, and education. Race was self-reported as American Indian or Alaska Native, Asian (including Native Hawaiian or Other Pacific Islander), Black or African American (hereinafter, Black), White, multiple races, or race unknown or not reported. These data were collected because sociocultural influences associated with race may also be associated with both cannabis use and brain activation patterns. We did not have sufficient data on ethnicity to analyze. Additional substance use variables were assessed with the SSAGA, including alcohol dependence diagnosis, and the Fagerström Test for Nicotine Dependence.36 To generate a single metric each for (1) alcohol use and (2) nicotine use that accounted for quantity and frequency, we created z score metrics as described previously.19 Episodic verbal memory was assessed with Form A of the Penn Word Memory Test.37 For crystallized intelligence, participants completed the National Institutes of Health Toolbox Picture Vocabulary Test,38 which assesses vocabulary knowledge and is associated with scholastic success39 and the “g” factor of intelligence.40
Seven tasks were used in this study, aiming to cover a broad range of behavioral processes.41 Tasks were chosen based on reliability of neural response and a well-characterized neurocognitive basis. The tasks examined neural response related to emotion, reward, motor function, working memory, language, relational or logical reasoning, and theory of mind or social information processing (task details are provided in the eMethods in Supplement 1). Neural response to the tasks has previously discriminated substance use history.42 For each task, we used the primary contrast, and we extracted activation levels from regions positively activated during the task (ie, regions engaged by the task, not regions that were deactivated by the task, such as nodes in the default mode network). Positive activation was defined as having significant activation (2-tailed P < .001), with most effect sizes (Cohen d) greater than 1.00 for the contrast (the task positive condition minus the control condition).43 We used the mean value of the activation levels across the listed regions for each task, so each participant had a single value representing the activation level for each task; the regions used are depicted in Figure 1. We chose to use a single value because (1) it reduced the number of outcomes for analysis, (2) it provided a more clinically interpretable metric, and (3) the activation levels across regions during a given task were not independent.44
The HCP investigators used a 3T Connectome Skyra scanner (Siemens) to acquire imaging data. For parameters, see the data acquisition plan published previously.45 Preprocessing was conducted as part of the HCP minimally preprocessed dataset with the FMRIB Software Library (Oxford University).46
Data were analyzed from January 31 to July 30, 2024. To examine participant characteristics, we compared nonusers, moderate users, and heavy users based on race, sex, age, education, income level, urine sample status, cannabis dependence diagnosis, age at first cannabis use, alcohol dependence diagnosis, nicotine dependence score, and alcohol z scores indicating quantity or frequency of use. We used the Wilcoxon rank sum test or the Fisher exact test to determine significance levels for the Table, and we report χ2 values in the Results. Statistical significance was set at a 2-tailed α ≤ .05.
For the primary analysis, we used 7 linear mixed-effects regression models (ie, 1 model for each of the 7 tasks). For each model, we used the lme4 package in R, version 4.4.0 (R Project for Statistical Computing), to assess associations of the tasks with (1) lifetime history of cannabis (linear and quadratic fit), (2) recent use, and (3) history of dependence. Lifetime history was coded as an ordered factor with 3 levels. Models were adjusted for effects of alcohol, nicotine, race, education, income, sex, and age at first cannabis use (reference: never used). Sibling status was assessed, and this nonindependence was accounted for by a random intercept of a single variable, MotherID, which was represented as a unique value for every biological family and coded as a factor. Given the 7 models, we performed false discovery rate correction for P values using the Benjamini-Hochberg method.
For graphical comparison, we calculated effect sizes (Cohen d) for activation during each task by (1) lifetime history of use (heavy users vs nonusers), (2) recent use (THC-positive vs THC-negative result), and (3) cannabis diagnosis (history vs no history of dependence). We report comparisons between nonusers and moderate users, as well as heavy users and moderate users, in eFigures 2 and 3 in Supplement 1.
For statistically significant results of brain activation, we conducted post hoc analysis to determine which regions were associated with the result. This involved running separate linear mixed-effects models as specified earlier for each brain region (eg, for the working memory task, the 4 regions included the anterior insula [AVI], the superior parietal lobule [7Pm], the medial PFC [8BM], and the dorsolateral PFC [i6-8]).
To examine how brain activation during the tasks are associated with cognitive performance, we fit the linear mixed-effects models described earlier for (1) accuracy during the working memory task, (2) performance on episodic verbal memory, and (3) accuracy during the theory of mind task. To confirm that outcomes associated with lifetime history of heavy use were not associated with recent cannabis use, we excluded all individuals who had a positive test result for THC and we reran the models for the working memory and theory of mind tasks. We assessed the assumption of no unmeasured confounders using the sensemakr package in R, whereby we fit a linear model as specified for our primary analyses and indicated that recent THC use was our benchmark covariate for confounding. This analysis indicates the level that an unmeasured confound would have to be associated with the primary independent variable (history of use) and outcome (activation) to reduce the estimate of the association to 0. We also examined correlation using Spearman rank-ordered correlation for activation during each of the 7 tasks, along with episodic verbal memory, crystallized intelligence, income, and education level (as an ordered factor). We conducted THC-by-sex interaction models for tasks that showed an association with recent use.
