Our investigation of the impact of cannabis abuse on various subcortical regions yielded results that add to a body of recent work using Human Connectome Project (HCP) data. For instance, recent cannabis use in this sample was negatively associated with hippocampal volume (47) and smaller left hippocampal volume mediated the association between frequency of cannabis use and working memory deficits in cannabis users (48). Additionally, HCP data has revealed an effect of THC exposure on amygdala microstructure organization (49). In line with these studies, we found that cannabis users had marginally smaller amygdala volumes than non-users. However, we only found a cannabis use-by-sex interaction in cerebellar cortex volumes, suggesting that females may be particularly susceptible to the effects of chronic cannabis use in this region. Finally, we observed that female cannabis users had poorer self-reported sleep quality than the other groups, which was particularly pronounced among females who began using cannabis in early adolescence. We discuss these findings in more detail below.

The cerebellum has traditionally been studied for its role in balance and motor coordination (50), nociception (51), and motor cognition (52, 53). Brain imaging studies in humans have shown that the cerebellum is sensitive to the acute and chronic effects of cannabis (8), including glucose metabolic activity (12, 54), volume, and resting-state activity (13, 55–57). Postmortem studies have found striking differences in the cerebellar structure of drug abusers relative to controls; one group showed increased autophagy biomarkers in the cerebellum of multi-substance drug abusers (58), while another found signs of neurodegeneration in the cerebellar cortex of people who were dependent on opioids, suggesting that drug addiction can negatively impact cerebellar structure (59). Recently, Gil-Miravet et al. found that the cerebellum modulates drug-cue associative memory in cocaine users (60), while Hung et al. showed increased functional connectivity between the pallidum and cerebellum of ketamine users, suggesting that the cerebellum has a fundamental role in the pathophysiology of addiction (61). The cerebellum is clearly affected by cannabis use as well; chronic cannabis users can experience cerebellar-dependent motor adaptation impairment (62), while synthetic cannabinoid users show reduced gray matter volume in the left cerebellum (63). These studies are consistent with several recent reviews published on the topic which note the cerebellum's role as a nexus between motor, reward, and cognitive processes crucial to drug seeking behavior (64–66). Compared to other brain regions, there is a relatively high concentration of CB1-Rs in the cerebellum (38, 67, 68). PET studies have shown that CB-1Rs are reversibly downregulated in people with a history of chronic cannabis consumption, which is likely to contribute to tolerance and dependence with repeated use (69). Previous studies have had mixed findings on the relationship between chronic cannabis use and cerebellar volumes with some studies suggesting that cannabis actually increases gray matter volumes (57, 70–73). Here we found that the smaller cerebellar cortical volumes in cannabis users relative to the controls were driven by the female cannabis users. This could explain the discrepancies in the literature since sex was not accounted for in prior investigations. Indeed, studies finding larger cerebellar volumes in cannabis users had very few or no female participants in the cannabis group: Wang et al.: 25% female (5F/15M); Cousijn et al.: 36% Female (12F/21M); Battistella et al.: 0% Female (0F/31M); Wu and Yang: 25% female (5F/15M); Koenders et al.: 25% female (5F/15M). Those findings contrast with the results in our 33% Female sample (56F/114M) and those of another related study (50% Female; 13F/13M) that found lower cerebellar microstructural integrity in adults at risk for CUD relative to controls (56). These findings underscore the importance of including an adequate number of female participants and of investigating sex differences in brain and behavioral outcomes for people with chronic substance use, for such differences appear to be prevalent throughout the addiction endophenotype (8, 74, 75).

We also observed a trend for a group effect on amygdala volume, with lower volumes in cannabis users compared to healthy controls, in agreement with prior studies (73, 76), and which correlated with amount of cannabis used and dependence severity (70, 73). In terms of sex differences, one study found that while adolescent female cannabis users had larger right amygdalar volumes than healthy controls, there was no such difference in males (77). However, our finding of lower amygdalar volume in cannabis users was not sex-dependent and likely would not contribute to the sex-dependent impairment in sleep quality that we observed. Nonetheless, it is possible that a deficit in amygdala volume could contribute to the overall poorer sleep quality observed in CUD compared to controls. The amygdala has been previously implicated in poor sleep quality; while functional connectivity between the amygdala and premotor cortex is negatively associated with sleep quality (78), it appears that sleep quality might modulate amygdalar functional connectivity and not vice versa (79). Additionally, patients with narcolepsy have lower GM volume in the amygdala relative to controls, suggesting a possible unidirectional relationship between sleep quality and amygdala volume (80).

