Investigation of acupoint specificity by multivariate granger causality analysis from functional MRI data.

Abstract

To investigate the acupoint specificity by exploring the effective connectivity patterns of the poststimulus resting brain networks modulated by acupuncture at the PC6, with the same meridian acupoint PC7 and different meridian acupoint GB37.
The functional MRI (fMRI) study was performed in 36 healthy right-handed subjects receiving acupuncture at three acupoints, respectively. Due to the sustained effects of acupuncture, a novel experimental paradigm using the nonrepeated event-related (NRER) design was adopted. Psychophysical responses (deqi sensations) were also assessed. Finally, a newly multivariate Granger causality analysis (mGCA) was used to analyze effective connectivity patterns of the resting fMRI data taken following acupuncture at three acupoints.
Following acupuncture at PC6, the red nucleus and substantia nigra emerged as central hubs, in comparison with the fusiform gyrus following acupuncture at GB37. Red nucleus was also a target following acupuncture at PC7, but with fewer inputs than those of PC6. In addition, the most important target following acupuncture at PC7 was located at the parahippocampus.
Our findings demonstrated that acupuncture at different acupoints may exert heterogeneous modulatory effects on the causal interactions of brain areas during the poststimulus resting state. These preliminary findings provided a clue to elucidate the relatively function-oriented specificity of acupuncture effects.

Full text

Page 1

Original Research
Investigation of Acupoint Specificity by
Multivariate Granger Causality Analysis From
Functional MRI Data
Yuanyuan Feng, PhD,1Lijun Bai, PhD,1Wensheng Zhang, PhD,1*Ting Xue, PhD,2
Yanshuang Ren, PhD,3,4Chongguang Zhong, PhD,1Hu Wang, PhD,1Youbo You, PhD,1
Zhenyu Liu, PhD,1Jianping Dai, MD,5Yijun Liu, PhD,6and Jie Tian, PhD1,2
Purpose: To investigate the acupoint specificity by explor-
ing the effective connectivity patterns of the poststimulus
resting brain networks modulated by acupuncture at the
PC6, with the same meridian acupoint PC7 and different
meridian acupoint GB37.
Materials and Methods: The functional MRI (fMRI) study
was performed in 36 healthy right-handed subjects
receiving acupuncture at three acupoints, respectively.
Due to the sustained effects of acupuncture, a novel ex-
perimental paradigm using the nonrepeated event-related
(NRER) design was adopted. Psychophysical responses
(deqi sensations) were also assessed. Finally, a newly mul-
tivariate Granger causality analysis (mGCA) was used to
analyze effective connectivity patterns of the resting fMRI
data taken following acupuncture at three acupoints.
Results: Following acupuncture at PC6, the red nucleus
and substantia nigra emerged as central hubs, in com-
parison with the fusiform gyrus following acupuncture at
GB37. Red nucleus was also a target following acupunc-
ture at PC7, but with fewer inputs than those of PC6. In
addition, the most important target following acupuncture
at PC7 was located at the parahippocampus.
Conclusion: Our findings demonstrated that acupunc-
ture at different acupoints may exert heterogeneous mod-
ulatory effects on the causal interactions of brain areas
during the poststimulus resting state. These preliminary
findings provided a clue to elucidate the relatively func-
tion-oriented specificity of acupuncture effects.
Key
delayed effect
J. Magn. Reson. Imaging 2011; 34:31–42.
V
Words:
acupuncturespecificity;mGCA;fMRI;
C 2011 Wiley-Liss, Inc.
ORIGINATED IN ANCIENT China, acupuncture has
currently been recognized in the Western medicine as
an important complementary therapy (1). However,
the mechanism underlying this traditional interven-
tion has not been clear and needs further investiga-
tions. In the past decades, functional MRI (fMRI) has
been used to study the anatomy and physiological
function underlying the acupuncture.
Previous studies mainly focused on the neural activ-
ities involving the acute effects of acupuncture (2–4).
Hui and colleagues reported that acupuncture can
exert modulatory effects on the limbic system and sub-
cortical gray structures in the human brain (3). The
modulation of cerebellar activities underlying the acu-
puncture stimulation was further proved later (4). One
recent report has provided additional evidence to sup-
port that acupuncture can modulate the limbic-para-
limbic-neocortical network (2). Overall, these studies
generally adopted the block-designed fMRI paradigm,
and therefore the temporal changes in the blood oxygen
level-dependent (BOLD) signal as predicted by the gen-
eralized linear model (GLM) conform to the ‘‘on-off’’
specifications. According to the theory of the traditional
Chinese medicine (TCM), acupuncture can induce
long-lastingeffectsevenaftertheneedling
1Key Laboratory of Complex Systems and Intelligence Science, Institute
of Automation, Chinese Academy of Sciences, Beijing, China.
2Life Science Research Center, School of Life Sciences and Technology,
Xidian University, Xi’an, Shaanxi, China.
3Department of Radiology, Xuanwu Hospital, Capital Medical
University, Beijing, China.
4Department of Radiology, Guang’anmen Hospital, Chinese Academy
of Traditional medicine, Beijing, China.
5Department of Radiology, Beijing Tiantan Hospital, Capital University
of Medical Sciences, Beijing, China.
6Departments of Psychiatry and Neuroscience, McKnight Brain
Institute, University of Florida, Gainesville, Florida, USA.
Contract grant sponsor: the Project for the National Key Basic
ResearchandDevelopmentProgram
2011CB707700, the Chinese Academy of Sciences; Contract grant
number: KGCX2-YW-129; Contract grant sponsor: the National High
Technology Research and Development Program of China (863
Program); Contract grant number: 2008AA01Z121; Contract grant
number: 2007AA01Z338;Contract
Natural Science Foundation of China; Contract grant number:
30873462;Contractgrantnumber:
number: 60901064; Contract grant number: 90924026; Contract
grant number: 81071137; Contract grant number: 81071217.
Drs. Feng and Bai contributed equally to this work.
*Address reprint requests to: W.Z., Institute of Automation, Chinese
Academy of Science, P.O. Box 2728, Beijing, 100190, China. E-mail:
wensheng.zhang@ia.ac.cn
Received May 20, 2010; Accepted March 3, 2011.
DOI 10.1002/jmri.22585
View this article online at wileyonlinelibrary.com.
(973)underGrantNo.
grantsponsor:theNational
30970774;Contractgrant
JOURNAL OF MAGNETIC RESONANCE IMAGING 34:31–42 (2011)
CME
V
C 2011 Wiley-Liss, Inc.
31

