The head-motion estimates calculated in the correction step were also placed within the corresponding confounds file. All resampling can be performed with a single interpolation step by composing all the pertinent transformations . Gridded resampling was performed using ants Apply Transforms , configured with Lanczos interpolation to minimize the smoothing effects of other kernels . Non-gridded resampling was performed using mri_vol2surf . Many internal operations of fMRIPrep use Nilearn 0.5.0, mostly within the functional processing workflow. For more details of the pipeline, see the section corresponding to workflows in fMRIP rep’s documentation. Data analysis was conducted in AFNI Version AFNI_20.0.18 ’Galba’ . The first level general linear model was conducted via 3dDeconvolve to generate contrast maps for each individual participant, including a regressor-of interest for each of the 4 task conditions . Six motion parameters were included as regressors of nointerest, as were the six a CompCor parameters.This generated beta-weight values at each voxel location for each of the four task conditions to carry forward to group analysis . Following first-level analysis, data were smoothed using a 6 mm gaussian kernel , for a final average smoothing level of 8.18 mm. For each of the following analyses, a whole-brain mask excluding the cerebellum was used. All analyses were performed using the AFNI function 3dLME , a group analysis program that performs linear mixed effects analysis on data with multiple measurements per participant. The primary analysis tested the effects of cannabis use and MDD diagnosis on emotion regulation. The model was specified as follows: task condition , cannabis use , MDD diagnosis , including two- and three-way interaction terms, were included as variables of interest. Medication use , age, and number of alcoholic drinks consumed in the last 28 days as regressors. Sex was not included as a regressor due to high collinearity with cannabis use. Numeric variables in this analysis and all subsequent analyses were mean-centered. A random effect of participant was included in the model, and a marginal sum of squares was used. Three secondary analyses were then conducted. First, we examined the interaction between emotion regulation style and task-condition in the full sample. Similar to the main analysis, an LME model was specified with a condition × ERQ score interaction term, and age, alcohol, and medication use included as regressors. The ERQ score involved subtracting the maladaptive emotional style from the adaptive style . Thus, higher ERQ scores indicated more adaptive emotion regulation than lower scores.
Two participants were excluded from this analysis due to missing ERQ score data. Next, we examined the relationship between HAM-D score and BOLD-signal activation during the emotion regulation task. Here, only individuals with an active MDD diagnosis were included . The LME model was specified with a condition × HAM-D score interaction, and age, alcohol, and medication were included as regressors. Finally, the effects of early-onset cannabis use on task-related BOLD signal activation were examined. Here, we only included individuals who actively used cannabis . We tested our hypothesis that early-onset cannabis use would have pronounced negative effects by grouping subjects into early-onset versus late onset . LME analysis is well-suited for such unbalanced groups . We then identified where early-onset cannabis users had greater or lower activation than late-onset users. The LME model was specified with a condition × age of onset interaction,vertical grow system and age, alcohol, and medication were included as regressors. For second-level analyses, the minimum cluster-size threshold was determined in two steps. First, we estimated the smoothness of the residuals for each subject output by 3dDeconvolve using the autocorrelation function option , and the mean smoothness level was calculated. Next, minimum cluster size was determined using a 10,000 iteration Monte Carlo simulation at a voxelwise alpha level of p = 0.05. Correction for multiple comparisons at p = 0.05 was achieved by setting a minimum cluster size of 64 voxels. Posthoc contrasts were FDR corrected. The current study used an fMRI paradigm of positively- and negatively-valenced emotional scenes to investigate the individual and combined effects of MDD and frequent cannabis use on emotion regulation. We also conducted several secondary analyses to explore how the various characteristics of emotion regulation, MDD, cannabis use and age of onset of cannabis use further contribute to emotion processing in the brain. Although we did not see a three-way interaction, both MDD and cannabis use showed a complete reversal of activity levels relative to their controls in response to the different conditions of the emotion regulation task. Specifically, while participants without MDD showed higher activation to the positive attend condition vs. the other three, those with MDD showed low activation to this condition, with the other three showing higher levels . Similarly, participants who did not use cannabis showed higher activation levels in response to the negatively vs. positively valenced conditions, while the opposite was true for cannabis users . The fact that we saw this reversal in all four conditions strongly suggests that both MDD and cannabis use affect several aspects of emotion processing. That is, we observed a change in both positive and negative, and effortful and passive emotion processing. Prior research has shown the effects of MDD and cannabis use on specific types of emotion processing, such as dysfunctional activity during active emotional reappraisal . The present results indicate that both MDD and cannabis use may have a more global effect than previously thought. Both of these effects were observed in the left temporal lobe. While these results were not predicted and are in need of replication, both theleft MTG and STG have frequently been associated with emotion processing , and have previously shown decreases in activity levels in individuals with MDD during emotion processing . Both regions are also involved in multisensory association . Given that the present stimuli were complex emotional scenes, it is possible that the interactions with MDD and cannabis use in each area reflect differences in multisensory representation. Individuals with MDD showed a reduced representation of positive stimuli during the attend condition, a difference that was eliminated with effortful emotion regulation. Thus, it is possible that individuals with MDD may be successfully augmenting positive representations, while being less successful in their attempt to regulate negative representations. In contrast, cannabis users showed an increased representation of positive stimuli and suppression of negative stimuli, and these mood-altering effects may reflect some of the participants’ motivation for ongoing cannabis use.
