Pain 130 (2007) 254–266

www.elsevier.com/locate/pain

Hypothalamus and amygdala response to acupuncture stimuli

in carpal tunnel syndrome
a,b,*
V. Napadow , N. Kettner b, J. Liu a, M. Li a, K.K. Kwong a, M. Vangel a,
N. Makris c, J. Audette d, K.K.S. Hui a
a
Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital, Charlestown, MA, United States
b
Department of Radiology, Logan College of Chiropractic, Chesterfield, MO, United States
c
Martinos Center for Biomedical Imaging, Department of Neurology, Massachusetts General Hospital, Charlestown, MA, United States
d
Spaulding Rehabilitation Hospital, Boston, MA, United States

Received 19 July 2006; received in revised form 21 November 2006; accepted 4 December 2006

Abstract

Brain processing of acupuncture stimuli in chronic neuropathic pain patients may underlie its beneficial effects. We used fMRI to
evaluate verum and sham acupuncture stimulation at acupoint LI-4 in Carpal Tunnel Syndrome (CTS) patients and healthy controls
(HC). CTS patients were retested after 5 weeks of acupuncture therapy. Thus, we investigated both the short-term brain response to
acupuncture stimulation, as well as the influence of longer-term acupuncture therapy effects on this short-term response. CTS
patients responded to verum acupuncture with greater activation in the hypothalamus and deactivation in the amygdala as com-
pared to HC, controlling for the non-specific effects of sham acupuncture. A similar difference was found between CTS patients
at baseline and after acupuncture therapy. For baseline CTS patients responding to verum acupuncture, functional connectivity
was found between the hypothalamus and amygdala – the less deactivation in the amygdala, the greater the activation in the hypo-
thalamus, and vice versa. Furthermore, hypothalamic response correlated positively with the degree of maladaptive cortical plastic-
ity in CTS patients (inter-digit separation distance). This is the first evidence suggesting that chronic pain patients respond to
acupuncture differently than HC, through a coordinated limbic network including the hypothalamus and amygdala.
 2006 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

Keywords: Chronic pain; Limbic; Brain; Complementary and alternative medicine

1. Introduction ies of acupuncture have noted stimulus-associated

response in several limbic structures including the amyg-
Acupuncture is a component of traditional Chinese dala, cingulate cortex (Wu et al., 1999; Hui et al., 2000;
medicine, and has evolved empirically over thousands Napadow et al., 2005), and hypothalamus (Wu et al.,
of years to treat a multitude of ailments (Kaptchuk, 1999; Hui et al., 2000; Hsieh et al., 2001). However,
2002). While the efficacy of acupuncture for many con- the majority of acupuncture neuroimaging studies have
ditions is still under debate, recent evaluation of this been performed on healthy adults. Conversely, an inter-
treatment modality has lent credence to the hypothesis esting PET study which controlled for expectancy,
that the brain and nervous system plays a leading role found verum acupuncture-specific response in the insula
in processing acupuncture stimuli. Neuroimaging stud- (Pariente et al., 2005). However, this study did not
directly contrast these results with healthy controls. His-
*
Corresponding author. Tel.: +1 617 724 3402; fax: +1 617 726
torically, it has been posited that acupuncture plays a
7422. homeostatic role (Zhu, 1954; Kaptchuk, 2000) and thus
E-mail address: [email protected] (V. Napadow). may have a greater effect on patient populations with a

0304-3959/$32.00  2006 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.pain.2006.12.003
 V. Napadow et al. / Pain 130 (2007) 254–266 255

pathological imbalance, compared to healthy individu- while 2 subjects had scans removed for excessive motion arti-
als. Hence, it remains to be seen if past acupuncture fact. Clinical evaluation was completed for both groups by
neuroimaging results in healthy adults will also apply an experienced physician [JA] at the Spaulding Rehabilitation
to patients with chronic pain, for which a direct compar- Hospital, and has been previously described (Audette et al., in
press). Subjects were screened and excluded if they were preg-
ison is necessary. Chronic neuropathic and inflammato-
nant or had a history of diabetes mellitus, rheumatoid arthri-
ry pain alters processing in several limbic brain regions
tis, thyroid dysfunction, wrist fracture or direct trauma to
(e.g., amygdala, insula, ACC) and may initiate and be median nerve, atrophy of the thenar eminence, carpal tunnel
maintained by sensitization in brain processing. surgery, current use of prescriptive opioid pain medications,
Evidence for this effect comes from both animal psychiatric and neurological disorders, head trauma with loss
(Neugebauer et al., 2004) and human (Stern et al., 2006) of consciousness, or other serious cardiovascular, respiratory
studies. Acupuncture may alter brain function through or renal illness. Subjects were also excluded if they had any
neuroplasticity mechanisms by a combination of afferent contraindication to undergoing MRI (e.g., pacemaker), or
somatosensory stimulus and affective evaluation, thereby any contraindication for acupuncture (e.g., anti-coagulation
modulating centrally maintained chronic pain states. therapy). CTS patients were included if they had experienced
Carpal tunnel syndrome (CTS) is the most common pain and/or paresthesias for greater than 3 months in the
median nerve distribution of the affected hand – digits 1–3,
entrapment neuropathy, with a prevalence of 3.72% in
and the radial aspect of digit 4. Further, Nerve Conduction
the United States (Papanicolaou et al., 2001). CTS etiol-
Study (NCS) findings needed to be consistent with mild to
ogy is characterized by compression of the distal median moderate CTS. Mild CTS was defined by delayed distal laten-
nerve due to an elevated interstitial fluid pressure in the cy of median sensory nerve conduction across the wrist
carpal tunnel. Ischemic injury to the median nerve leads (>3.7 ms and/or median – ulnar >0.5 ms) with normal motor
to anoxic capillary damage, which then leads to nerve conduction. Moderate CTS was defined by mild CTS
increased membrane permeability, exudative edema and with delayed distal latency of median motor nerve conduc-
and subsequent fibrosis (Keir and Rempel, 2005; Sud tion across the wrist (>4.2 ms), but with normal motor ampli-
and Freeland, 2005). CTS is associated with a range of tudes. Subjects with severe CTS, defined by prolonged median
symptoms primarily in the first through fourth digit, sensory and motor latencies with either absent sensory nerve
including paresthesias, pain, and weakness. action potentials and/or reduced (50%) median motor ampli-
tudes, were excluded. Patients were also excluded if they dem-
We have previously demonstrated that CTS is associ-
onstrated any evidence on NCS of generalized peripheral
ated with cortical hyperactivation to simple (non-acu-
neuropathy or localized ulnar nerve entrapment.
puncture) somatosensory stimuli and maladaptive
somatotopic plasticity in contralateral primary somato- 2.2. Acupuncture treatment
sensory cortex (Napadow et al., 2006). This pathological
central correlate of the peripheral CTS lesion was found Acupuncture was performed by experienced practitioners
to be altered after successful acupuncture treatment on CTS patients over a 5-week period, after baseline clinical
(Napadow et al., in press). In the current study, we and fMRI evaluation. Treatments were provided 3 times per
explored brain processing of acupuncture stimulation week for three weeks and two times per week for the remaining
in CTS patients compared to healthy controls (HC), two weeks. A semi-individualized approach was used wherein
controlling for cutaneous somatosensory/cognitive every subject was treated for 10 min with 2 Hz electro-acu-
puncture at common acupoints – unilateral TW-5 (triple-
effects with sham acupuncture. Our hypothesis was that
warmer 5, dorsal aspect of forearm) to PC-7 (pericardium 7,
chronic pain patients would demonstrate greater
1st wrist crease). This was followed by manual needling at acu-
response to verum acupuncture stimulation in limbic points chosen by the practitioner that were based on the indi-
brain regions which have been associated with the main- vidual symptoms of the presenting patient. Three acupoints
tenance of a persistent pain state. were chosen out of the following six: HT-3 (heart 3, medial
aspect of elbow), PC-3 (pericardium 3, medial aspect of elbow
2. Methods crease), SI-4 (small intestine 4, ulnar aspect of wrist), LI-5
(large intestine 5, radial aspect of wrist), LI-10 (large intestine
2.1. Subject recruitment and evaluation 10, lateral aspect of forearm), LU-5 (lung 5, lateral aspect of
elbow crease). These acupoints were stimulated with a manual
All participants in the study provided written informed con- even needle technique where a deqi response was obtained.
sent in accordance with the Human Research Committee of
the Massachusetts General Hospital and also took part in a 2.3. FMRI image acquisition and stimulation protocol
study of cortical plasticity and somatotopy. A total of 25 sub-
jects were enrolled in this study; 13 adults affected by CTS, and FMRI test–retest scanning was completed on 10 CTS
12 age and gender-matched HC. Of the 13 CTS patients initial- patients (6 female; mean age: 51.1, range 31–60) before (base-
ly studied, one was removed for excessive motion during fMRI line) and following 5 weeks of acupuncture, with an interval of
scanning, and two were not imaged during acupuncture stimuli 40.6 ± 4.4 days (l ± SD) between scan sessions. FMRI scan-
due to claustrophobia. Of the 12 healthy adults recruited, one ning was also completed on 9 age and gender-matched HC
subject was removed for suspected sub-clinical neuropathy, subjects (6 female; mean age: 46.9, range 32–59) spaced at least
256 V. Napadow et al. / Pain 130 (2007) 254–266

