Low Frequency Oscillations in the Medial Orbitofrontal Cortex Mediate Widespread Hyperalgesia Across Pain Conditions

Figures

• 

  Figure 1.. Schematic overview of the experimental design for concurrent QST and EEG. Left panels: Participants with chronic low back pain (cLBP) and chronic pancreatitis (CP) were enrolled. Middle panels: EEG was recorded during pinprick stimulation at the lower back and dorsum of the right hand (for cLBP), and at the upper abdomen and dominant forearm (for CP). Right panels: Participants were pain-phenotyped, and EEG source localization was performed, focusing on mOFC and dlPFC as regions of interest (ROIs).

• 

  Figure 2.. Pain phenotype effects on time-frequency responses in the mOFC (top two rows) and dlPFC (bottom two rows) based on cLBP data (n=67). Left panels display the data-driven time-frequency basis for the power; right panels show posterior summaries of group-by-condition effects projected onto that basis (means and 90% credible intervals), adjusted for pain intensity, age, and sex. Results are shown by stimulus condition (Back 32 mN, Back 256 mN, Hand 32 mN, Hand 256 mN) and laterality (top: left hemisphere (lh); bottom: right hemisphere (rh)).

• 

  Figure 3.. Time-frequency projection pattern for the mOFC used in biosignature. The pattern highlights increased delta/theta (1–7 Hz) (weighted by +1) and decreased alpha (8–12 Hz) (weighted by - 1) oscillatory power during the 0.4–0.7 second post-stimulus window. This weighting scheme was applied to each participant’s mOFC time-frequency response (left and right hemispheres, averaged) to compute a scalar biosignature score reflecting expression of the widespread hyperalgesia profile.

• 

  Figure 4.. Top panels: Projection scores onto the mOFC biosignature in the CP sample (n=18), by 3 pain phenotype groups, for the left and right mOFC (left and right panels), respectively. Group means ± standard error of the mean (SEM) are shown for CP participants, grouped by pain phenotype (Widespread, Segmental, No Hyperalgesia). The mOFC projection scores successfully differentiated pain phenotypes, with consistent patterns observed across the both hemispheres. Bottom panels: Projection scores onto the mOFC biosignature based on the pooled (cLBP + CP) sample (n=85), by 3 pain phenotype groups, for the left and right mOFC (left and right panels), respectively. Group means ± SEM for participants in the pooled sample, grouped by pain phenotype (Widespread Hyperalgesia, Localized Pain, No Pain). The mOFC projection scores successfully distinguished pain phenotypes, with consistent patterns observed across the both hemispheres.

• 

  Figure 5.. Pain phenotype effects on time-frequency responses in the mOFC and dlPFC based on pooled data (cLBP and CP; n=85). Left panels display the data-driven time-frequency basis; right panels show posterior summaries of group-by-condition effects projected onto that basis (means and 90% credible intervals), adjusted for pain intensity, age, sex and cohort. Results are shown by stimulus condition (Affected/Unaffected site and High/Low intensity) and laterality (top: left hemisphere; bottom: right hemisphere). In the pooled analysis, Localized Pain refers to regional pain with either localized hypersensitivity or no measurable hyperalgesia.

• 

  Figure 6.. Left panel: Change in Brief Pain Inventory (BPI) pain severity subscale at 3 months after a technically successful ERP, stratified by pain phenotypes. Negative values indicate improvement in pain. Values are median and interquartile range (IQR). Overall, participants with widespread or segmental hyperalgesia showed minimal or no responses while participants with no hyperalgesia exhibited better responses, albeit with greater variability. Right panel: Post-treatment pain change (negative values indicate pain improvement) vs. baseline EEG biomarker scores. Linear regression lines are shown for participants with and without widespread hyperalgesia. The interaction model (EEG × widespread hyperalgesia) that incorporated the EEG marker outperformed models based on the widespread hyperalgesia indicator alone or the 3-level pain phenotypes (= 71% vs. 21% and 27%) and achieved LOOCV AUC = 88% (vs. 63% and 71%).