Altered small-world efficiency of brain functional networks in acupuncture at ST36: a functional MRI study

Abstract

Background: Acupuncture in humans can produce clinical effects via the central nervous system. However, the neural substrates of acupuncture's effects remain largely unknown.

Results: We utilized functional MRI to investigate the topological efficiency of brain functional networks in eighteen healthy young adults who were scanned before and after acupuncture at the ST36 acupoints (ACUP) and its sham point (SHAM). Whole-brain functional networks were constructed by thresholding temporal correlations matrices of ninety brain regions, followed by a graph theory-based analysis. We showed that brain functional networks exhibited small-world attributes (high local and global efficiency) regardless of the order of acupuncture and stimulus points, a finding compatible with previous studies of brain functional networks. Furthermore, the brain networks had increased local efficiency after ACUP stimulation but there were no significant differences after SHAM, indicating a specificity of acupuncture point in coordinating local information flow over the whole brain. Moreover, significant (P<0.05, corrected by false discovery rate approach) effects of only acupuncture point were detected on nodal degree of the left hippocampus (higher nodal degree at ACUP as compared to SHAM). Using an uncorrected P<0.05, point-related effects were also observed in the anterior cingulate cortex, frontal and occipital regions while stimulation-related effects in various brain regions of frontal, parietal and occipital cortex regions. In addition, we found that several limbic and subcortical brain regions exhibited point- and stimulation-related alterations in their regional homogeneity (P<0.05, uncorrected).

Conclusions: Our results suggest that acupuncture modulates topological organization of whole-brain functional brain networks and the modulation has point specificity. These findings provide new insights into neuronal mechanism of acupuncture from the perspective of functional integration. Further studies would be interesting to apply network analysis approaches to study the effects of acupuncture treatments on brain disorders.

Figure 1. A schematic illustration of the design paradigm and brain network construction.

(A), The points of stimulation used in the present study: verum acupuncture at ST36 (ACUP) and sham points (SHAM); (B), Two scans were performed before and after the stimulation at both ACUP and SHAM, which were used to construct brain networks, respectively. (C) Top, under each condition, a correlation matrix was obtained for each subject by calculating inter-regional Pearson correlation coefficient of mean time series among 90 regions; Middle, these correlation matrices were further converted into binary versions (i.e., adjacency matrices) by applying a thresholding procedure such that the elements were set to 1 if their absolute correlation coefficients were larger than a predefined threshold and 0 otherwise; Bottom, the obtained binary matrices could be finally represented as networks or graphs that were composed of brain nodes and edges.

Figure 2. The local and global efficiency of random, regular and actual functional brain networks as a function of cost.

The brain networks under each condition showed higher local efficiency than the matched random networks (A) and higher global efficiency than the matched regular networks (B) at the whole cost range between 0.1 and 0.4 used in the present study. Thus, the brain networks under each condition exhibited small-world properties. The brain networks were also found to be economical because both the local and global efficiency were much higher than the required cost.

Figure 3. Between-condition differences in the integrated global network parameters.

, , , and denote the local efficiency, global efficiency, normalized local efficiency, normalized global efficiency, and efficiency-based small-worldness, respectively. Note that the local efficiency () was greater after acupuncturing ST36, but not in the case of SHAM. n.s., non-significant, *, P<0.05.

Figure 4. The hubs of functional brain networks.

The nodal sizes indicate their relative nodal degree within each condition. Regions with normalized nodal degree greater than mean + SD were identified as hubs. Note that the connectivity backbone (sparsity  = 5%) was obtained by thresholding the mean correlation matrix under each condition. For more details, see Table 2. L, left; R, right.

Figure 5. Regions showing significant point- (A) and stimulation- (B) related differences in regional nodal degree.

The node sizes indicate the effects (i.e., t values) of interest on nodal degree. The threshold was P<0.05 (uncorrected). Note that the connectivity backbone (sparsity  = 5%) was obtained by thresholding the mean correlation matrix across all subjects and conditions. For more details, see Table 3. ACUP, acupuncture at ST36; SHAM, acupuncture at sham point; Before, before stimulation; After, after stimulation; L, left; R, right.

Figure 6. Regions showing significant point- (A) and stimulation- (B) related differences in regional nodal homogeneity.

The node sizes indicate the effects (i.e., t values) of interest on nodal degree. The threshold was P<0.05 (uncorrected). Note that the connectivity backbone (sparsity  = 5%) was obtained by thresholding the mean correlation matrix across all subjects and conditions. For more details, see Table 4. ACUP, acupuncture at ST36; SHAM, acupuncture at sham point; Before, before stimulation; After, after stimulation; L, left; R, right.