Article Text


Additional evidence for the sustained effect of acupuncture at the vision-related acupuncture point, GB37
  1. Jixin Liu,
  2. Jiaofen Nan,
  3. Shiwei Xiong,
  4. Guoying Li,
  5. Wei Qin,
  6. Jie Tian
  1. Life Sciences Research Center, School of Life Sciences and Technology, Xidian University, Xi'an, China
  1. Correspondence to Dr Jixin Liu, Life Sciences Research Center, School of Life Sciences and Technology, Xidian University, Xi'an, Shaanxi 710071, China; liujixin2010{at}


Objective To investigate the dynamics underlying the sustained effect of acupuncture as a possible explanation of earlier findings that acupuncture stimulation at the vision-related acupuncture point, GB37, cannot specifically change the functional MRI (fMRI) signals of the visual cortex compared with stimulation at an adjacent non-meridian point.

Methods The ‘on–off’ experimental design was separated into four series conditions: 1 min of baseline scanning at the beginning, then two stimulation epochs separated by a 50 s ‘rest’ period, and then a 1 min ‘rest’ epoch. The standard General Linear Model (GLM) approach and multi-conditions analysis were used.

Results Results from the multi-conditions analysis were different from those from the standard GLM analysis. We found that the neural signal of the limbic–paralimbic–neocortical system after acupuncture stimulus lasted longer than the putative period. Furthermore, the fMRI signal changes in the occipital cortex showed different temporal patterns between GB37 and the non-meridian point.

Conclusions Owing to the sustained effect of acupuncture, standard GLM analysis may be unsuitable for ‘on–off’ design acupuncture studies and lead to uncertain and contradictory results. The findings from this study suggest that acupuncture at GB37 can induce complex brain activity in the vision cortex. The state-related neural signal may reflect one of the significant characteristics underlying acupuncture.

  • Acupuncture

Statistics from


Clinical acupuncture typically has an effective treatment phase several hours after the needling session. Psychophysical analysis by Mayer1 and Price et al2 showed a sustained acupuncture analgesic effect during the resting state after the needling session. In recent years, increasing numbers of acupuncture studies on functional MRI (fMRI) have focused on the delayed effects of acupuncture. Our group has already explored sustained responses of the brain networks induced by acupuncture.3–10 We have previously discussed the relationship between the immediate and delayed response of acupuncture and found that acupuncture at an acupuncture point induced more robust response patterns after a period of acupuncture stimulation than acupuncture at an adjacent non-meridian point (NMP).4 Furthermore, these results have been confirmed in analysis of brain networks development.5 Therefore, this sustained effect should be taken into account when designing paradigms and choosing the statistical model to use for acupuncture studies.

Kong et al11 have concluded that electroacupuncture stimulation at both vision-related acupuncture points and a non-acupuncture point can produce moderate comparable decreases in fMRI signal in the occipital cortex between groups, but not changes in spatial distribution. Similar results were obtained by Gareus et al.12 These results do not support the traditional Chinese medicine (TCM) theory of acupuncture point specificity. In these studies, the ‘on–off’ fMRI paradigm and the General Linear Model (GLM) approach were used.11 The premise of these models is that the effect of the stimulus during the ‘on’ period will have little effect during the ‘off’ period.13 The exact timing and duration of psychological events must be specified a priori.14 However, too little attention has been paid to the sustained effect in previous studies.11 ,12 ,15 If the effects of acupuncture peak long after the needling session, its temporal features would not concur with the ‘on–off’ response patterns; in which case, the estimation in the GLM contrast analysis may lead to statistical errors. To address these potential problems, in this study we took into account the sustained effect of acupuncture to reconsider whether or not acupuncture stimulation at a vision-related acupuncture point can specifically change the fMRI signals in the visual cortex.

Bai et al7 separated the ‘on–off’ experimental design into different conditions across each subject, and investigated the changes in fMRI signal between each condition and the baseline (BL). They found that the activity pattern during rest was significantly associated with the stimulus-related effects, some of which were even higher than that of the stimulation phases.7 In the present study, a similar statistical model was used. We paid more attention to the temporal characteristics of the neural signal changes during and after acupuncture stimulation at a vision-related acupuncture point and an NMP. We also compared our results with those from standard GLM analysis.

Materials and methods

All research procedures were approved by the West China Hospital Subcommittee on Human Studies and were conducted in accordance with the Declaration of Helsinki. All subjects were fully informed of the nature of the research and had given written consent.


All the participants were recruited from a group of 22 college students (11 male and 11 female; mean±SD age 22.3±1.7years). All were right-handed. They were all acupuncture naïve, did not have a history of major medical illnesses, head trauma or neuropsychiatric disorders, had not used prescription medications within the preceding month, and did not have any contraindications to exposure to a high magnetic field.