Page 2

manipulation being terminated (5). Considering such
delayed effect, the ‘‘off-state’’ in the block design may
remain some effects of acupuncture, and cannot ideally
return to the baseline level (6). In addition, one recent
study has already pointed that neural responses
induced by acupuncture have the saliently time-vary-
ing characteristics (6–9). This temporal aspect of the
BOLD response to acupuncture may violate the
assumptions of the block-designed GLM estimates,
which may be susceptible to errors of statistical signifi-
cance (6). Therefore, a newly experimental paradigm
namely the NRER design was adopted to explore the
neural responses induced by acupuncture. This design
incorporates the sustained phase which focuses on the
dynamic of acupuncture-related brain activities. Con-
sidering the sustained effects of acupuncture, the
NRER design may be more suitable to elucidate its spe-
cific mechanism underlying acupuncture.
Several studies have paid attention to the sustained
effect of the acupuncture and its influence on the
poststimulus resting state networks (8–12). Dhond et
al revealed that acupuncture can influence intrinsic
connectivity in the resting brain networks, i.e., default
mode network (11). Our group has investigated the
functional connectivity of brain networks involved in
acupuncture and demonstrated that acupuncture can
exert modulatory effect on the insula-anchored brain
network during the poststimulus resting period (8).
One recent study further proved the existence of a
large anti-correlated functional connectivity networks
of resting brain modulated by acupuncture (12).
These studies demonstrated the existence of function-
guided brain networks underlying the specific effect of
acupuncture. Most of them adopted the standard
functional connectivity analysis, which was a kind of
undirected graph analysis of temporal correlations
between time series in different brain regions. How-
ever, little was knownabout
strength of the information flow between these brain
regions modulated by acupuncture. Further investiga-
tion of the interregional causal interactions may be
helpful to explainthe
underlying acupuncture. Recently, a newly multivari-
ate Granger causality analysis (mGCA) has been intro-
duced as an effective connectivity method to analyze
causal relations among multiple brain areas from
fMRI data (13–15). By exploring this approach to ana-
lyze the causal influences of the activated regions
modulated by acupuncture, we can account for the
acupuncture modulatory effects on multiple relevant
regions simultaneously. Unraveling the organization
of the large-scale cortical networks following acupunc-
ture would pave the way to better understand the
neurophysiological function underlying acupuncture.
In the present study, we combined mGCA (14) with
the NRER design to evaluate the effective connectivity
patterns among multiple brain regions during the
poststimulus resting state following acupuncture at
the PC6, with the same meridian acupoint PC7 and
different meridian acupoint GB37. We examined the
directionalityandstrength
between multiple brain regions following acupuncture
at three acupoints to find whether there was the rela-
thedirectionand
neurophysiological action
ofcausalinfluence
tively specific modulatory effect underlying acupunc-
ture at different acupoints. By detecting the specificity
of brain effective connectivity underlying the same
meridian and different meridian acupoints, we can
provide further evidence to explore the functional
specificity of acupuncture.
MATERIALS AND METHODS
Subjects
Thirty-six healthy right-handed college students (18
males, aged 21.4 6 1.3) were recruited from a homo-
geneous group to reduce intersubject variabilities. All
subjects were acupuncture naı ¨ve according to the
Edinburgh Handedness Inventory (16). After given a
complete description of the study, each of the subjects
signed the informed consent form. All subjects were
free of any major medical illnesses, any head trauma,
any neuropsychiatric disorders, any intake of pre-
scription medications within the last month, as well
as any contraindications for exposure to a high mag-
netic field. Subjects were allocated to receive once
acupuncture at one of the three acupoints in a ran-
dom sequence by using an envelope method, with the
gender ratio balanced (analysis of variance, P > 0.6)
among different groups. All protocols were approved
by a local subcommittee on Human Studies.
Experimental Paradigm
All subjects underwent a resting state scan. The
NRER scanning for every acupoint incorporated 2-min
needling manipulations, preceded by a 1-min rest,
and followed by another 1-min rest scanning (shown
in Fig. 1). During the experiment, the subjects were
instructed to keep their eyes closed and remain
relaxed without engaging in any mental tasks. Each
subject only received once acupuncture at one of the
three acupoints in a random order: PC6 (Neiguan),
located between the tendon of palmaris longus and
flexor carpi radialis at a depth approximately 3 cm
above the midpoint of the transverse crease of the
wrist (17); PC7 (Daling), located between the tendons
of the long palmar muscle and radial flexor muscle at
the midpoint of the crease of the wrist (18); GB37
(Guangming), located in the lateral aspect of the lower
leg with needle being inserted on the anterior border
of the fibula (19). A sterile disposable 38 gauge
Figure 1. Experimental paradigm. The entire run for each acu-
point lasted for 4 min. For statistical analyses, the signal inten-
sity during the 1 min rest phase served as a control baseline for
detecting the changes in signal intensity during acupuncture
stimulation, thereby functionally defining the regions of interest.
In addition, the data from the 1 min rest phase were used for
mGCA. [Color figure can be viewed in the online issue, which is
available at wileyonlinelibrary.com.]
32Feng et al.

Page 3

stainless steel acupuncture needle (0.2 mm in diame-
ter and 40 mm in length) was used to deliver acupunc-
ture stimulation. During acupuncture, the needle was
rotated manually clockwise and counterclockwise for 1
min at a rate of 60 times per min by a balanced ‘‘tonify-
ing and reducing’’ technique (3). The precise locations
of needling, the presumed acupuncture effects, and
the stimulation paradigm were not divulged. The pro-
cedure was performed by the same experienced and li-
censed acupuncturist on all subjects.
The subjects were questioned about aching, pres-
sure, soreness, heaviness, fullness, warmth, coolness,
numbness, tingling, dull or sharp pain and any other
sensations they felt during the scan after each fMRI
scanning (20,21). The questionnaire used a 10-point
visual analogue scale (VAS), which was scaled at 0 ¼
no sensation, 1–3 ¼ mild, 4–6 ¼ moderate, 7–8 ¼
strong, 9 ¼ severe and 10 ¼ unbearable sensation.
The subjects were excluded from further analysis if
they experienced sharp pain (greater than the mean
by more than 2 standard deviations) (20). None expe-
rienced the sharp pain among the 36 participants.
Data Acquisition and Analysis
MRI data were acquired using a 3.0 Tesla (T) Signa
(GE) MR scanner. Head movements were prevented by
a custom-built head holder. The images were parallel
to the anterior commissure–posterior commissure line
and covered the whole brain. Thirty-two axial slices
were obtained using a T2*-weighted single-shot, gradi-
ent-recalled echo planar imaging (EPI) sequence (field
of view [FOV] ¼ 240 mm ? 240 mm, matrix ¼ 64 ?
64, thickness ¼ 5 mm, repetition time [TR] ¼ 2000
ms, echo time [TE]¼ 30 ms, flip angle ¼ 90?). After the
functional run, high-resolution structural information
on each subject was also acquired using three-dimen-
sional (3D) MRI sequences with a voxel size of 1 mm3
for anatomical localization (TR ¼ 2.7 s, TE ¼ 3.39 ms,
matrix ¼ 256 ? 256, FOV ¼ 256 mm ? 256 mm, flip
angle ¼ 7?, slice thickness ¼ 1 mm).
After that, all images were preprocessed using sta-
tistical parametric mapping (SPM5, http://www.fil.io-
n.ucl.ac.uk/spm/). First, the image data underwent
slice-timingcorrectionand
motions using least-squares minimization. None of
the subjects had head movements exceeding 1 mm on
any axis and head rotation greater than one degree.
Thus all subjects (n ¼ 36) were left. A mean image
created from the realigned volumes was coregistered
with the subject’s individual structural T1-weighted
volume image. Then, the standard MNI template pro-
vided by SPM5 was used in spatial normalization with
resampling at 2 mm ? 2 mm ? 2 mm (22). In the end,
the functional images were spatially smoothed with a
3D Gaussian kernel (FWHM ¼ 6 mm).
realignmentforhead
Definition of Regions of Interest
Taking into account the sustained effects of acupunc-
ture, the mean signal intensity of the resting period
preceded by the active stimulation was taken as the
baseline. For each subject, the difference in the BOLD
response between stimulus and baseline conditions
was estimated at every voxel across the whole brain
volume by using the general linear model (GLM) in
SPM5. The obtained t-maps at individual levels were
then entered into the ‘‘random effect’’ group analysis
framework by the one-sample t-test (d.f. ¼ 11) sum-
mary statistic (P < 0.001, uncorrected). The statistical
maps indicated the brain activation in response to
acute effects of acupuncture stimuli, thereby func-
tionally defining the regions of interest (ROIs). Each
peak voxel with its nearest 10 neighbors was defined
as a group ROI.
Considering the anatomical variance across sub-
jects, subject-specific peak voxels and subject-specific
ROIs were defined on individual t-maps as follows.
The given group ROI was used as a mask and then,
based on individual t-maps, the voxel with the largest
t-value within this mask served as the subject-specific
peak voxel. ROIs were selected based on the acupunc-
ture-stimulation results. First, the time series corre-
sponding to the poststimulus period of BOLD signal
intensities from these selected ROIs were selected.
Then, the time series were averaged across voxels
within each ROI and normalized across subjects sepa-
rately for each group to form a single vector per ROI.
In this manner, we obtained a total of 25 ROI time se-
ries related to acupuncture stimulation for each acu-
point for further analysis.
mGCA
In this study, we used mGCA to describe the effective
connectivity in the postacupuncture resting brain
(14).Thisapproachdetected
between brain regions by computing directed transfer
function (DTF) from a multivariate autoregressive
model of the time-course of selected ROIs. Based on
the principle of Granger causality, the DTF was ren-
dered in a multivariate formulation (23). Therefore the
DTF can effectively model the inherently multivariate
nature of neuronal networks. Preprocessed dataset
were implemented using this method. The algorithm
was coded in MATLAB7 (The MathWorks Inc.) and
details were as follows.
First, we need to construct the multivariate autore-
gressive (MVAR) model of the times series. Let X(t) ¼
[x1(t), x2(t),..., xM(t)] be a matrix representing data
from M ROIs, in which xi(t) is the time series corre-
sponding to ith ROI. The MVAR model of order p is
given by the following:
causalinteractions
XðtÞ ¼
X
p
x?1
AðnÞXðt ? nÞ þ EðtÞ½1?
Traditionally, the model order was determined using
the Akaike information criterion (24). Here, a model
order of one was chosen (14). A(n) is the matrix of pre-
diction coefficients composed of elements aij(n). E(t) is
the vector corresponding to the residual errors. Then
applying the Fourier transform to Eq. [1] is given as
follows:
Xðf Þ ¼ A?1ðf ÞEðf Þ ¼ Hðf ÞEðf Þ½2?
Acupoint Specificity by Granger Causality Analysis33

