Long-term effects of marijuana use on the brain

Francesca M. Filbeya,1, Sina Aslana,b, Vince D. Calhounc,d, Jeffrey S. Spencea, Eswar Damarajuc, Arvind Caprihanc,
and Judith Segallc
a
Center for BrainHealth, University of Texas, Dallas, TX 75235; bAdvance MRI, LLC, Frisco, TX 75034; cThe Mind Research Network, Albuquerque, NM 87106;
and dUniversity of New Mexico, Albuquerque, NM 87131

Edited by Cameron Carter, University of California Davis Centre for Neuroscience, Sacramento, CA, and accepted by the Editorial Board October 13, 2014
(received for review August 8, 2014)

MRI orbitofrontal cortex

diffusion tensor imaging

| functional connectivity | resting state fMRI |

he rate of marijuana use has had a steady increase since 2007

(1). Among >400 chemical compounds, marijuanas effects
are primarily attributed to -9-tetrahydrocannabinol (THC),
which is the main psychoactive ingredient in the cannabis plant.
THC binds to cannabinoid receptors, which are ubiquitous in the
brain. Consequently, exposure to THC leads to neural changes
affecting diverse cognitive processes. These changes have been
observed to be long-lasting, suggesting that neural changes due
to marijuana use may affect neural architecture (2). However, to
date, these brain changes as a result of marijuana use remains
equivocal. Specifically, although functional changes have been
widely reported across cognitive domains in both adult and
adolescent cannabis users (36), structural changes associated with
marijuana use have not been consistent. Although some have
reported decreases in regional brain volume such as in the hippocampus, orbitofrontal cortex, amygdala, and striatum (712),
others have reported increases in amygdala, nucleus accumbens,
and cerebellar volumes in chronic marijuana users (1315).
However, others have reported no observable difference in
global or regional gray or white matter volumes in chronic
marijuana users (16, 17). These inconsistencies could be attributed to methodological differences across studies pertaining to
study samples (e.g., severity of marijuana use, age, sex, comorbidity with other substance use or psychiatric disorders) and/or
study design (e.g., study modality, regions of interest).
Because THC binds to cannabinoid 1 (CB1) receptors in the
brain, when differences are observed, these morphological
changes associated with marijuana use have been reported in
www.pnas.org/cgi/doi/10.1073/pnas.1415297111

CB1 receptor-enriched areas such as the orbitofrontal cortex,

anterior cingulate, striatum, amygdala, insula, hippocampus, and
cerebellum (2, 11, 13, 18). CB1 receptors are widely distributed
in the neocortex, but more restricted in the hindbrain and the
spinal cord (19). For example, in a recent study by Battistella
et al. (18), they found significant brain volume reductions in the
medial temporal cortex, temporal pole, parahippocampal gyrus,
insula, and orbitofrontal cortex (OFC) in regular marijuana users
compared with occasional users. Whether these reductions in
brain volume lead to downstream changes in brain organization
and function, however, is still unknown.
Nevertheless, emergent studies have demonstrated a link between brain structure and connectivity. For example, Van den
Heuvel et al. and Greicius et al. demonstrated robust structural
connections between white matter indexes and functional connectivity strength within the default mode network (20, 21).
Similarly, others have reported correlated patterns of gray matter structure and connectivity that are in many ways reflective of
the underlying intrinsic networks (22). Thus, given the literature
suggesting a direct relationship between structural and functional
connectivity, it is likely that connectivity changes would also be
present where alterations in brain volume are observed as a result of marijuana use.
The goal of this study was to characterize alterations in brain
morphometry and determine potential downstream effects in
connectivity as a result of chronic marijuana use. To address the
existing inconsistencies in the literature that may be in part due
to methodological issues, we (i) used three different MRI techniques to investigate a large cohort of well-characterized chronic
cannabis users with a wide age range (allowing for characterization without developmental or maturational biases) and
compared them to age- and sex-matched nonusing controls;
(ii) examined observable global (rather than select) gray matter
Significance
The existing literature on the long-term effects of marijuana on
the brain provides an inconsistent picture (i.e., presence or
absence of structural changes) due to methodological differences across studies. We overcame these methodological issues
by collecting multimodal measures in a large group of chronic
marijuana using adults with a wide age range that allows
for characterization of changes across lifespan without developmental or maturational biases as in other studies. Our
findings suggest that chronic marijuana use is associated with
complex neuroadaptive processes and that onset and duration
of use have unique effects on these processes.
Author contributions: F.M.F. and V.D.C. designed research; F.M.F. and J.S. performed research; S.A., J.S.S., E.D., and A.C. analyzed data; and F.M.F. and S.A. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. C.C. is a guest editor invited by the Editorial
Board.
Freely available online through the PNAS open access option.
1

To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.

1073/pnas.1415297111/-/DCSupplemental.

