Do mindfulness-based interventions change brain function in people with substance dependence? A systematic review of the fMRI evidence

The emerging evidence from 7 studies indicates that brain functional changes associated with MBI may occur in individuals diagnosed with SUDs in pathways relevant for reward processing [4] and for mindfulness, most consistently the ACC and the striatum [10, 33, 34]. Brain function was largely unaffected by control interventions. The reviewed MBIs reduced quantity of cigarette use and cravings, and brain functional changes were associated with lower quantity of use. The evidence is insufficient to confirm whether MBIs in the reviewed studies changed mindfulness states or traits or both, or mental health symptom scores. Further, there is little-to-no evidence about how MBI affects psychological wellbeing in SUD, or how brain functional changes in individuals diagnosed with SUD correlates with measures of mental health symptom severity.

The evidence to date shows that MBIs are associated with brain functional changes (activity and connectivity), particularly within the ACC and the striatum. Notably, these regions are implicated in neuroscientific theories of addiction [4] and meta-analytic evidence on the effects of mindfulness training [10, 35]. The ACC is implicated in addiction-relevant cravings, motivation and preoccupation with substance use [4]; and also is targeted by mindfulness to aid self-regulation and conflict-monitoring [4, 10, 36].

The striatum is a core component of reward processing, sensitivity to reward and compulsive substance use [35]; and meta-analytic evidence from healthy samples indicates that it is engaged by meditative states [10]. Thus, MBIs may regulate the function of key cortical/inhibitory regions within the mindfulness network that are aberrant in addiction (e.g., ACC). This might occur by MBIs mitigating cortical dysfunction (e.g., within the ACC). Specifically, MBIs may boost cortical inhibitory projections that downregulate reactivity of striatal regions implicated in craving – thereby changing the connectivity between these regions or the activity of the regions, or both. Indeed, MBIs were consistently associated with lower craving and drug quantity consumption across the behavioral literature to date [18] and in the current review of fMRI studies.

The reviewed nascent literature showed that MBIs were also associated with changes (mostly with small-effect sizes) in the function of additional and interconnected regions implicated in cognitive processes that are core to both reward dysregulation and to mindfulness. They include: the insula, cerebellum, precuneus, inferior frontal gyrus, and PCC. MBIs in the examined SUDs were associated with changes in the function of the insula, and some of these effects were reported to be robust (e.g., whole-brain family-wise corrected) [29]. The insula is a primary hub for interoception (i.e., awareness of bodily states, which in addiction could be affected by cravings, withdrawal). Notably, altered interoception and cravings in addiction [37‐39], as well as relapse [40], have been ascribed to the insula. Indeed, the insula has been a target of brain stimulation to diminish appetitive behavior [41, 42] and its function during inhibition to smoking cues has been correlated with abstinence (i.e. to tobacco) [43]. Concurrently, the insula has been recognized as one of the primary neural mechanisms of action for mindfulness practice in healthy samples [44]. For example, mindfulness practice has been shown to down-regulate the activity of the insula (voxel-wise P < .01, exceeding volume threshold of 256 µl for cluster wise probability of 0.05 for the insula), as tested during an inspiratory breathing fMRI task in a sample of marines before and after undergoing 20 h of mindfulness-based fitness training or training as usual [36]. Also, mindfulness practice has been associated with medium-effect size changes of the structural connectivity of the insula network, compared to a control cognitive training intervention [45]. Overall, early evidence that MBIs in SUD may involve modulation of the insula is consistent with existing emerging findings from addiction and mindfulness literature. Interestingly, while insular function decreased pre-to-post MBI in the examined SUDs, in healthy samples findings pertaining to MBI-associated changes in insular function have been mixed. For example, in a recent systematic review and meta-analysis on mindfulness fMRI tasks in healthy samples, only 62% of fMRI studies using MBI reported increased activity of the insula [46]. The direction of changes in insular function in the examined SUDs and healthy samples may be (partly) due to distinct baseline levels of insular function, which are aberrant in SUD compared to controls and may be more heterogeneous in healthy samples.

