Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jun 10.
Published in final edited form as: Addict Biol. 2020 Jan 28;26(1):e12874. doi: 10.1111/adb.12874

Recent Cannabis Use is Associated with Smaller Hippocampus Volume: High-Resolution Segmentation of Structural Subfields in a Large Non-Clinical Sample

Max M Owens 1,2, Lawrence H Sweet 2,3, James MacKillop 2,4,5
PMCID: PMC9187039  NIHMSID: NIHMS1783088  PMID: 31991525

Abstract

There is mixed evidence that individuals who use cannabis have reduced hippocampal and amygdalar gray matter volume, potentially because of small sample sizes and imprecise morphological characterization. New automated segmentation procedures have improved the measurement of these structures and allow better examination of their subfields, which have been linked to distinct aspects of memory and emotion. The current study applies this new segmentation procedure to the Human Connectome Project Young Adult dataset (N = 1080) to investigate associations of cannabis use with gray matter volume in the hippocampus and amygdala. Results revealed significant bilateral inverse associations of hippocampal volume with recent cannabis use (THC+ urine drug screen; p<.005). Hippocampal subfield analyses indicated these associations were primarily driven by the head of the hippocampus, the first section of the Cornu Ammonis (CA1), the subicular complex, and the molecular layer of the hippocampus. No associations were detected for age of cannabis initiation, the frequency of cannabis use across the lifespan, or the lifetime presence of cannabis use disorder. In one of the largest studies to date, these results support the hypothesis that recent cannabis use is linked to reduced hippocampal volume, but that this effect may dissipate following prolonged abstinence. Furthermore, these results clarify the specific subfields which may be most associated with recent cannabis use.

INTRODUCTION

Cannabis is one of the most widely used psychoactive substances worldwide1 and is increasingly available in legal markets. Because of this, it is important that the effects of cannabis on the brain be fully understood and the literature on these effects be robust and replicable. Furthermore, research suggests that, as cannabis has increased in potency (i.e., increasing content of tetrahydrocannabinol [THC] and decreasing content of cannabidiol [CBD])2, that the effects of cannabis on the brain structure may be increasing as well3. One apparent effect of cannabis on the brain is reduced gray matter volume in the hippocampus in cannabis users4. This has been found across contexts and populations, albeit typically in very heavy users in small samples57. Additionally, a larger study (N = 111) has shown that differences in in hippocampal gray matter volume in cannabis users do not extend to former users, suggesting that there may be changes to hippocampal volume caused by cannabis which reverse over time8. However, findings of reduced hippocampal volume in cannabis users are not uniform across the literature, as a recent mega-analysis of 185 cannabis dependent individuals and 246 controls failed to find differences in hippocampal or amygdalar volume9, as did a 2017 study of 20 young adult heavy cannabis users and 23 matched non-cannabis using healthy controls10.

The hippocampus is not a homogenous structure and new research has begun to investigate the effects of cannabis use on specific subfields of the hippocampus. Anterior-to-posterior the hippocampus is divided into its head, body, and tail11. Spanning across these divisions, the hippocampus consists of several histologically distinct subfields that are interconnected through a number of pathways that pass information in and out of the hippocampus through the entorhinal cortex, and among the subfields of the hippocampus11,12. For example, one major pathway, the perforant path, has neural signal (i.e., action potentials) pass from the entorhinal cortex to the dentate gyrus (DG), to the third section of the Cornu Ammonis (CA3), then to the second section (CA2), then to the first (CA1), and then to the subicular complex which projects back to the entorhinal cortex13. In animal models, post-mortem, and some early in vivo human imaging studies, the hippocampal subfields have been shown to fulfill different roles in the memory processes conducted in the hippocampus1418. In the first study to examine the association of cannabis with hippocampal subfields, Chye et al.19 used Freesurfer’s automated segmentation to find several subfields of the hippocampus differentiated 39 cannabis dependent individuals from 35 non-users: CA1, CA2, CA3, CA4 and the dentate gyrus (DG). This study also found the CA2, CA3, CA4, and DG to be associated with lifetime amount of cannabis used. Another study in 89 young adults found similar results, identifying differences between cannabis users and controls in the CA1, CA2/3, and CA4 as mapped by Freesurfer. However, since that time, due in part to concerns about earlier hippocampal subfield segmentation procedures20, new subfield segmentation programs for the hippocampus and amygdala have been released in Freesufer version 6.021,22, which significantly improved the resolution and accuracy of hippocampal and amygdalar subfield segmentation, while also allowing the ability to quantify the volume of several smaller structures in the hippocampus including the parasubiculum, HATA, fimbria, and hippocampal fissure. The new hippocampus segmentation was used by Beale et al.23 to investigate changes in hippocampal subfield volume in 20 heavy cannabis users who received CBD for 10-weeks, finding an apparently restorative effect. However, this study did not compare these cannabis users to healthy controls, limiting the ability of this study to make inferences about the association of cannabis with the hippocampal subfields identified by Freesufer 6.0.

There have also been a number of studies linking cannabis use to smaller amygdalar volume2426. Similar to the hippocampus, the amygdala is not one homogenous brain structure, but rather composed of subfields that are unique in cytoarchictecture and histology27. The amygdala’s subfields are typically grouped into the basolateral complex, made up of the lateral nucleus, basal nucleus, accessory-basal nucleus, paralaminar nucleus, and corticoamygdaloid transition; and the corticomedial nuclei made up of the cortical nucleus, central nucleus, medial nucleus, and the anterior amygdaloid area27,28. These subfields are interconnected through various pathways which are responsible for emotional processes such as fear conditioning, reward processing, aggression, and addictive behavior28. The best understood of these pathways is for fear conditioning, which occurs via a pathway entering into the lateral nucleus, then connecting to the basal nucleus, then to the central nucleus, and then out to other subcortical structures28. Other specific nuclei have been linked with reward, motivation, and addiction28. There has not been, to our knowledge, any published research on the association of cannabis with the volume of these amygdalar subfields.