PNAS | November 25, 2014 | vol. 111 | no. 47 | 1691316918

NEUROSCIENCE

Questions surrounding the effects of chronic marijuana use on

brain structure continue to increase. To date, however, findings
remain inconclusive. In this comprehensive study that aimed to
characterize brain alterations associated with chronic marijuana
use, we measured gray matter (GM) volume via structural MRI
across the whole brain by using voxel-based morphology, synchrony among abnormal GM regions during resting state via
functional connectivity MRI, and white matter integrity (i.e.,
structural connectivity) between the abnormal GM regions via
diffusion tensor imaging in 48 marijuana users and 62 age- and
sex-matched nonusing controls. The results showed that compared with controls, marijuana users had significantly less bilateral
orbitofrontal gyri volume, higher functional connectivity in the
orbitofrontal cortex (OFC) network, and higher structural connectivity in tracts that innervate the OFC (forceps minor) as measured
by fractional anisotropy (FA). Increased OFC functional connectivity in marijuana users was associated with earlier age of onset.
Lastly, a quadratic trend was observed suggesting that the FA
of the forceps minor tract initially increased following regular
marijuana use but decreased with protracted regular use. This
pattern may indicate differential effects of initial and chronic
marijuana use that may reflect complex neuroadaptive processes
in response to marijuana use. Despite the observed age of onset
effects, longitudinal studies are needed to determine causality of
these effects.

Table 1. Subject characteristics (mean SD)

Demographic variable
Participants (n)
Sex (M/F)
Age
Race
White
Latino
Native American
Black
Asian
Education, y
IQ*
Age of onset
Years of use
Weekly use
Daily use
Marijuana-related
problems (MPS)
Marijuana
dependence (N)

Control

Cannabis users

Exclusively
cannabis users

62
39/23
30.0 7.4

48
33/15
28.3 8.3

27
17/10
28.1 8.9

25
27
5
4
1
13.9 1.7
110.9 11.6

24
16
3
3
0
14.2
105.8
18.1
9.8
11.1
3.1
3.4

2.4
12.2
3.4
8.0
1.4
1.6
4.1

25

11
11
2
3
0
14.3
104.0
18.7
8.7
11.2
2.8
2.5

2.5
1.4
2.9
8.7
1.4
1.4
2.4

14

*Control and cannabis users IQ were different (P < 0.05). IQ was assessed by
using the Wechsler Abbreviated Scale of Intelligence (54).
Marijuana dependence was assessed via Structured Clinical Interview for
DSM IV Disorders (55).

differences between marijuana users and nonusing controls; and

(iii) performed subsequent analyses to determine how these
changes relate to functional and structural connectivity, as well
as behavior. Given the existing literature on morphometric
reductions associated with long-term marijuana use, we expected
gray matter reductions in THC-enriched areas in chronic marijuana users that will be associated with changes in brain connectivity and marijuana-related behavior.
Methods
Participants. A total of 110 participants consisting of 62 nonusing controls and
48 marijuana users were recruited through fliers and media advertisement in
the Albuquerque, NM, metro area. We previously presented results on
subgroups of these participants (8, 23, 24). Written informed consent was
obtained from all participants in accordance with the Institutional Review
Board (IRB) of The University of New Mexico. The inclusion criteria for all of
the participants were as follows: (i) English as the primary language; and (ii)
no current or history of psychosis, traumatic brain injury, or neurological disorder. Marijuana users (cannabis group) were included if they currently use
marijuana regularly (at least four times per week) over the last 6 mo (confirmed via positive THC-COOH urinalysis). Nonusing controls (control group)
had no self-reported regular use of marijuana and had a negative urine drug
screen at baseline. Table 1 summarizes the demographic information, behavioral measures, and total number of participants per cohort.
MRI Acquisition. MRI scans were performed on a Siemens 3 Tesla Trio scanner
by using the standard 12-channel phased array head coil. We used different
MRI techniques to investigate brain changes between cannabis users and
control groups: (i) a high resolution T1-weighted image to measure gray
matter volume, (ii) a resting state functional MRI scan was collected to assess
functional connectivity of the brain, and (iii) a diffusion tensor imaging scan
was collected to provide an assessment of structural connectivity between
brain regions via white matter tracts. The details of the imaging parameters
and their processing techniques are provided below:
Whole brain high-resolution T1-weighted anatomical images were collected by using a multiecho magnetization prepared rapid gradient echo
(MPRAGE) sequence with the following parameters: repetition time (TR)/echo
time (TE)/inversion time (TI) = 2,530/1.64, 3.5, 5.36, 7.22, 9.08/1,200 ms, flip
angle = 7, field of view (FOV) = 256 256 192 mm3, voxel size = 1 1
1 mm3, and number of excitations (NEX) = 1. The sequence parameters for
functional MRI (fcMRI) were: FOV = 240 240, matrix = 64 64, slice
thickness = 4.55 mm, no gap between slices, voxel size = 3.75 3.75 4.55 mm2,