Page 4

Where
aijðf Þ ¼ dij?
X
p
n¼1
aijðnÞe?i2pfnandHðf Þ ¼ A?1ðf Þ½3?
Here, dij is the Dirac-delta function, which is 1
when i ¼ j and 0 otherwise. Also, i ¼ 1. . .M, j ¼ 1. .
.M. H(f) is the frequency domain transfer matrix and
hij(f) represents its ith row and jth column element.
hij(f) is defined as the nonnormalized DTF (25) corre-
sponding to the influence of ROI j onto ROI i. The
direct DTF (dDTF) was obtained by multiplying hij(f)
with the partial coherence between ROIs i and j (25).
This operation ensures that direct connections are
emphasized and mediated influences are de-empha-
sized. To calculate the partial coherence, the cross-
spectra were computed as follows:
Sðf Þ ¼ Hðf ÞVH?ðf Þ½4?
Here, V is the variance of the matrix E(f), and the
asterisk denoted transposition and complex conjuga-
tion. Then we obtained the partial coherence between
ROIs i and j:
vijðf Þ ¼
M2
ijf ð Þ
Mii f ð ÞMij f ð Þ
½5?
where Mij(f) is the minor obtained by removing the ith
row and jth column from the matrix S (26). The par-
tial coherence lies in the range of (0, 1). Here, a value
of 0 indicates no direct association between the ROIs
with the influence of all other ROIs removed. While if
the value is 1 indicates there exists complete direct
association. Finally the dDTF is defined as the sum of
all frequency components of the product of the non-
normalized DTF and partial coherence as given in the
equation below:
dDTFij¼
X
f
hij f ð Þvij f ð Þ½6?
In the end, we obtained the value of dDTF, which
onlyreflectsthemagnitude
between the ROIs.
The statistical significance of the path weights was
ascertained using surrogate data (25,27). Surrogate
data were generated by randomizing the phase of the
original time series spectrum while retaining its mag-
nitude. A null distribution was obtained by generating
2500 sets of surrogate data and calculating the dDTF
(for every connection) from these 2500 datasets. The
dDTF value obtained from the original time series was
verified using with the null distribution for a one-
tailed test with the significant a P-value of 0.05. The
effective connectivity network of the 25 ROIs was con-
structed by visualizing the significant dDTF (P < 0.05)
obtained after running the statistical significant test.
Additionally, we adopted In-Out degree to compare
the effective connectivity network for three acupoints
(28). The In-Out degree of a node is defined as the dif-
ference between its In-degree (the number of causal
in-flows) and Out-degree (the number of causal out-
ofcausalinfluence
flows). In-Out degree can identify hub regions that have
important causal effects on network dynamics (28).
RESULTS
Psychophysical Results
In this study, the prevalence of deqi sensation was
expressed as the percentage of individuals in the
group that reported the given sensations (Fig. 2A). No
difference was found in regard to the prevalence of
the listed sensations elicited by acupuncture stimula-
tion (P > 0.05) when grouped across all acupoints.
However, there were some differences with respect to
the type of sensations. In the case of PC6, soreness
was 53%, seemed more frequent than the case of PC7
and GB37 which were 28% and 33%. Numbness was
in the same case with soreness (65% versus 43% and
35%). Following acupuncture at PC7, fullness was
67%, occurred more commonly than PC6 and GB37
which were 48% and 42%. Compared with PC6 and
GB37, dull pain was also in the same case with
Figure 2. Averaged psychophysical response (N ¼ 36). A: The
percentage of subjects who reported having experienced the
given sensation (at least one subject experienced the seven sen-
sations listed). B: The intensity of reported sensations measured
by an average score (with standard error bars) on a scale from 0
denoting no sensation to 10 denoting an unbearable sensation.
*Denoting significant differences under Fisher’s exact test (P <
0.005). [Color figure can be viewed in the online issue, which is
available at wileyonlinelibrary.com.]
34Feng et al.

Page 5

Table 1
Coordinates and t Score of the Peak Voxel Within-Group ROIs Following Acupuncture at PC6, PC7 and GB37 (P < 0.001, Uncorrected)
Regions
PC6
PC7
GB37
Talairach
t
V
Talairach
t
V
Talairach
t
V
Hem
x
y
z
value
voxels
Hem
x
y
z
value
voxels
Hem
x
y
z
Value
Voxels
FGM
R
8
-61
-19
-3.76
21
R
6
-56
-22
-2.73
3
R
6
-48
-19
3.38
8
Culmen
R
12
-53
-12
-4.82
413
R
10
-34
-15
-5.20
443
L
-44
-44
-20
4.49
143
Uvula
R
32
-77
-26
4.03
57
L
-6
-71
-27
-3.58
23
R
32
-77
-26
4.03
57
Nodule
R
10
-63
-24
-2.99
22
R
2
-56
-27
-4.02
82
R
2
-56
-27
-4.02
12
AMY
R
22
-4
-12
-4.28
87
L
-18
-8
-11
5.49
51
L
-18
-8
-11
5.49
51
PH
L
-26
-53
-6
-2.82
4
L
-12
-37
-2
-4.88
86
L
-22
-33
-2
6.39
33
HIPP
L
-28
-41
4
-2.84
1
L
-28
-35
-3
-2.74
2
L
-28
-14
-11
3.22
15
Insula
R
40
-6
2
-6.68
146
R
42
-38
18
-3.38
34
R
32
-27
11
3.77
20
RN
R
8
-22
-4
-2.95
4
R
4
-22
-4
-4.96
31
L
-6
-22
-4
3.33
3
SN
R
12
-22
-7
-3.21
6
R
8
-12
-8
-3.37
14
R
16
-20
-4
5.00
25
PTM
R
28
-16
-6
-3.59
19
L
-24
-11
13
-2.74
1
L
-24
-11
13
6.31
249
VPLN
L
-18
-17
3
-4.31
45
L
-22
-21
8
-2.86
4
L
-18
-19
1
5.08
39
VPMN
L
-16
-23
5
5.37
13
L
-16
-21
8
-2.85
3
L
-16
-23
5
5.37
13
CD
R
22
-34
16
-5.25
17
R
24
-36
13
-3.15
13
L
-32
-33
2
3.38
5
HYPO
L
-8
-6
-3
-2.84
3
L
-8
-6
-3
-2.84
3
L
-8
-6
-3
-2.84
3
Cuneus
L
-24
-55
-7
-3.43
19
R
16
-88
36
-4.75
40
L
-16
-97
9
3.83
30
FG
L
-36
-47
-14
-3.77
55
R
-42
-87
-26
-5.37
41
L
-44
-40
-18
5.61
59
MOG
L
-18
-87
17
3.77
16
L
38
-17
6
-3.28
8
L
-18
-87
17
3.77
16
SII
R
65
-20
36
-4.54
144
R
67
-19
18
-3.33
5
R
51
-15
54
5.04
53
SMA
R
61
-17
41
-5.30
20
R
20
-25
53
-3.85
3
L
-51
-4
35
7.91
36
DLPFC
R
57
18
3
-2.68
2
R
57
18
3
-2.68
2
R
59
22
19
4.73
67
IPL
L
-40
-29
38
-3.42
25
R
65
-42
22
-3.74
14
R
32
-50
54
4.18
5
ACC
L
-20
-16
38
-3.00
3
R
4
11
22
-3.05
4
R
4
11
22
-3.05
4
PCC
L
-46
-42
-15
-2.81
5
R
6
-28
25
-2.95
4
L
-6
-34
22
8.05
17
MTG
L
-65
-29
-4
-4.95
23
L
-65
-29
-4
-4.95
23
R
40
-55
25
5.54
116
Hem ¼ hemisphere; FGM ¼ fastigium; AMY ¼amygdala; PH ¼ parahippocampal gyrus; HIPP ¼ hippocampus; RN ¼ red nucleus; SN ¼ substania nigra; PTM ¼ putamen; VPLN ¼ ventral pos-
terior lateral nucleus; VPMN ¼ ventral posterior medial nucleus; CD ¼ caudate; HYPO ¼ hypothalamus; FG ¼ fusiform gyrus; MOG ¼ middle occipital gyrus; SII ¼ secondary somatosensory
cortex; SMA ¼ supplementary motor area; DLPFC ¼ dorsolateral prefrontal cortex; IPL ¼ inferior parietal lobule; ACC ¼ anterior cingulate cortex; PCC ¼ posterior cingulate cortex; MTG ¼ mid-
dle temporal gyrus.
Acupoint Specificity by Granger Causality Analysis 35