5 weeks apart, with an interval of 44.0 ± 7.7 days between scan (twirled) until the subjects felt a sensation that was not
sessions. Five patients presented with CTS symptoms in both described as sharp pain to minimize the risk of inducing sharp
hands, while five presented with only unilateral symptomato- pain during the scan run (see below). During the fMRI scan,
logy. For patients with bilateral CTS symptoms, testing was the needle was manually stimulated (twirled, ±90) at 1 Hz
done on the more affected hand (self-report). In all cases, the during the stimulation blocks. In all scan runs, subjects were
more affected hand was also the patient’s dominant hand. instructed to keep their eyes closed, and to pay close attention
The chronicity of symptoms (self-reported) ranged from 4 to the sensations felt at the stimulated hand.
months to 10 years, with 10 of 11 patients having symptoms Sham acupuncture consisted of a non-insertive cutaneous
for longer than 1 year. stimulation over the acupoint. Sham controls for acupuncture
Functional scans were acquired using a 3.0 T Siemens Allegra represent one of the most controversial topics in acupuncture
MRI System equipped for echo planar imaging with quadrature research. We chose to impart a sham acupuncture control
head coil. The subject lay supine in the scanner with the head via repetitive tactile stimulation (tapping at 1 Hz) at acupoint
immobilized using cushioned supports. Two sets of structural LI-4 using a 5.88 von Frey monofilament. Subjects were acu-
images were collected using a T1-weighted MPRAGE sequence puncture-naı̈ve and were not informed of a sham acupuncture
(TR/TE = 2.73/3.19 ms, flip angle = 7, FOV = 256 · 256 mm; condition, only that there would be ‘‘different forms’’ of acu-
slice thickness = 1.33 mm). Blood oxygenation level-dependent puncture during fMRI. Subjects lay supine in the scanner with
(BOLD) functional imaging was performed using a gradient their eyes closed and their vision of distal body regions blocked
echo T2*-weighted pulse sequence (TR/TE = 3000/30 ms, flip by the volumetric head coil. Thus, they could not see the inter-
angle = 90, FOV = 200 · 200 mm, 38 sagittal slices, slice thick- vention occurring at their periphery. Sham acupuncture was
ness = 3.0 mm with 0.6 mm interslice gap, matrix = 64 · 64, 140 always performed before verum acupuncture to keep subjects
time points). Image collection was preceded by 4 dummy scans from speculating as to whether the control stimulation was tru-
to allow for equilibration of the MRI signal. ly acupuncture. As order effects would most likely be the same
During an fMRI session, two different types of acupuncture for all subgroups, we felt it was more important to keep sub-
intervention were performed. Both interventions adopted a jects from suspecting that a sham procedure was being used,
7-min block paradigm including a 2-min rest, 1-min stimula- and did not randomize the order of sham and verum stimula-
tion, 2-min rest, 1-min stimulation, 1-min rest block (Fig. 1). tion. In sum, our sham acupuncture stimulation controlled for
Both interventions were performed at acupoint LI-4 (hegu, both superficial, cutaneous somatosensory effects over the acu-
see Fig. 1) over the 1st interosseus m. on the affected hand point, as well as the cognitive processing induced by subjects
for CTS patients (right hand: 9/10) and dominant hand for expecting an ‘‘acupuncture’’ stimulation.
HC subjects (right hand: 8/9). This acupoint was chosen In order to conduct a concurrent psychophysical analysis,
because it is distal to the CTS lesion, is a very common acu- we gathered data by verbal analog scale after each scan run
point used clinically for many chronic pain conditions (Napa- as to the existence of acupuncture sensation, or deqi. The sen-
dow et al., 2004), and has been used in several neuroimaging sations included aching, soreness, pressure, heaviness, fullness,
studies (including our own group) in the past (Hui et al., warmth, cool, numbness, tingling, spreading and dull pain.
2000; Pariente et al., 2005). This acupoint is innervated over The perception of sharp pain was also recorded, though this
the skin surface by the superficial branch of the radial nerve, sensation is not considered to be characteristic of deqi. We also
and at its depth by the deep branch of the ulnar nerve (dorsal asked subjects if they were anxious or relaxed during the scan
interosseus, adductor pollicis muscles) and, if the needle is run. This procedure has been successfully used by our group in
inserted deep enough, the recurrent branch of the median the past to decipher psychophysical response in conjunction
nerve (flexor pollicis brevis muscle). Our insertion depth was with neuroimaging (Hui et al., 2005; Napadow et al., 2005).
chosen for stimulation of deep muscle afferents, as these ulnar Frequency counts of different sensations were compared
and median nerve fibers have been implicated in experimental between different groups with a chi-squared test, significant
analgesia (Chiang et al., 1973). The first two scan runs evalu- at p < 0:05. The intensity of different sensations was analyzed
ated sham acupuncture (see below), while the next two scan with a 2 · 2 ANOVA for factors GROUP and TIME.
runs evaluated verum acupuncture at LI-4.
Verum acupuncture was performed by first inserting a non- 2.4. Single subject fMRI data analysis
magnetic (pure silver), 0.23 mm diameter, 30 mm long acu-
puncture needle (MAEDA Toyokichi Shoten, Tokyo, Japan) Analysis was carried out using a combination of software
into LI-4 before the fMRI scan run began. The insertion depth including FEAT (FMRI Expert Analysis Tool) Version 5.1,
was approximately 1.5 cm. The needle was briefly stimulated part of FSL (FMRIB’s Software Library, www.fmrib.ox.
ac.uk/fsl), and AFNI (Cox, 1996). Data collected from
left-hand dominant HC subjects (1 subject) or predominant
left-handed CTS patients (1 patient) were mirror reversed
across the mid-sagittal plane. The following pre-statistics
processing was applied: motion correction using MCFLIRT;
non-brain removal using Brain Extraction Tool, BET and
FMRIB’s Automated Segmentation Tool, FAST; spatial
smoothing using a Gaussian kernel of FWHM 5 mm; mean-
Fig. 1. Verum and sham acupuncture during fMRI scanning was based intensity normalization; and highpass temporal filtering
accomplished at LI-4 on the hand. The stimulation paradigm consisted (f = 1/180.0 s). Time-series statistical analysis was carried out
of two stimulus blocks and three rest blocks. using FILM (FMRIB’s Improved Linear Model) with local
 V. Napadow et al. / Pain 130 (2007) 254–266 257

