Acupuncture analgesia involves modulation of pain-induced gamma oscillations and cortical network connectivity

Figure 1. Pain ratings. The top panel shows the grand mean time course of pain-rating change (NRS change). All pain ratings were normalized to the mean pain rating value of the pre-block. Significant differences based on a running t-test are shaded in grey. The bottom panel shows the mean NRS change for each block. The asterisks highlight significant differences based on a paired t-test. Besides a significant effect with lower pain ratings after successive block repetition, acupuncture treatment induced a stronger reduction of pain ratings compared to placebo treatment in each treatment block.

Figure 2. Heart rate variability. Top panel: ECG spectra for sham and verum condition revealed LF (0.04–0.15 Hz) and HF (0.15–0.4 Hz) peaks. Short-term parameters such as the HF-Spectrum, SD1, DFA-alpha-1 and RMSSD indicate a decrease in vagal activity over time for both conditions. This effect was significant stronger, using a paired t-test, during verum treatment for treatment block 1 and 2 as indicated by asterisks. Long-term parameters such as the LF-spectrum, SD2, DFA-alpha-2 and SDNN were unaffected by repetition and treatment manipulation.

Figure 3. Gamma oscillations. Top left: Our analysis revealed laser-pain induced gamma oscillations with a centro-frontal topographical distribution. Bottom Left: Time-frequency analysis of signals within this centro-frontal ROI showed a maximum at 74 Hz and 240ms for the grand mean data. This gamma activity decreased over time and the gamma decrease was significantly stronger, based on paired t-tests, during the last treatment block for verum acupuncture compared to the sham condition (right panel), as indicated by an asterisk.

Figure 4. Source localization of pain-induced gamma power. Using eLORETA five significant gamma-activation clusters were localized in the contralateral insula (Ins), mid-cingulate cortex (MCC), contralateral primary somatosensory cortex (SI) and prefrontal cortex (PFC). Colour-scaled Z-scores based on a cluster randomization approach are projected on a MNI-standard brain. Power analysis within these anatomical locations revealed a significant repetition effect (based on a two-way ANOVA) within MCC and SI. No significant power effect was found between sham and verum treatment on the source level.

Figure 5. Coupling of gamma oscillations. Using the multivariate interaction measure (MIM), neuronal coupling was analyzed between the five regions obtained from source reconstruction results, as shown on the left panel. Only statistically significant connections (based on paired t-tests using z-normalization of MIM values) are shown and the strength of connectivity is expressed by the thickness of the line. The top row show data for verum treatment, the bottom row those for sham treatment. Before the intervention (Pre) connections were identical for both days. Comparing the last block (Treat 3), more connections were present in the verum condition between both prefrontal cortices (PFC) and mid cingulate cortex (MCC), insula (Ins) and primary somatosensory cortex (SI).

Figure 6. Correlation between cortical connectivity, pain ratings and HRV parameters. Change in pain ratings (NRS change) was significant negatively correlated (blue line) with the strength of pain induced coupling (multivariate interaction measure, MIM) between contralateral insula (Ins) and mid-cingulate cortex (MCC). The same negative correlation was found between MIM and the HF spectral power value (HF). Furthermore, we identified positive correlations (red lines) between HF parameters and MIM for connections between insula and ipsilateral prefrontal cortex (PFCi) and between both prefrontal cortices (PFCc and PFCi).