16914 | www.pnas.org/cgi/doi/10.1073/pnas.1415297111

32 axial slices, TR/TE = 2,000/29 ms, flip angle = 60, 158 image volumes, and
scan duration = 5.5 min. The diffusion tensor imaging (DTI) MRI scans (b =
800 s/mm2) were acquired by using a twice-refocused spin echo sequence
with 30 diffusion gradients and the b = 0 experiment repeated five times
with the following parameters: TE/TR = 84/9,000 ms, flip angle = 90, FOV =
256 256 144 mm3, voxel resolution = 2 2 2 mm3 and NEX = 1. The
sequence parameters for fcMRI were FOV = 240 240 mm2, matrix = 64
64, slice thickness = 4.55 mm, voxel size = 3.75 3.75 4.55 mm3, 31 axial
slices, TR/TE = 2,000/29 ms, flip angle = 90, 158 image volumes, and scan
duration = 5.3 min.
MRI Data Processing. We used the voxel-based morphology (VBM) technique
to investigate whole brain structural abnormalities. High-resolution T1
images were processed by using the Diffeomorphic Anatomical Registration
Through Exponentiated Lie Algebra (DARTEL), an improved VBM method
that can achieve intersubject brain images registration more accurately in
SPM 8 (www.fil.ion.ucl.ac.uk/spm). Briefly, the following steps were performed on the T1 images: (i) MR images were segmented into gray matter
(GM), white matter (WM), and cerebrospinal fluid; (ii) customized GM
templates were created from the images of study by using DARTEL technique; (iii) after an initial affine registration of the GM DARTEL templates to
the tissue probability maps in MNI space, nonlinear warping of GM images
was performed to the DARTEL GM template and then used in the modulation step to ensure that relative volumes of GM were preserved following
the spatial normalization procedure; (iv) the modulated, normalized GM
images were smoothed with a 6-mm full width at half maximum (FWHM).
Next, we conducted a two-sample t test with intelligence quotient (IQ) as
a covariate. A voxel level threshold of P < 0.01 (FWE-corrected) and cluster
size 15,936 mm3 was determined based on AFNI softwares 3dClustSim
[National Institute of Mental Health (NIMH) Scientific and Statistical Computing Core]. For the analyses exploring relationships between activation
maps and behavioral measures, 10-mm sphere masks were defined around
the peak voxels of the significant gray matter clusters.
Resting state fMRI (rsfMRI) images were analyzed by using AFNI (NIMH
Scientific and Statistical Computing Core) and in-house MATLAB scripts. The
dataset was preprocessed with motion correction (realignment), slice timing
correction, removal of the linear trend, transformation to standard MNI space
(matrix = 53 63 46, resolution = 3 3 3 mm3), and smoothing by
a Gaussian filter with a full width half maximum (FWHM) of 10 mm. Next,
the images were band-pass filtered (0.010.1 Hz) on a voxel-by-voxel basis to
keep only the appropriate frequency fluctuations. Next, the signals in white
matter and cerebrospinal fluid were regressed out by using averaged signals
from the white matter and the ventricles from each voxel time series.
Functional connectivity was measured by using a seed-based approach by
choosing bilateral orbitofrontal gyri cluster peaks from VBM analysis, [+26
+54 8] and [16 +58 10] in MNI template. The cross-correlation coefficient
between these seed voxels and all other voxels was calculated to generate
a correlation map. Then, the correlation maps were transformed to a z-score
map by using Fishers inverse hyperbolic tangent transformation. Next,
a region of interest (ROI) analysis was performed. Each regions anatomical
region was defined based on automated anatomical labeling (AAL) database. Then, the orbitofrontal and temporal functional masks were defined
as the top 200 voxels according to their z score in their functional connectivity maps as described by Chapman and coworkers (25).
Diffusion-weighted data were processed by using the University of
Oxfords Center for Functional Magnetic Resonance Imaging of the Brain
Software Library release 4.0 (www.fmrib.ox.ac.uk/fsl). First, the data were
corrected for head movement and eddy current distortions by using Eddycorrect,

Fig. 1. Group comparison of the gray matter volume by SPM8 plus DARTEL
analysis demonstrates significant reduction of gray matter volume in bilateral orbitofrontal gyri (AAL atlas) in marijuana users compared with controls.
Right side of the image represents the right hemisphere in axial view.

Filbey et al.

users was significantly lower than the control group (P < 0.05).
Table 1 summarizes participants demographic data.

Fig. 2. (A) The average functional connectivity maps (i.e., OFC network;
bilateral OFC and temporal gyri) of the control and cannabis groups are
superimposed on their average T1-weighted image. For illustration purposes,
the z-score maps were arbitrarily thresholded (z score 2, k 50) to qualitatively visualize the difference in the intensity and cluster size. (B) Mean
fcMRI z scores are shown for the orbitofrontal network for cannabis and
controls groups. The cannabis group showed higher resting activity in the
bilateral OFC and temporal gyri compared with the control group.

which aligns all of the volumes. Next, DTIfit was used to independently fit
diffusion tensors to each voxel, with the brain mask limiting the fitting of
tensors to brain space. The output of DTIfit yielded voxel-wise maps of
fractional anisotropy (FA), axial diffusivity (AD) (1), radial diffusivity (RD)
(average of 2 and 3) and mean diffusivity (MD) (average of 1, 2, and 3)
for each participant. Finally, in the tractography analysis, white matter tracts
were constructed with minimum FA of 0.20 and maximum turning angle of
50. Because the orbitofrontal cortex is innervated by the forceps minor,
and, therefore, plays a role in decision-making processes, the forceps minor
tract was delineated via two techniquesmanual and automatic tractographywhereas the forceps major tract was delineated only manually as
a control. In manual tractography, the forceps minor and forceps major
tracts were delineated by drawing manual ROI per methods described in
Wakana et al. (26). In automatic tractography, the VBM clusters (see Fig. 3)
were coregistered to each participants native DTI space and used as an ROI
to delineate the fiber tract. Specifically, these regions were dilated five
times, using 3dAutomask in AFNI, to ensure the clusters were expanded into
the white matter tissue. Last, an AND operation between the two clusters
was performed and the resultant fiber was the forceps minor from the right
middle orbitofrontal and left superior orbitofrontal gyri.