autocorrelation correction. The hemodynamic response func- also a greater activation or deactivation in another brain
tion utilized in the general linear model (GLM) analysis was region? A correlation matrix was computed (MATLAB, Math-
defined by the block design paradigm convolved with a pre- works Inc., Natick, MA) and resultant p-values Bonferroni
scribed gamma function (standard deviation = 3 s, mean corrected for multiple comparisons. Regions that demonstrat-
lag = 6 s). ed significant correlation were then correlated to other metrics
of clinical and brain response collected from our CTS patient
2.5. Group fMRI data analysis cohort.

Functional images from each subject were co-registered to

their averaged T1-weighted MPRAGE structural volume, 3. Results
which was co-registered to standard MNI space using affine
transformation with FMRIB’s Linear Image Registration Tool Our neuroimaging results demonstrate differences in
(FLIRT). Standard space parameter estimates from individual how chronic pain patients with CTS process acupunc-
subjects were imported to a 2nd level analysis with FLAME ture compared to HC. The clinical results of acupunc-
(FMRIB’s Local Analysis of Mixed Effects, stage 1 only) in ture intervention in our CTS cohort have been
order to average multiple runs for the same subject. Parameter presented elsewhere (Audette et al., in press; Napadow
estimates from this 2nd level analysis were then used in a full
et al., in press), and demonstrated improvement in sub-
3rd level Markov Chain Monte Carlo (MCMC)-based mixed
jective (Boston CTS questionnaire) as well as objective
effects analysis with FLAME, which takes into account inter-
subject variance. (nerve conduction studies, grip strength) measures.
In order to test for cross-sectional differences between CTS In order to derive the differences in how CTS patients
patients and HC at baseline, the 3rd level analysis was set up as process verum deep intra-muscular acupuncture com-
a 2 · 2 mixed effects ANOVA model with fixed factors pared to HC, controlling for non-specific cutaneous
GROUP (levels include CTS and HC) and STIM (levels somatosensory and cognitive elements of acupuncture
include verum and sham acupuncture) and subjects treated stimulation (sham), we performed a mixed effects 2 · 2
as a nested random factor. The interaction from this analysis ANOVA whose interaction was defined as: [CTS
was defined as [CTS(acup)–CTS(sham)]–[HC(acup)– (acup)–CTS(sham)]–[HC(acup)–HC(sham)]. A signifi-
HC(sham)]. Thus, the interaction tests for whether there were cant interaction was found in the contralateral amygda-
differences between CTS patients and HC subjects in how they
la and lateral hypothalamic area (LHA), in addition to
respond to verum acupuncture, controlling for somatosensory/
other limbic, somatosensory, and cognitive processing
cognitive stimulation. In order to test for longitudinal differ-
ences in brain response for CTS patients before versus after brain regions. These regions included pregenual and
5 weeks of acupuncture treatment, we performed a 2 · 2 mixed anterior-middle anterior, and retrosplenial posterior cin-
effects ANOVA 3rd level analysis with factors TIME (levels gulate cortices (pgACC, amACC, rspPCC), dorsolateral
include baseline and post-acupuncture) and STIM (levels and ventromedial prefrontal cortices (DLPFC,
include verum and sham acupuncture). Here, the interaction VMPFC), anterior insula, septal area, SI, supplementa-
was defined as [CTS.baseline(acup)–CTS.baseline(sham)]– ry motor area (SMA), and thalamus (Table 1). All of the
[CTS.postacup(acup)–CTS.postacup(sham)]. Thus, the inter- interactions were negative except for the LHA. An anal-
action tests for whether there were differences for CTS patients ysis of fMRI signal response in each of the four interac-
before versus after 5 weeks of acupuncture treatment in how tion subgroups found that the negative interaction in the
their brain responds to verum acupuncture, controlling for
left amygdala was mostly due to fMRI signal decrease in
somatosensory/cognitive stimulation. A confounding issue in
CTS patients in response to verum acupuncture (Fig. 2,
presenting a 2 · 2 ANOVA interaction formed by a difference
of differences is that a significant interaction effect may arise Table 1). Similarly, the positive interaction in the LHA
from many different permutations in super or sub-threshold was mainly due to fMRI signal increase in CTS patients
response on the part of any of the 4 subgroups tested. In order in response to verum acupuncture (Fig. 2, Table 1). This
to clarify the origin of a significant interaction effect, imaging was due to the fact that the greatest effect size (%-signal
results were presented with interaction plots and tables in change) in the interaction was by the subgroup: CTS
order to denote which subgroup(s) in the interaction produced patients at baseline with verum acupuncture
the largest %-signal change. stimulation.
Resultant statistical parametric maps were corrected for CTS patients and healthy adults respond differently
multiple comparisons at a false discovery rate (FDR) of 0.01 to sham acupuncture. We found that CTS patients
(Genovese et al., 2002). Data were then clustered with a mini-
respond with greater fMRI signal increase in somatosen-
mum volume of 5 image voxels.
sory regions such as contralateral SI, SII, and posterior
Functional connectivity in CTS patients’ baseline fMRI sig-
nal response to verum acupuncture was explored by calculat- insula; cognitive regions such as DLPFC; and affective
ing a within-subject correlation matrix. Brain regions in the regions such as pre-genual ACC (pgACC, BA24/32)
correlation matrix were limited to those with significant mod- and VMPFC (Fig. 3, Table 2).
ulation under the baseline ANOVA interaction (Table 1). In In order to derive the differences in how CTS patients
other words, we asked the following question: when a patient process acupuncture stimuli before versus after 5 weeks
responds with greater activation in one brain region, is there of clinical acupuncture treatments (again controlling for
 258 V. Napadow et al. / Pain 130 (2007) 254–266

non-specific somatosensory and cognitive elements), we

Summary of the baseline GROUP · STIM interaction which defines the difference between healthy adults and CTS patients in their different fMRI responses to verum and sham acupuncture

%D Sham (l ± r)

HC: healthy controls; pgACC: pre-genual subdivision of anterior cingulate cortex; amACC: anterior-middle subdivision of anterior cingulate cortex; rspPCC: retrosplenial posterior cingulate cortex;
Amyg: amygdala; DLPFC: dorsolateral prefrontal cortex; LatHypothA: lateral hypothalamic area; aInsula: anterior Insula; SI: primary somatosensory cortex; Z-score: most significant voxel in
performed another mixed effects 2 · 2 ANOVA whose

