What Is the de-qi-Related Pattern of BOLD Responses? A Review of Acupuncture Studies in fMRI

de-qi, comprising mostly subjective sensations during acupuncture, is traditionally considered as a very important component for the possible therapeutic effects of acupuncture. However, the neural correlates of de-qi are still unclear. In this paper, we reviewed previous fMRI studies from the viewpoint of the neural responses of de-qi. We searched on Pubmed and identified 111 papers. Fourteen studies distinguishing de-qi and sharp pain and eight studies with the mixed sensations were included in further discussions. We found that the blood oxygenation level-dependent (BOLD) responses associated with de-qi were activation dominated, mainly around cortical areas relevant to the processing of somatosensory or pain signals. More intense and extensive activations were shown for the mixed sensations. Specific activations of sharp pain were also shown. Similar BOLD response patterns between de-qi evoked by acupuncture stimulation and de-qi-like sensations evoked by deep pain stimulation were shown. We reckon that a standardized method of qualification and quantification of de-qi, deeper understanding of grouping strategy of de-qi and sharp pain, and making deep pain stimulation as a control, as well as a series of improvements in the statistical method, are crucial factors for revealing the neural correlates of de-qi and neural mechanisms of acupuncture.

Many researchers have begun to qualify and quantify sensations of de-qi [15][16][17][18][19][20][21][22][23][24][25][26][27], because they argued that the lack of a competent measure of de-qi affected the association between de-qi and the clinical outcomes [1,18,19]. Most studies reported that de-qi sensations involved numbness, heaviness, aching, dull pain, and tingling [1,16,18,19,23,25,27]. Other sensations such as fullness/distention, soreness, and pressure were also included in part of the studies [1,16,19,27]. The dimensions of sensations of de-qi were largely different, which ranged from four [19] to seventeen [25]. Although there have not a consensus that which sensations were involved in de-qi, researchers focused on qualifying and quantifying sensations of de-qi, and they agreed that deqi includes specific sensations for acupuncture stimulation [1,16,18,27]. Besides, another consensus was reached that as a conventional component of the needling sensations presented during acupuncture manipulation, sharp pain is noxious and is what acupuncturists try to avoid during needle manipulation [1,16,18,25]. It is often excluded from the 2 Evidence-Based Complementary and Alternative Medicine components of de-qi and regarded as the sensation that is irrelevant to the acupuncture effect [1,16,18,25].
With the aid of functional magnetic resonance imaging (fMRI) techniques, it is possible to explore the neural responses of de-qi and further interpret the neural mechanisms of acupuncture. More than one hundred studies have been published to explore the neurobiological mechanisms of acupuncture in the past few decades [28][29][30][31][32][33][34][35][36]. However, few of them paid close attention to the neural correlation of de-qi. Therefore, several fundamental issues about the central responses of de-qi remain open to debate. First of all, although researchers agreed that de-qi has specific sensations for acupuncture stimulation [16], the sensory quality of de-qi was similar to that of deep pain constructed by intramuscular injections of hypertonic saline, which mainly included tenderness, heaviness, aching, cramping, throbbing, and gnawing [25,37]. Thus, the quality of the sensation of deqi was likely nonspecific [38,39]. Up to now, however, direct comparisons of blood oxygenation level-dependent (BOLD) responses between de-qi sensations evoked by acupuncture stimulation and de-qi-like sensations evoked by deep pain stimulation have not yet been investigated. Only one recent study involved the indirect comparisons of these two stimulations, which had a similar sensory quality, in the discussion [40]. Secondly, in the domain of pain studies, it is generally known that differences in the quality and origin between sharp pain and deep pain are remarkable [41]. Abundant evidence from animal investigations [42,43], clinical data [44][45][46], and human neuroimage studies [37,[47][48][49][50][51][52] indicated different brain processing pathways for acute superficial pain (sharp pain) and deep pain. The sharp pain following acupuncture stimulation is sometimes perceived by subjects. However, most previous acupuncture fMRI studies did not explicitly distinguish the needle sensations into de-qi and sharp pain. To make matters worse, significant incompatibilities were shown across the results of several studies which excluded subjects who experienced sharp pain [22,23,25] or divided subjects into two groups according to whether subjects experienced sharp pain during acupuncture manipulation [33,[53][54][55].
In this paper, we would like to summarize the results of fMRI-based acupuncture studies from the perspective of the general pattern of central BOLD responses of de-qi. We aim to organize the evidence about three fundamental questions. First, what is the pattern of central responses of de-qi evoked by during acupuncture stimulation? Second, how are the distinct patterns of central BOLD responses associated with de-qi and sharp pain? Thirdly, are the patterns of central BOLD responses associated with de-qi specific from those of deep pain in pain studies? In addition, we hope to offer several suggestions for future studies of the neural correlates of de-qi evoked by acupuncture stimulation.