Along with a general lack of research on the subfields of the amygdala and hippocampus in cannabis users, another limitation of the existing literature is a major focus on individuals with very heavy cannabis use rather than those using in a way that is more typical in the population. One exception includes a study by Chye et al. that examined 22 non-dependent cannabis users, finding them not to differ from healthy controls in hippocampal volume19. Additionally, the largest previous investigation of cannabis use in a non-clinical sample to date was a two study investigation, with study one including 622 young adults and study two including 474 middle-aged men drawn from the general population; this investigation also did not find associations between hippocampal or amygdalar volume and cannabis use in either study29. However, one concern in concluding that there are no differences in hippocampal or amygdalar volume in recreational cannabis users from is this study is that the study included only a coarse single-item measure of cannabis use, “Have you ever used cannabis once a week or more?” which allows for the possibility that important features of cannabis use were overlooked, such as heaviness of use, age of initiation, and recency of use. Additionally, data from this study were analyzed using an older version of Freesufer (v3.0) and were collected with a 1.5-Tesla MRI scanner, both of which may have resulted in less sensitivity to detect effects than the current study which used the most advanced data acquisition and analysis approaches available.

Thus, the aim of the current study was to investigate which hippocampal and amygdalar subfields are most associated with cannabis consumption using more nuanced measures of cannabis use, the improved Freesurfer 6.0 segmentation procedures, and data from a 3-Tesla MRI scanner. This was done by utilizing data from the Young Adult Human Connectome Project (HCP) to test these associations in one of the largest studies to date. To determine if specific cannabis use features are most related to hippocampal and amgydalar volume, the current study examined four cannabis use variables: total lifetime consumption, age of initiation, lifetime cannabis use disorder status, and recent use, defined by the presence of THC in a urine drug screen administered on the day of the scan.

METHODS

Participant Characteristics

MRI data were collected from 1113 participants at Washington University in St. Louis over the course of two days as part of the Human Connectome Project between August 2012 and October 2015, and released in full on March 1, 2017. Informed consent was obtained for all participants. Participants were 22–35 years old and had no significant history of psychiatric disorder, substance abuse, neurological disorder or damage, cardiovascular disease, or Mendelian genetic disease (e.g., cystic fibrosis). They also did not have any contraindications for receiving an MRI such as metal devices in the body or claustrophobia. For full details of inclusion and exclusion criteria, see van Essen et al.30. Participants were excluded from the current analysis if they had missing data on cannabis use or for any covariates, resulting in loss of 33 participants. Thus, the final sample for this study comprised 1080 participants (Table 1).

Table 1.

Sample demographic characteristics (N = 1080).

Variable Full Sample

Sex (%Female) 54.5%
Years of Age (mean) 28.8(3.7)
Race (% Non-White) 25.5%
Ethnicity (% Hispanic) 8.6%
Income (median/year) $50,000–$74,999
Years of Education (mean) 14.9(1.8)
Heaviest Drinks per Day (mean) 3.4 (1.7)
Ever Regular Smoker (%) 26.7%
Ever Used Illicit Drugs other than Cannabis (%) 21.5%
THC+ status (% positive) 11.0%
Times Used Cannabis (median) 1–5
Age 1st Use (% used before age 18) 28.1%
Lifetime CUD Diagnosis (% yes) 9.2%

CUD= Cannabis Use Disorder.

Measures

Semi-Structured Assessment for the Genetics of Alcoholism (SSAGA-II)

The cannabis use questions from the SSAGA-II31 were used to assess lifetime exposure to cannabis. These items included Times Used, Age of First Use, and lifetime diagnosis of cannabis use disorder (Lifetime CUD DX). These responses were binned ordinally by the HCP research team prior to release to the public. Times Used was binned into six categories: 1) never; 2) 1 to 5 times; 3) 6–10 times; 4) 11–100 times; 5) 101–999 times; and 6) 1000 or more times. Age of First Use was binned into five categories 1) under 15; 2) 15–17; 3) 18–20; 4) over 20; and 5) never used. Lifetime CUD DX was coded as a binary variable.

THC Urine Screen

Participants completed a urine drug screen on the day of testing that assessed for recent exposure to tetrahydrocannabinol (THC), the psychoactive component of cannabis. From this a binary variable was created called THC+ status indicating whether participants were positive for the presence of THC in their urine. Prior research suggests that the amount of time an individual remains positive for THC is dependent on several factors including heaviness of cannabis use. Detection is possible for between one and three days after a single use and three to four weeks on average for heavy users, although it may be detectable up to three months later in some individuals32.

Retrospective Report of Tobacco and Alcohol Use

A retrospective report was given for tobacco and alcohol use. From this, heaviest alcoholic drinks per week and lifetime cigarette smoking history were derived for use as covariates.