Page 6

fullness (67% versus 25% and 33%) following acu-
puncture at PC7. In contrast, tingling was saliently
more frequent for GB37 than PC6 and PC7 (58% ver-
sus 25% and 33%). The average stimulus intensities
(mean 6 SE) were approximately similar during acu-
puncture on PC6 (1.85 6 1.80), PC7 (1.93 6 2.03)
and GB37 (1.83 6 1.79) (Fig. 2B). However, soreness
was more intense for PC6 than PC7 and GB37 (P <
0.005) while dull pain was stronger for PC7 and GB37
compared with PC6 (P < 0.005). In addition, numb-
ness was more intense for PC6 and PC7 than GB37 (P
< 0.005) while tingling exhibited more intense for
GB37 compared with PC6 and PC7 (P < 0.005).
mGCA Result of Resting Brain Networks Modu-
lated by Acupuncture at Three Acupoints
For bilaterally activated regions, we only selected the
hemisphere anatomical area with a more significant t
value as the representative ROI (shown in Table 1).
The mGCA was performed on the averaged time series
of voxels within each ROI, separately for each acu-
point during the poststimulus phase. Effective con-
nectivity graphs were constructed using the thickness
ofconnectinglines and
strength and direction of the causal influences (indi-
cated in Fig. 3). Only links that showed significant
effective connectivity were presented in the network (P
< 0.05). Graphs were visualized using Pajek software
(www.vlado.fmf.uni-lj.si/pub/networks/pajek).
In-Out degrees of three effective connectivity networks
were sorted in a descending order as shown in Figure
4. Only the brain regions with a nonzero value of In-
Out degree were listed.
Following acupuncture at PC6, the mGCA result
showed that the substantia nigra and red nucleus
emerged ascentralhubs.
received causal inflows from almost all other nodes in
the brain network, including the amygdala, caudate,
fastigium,hippocampus,
(PH), red nucleus, secondary somatosensory cortex
arrowstoindicate the
The
The substantianigra
parahippocampalgyrus
(SII) and uvula. In addition, the red nucleus received
causal inflows from more than half regions in the
brain network, primarily from the caudate, fastigium,
substantia nigra, and SII. Of interest, we found that
the red nucleus and substantia nigra were not only
central hubs but also have significant causal influ-
ence with each other. The path weights of mGCA
result for PC6 were tabulated in Table 2 with signifi-
cant connections shown in red color.
The brain regions with extensive causal interactions
following acupuncture at PC6 were mainly located at
thelimbic/paralimbic-cerebellum
areas as mentioned above. Of interest, several of these
brain regions (the fastigium, hippocampus, PH and
uvula) also have remarkably causal interactions fol-
lowing acupuncture at PC7. Furthermore, the red nu-
cleus received two significant causal inflows from the
culmen and uvula. In addition, the PH was a signifi-
cant target, receiving causal inflows from the hippo-
campus, uvula, culmen and fastigium. In contrast,
the brain regions with extensive causal interactions
following acupuncture at GB37 were primarily located
at the vision-related cortex (the fusiform gyrus), with
limited extent of the limbic system. The fusiform
gyrus received causal inflows primarily from the SII,
thalamus (ventral posterior lateral nucleus [VPLN],
ventral posterior medial nucleus [VPMN]). Moreover,
we identified that several brain areas (the insula and
nodule) only have causal interactions following acu-
puncture at PC6 compared with PC7 and GB37. The
path weights of mGCA result for PC7 and GB37 were,
respectively, tabulated in Tables 3 and 4 with signifi-
cant connections shown in red color.
and subcortical
DISCUSSION
In this study, a newly mGCA combined with a novel
NRER design were used to investigate the specific
effective connectivity during the poststimulus resting
period following acupuncture at PC6, PC7, and GB37.
Our resultsdemonstratedthat acupuncture at
Figure 3. Multivariate Granger causality relationships with significant connections (P < 0.05) following acupuncture at PC6, PC7,
and GB37. Relative strength of path weights (in arbitrary units) was indicated by the width of the arrows. Abbreviations: Hem, hemi-
sphere; ACC, anterior cingulate cortex; AMY, amygdala; CD, caudate; DLPFC, dorsolateral prefrontal cortex; FGM, fastigium; FG,
fusiform gyrus; HIPP, hippocampus; HYPO, hypothalamus; IPL, inferior parietal lobule; MOG, middle occipital gyrus; MTG, middle
temporal gyrus; PCC, posterior cingulate cortex; PH, parahippocampal gyrus; PTM, putamen; RN, red nucleus; SII, secondary soma-
tosensory cortex; SMA, supplementary motor area; SN, substania nigra; VPLN, ventral posterior lateral nucleus; VPMN, ventral poste-
rior medial nucleus. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
36Feng et al.

Page 7

different acupoints may exert heterogeneous modula-
tory effects on the causal interactions of brain areas
during the poststimulus resting state. As a peripheral
input to transducing signals into the brain, acupunc-
ture may induce the reorganization of the effective
connectivityacrossdifferent
These different effective connectivity patterns may be
related to the special effects of acupuncture in clinical
neural subsystems.
settings. Our findings may be helpful to understand
the basic neurophysiological mechanisms underlying
the functional specificity of acupuncture.
Several studies have begun to gaze on the modula-
tory effect of acupuncture on the resting brain net-
works, primarily using the functional connectivity
analysis (8–12). Popular functional connectivity analy-
sis primarily focused on the correlation patterns
Figure 4. Ranking of the In-
Out degrees of the effective
connectivity network following
acupuncture at each acupoint
in a descending order. The In-
Outdegreeof
definedas the
betweenitsIn-degree
number ofcausal
and Out-degree (the number
of causal out-flows). Only the
brain regions with a nonzero
value of In-Out degree are
listed here. Abbreviations: L,
left hemisphere; R, right hemi-
sphere; MOG, middle occipital
gyrus; DLPFC,
prefrontal cortex; SII, second-
arysomatosensory
VPMN,ventral
medial nucleus; PCC, poste-
rior cingulate cortex; ACC, an-
terior cingulate cortex; VPLN,
ventral posterior lateral nu-
cleus;IPL,inferior
lobule; PH, parahippocampal
gyrus;SMA,
motor area; MTG, middle tem-
poral gyrus.
anode
difference
is
(the
in-flows)
dorsolateral
cortex;
posterior
parietal
supplementary
Acupoint Specificity by Granger Causality Analysis37