0.20 ± 0.04
0.03 ± 0.05
0.06 ± 0.04
0.23 ± 0.04
0.11 ± 0.08
0.01 ± 0.03
0.60 ± 0.02
0.20 ± 0.05
0.04 ± 0.09
0.17 ± 0.12
0.02 ± 0.05
0.20 ± 0.05
0.07 ± 0.05
Healthy interaction was defined as: [CTS.baseline(acup)–
CTS.baseline(sham)]–[CTS.postacup(acup)–CTS.posta-
cup(sham)]. A significant interaction was again found in
the left amygdala and LHA, in addition to ACC, insula,
and SI (Table 3). The greatest %-signal change in the
interaction was again by the subgroup: CTS patients
%D Acup (l ± r)

at baseline with verum acupuncture stimulation for both

0.00 ± 0.06
0.28 ± 0.06
0.28 ± 0.07
0.07 ± 0.02
0.06 ± 0.03
0.01 ± 0.07
0.28 ± 0.01
0.17 ± 0.22
0.15 ± 0.09
0.28 ± 0.19
0.25 ± 0.07
0.01 ± 0.04
0.14 ± 0.03
the LHA and amygdala (Fig. 4, Table 3). While the lon-
Healthy

gitudinal CTS analysis yielded significant interactions in

limbic and primary somatosensory cortical regions, the
HC test–retest analysis yielded significant ANOVA
interactions in mainly higher associative and cognitive
cortical regions (Table 3). These regions included the
%D Sham (l ± r)

cuneus, superior parietal lobule (BA 7), supramarginal

0.21 ± 0.09
0.29 ± 0.07
0.11 ± 0.03
0.24 ± 0.12
0.11 ± 0.08
0.14 ± 0.06
0.17 ± 0.01
0.18 ± 0.05
0.16 ± 0.06
0.10 ± 0.02
0.30 ± 0.08
0.16 ± 0.09
0.25 ± 0.04

gyrus (BA 40), and ventromedial prefrontal cortex

CTSbase

(VMPFC).
Functional connectivity for CTS patients at baseline
was explored by comparing fMRI signal response to
verum acupuncture in brain regions noted in Table 1,
the baseline cross-sectional ANOVA interaction. Of
%D Acup (l ± r)

the 13 brain regions modulated by this interaction, only

0.12 ± 0.09
0.09 ± 0.08
0.18 ± 0.02
0.21 ± 0.03
0.37 ± 0.09
0.31 ± 0.06
0.36 ± 0.02
0.38 ± 0.03
0.04 ± 0.06
0.31 ± 0.03
0.04 ± 0.03
0.14 ± 0.07
0.03 ± 0.03

one correlation was found to be significant. A positive

CTSbase

correlation (r = 0.89, p < 0:05, corrected) was found

between signal change in the amygdala and LHA
(Fig. 5A). Specifically, those patients that responded to
verum acupuncture with less deactivation in the amyg-
dala had greater activation in the LHA, and vice versa
Z-score
3.30
3.70
3.42
3.30
4.33
3.00
4.15
2.88
3.16
3.57
4.15
4.08
2.85
Baseline GROUP · STIM interaction: (CTSbase.acup–CTSbase.sham)–(HC.acup–HC.sham)

(Fig. 5B).
Furthermore, we found that baseline CTS patients’
fMRI signal response in the LHA in response to verum
cluster; %D: percent signal change expressed as mean ± standard deviation in cluster.

acupuncture was significantly correlated (r = 0.85,

Z (mm)

p < 0:05, Fig. 6) with the separation of somatotopic

6

9
6
8
4
8
34
27
25
13

51
45
13

representations of digits 2 and 3 in contralateral SI.

This separation distance was found to be a marker of
maladaptive cortical plasticity in CTS patients in a
Y (mm)

previous publication (Napadow et al., 2006). No other

40
13
26
53

49
41

19
32
12
8

6
7
0

neuroimaging or clinical biomarker correlated with

patient limbic response to verum acupuncture. Howev-
er, a negative trend was found between the fMRI signal
X (mm)

response in the hypothalamus and the intensity of sharp

pain elicited by verum acupuncture stimulation
13

21
20

34

42
13
3
8
7

8
7

2
6

(r = 0.66, p = 0.055) – i.e., the more sharp pain, the

Talairach

less fMRI signal increase in the LHA. While some

authors have suggested that acupuncture works as
Side
R

R
L

L
L
L

L
L

L
L

stress-induced analgesia (Bragin et al., 1983), this result

suggests that activation of the LHA in CTS patients at
the group level was not likely due to a sharp pain/stress
Region (Brodmann area)

response.
A statistical analysis found no difference between
CTS patients and HC in regard to the prevalence of dif-
LatHypothA
amACC (24)
amACC (24)

ferent sensations elicited by verum acupuncture stimula-

rspPCC (31)
pgACC (24)

Septal Area
stimulation

Thalamus

tion. There were also no statistically significant

SI (3b/1)
VMPFC
DLPFC
Table 1

aInsula
Amyg

differences in sensations elicited at baseline versus after

SMA

5 weeks of acupuncture for CTS patients. Moreover,

 V. Napadow et al. / Pain 130 (2007) 254–266 259

Fig. 2. Cross-sectional fMRI analysis of CTS patients at baseline and healthy controls (HC). The ANOVA interaction tests: (CTSbase.acup–
CTSbase.sham)–(HC.acup–HC.sham). A positive interaction was found in the hypothalamus (A), while a negative interaction was found in the
amygdala (B). Interaction plots demonstrated that in both cases the greatest %-signal change in the interaction was by the subgroup: CTS patients at
baseline with verum acupuncture stimulation.

Fig. 3. Cross-sectional fMRI contrast between CTS patients and healthy controls at baseline in processing sham acupuncture. CTS patients
responded with greater fMRI signal increase in (A) primary sensory, SI; (B) cognitive, dorsolateral prefrontal cortex (DLPFC); and (C) affective/
motivational, rostral ACC (rACC) and ventromedial prefrontal cortex (VMPFC).

there were no statistically significant differences in HC re-test: 1/11) or sham acupuncture (CTS baseline:
reported anxiety during verum acupuncture (CTS base- 1/20; CTS postAcup: 0/15; HC baseline: 1/14; HC
line: 4/15 runs; CTS postAcup: 1/11; HC baseline: 3/16; re-test: 0/11). The mean intensity (across subjects) of
260 V. Napadow et al. / Pain 130 (2007) 254–266

Table 2
Summary of significant clusters of activation difference between CTS patients and healthy controls in processing sham acupuncture
Sham acupuncture processing at baseline
Region (Brodmann area) Talairach %D CTS (l ± r) %D HC (l ± r)
Side X (mm) Y (mm) Z (mm) Z-score
CTS > Healthy controls
pgACC (24/32)/VMPFC L 3 34 3 3.14 0.13 ± 0.07 0.43 ± 0.08
Hippocampus L 30 21 9 3.17 0.00 ± 0.02 0.27 ± 0.02
pInsula L 38 10 10 2.83 0.50 ± 0.09 0.16 ± 0.07
SI L 25 27 59 4.80 0.19 ± 0.08 0.23 ± 0.07
SII L 54 16 14 3.66 0.88 ± 0.15 0.31 ± 0.08
Healthy controls > CTS
None
pgACC: pre-genual subdivision of anterior cingulate cortex; pInsula: posterior insula; SI: primary somatosensory cortex; SII: secondary somato-
sensory cortex; VMPFC: ventromedial prefrontal cortex; Z-score: most significant voxel in cluster; %D: percent signal change expressed as
mean ± standard deviation in cluster.

