2.1.

Pain Threshold Measurements

To determine the amount of nociceptive stimuli to be delivered during the experiment, we assessed individual pain thresholds obtained from each thumb of each subject. We applied a quantitative sensory testing (QST) method by using electronic von Frey (eVF) anesthesiometer (Ugo Basile Co., Italy), a precise and accurate method for pain threshold application.^43,44 In the QST method with eVF, a continuously variable force transducer using a rigid filament directly contacted the thumb, and an increased mechanical pressure was applied five times with increasing force of 0.1 g steps by using a single nonflexible filament that has a diameter of 0.5 mm. Stimulus was delivered perpendicularly to the interphalangeal joint of thumbs in every participant. For every five application, while stimulus was applied in a linearly increasing intensity trend, participants gave a verbal sign when they felt an unpleasant feeling caused by sharp painful sensation.⁴⁴ This pressure value was noted as the least pressure intensity in which participants perceived pain.⁴⁵ Between every measurement, there was a 20-s interval to prevent habituation. Then, these five measurements were averaged and recorded as individual pressure pain threshold.

2.2.

Channel Positioning and the fNIRS System

In our study, fNIRS was performed at Ankara University Brain Research Center with Hitachi ETG 4000 continuous wave fNIRS system (Hitachi Co., Japan). Optical near-infrared lights with two different wavelengths (695 and 830 nm) were sent to the head surface via a source optode and captured back by a detector optode. Sampling frequency was 10 Hz. Optical light signals were converted to $\mathrm{\Delta }{\mathrm{HbO}}_2$ and $\mathrm{\Delta }\mathrm{Hb}$ by using modified Beer–Lambert law.⁴⁶ We used the $2\times 3\times 3$ optode configuration including 24 channels.

We utilized the electroencephalography (EEG) 10-20 electrode positioning system to position the fNIRS source and detectors onto the head surface.⁴⁷ In this system, half of the distance from nasion to inion corresponds to the channel $\mathrm{C}z$. After defining the position of $\mathrm{C}z$, we set the $3\times 3$ probe holders for each hemisphere over the line passing through both ears. Then, we defined positions of C3 and C4 by measuring the distance between both tragi, and 30% of this value gave us the position of C3 from left tragus and C4 from the right tragus, which corresponded to the left and right SI, respectively, as described previously.^48,49 Source numbers 18 and 13 were placed onto the points C3 and C4 in both hemispheres, respectively (Fig. 1). After the probe holder placement, we marked optode positions by using a three-dimensional digitizer (Polhemus Co., Vermont) to determine the exact position of each channel. Then, we obtained the position file to use it for registering to the Montreal Neurological Institute (MNI) space to determine the landmarks that correspond to every channel position by using NIRS analysis package.⁵⁰ Then, we averaged coordinate values of all participants.⁵¹ To obtain brain regions corresponding to MNI coordinates, we used LONI Probabilistic Brain Atlas.⁵² Channel configuration on head is shown in Fig. 1. Channel numbers and corresponding cortical regions with MNI coordinates are shown in Table 1.

Fig. 1

Channel and optode configuration of $2\times 3\times 3$ probe setting. Locations that are represented as squares are channels. White circles that include numbers in blue are detectors. Pink circles that include numbers in black are sources. R, right and L, left.

Table 1

Channel numbers and average coordinate positions with corresponding cortical structures registered onto MNI space after using LPBA 40 cortical atlas. Probability values were obtained from LPBA 40 cortical atlas. (L, left and R, right).

2.3.

Experiment 1: Median Nerve Stimulation with TENS

We performed two different experiments with left and right thumbs in separate runs. Our first experiment was median nerve stimulation by using TENS. For TENS application, we used the Intelect TENS device (Chattanooga Co., Tennessee) with a pulse width of $60\mu \mathrm{s}$ and 115 Hz frequency via square Dura-Stick Plus Self-Adhesive electrodes with 5-cm size. A predetermined amplitude threshold ($\approx 30\mathrm{mA}$) was applied to each participant, which caused a tingling effect on his/her distal forearms/hands. Experimental paradigm included three consequent blocks [Fig. 2(a)]. After a 20-s “rest” condition, 20-s “TENS” was applied and consequent every 20-s “TENS” condition, 20-s “rest” condition was applied.