MRI Measurements. Voxelwise comparison of the high-resolution

T1 images showed a significant lower gray matter volume in
marijuana users in the right middle orbitofrontal (MNI coordinates: [+26 +54 8]; t score = 3.37) and left superior orbitofrontal gyri (MNI coordinates: [16 +58 10]; t score = 3.19)
[P < 0.01 (FWE corrected) and cluster 15,936 mm3] per anatomical automatic labeling (AAL), shown in Fig. 1. The reverse
contrast, marijuana > control, did not yield any significant voxels.
Following these observed structural alterations in the orbitofrontal region, we then characterized the functional connectivity of
the orbitofrontal network. The components of this network consist
of bilateral orbitofrontal and bilateral temporal gyri (28). Fig. 2A
shows the average functional connectivity maps in the orbitofrontal
network for the control and cannabis groups. These maps show
qualitative differences between the groups such that the cannabis
group had higher functional connectivity compared with the control
group. Fig. 2B shows that, quantitatively, marijuana users had significantly higher connectivity in all four nodes (i.e., bilateral OFC
and bilateral temporal lobe) compared to the control group.
We also measured the structural connectivity of the forceps
minor tract, which connects the orbitofrontal regions, both manually and automatically (shown in Fig. 3). We found that the
forceps minors FA of the cannabis group was significantly higher
than the control group in both automatic and manual methods,
P = 0.003 and P < 0.001, respectively. As a manipulation check, we
also measured the FA of the forceps major, which did not show
any significant difference between control and cannabis groups.
Additionally, we examined which component of FA may be driving this effect. To that end, we also carried out one-way ANOVA
comparisons of mean diffusivity (MD), radial diffusivity (RD), and
axial diffusivity (AD) between the groups. We found that RD of
the cannabis users was significantly lower than that of the controls

Behavioral Measures. Behavior related to marijuana use was captured by

using the Marijuana Problem Survey (MPS) (27). The MPS is a 19-item measure that assesses the negative psychological, social, occupational, and legal
consequences of marijuana use in the last 90 d (e.g., problems with family
and significant others, missing work or losing a job, feeling bad about
marijuana use). Each problem is rated from 0 (no problem) to 2 (serious
problem), and the number of items endorsed as 1 or 2 is summed to create
an index of the total number of problems (range = 019). Treatment-seeking
marijuana users report an average of 910 problems.
Statistical Analysis. A general statistical linear model was applied to assess the
contribution of chronic marijuana use on cognition, gray matter volume,
functional connectivity, and structural connectivity measures. The model
included two groups (i.e., marijuana users and controls) and IQ as a covariate.
Two sample t tests were performed to assess to identify how groups differed
in the aforementioned measures, and we hypothesized that the cannabis
group would show alterations in gray matter volume, functional connectivity, and structural connectivity. Last, parametric regression models were
tested to examine the relationship among gray matter volume, functional
connectivity, white matter integrity, and neurocognitive measures within
the cannabis group. To ensure the best parametric regression fit, we performed Akaike information criterion (AIC) to provide a means for model
selection (i.e., linear vs. quadratic).

Sample Characteristics. All MR images were visually inspected for

possible artifacts. Of 110 participants, three control participants
did not complete the functional connectivity MRI protocol. Nine
participants (seven in the control group; two in the cannabis
group) did not produce the forceps minor tract via the automated technique. No participant was excluded based on motion
criteria of >3 mm and >3. There was no significant difference in
age or sex between the groups. However, the IQ of the marijuana
Filbey et al.

Fig. 3. A representative participants forceps minor tract (in red) and gray
matter nodes (in blue) is overlaid on its corresponding fractional anisotropy map.

PNAS | November 25, 2014 | vol. 111 | no. 47 | 16915

NEUROSCIENCE

Results

Table 2. Forceps minor and forceps major tracts DTI parameters

(mean SD)
DTI measures
Forceps minor
(automated)
FA
AD
RD
MD
Forceps minor
(manual)
FA
AD
RD
MD
Forceps major
(manual)
FA
AD
RD
MD

Control

Cannabis

P value

0.551
1.39e03
5.32e04
8.19e04

0.028
6.68e05
4.60e05
4.79e05

0.570
1.40e03
5.14e04
8.09e04

0.032
4.27e05
4.11e05
3.45e05

0.003
0.57
0.05
0.25

0.549
1.34e03
5.19e04
7.94e04

0.025
5.36e05
3.47e05
3.53e05

0.569
1.35e03
4.99e04
7.82e02

0.025
3.56e05
3.14e05
2.75e05

<0.001
0.77
0.004
0.051

0.643
1.59e03
4.70e04
8.41e04

0.029
5.75e05
3.84e05
3.27e05

0.651
1.59e03
4.60e04
8.34e04

0.020
5.38e05
3.14e05
3.22e05

0.07
0.88
0.09
0.17

in both automatic and manual tractography, P = 0.05 and P =

0.004, respectively (shown in Table 2). Differences between
AD and MD were not significant between the groups.
BrainBehavior Correlations. Our analyses showed significant correlations between forceps minor integrity/OFC functional connectivity and behavioral indicators of marijuana use (Table 3).
The forceps minors FA and RD showed gains with initial
heavy use but declined after chronic use, as shown in Fig. 4. The
functional connectivity of the bilateral OFC showed similar
patterns where there was an inverse correlation with age of onset
of use such that earlier age of onset leads to higher functional
connectivity of the bilateral OFC. Finally, there was an inverse
correlation between the left temporal lobe functional connectivity and problems related with marijuana use, such that the
greater the functional connectivity of the left temporal cortex to
bilateral OFC, the lower the total score on MPS.
Post Hoc Analyses. Because of the high comorbidity between
marijuana, tobacco, and alcohol use, we performed additional
analyses to control for potential confounding effects of tobacco
and alcohol use. In these analyses, we excluded marijuana users
who reported use of other substances. This resulted in 27 exclusively marijuana users as shown in Table 1. The association
between the neuroimaging results and behavioral measures were
also assessed as described previously. Similar to the main findings, the exclusively marijuana users showed significantly lower
OFC gray matter density, and significantly higher OFC network
functional connectivity and FA/RD of the forceps minor tract