the maximum deqi sensation was not significantly differ- acupuncture, and HC test–retest. In general, the fact
ent between any subgroups (CTSbase.verum: 3.6 ± 2.1; that some sensations were more common for VA com-
CTSbase.sham: 2.9 ± 1.6; CTSpost.verum: 3.7 ± 1.6; pared to SA should not be surprising. Several studies
CTSpost.sham: 2.7 ± 1.0; HCbase.verum: 4.1 ± 2.0; have noted that non-insertive forms of sham acupunc-
HCbase.sham: 3.3 ± 2.0; HCpost.verum: 4.7 ± 2.2; ture (e.g., placebo needles) are better at mimicking
HCpost.sham: 3.2 ± 2.0; mean ± SD). The mean inten- verum needle insertion, rather than needle manipulation
sity of sharp pain (when experienced) in response to (Tsukayama et al., 2006). This observation applies to
verum acupuncture at LI-4 was in the mild to low mod- our protocol design as well, since our acupuncture stim-
erate range for both groups at baseline (CTS baseline: ulation is best characterized as needle manipulation and
3.2 ± 0.8; HC: 3.4 ± 0.7, mean ± SEM), as well as for identical sensations for VA and SA would not be expect-
both groups after clinical acupuncture treatment (CTS ed. In sum, our results suggest there was little difference
postAcup: 4.0 ± 1.6) or 5 weeks re-test (HC: 2.7 ± 1.0). in psychophysical response between different subgroups
Differences did exist between verum (VA) and sham and thus, the significant imaging findings were likely not
(SA) acupuncture in regard to the types of sensations the result of differences in the sensations elicited.
elicited (Table 4). Specifically, the prevalence of aching
(VA: 75.0% of runs, SA: 21.4%, p = 0.003), soreness 4. Discussion
(VA: 62.5%, SA: 14.3%, p = 0.007), dull pain (VA:
75.0%, SA: 14.3%, p < 0:001), sharp pain (VA: 68.8%, Brain response to acupuncture stimulation may
SA: 14.3%, p = 0.003), and spreading (VA: 50.0%, SA: underlie the efficacy of this age-old healing modality.
14.3%, p = 0.038) was greater during VA for HC sub- In this study we used neuroimaging to explore cross-
jects. For re-tested HC subjects, the prevalence of aching sectional and longitudinal differences in acupuncture
(VA: 90.9%, SA: 9.1%, p < 0:001), soreness (VA: 81.8%, processing between CTS patients (before versus after
SA: 18.2%, p = 0.003), sharp pain (VA: 63.6%, SA: clinical acupuncture treatment) and HC, controlling
18.2%, p < 0:030), and spreading (VA: 63.6%, SA: for superficial somatosensory/cognitive elements of
9.1%, p = 0.008) was greater for VA. For baseline stimulation. Our results demonstrated that baseline
CTS patients, the prevalence of aching (VA: 80.0%, CTS patients respond to acupuncture with more pro-
SA: 15.0%, p < 0:001), fullness (VA: 33.3%, SA: 5.0%, nounced modulation in a network of limbic brain
p = 0.028), dull pain (VA: 66.7%, SA: 0.0%, regions including the amygdala and hypothalamus. Fur-
p < 0:001), sharp pain (VA: 60.0%, SA: 15.0%, thermore, brain processing of acupuncture stimulation
p = 0.006), and spreading (VA: 53.3%, SA: 20.0%, may change after a course of clinical acupuncture
p = 0.040) was greater during VA compared to SA. treatment.
For CTS patients after acupuncture therapy, the preva-
lence of aching (VA: 72.7%, SA: 6.7%, p < 0:001), sore- 4.1. Amygdala response to acupuncture
ness (VA: 54.5%, SA: 0.0%, p < 0:001), dull pain (VA:
72.7%, SA: 6.7%, p < 0:001), and spreading (VA: The amygdala is a critical component of the limbic
45.5%, SA: 6.7%, p = 0.020) was greater for VA. Thus, system which can translate somatosensory stimuli (e.g.,
for the most part, VA and SA differed consistently for acupuncture) into affective states. The amygdala is
the different subgroups – CTS baseline, CTS post- important in processing emotions, especially fear and
Table 3
Summary of the STIM · TIME interaction for CTS patients and HC, which defines the difference between baseline and post-acupuncture CTS patients (or re-test for HC) in their different fMRI
responses to verum and sham acupuncture
STIM · TIME interaction:
Carpal tunnel syndrome: (CTSbase.acup–CTSbase.sham)–(CTSpost.acup–CTSpost.sham)
Region (Brodmann Area) Talairach CTSbase CTSbase CTSpost CTSpost
%D Acup (l ± r) %D Sham (l ± r) %D Acup (l ± r) %D Sham (l ± r)
Side X (mm) Y (mm) Z (mm) Z-score
amACC (24) R 7 28 23 3.27 0.21 ± 0.02 0.38 ± 0.03 0.22 ± 0.04 0.03 ± 0.06
Amygdala L 22 8 11 3.79 0.45 ± 0.09 0.20 ± 0.11 0.03 ± 0.05 0.11 ± 0.05
VMPFC R 12 48 11 3.37 0.46 ± 0.05 0.19 ± 0.12 0.08 ± 0.05 0.28 ± 0.07
LatHypothA L 8 8 7 3.37 0.29 ± 0.12 0.12 ± 0.05 0.06 ± 0.07 0.16 ± 0.03
aInsula R 37 2 3 3.54 0.22 ± 0.02 0.22 ± 0.01 0.21 ± 0.01 0.00 ± 0.03
a/pInsula L 37 3 7 3.95 0.06 ± 0.05 0.32 ± 0.10 0.23 ± 0.11 0.12 ± 0.06
SI (3b/1) L 21 28 55 4.37 0.21 ± 0.05 0.10 ± 0.06 0.10 ± 0.03 0.16 ± 0.04
Healthy Controls: (HCbase.acup–HCbase.sham)–(HCpost.acup–HCpost.sham)
Region (Brodmann area) Talairach HCbase HCbase HCpost HCpost
%D Acup (l ± r) %D Sham (l ± r) %D Acup (l ± r) %D Sham (l ± r)
Side X (mm) Y (mm) Z (mm) Z-score
V. Napadow et al. / Pain 130 (2007) 254–266

cuneus L 16 74 28 3.66 0.04 ± 0.01 0.02 ± 0.02 0.13 ± 0.01 0.20 ± 0.02
SMG (40) R 59 36 37 3.96 0.32 ± 0.09 0.16 ± 0.03 0.06 ± 0.02 0.46 ± 0.06
SPL (7) R 29 50 60 3.55 0.06 ± 0.10 0.30 ± 0.10 0.05 ± 0.05 0.29 ± 0.02
VMPFC L 1 43 7 2.85 0.30 ± 0.03 0.64 ± 0.10 0.44 ± 0.10 0.01 ± 0.05
amACC: anterior-middle subdivision of anterior cingulate cortex; LatHypothA: lateral hypothalamic area; aInsula: anterior insula; a/pInsula: anterior/posterior insula; SI: secondary somato-
sensory area; SMG: supramarginal gyrus; SPL: superior parietal lobule; VMPFC: ventromedial prefrontal cortex. Z-score: most significant voxel in cluster; %D: percent signal change expressed as
mean ± standard deviation in cluster.
261
262 V. Napadow et al. / Pain 130 (2007) 254–266

Fig. 4. Longitudinal fMRI analysis of CTS patients at baseline and after 5 weeks of acupuncture treatment. The ANOVA interaction tests:
(CTSbase.acup–CTSbase.sham)–(CTSpost.acup–CTSpost.sham). A positive interaction was found in the hypothalamus (A), while a negative
interaction was found in the amygdala (B). Interaction plots demonstrated that the greatest %-signal change in the interaction was by the subgroup:
CTS patients at baseline with verum acupuncture stimulation.

