With a shift towards legalisation and decriminalisation of recreational cannabis use, the prevalence of cannabis use disorder (CUD) appears to be increasing (1). CUD is associated with adverse effects on mental health, including increased risk of mood and anxiety disorders, cognitive dysfunction, and social impairment (2). As such, there has been considerable focus in recent years on assessing prognostic markers of CUD, which may aid in understanding and optimising long-term clinical outcomes (3). Numerous structural MRI studies have investigated CUD induced cortical and subcortical changes, yet evidence remains inconsistent with a meta-analysis revealing that about 50% of studies reported no significant grey matter (GM) alterations (3, 4). Among significant findings, morphological alterations have been observed across hippocampus (5–7), frontal cortex (8–11), and amygdala (7, 12, 13). Besides morphological changes, CUD has also been linked to alterations in brain WM organization (14). Previous diffusion MRI studies have identified heterogeneous alterations in fractional anisotropy (FA; a measure of WM integrity) across the corpus callosum (15–17), frontal regions (6, 16), cingulum (18, 19), and cerebellum (20). Plausibly, some of the variability may be attributable to small sample sizes or variation in sample demographics (e.g., inclusion of comorbid diagnoses or polysubstance use, and discrepancy in frequency of use) (14).
In addition to demographic variability, the mixed findings may be explained by the employed model. Specifically, most of the previous studies assessing WM changes have employed diffusion tensor imaging (DTI) model which is known to be limited in its capacity to identify intricate and diverse WM changes, and therefore may contribute to inconsistent outcomes across studies (3, 21). Moreover, the measure of FA lacks the microstructural specificity to fully characterise the organization of the white matter tracts. As such, advanced diffusion models using the FBA framework can help reconcile these discrepancies by measuring morphological information from both microscopic fibre density (FD) and macroscopic fiber cross section (FC, the calibre of a fibre bundle) for individual WM fibre populations within each voxel - known as fixels (22). To date, no prior stud has utilised these complimentary analytical techniques and advanced metrics to investigate WM alterations in CUD.
To date, only a handful of studies (17, 19, 23, 24) have investigated WM changes in CUD using network-based modelling (e.g., network-based statistics, connectomics). Among significant findings, altered structural connectivity has been observed across hippocampus, caudate, pallidum (24), cingulate (19), splenium of CC, and right hippocampus (17). These studies, and their finding, lack the microstructural specificity to fully characterise the integrity of the structural network for several reasons – they did not :1) used advanced diffusion sequences (e.g., high b-value and high angular resolution) that significantly affect the resolution of the acquired data and is required to resolve for the crossing fibre orientation using constrained spherical deconvolution (CSD) model (25, 26); 2) employ advanced pre-processing such as outlier replacement (27) and slice to volume motion correction (28); 3) filter reconstructed WM streamlines to be more biologically plausible (29); 4) estimate specific measures of WM microstructure such as fiber density and cross section (22).
As such, further work incorporating advanced diffusion acquisition and pre-processing techniques can help better understand the biological underpinnings of the WM changes in CUD.
In this study, we investigated differences in WM connectivity and microstructure between CUD and healthy controls, using whole-brain connectome and fixel-based analysis. Importantly, we overcome the limitations of previous DTI studies by using the CSD model capturing properties of individual fixels in the presence of crossing fibre bundles (21, 30), which are known to be present in almost 90% of WM voxels (31). We also assessed potential correlations between WM parameters (i.e., strength of connectivity between GM nodes, density and cross section of fibre bundles) and measures of cannabis use, cognition and wellbeing in CUD. We hypothesized that CUD would be associated with alterations in frontal regions, hippocampus, and corpus callosum compared to healthy controls.