Abstract

Introduction

Brain function supporting error-monitoring has rarely been examined among persons living with HIV (PLWH) despite its importance for recognizing and preventing maladaptive behavior (e.g., medication non-adherence) that could lead to worsened health outcomes among this vulnerable population. As medicinal and recreational cannabis use is prevalent among PLWH [1], we aimed to assess interactive impacts of HIV infection and chronic cannabis (CB) use on error-processing brain activity and investigate implications for clinically relevant disease management behaviors.

Methods

Our sample of 103 participants (68.0% male, mean age=35.610.9 years) was stratified into four groups based on HIV serostatus and CB use history (HIV+/CB+, n=30; HIV+/CB-, n=25; HIV-/CB+, n=26; HIV-/CB-, n=22). To probe error-processing mechanisms, participants underwent fMRI scanning while completing a Go/NoGo, motor inhibition paradigm called the Error Awareness Task (EAT). Participants also completed a battery of well-validated instruments including the Revised Medication Management Test (MMT-R), an objective behavioral measure of medication management abilities [2], and self-reports of cannabis use history. Following preprocessing (FMRIPREP v1.1.1 [3]), six EAT runs were entered into a subject-level general linear model (GLM) including nuisance regressors and three task-related regressors. To characterize brain activity linked with cognitive control/failures, NoGo-correct minus NoGo-error [C-E] contrast values were assessed with a whole-brain, one-sample t-test (3dTtest++: two-tailed, pvoxel-wise=1.0e-10, cluster extent: 20 voxels). To assess HIV and CB main and interaction effects on these [C-E] contrast values, a whole-brain, 2(HIV) x 2(CB) ANOVA (3dMVM), including sex, age, and IQ as covariates, was performed. To assess brain-behavior relationships, we conducted Pearson’s correlations between averaged error-related  coefficients, extracted from identified clusters/regions of interest (ROIs), and error frequency in the EAT (ERROR COUNT). To probe clinically-relevant implications of HIV-associated alterations, we examined relationships between error-related brain activity and behavioral performance on the MMT-R (MMT SCORE) by conducting HIV-status x MMT SCORE ANCOVAs. Additionally, we tested a meditation model in which error-related brain activity (M) mediated the effect of HIV-status (X) on MMT SCORE (Y) [4, 5]. Finally, to evaluate the impact of lifetime cannabis use amount (AMOUNT), we assessed HIV-status x AMOUNT interactions on error-related brain activity and MMT SCORE among cannabis using participants (n=55).

Results

We observed error-related brain activity in the anterior insula that was associated with better EAT performance across the full sample. Regarding group effects, PLWH displayed a lack of error-related deactivation in two default mode network (DMN) hub regions (the posterior cingulate cortex [PCC] and medial prefrontal cortex [mPFC]) that was contrarily observed among HIV- controls (Fig.1A). Additionally, degree of PCC suppression was associated with improved EAT performance among HIV- controls but not among PLWH (Fig.1B). CB main and interaction effects were not detected. Across all groups, reduced error-related PCC deactivation was associated with poorer medication management performance (Fig.2A) and mediated the effect of HIV-status on medication management abilities (Fig.2B). Finally, amount of CB used over the lifetime was associated with reduced mPFC deactivation to errors among CB using, HIV- controls, and poorer medication management abilities across all CB users.

Conclusions

Our results demonstrate diminished error-related DMN suppression among PLWH linked to poor medication management. Identifying this HIV-associated, neurobiological alteration, which may contribute to high rates of medication nonadherence among this population [6, 7], could inform treatment planning and tailor self-care education.