Fig. 5. Functional connectivity was explored by calculating the percent signal change correlation matrix for CTS patients at baseline responding to
verum acupuncture. Potential brain regions (13) were drawn from those demonstrating a significant cross-sectional ANOVA interaction (Table 1).
(A) The only two regions which demonstrated a significant correlation were the amygdala and hypothalamus. (B) The less deactivation in the
amygdala, the more activation in the hypothalamus for CTS patients at baseline in response to acupuncture.

defensive behavior (Zald, 2003). Neuroimaging studies potentiation (LTP) which induces synaptic strengthen-
have demonstrated amygdala activation in response to ing for connections within the amygdala as well as
acute pain (Bingel et al., 2002; Bornhovd et al., 2002). between the amygdala and other limbic regions
Electrophysiological studies in rats have corroborated (Chapman et al., 1990; Neugebauer et al., 2004).
these findings (Bernard et al., 1992), and have also pro- In our data, the amygdala was found to be deactivat-
vided evidence of sensitization and hyperactivation in ed in CTS patients in response to verum acupuncture.
the amygdala following induction of an inflammatory As chronic pain has been associated with high
chronic pain state (Neugebauer and Li, 2003). Such background or baseline activity in the amygdala
neuroplastic change is thought to result from long term (Neugebauer and Li, 2003), we propose that in some
 V. Napadow et al. / Pain 130 (2007) 254–266 263

Fig. 6. The magnitude of hypothalamic activation in response to verum acupuncture for CTS patients at baseline was found to correlate negatively
with their digit 2/digit 3 somatotopic separation in contralateral SI. The worse a patient’s central maladaptive neuroplasticity, the more they
responded to acupuncture with hypothalamic activation.

patients, acupuncture may function to ameliorate the limbic activity, particularly in the amygdala, is part of a
affective component of chronic pain by deactivating a cognitive coping mechanism for pain suppression (Hsieh
sensitized, hyperactive amygdala. A short-term deacti- et al., 1999; Petrovic and Ingvar, 2002). Typically, this
vation within the amygdala may initiate a progressive occurs when the subject is unable to make a behavioral
normalization of activity via neuroplasticity, leading to response to avoid an aversive context (e.g., acupuncture
long-term physiologically and clinically relevant or pain experiments inside an MRI scanner). In some of
response. our patients, subjective expectation of pain from acu-
Interestingly, while experimental pain typically puncture may have evoked cognitive coping mechanisms
produces signal increase in the amygdala, several neuro- manifest by amygdala deactivation. This may have been
imaging studies have also noted signal decrease in this particularly amplified for chronic pain patients, naı̈ve to
limbic region (Derbyshire et al., 1997; Becerra et al., acupuncture at baseline fMRI testing. As therapeutic
1999; Petrovic et al., 1999; Becerra et al., 2001; Petrovic acupuncture rarely produces analgesia from a single
et al., 2004). It has been theorized that down-regulating exposure, rather working through accumulative,

Table 4
Psychophysics analysis for different subgroups
Ache Sore Press Heavy Full Warm Cool Numb Tingle Pain (dull) Pain (sharp) Spread
CTSbase.verum vs. HCbase.verum n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s.
CTSbase.verum vs. CTSpost.verum n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s.
HCbase.verum vs. HCpost.verum n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s.
CTSbase.verum vs. CTSbase.sham 0.001 n.s. n.s. n.s. 0.028 n.s. n.s. n.s. n.s. 0.001 0.006 0.040
Verum 80.0% 33.3% 66.7% 60.0% 53.3%
Sham 15.0% 5.0% 0.0% 15.0% 20.0%
HCbase.verum vs. HCbase.sham 0.003 0.007 n.s. n.s. n.s. n.s. n.s. n.s. n.s. 0.001 0.003 0.038
Verum 75.0% 62.5% 75.0% 68.8% 50.0%
Sham 21.4% 14.3% 14.3% 14.3% 14.3%
CTSpost.verum vs. CTSpost.sham 0.001 0.001 n.s. n.s. n.s. n.s. n.s. n.s. n.s. 0.001 n.s. 0.020
Verum 72.7% 54.5% 72.7% 45.5%
Sham 6.7% 0.0% 6.7% 6.7%
HCpost.verum vs. HCpost.sham 0.001 0.003 n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. 0.030 0.008
Verum 90.9% 81.8% 63.6% 63.6%
Sham 9.1% 18.2% 18.2% 9.1%
Significant differences in prevalence of deqi sensations elicited by acupuncture were evaluated using a chi-square test. Prevalence was defined by the
proportion of total imaging runs in which a subject experienced each different sensation. The table shows the Pearson chi-square 2-sided p-value (in
bold) and percentage of runs for individual subgroup. n.s., Non-significant.
264 V. Napadow et al. / Pain 130 (2007) 254–266

regularly spaced treatments (Carlsson, 2002), a learned may modulate sympathetic response local to the CTS
cognitive coping strategy may become more entrained lesion via central modulation of hypothalamic activity.
with repeated exposure to acupuncture, leading to a Other studies have found either fMRI signal increase
longer-term strategy for pain suppression. (Wu et al., 1999) or signal decrease (Hui et al., 2000) in
the hypothalamus in response to LI-4 manual acupunc-
4.2. Hypothalamus response to acupuncture ture stimulation in healthy adults. Our results demon-
strated that the significant positive interaction in the
We also found that acupuncture is associated with LHA was formed by a signal increase for CTS patients
activity in LHA for CTS patients. Somatosensory input and signal decrease for HC – more consistent with Hui
to the hypothalamus is carried by both direct (unmodu- et al., (i.e., signal decrease for healthy adults). The hypo-
lated, monosynaptic) and indirect (modulated by corti- thalamic signal increase seen by Wu et al., may have
cal and sub-cortical regions, polysynaptic) pathways been due to the type of needle stimulation (combined
(Bernard et al., 1996; Burstein, 1996). Limbic input to twisting and lifting-thrusting) which produced more
the LHA from the amygdala is transmitted by the ven- intense psychophysical response (mean of 6 on a 0–10
tral amygdalofugal pathway (Parent, 1996), thereby scale) compared to our study (HC: 4.1; CTS: 3.6).
translating affect and emotion into autonomic response. Ultimately, acupuncture most likely affects CTS
Interestingly, our data suggested functional connectivity pathology and symptomatology through both periphe-
between the amygdala and LHA for baseline CTS ral and central mechanisms. As CTS is due to ischemic
patients (see Fig. 5B). In addition, the greater the mal- neuropathology, the ability of acupuncture to induce
adaptive plasticity in contralateral SI (as measured by increased blood flow both superficially and deep to the
digit 2/digit 3 representation blurring), the greater the surface (Sandberg et al., 2003) may improve microcircu-
hypothalamic signal increase in response to acupuncture lation to the impacted median nerve within the carpal
(see Fig. 6). Hence, patients with central manifestations tunnel. As part of our therapeutic acupuncture protocol,
of their peripheral CTS lesion may preferentially CTS patients received electro-acupuncture at acupoint
respond to acupuncture stimulation with hypothalamic PC-7, which is centered over the median nerve at the
activity. wrist crease. It has been suggested that the vasodilatory
The LHA is capable of producing anti-nociception effect of acupuncture is due to the release of CGRP and
through efferents to modulatory serotonergic neurons other vasodilatory neuropeptides (Sato et al., 2000).
in the rostroventral medulla (RVM) acting on the spinal Future studies should assess peripheral effects of acu-
dorsal horn (Holden et al., 2005). The hypothalamus is puncture concurrently with central effects.
also an important component of the endogenous opioid-
ergic pain control system, which functions through con- 4.3. Processing of sham acupuncture
nections from the hypothalamus to periaqueductal gray
to RVM, and has been implicated as a potential mecha- Deactivations in response to verum acupuncture in
nism of acupuncture action (Pomeranz and Chiu, 1976; baseline CTS patients were in stark contrast to the greater
Takeshige et al., 1992; Han, 2004). However, this system degree of activation in response to a sham acupuncture
is associated more with arcuate nucleus than LHA control for CTS patients at baseline compared to HC.
activity. More pronounced activity in primary sensorimotor
The hypothalamus is also an important component of regions corroborated our previous data for sensorimo-
the central autonomic network. Through the cholinergic tor cortical hyperactivation in CTS patients in response
anti-inflammatory pathway, hypothalamic (autonomic) to non-noxious stimulation of affected digits (Napadow
activity can inhibit inflammation (Tracey, 2002). Acu- et al., 2006). Additionally, our data demonstrated that
puncture may be able to down-regulate inflammatory when the stimulus was a context-laden sham acupunc-
response by tilting sympathovagal balance towards ture intervention, cognitive (DLPFC) and affective
vagal enhancement. In fact, acupuncture has been noted (rACC, VMPFC) brain regions were also activated
to affect sympathovagal balance in human studies more for CTS patients compared to HC. Interestingly,
assessing heart rate and heart rate variability (Nishijo these regions have also been implicated in executive con-
et al., 1997; Haker et al., 2000). However, CTS is more trol of placebo analgesia (Zubieta et al., 2005; Kong
typically characterized by non-inflammatory fibrosis et al., 2006).
and it remains to be seen if a general anti-inflammatory
effect could impact CTS. Even so, peripheral autonomic 4.4. Methodological issues and limitations
pathology has been noted in CTS via thermography
(Ming et al., 2005; Orlin et al., 2005) and sympathetic To explore the specificity of acupuncture limbic mod-
skin response (SSR) (Kiylioglu et al., 2005). As the ulation in a longitudinal CTS analysis, a similar test–
hypothalamus is one of the principal SSR effector retest analysis was performed in HC (see Table 3). In
regions in the brain (Vetrugno et al., 2003), acupuncture contrast to the limbic modulations, the HC test-rested
 V. Napadow et al. / Pain 130 (2007) 254–266 265