compared with the control group (Figs. S1 and S2 and Table S1).
There was also a quadratic trend in the correlation between the
forceps minors FA/RD and use duration, t score = 2.05, P =
0.05 and t score = 2.60, P = 0.016, respectively (Table 4). Additionally, the exclusively marijuana users showed an inverse
relationship between bilateral OFC gray matter volume and
problems related to marijuana use. That is, the lower the OFC
gray matter volume in these participants, the higher their MPS
total scores.
Lastly, to partially address how these abnormalities are related
to cognitive processes, we conducted a mediation analysis to
assess whether neural abnormalities (OFC gray matter volumes,
OFC/temporal lobe functional connectivity, FA/RD of forceps
minor) mediate lower scores on IQ in marijuana users. We did
not find that the causal variable (i.e., marijuana use) was
significantly correlated with the mediator variable (i.e., OFC
gray matter volume, OFC/temporal functional connectivity, and
FA/RD of forceps minor) and outcome variable (i.e., IQ). We,
therefore, suggest that the path from marijuana use to neural
abnormalities to decreases in IQ is more complex and, perhaps,
include other mediators such as environmental (i.e., age of onset) and/or genetic factors.
Discussion
Unlike the animal literature, whether exposure to marijuana
leads to long-term changes in human brain structure has been
equivocal. To address this limitation, we evaluated brain
structural changes associated with chronic marijuana use in
a large group of well-characterized marijuana users relative to
age- and sex-matched nonusing controls. Our findings provide
evidence that heavy, chronic marijuana users have lower OFC
gray matter volumes compared with nonusing controls. This
finding remained even in the smaller sample of exclusively
marijuana users (n = 27, i.e., no comorbid substance use),
demonstrating that this effect (i) is robust and (ii) is greater
than potential effects of comorbid substance use. Similar
decreases in OFC volume have been reported in marijuana using
adults (29) and adolescents (12) compared with nonusing controls. Interestingly, a prospective study also found that smaller
OFC volumes at 12 y of age predicted initiation of marijuana use
at 16 y of age (30). These effects on the OFC are not surprising
given that the OFC is a primary region in the reward network, is
enriched with CB1 receptors, and is highly implicated in addictive behaviors (23, 24, 31, 32) such as those related to disruptions
in motivation (33) and decision making (34, 35). Whereas others
have reported alterations in various CB1-enriched regions such
as the amygdala, hippocampus, ventromedial prefrontal, OFC,
insula, and striatum, our findings are specific to the OFC. Several animal and human studies have demonstrated greater THCinduced down-regulation of CB1 receptors in cortical areas relative to subcortical areas, which support our findings. Given that
CB1 receptors are found on excitatory terminals of cortical
projection neurons, this alteration in endocannabinoid signaling

Table 3. Associations between gray matter volume, functional connectivity, white matter
integrity, and neurocognitive measures within the cannabis group
Neuroimaging measures
DTI MRI
Forceps minorFA
Forceps minorRD
fcMRI
Left orbitofrontal cortex
Right orbitofrontal cortex
Left temporal cortex

Behavioral measures

Parametric model

t score

P value

Duration of use
Duration of use

Quadratic
Quadratic

2.7
2.5

0.009
0.015

Linear
Linear
Linear

2.1
2.1
2.7

0.046
0.042
0.010

Age of onset
Age of onset
MPS total

The curve fitting for the general linear model was controlled by AIC.

16916 | www.pnas.org/cgi/doi/10.1073/pnas.1415297111

Filbey et al.

Fig. 4. The relationship between duration of marijuana use and forceps

minors FA (A) and RD (B). The quadratic curve showed the best fit per AIC.
The x axis has been transformed to square root of years of use because of
gap between participants years of use.