The emerging evidence synthesized in our review revealed SUD-related changes in additional higher order cognitive control regions (precuneus, IFG). Importantly, these regions are reportedly implicated in executive dysfunction in SUDs (e.g., disinhibition, self-control, salience attribution and awareness, compulsive use) [33]. In addition, they have been implicated in substance use behavior (e.g., smoking and drinking quantities) [47]. The inferior frontal gyrus has been implicated in inhibitory control during withdrawal (i.e., from smoking) [48], disinhibition [49], intoxication [50], and in relapse [40]; while the precuneus is implicated in reactivity to drug cues (i.e., tobacco and alcohol; [51]) and impulsivity [52]. In addition, mindfulness practice has been shown to modulate higher cognitive control regions [53, 54]. Together, the findings provide support for the hypothesis that MBIs affect inferior frontal/precuneus regions implicated in addiction.

The reviewed evidence also showed that MBIs change the function of the PCC [9, 28]. Notably, the PCC underlies the cognitive processes of self-reference [55] and is a core component of the default mode network, a brain system that activates ‘when individuals are not focused on the external environment’ [56, 57]. Previous work shows that the function of the PCC is altered in addiction, including in the population diagnosed with SUDs as examined in this review (i.e., nicotine, opiate disorders, alcohol use disorders) [58] and while performing fMRI tasks used in most of the reviewed studies (i.e., resting state and cue reactivity) [59, 60]. The emerging findings from the literature are aligned with theories that meditation changes the function of the PCC [11], and with growing evidence that MBIs change the function of the PCC in healthy samples (e.g., connectivity) [12‐14].

Further, the evidence from this review indicates that changes in PCC function correlates with cigarette reduction. This finding is consistent with previous work that structural damage of the PCC disrupts cigarette smoking [61] and functional connectivity predicts relapse [62]. Future work is required to confirm that the PCC is implicated in reduced use of other substances and to explore its role in predicting relapse post MBIs.

Changes in the function of the cerebellum also emerged in fMRI studies of mindfulness in the examined SUDs, consistent with findings that it is implicated in meditative states [10]. These findings are also in line with evidence that the cerebellum is an integral part of the addiction circuitry [63] and plays a key role in reward processing, motivation and cognitive control.

In sum, the reviewed evidence suggests that brain functional changes are associated with MBIs in the examined SUDs (i.e., opiates, nicotine; see Fig. 1), most consistently the ACC, followed by the striatum, and other regions (e.g., insula, precuneus/inferior frontal gyrus, PCC, cerebellum). This network of regions has previously been implicated in SUDs [33, 64], MBIs and mindfulness practice in healthy samples and cognitive-based interventions in SUDs [65]. Therefore, it may be that MBIs target the function of core regions of the addiction neurocircuitry.

Changes in the ACC were reported also by Fahmy and colleagues, who conducted the only study in opiate users, who were in treatment (with mood stabilizers, antipsychotics and sedatives) [26]. Thus, MBIs may ameliorate dysregulated reward function of the ACC across the examined SUDs, and in response to natural and drug related rewards. At this point there is insufficient evidence to conclude that MBIs affect distinct brain pathways that might be relevant to specific SUD populations, which is worthy of investigation given that there are important differences in the psychopharmacological signatures of specific drugs. Future well-controlled intervention studies that deliver consistent MBIs to distinct SUD groups are required to unpack this.

Several studies reported correlations between brain functional changes associated with MBI (i.e., striatum, ACC, PCC, mPFC) and reduction in cigarettes exposure [9, 27‐29]. Thus, brain functional changes associated with MBI may underlie reduction in substance use quantity as found in this review (i.e., decreased cigarette use) [27, 29], and previous meta-analyses (i.e., several substances) [18]. However, as the nature of the association is correlational the opposite may be true – reduction in quantity of substance use may drive changes in brain function.

