Eeg-Based Functional Connectivity – A Possible Biomarker for Neuropathic Pain in DPN

Abstract

Background: The reason some diabetic polyneuropathy (DPN) patients develop neuropathic pain while others present with painless symptoms is unknown. Altered central pain processing has been associated with neuropathic pain. Previous EEG studies found unique cortical patterns in patients with neuropathic pain however these studies were limited to univariate analysis of locally activated areas of cortex. It has been proposed that pain is encoded by complex activity and connectivity patterns which can be explored by application of Machine Learning techniques on the EEG signal. We aimed to investigate whether EEG based functional connectivity can be a potential biomarker for distinction between painful and non-painful DPN. Methods: We recorded contact-heat pain evoked potentials (CHEPs) of 120 painful DPN patients (33F, 63.2 ± 9.9 yrs) and 38 non-painful DPN patients (7F, 64.3 ± 9.5 yrs). Connectivity analysis was conducted based on a measure of synchrony (Phase Locking Value) between the recording sites. We then applied a data-driven analysis scheme to identify the most differentiating functional connections in each frequency band. These connections were validated and used to predict neuropathic pain. Results: We found overall higher cortical functional connectivity between areas of the pain matrix among painful DPN patients. In alpha and theta bands, the connectivity values were significantly different, and differentiated between the two groups with fair-to-excellent specificity (0.797) and sensitivity (0.9). Conclusion: Patients with DPN can be successfully classified into painful and non-painful based solely on EEG functional connectivity data. Increased connectivity between areas of the pain matrix in alpha and tetha band might be a biomarker for neuropathic pain.

Keywords

• Painful diabetic neuropathy
• cortical connectivity
• EEG