Fig. 2

Experimental design. (a) Median nerve stimulation and (b) painful stimulation with TENS I, initial baseline; A, pain + placebo TENS; B, pain + TENS; R, rest; and T, TENS.

2.4.

Experiment 2: Painful Stimulation with TENS

We performed two different experiments with left and right thumbs in separate runs. A single experiment consisted of six pseudorandom blocks with three conditions: “pain + placebo TENS,” “pain + TENS,” and “rest.” The experimental paradigm is shown in Fig. 2(b). “Pain + placebo TENS” and “pain + TENS” conditions were both applied both for 20 s. “Rest” period was applied for 40 s. The experimental run began with a 50-s “initial rest” period. In “pain + placebo TENS” and “pain + TENS” conditions, we applied nociceptive stimuli to the thumb. In the “pain + TENS” condition, the TENS device was turned on during the administration of nociceptive stimuli by applying the conventional TENS parameters mentioned above. During the “pain + placebo TENS” condition, TENS pads were still on the forearms of participants, but the device was not operating. The participants were blind about the timing of the TENS application. The TENS device was connected during the entire experiment and the participants were unaware of the blocks for which the TENS was turned on. TENS application was shown in Fig. 3.

Fig. 3

TENS application to right forearm of a participant.

While applying nociceptive stimuli in both conditions, eVF was pressed four times onto the interphalangeal joint of the thumb. Each time eVF is used, pain delivery lasted 4-s, at the individual pressure pain threshold value, which was previously measured before the experiment. Interstimulus interval (ISI) was 1 s. Hence, each repetition of pain plus ISI took 5 s. Each pain block consisted of four repetitions of pain delivery, summing up to a block of 20 s.

Before the experiment, each participant was informed to estimate pain ratings within a scale of 0 (no pain) to 100 (extreme pain). After every block, participants were stipulated to verbally quantify their own subjective pain experience according to this scale during resting periods. During the experimental design, we targeted 70 out of 100 as target pain sensation level for both groups. To reach this level, we applied different nociceptive stimulations for each participant, based on the predetermined individual pain threshold. This type of experiment was previously applied in several studies, referred as “subjective pain control condition.”^27,33

2.5.

Data Processing

We used MATLAB for preprocessing (MathWorks, Inc., Natick, Massachusetts). Preprocessing pipeline included baseline correction, detrending to eliminate low frequency drift, filtering for removal of systemic artifacts, and averaging of blocks in the same condition. Initially, baseline correction was applied to detect the activity trend by observing negative peaks in $\mathrm{\Delta }\mathrm{Hb}$ and positive peaks in $\mathrm{\Delta }{\mathrm{HbO}}_2$ during active blocks. Wavelet-based detrending filter was used to remove this drift.⁵³ Finally, a low-pass filter with a cut-off at 0.05 Hz was applied to remove high-frequency noise that possibly include Mayer waves (0.05 to 0.2 Hz), cardiac pulsations (0.6 to 2.0 Hz), and respiratory artefact (0.15 to 0.4 Hz).⁵⁰ We applied amplitude thresholding method⁵⁰ either for $\mathrm{\Delta }{\mathrm{HbO}}_2$ and $\mathrm{\Delta }\mathrm{Hb}$ data in order to remove out motion artifacts. If amplitude exceeds threshold value for $\mathrm{\Delta }\mathrm{Hb}$ ($-0.5\mu \mathrm{M}$) and $\mathrm{\Delta }{\mathrm{HbO}}_2$ ($0.5\mu \mathrm{M}$), amplitude thresholding method was applied to related $\mathrm{\Delta }{\mathrm{HbO}}_2$ and $\mathrm{\Delta }\mathrm{Hb}$ channel.

2.6.

Statistical Analysis of Clinical Data

Statistical analyses were performed with SPSS 20.0. We used Shapiro-Wilk test for normality of variables. We applied a $2\times 2$ (group × hand) ANOVA to compare pain thresholds between the groups. A $2\times 2$ (group × condition) between/within subject design of ANOVA was applied for pain ratings. Group was between FM patients and HC and condition was within “pain + placebo TENS” and “pain + TENS.” Statistics for pain ratings were run separately for each hand. Pearson correlation coefficient was used to associate fNIRS and behavioral data. To compare gender distribution, we used chi-square test. A $p$ $\text{value}\le 0.05$ was considered for statistical significance.