could affect the plasticity of OFC circuits (36). Unfortunately,

the cross-sectional nature of the present study cannot directly
address whether these reductions are the cause or the consequence of marijuana use. However, neurotoxic effects of cannabis have been widely reported in the animal literature. Based
on the animal literature, potential mechanisms that may lead to
OFC reductions due to cannabis neurotoxicity may, therefore,
include neuronal loss, changes in cell size, or a reduction in CB1
density. It is possible, however, that these OFC abnormalities
may reflect preexisting pathophysiology related to vulnerability
to marijuana abuse and dependence.
To determine the potential downstream effects of OFC volume
reduction, we evaluated OFC functional (fcMRI) and structural
connectivity (DTI). Functional connectivity analysis revealed
greater connectivity within the OFC network in marijuana users
compared with controls, which is concordant with existing restingstate studies (37) and task-based studies (38, 39). This increased
functional connectivity in users may suggest a compensatory
mechanism whereby greater network recruitment is engaged to
compensate for OFC liability (40). Tomasi et al. (41) illustrated
how greater functional connectivity requires higher glucose consumption (70% of brains energy consumption), and, consequently, hubs of higher functional connectivity must be efficient. In
their report, the OFC was described as having high glucose efficiency as measured by the ratio between the strength of functional
integration (based on rsfMRI and the number of connections of
the network nodes) and cerebral metabolic rate of glucose. Taken
together, because the OFC is a network hub, observed increase in
OFC functional connectivity concomitant with reductions in OFC
gray matter may suggest neuroadaptive plasticity.
The findings of greater functional connectivity in OFC network
in marijuana users were echoed by increased structural connectivity
(i.e., FA) of the forceps minor in marijuana users relative to controls. Greater FA has been suggested to reflect better myelination
and/or intact axons (42). Based on RD and AD measurements, it
appears that the FA difference between the groups in the forceps
minor was driven by lower RD, suggesting greater myelination in
the marijuana users. Although not as widely reported, greater
white matter microstructure in marijuana users has also been

reported by DeLisi et al. (43) in adolescent moderate marijuana

users; however, the difference from controls did not reach significance. Greater FA has also been reported in alcohol users
(44, 45), which was posited to reflect a premorbid vulnerability
for accelerated PFC myelin maturation in those at risk for alcohol use disorders. Among possible explanations for these
findings of greater FA in marijuana users include differential
effects of cannabis depending on the specific fiber tract. Specifically, because the forceps minor connects the OFC, which is
enriched with CB1, it is possible that there are unique neural
adaptations to the forceps minor that are unlike other white
matter tracts in the brain (e.g., corpus callosum). Others have
also reported antiinflammatory properties of cannabis constituents such as cannabidiol (CBD). DTI is sensitive to increased tissue water resulting in decreased FA as a result of
inflammation; therefore, it is possible that any antiinflammatory
effects of cannabis would lead to greater FA. Lastly, it is also
possible that the effects of cannabis (i.e., CBD) may be beneficial
to white matter in terms of regulation of mitochondrial activity,
antioxidant processes, and modulation of clearance processes that
protect neurons on the molecular level (46). Future studies are
needed to examine these specific effects on white matter.
Altogether, if these effects are indeed due to neurotoxic effects
of cannabis, the inverse relationship between OFC structure and
connectivity suggests that OFC gray matter (vs. white matter) is
more vulnerable to the effects of THC. Endogenous cannabinoids
play an important role in synaptic pruning (47), therefore, introduction of exogenous cannabinoids such as THC might disrupt
this system by competing for the receptors and, thereby, inhibiting
synaptic pruning particularly in receptor-enriched areas such as the
OFC (48). In other words, any premorbid developmental trajectory
may be modified by exposure to cannabis, resulting in accelerated
OFC myelin maturation. However, although the majority of the
animal literature and emergent human studies illustrate the downregulation of CB1 receptors as a result of THC, we acknowledge
that longitudinal studies are needed to address causality of these
neural abnormalities (4951).
Our findings of negative correlations between connectivity indexes and measures of marijuana use suggest a cumulative deleterious effect of marijuana on OFC connectivity. There was
a relationship between functional connectivity and onset of use
that suggested that greater functional connectivity was associated
with earlier onset of regular use, whereas chronic marijuana use
showed lower structural connectivity (i.e., FA). This dissociation
demonstrates the complexity of marijuanas effects on the brain,
particularly on marijuanas interaction with neurodevelopmental
periods. Along with the important findings by Cheetham et al.
(30) suggesting that lower OFC volume predates the onset of
marijuana use, we suggest that greater functional connectivity
observed at the onset of marijuana use that then dissipates
with chronic use may be a form of neural scaffolding. This
comprehensive pattern of neural response to marijuana is
of particular importance in terms of treatment and even
policy. Future studies should focus on the nuances of these
complex interactions.

Neuroimaging measures
DTI MRI
Forceps minorFA
Forceps minorRD
Gray matter volume
Left middle OFC
Right superior OFC

Filbey et al.

Behavioral measures

Parametric model

t score

P value

Duration of use in years

Duration of use in years

Quadratic
Quadratic

2.05
2.60

0.05
0.016

2.02
2.30

0.056
0.032

MPS total score

MPS total score

Linear
Linear

PNAS | November 25, 2014 | vol. 111 | no. 47 | 16917

NEUROSCIENCE

Table 4. Brainbehavior correlations in exclusively marijuana users (n = 27)

users may return to normal values due to neuroadaptive phenomena occurring after periods of abstinence (5153). Although
our study cannot address whether the structural alterations observed are permanent or reversible, such an investigation would
provide important information as to the trajectory of these
effects. Given the indication that a quadratic trend may fit the
trajectory of these alterations, it would be important to verify
these findings with a longitudinal approach.

To date, treatment and prognosis of cannabis use disorders is

hampered by the inconclusive underlying pathophysiology associated with marijuana use. In this study, we found that chronic
exposure to marijuana (i) reduces OFC gray matter volume, (ii)
increases structural and functional connectivity, and (iii) leads to
neural alterations that are modulated by age of onset and duration of use. All in all, these findings suggest that chronic
marijuana use results in complex neuroadaptive processes. Future studies are needed to determine whether these changes
revert back to normal following protracted abstinence from
marijuana use. Existing literature shows that cognitive alterations and CB1 receptor down-regulation in regular marijuana

ACKNOWLEDGMENTS. We thank Tim McQueeny for assistance with the

behavioral data. This work was supported by National Institute on Drug
Abuse Grant K01 DA021632.