The difference between the observed effects, namely regulation versus representation of valence, could be why the specific area of temporal lobe differs. Finally, although both MDD and cannabis use affected emotional processing within the temporal lobe, the difference in specific regions may account for why we did not observe a threeway interaction. Although several regions of the frontal cortex showed activation differences among emotion regulation task conditions, there were no interactions with MDD or cannabis use. Models of both depression and of cannabis use predict the under-activation of frontal regions, specifically the vlPFC, dlPFC, and dmPFC. During healthy emotion regulation, we also observe suppression of these areas . Because the individuals with MDD are already experiencing suppression in these regions, it is possible that the amount of change during the emotion regulation task was not enough to appear different from non-depressed participants. We also found that higher ERQ scores, which represent a greater ability to adaptively control one’s emotions, correlated with less activity in the right frontal lobe. This was observed across all task conditions, indicating that better emotional control leads to less effortful emotion processing overall. While this may seem intuitive,mobile grow systems it may be surprising that there was no interaction with condition; for example, Greening and colleagues found suppressed BOLD activity in individuals with MDD during negative regulation compared to healthy controls, but no difference in positive regulation. However, here, even in the ‘attend’ conditions, individuals with low ERQ scores showed more effortful processing than those with high scores. This consistency may reflect that emotion processing occurs even when passively viewing emotionally laden images . Poorer emotional regulation has been linked to MDD , and correlates with increases in activity in frontal regions when viewing emotional images . Thus, these results fit well with previous literature, and suggest that even passive emotional processing is more effortful for those with poorer regulation, which may be a neural representation of less adaptive emotion regulation strategies . The relationship in the left MTG between HAM-D and task condition in individuals with MDD was driven by the steep increase in activity in response to the ‘negative reduce’ condition with increasing score. This relationship echoes the results found when comparing individuals with and without MDD , which showed a similar increase in activity in this condition. Notably, a similar relationship was not found in the other three conditions, highlighting the fact that even within a group of persons with MDD, there are individual differences in levels of depressive symptoms that affect different aspects of emotion regulation. Finally, our emotion regulation task showed activation within the expected network of regions involved in emotion processing, specifically the left inferior parietal lobe, the left middle frontal gyrus, the right insula, and the left inferior frontal gyrus. In both the left inferior parietal lobe and left inferior frontal gyrus, the ‘negative attend’ condition had significantly lower levels of activation than the other conditions. The left middle frontal gyrus showed lower activity to negative versus positive conditions, and the insula showed increased activation in the ‘negative reduce’ condition relative to the others. All four regions have shown differential activation during viewing of emotionally negative stimuli compared to neutral stimuli , and are thought to belong to a larger network of regions involved in the initial appraisal , regulation , and the final generation of regulated emotional states. However, although the regions showing an effect of task condition were part of the well-studied emotion processing network, the areas we found to be modulated by MDD, cannabis use, or characteristics of these two factors were outside of this network. The fact that these effects extended beyond typical emotion processing areas during the present task indicates that both MDD and cannabis use have far-reaching consequences for the brain, perhaps affecting domain-general processes .
One limitation of the present study is that our analysis of early-onset cannabis use did not identify any significant effects of age of onset, with only a main effect of condition, implying these results were similar across age groups. This was surprising, as early-onset cannabis use was previously associated with increased connectivity between the default mode network and reward-processing areas in the same sample , though the early age of onset group was defined differently. Additionally, a recent review paper reported that adolescent exposure to cannabinoids can lead to dysregulation of emotion and reward processing in rats . One possible explanation for the lack of effects in this area is the low number of participants in this analysis; only 12 individuals were considered “early” cannabis users, which may not have been a large enough sample to detect differences between early and late cannabis use. A second limitation is that we did not study the effects of comorbidity with other psychiatric illnesses. Data on comorbidities were collected and reported; as can be seen in Table S1, there was a large range of psychiatric comorbidities within the sample of individuals with MDD. Because of the large variation in the type of comorbidities observedwithin the sample, we do not have reason to believe that any one diagnosis could be driving the results observed here. However, comorbidity of MDD with other psychopathologies can impact emotion regulation and should be considered in future work. Tobacco and cannabis are among the most commonly used substances by adolescents worldwide. In 2019, 27.1% U.S. high school students and 22.3% of U.S. high school seniors reported past-30-days use of tobacco products and cannabis, respectively, with 2.4% and 6.4% of U.S. high school seniors using cigarettes and cannabis on a daily basis, respectively . Cannabis is often used in combination with combustible tobacco by young people. Approximately 14% of young adults in the U.S. report combustible tobacco and cannabis co-use within the past month .