Identification of the Relevant Literature
The acupuncture studies involved in this paper were first identified by searching on Pubmed using the key words "acupuncture" or "electroacupuncture" in the title and "fMRI" or "functional magnetic resonance imaging" in the title/abstract (four combinations of key words: "acupuncture" and "fMRI, " "acupuncture" and "functional magnetic resonance imaging, " "electroacupuncture" and "fMRI, " and "electroacupuncture" and "functional magnetic resonance imaging"). About 111 original studies published in peerreviewed journals in English were included in the search (two studies in the Evidence-Based Complementary and Alternative Medicine with the state of "in press" were also involved). Firstly, 17 review articles were excluded. Secondly, 23 studies which did not evaluate and record subjects' needling sensations were discarded from further discussions. Thirdly, 13 studies that were based on data-driven methods were excluded. Fourthly, 4 studies that aimed to explore the continuous effect of acupuncture were also excluded. Fifthly, other 15 studies were also not included because of the lack of de-qi-related BOLD responses (2 were behavioral studies, 2 were magnetoencephalogram-based studies, 4 studies were associated with the therapeutic/placebo effect of acupuncture, 1 study was laser-acupuncture-based, and 5 studies lacked the whole brain's one-sample t-test results). In the remaining 39 studies, 17 studies asked or evaluated the subjects' de-qi but did not mention whether subjects underwent sharp pain [36,38,[56][57][58][59][60][61][62][63][64][65][66][67][68][69][70]. Eight studies which asked or evaluated the subjects' de-qi and sharp pain but did not distinguish de-qi and sharp pain in the fMRI analysis [53,[71][72][73][74][75][76][77] were named as "mixed pattern studies" and summarized in Table 3. Fourteen studies that excluded subjects who experienced sharp pain for the fMRI data analysis [32,34,[78][79][80][81][82][83][84][85] or divided subjects into two groups according to whether subjects experienced sharp pain during acupuncture manipulation [33,40,54,55] were named as "pure de-qi pattern studies" and are summarized in Table 2. The experimental details and the methodological details of the 14 de-qi-related studies are shown in Table 1 and Table 5, respectively.

General Observations
Reviewing the studies of acupuncture in fMRI, we found that de-qi and sharp pain were not distinguished in most studies (97 out of 111, 87%). Therefore, limited resources could be used for summarizing the de-qi BOLD response patterns. We suggest that it is necessary to differentiate de-qi and sharp pain in further fMRI acupuncture studies. In the following part, we will summarize the BOLD response patterns of pure de-qi, the BOLD response patterns of mixed sensations, the similarities/differences of BOLD response patterns between pure de-qi and mixed sensations, similarities/differences of BOLD response patterns between pure de-qi and de-qi-like sensations evoked by deep pain stimulation, and the comparison of de-qi-related regions with regions generally activated in acupuncture fMRI studies, which were summarized in other reviews [39,86].

The Similarities/Differences of BOLD Response Patterns between Pure de-qi and Mixed Sensations.
In the 14 "pure deqi pattern studies, " only two studies performed the betweengroup analysis that statistically compared the de-qi group and mixed group (de-qi + sharp pain) based on a twosample unpaired t-test [40,55]. In Hui et al., 2009, significant differences in the posterior cingulate/precuneus, pregenual cingulate/frontal pole, subgenual area, orbitofrontal cortex, temporal pole, amygdala, hippocampus, parahippocampus, hypothalamus, cerebellar vermis (lobules VII and VIII), SII, and anterior middle cingulate were shown between the de-qi group and mixed group. Specifically, activity of BOLD signals was significantly less activated or more deactivated in the deqi group than that in the mixed group. In Sun et al., 2012, significantly stronger activations of the mixed group were presented in the bilateral putamen, the bilateral thalamus, and the bilateral cerebellum (CrusI and CrusII) [40]. In these regions, BOLD signals of the de-qi group were barely changed, while significant BOLD responses were shown in the mixed group.