Our finding of sex-specific differences in cerebellar volume among cannabis users was not present in other regions with high CB1-R density, such as the amygdala and the hippocampus. Preclinical studies in rats have shown that chronic THC caused downregulation and desensitization of CB1-R in cerebellum, and these decreases were especially large in females (9). Other studies have reported higher baseline CB1-R density in female compared to male rats, although the cerebellum was not examined (10). Human PET studies have similarly found that females have higher baseline CB1-R availability than males in many brain regions (81), including cerebellar cortex (82). Given that CB1-R density was influenced by the estrous cycle in preclinical studies, it is possible that female sex hormones play a role in sex differences in CB1-R availability as well as sex differences on cannabis effects in brain and behavior (10). Animal models could be used to test if sex differences in CB1-R density prior to and after chronic THC exposures may confer female vulnerability to potential neurotoxic effects of cannabis on cerebellar structure and function. Additionally, while initial human PET studies found that CB1-Rs in both sexes were downregulated in response to chronic cannabis use (69), future longitudinal studies should examine whether there are sex differences on the association of CB1-R downregulation with the severity of CUD. This is especially important given that there are currently no FDA-approved pharmacological treatments for CUD, and that one promising candidate, the fatty-acid amide hydrolase (FAAH) inhibitor PF-04457845, was recently shown to reduce cannabis withdrawal severity and promote abstinence, but only men with CUD were included in the trial (83). Thus, much work remains to be done to see if treatments show similar improvements in females and if they do so in part via cerebellar mechanisms. Our group has previously proposed that downregulation of CB1R in subjects with cannabis dependence might increase vulnerability to cortical thinning, suggesting that CB1R availability can lead to structural changes in the brain (21). Another study found that some heavy cannabis users have a genetic predisposition toward cannabis dependence due to a functional single nucleotide polymorphism affecting cannabis receptor-1 gene expression; among cannabis users, minor relative to major allele carriers had lower volume in the nearby hippocampus, but not the amygdala (84). However, cerebellum volume was not examined in this study, and it is possible that there is a similar connection between CB1R and cerebellar volume. Future studies should examine this possibility to uncover the link between CB1R activation/availability and amygdalar/cerebellar volume.

We also observed that females' self-reported sleep quality (as indexed by the global PSQI score) was similarly more vulnerable to the negative effects of cannabis use than for males. Given that patients with degenerative diseases of the cerebellum such as cerebellar ataxia commonly report sleep disturbances, poor subjective sleep quality, restless leg syndrome, and REM behavior disorder, it is plausible that the cerebellar volume loss in female cannabis users contributed to their poor sleep quality (85, 86). However, in our study the effects of cannabis on cerebellar volumes did not mediate the effects of cannabis on sleep quality, which is likely to reflect a more complex association between cannabis effects in brain structure and function. Similarly, the effects of cannabis on sleep quality did not mediate its effects on brain volume, which might also indicate distinct neurobiological processes underlying these two effects. Given the observational nature of this study, we are unable to rule out the possibility that women to start with had lower sleep quality than men as has been reported by other studies (25–28), though in our current study sleep scores in control males did not significantly differ from those in control females. It is also possible that a mismatch between expectation and reality in how cannabis helps with sleep may play a role in self reports; in a majority female (67%) sample of cannabis users, while both frequency and presence of cannabis use were associated with the expectation of improved sleep, cannabis use was actually associated with poorer subjective sleep quality (87). Finally, we observed that females who reported first using cannabis in early adolescence tended to report the worst sleep quality, which aligns with a recent large-scale twin study (n = 1,656) that reported that regular cannabis use at a young age correlated with shorter sleep duration in adulthood (88). Given the differences in socialization, development, and expectations associated with cannabis use, women may be more vulnerable to the negative sleep effects of cannabis abuse at younger ages than men. These data complement a large body of literature suggesting that early-onset cannabis use is strongly associated with poor neuropsychiatric outcomes (89), and again highlight sex differences as an important future avenue of investigation.