1. SAMHSA (2012) Results from the 2011 National Survey on Drug Use and Health:
National Findings (Subst Abuse Ment Health Serv Admin, Rockville, MD), NSDUH
Series H-30, DHHS Publication No. SMA 06-4194.
2. Lorenzetti V, Lubman DI, Whittle S, Solowij N, Ycel M (2010) Structural MRI findings
in long-term cannabis users: What do we know? Subst Use Misuse 45(11):17871808.
3. Sneider JT, Gruber SA, Rogowska J, Silveri MM, Yurgelun-Todd DA (2013) A preliminary study of functional brain activation among marijuana users during performance of a virtual water maze task. J Addict 2013:461029.
4. Cousijn J, et al. (2013) Effect of baseline cannabis use and working-memory network
function on changes in cannabis use in heavy cannabis users: A prospective fMRI
study. Hum Brain Mapp 35(5):24702482.
5. Smith AM, Longo CA, Fried PA, Hogan MJ, Cameron I (2010) Effects of marijuana on
visuospatial working memory: An fMRI study in young adults. Psychopharmacology
(Berl) 210(3):429438.
6. van Hell HH, et al. (2010) Chronic effects of cannabis use on the human reward system: An fMRI study. Euro Neuropsychopharmacol 20(3):153163.
7. Batalla A, et al. (2013) Modulation of brain structure by catechol-O-methyltransferase
Val(158) Met polymorphism in chronic cannabis users. Addict Biol. 19(4):72232.
8. Schacht JP, Hutchison KE, Filbey FM (2012) Associations between cannabinoid
receptor-1 (CNR1) variation and hippocampus and amygdala volumes in heavy cannabis users. Neuropsychopharmacology 37(11):23682376.
9. Demirakca T, et al. (2011) Diminished gray matter in the hippocampus of cannabis
users: Possible protective effects of cannabidiol. Drug Alcohol Depend 114(2-3):
242245.
10. Ashtari M, et al. (2011) Medial temporal structures and memory functions in adolescents with heavy cannabis use. J Psychiatr Res 45(8):10551066.
11. Ycel M, et al. (2008) Regional brain abnormalities associated with long-term heavy
cannabis use. Arch Gen Psychiatry 65(6):694701.
12. Churchwell JC, Lopez-Larson M, Yurgelun-Todd DA (2010) Altered frontal cortical
volume and decision making in adolescent cannabis users. Front Psychol 1:225.
13. Cousijn J, et al. (2012) Grey matter alterations associated with cannabis use: Results of
a VBM study in heavy cannabis users and healthy controls. Neuroimage 59(4):
38453851.
14. McQueeny T, et al. (2011) Gender effects on amygdala morphometry in adolescent
marijuana users. Behav Brain Res 224(1):128134.
15. Gilman JM, et al. (2014) Cannabis use is quantitatively associated with nucleus accumbens and amygdala abnormalities in young adult recreational users. J Neurosci
34(16):55295538.
16. Block RI, et al. (2000) Effects of frequent marijuana use on brain tissue volume and
composition. Neuroreport 11(3):491496.
17. Tzilos GK, et al. (2005) Lack of hippocampal volume change in long-term heavy
cannabis users. Am J Addict 14(1):6472.
18. Battistella G, et al. (2014) Long-term effects of cannabis on brain structure. Neuropsychopharmacology 39(9):20412048.
19. Tsou K, Brown S, Saudo-Pea MC, Mackie K, Walker JM (1998) Immunohistochemical
distribution of cannabinoid CB1 receptors in the rat central nervous system. Neuroscience 83(2):393411.
20. van den Heuvel M, Mandl R, Luigjes J, Hulshoff Pol H (2008) Microstructural organization of the cingulum tract and the level of default mode functional connectivity.
J Neurosci 28(43):1084410851.
21. Greicius MD, Supekar K, Menon V, Dougherty RF (2009) Resting-state functional
connectivity reflects structural connectivity in the default mode network. Cereb
Cortex 19(1):7278.
22. Segall JM, et al. (2012) Correspondence between structure and function in the human
brain at rest. Front Neuroinform 6:10.
23. Filbey FM, Schacht JP, Myers US, Chavez RS, Hutchison KE (2009) Marijuana craving in
the brain. Proc Natl Acad Sci USA 106(31):1301613021.
24. Filbey FM, Schacht JP, Myers US, Chavez RS, Hutchison KE (2010) Individual and additive effects of the CNR1 and FAAH genes on brain response to marijuana cues.
Neuropsychopharmacology 35(4):967975.
25. Tung KC, et al. (2013) Alterations in resting functional connectivity due to recent
motor task. Neuroimage 78:316324.
26. Wakana S, et al. (2007) Reproducibility of quantitative tractography methods applied
to cerebral white matter. Neuroimage 36(3):630644.
27. Stephens RS, Babor TF, Kadden R, Miller M; Marijuana Treatment Project Research
Group (2002) The Marijuana Treatment Project: Rationale, design and participant
characteristics. Addiction 97(Suppl 1):109124.