The Similarities/Differences of BOLD Response Patterns between Pure de-qi and de-qi-Like Sensations Evoked by Deep
Pain Stimulation. In contrast to superficial (cutaneous) pain, deep pain (originating from muscle, joints or viscera) is dull, diffuse, and difficult to localize [37,50]. In Henderson et al., 2006, the researchers used intramuscular injections of hypertonic saline to construct a deep pain model [37]. The subjective sensations under this deep pain mainly included tenderness, heaviness, aching, cramping, throbbing, and gnawing, which were similar to that of de-qi evoked by acupuncture stimulation. Table 4 summarizes the activations and deactivations from the deep pain studies [37,48,50,87,88]. Common activations (frequency > 2 or 40%) were seen in the anterior cingulate, posterior cingulate, SI, SII, MI, insula, cerebellar cortices, inferior parietal, claustrum, and thalamus. A common deactivation was in the perigenual cingulate. Most of the activations for deep pain were consistent with that of de-qi evoked by acupuncture stimulation, except for the anterior cingulate and posterior cingulate, which were commonly deactivated during de-qi. Besides, the amygdala, hippocampus, parahippocampus, hypothalamus, and temporal pole were deactivated for de-qi but not for deep pain.

The Comparison of de-qi-Related Regions with Regions
Generally Activated in Acupuncture fMRI Studies. Huang et al., 2012, summarized the general BOLD responses in acupuncture fMRI studies [86]. The supramarginal gyrus/insula/SII, presupplementary motor area/middle cingulate, thalamus, and precentral gyrus were the most commonly found activations. Besides, the anterior cingulate, subgenual cortex, amygdala/hippocampal formation, ventromedial prefrontal cortex, and posterior cingulate were

Discussion
This paper summarized the results of fMRI-based acupuncture studies from the perspective of the general pattern of central BOLD responses of de-qi. Three fundamental issues are discussed later. In addition, several suggestions for future studies of the neural correlates of de-qi evoked by acupuncture stimulation were offered.

Activation-Dominated BOLD Responses Associated with
de-qi during Acupuncture Stimulation. The most robust BOLD responses evoked by acupuncture stimulation were in the SII, insula, SI, cerebellum, thalamus, MI, STG, visual cortices, IFG, SMA/pre-SMA, basal ganglia, MTG, and ACC. Under the methodological inclusion criteria [39], only the deactivation of the occipital cortex was shown. For the deqi-related BOLD responses, a similar activation pattern was shown, including in the SI, SII, thalamus, MI, cerebellum, insula, inferior parietal lobe, and anterior MCC. Besides, several deactivations were seemingly the common pattern of de-qi. However, we argued that the reliability of these deactivations was poor. We inferred that several possible reasons might contribute to these deactivation patterns. Firstly, the average of the repeated runs for each subject was applied. Repetition may itself alter the distribution of the activated regions due to the influence of memory and expectation [48]. Secondly, global normalization, a questionable data processing step adopted in several fMRI-based acupuncture studies, can introduce an artificially negative relationship with the task [83,[89][90][91][92]. Particularly, our recent study clarified that for the fMRI-based acupuncture data, global normalization significantly changed the activation-dominated results to deactivation-dominated ones [83]. Therefore, we suggested that BOLD responses associated with de-qi during acupuncture stimulation should be activation dominated.

Distinct Patterns of Central BOLD Responses Associated with de-qi and Sharp
Pain. Since subjects with pure sharp pain during acupuncture stimulation were difficult to obtain, the differences between de-qi and sharp pain evoked by acupuncture stimulation could not be presented directly. Therefore, we had to infer their differences between the de-qi and mixed sensations. However, studies which were statistically compared with the BOLD responses between de-qi and mixed sensations were also a scarcity [40,55]. Our recent study focused on these issues and indicated that both the quantitative and qualitative differences of BOLD responses between de-qi and mixed sensations evoked by acupuncture stimulation were distinct [40]. We inferred that the pattern of BOLD responses for sharp pain might be partly separated from that of de-qi in the spatial distribution.
The subjects with sharp pain should be excluded from those with only de-qi when exploring the central BOLD responses during acupuncture stimulation [40].   [76,77] sensations. Due to individual variability of brain morphology and differences in experimental design, the central patterns of activation during deep pain and acupuncture stimulation were difficult to compare between studies. Therefore, we suggested that a specific central effect of de-qi during acupuncture stimulation might be illustrated after comparing it directly to deep pain stimulation.

