Methamphetamine-related working memory difficulties underpinned by reduced frontoparietal responses

Abstract

Working memory difficulties are common, debilitating, and may pose barriers to recovery for people who use methamphetamine. Yet, little is known regarding the neural dysfunctions accompanying these difficulties. Here, we acquired cross-sectional, functional magnetic resonance imaging while people with problematic methamphetamine-use experience (MA⁺, n = 65) and people without methamphetamine-use experience (MA^-, n = 44) performed a parametric n-back task (0-back through 2-back). Performance on tasks administered outside of the scanner, together with n-back performance, afforded to determine a latent dimension of participants' working memory ability. Behavioural results indicated that MA⁺ participants exhibited lower scores on this dimension compared to MA^- participants (d = -1.39, p < .001). Whole-brain imaging results also revealed that MA⁺ participants exhibited alterations in load-induced responses predominantly in frontoparietal and default-mode areas. Specifically, while the MA^- group exhibited monotonic activation increases within frontoparietal areas and monotonic decreases within default-mode areas from 0-back to 2-back, MA⁺ participants showed a relative attenuation of these load-induced activation patterns (d = -1.55, p < .001). Moreover, increased activations in frontoparietal areas from 0- to 2-back were related to greater working memory ability among MA⁺ participants (r = .560, p = .004). No such effects were observed for default-mode areas. In sum, reductions in working memory ability were observed alongside load-induced dysfunctions in frontoparietal and default-mode areas for people with problematic methamphetamine-use experience. Among them, load-induced activations within frontoparietal areas were found to have a strong and specific relationship to individual differences in working memory ability, indicating a putative neural signature of the working memory difficulties associated with chronic methamphetamine use.

Keywords: default mode; fMRI; frontoparietal; methamphetamine; working memory.

FIGURE 1

N‐back illustration and working memory group effects. (A) Example of 0‐back (left) and 2‐back (right) load conditions. For the 0‐back condition, the valid probe was fixed stimulus (i.e., the letter “W”). For 1‐ and 2‐back conditions, valid probe‐stimuli were varied. In the 2‐back condition, a valid probe was any letter that was displayed two screens prior (e.g., “K” in the provided example). (B) Group comparisons on the 5 targeted working memory performance measures and the latent working memory dimension (WM _L ). MA⁻ = people without methamphetamine experience; MA⁺ = people with problematic methamphetamine‐use experience. d = Cohen's d group effect size; ** = p < .01; *** = p < .001.

FIGURE 2

Whole‐brain group $\times$ Load interaction effects. (A) Cortical surface projections of significant group $\times$ load interactions. Nine significant voxel clusters were observed (see Table 2, Figure 3). Significance was determined via p ≤ .001 and k > 62 faces‐touching voxel thresholds, yielding an estimated family‐wise error rate < .05. (B) Example of group $\times$ load directional effects. Group effects from 2‐back > 0‐back load contrast shown within significant clusters from the group $\times$ load interaction. Warmer colours indicate MA⁺ exhibited a more positive (i.e., less negative) load‐induced activation effect compared to MA⁻. Cooler colours indicate MA⁺ exhibited less positive load‐induced activation effect compared to MA⁻. Whole‐brain volume‐based maps of group and load main effects, group $\times$ load interaction effects, are provided in the Supplemental Materials along with load effects separated by group.

FIGURE 3

Effects by network label. Cortical surface projections of significant voxel clusters separated by frontoparietal and default‐mode network labels (smoothed for display). (Top/warm colours) Dorsal view of the four frontoparietal voxel clusters and group effects on cluster activations by load. (Bottom/cool colours) Medial view of the four default‐mode voxel clusters and group effects on cluster activations by load. Post‐hoc tests indicated significant group differences emerged for clusters in both networks during the 2‐back condition. d = Cohen's d group effect size; p _Bonferroni indicates p‐value corrected for the number of post‐hoc tests. Voxels = voxels within each cluster. See Table 2 for description of area labels and expanded cluster details.

FIGURE 4

Test of maximum network difference effects. Y‐axis reflects combined changes in frontoparietal and default‐mode cluster activations from 0‐ to 2‐back. Greater maximum network difference coefficients indicate larger combined load‐induced increases in frontoparietal activations and load‐induced decreases in default‐mode cluster activations. d = Cohen's d effect size. *** = p < .001.

FIGURE 5

Activations within network clusters and MA⁺ Participants' working memory ability. (A) Greater increases in activation within frontoparietal clusters from 0‐back to 2‐back was correlated with greater working memory ability (i.e., WM _L ) among MA⁺ participants. (B) No significant effect was observed for activations in default‐mode clusters on MA⁺ participants' working memory ability. Shaded areas reflect 95% confidence bands. Removal of potential leverage point on Y‐axes (lowest WM _L score; −1.62 SD) did not impact the significance of results: (A) r = .570, p = .003; (B) r = −.374, p = .072.