Procedures

MRI Data Acquisition and Processing

High-resolution T1-weighted and T2-weighted structural images were acquired on a 3T Siemens Skyra scanner (Siemens AG, Erlanger, Germany) with a 32-channel head coil at a resolution of 0.7 mm3 isotropic. Scanning parameters for the T1 scan were FOV = 224 × 240, matrix = 320 × 320, 256 sagittal slices; TR = 2400 ms and TE = 2.14 ms; parameters for the T2 scan were FOV = 224 × 240, matrix = 320 × 320, 256 sagittal slices; TR = 3200 ms and TE = 565 ms. T1 and T2 scans were immediately reviewed for quality by the MRI technician after acquisition and then reviewed again within 1 business day by another trained rater for crispness, blurriness, motion, and artifacts33. Scans were re-acquired if scan quality was rated as unsatisfactory in either of these reviews. Data were reconstructed and preprocessed by the HCP team using the Freesurfer pipeline34,35 in FreeSurfer Image Analysis Suite version 5.3 (http://surfer.nmr.mgh.harvard.edu). In this process, Freesurfer models boundaries between the cortical white matter, gray matter, and pial surfaces in order to create a two-dimensional representation of the surface of the gray matter ribbon of the cortex. Additionally, after Freesurfer pipelines were completed image reconstruction, preprocessing, and segmentation were checked visually for major errors or irregularities. See the HCP publication on their preprocessing pipeline for more details of acquisition, reconstruction, and preprocessing33,36. Additional quality control was performed for the Freesurfer version 6.0 hippocampal and amygdalar segmentation including checking for outliers and spot-checking individual subjects. Using an outlier threshold of Z = 4, one outlier was found for the whole amygdala; none were found for whole hippocampus or hippocampal subfields. The outlying value for whole amygdala was winsorized to the highest non-outlying value. Spot-checking did not uncover any aberrant segmentations.

Segmentation of Hippocampus and Amygdala

The FreeSurfer version 5.3 pipeline was utilized to process the T1 and T2 data and to derive intracranial volume (ICV) for use in this study34,35. Hippocampal and amygdalar subfield segmentation was derived using the new automated algorithm available in FreeSurfer 6.021,22; these segments were the basis of the whole hippocampus and amygdala data used in the current study as well as all the hippocampal and amygdalar subfields. To create bilateral hippocampal and amygdalar volumes for each subject, volumes of the left and right hippocampus were averaged for each subfield and for the whole hippocampus and amygdala. The hippocampal and amygdalar segmentations are based on refined probabilistic atlases constructed by three independent neuroanatomists from both in vivo and ex vivo human brains using both manual and automated methods. Using Bayesian inference, the atlas is used to automatically segment the hippocampus and amygdala into 9 bilateral nuclei (i.e., 18 total). These segmentation procedures were both validated on large publicly available datasets (ADNI and ABIDE) and both demonstrated greater accuracy in identifying individuals from special populations than prior versions of FreeSurfer.22,37

Data Analysis

Before beginning the primary analyses, all cannabis use variables were tested for association with each other using Pearson’s correlations. Additionally, hippocampal and amygdalar subfields were examined for association using Pearson’s correlations. In the primary analyses, each cannabis use variable was tested for association with volume in the hippocampus and amygdala using a two-step hierarchical approach. Both steps of this strategy used linear mixed effect modeling conducted in SPSS version 25. In these models, cannabis use variables were each separately tested as predictors of hipoocampal and amygdalar volume. Covariates included in all models as fixed effects were estimated total intracranial volume (eICV), age, gender, income, education, drinks per week, tobacco use days per week, and recent use of illicit drugs other than cannabis. These covariates were chosen for the known relationship with cannabis use and brain volume. Differences in covariates between cannabis users and non-users are shown in Supplemental Table 1. Additionally, family was used as random effect in order to account for the similarity of members of the same family with each other (i.e., siblings). The first step of the primary analysis was to test the associations of each cannabis use variable with volume of the whole hippocampus and volume of the whole amygdala. In this step a threshold of p < .005 was used in accordance with recommendations from the literature38. This was done in order to constrain the subfields tested in the second step to only those in which the entire structure was associated with cannabis use. In the second step, the hippocampus and amygdala subfields produced by Freesurfer 6.0 were used as the dependent variables in separate models. These included nine subfields per hemisphere for the hippocampus and nine per hemisphere for the amygdala, as well as the hippocampal head, hippocampal body, and hippocampal tail which are divisions that are orthogonal to the nine primary subfields derived. Then each of these subfields was examined for association with any of the cannabis use variables that were significantly associated with the whole amygdala or whole hippocampus for that hemisphere. In this step a false discovery rate threshold of q < .05 was used to account for family-wise error.

RESULTS

Preliminary Correlational Analyses

Cannabis use variables were all significantly correlated (p < .01; Supplemental Table 2). Figure 1 for distributions of cannabis use variables. See Supplementary Table 1 for distributions of cannabis use variables. Additionally, bilaterally the whole hippocampus and whole amygdala were significantly associated with each of their own subfields (p < .01; Supplemental Table 3).

Figure 1.

Figure 1.

Example of hippocampal segmentation from one participant in the current study. Left column of labels represents amygdalar subfields; right column represents hippocampal subfields.

Whole Hippocampus and Amygdala Associations with Cannabis Use

THC+ status was negatively associated with the whole hippocampus at p < .005 with a small effect size (d ~ .1). Neither Times Used, Age of First Use, nor lifetime CUD DX were associated with gray matter volume in the hippocampus (Table 2). No cannabis use measures were associated with gray matter volume in the amygdala at p < .005. When testing the associations of THC+ status with hippocampal volume, gender and eICV were also significantly related to hippocampal volume (Supplemental Table 4); when testing the associations of THC+ status with amygdalar volume, gender and eICV were also significantly related to hippocampal volume (Supplemental Table 5).

Table 2.

Associations of individual cannabis variables with whole hippocampus and amygdala volumes, covarying for estimated total intracranial volume, age, gender, income, education, twin status, drinks per week, tobacco use days per week, and recent use of illicit drugs other than cannabis.