Of note, variables associated with changes in drug quantity were not systematically measured or accounted for (e.g., withdrawal, craving, nicotine receptors), and their role remains to be clarified. Further in some studies, changes pre-to-post MBIs were noted in brain function but not in problems with substance use (e.g., craving). This trend may reflect a dissociation between neurocognitive and behavioral changes in MBI – whereby changes may occur in brain function but not in behavior—as previously suggested [66]. Alternatively, there could be a threshold of change required for brain related change to drive behavioral change. To test this hypothesis, growth models that allow for random intercepts and slopes are warranted. A strength of the literature was that all sample characteristics (e.g., demographics, mindfulness, personality measures, craving, substance exposure) were reportedly not significantly different between intervention groups at baseline. Therefore, the group differences in brain function are unlikely to reflect characteristics predating MBIs. However, non- random assignment of participants to groups in some studies, and other potential biases need to be accounted for when drawing conclusions about the strength of the evidence.

Interestingly, trait mindfulness has been seldom examined in the reviewed literature and in relation to brain function. In one study conducted by Kober and colleagues (2019), non-significant trends emerged for an intervention-by-time effect on trait mindfulness (FFMQ [67]. Specifically, participants in the MT group (but not those in the FFS group) showed increased scores on trait mindfulness facets including “observe”, “non-react” and total score. Meanwhile, the MT group (vs FFMQ and CBT) showed lower stress reactivity of the amygdala and insula, which was associated with reduced smoking post intervention and at 3-month follow-up. Thus, it cannot be excluded that changes in trait mindfulness specific to the MT group might have moderated reduced stress brain reactivity. In line with this notion, greater trait mindfulness has been associated with reduced self-reported distress and negative affect in response to laboratory stressors [68]; and amygdala’ reactivity to negative faces [69] and amygdala function during rest [70]. Given that only a single fMRI study has examined trait mindfulness in relation to MBI related brain functional changes in the examined SUDs, more evidence is required to demonstrate if trait mindfulness moderates how MBIs affect brain function in SUD.

There is insufficient evidence to demonstrate if brain functional changes associated with MBIs cause changes in mindfulness. Only one study that reported a group-by-time effect on brain function, explored and demonstrated correlations between mindfulness changes pre-to-post MBI and brain functional changes in higher order cognitive control regions (IFG) [26]. Further, there were limited and mixed findings regarding how MBIs affected mindfulness levels in the reviewed studies, with lack of changes and increases in mindfulness levels in both MBI and TAU [26, 30]. The evidence synthesized in this review contrasted meta-analytic findings of increased mindfulness scores with MBIs in people diagnosed with a SUD [18]. The lack of robust change in mindfulness in the reviewed samples may be due to methodological limitations of the fMRI studies in SUDs. First, only two studies examined changes in mindfulness scores [26, 30] and their sample sizes were inadequately powered to robustly detect changes (n = 5 and 16 participants per group). Thus, there is insufficient evidence to determine if the reviewed interventions changed mindfulness levels.

Second, mindfulness was measured via two distinct scales that did not assess mindfulness in relation to the period of time that synchronized with the intervention (e.g., past week, 3 weeks) or did not specify an index period of time or both; therefore, they might have not been sensitive to any changes over the specific intervention period. Changes in mindfulness may have been evident if appropriate measures had been used immediately before and after each mindfulness session, such as the State Mindfulness Scale, which has been shown to be sensitive to mindfulness changes in healthy samples [71].

Overall, there is insufficient evidence to determine if MBIs in the reviewed studies increased mindfulness, and if brain changes in SUDs associated with MBIs correlate with or are caused by increased mindfulness. These hypotheses require testing through future repeated measure fMRI studies with active and passive control conditions, that assess dimensions of mindfulness relevant for the intervention type and duration, to detect mindfulness changes with precision. Similarly, the evidence on brain functional changes associated with MBI and urges, stress, dimensions of personality, mental health and wellbeing consisted mostly of single studies. The fMRI evidence to date is insufficient to profile the relevance of brain functional changes of MBIs in addiction and warrants detailed assessment of measures of mental health and wellbeing and their relationship to brain function.