2.7.

Brain Activity Data Analysis

After preprocessing, we obtained mean $\mathrm{\Delta }{\mathrm{HbO}}_2$ value for each condition. We performed the following analyses for each channel separately. In the first experiment, we ran $2\times 2$ (group × hand) repeated measures ANOVA by using block averaged mean $\mathrm{\Delta }{\mathrm{HbO}}_2$ of “TENS” condition. In the second experiment, we ran $2\times 2$ (group × condition) repeated measures ANOVA using mean $\mathrm{\Delta }{\mathrm{HbO}}_2$ values of “pain + placebo TENS” and “pain + TENS” conditions. For each channel, post hoc analyses were performed separately for significant factors/interactions by using Bonferroni correction similarly to the methodology used in several studies.^20,54 A $p$ $\text{value}\le 0.05$ was considered for statistical significance.

3.1.

Brain Activation Results: Quality Control

Before statistical analysis, data from each channel and each participant were checked for quality control. One healthy subject was discarded due to no hemodynamic response based on correlation analysis with the applied task waveform. All results for median nerve stimulation are shown in Fig. 4, and for LHS and RHS of painful stimulation with TENS in Figs. 5 and 6, respectively. Anatomical locations and all statistical experimental results for all channels with their corresponding regions are shown in Table 3.

Fig. 4

Channels that show group main effect in median nerve stimulation with TENS marked on MNI space. (a) Left hemisphere and (b) right hemisphere (L, left and R, right). Blue dots represent channels and red dots represent the sources which are considered as C3 and C4 in EEG 10-20 system (13 and 18). The difference found in group main effect was circled in red. FM, fibromyalgia patients, and HC, healthy controls.

Fig. 5

Channels that show group and condition main difference for LHS. The difference found in group main effect and condition main effect was circled in red and yellow, respectively. (a) Left hemisphere and (b) right hemisphere. FM, fibromyalgia patients, and HC, healthy controls.

Fig. 6

Channels that show group main difference and interaction between group and condition for RHS. The difference found in group main effect and interaction between group and condition was circled in red and white, respectively. (a) Left hemisphere and (b) right hemisphere. FM, fibromyalgia patients, and HC, healthy controls.

Table 3

Results of median nerve stimulation with TENS and painful stimuli with TENS. Post hoc comparison results explained below the p value. FM, fibromyalgia HC, healthy controls; n.s., not significant; LHS, left-hand stimulation; and RHS, right-hand stimulation.

3.2.

Correlations Between fNIRS and Behavioral Data

In RHS, among these channels, significant negative correlations were observed between dominant hand pain threshold and dominant hand mean $\mathrm{\Delta }{\mathrm{HbO}}_2$ of “pain + TENS” condition in the ipsilateral supramarginal (channel 16; $r=-0.33$, $p=0.05$) and between dominant hand mean $\mathrm{\Delta }{\mathrm{HbO}}_2$ of “pain + TENS” condition in the ipsilateral SI (channel 19; $r=-0.45$, $p=0.006$) gyri. This indicates that the subjects with a higher pain threshold had lower mean brain activity in the pain + TENS condition compared to the other subjects. All these correlation results are shown in Fig. 7.

Fig. 7

Correlation plots of (a) right-hand pain threshold and block averaged mean $\mathrm{\Delta }{\mathrm{HbO}}_2$ of right-hand stimulation of “pain + TENS” for channel 16 and (b) right-hand pain threshold and block averaged mean $\mathrm{\Delta }{\mathrm{HbO}}_2$ of right-hand stimulation of “pain + TENS” for channel 19.

For LHS, we determined to focus on channels that had group or condition main effect (channels 5, 6, 7, 18, and 23). Among these channels, there was no significant correlation between pain ratings and pain thresholds.

4.1.