Cannabis Index Structure F pF Category B SE t pt d

THC+ Status Whole Hippocampus 9.05 0.003 THC+ −154.46 51.34 3.01 0.003 0.09
THC− - - - - -

Whole Amygdala 3.76 0.053 THC+ −59.38 30.64 1.94 0.053 0.06
THC− - - - - -

Times Used Whole Hippocampus 0.79 0.56 Never 119.15 70.68 1.69 0.09 0.05
1–5 120.36 70.84 1.70 0.09 0.05
6–10 105.94 76.93 1.38 0.17 0.04
11–100 52.13 70.23 0.74 0.46 0.02
101–999 46.32 72.64 0.64 0.52 0.02
100+ - - - - -

Whole Amygdala 1.26 0.28 Never 58.09 42.05 1.38 0.17 0.04
1–5 50.19 42.11 1.19 0.23 0.04
6–10 92.88 45.85 2.03 0.04 0.06
11–100 9.28 41.80 0.22 0.82 0.01
101–999 15.55 43.38 0.36 0.72 0.01
100+ - - - - -

Lifetime CUD DX Whole Hippocampus 0.03 0.87 CUD+ 9.99 59.65 −0.17 0.87 −0.01
CUD−

Whole Amygdala 0.04 0.84 CUD+ −7.00 35.64 0.20 0.84 0.01
CUD−

Age of First Use Whole Hippocampus 0.69 0.60 Never −24.25 48.95 −0.50 0.62 −0.02
<=14 −67.07 77.77 −0.86 0.39 −0.03
15–17 −45.85 55.10 −0.83 0.41 −0.03
18–20 −80.90 53.70 −1.51 0.13 −0.05
>=21 - - - - -

Whole Amygdala 0.80 0.52 Never −19.48 29.20 0.67 0.51 0.02
<=14 −21.55 46.31 0.47 0.64 0.01
15–17 −51.31 32.88 1.56 0.12 0.05
18–20 −41.39 32.08 1.29 0.20 0.04
>=21 - - - - -

Bold indicates significant associations at p < .005. F and pf represent the F test and p-value for the whole factor variable. B, SE, t, pt, and d represent the post-hoc test for each of the dummy coded variables in each factor. d = effect size calculated as the unstandardized coefficient (B) divided by the pooled standard deviation, which functions as a pseudo-Cohen’s d. Categories with dashes (-) in place of statistics indicate reference category.

Hippocampal Subfield Associations with THC+ Status

Since only THC+ status was associated with gray matter volume of the whole hippocampus, only THC+ status was evaluated for associations with the hippocampal head/body/tail and the hippocampal subfields. THC+ status was significantly negatively associated with the head of the hippocampus with a small effect size (d ~ .1), though not the hippocampal body or tail (Table 3). Regarding the hippocampal subfields, THC+ status was negatively associated with the CA1, the subiculum and presubiculum, and the molecular layer, all with small effect sizes (d ~ .07 - .1). Since no cannabis use variables were associated with gray matter volume of the whole amygdala, amygdalar subfields were not tested for associations with cannabis use variables.

Table 3.

Associations of THC+ Status with hippocampal subfields, covarying for estimated total intracranial volume, age, gender, income, education, twin status, drinks per week, tobacco use days per week, and recent use of illicit drugs other than cannabis.

Region (DV) F B SE t p d

Hippocampal Head 11.03 102.06 30.73 3.32 0.001 0.10
Hippocampal Body 1.71 24.43 18.66 1.31 0.19 0.04
Hippocampal Tail 2.43 20.21 12.95 1.56 0.12 0.05
CA1 9.73 36.72 11.78 3.12 0.002 0.09
CA2/CA3 0.22 2.28 4.86 0.47 0.64 0.01
CA4 2.95 7.50 4.37 1.72 0.09 0.05
Dentate Gyrus 3.58 17.58 9.30 1.89 0.06 0.06
Subiculum 5.64 18.23 7.68 2.37 0.02 0.07
Presubiculum 6.35 15.39 6.11 2.52 0.01 0.08
Parasubiculum 3.77 4.24 2.18 1.94 0.052 0.06
Molecular Layer 7.66 24.27 8.77 2.77 0.006 0.08
Fimbria 4.67 6.43 2.97 2.16 0.03 0.07
HATA 3.71 3.15 1.64 1.93 0.05 0.06
Hippocampal-Fissure 0.39 2.34 3.75 0.62 0.53 0.02

Bold indicates significant at FDR q < .05. d = effect size calculated as the unstandardized coefficient (B) divided by the pooled standard deviation, which functions as a pseudo-Cohen’s d.

DISCUSSION

The current results provide one of the highest resolution examinations of the links of cannabis use on the volume of the hippocampus and amygdala to date. The results indicate negative associations of recent cannabis use, as indicated by positive urine screen for THC, with volume the hippocampus, but not in the amygdala. Subfield analyses of the hippocampus indicate that these effects were being driven by negative associations between recent cannabis use and volume in the several hippocampal structures: the CA1, presubiculum, subiculum and molecular layer. Additionally, on the anterior/posterior axis of the hippocampus, cannabis use was associated with the hippocampal head, but not the body or tail. Effects sizes of these associations were small, with the largest effects found in the hippocampal head and CA1 (d = .10 and d = .09, respectively).