Page 8

Table 2
Path Weights (Arbitrary Units) From Multivariate Granger Causality Analyses of PC6
ACC AMY
CD
Culmen Cuneus DLPFC FGM
FG
HIPP HYPO
IPL
Insula MOG MTG Nodule PCC
PH
PTM
RN
SII
SMA
SN
VPLN VPMN Uvula
ACC
0.30
0.03
0.37
0.06
0.04
0.50
0.19
0.21
0.03
0.00
0.01
0.00
0.00
0.02
0.02
0.00
0.06
0.01
0.50 0.08
0.07
1.34
0.01
0.19
0.04
AMY
0.04
0.24
0.59
0.07
0.05
0.80
0.24
0.32
0.13
0.00
0.03
0.00
0.21
0.03
0.27
0.00
0.15
0.04
0.90 0.28
0.01
1.92
0.02
0.14
0.16
CD
0.04
0.05
2.79
0.07
0.23
0.67
0.89
0.27
0.12
0.00
0.03
0.04
0.56
0.00
0.07
0.24
0.19
0.09
2.00 0.87
0.03
4.61
0.15
1.19
0.28
Culmen
0.02
0.02
0.26
0.78
0.01
0.09
0.12
0.10
0.09
0.00
0.00
0.01
0.07
0.01
0.00
0.01
0.01
0.01
0.33 0.24
0.05
0.19
0.18
0.38
0.00
Cuneus
0.04
0.04
2.06
0.02
0.33
0.63
0.64
0.31
0.24
0.01
0.03
0.02
0.41
0.00
0.03
0.16
0.26
0.03
1.66 0.61
0.02
4.07
0.12
0.99
0.24
DLPFC
0.06
0.08
0.80
0.03
0.08
2.37
0.34
0.16
0.03
0.01
0.02
0.00
0.20
0.03
0.06
0.06
0.13
0.06
1.56 0.56
0.02
2.72
0.31
0.89
0.09
FGM
0.05
0.05
2.40
0.09
0.19
0.78
1.03
0.23
0.08
0.00
0.03
0.04
0.43
0.00
0.11
0.17
0.15
0.10
2.20 0.90
0.02
4.75
0.21
1.21
0.27
FG
0.10
0.12
1.20
0.12
0.16
0.59
0.39
0.65
0.21
0.00
0.03
0.00
0.16
0.03
0.05
0.02
0.20
0.02
0.74 0.29
0.01
2.48
0.01
0.46
0.10
HIPP
0.02
0.06
0.68
0.13
0.15
0.14
0.17
0.26
0.51
0.00
0.02
0.01
0.22
0.00
0.07
0.13
0.24
0.01
0.61 0.12
0.00
1.87
0.00
0.07
0.20
HYPO
0.00
0.00
0.04
0.00
0.03
0.19
0.03
0.02
0.00
0.09
0.00
0.00
0.00
0.01
0.04
0.05
0.00
0.00
0.01 0.04
0.00
0.17
0.12
0.36
0.01
IPL
0.07
0.13
2.18
0.00
0.20
1.13
0.78
0.35
0.18
0.00
0.05
0.02
0.54
0.02
0.19
0.10
0.26
0.11
2.28 0.84
0.00
4.94
0.15
0.96
0.41
Insula
0.01
0.00
1.16
0.05
0.07
0.07
0.41
0.02
0.04
0.00
0.01
0.09
0.19
0.03
0.03
0.47
0.03
0.03
0.60 0.22
0.09
1.30
0.00
0.18
0.22
MOG
0.00
0.06
1.94
0.07
0.16
0.59
0.54
0.13
0.14
0.00
0.03
0.02
0.81
0.00
0.12
0.26
0.20
0.10
1.89 0.71
0.04
3.50
0.16
0.69
0.37
MTG
0.02
0.03
0.02
0.04
0.00
0.32
0.00
0.07
0.00
0.00
0.00
0.01
0.00
0.26
0.00
0.07
0.01
0.00
0.01 0.02
0.02
0.06
0.00
0.03
0.00
Nodule
0.01
0.15
0.52
0.00
0.02
0.38
0.29
0.09
0.08
0.01
0.03
0.01
0.25
0.00
0.46
0.07
0.11
0.07
1.12 0.31
0.00
1.82
0.05
0.12
0.26
PCC
0.00
0.00
0.70
0.01
0.05
0.14
0.19
0.02
0.07
0.01
0.01
0.04
0.22
0.02
0.03
0.95
0.04
0.02
0.40 0.08
0.07
0.81
0.00
0.07
0.16
PH
0.05
0.10
1.58
0.02
0.24
0.88
0.47
0.37
0.34
0.00
0.04
0.01
0.48
0.01
0.12
0.10
0.35
0.05
1.78 0.48
0.00
4.12
0.09
0.60
0.43
PTM
0.01
0.05
1.41
0.05
0.05
0.81
0.59
0.06
0.04
0.00
0.03
0.01
0.44
0.00
0.16
0.08
0.10
0.19
1.66 0.65
0.01
3.41
0.26
0.82
0.31
RN
0.05
0.07
1.98
0.09
0.18
1.29
0.80
0.17
0.11
0.00
0.04
0.02
0.54
0.00
0.16
0.13
0.22
0.10
2.87 0.86
0.00
4.88
0.27
1.03
0.43
SII
0.02
0.06
2.27
0.17
0.18
1.22
0.87
0.17
0.05
0.00
0.04
0.02
0.53
0.00
0.11
0.07
0.15
0.11
2.26 1.08
0.03
4.37
0.29
1.33
0.22
SMA
0.09
0.01
0.38
0.15
0.03
0.22
0.08
0.01
0.00
0.00
0.00
0.03
0.14
0.02
0.00
0.25
0.00
0.01
0.04 0.12
0.26
0.09
0.01
0.10
0.02
SN
0.07
0.08
2.38
0.03
0.24
1.18
0.91
0.29
0.17
0.00
0.04
0.02
0.52
0.00
0.13
0.14
0.27
0.11
2.56 0.87
0.00
5.66
0.24
1.22
0.41
VPLN
0.01
0.01
0.71
0.23
0.06
1.23
0.37
0.01
0.00
0.02
0.01
0.00
0.22
0.00
0.03
0.00
0.05
0.08
1.33 0.54
0.00
2.19
0.59
1.22
0.09
VPMN
0.03
0.02
1.84
0.16
0.17
1.16
0.69
0.16
0.02
0.02
0.03
0.01
0.31
0.00
0.03
0.03
0.11
0.08
1.62 0.79
0.01
3.64
0.39
1.84
0.12
Uvula
0.01
0.04
0.94
0.00
0.09
0.24
0.34
0.08
0.12
0.00
0.02
0.02
0.36
0.00
0.12
0.18
0.18
0.07
1.45 0.28
0.01
2.64
0.06
0.25
0.83
Significant paths (P < 0.05) are shown in bold.
Hem ¼ hemisphere; ACC ¼ anterior cingulate cortex; AMY ¼ amygdala; CD ¼ caudate; DLPFC ¼ dorsolateral prefrontal cortex; FGM ¼ fastigium; FG ¼ fusiform gyrus; HIPP ¼ hippocampus;
HYPO ¼ hypothalamus; IPL ¼ inferior parietal lobule; MOG ¼ middle occipital gyrus; MTG ¼ middle temporal gyrus; PCC ¼ posterior cingulate cortex; PH ¼ parahippocampal gyrus; PTM ¼ puta-
men; RN ¼ red nucleus; SII ¼ secondary somatosensory cortex; SMA ¼ supplementary motor area; SN ¼ substania nigra; VPLN ¼ ventral posterior lateral nucleus; VPMN ¼ ventral posterior medial
nucleus.

