Data-driven Evaluation of Brain Network Engagement with Reward Task Gain Anticipation

769 

Abstract Submission 

Su Hyoun Park^1,2, Andrew Michael³, Anne Baker^1,2, Carina Lei^1,2, Katherine Martucci^1,2,3

¹Department of Anesthesiology, Human Affect and Pain Neuroscience Laboratory, Duke University School, Durham, NC, ²Center for Translational Pain Medicine, Duke University Medical Center, Durham, NC, ³Duke Institute for Brain Sciences, Duke University, Durham, NC

Su Hyoun Park  
Department of Anesthesiology, Human Affect and Pain Neuroscience Laboratory, Duke University School|Center for Translational Pain Medicine, Duke University Medical Center
Durham, NC|Durham, NC

Andrew Michael  
Duke Institute for Brain Sciences, Duke University
Durham, NC

Anne Baker  
Department of Anesthesiology, Human Affect and Pain Neuroscience Laboratory, Duke University School|Center for Translational Pain Medicine, Duke University Medical Center
Durham, NC|Durham, NC

Carina Lei  
Department of Anesthesiology, Human Affect and Pain Neuroscience Laboratory, Duke University School|Center for Translational Pain Medicine, Duke University Medical Center
Durham, NC|Durham, NC

Katherine Martucci  
Department of Anesthesiology, Human Affect and Pain Neuroscience Laboratory, Duke University School|Center for Translational Pain Medicine, Duke University Medical Center|Duke Institute for Brain Sciences, Duke University
Durham, NC|Durham, NC|Durham, NC

Reward motivation is essential in shaping human behavior and cognition¹. Brain reward systems are critical in modulating pain experiences^2,3,4 and consistently appear to be dysregulated among individuals with chronic pain conditions such as fibromyalgia^5,6,7,8. Fibromyalgia is characterized by widespread musculoskeletal pain, fatigue, cognitive problems, and mood-related symptoms. To gain a more comprehensive understanding of how brain reward circuits are altered in chronic pain, we used Independent Component Analysis (ICA) to investigate how brain networks contribute to altered reward processing in fibromyalgia.

From female individuals with fibromyalgia (N=24) and female healthy controls (N=24), we acquired fMRI data while participants performed a monetary incentive delay (MID) reward task. A group ICA (GICA) was performed on the functional data for the two MID task scans using GIFT v4.0 software. GICA computes brain functional networks and their timecourses in a data-driven manner. The denoised preprocessed functional data were provided as input into the GICA toolbox. The fMRI data were decomposed into 30 functional networks and were visually inspected. Networks of interest were selected based on their relevance to reward processing and included the left motor network, basal-ganglia network, and value-driven attention network. These functional networks were selected prior to conducting ICA. To measure patients vs. healthy control group differences, we evaluated each functional network's temporal correlation with the gain anticipation task timecourse. First, we modeled the gain anticipation timecourse based on the task's timing. Using the FMRIB Software Library⁹, we constructed the task timecourse model by convolving the timecourse of gain anticipation with the hemodynamic response function. Then, we evaluated the correlation between the gain anticipation timecourse vs. each ICA-derived network timecourse. At this step, we regressed out six motion parameters (three translation and three rotation) from each participant's functional network timecourse. The resulting correlation coefficients indicated the degree of similarity between each ICA network's timecourse and the gain anticipation timecourse at the individual level. These correlation coefficients were then Fisher's r-to-z transformed before group analyses. We repeated this process for all participants. Finally, we performed the group comparison of correlation coefficients between patients vs. healthy controls (Figure 1).

Compared to controls, the fibromyalgia cohort demonstrated significantly stronger correlation between the left motor network timecourse and the task-timecourse, indicating the left motor network was more engaged with gain anticipation in fibromyalgia. In an exploratory analysis, we compared motor network engagement during early vs. late phases of gain anticipation. Across both cohorts, greater motor network engagement (i.e., stronger correlation between network timecourse and task-timecourse) occurred during the late timepoint, which reflected enhanced motor preparation immediately prior to target response. Consistent with the main results, patients exhibited greater engagement of the motor network during both early and late phases as compared with healthy controls (Figure 2). Visual-attention and basal ganglia networks revealed similar engagement in the task across groups. As indicated by post-hoc analyses, motor network engagement was positively related to anxiety (r = 0.434, p = 0.043) and negatively related to reward responsiveness (r = -0.429, p = 0.036).

In summary, by using a novel data-driven ICA approach to analyze task-based fMRI data, we identified enhanced reward-task related engagement of the motor network in fibromyalgia. This study provided broader insights to how brain networks are altered during performance of a reward task in chronic pain.

Reward and Punishment ¹

fMRI Connectivity and Network Modeling ²

Perception: Pain and Visceral

Anxiety

FUNCTIONAL MRI

Pain

Other - chronic pain; brain activity; value; motor network; visual network; motivation; monetary incentive delay task