Recent cannabis use was associated with smaller volume of the head of the hippocampus but was not associated with volume of the body or tail. The hippocampal subfield most clearly associated with cannabis use was CA1, which was bilaterally associated with THC+ status. CA1 receives inputs from the DG and CA3 along the perforant path. Additionally, two subfields that make up the subicular complex (subiculum and presubiculum) were also associated with cannabis use. The subicular complex is the final stop in the perforant path receiving inputs primarily from CA1, but also from CA3 and DG. It is where outputs from the hippocampus are initiated, projecting outputs not only to the entorhinal cortex and perirhinal cortex, but also to the medial orbitofrontal cortex and anterior cingulate cortex39,40. It has been suggested that one of its roles is to amplify the signal from the rest of the hippocampus that is output to these regions39,40

The differences in results between the cannabis use variables are notable. THC+ status, indicating recent use, was the only cannabis variable linked to hippocampal or amygdalar volume. No association was found between hippocampal or amygdalar volume and Age of First Use, which marks how early an individual began using cannabis but not about how much they have used, CUD DX, which is indicative of having ever met the diagnosis for cannabis use disorder according to a clinical interview, or time used, which indicates the total amount of cannabis used across the lifespan. Thus, the current results suggest that the more important feature to the volume of these two structures is whether cannabis was used recently than the total amount used in the past, the severity of cannabis use problems, or the age at which use was initiated. This is consistent with prior literature, which has found that differences in hippocampal volume found in cannabis users were not present in former cannabis users8. The current results provide some support for the hypothesis that hippocampal morphometry may revert to that of a non-user following prolonged abstinence from cannabis.

The present results differ from those of Gillespie et al.29 and Chye et al.19, in which results indicated an absence of association between non-clinical levels of cannabis use and hippocampal or amygdalar morphometry. One likely reason for this is the improved resolution of the measurement of cannabis use in the current study, which measured cannabis use using both self-reported and biological measures, assessed for use both past and recent used, and distinguished between problematic and non-problematic cannabis use. The use of multiple cannabis measures presents one major improvement of this study compared to prior literature. Another improvement of the current study was the use of more advanced neuroimaging assessment using 3T MRI and Freesurfer’s new automated hippocampal and amgydalar segmentation procedure21,22, which was designed to address noted issues in existing automation segmentation procedures20. Additionally, this was the largest study to date to investigate these effects, which provided greater statistical power than previous studies. Each of these improvements may have allowed greater sensitivity to detect small effects. Indeed, the effect sizes of the associations of cannabis use and hippocampal volume were small, making prior lack of detection due to lack of power a possibility. Future large studies of heavy cannabis users may find larger effects, although this is conjecture.

Whole hippocampus and amygdala results are only somewhat consistent with prior reports using partial samples of the HCP dataset. Pagliaccio et al.41 found associations of Times Used with the left amygdala volume and left hippocampal volume in a partial HCP sample. Likewise, Orr et al.42 also found that the shape of the left hippocampus and left amygdala was associated with Times Used cannabis in a partial HCP sample. It also was unique in that it found THC+ status to be the variable that was associated with hippocampal volume. Furthermore, unlike prior studies, the current study did not find associations of amygdala volume nor did it find associations of Times Used with hippocampal volume that were significant beyond multiple comparison correction. There are several potential reasons for this discrepancy. The current study differed from these by analyzing hippocampal and amygdalar data that was combined across hemispheres, given that there was no a priori reason to suspect that cannabis would affect one hemisphere exclusively. It is also possible that the differences are the result of the new Freesurfer hippocampal and amygdala segmentation software (i.e., version 6.0), while the other studies used Freesurfer version 5.341 and FSL for VBM42 respectively. One final reason for the difference may be the differences in sample size. The current study used the full HCP sample (n = 1068) while the previous studies used an earlier release of the HCP dataset resulting in sample sizes of 483 (Pagliaccio et al.) and 466 (Orr et al.). Recent research has suggested that smaller sample sizes are associated with greater risk of type I and type II error 43 and while the prior studies would not be called small, this does provide an additional potential reason for the differences in findings.

One important issue relating to this analysis is the issue of directionality. Since this study uses a pseudo-experimental design, it cannot be definitively determined if cannabis use modifies the volume of the hippocampus, having smaller hippocampal volume is associated with a predisposition to use cannabis, or a third variable is associated with both cannabis use and volume of the hippocampus. However, there is reason to suspect that hippocampal effects may be caused by cannabis use, rather than representing a predisposing factor to cannabis use. Associations of urinary THC+ status with hippocampal volume could suggest that cannabis use is causing changes to the volume of the hippocampus that is reversible following abstinence. This would be consistent with prior work that found that former cannabis users did not show differences in hippocampal volume8 and with other work from the HCP sample that found differences in working memory performance and associated underlying neural activation differed by THC+ status, but not by other measures of cannabis use. If the association of cannabis use and hippocampal volume dissipates after cessation that would suggest it is not a predisposing factor but rather a consequence of use.

This study has several important implications. First, the study’s broader set of findings provide further evidence for a long-held hypothesis that cannabis use is associated with smaller volumes in the hippocampus, the key structure in the formation and storage of memories. This study provides one of the largest and most thorough tests of this hypothesis to date, making it a meaningful contribution to the literature for this reason alone. Second, the results of subfield analyses provide clues for future research regarding the specific mechanisms by which cannabis use affects cognitive function. There is strong evidence that cannabis is associated with neuropsychological deficits44,45 and understanding the specific nuclei in the hippocampus that differ in cannabis users provides a theoretical framework from which to base future research on the neural mechanisms by which cannabis affects these cognitive functions. Third, the differential results among cannabis use variables provide evidence about the nature of the effects cannabis may have on the hippocampus and amygdala. That recent use was the most strongly associated feature of cannabis use with hippocampal volume and that age of cannabis initiation and history of cannabis use disorder were not associated with hippocampal or amygdalar volume suggests that cannabis may have acute effects on the volume of the hippocampus that dissipate over time rather than chronic effects that result from early initiation of use or problematic use in the past.