Page 9

Table 3
Path Weights (Arbitrary Units) From Multivariate Granger Causality Analyses of PC7
ACC AMY
CD
Culmen Cuneus DLPFC FGM
FG
HIPP HYPO
IPL
Insula MOG MTG Nodule PCC
PH
PTM
RN
SII
SMA
SN
VPLN VPMN Uvula
ACC
0.12
0.03
0.04
0.00
0.00
0.00
0.01
0.00
0.00
0.00
0.00
0.04
0.00
0.18
0.02
0.00
0.03
0.01
0.10 0.01
0.00
0.01
0.02
0.03
0.04
AMY
0.00
0.85
0.02
0.01
0.00
0.14
0.05
0.00
0.00
0.01
0.07
0.35
0.00
0.07
0.00
0.01
0.03
0.02
0.26 0.00
0.14
0.09
0.26
0.17
0.01
CD
0.01
0.02
0.88
0.00
0.02
0.06
0.24
0.07
0.04
0.01
0.01
0.00
0.25
0.05
0.10
0.17
0.06
0.08
0.01 0.04
0.04
0.12
0.28
0.32
0.00
Culmen
0.00
0.16
0.05
0.08
0.01
0.09
0.36
0.06
0.20
0.00
0.00
0.00
0.00
0.20
0.02
0.11
1.88
0.00
1.49 0.00
0.18
0.01
0.02
0.00
0.41
Cuneus
0.00
0.00
0.32
0.01
0.02
0.01
0.17
0.05
0.06
0.02
0.02
0.01
0.06
0.00
0.05
0.35
0.39
0.08
0.24 0.01
0.07
0.06
0.32
0.21
0.04
DLPFC
0.00
0.10
0.04
0.01
0.00
1.18
0.03
0.13
0.06
0.01
0.11
0.29
0.00
0.19
0.01
0.00
1.01
0.03
0.13 0.00
0.01
0.12
0.15
0.12
0.24
FGM
0.00
0.06
0.36
0.04
0.01
0.05
0.61
0.12
0.23
0.00
0.03
0.00
0.06
0.06
0.08
0.27
1.51
0.01
0.48 0.00
0.21
0.06
0.06
0.04
0.24
FG
0.00
0.01
0.14
0.01
0.01
0.33
0.16
0.46
0.05
0.01
0.11
0.18
0.23
0.18
0.03
0.05
0.36
0.03
0.14 0.03
0.00
0.35
0.09
0.13
0.03
HIPP
0.00
0.00
0.09
0.05
0.01
0.22
0.39
0.07
0.35
0.00
0.04
0.02
0.01
0.05
0.07
0.15
2.94
0.01
0.80 0.01
0.11
0.05
0.01
0.01
0.42
HYPO
0.00
0.06
0.08
0.00
0.01
0.11
0.00
0.02
0.01
0.16
0.00
0.00
0.03
0.04
0.01
0.00
0.23
0.07
0.07 0.02
0.00
0.04
0.38
0.36
0.00
IPL
0.00
0.24
0.05
0.00
0.00
0.50
0.07
0.20
0.06
0.00
0.25
0.43
0.16
0.04
0.03
0.02
0.44
0.00
0.00 0.01
0.02
0.37
0.01
0.00
0.06
Insula
0.01
0.30
0.00
0.00
0.00
0.34
0.00
0.08
0.01
0.00
0.11
1.00
0.07
0.05
0.00
0.05
0.25
0.03
0.00 0.00
0.12
0.19
0.20
0.11
0.01
MOG
0.00
0.01
0.35
0.00
0.01
0.00
0.06
0.17
0.01
0.01
0.07
0.11
0.64
0.01
0.01
0.04
0.01
0.02
0.16 0.02
0.03
0.35
0.07
0.19
0.02
MTG
0.03
0.07
0.05
0.02
0.00
0.27
0.04
0.10
0.02
0.01
0.01
0.05
0.01
0.85
0.00
0.00
0.24
0.00
0.97 0.00
0.02
0.04
0.02
0.01
0.23
Nodule
0.01
0.00
0.47
0.01
0.01
0.07
0.27
0.06
0.13
0.00
0.05
0.00
0.04
0.00
0.19
0.18
0.70
0.03
0.05 0.01
0.11
0.10
0.09
0.09
0.18
PCC
0.00
0.01
0.25
0.01
0.02
0.00
0.27
0.04
0.09
0.00
0.01
0.08
0.04
0.00
0.05
0.62
0.72
0.03
0.17 0.00
0.09
0.08
0.14
0.07
0.12
PH
0.00
0.01
0.02
0.04
0.01
0.35
0.27
0.05
0.30
0.01
0.03
0.07
0.00
0.06
0.04
0.13
3.38
0.03
0.92 0.01
0.06
0.05
0.08
0.10
0.48
PTM
0.00
0.08
0.28
0.00
0.02
0.13
0.03
0.06
0.01
0.05
0.00
0.13
0.06
0.01
0.02
0.08
0.46
0.23
0.00 0.05
0.04
0.02
0.79
0.67
0.03
RN
0.01
0.09
0.00
0.04
0.01
0.07
0.13
0.03
0.12
0.00
0.00
0.00
0.04
0.36
0.00
0.05
1.35
0.00
2.26 0.01
0.10
0.00
0.03
0.00
0.36
SII
0.01
0.00
0.27
0.00
0.01
0.01
0.02
0.11
0.02
0.02
0.01
0.00
0.11
0.00
0.02
0.00
0.37
0.11
0.16 0.11
0.00
0.03
0.29
0.33
0.08
SMA
0.00
0.25
0.08
0.03
0.01
0.03
0.27
0.00
0.08
0.00
0.01
0.27
0.03
0.03
0.04
0.12
0.41
0.02
0.51 0.00
0.45
0.09
0.12
0.04
0.19
SN
0.00
0.10
0.15
0.00
0.00
0.18
0.05
0.22
0.03
0.01
0.12
0.26
0.30
0.04
0.02
0.07
0.23
0.01
0.00 0.00
0.06
0.68
0.03
0.06
0.04
VPLN
0.00
0.22
0.25
0.00
0.03
0.20
0.04
0.04
0.00
0.06
0.00
0.21
0.04
0.02
0.02
0.09
0.28
0.20
0.06 0.03
0.06
0.02
0.97
0.79
0.01
VPMN
0.00
0.17
0.32
0.00
0.02
0.16
0.03
0.07
0.00
0.07
0.00
0.14
0.14
0.01
0.02
0.05
0.41
0.19
0.00 0.04
0.02
0.05
0.87
0.86
0.03
Uvula
0.01
0.01
0.00
0.04
0.00
0.36
0.19
0.02
0.19
0.00
0.02
0.01
0.01
0.25
0.04
0.09
2.08
0.01
1.05 0.01
0.11
0.04
0.01
0.03
0.78
Significant paths (P < 0.05) are shown in bold.
Hem ¼ hemisphere; ACC ¼ anterior cingulate cortex; AMY ¼ amygdala; CD ¼ caudate; DLPFC ¼ dorsolateral prefrontal cortex; FGM ¼ fastigium; FG ¼ fusiform gyrus; HIPP ¼ hippocampus;
HYPO ¼ hypothalamus; IPL ¼ inferior parietal lobule; MOG ¼ middle occipital gyrus; MTG ¼ middle temporal gyrus; PCC ¼ posterior cingulate cortex; PH ¼ parahippocampal gyrus; PTM ¼ puta-
men; RN ¼ red nucleus; SII ¼ secondary somatosensory cortex; SMA ¼ supplementary motor area; SN ¼ substania nigra; VPLN ¼ ventral posterior lateral nucleus; VPMN ¼ ventral posterior medial
nucleus.