Left (Nondominant) Hand Stimulation

LHS data indicated that significant differences were found in only group and condition main effects. We observed group main effect in contralateral SMG and ipsilateral SI. Post hoc analysis revealed that FM patients showed higher brain activity than HC. On the other hand, condition main effect was observed in ipsilateral SI and SMG and contralateral MI. Post hoc analysis showed that “pain + TENS” condition showed decreased hemodynamic activity compared to “pain + placebo TENS” condition in these channels. These findings reflect that TENS does have an effect to reduce somatosensory brain activity in both FM and healthy groups. In support of our findings, recent studies on FM patients also demonstrated that TENS causes a significant pain reduction.^1,4,5,7 Activity reduction via TENS was found in right SMG located in IPL. Such a reduction in IPL was also reported in a recent study performed on healthy participants.²²

The TENS effect we observed in ipsilateral SMG (BA40) located in IPL is a replication with a previous fMRI study that showed the TENS effect in IPL while applying nociceptive stimuli in controls.²² IPL was found active when nociceptive stimulus was applied both in controls and FM patients.^27,30,33 However, the roots of pain perception with respect to IPL have not been clarified. IPL has an important role in attention to the noxious stimuli. Decrease of hemodynamic activity in “pain + TENS” condition might be related to a decrease in hypervigilance to nociceptive stimuli.

The LHS pain ratings showed that groups sensed nociceptive stimulation similarly because there was no difference in pain ratings. However, significant difference was found in condition main effect of the pain ratings. Post hoc analysis revealed that TENS efficiency was perceived by subjects in subjective pain ratings. This behavioral finding supports our condition main effect in hemodynamic activity results. We conclude that for the nondominant hand, TENS efficiency can be represented both in HC and FM patients in parietal and somatosensory cortices, which have important roles in pain perception.

4.2.

Right (Dominant) Hand Stimulation

For RHS, when compared with HC, increased hemodynamic activity was found in FM group in the ipsilateral SMG, which is one of the two main components of IPL. Several studies showed that in this area, higher hemodynamic activities were observed in FM patients compared to HC.^{27–29,33,57} This might be related with hypersensitivity to nociceptive stimuli. Furthermore, IPL was found active in pain processing when equal pain sensation experiment was conducted.^27,33 Activity differences of both groups in the ipsilateral SMG might be related to augmented pain processing among patients in this area.

On the other hand, we could not observe any difference for the condition main effect for RHS. This was obviously due to the significant interactions observed between group and condition in bilateral SI, MI, and SPG, and contralateral SMG. Post hoc results revealed that “pain + TENS” activity in FM patients was higher than the “pain + placebo TENS” condition, whereas in HC, a reverse pattern was observed. Several studies showed that brain activity to nociceptive stimulation was higher in FM patients than controls in SI, IPL (including AG),^27,33 and MI.³³ As superior parietal lobe is classified as sensory association cortex, an increase in hemodynamic activation in FM patients in this region is noteworthy.

In FM patients, contradictory to the decrease we expected with the use of TENS, the increase in hemodynamic response might reflect a malfunction of the gate control mechanism. Woolf et al. suggested that nonnociceptive stimulation causes allodynia and secondary hyperalgesia that might occur due to peripheral sensitization in individuals with central sensitization.⁴¹ According to Cook et al.,²⁸ central nervous system dysregulation was found independent of stimulus type in FM patients. FM patients may consider TENS as a nociceptive stimulation, which may have introduced an increase in hemodynamic activation. Such alteration in pain processing mechanisms due to changes in peripheral and central nervous systems was reported previously.⁵⁸ Increased activations of FM patients in “pain + TENS” condition within bilateral MI might also be due to attentional mechanisms triggered by TENS. However, this reasoning is somehow speculative as pain ratings did not change between the two conditions. On the other hand, the negative correlation between hemodynamic activity obtained by RHS and right-hand pain threshold in supramarginal and SPG indicate a clear association between the lower pain thresholds and higher “pain + TENS” activity. In addition, a negative correlation between “pain + placebo TENS” condition ratings and block averaged mean $\mathrm{\Delta }{c}_{\mathrm{HBO}2}$ was observed in left SPG (channel 3). In a recent study, negative correlation between painful stimulation and pain ratings was observed in FM patients before painful stimulation and subsequent pain in MI, SMA, and midcingulate cortex.³⁴ Our finding in SPG by using both groups (FM and HC) is complementary to this finding.