One notable consideration of the current study is that of effect sizes. All associations demonstrated between cannabis use and volume of the hippocampus were small. Despite findings that were significant substantially beyond appropriate multiple comparison correction, no cannabis use variable had an effect size of over .1 for its association with the hippocampus or in any subfield. However, even small effect sizes may have meaningful effects in some contexts, such as those already struggling academically or professionally due to limitations in memory who may be further impacted by these changes to the hippocampus. In addition, the effects are likely larger for individuals with very heavy cannabis use.

Another consideration is of specificity among subfield results. Given that the hippocampal subfields are all a part of one larger structure, it was expected that they would be highly correlated. However, this inherently creates challenges for specificity, as the question is raised if certain parts of the hippocampus are indeed uniquely linked to cannabis use or if there is simply be a global effect on the hippocampus and identified regional distinctions represent only statistical noise. The answer to this question likely varies by subfield, with some being clearly associated with THC or not, and others being more ambiguous. For example, it seems very likely that the hippocampal head, CA1, subiculum, presubiculum, and molecular layer represent subfields are truly linked to recent cannabis use; they were clearly significant and had the largest effect sizes. Equally clear was the lack of association of THC+ status with the CA2/CA3 and hippocampal fissure, which was far from significant and had effect sizes which were 70–80% smaller than the key subfields identified above. However, there were several subfields in which greater caution should be exercised in making claims about their relationship to recent cannabis use. In addition to being non-significant in even a standard p < .05 thresholding, the hippocampal body and tail, the CA4, and the HATA had effect sizes that were 40–50% smaller than the regions which were most clearly associated with THC+ status; they seem likely but not certain to be true negatives. Furthermore, there were three regions (DG, parasubiculum, and fimbria) which were nominally significant and were 35% smaller than the regions clearly linked to THC+ status. These regions represent those most likely to be false negatives and it may be that these regions would be found to be linked to THC+ status as well in a larger sample, though with a smaller effect size. Given the lack of existing studies using hippocampal subfield analyses in the context of cannabis use, replication of the current subfield results is needed to increase confidence that the hippocampal head, CA1, subiculum, and presubiculum are the only subfields of the hippocampus linked to recent cannabis use.

A limitation of this study was its inability to assess the specific behavioral and cognitive factors associated with volume of the hippocampal and amygdalar subfields. Doing so would have required more nuanced tests of memory and emotion than were available in the current dataset. Future work exploring the differences in functional importance for these subfields is called for. Additionally, cannabis use is associated with numerous individual differences, which may serve as confounds in quasi-experimental research such as the current study. Fortunately, the current sample was well-powered enough to allow for extensive use of covariates, though it would be ideal to conduct this research in perfectly matched sample of cannabis users and non-users. Additionally, the fact that cannabis use variables were only available as binned, ordinal measures represents another limitation as converting to ordinal data loses some of the variance captured in continuous data. However, this limitation was unavoidable to us, as the decision to make the data ordinal was made by the HCP investigational team. Generally, the inability to make decisions specific to each individual analysis/project that is conducted with a given dataset represents one of the fundamental tradeoffs of new initiatives to generate large, shared datasets in neuroimaging (e.g., HCP, the Adolescent Brain and Cognitive Development study [ABCD], the UK BioBank). However, because of this ‘open science’ approach the current study was able to perform the largest investigation to date of the association between cannabis use and the subfields of hippocampus. Furthermore, it used multiple cannabis use variables to determine which aspects of cannabis use were most linked to differences in these regions. Additionally, this is one of the first studies to investigate the effects of cannabis use on hippocampal and amygdalar subfields drawn using the new Freesurfer 6.0 segmentation procedure21,22 which improves upon issues in prior versions of Freesurfer in segmenting these structures20.

In summary, the current study demonstrates small but significant inverse associations of recent cannabis use with the volume of the hippocampus, with subfield analyses finding significant association recent use with the head of the hippocampus, CA1, subicular complex, and molecular layer. These results differ from much of the prior literature in meaningful ways, perhaps due to higher resolution methods, but nonetheless demonstrate that cannabis’ effects on the hippocampus are specific to certain subfields of the hippocampus and may dissipate over time.

Supplementary Material

Supplemental Materials

Figure 2.

Figure 2.

Figure 2.

Figure 2.

Bar graph of mean volume of structures with significant differences between THC+ and THC−.

ACKNOWLEDGMENTS

These data are from the Human Connectome Project, WU-Minn Consortium (Principal Investigators: David Van Essen and Kamil Ugurbil; 1U54MH091657) funded by the 16 NIH Institutes and Centers that support the NIH Blueprint for Neuroscience Research; and by the McDonnell Center for Systems Neuroscience at Washington University in St. Louis. The authors are deeply appreciative to the Human Connectome Project for open access to its data. In addition, the work was supported by NIH grant R01 AA025911 (PIs: JM & LS), the Peter Boris Chair in Addictions Research (JM) and the Gary Sperduto Endowed Professorship in Clinical Psychology (LS). No funding sources were involved in study design or collection, analysis, and interpretation of the data. These findings do not reflect the official position of the National Institutes of Health.

CONFLICTS OF INTEREST

MMO has no potential financial conflicts of interest to declare. LHS receives research funding from the National Institutes of Health. JM receives research funding from the National Institutes of Health, the Canadian Institutes of Health Research, and Correctional Service of Canada, and is a principal in BEAM Diagnostics, Inc.