analysis yielded significant interactions in mostly higher Becerra L, Breiter HC, Wise R, Gonzalez RG, Borsook D. Reward
associative and cognitive cortical regions, perhaps corre- circuitry activation by noxious thermal stimuli. Neuron
2001;32:927–46.
sponding to a learning effect inherent in a test–retest Bernard JF, Huang GF, Besson JM. Nucleus centralis of the amygdala
experiment. The one exception was the VMPFC, impli- and the globus pallidus ventralis: electrophysiological evidence for
cated as a para-limbic region processing affective states. an involvement in pain processes. J Neurophysiol 1992;68:551–69.
However, this region is also part of the task-non-specific Bernard JF, Bester H, Besson JM. Involvement of the spino-
default network (Gusnard et al., 2001), and fMRI signal parabrachio-amygdaloid and -hypothalamic pathways in the
autonomic and affective emotional aspects of pain. Prog Brain
decreases were found in 3 of the 4 HC subgroups (Table Res 1996;107:243–55.
3). Hence, this particular interaction may not be specif- Bingel U, Quante M, Knab R, Bromm B, Weiller C, Buchel C.
ically related to differences in acupuncture processing Subcortical structures involved in pain processing: evidence from
but simply learned differences in task salience and atten- single-trial fMRI. Pain 2002;99:313–21.
tion (McKiernan et al., 2003). Bornhovd K, Quante M, Glauche V, Bromm B, Weiller C, Buchel C.
Painful stimuli evoke different stimulus-response functions in the
Multiple controls were used in this fMRI study. amygdala, prefrontal, insula and somatosensory cortex: a single-
When fMRI results are presented by an interaction trial fMRI study. Brain 2002;125:1326–36.
which is composed of a difference of differences (not Bragin EO, Vasilenko GF, Durinjan RA. The study of the central grey
uncommon in neuroimaging studies), it is important to matter in mechanisms of different kinds of analgesia: effects of
appreciate which subgroups were the prime delimiters lesions. Pain 1983;16:33–40.
Burstein R. Somatosensory and visceral input to the hypothalamus
of the significant interaction. We chose to report this and limbic system. Prog Brain Res 1996;107:257–67.
information through interaction plots in our figures Carlsson C. Acupuncture mechanisms for clinically relevant long-term
and as separate data columns in our tables. This effects–reconsideration and a hypothesis. Acupunct Med
approach allowed us to focus our discussion on brain 2002;20:82–99.
regions where interaction effects were mainly due to Chapman PF, Kairiss EW, Keenan CL, Brown TH. Long-term
synaptic potentiation in the amygdala. Synapse 1990;6:271–8.
verum acupuncture processing in CTS patients – i.e., Chiang C, Chang C, Chu H, Yang L. Peripheral afferent pathway for
the amygdala and LHA. acupuncture analgesia. Sci Sinica 1973;16:210–7.
In conclusion, we have presented evidence that acu- Cox RW. AFNI: software for analysis and visualization of functional
puncture processing in the brains of CTS patients differs magnetic resonance neuroimages. Comput Biomed Res
from that of HC. Patients with CTS respond to acu- 1996;29:162–73.
Derbyshire SW, Jones AK, Gyulai F, Clark S, Townsend D, Firestone
puncture with more pronounced fMRI signal decrease LL. Pain processing during three levels of noxious stimulation
in the amygdala and signal increase in the LHA. The produces differential patterns of central activity. Pain
functional connectivity found between the amygdala 1997;73:431–45.
and hypothalamus suggests that a coordinated limbic Genovese CR, Lazar NA, Nichols T. Thresholding of statistical maps
response to acupuncture stimuli may underlie the effica- in functional neuroimaging using the false discovery rate. Neuro-
image 2002;15:870–8.
cy of this controversial healing modality. Gusnard DA, Akbudak E, Shulman GL, Raichle ME. Medial prefrontal
cortex and self-referential mental activity: relation to a default mode
of brain function. Proc Natl Acad Sci USA 2001;98:4259–64.
Acknowledgements Haker E, Egekvist H, Bjerring P. Effect of sensory stimulation
(acupuncture) on sympathetic and parasympathetic activities in
healthy subjects. J Auton Nerv Syst 2000;79:52–9.
This research was supported by grants from the NIH: Han JS. Acupuncture and endorphins. Neurosci Lett 2004;361:258–61.
NCCAM (K01-AT002166-01, P01-AT002048-02), NCRR Holden JE, Farah EN, Jeong Y. Stimulation of the lateral hypothal-
(P41RR14075), and the Mental Illness and Neurosci- amus produces antinociception mediated by 5-HT1A, 5-HT1B and
ence Discovery (MIND) Institute, as well as the 5-HT3 receptors in the rat spinal cord dorsal horn. Neuroscience
Department of Physical Medicine and Rehabilitation 2005;135:1255–68.
Hsieh JC, Stone-Elander S, Ingvar M. Anticipatory coping of pain
Mini-grant Program, Harvard Medical School expressed in the human anterior cingulate cortex: a positron
(#R02034). We would also like to thank Richard Harris emission tomography study. Neurosci Lett 1999;262:61–4.
and Bruce Rosen for constructive criticism of this Hsieh JC, Tu CH, Chen FP, Chen MC, Yeh TC, Cheng HC, et al.
manuscript. Activation of the hypothalamus characterizes the acupuncture
stimulation at the analgesic point in human: a positron emission
tomography study. Neurosci Lett 2001;307:105–8.
Hui KK, Liu J, Makris N, Gollub RL, Chen AJ, Moore CI, et al.
References Acupuncture modulates the limbic system and subcortical gray
structures of the human brain: evidence from fMRI studies in
Audette J, Ryan A, Liu J, Li M, Kettner NW, Kwong KK, et al. normal subjects. Hum Brain Mapp 2000;9:13–25.
Acupuncture for chronic carpal tunnel syndrome: a pilot prospec- Hui KK, Liu J, Marina O, Napadow V, Haselgrove C, Kwong KK,
tive controlled outcome study. Am J Phys Med Rehabil, in press. et al. The integrated response of the human cerebro-cerebellar and
Becerra LR, Breiter HC, Stojanovic M, Fishman S, Edwards A, limbic systems to acupuncture stimulation at ST 36 as evidenced by
Comite AR, et al. Human brain activation under controlled fMRI. Neuroimage 2005;27:479–96.
thermal stimulation and habituation to noxious heat: an fMRI Kaptchuk T. The web that has no weaver: understanding Chinese
study. Magn Reson Med 1999;41:1044–57. medicine. Chicago: Contemporary Books; 2000.
266 V. Napadow et al. / Pain 130 (2007) 254–266