Acupuncture experiment procedures

In the acupuncture experiment, all subjects were blinded to the type and order of stimulations. During scanning, subjects were instructed to keep their eyes closed in order to prevent them from observing the procedures. The presentation sequence of the acupuncture and sham protocols was randomised across the fMRI runs, and the order of presentation was counterbalanced across subjects.

Acupuncture was performed at the acupuncture point, GB37, which is widely used to treat eye disorders.16 The NMP was located about 1.5 cm posterior and inferior to GB37 based on the TCM theory, where there is neither an acupuncture point nor a meridian passing through. Acupuncture stimulation was delivered using a sterile disposable stainless-steel acupuncture needle 0.25 mm in diameter and 40 mm in length. The needle was inserted perpendicularly to a depth of 20–30 mm. The acupuncturist then rotated the needle clockwise and counterclockwise for 1 min at a rate of 60 times per min. The experiment was performed by the same experienced and licensed acupuncturist on all subjects.

We used the standard ‘on–off’ experimental paradigm during the fMRI scan, in which the needle was kept in the point for 1 min before manipulation (the acupuncturist did not feel deqi at these time points), followed by two separate stimulation blocks. The ‘on–off’ design included an initial 1 min of BL scanning, then two stimulation epochs (S1 and S2) separated by a 50 s ‘rest’ period (R1), followed by another 1 min ‘rest’ epoch (R2) (figure 1). The acupuncturist felt deqi on S1 and S2. The entire scan lasted for 220 s per run. The sham stimulation was performed identically with the manual acupuncture condition—for example, needle depth, stimulation intensity and manipulation method.

Figure 1

Experimental paradigm. In our different-conditions analysis, the signal intensity of the baseline (BL) phase served as a control BL for detecting signal changes in the four sequential epochs: stimulation 1 (S1), rest 1 (R1), stimulation 2 (S2) and rest 2 (R2).

At the end of each fMRI scan, subjects were asked to quantify their sensations using a 10-point visual analogue scale (VAS). Different data intervals of the VAS represented the extent of the different sensations of the subjects, with 0=no sensation, 1–3=mild, 4–6=moderate, 7–8=strong, 9=severe, and 10=unbearable. Because sharp pain was considered an inadvertent noxious stimulation, the subject whose scale feeling of sharp pain was greater than the mean by more than 2 SDs was excluded from further analysis.17

fMRI scanning procedure

The experiments were carried out in a 3T GE scanner. A gradient echo T2*-weighted sequence with in-plane resolution of 3.75×3.75 mm (time to echo (TE) 30 ms, repetition time (TR) 2 s, matrix 64×64, field of view (FOV) 240 mm, flip angle 90°) and a set of T1-weighted high-resolution structural images (TE 3.39 ms, TR 2.7 s, matrix 256×256, FOV 256 mm, flip angle 7°, in plane resolution 1×1 mm, slice thickness 1 mm) were acquired.

fMRI data analysis

The data were preprocessed by removing the first 10 time points to eliminate non-equilibrium effects of magnetisation. The remaining time points (105 volumes) were used for further analysis. All functional images were processed using the following steps: (1) rigid body correction for geometrical displacements caused by head movement; (2) co-registration with the Montreal Neurological Institute (MNI) EPI template image; (3) images were smoothed with a 6 mm Gaussian kernel to decrease spatial noise. Taking into consideration the sustained effect of acupuncture, we defined four different conditions per session after the BL (as shown in four colours in figure 1). Naturally, four t-contrasts were defined: S1 minus BL, R1 minus BL, S2 minus BL, and R2 minus BL. Each comparison of the contrast was used to find voxel-wise signal changes for a particular condition relative to the BL. The standard GLM analysis was also performed in the same manner as in Kong et al's study ((S1+S2)−(BL+R1+R2)).11 The preprocessing steps and the statistical analysis were all performed with SPM5 software.


No participants reported feeling sharp pain during any of the scans. No significant difference was found between the sensation intensity of acupuncture at GB37 and NMP (paired t test, p>0.05).