REFERENCES

  • 1.Winstock A, Barratt M, Ferris J, Maier L. Global Drug Survey 2017. Glob Drug Surv. 2017. [Google Scholar]
  • 2.ElSohly MA, Mehmedic Z, Foster S, Gon C, Chandra S, Church JC. Changes in cannabis potency over the last 2 decades (1995–2014): Analysis of current data in the United States. Biol Psychiatry. 2016. doi: 10.1016/j.biopsych.2016.01.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Lorenzetti V, Solowij N, Yücel M. The role of cannabinoids in neuroanatomic alterations in cannabis users. Biol Psychiatry. 2016. doi: 10.1016/j.biopsych.2015.11.013 [DOI] [PubMed] [Google Scholar]
  • 4.Nader DA, Sanchez ZM. Effects of regular cannabis use on neurocognition, brain structure, and function: a systematic review of findings in adults. Am J Drug Alcohol Abuse. 2018. doi: 10.1080/00952990.2017.1306746 [DOI] [PubMed] [Google Scholar]
  • 5.Cousijn J, Wiers RW, Ridderinkhof KR, Van den Brink W, Veltman DJ, Goudriaan AE. Grey matter alterations associated with cannabis use: Results of a VBM study in heavy cannabis users and healthy controls. Neuroimage. 2012;59(4):3845–3851. doi: 10.1016/j.neuroimage.2011.09.046 [DOI] [PubMed] [Google Scholar]
  • 6.Schacht JP, Hutchison KE, Filbey FM. Associations between cannabinoid receptor-1 (CNR1) variation and hippocampus and amygdala volumes in heavy cannabis users. Neuropsychopharmacology. 2012;37(11):2368–2376. doi: 10.1038/npp.2012.92 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Lorenzetti V, Solowij N, Whittle S, et al. Gross morphological brain changes with chronic, heavy cannabis use. Br J Psychiatry. 2015;206(1):77–78. doi: 10.1192/bjp.bp.114.151407 [DOI] [PubMed] [Google Scholar]
  • 8.Yücel M, Lorenzetti V, Suo C, et al. Hippocampal harms, protection and recovery following regular cannabis use. Transl Psychiatry. 2016;6(1):e710. doi: 10.1038/tp.2015.201 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Mackey S, Allgaier N, Chaarani B, et al. Mega-Analysis of Gray Matter Volume in Substance Dependence: General and Substance-Specific Regional Effects. Am J Psychiatry. 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Koenders L, Lorenzetti V, De Haan L, et al. Longitudinal study of hippocampal volumes in heavy cannabis users. J Psychopharmacol. 2017. doi: 10.1177/0269881117718380 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Duvernoy HM. The Human Hippocampus: Functional Anatomy, Vascularization and Serial Sections with MRI. 4th ed. Springer Science & Business Media; 2013. [Google Scholar]
  • 12.Yau SY, Li A, So KF. Involvement of Adult Hippocampal Neurogenesis in Learning and Forgetting. Neural Plast. 2015. doi: 10.1155/2015/717958 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Knierim JJ. The hippocampus. Curr Biol. 2015. doi: 10.1016/j.cub.2015.10.049 [DOI] [PubMed] [Google Scholar]
  • 14.Nakazawa K, McHugh TJ, Wilson MA, Tonegawa S. NMDA receptors, place cells and hippocampal spatial memory. Nat Rev Neurosci. 2004;5(5):361–372. doi: 10.1038/nrn1385 [DOI] [PubMed] [Google Scholar]
  • 15.O’Reilly RC, Rudy JW. Conjunctive representations in learning and memory: Principles of cortical and hippocampal function. Psychol Rev. 2001;108(2):311–345. doi: 10.1037/0033-295X.108.2.311 [DOI] [PubMed] [Google Scholar]
  • 16.Hunsaker MR, Kesner RP. Evaluating the differential roles of the dorsal dentate gyrus, dorsal CA3, and dorsal CA1 during a temporal ordering for spatial locations task. Hippocampus. 2008. doi: 10.1002/hipo.20455 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Kesner RP, Rolls ET. A computational theory of hippocampal function, and tests of the theory: New developments. Neurosci Biobehav Rev. 2015. doi: 10.1016/j.neubiorev.2014.11.009 [DOI] [PubMed] [Google Scholar]
  • 18.Kesner RP, Hunsaker MR. The temporal attributes of episodic memory. Behav Brain Res. 2010;215(2):299–309. doi: 10.1016/j.bbr.2009.12.029 [DOI] [PubMed] [Google Scholar]
  • 19.Chye Y, Suo C, Yücel M, den Ouden L, Solowij N, Lorenzetti V. Cannabis-related hippocampal volumetric abnormalities specific to subregions in dependent users. Psychopharmacology (Berl). 2017. doi: 10.1007/s00213-017-4620-y [DOI] [PubMed] [Google Scholar]
  • 20.Wisse LEM, Biessels GJ, Geerlings MI. A critical appraisal of the hippocampal subfield segmentation package in FreeSurfer. Front Aging Neurosci. 2014. doi: 10.3389/fnagi.2014.00261 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Saygin ZM, Kliemann D, Iglesias JE, et al. High-resolution magnetic resonance imaging reveals nuclei of the human amygdala: manual segmentation to automatic atlas. Neuroimage. 2017;155:370–382. doi: 10.1016/j.neuroimage.2017.04.046 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Iglesias JE, Augustinack JC, Nguyen K, et al. A computational atlas of the hippocampal formation using ex vivo, ultra-high resolution MRI: Application to adaptive segmentation of in vivo MRI. Neuroimage. 2015;115:117–137. doi: 10.1016/j.neuroimage.2015.04.042 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Beale C, Broyd SJ, Chye Y, et al. Prolonged Cannabidiol Treatment Effects on Hippocampal Subfield Volumes in Current Cannabis Users. Cannabis Cannabinoid Res. 2018;3(1):94–107. doi: 10.1089/can.2017.0047 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Gilman JM, Kuster JK, Lee S, et al. Cannabis Use Is Quantitatively Associated with Nucleus Accumbens and Amygdala Abnormalities in Young Adult Recreational Users. J Neurosci. 2014. doi: 10.1523/JNEUROSCI.4745-13.2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Lorenzetti V, Solowij N, Fornito A, et al. The impact of regular cannabis use on human brain structure. Eur Neuropsychopharmacol. 2012. [Google Scholar]
