Rewiring Chronic Pain

Chronic pain is estimated to affect 30% of adults and remains poorly managed by pharmacological approaches, which yield only modest pain reductions (~0.7-0.9 points on a 0-10 scale) and carry risks of sedation, misuse, and dependence. Short-term psychological interventions, particularly Pain Reprocessing Therapy (PRT), have shown large effects in randomized trials, yet the mechanisms driving their efficacy, and the specific contributions of cognitive versus somatic components, remain unclear.

Chronic pain has been consistently associated with structural and functional brain alterations in regions supporting sensory, affective, and interoceptive processing, including the ventromedial prefrontal cortex, anterior cingulate cortex, insula, and nucleus accumbens. Contemporary frameworks (e.g., the dynamic pain connectome) conceptualize chronic pain as a disorder of distributed network dysregulation rather than isolated regional dysfunction, motivating whole-brain functional connectivity, gray matter, and white matter analyses. At the electrophysiological level, slow-wave EEG activity, particularly theta (4-8 Hz) and alpha (8-12 Hz) power, has been linked to low-arousal states induced by hypnosis and mindfulness meditation, both of which produce clinical pain reduction. Animal evidence further links theta activity to neuroplasticity. On this basis, we hypothesize that arousal-state shifts indexed by EEG slow-wave activity and HRV coherence during PRT sessions will predict clinical improvement, and that pre-post MRI will reveal corresponding network and structural changes.

The proposed study is a single-blind, three-arm randomized controlled trial that will compare: (1) Full PRT integrating cognitive (psychoeducation) and somatic components (see full protocol at www.painreprocessingtherapy.com/wp-content/uploads/2024/10/PRT-Group-Protocol.pdf); (2) Partial PRT delivering only the cognitive component; and (3) Treatment as Usual (TAU). Both intervention arms will consist of eight 60-minute individual in-person sessions delivered twice weekly by trained providers, with real-time EEG monitoring during each session accessible to the provider to dynamically inform the therapeutic process. Randomization will be stratified by age and medication use, and participants will also be randomly assigned across providers to prevent provider-related confounding. A pilot phase (3-5 participants per intervention arm) will be conducted with EEG but without MRI to refine procedures.

Home practice will be standardized within each arm via a unified 8-10 minute audio recording: somatic tracking for the Full PRT arm, and an external-attention mindfulness exercise (sounds, smells, and visual imagery, deliberately avoiding bodily focus) for the Partial PRT arm, isolating the somatic component. Adherence will be self-reported. Assessment will occur at baseline, mid-treatment (abbreviated battery), post-treatment, and at follow-up (MRI up to 6 months; questionnaires up to 5 years). Twice-daily pain ratings and pre/post-session relaxation and valence ratings (0-10) will be collected throughout treatment.

Multimodal physiological and neural data will be collected as follows. MRI scans (pre- and post-treatment, ~40 min each) will include high-resolution T1-weighted anatomical imaging, resting-state fMRI, diffusion MRI, and 10 minutes of naturalistic audio-guided fMRI (5 min somatic tracking, 5 min neutral book chapter audio, both with eyes-closed via noise-cancelling headphones). EEG will be recorded during every intervention session using the g.tec Unicorn BCI Core-8 (8 channels, flexible montage). Continuous physiological monitoring via Garmin Vivosmart 5 wearables will capture heart rate variability (beat-to-beat intervals), sleep architecture, and daily activity levels for up to 6 weeks. Blood and saliva samples may also be collected.

Analyses will be conducted in Python. Clinical and physiological outcomes will be modeled using Bayesian mixed-effects general linear models to test whether markers of low-arousal states (EEG alpha/theta activity and HRV coherence) predict pain reduction, accounting for repeated measures and covariates. fMRI data will be preprocessed with fMRIPrep and analyzed using parcellation-based connectivity and Network-Based Statistic; structural MRI with FreeSurfer (gray matter volume); diffusion MRI with DIPY (fractional anisotropy); naturalistic fMRI with voxelwise encoding models; and EEG with MNE-Python. Exploratory multivariate prediction of clinical response will use machine learning classifiers.