Several Suggestions for Future Studies of the Neural
Correlates of de-qi Evoked by Acupuncture Stimulation. First of all, for better understanding what de-qi is and exploring its possible role in acupuncture, many researchers have been engaged in qualifying and quantifying de-qi [1,[15][16][17][18][19][20][21][22][23][24][25][26][27]. However, in the aforesaid fMRI-based acupuncture studies about de-qi, different sensation questionnaires were used to quantify subjects' acupuncture sensations (see Table 1 for details). This may lead to two results. One is that the kinds of acupuncture sensations recorded by different studies are partially different. The other is that the same score of an acupuncture sensation between different studies refers to a different subjective intensity experienced by subjects due to the various definitions of the same score between different studies. Thus, acupuncture sensations recorded by different studies are inappropriate to compare. We suggest that further studies should pay more attention to the quantification questionnaire of de-qi and try to apply a standardized quantification method to better control the experimental conditions [38,76]. Secondly, another important discrepancy among studies is the definition of grouping. In Hui et al. and Asghar et al. [33,53], they both grouped subjects into two groups (the deqi group and mixed group). But the definition of the "deqi group" was totally different in the two studies. In the studies of Hui, the de-qi group referred to subjects who only experienced de-qi during acupuncture, while in Asghar et al., the de-qi group referred to subjects whose de-qi scores were greater than sharp pain, which were similar to the Sun et al. 's sharp pain group (de-qi sensation mixed with sharp pain). Sun et al. 's study indicated that even a little sharp pain mixed in, both the quantitative and qualitative differences of BOLD responses between de-qi and mixed sensations evoked by acupuncture stimulation were significant [40]. Therefore, we proposed that subjects with sharp pain should be separated from those with only de-qi when exploring the central BOLD responses during acupuncture stimulation. Further comparisons between de-qi and sharp pain should be accumulated. Besides, the deep pain stimulations, which had a similar sensory quality to de-qi evoked by acupuncture stimulation, could serve as a valid control for acupuncture.
At last, methodological problems and differences may partly contribute to the discrepancies among studies which were reviewed here. Beissner and Henke found that most acupuncture studies lacked adequate strict statistical methods and suggested that several methodological problems should be solved to facilitate acupuncture studies [93]. The improvements for obtaining more reliable results included a larger sample size, corrected threshold, and a more robust method of statistical inference [94]. With regards to the studies reviewed in this paper, an inappropriate group analysis method, a too small sample size, and the liberal threshold applied by the studies could be the most important problems (see Table 3 for details). For instance, first, Hui et al. [33] applied the fixed-effect model in the group analysis, which could produce positive group results even when only a single subject has strong activations [93]. Second, Hui et al. recruited 13 subjects in their study, and only two subjects experienced the mixed sensation [54]. Hui et al. recruited 15 subjects and a mixed group with 4 subjects [33]. Such a sample size (2 subjects and 4 subjects) is too small to obtain group results with sufficient statistical power. Third, several studies adopted global normalization [33,53,55], which is a questionable data processing step and can introduce an artificially negative relationship with the task [89][90][91][92][95][96][97][98]. Particularly, our recent study clarified that for the fMRI-based acupuncture data, global normalization significantly changed the activation-dominated results to deactivation-dominated ones [83]. As a whole, solving these crucial methodological problems could greatly help studies on neural correlates of de-qi to obtain more repeatable and reliable results. Thus, we suggest that future studies should focus on improvements in their methodological problems which we believe could shed light on studies of neural correlates of de-qi.

Synopsis and Possible Solutions
de-qi, comprising mostly of subjective sensations during acupuncture, is traditionally considered as a very important component of the possible therapeutic effects of acupuncture. Thus, it is of great importance to reveal the neural correlates of de-qi which may benefit the understanding of neural mechanisms of acupuncture. However, previous studies in fMRI involving the BOLD response of de-qi were limited and did not reach consistent conclusions on the neural response pattern of de-qi in the brain. In this paper, we summarized previous fMRI studies on the neural responses of de-qi and answered three fundamental questions. For the first question concrning the pattern of the central responses of de-qi evoked by acupuncture stimulation, our answer was that the BOLD responses associated with de-qi during acupuncture stimulation were activation dominated, mainly around cortical areas relevant to the processing of somatosensory or pain signals. For the second question on how the distinct patterns of central BOLD responses are associated with de-qi and sharp pain, our answer was that more intensive and extensive activations were shown for the mixed sensations evoked by acupuncture stimulation. Specific activations of sharp pain 8 Evidence-Based Complementary and Alternative Medicine were also shown. For the third question asking if the patterns of central BOLD responses were associated with de-qi specifically from those of deep pain in pain studies, our answer was that similar BOLD response patterns between de-qi evoked by acupuncture stimulation and de-qi-like sensations evoked by deep pain stimulation were shown. Finally, we reckon that a standardized method of qualification and quantification of de-qi, a deeper understanding of grouping strategy of de-qi and sharp pain, and marking the deep pain stimulation as a control, as well as a series of improvements in the statistical method, are crucial factors for revealing the neural correlates of de-qi and neural mechanisms of acupuncture.