Page 10

Table 4
Path Weights (Arbitrary Units) From Multivariate Granger Causality Analyses of GB37
ACC AMY
CD
Culmen Cuneus DLPFC FGM
FG
HIPP HYPO
IPL
Insula MOG MTG Nodule PCC
PH
PTM
RN
SII
SMA
SN
VPLN VPMN Uvula
ACC
0.05
0.15
0.00
0.04
0.05
0.12
0.02
0.06
0.10
0.00
0.00
0.02
0.02
0.02
0.18
0.14
0.01
0.00
0.13 0.00
0.01
0.09
0.00
0.02
0.05
AMY
0.01
0.73
0.00
0.08
0.15
0.10
0.00
0.07
0.03
0.05
0.05
0.01
0.04
0.05
0.02
0.01
0.37
0.01
0.03 0.11
0.09
0.06
0.04
0.00
0.06
CD
0.00
0.00
0.65
0.04
0.26
0.23
0.01
0.29
0.04
0.00
0.29
0.00
0.01
0.13
0.15
0.01
0.00
0.04
0.34 0.02
0.52
0.15
0.38
0.00
0.01
Culmen
0.01
0.26
0.13
0.21
0.01
0.04
0.01
0.05
0.01
0.01
0.12
0.00
0.01
0.00
0.00
0.00
0.03
0.03
0.10 0.01
0.07
0.00
0.06
0.00
0.00
Cuneus
0.00
0.08
0.12
0.00
1.34
0.07
0.02
0.32
0.00
0.01
0.16
0.02
0.05
0.21
0.24
0.22
0.12
0.00
0.00 0.06
0.46
0.33
0.28
0.00
0.07
DLPFC
0.01
0.08
0.18
0.01
0.10
0.83
0.00
0.50
0.06
0.04
0.02
0.00
0.00
0.06
0.00
0.05
0.33
0.00
0.30 0.11
0.50
0.49
0.33
0.01
0.05
FGM
0.01
0.03
0.04
0.01
0.27
0.01
0.12
0.03
0.06
0.00
0.03
0.01
0.04
0.04
0.12
0.06
0.00
0.00
0.03 0.00
0.02
0.07
0.16
0.04
0.04
FG
0.00
0.06
0.21
0.01
0.48
0.48
0.00
0.88
0.01
0.02
0.11
0.00
0.03
0.18
0.05
0.00
0.28
0.00
0.12 0.15
0.86
0.62
0.54
0.00
0.18
HIPP
0.01
0.04
0.06
0.00
0.00
0.12
0.01
0.01
0.46
0.05
0.08
0.00
0.01
0.00
0.12
0.06
0.29
0.02
0.01 0.03
0.01
0.19
0.04
0.02
0.08
HYPO
0.00
0.21
0.00
0.01
0.09
0.20
0.00
0.11
0.11
0.19
0.03
0.00
0.00
0.05
0.00
0.00
0.36
0.00
0.02 0.13
0.14
0.30
0.13
0.00
0.08
IPL
0.00
0.07
0.36
0.05
0.41
0.04
0.01
0.20
0.06
0.01
0.52
0.01
0.00
0.13
0.21
0.10
0.03
0.04
0.15 0.00
0.39
0.08
0.19
0.00
0.02
Insula
0.02
0.09
0.04
0.00
0.79
0.00
0.01
0.07
0.04
0.02
0.11
0.03
0.04
0.17
0.37
0.30
0.01
0.00
0.09 0.03
0.18
0.06
0.14
0.01
0.00
MOG
0.00
0.14
0.03
0.01
0.32
0.01
0.02
0.15
0.02
0.00
0.00
0.01
0.20
0.11
0.10
0.07
0.11
0.00
0.03 0.05
0.16
0.10
0.23
0.01
0.09
MTG
0.00
0.10
0.23
0.00
0.81
0.14
0.01
0.44
0.00
0.02
0.19
0.02
0.06
0.36
0.29
0.13
0.19
0.00
0.00 0.13
0.80
0.42
0.36
0.00
0.03
Nodule
0.02
0.02
0.17
0.00
0.57
0.00
0.02
0.07
0.09
0.00
0.19
0.03
0.04
0.18
0.59
0.18
0.00
0.01
0.00 0.02
0.31
0.03
0.16
0.01
0.00
PCC
0.01
0.01
0.02
0.00
0.65
0.10
0.01
0.01
0.06
0.00
0.12
0.03
0.03
0.10
0.23
0.44
0.01
0.00
0.18 0.00
0.06
0.00
0.03
0.01
0.00
PH
0.00
0.33
0.00
0.01
0.19
0.34
0.00
0.30
0.15
0.08
0.02
0.00
0.03
0.08
0.00
0.01
0.84
0.02
0.00 0.16
0.31
0.45
0.14
0.00
0.21
PTM
0.00
0.13
0.26
0.06
0.03
0.00
0.00
0.00
0.09
0.01
0.27
0.00
0.01
0.01
0.08
0.02
0.24
0.07
0.22 0.01
0.05
0.01
0.06
0.00
0.14
RN
0.01
0.02
0.22
0.02
0.00
0.26
0.00
0.10
0.01
0.00
0.08
0.00
0.01
0.00
0.00
0.08
0.00
0.02
0.99 0.00
0.13
0.10
0.09
0.01
0.00
SII
0.00
0.33
0.04
0.01
0.34
0.39
0.00
0.52
0.05
0.10
0.00
0.01
0.04
0.18
0.04
0.00
0.52
0.00
0.00 0.25
0.70
0.53
0.32
0.00
0.15
SMA
0.00
0.06
0.30
0.01
0.56
0.38
0.00
0.68
0.00
0.02
0.18
0.01
0.03
0.25
0.16
0.02
0.24
0.00
0.11 0.16
1.11
0.47
0.46
0.00
0.05
SN
0.01
0.05
0.11
0.00
0.47
0.44
0.01
0.58
0.09
0.06
0.04
0.00
0.02
0.17
0.02
0.00
0.41
0.00
0.10 0.15
0.56
0.96
0.43
0.00
0.19
VPLN
0.00
0.03
0.29
0.01
0.44
0.33
0.02
0.54
0.02
0.03
0.12
0.01
0.05
0.15
0.11
0.01
0.13
0.01
0.10 0.09
0.59
0.45
0.85
0.01
0.07
VPMN
0.01
0.02
0.00
0.00
0.03
0.07
0.05
0.02
0.12
0.00
0.01
0.00
0.02
0.00
0.05
0.05
0.01
0.00
0.08 0.00
0.05
0.02
0.07
0.08
0.01
Uvula
0.00
0.08
0.01
0.00
0.19
0.09
0.01
0.31
0.07
0.03
0.02
0.00
0.03
0.02
0.00
0.00
0.35
0.02
0.00 0.07
0.10
0.34
0.13
0.00
0.50
Significant paths (P < 0.05) are shown in bold.
Hem ¼ hemisphere; ACC ¼ anterior cingulate cortex; AMY ¼ amygdala; CD ¼ caudate; DLPFC ¼ dorsolateral prefrontal cortex; FGM ¼ fastigium; FG ¼ fusiform gyrus; HIPP ¼ hippocampus;
HYPO ¼ hypothalamus; IPL ¼ inferior parietal lobule; MOG ¼ middle occipital gyrus; MTG ¼ middle temporal gyrus; PCC ¼ posterior cingulate cortex; PH ¼ parahippocampal gyrus; PTM ¼ puta-
men; RN ¼ red nucleus; SII ¼ secondary somatosensory cortex; SMA ¼ supplementary motor area; SN ¼ substania nigra; VPLN ¼ ventral posterior lateral nucleus; VPMN ¼ ventral posterior medial
nucleus.