Acupuncture at GB37 versus BL epochs

Group analysis averaged across all subjects in different epochs of acupuncture stimulation is shown in table 1 (p<0.001). Two needling epochs compared with the control fixation (S1/S2 minus BL) produced an extensive signal increase in the neocortical, brainstem, limbic/paralimbic areas and cerebellum, including the prefrontal/temporal cortex, occipital cortex, bilateral secondary/primary somatosensory cortex (SII/SI), bilateral substantia nigra, red nucleus, pons, ipsilateral amygdala, contralateral parahippocampus (PH), hippocampus (Hipp), anterior cingulated cortex, middle cingulated cortex, ventral striatum, and anterior/posterior part of the cerebellum (figure 2 and table 1). Furthermore, the temporal signals of some active regions did not decrease after the end of acupuncture stimulation and exhibited a sustained effect during the whole session (figure 2). It was an intriguing finding that such noticeably similar spatial distributions were also found in R1 and R2 to those at BL; these periods were formerly treated as the same as the BL, according to the conventional theory of the ‘on–off’ experimental design (figure 2 and table 1). Direct comparisons of brain activity in the occipital cortex showed the characteristic of time variation during the sequential epochs; the brain responses in the first and second epochs (S1 and R1) were predominantly negative, but positive activation was seen during the following two epochs (S2 and R2) (figure 3 and table 1). In the present findings, the general content of the temporal patterns of the sequential epochs did not follow the ‘on–off’ paradigm that the experimental design suggested; instead, they remained similar to each other.

Table 1

Foci with significant alterations in the signal changes obtained by applying the multi-GLM and standard GLM (s-GLM) analysis at the GB37 (p<0.001)

Figure 2

Group results of brain activation patterns for GB37 and the non-meridian point from different-conditions analysis and standard General Linear Model (s-GLM) analysis. Statistical significance reached threshold at p<0.001 (uncorrected) with a minimum cluster size of 5 voxels. BL, baseline; R1 and R2, rest periods 1 and 2; S1 and S2, stimulation periods 1 and 2.

Figure 3

Activity patterns of the occipital cortex between the different-conditions analysis and standard General Linear Model (s-GLM) analysis. Regions of the vision cortex are shown in different epochs at the GB37 and non-meridian point. BL, baseline; R1 and R2, rest periods 1 and 2; S1 and S2, stimulation periods 1 and 2.

Acupuncture at NMP versus BL epochs

The spatial distribution and intensity of the signal change in the different epochs of acupuncture at NMP are shown in table 2 (p<0.001). It was obvious that the temporal feature of the blood oxygenation level dependent (BOLD) signal changes exhibited sustained effects in certain brain regions, such as the sensorimotor cortex, frontal cortex and cerebellum (figure 2). Compared with acupuncture at GB37, a small extent of spatial distribution signal changes was found with stimulation at NMP. In the occipital cortex, the BOLD signal decreased in the first and second epochs, but did not exhibit significant changes in the sequential epochs. It showed a different temporal pattern from stimulation at GB37.

Table 2

Foci with significant alterations in the signal changes obtained by applying the multi-GLM and standard GLM (s-GLM) analysis at the non-meridian point (p<0.001)

Standard GLM design matrix analysis

To test if the estimation of the standard contrast analysis ((S1+S2)−(BL+R1+R2)) was different from our results, we also analysed brain activation with standard GLM analysis. From the group results (p<0.001), we found that the intensity and extent of neural activity from stimulation at GB37 determined using standard GLM were much smaller. It only produced an increased signal in the superior temporal gyrus (STG), SI/SII and thalamus, and a decreased signal in the pons, PH/Hipp, putamen, occipital cortex and anterior/posterior part of the cerebellum (table 1); there were also more negatively activated regions. In contrast, the standard contrast analysis of NMP exhibited greater spatial distribution signal changes compared with our results, predominantly in the brainstem and subcortical and limbic systems (table 2). The fMRI signal deactivations observed in the occipital regions from the standard contrast analysis were similar to the results of Kong et al.11


In this study, we have provided a new angle on investigating the dynamics underlying the sustained effect of acupuncture. We separated the experimental design into a series of conditions and attempted to detect systematic temporal patterns following acupuncture stimulation at a vision-related acupuncture point, GB37, and an NMP. In comparison with standard GLM analysis, our analytical procedure was more sensitive in detecting brain responses from the sustained effect of acupuncture. Furthermore, we found distinct temporal characteristics of the BOLD signal in the visual cortex between GB37 and NMP.

We found that the sustained neural signals of the limbic–paralimbic–neocortical system increased in the four epochs (table 1), which has been reported in several previous acupuncture studies.7 ,8 However, the standard GLM analysis showed a more extensive spatial distribution signal decrease in the limbic–cerebellar regions (such as the occipital cortex, pons, PH/Hipp, putamen and cerebellum), but with a smaller signal increase (such as in the STG, SII and thalamus) (figure 2 and table 1). Furthermore, compared with our results, the sign of the fMRI signal was reversed in some of the brain regions (pons and PH/Hipp) in the standard GLM analysis. These paradoxical results are difficult to explain, but they may be related to the sustained effect of acupuncture. As far as we know, most acupuncture studies used the well-known logical ‘cognitive subtraction’ model, which compares the different brain responses between a range of needling conditions and the BL, which has a premise that the BOLD signal evoked by the stimulation condition should follow the ‘on–off’ pattern depicted by the experimental design.11 ,15 Neter et al have pointed out that the GLM has the potential disadvantage of being misleading with regard to state-related activity that does not conform to the ‘on–off’ specifications set. In that case, the GLM would reduce sensitivity and lead to incorrect inferences and limitations in reproducibility.18–20 We found that the neural response evoked by acupuncture did not turn on and off rapidly but lasted longer, violating the basic assumption of standard GLM analysis. Consequently, because of the sustained effect of acupuncture, contradictory and uncertain results may be obtained when using standard GLM analysis for ‘on–off’ design acupuncture studies.11 ,12 Clearly, these erratic statistical results cannot be used to reveal the actual role of acupuncture.

