Towards developing personalized, adaptive deep brain stimulation (aDBS) for chronic pain, we successfully predict acute and chronic pain states from the first-in-human chronic brain recordings in chronic pain patients.

DBS for chronic pain has variable efficacy and is prone to loss of effect over time. DBS may be improved by adaptively stimulating more effective brain regions with optimized parameters in response to individualized brain states. A major gap in knowledge is the underlying neural signals supporting acute (experimental) and chronic (spontaneous) neuropathic pain states.  Here, we target novel cortical regions including anterior cingulate (ACC) and orbitofrontal cortex (OFC) which have been implicated in the affective dimension of pain.

Four  patients with refractory chronic pain syndromes have received the first-in-human chronic neural recording implants for chronic pain using an investigational DBS device.  Chronic pain states were studied using frequent at-home reporting of pain scores coincident with patient-triggered neural recordings of local field potentials (LFP). To derive biomarkers, we calculate local field potential power spectra and interregional coherence between ACC and OFC. Pain state prediction is performed with support vector machine classification. Neural responses to acute pain stimuli are tested using quantitative sensory testing.

Patient reported pain scores exhibit diurnal trends and individual variability. Delta (2-4 Hz), theta (4-8 Hz) and gamma-band (30-100 Hz) activity the ACC and OFC, together with ACC-OFC coherence, predict chronic pain scores with up to 80% accuracy within individual. Biomarkers of acute pain states are distinct from chronic pain.

Neural signatures of chronic pain can be used to predict an individual’s chronic pain state and may be used to control aDBS. These signatures are distinct from experimental acute pain states, suggesting distinct underlying circuits. Neural signatures of pain are also differentially modulated by stimulation and cognitive tasks  (abstract by Chin et. al).