Page 11

between a seed region and any other brain structures
throughout the whole brain, depicting brain networks
related to certain functions. However, this method
was limited to assess brain regions functionally con-
nected to the initially selected seed and was unable to
directly characterize interactions between multiple
brain regions. In this study, we addressed this prob-
lem by introducing the mGCA to detect causal influ-
ences between multiple brain areas. By visualizing the
effective connectivity, we can obtain both the direction
and strength of the information flow between multiple
brain regions in the resting state network.
From the mGCA results, we identified that brain
regions have extensive causal interactions following
acupuncture at PC6, mainly locating at the amygdala,
caudate, fastigium, hippocampus, insula, PH, and
flocculonodularlobe of
uvula). Previous studies from Hui and colleagues also
demonstrated that these brain regions can be modu-
lated by acupuncture (3,20). These limbic-cerebellum
and subcortical areas are more engaged in affective
motivation and autonomic drive of bodily responses
(29). Therefore, the causal influences between these
brain regions may be related to the sedative or tran-
quilizing effect of PC6 after a variety of stressors or
depression (17). It was also worth noting that several
limbic/paralimbic-cerebellum
hippocampus, PH, uvula) also have causal interac-
tions underlying acupuncture at PC7 in the poststi-
mulus rest brain networks. Even with the relative
similarity, the PH was presented to be the most im-
portant brain areas, receiving causal inflows mainly
from the hippocampus. These stress-sensitive brain
regions have previously been reported for their central
role in depressive illness (30). The causal influences
between the PH and other brain regions underlying
PC7 may provide an explanation for the mediate effect
of PC7 to treat depressive disorders (31).
In addition, we identified the insula and nodule
have causal interactions with other brain regions only
following acupuncture at PC6. Previous studies in
humans have reported that resection of the insula
resulted in visceral motor dysfunction, nausea, vomit-
ing and gastrointestinal disorders (32). Therefore, the
causal influences anchored by insula may help better
understand the specific treatment effect of PC6 on the
nausea and vomiting. The flocculonodular lobe of the
nodule, as part of vestibule cerebellum, receives ves-
tibular projections from primary and secondary vesib-
ular afferents, as well as vestibular climbing fibers
(33,34). Our study demonstrated that the causal pat-
terns underlying acupoint PC6 in the treatment of
nausea can selectively modulate specific neural sub-
strates—the flocculonodular lobe of nodule related to
vestibularfunction,subsequently
symptoms with motion-related sickness and nausea
(4).
Along the different meridian with respect to PC6,
the GB37 was an acupoint belonging to the gallblad-
der channel. As its name ‘‘brightness’’ suggests, GB37
was described as a very effective acupoint influencing
multiple vision-related disorders, such as the cata-
racts, night blindness and optic atrophy (19). The
cerebellum(noduleand
areas(the fastigium,
alleviatingthe
mGCA result demonstrated that acupuncture at this
acupoint saliently elicited different effective connectiv-
ity compared with PC6 and PC7. The brain regions
with extensive causal interactions were primarily
located at the vision-related cortex (the fusiform
gyrus), with limited extent of the limbic system. The
fusiform gyrus became an important site of target,
receiving major inputs from both the SII and thala-
mus (VPLN, VPMN). The drive from the SII to the fusi-
form gyrus intuitively depicted that there existed a
control signal starting from the somatosensory cortex
to the vision-related cortex, thereby elucidating the
treatment of GB37 in vision-related disorders. Results
indicated that acupuncture at acupoints along differ-
ent meridian may induce distinct reorganization of
theeffectiveconnectivity
subsystems.
In conclusion, our findings demonstrated that there
existed different effective connectivity patterns during
the poststimulus resting state following acupuncture
at acupoint PC6, compared with the same meridian
acupoint PC7 and different meridian acupoint GB37.
We suggested that the distinct modulation patterns of
the resting brain attributed to the specific effects
evoked by the three acupoints. We also identified that
some important brain regions emerged as central
hubs in the poststimulus resting brain. These brain
regions may be related to the effects of acupoint on
modulating special disorder treatment. This prelimi-
nary finding may provide a new clue to understand
the relatively function-oriented specificity of acupunc-
ture effects.
acrossdifferent neural
REFERENCES
1. NIH. NIH consensus conference statement acupuncture. JAMA
1998;280:1518–1524.
2. Fang JL, Jin Z, Wang Y, et al. The salient characteristics of the
central effects of acupuncture needling: limbic-paralimbic-neo-
cortical networkmodulation.
1196–1206.
3. Hui KK, Liu J, Makris N, et al. Acupuncture modulates the limbic
system and subcortical gray structures of the human brain: evi-
dence from fMRI studies in normal subjects. Hum Brain Mapp
2000;9:13–25.
4. Yoo SS, Teh EK, Blinder RA, Jolesz FA. Modulation of cerebellar
activities by acupuncture stimulation: evidence from fMRI study.
Neuroimage 2004;22:932–940.
5. Beijing S. Nanjing colleges of traditional Chinese medicine.
Essentials of Chinese acupuncture. Beijing: Foreign Language
Press; 1980:36.
6. Bai LJ, Qin W, Tian J, et al. Time-varied characteristics of acu-
puncture effects in fMRI studies. Hum Brain Mapp 2009;30:
3445–3460.
7. Bai LJ, Yan H, Li L, et al. Neural specificity of acupuncture stim-
ulation at pericardium 6: evidence from an FMRI study. J Magn
Reson Imaging 2010;31:71–77.
8. Bai LJ, Qin W, Tian J, et al. Acupuncture modulates spontane-
ous activities in the anticorrelated resting brain networks. Brain
Res 2009;1279:37–49.
9. Bai LJ, Tian J, Zhong C, et al. Acupuncture modulates temporal
neural responses in wide brain networks: evidence from fMRI
study. Mol Pain 2010;6:73.
10. Bai LJ, Qin W, Liang JM, Tian J, Liu YJ. Spatiotemporal modula-
tion of central neural pathway underlying acupuncture action: a
systematic review. Curr Med Imaging Rev 2009;5:167–173.
11. Dhond RP, Yeh C, Park K, Kettner N, Napadow V. Acupuncture
modulates resting state connectivity in default and sensorimotor
brain networks. Pain 2008;136:407–418.
Hum BrainMapp2009;30:
Acupoint Specificity by Granger Causality Analysis 41

Page 12

12. Bai LJ, Qin W, Tian J, Dai JP, Yang WH. Detection of dynamic
brain networks modulated by acupuncture using a graph theory
model. Prog Nat Sci 2009;19:827–835.
13. Deshpande G, LaConte S, James GA, Peltier S, Hu X. Multivari-
ate Granger causality analysis of fMRI data. Hum Brain Mapp
2009;30:1361–1373.
14. Deshpande G, LaConte S, Peltier S, Hu X. Directed transfer func-
tion analysis of fMRI data to investigate network dynamics. Conf
Proc IEEE Eng Med Biol Soc 2006;1:671–674.
15. Stilla R, Deshpande G, LaConte S, Hu X, Sathian K. Posterome-
dial parietal cortical activity and inputs predict tactile spatial
acuity. J Neurosci 2007;27:11091–11102.
16. Oldfield RC. The assessment and analysis of handedness: the
Edinburgh inventory. Neuropsychologia 1971;9:97–113.
17. Stux. G, Pomeranz. B. Acupuncture: textbook and atlas. Berlin:
Springer-Verlag; 1987. p 231–244.
18. Dundee JW, Chestnutt WN, Ghaly RG, Lynas AG. Traditional
Chinese acupuncture: a potentially useful antiemetic. BMJ 1986;
293:583–584.
19. Liu GW. Acupoints of three Yang meridians of foot. In: Liu GW,
editor. A complement work of present acupuncture and moxibus-
tion. Tianjin: Hua Xia Publishing House; 1997:327–479.
20. Hui KK, Liu J, Marina O, et al. The integrated response of the
human cerebro-cerebellar and limbic systems to acupuncture
stimulation at ST 36 as evidenced by fMRI. Neuroimage 2005;27:
479–496.
21. Kong J, Gollub R, Huang T, et al. Acupuncture de qi, from quali-
tative history to quantitative measurement. J Altern Complement
Med 2007;13:1059–1070.
22. Ashburner J, Friston KJ. Nonlinear spatial normalization using
basis functions. Hum Brain Mapp 1999;7:254–266.
23. Blinowska KJ, Kus R, Kaminski M. Granger causality and infor-
mation flow in multivariate processes. Phys Rev E Stat Nonlin
Soft Matter Phys 2004;70:50902–50906.
24. Akaike H. New look at statistical-model identification. IEEE Trans
Automat Contr 1974;19:716–723.
25. Kus R, Kaminski M, Blinowska KJ. Determination of EEG activity
propagation: pair-wise versus multichannel estimate. IEEE Trans
Biomed Eng 2004;51:1501–1510.
26. Strang G. Introduction to linear algebra. Cambridge: Wellesley-
Cambridge Press; 1998.
27. Theiler J, Eubank S, Longtin A, Galdrikian B, Farmer JD. Testing
for nonlinearity in time-series - the method of surrogate data.
Physica D 1992;58:77–94.
28. Jiao Q, Lu G, Zhang Z, et al. Granger causal influence predicts
BOLD activity levels in the default mode network. Hum Brain
Mapp 2011;32:154–161.
29. Mann F. Reinventing acupuncture: a new concept of ancient
medicine. Great Britain: Biddles Ltd; 1992. 44 p.
30. Campbell S, Macqueen G. The role of the hippocampus in the
pathophysiology of major depression. J Psychiatry Neurosci
2004;29:417–426.
31. McDonald B, Highley JR, Walker MA, et al. Anomalous asymme-
try of fusiform and parahippocampal gyrus gray matter in schizo-
phrenia: a postmortem study. Am J Psychiatry 2000;157:40–47.
32. Penfield W, Jasper HH. Epilepsy and the functional anatomy of
the human brain. New York: Little, Brown and Company; 1954.
33. Ito M. Neurophysiology of the nodulofloccular system. Rev Neurol
(Paris) 1993;149:692–697.
34. Pettorossi VE, Grassi S, Errico P, Barmack NH. Role of cerebellar
nodulus and uvula on the vestibular quick phase spatial con-
stancy. Acta Otolaryngol Suppl 2001;545:155–159.
42 Feng et al.