28. Cohen MX, Heller AS, Ranganath C (2005) Functional connectivity with anterior cingulate and orbitofrontal cortices during decision-making. Brain Res Cogn Brain Res
23(1):6170.
29. Cousijn J, Goudriaan AE, Wiers RW (2011) Reaching out towards cannabis: Approachbias in heavy cannabis users predicts changes in cannabis use. Addiction 106(9):
16671674.
30. Cheetham A, et al. (2012) Orbitofrontal volumes in early adolescence predict initiation of cannabis use: A 4-year longitudinal and prospective study. Biol Psychiatry
71(8):684692.
31. Cousijn J, et al. (2013) Cannabis dependence, cognitive control and attentional bias
for cannabis words. Addict Behav 38(12):28252832.
32. Cousijn J, et al. (2013) Individual differences in decision making and reward processing predict changes in cannabis use: A prospective functional magnetic resonance
imaging study. Addict Biol 18(6):10131023.
33. Kringelbach ML, ODoherty J, Rolls ET, Andrews C (2003) Activation of the human
orbitofrontal cortex to a liquid food stimulus is correlated with its subjective pleasantness. Cereb Cortex 13(10):10641071.
34. Bolla KI, et al. (2003) Orbitofrontal cortex dysfunction in abstinent cocaine abusers
performing a decision-making task. Neuroimage 19(3):10851094.
35. Bechara A, Damasio H, Damasio AR (2000) Emotion, decision making and the orbitofrontal cortex. Cereb Cortex 10(3):295307.
36. Gremel CM, Luo G, Lovinger D, Costa R (2013) Endocannabinoid signaling in orbitofrontal cortex modulates habit formation. Society for Neuroscience Conference (Society for Neuroscience, San Francisco).
37. Orr C, et al. (2013) Altered resting-state connectivity in adolescent cannabis users.
Am J Drug Alcohol Abuse 39(6):372381.
38. Filbey F, Yezhuvath U (2013) Functional connectivity in inhibitory control networks
and severity of cannabis use disorder. Am J Drug Alcohol Abuse 39(6):382391.
39. Harding IH, et al. (2012) Functional connectivity in brain networks underlying cognitive control in chronic cannabis users. Neuropsychopharmacology 37(8):19231933.
40. Zhang D, Raichle ME (2010) Disease and the brains dark energy. Nat Rev Neurol 6(1):
1528.
41. Tomasi D, Wang GJ, Volkow ND (2013) Energetic cost of brain functional connectivity.
Proc Natl Acad Sci USA 110(33):1364213647.
42. Song SK, et al. (2002) Dysmyelination revealed through MRI as increased radial (but
unchanged axial) diffusion of water. Neuroimage 17(3):14291436.
43. Delisi LE, et al. (2006) A preliminary DTI study showing no brain structural change
associated with adolescent cannabis use. Harm Reduction Journal, 10.1186/14777517-3-17.
44. Cardenas VA, et al. (2013) Not lesser but Greater fractional anisotropy in adolescents
with alcohol use disorders. NeuroImage Clinical 2:804809.
45. De Bellis MD, et al. (2008) Diffusion tensor measures of the corpus callosum in adolescents with adolescent onset alcohol use disorders. Alcohol Clin Exp Res 32(3):
395404.
46. Bilkei-Gorzo A (2012) The endocannabinoid system in normal and pathological brain
ageing. Philos Trans R Soc Lond B Biol Sci 367(1607):33263341.
47. Kim HJ, Waataja JJ, Thayer SA (2008) Cannabinoids inhibit network-driven synapse
loss between hippocampal neurons in culture. J Pharmacol Exp Ther 325(3):850858.
48. Bossong MG, Niesink RJ (2010) Adolescent brain maturation, the endogenous cannabinoid system and the neurobiology of cannabis-induced schizophrenia. Prog
Neurobiol 92(3):370385.
49. Sim-Selley LJ (2003) Regulation of cannabinoid CB1 receptors in the central nervous
system by chronic cannabinoids. Crit Rev Neurobiol 15(2):91119.
50. Burston JJ, Wiley JL, Craig AA, Selley DE, Sim-Selley LJ (2010) Regional enhancement
of cannabinoid CB receptor desensitization in female adolescent rats following repeated Delta-tetrahydrocannabinol exposure. Br J Pharmacol 161(1):103112.
51. Hirvonen J, et al. (2012) Reversible and regionally selective downregulation of brain cannabinoid CB1 receptors in chronic daily cannabis smokers. Mol Psychiatry 17(6):642649.
52. Bosker WM, et al. (2012) Medicinal (9) -tetrahydrocannabinol (dronabinol) impairs
on-the-road driving performance of occasional and heavy cannabis users but is not
detected in Standard Field Sobriety Tests. Addiction 107(10):18371844.
53. Hanson KL, et al. (2010) Longitudinal study of cognition among adolescent marijuana
users over three weeks of abstinence. Addict Behav 35(11):970976.
54. Wechsler D (1999) Manual for the Wechsler Abbreviated Scale of Intelligence (The
Psychological Corporation, San Antonio).
55. First MB, Spitzer RL, Gibbon M, Williams JBW (1997) Users Guide for the Structured
Clinical Interview for DSM-IV Axis I Disorders (SCID) (American Psychiatric Press,
Washington, DC).

16918 | www.pnas.org/cgi/doi/10.1073/pnas.1415297111

Filbey et al.