When results from the two experiments are merged for the RHS, we encountered a very important finding in bilateral SPG (channels 3 and 17). In these regions, while TENS activity in FM patients was found higher than HC in the median nerve stimulation experiment, RHS experiment showed that “pain + TENS” activity was found higher in FM patients than HC. Therefore, this activity increase in FM patients compared to HC during painful stimulation with TENS seems to be related to hypersensitivity of FM patients to TENS. Perhaps, this can be explained with allodynia in FM patients. On the contrary, the activity decrease in HC during painful stimulation with TENS compared to painful stimulation alone is an expected result, which can be explained the “gate control theory of pain.” Such a striking difference between groups is not present in the LHS experiments we performed. There is not enough knowledge in FM literature that explains such a dichotomy between the pain processing in right versus left hands.

Although hand dominancy is an active research area in pain perception, there is no common agreement about the effect of hand dominancy in this domain. Some studies show that pain perception is different between hands,^59–62 but others do not agree about this difference.^63–66 However, if there is a difference between hands, nondominant hand shows a higher sensitivity to nociceptive stimulus.^67–73 This conflict shows that while analyzing pain perception on either psychophysical studies or neuroimaging studies, hand dominancy effect should be considered as an effective factor. Moreover, some studies showed that pain threshold measured by a pressure algometer is greater in dominant hand than nondominant hand of healthy participants.^{59,60,68–71,38}

On the other hand, there are several neuroimaging studies that include painful stimulation to both hands.^23–26 These studies were performed by using noxious laser,^23,24 electrical stimulation,²⁵ and heat stimulation.²⁶ These studies showed that there is bilateral activation and contralateral bias in SI, SII, insula, and thalamus^24,26 and also motor output related structures such as putamen and cerebellum.²³ Also, these studies suggested that pain processing was strongly lateralized to the right hemisphere especially in MFG, ACC, inferior frontal gyrus, medial/superior frontal gyrus, and IPL.²⁵ Furthermore, a meta-analysis was performed by using activation likelihood estimate method.⁷⁴ In this meta-analysis, hemispheric lateralization of pain perception was analyzed by comparing two groups of studies on right- and left-hand stimulations. Results showed that without considering the left or right stimulation, insular cortex and right ACC showed the most significant probabilistic values, which are generally observed in several pain studies.¹³ For LHS, the likelihood of activation was found significant in right SI, MI, posterior parietal cortex, and superior frontal gyrus, and for RHS, left SI, ACC, MI, IPL, and MFG. The likelihood of activation in ipsilateral side was found significant in midbrain for LHS. For RHS, ACC, IPL, and MFG showed the significant likelihood activation.

A recent study conducted a similar study to ours (on feet rather than hands) in healthy participants reported an increase in “pain + TENS “activity compared to “pain only” condition in SI.²² This would have meant that dominant-side effect of TENS is similar in both patient and healthy groups, but our finding in the healthy group is reverse. A recent study in healthy subjects showed that TENS efficiency on pain site might be related to stimulation intensity.⁷⁵ While Choi et al.²² applied a mean of 16-mA TENS intensity, we applied about 30 mA. Hence, stimulus intensity might be a factor just as important as the stimulated side.

The RHS findings showed that a subjectively adjusted amount of stimulation based on pain thresholds caused similar pain ratings for both groups consistent with the previous studies.^27,33 Surprisingly, we found no difference between the ratings of the conditions. According to our results, none of the groups indicated differential pain perception during the “pain + TENS” condition, compared to the “pain + placebo TENS” condition. This indicates that the brain activity reduction in HC as well as brain activity increase in FM patients during “pain + TENS” condition was not carried on to conscious levels. Further studies are needed to isolate the underlying factors for this phenomenon. On the contrary, based on RHS mean ${\mathrm{HBO}}_2$ activity findings, there was a negative and moderate correlation between pain thresholds and brain activity, which indicates that increasing pain thresholds produce less brain activity showing pain processing efficiency [Figs. 7(a) and 7(b)].

4.3.

Limitations

While performing statistical analyses, we assumed that the channels are independent. This means that any probable cross-talk between the channels was disregarded. The perceived pain ratings between the conditions were not significant. Therefore, the neuroimaging findings presented here only indicate unconscious aspects of pain. Our findings are generalizable insofar as female patients are of concern.