  • 26.Lorenzetti V, Solowij N, Fornito A, et al. Brain morphology in cannabis users: Region-dependent reductions and role of use patterns. Biol Psychiatry. 2013. [Google Scholar]
  • 27.Sah P, Faber ESL, Lopez De Armentia M, Power J. The Amygdaloid Complex: Anatomy and Physiology. Physiol Rev. 2003. doi: 10.1152/physrev.00002.2003 [DOI] [PubMed] [Google Scholar]
  • 28.LeDoux J The amygdala. Curr Biol. 2007. doi: 10.1016/j.cub.2007.08.005 [DOI] [Google Scholar]
  • 29.Gillespie NA, Neale MC, Bates TC, et al. Testing associations between cannabis use and subcortical volumes in two large population-based samples. Addiction. 2018;113(9):1661–1672. doi: 10.1111/add.14252 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Van Essen DC, Smith SM, Barch DM, Behrens TEJ, Yacoub E, Ugurbil K. The WU-Minn Human Connectome Project: An overview. Neuroimage. 2013;80:62–79. doi: 10.1016/j.neuroimage.2013.05.041 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Bucholz KK, Cadoret R, Cloninger CR, et al. A new, semi-structured psychiatric interview for use in genetic linkage studies: a report on the reliability of the SSAGA. J Stud Alcohol. 1994;55(2):149–158. doi: 10.15288/jsa.1994.55.149 [DOI] [PubMed] [Google Scholar]
  • 32.Musshoff F, Madea B. Review of biologic matrices (urine, blood, hair) as indicators of recent or ongoing cannabis use. Ther Drug Monit. 2006;28(2):155–163. doi: 10.1097/01.ftd.0000197091.07807.22 [DOI] [PubMed] [Google Scholar]
  • 33.Marcus DS, Harms MP, Snyder AZ, et al. Human Connectome Project informatics: Quality control, database services, and data visualization. Neuroimage. 2013;80:202–219. doi: 10.1016/j.neuroimage.2013.05.077 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Dale AM, Fischl B, Sereno MI. Cortical surface-based analysis. I. Segmentation and surface reconstruction. Neuroimage. 1999;9(2):179–194. doi: 10.1006/nimg.1998.0395 [DOI] [PubMed] [Google Scholar]
  • 35.Fischl B, Sereno MI, Dale AM. Cortical Surface-Based Analysis: II: Inflation, Flattening, and a Surface-Based Coordinate System. Neuroimage. 1999;9(2):195–207. doi: 10.1006/nimg.1998.0396 [DOI] [PubMed] [Google Scholar]
  • 36.Glasser MF, Sotiropoulos SN, Wilson JA, et al. The minimal preprocessing pipelines for the Human Connectome Project. Neuroimage. 2013;80:105–124. doi: 10.1016/j.neuroimage.2013.04.127 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Saygin ZM, Kliemann D, Iglesias JE, et al. High-resolution magnetic resonance imaging reveals nuclei of the human amygdala: manual segmentation to automatic atlas. Neuroimage. 2017;155:370–382. doi: 10.1016/J.NEUROIMAGE.2017.04.046 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Benjamin DJ, Berger JO, Johannesson M, et al. Redefine statistical significance. Nat Hum Behav. 2018. doi: 10.1038/s41562-017-0189-z [DOI] [PubMed] [Google Scholar]
  • 39.O’Mara SM, Commins S, Anderson M, Gigg J. The subiculum: A review of form, physiology and function. Prog Neurobiol. 2001;64(2):129–155. doi: 10.1016/S0301-0082(00)00054-X [DOI] [PubMed] [Google Scholar]
  • 40.Matsumoto N, Kitanishi T, Mizuseki K. The subiculum: Unique hippocampal hub and more. Neurosci Res. 2019. doi: 10.1016/j.neures.2018.08.002 [DOI] [PubMed] [Google Scholar]
  • 41.Pagliaccio D, Barch DM, Bogdan R, et al. Shared Predisposition in the Association Between Cannabis Use and Subcortical Brain Structure. JAMA psychiatry. 2015;72(10):994–1001. doi: 10.1001/jamapsychiatry.2015.1054 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Orr JM, Paschall CJ, Banich MT. Recreational marijuana use impacts white matter integrity and subcortical (but not cortical) morphometry. NeuroImage Clin. 2016;12:47–56. doi: 10.1016/j.nicl.2016.06.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Button KS, Ioannidis JPA, Mokrysz C, et al. Power failure: why small sample size undermines the reliability of neuroscience. Nat Rev Neurosci. 2013;14(5):365–376. doi: 10.1038/nrn3475 [DOI] [PubMed] [Google Scholar]
  • 44.Petker T, Owens M, Amlung M, Oshri A, Sweet L, MacKillop J. Cannabis involvement and neuropsychological performance: findings from the Human Connectome Project. J psychiatry Neurosci. 2019;44(5):1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Scott JC, Slomiak ST, Jones JD, Rosen AF, Moore TM, Gur RC. Association of Cannabis With Cognitive Functioning in Adolescents and Young Adults: A Systematic Review and Meta-analysis. JAMA Psychiatry. 2018. doi:doi: 10.1001/jamapsychiatry.2018.0335 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Materials

RESOURCES