Previous studies did not find a significant difference in the occipital cortex between the vision-related acupuncture point and the NMP.11 ,12 As shown in the standard GLM analysis (figure 3), signal decreases were found at both GB37 and NMP, and the activations exhibited a similar spatial distribution pattern. This result was similar to that of Kong et al.11 The standard GLM analysis only reflected a ‘snapshot’ of the regional brain activity in response to acupuncture stimulus. In other words, it emphasised the spatial distribution of the brain areas where BOLD signals were altered in response to a task or condition, but was limited in the detection of the sustained effect of acupuncture. As shown in our different-conditions results (figure 3), the BOLD signal in the occipital cortex decreased in a similar manner to that in the first and second epochs of stimulation at both GB37 and NMP; the sign of the signal changes was reversed and showed an increasing pattern in S2 and R2 at GB37 (table 1 and figure 3), which was not found in NMP.

Kong et al11 proposed that the BOLD signal decreases in the occipital cortex were due to selective attention in acupuncture stimulation, which decreased the activity for other sensory modalities. We do not believe that such an explanation can fully account for the findings observed in acupuncture at the vision-related acupuncture point. As discussed above, the BOLD signal in the vision cortex responded to the initial administration and showed no significant activity in the following epochs in the NMP (table 2 and figure 3). However, the signal was transformed from a decreasing pattern to an increasing one, as shown in the subsequent epochs at GB37 (table 1 and figure 3). As the same somatosensory manipulation was performed at both GB37 and NMP, it was not surprising that similar decreased signal responses were primarily located in the occipital cortex during the first epoch, which may be involved in nociceptive processing and pain perception. After the ‘resting’ epoch (R1), subjects may adapt to the change, so that the neural signal shows a smaller decrease in the occipital cortex (figure 3). In the following epochs (S2 and R2), the temporal characteristics of the BOLD signal evoked by acupuncture were significantly different between GB37 and NMP. The neural signal of the occipital cortex exhibited a more complex pattern at GB37. We suggest that acupuncture at the vision-related acupuncture point may exert a direct effect on the vision-related cortex to modulate its activity.

Several issues in this study need to be addressed further. First, it is only a preliminary study into acupuncture at vision-related acupuncture points. The relationship between the limbic/paralimbic areas and the visual cortex when acupuncture is performed at vision-related acupuncture points could not be determined. Comprehensive investigation of brain responses is needed for a more general, integrated understanding of the mechanisms underlying diseased mental states for acupuncture. Second, the different time courses in the different brain regions may reflect how acupuncture is conducted and mediated through the central nervous system. Hence, neural responses during the sustained phase of acupuncture need to be further studied to obtain its detailed temporal profile, considering how multiple levels of dynamic activity in concert cause the processing of acupuncture.

In conclusion, this study has focused on the temporal pattern of signal changes following acupuncture at a vision-related acupuncture point and an NMP. Our different-conditions analysis had advantages in discovering activation patterns related to a relatively unconstrained psychological model; thus these results may reveal the true neural response underlying the acupuncture stimulus. Following acupuncture at the vision-related acupuncture point and the NMP, their temporal characteristics of the BOLD signal exhibited major differences. We hope our findings may provide deeper insight into the neural architecture of acupuncture at a vision-related acupuncture point.

Summary points

  • By TCM theory, GB37 relates to the visual cortex though previous studies have not proved this.

  • We used a novel analysis of fMRI by including the rest periods.

  • We showed an effect of GB37 compared with a non-point.


View Abstract


  • Contributors JL: conception and design, drafting the article. JN, SX and GL: analysis of data. JL, WQ and JT: final approval of the version to be published.

  • Funding This study was supported by the Project for the National Key Basic Research and Development Program (973) under Grant Nos 2012CB518501 and 2011CB707702, the National Natural Science Foundation of China under Grant Nos 30930112, 30970774, 81000640, 81000641, 81101036, 81101108, 31150110171, 30901900, 81271644, 31200837 and K5051210003, and the Fundamental Research Funds for the Central Universities.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval All research procedures were approved by the West China Hospital Subcommittee on Human Studies and were conducted in accordance with the Declaration of Helsinki.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.