Kaptchuk TJ. Acupuncture: theory, efficacy, and practice. Ann Intern Pariente J, White P, Frackowiak RS, Lewith G. Expectancy and belief
Med 2002;136:374–83. modulate the neuronal substrates of pain treated by acupuncture.
Keir PJ, Rempel DM. Pathomechanics of peripheral nerve loading. Neuroimage 2005;25:1161–7.
Evidence in carpal tunnel syndrome. J Hand Ther 2005;18:259–69. Petrovic P, Ingvar M. Imaging cognitive modulation of pain process-
Kiylioglu N, Akyol A, Guney E, Bicerol B, Ozkul A, Erturk A. ing. Pain 2002;95:1–5.
Sympathetic skin response in idiopathic and diabetic carpal tunnel Petrovic P, Ingvar M, Stone-Elander S, Petersson KM, Hansson P. A
syndrome. Clin Neurol Neurosurg 2005;108:1–7. PET activation study of dynamic mechanical allodynia in patients
Kong J, Gollub RL, Rosman IS, Webb JM, Vangel MG, Kirsch I, with mononeuropathy. Pain 1999;83:459–70.
et al. Brain activity associated with expectancy-enhanced placebo Petrovic P, Carlsson K, Petersson KM, Hansson P, Ingvar M.
analgesia as measured by functional magnetic resonance imaging. J Context-dependent deactivation of the amygdala during pain. J
Neurosci 2006;26:381–8. Cogn Neurosci 2004;16:1289–301.
McKiernan KA, Kaufman JN, Kucera-Thompson J, Binder JR. A Pomeranz B, Chiu D. Naloxone blockade of acupuncture analgesia:
parametric manipulation of factors affecting task-induced deactiva- endorphin implicated. Life Sci 1976;19:1757–62.
tion in functional neuroimaging. J Cogn Neurosci 2003;15:394–408. Sandberg M, Lundeberg T, Lindberg LG, Gerdle B. Effects of
Ming Z, Zaproudina N, Siivola J, Nousiainen U, Pietikainen S. acupuncture on skin and muscle blood flow in healthy subjects. Eur
Sympathetic pathology evidenced by hand thermal anomalies in J Appl Physiol 2003;90:114–9.
carpal tunnel syndrome. Pathophysiology 2005;12:137–41. Sato A, Sato Y, Shimura M, Uchida S. Calcitonin gene-related peptide
Napadow V, Liu J, Kaptchuk TJ. A systematic study of acupuncture produces skeletal muscle vasodilation following antidromic stim-
practice: acupoint usage in an outpatient setting in Beijing, China. ulation of unmyelinated afferents in the dorsal root in rats.
Complement Ther Med 2004;12:209–16. Neurosci Lett 2000;283:137–40.
Napadow V, Makris N, Liu J, Kettner NW, Kwong KK, Hui KK. Stern J, Jeanmonod D, Sarnthein J. Persistent EEG overactivation in
Effects of electroacupuncture versus manual acupuncture on the the cortical pain matrix of neurogenic pain patients. Neuroimage
human brain as measured by fMRI. Hum Brain Mapp 2006.
2005;24:193–205. Sud V, Freeland AE. Biochemistry of carpal tunnel syndrome.
Napadow V, Kettner N, Ryan A, Kwong KK, Audette J, Hui KK. Microsurgery 2005;25:44–6.
Somatosensory cortical plasticity in carpal tunnel syndrome–a Takeshige C, Sato T, Mera T, Hisamitsu T, Fang J. Descending pain
cross-sectional fMRI evaluation. Neuroimage 2006;31:520–30. inhibitory system involved in acupuncture analgesia. Brain Res
Napadow V, Liu J, Li M, Kettner N, Ryan A, Kwong KK, et al. Bull 1992;29:617–34.
Somatosensory Cortical Plasticity in Carpal Tunnel Syndrome Tracey KJ. The inflammatory reflex. Nature 2002;420:853–9.
Treated by Acupuncture. Human Brain Mapp, in press. Tsukayama H, Yamashita H, Kimura T, Otsuki K. Factors that
Neugebauer V, Li W. Differential sensitization of amygdala neurons to influence the applicability of sham needle in acupuncture trials: two
afferent inputs in a model of arthritic pain. J Neurophysiol randomized, single-blind, crossover trials with acupuncture-expe-
2003;89:716–27. rienced subjects. Clin J Pain 2006;22:346–9.
Neugebauer V, Li W, Bird GC, Han JS. The amygdala and persistent Vetrugno R, Liguori R, Cortelli P, Montagna P. Sympathetic skin
pain. Neuroscientist 2004;10:221–34. response: basic mechanisms and clinical applications. Clin Auton
Nishijo K, Mori H, Yosikawa K, Yazawa K. Decreased heart rate by Res 2003;13:256–70.
acupuncture stimulation in humans via facilitation of cardiac vagal Wu MT, Hsieh JC, Xiong J, Yang CF, Pan HB, Chen YC, et al.
activity and suppression of cardiac sympathetic nerve. Neurosci Central nervous pathway for acupuncture stimulation: localization
Lett 1997;227:165–8. of processing with functional MR imaging of the brain –
Orlin JR, Stranden E, Slagsvold CE. Effects of mechanical irritation on preliminary experience. Radiology 1999;212:133–41.
the autonomic part of the median nerve. Eur J Neurol Zald DH. The human amygdala and the emotional evaluation of
2005;12:144–9. sensory stimuli. Brain Res Brain Res Rev 2003;41:88–123.
Papanicolaou GD, McCabe SJ, Firrell J. The prevalence and charac- Zhu L. New acupuncture (Xian Zhen-jiu Xue). Beijing: People’s
teristics of nerve compression symptoms in the general population. Press; 1954.
J Hand Surg [Am] 2001;26:460–6. Zubieta JK, Bueller JA, Jackson LR, Scott DJ, Xu Y, Koeppe RA,
Parent A. Carpenter’s human neuroanatomy. Baltimore: Williams & et al. Placebo effects mediated by endogenous opioid activity on
Wilkins; 1996. mu-opioid receptors. J Neurosci 2005;25:7754–62.