Prolonged Continuous Theta Burst Stimulation Increases Motor Corticospinal Excitability and Intracortical Inhibition in Patients with Neuropathic pain: An Exploratory, Single-Blinded, Randomized Controlled Trial

Structured Abstract

Objectives

A new paradigm for Transcranial Magnetic Stimulation (TMS), referred to as prolonged continuous theta burst stimulation (pcTBS), has recently received attention in the literature because of its advantages over high frequency repetitive TMS (HF-rTMS). Clinical advantages include less time per intervention session and the effects appear to be more robust and reproducible than HF-rTMS to modulate cortical excitability. HF-rTMS targeted at the primary motor cortex (M1) has demonstrated analgesic effects in patients with neuropathic pain but their mechanisms of action are unclear and pcTBS has been studied in healthy subjects only. This study examined the neural mechanisms that have been proposed to play a role in explaining the effects of pcTBS targeted at the M1 and DLPFC brain regions in neuropathic pain (NP) patients with Type 2 diabetes.

Methods

Forty-two patients with painful diabetic neuropathy were randomized to receive a single session of pcTBS targeted at the left M1 or left DLPFC. pcTBS stimulation consisted of 1,200 pulses delivered in 1 minute and 44 seconds with a 35-45 minute gap between sham and active pcTBS stimulation. Both the activity of the descending pain system which was examined using conditioned pain modulation and the activity of the ascending pain system which was assessed using temporal summation of pain were recorded using a handheld pressure algometer by measuring pressure pain thresholds. The amplitude of the motor evoked potential (MEP) was used to measure motor corticospinal excitability and GABA activity was assessed using short (SICI) and long intracortical inhibition (LICI). All these measurements were performed at baseline and post-pcTBS stimulation.

Results

Following a single session of pcTBS targeted at M1 and DLPFC, there was no change in BPI-DN scores and on the activity of the descending (measured using conditioned pain modulation) and ascending pain systems (measured using TSP) compared to baseline but there was a significant improvement of >13% in perception of acute pain intensity, increased motor corticospinal excitability (measured using MEP amplitude) and intracortical inhibition (measured using SICI and LICI).

Conclusion

In patients with NP, a single session of pcTBS targeted at the M1 and DLPFC modulated the neurophysiological mechanisms related to motor corticospinal excitability and neurochemical mechanisms linked to GABA activity, but it did not modulate the activity of the ascending and descending endogenous modulatory systems. In addition, although BPI-DN scores did not change, there was a 13% improvement in self-reported perception of acute pain intensity.

Introduction

Neuropathic pain (NP) affects 7%–10% of the general population, and accounts for almost 20-25% of patients with chronic pain [9,111]. NP is caused by a lesion or disease affecting the somatosensory nervous system, has a considerable impact on quality of life, and is associated with substantial morbidity and high economic burden on the individual and society [2,22,103]. Prior work suggests that current treatments for NP are ineffective, including pharmacological treatments that are inadequate due to both poor efficacy and tolerability [2,10,111]. Despite the increasing focus on mechanism-based classification approaches, and the identification of disease-based phenotypes [1,43,69,109], the poor understanding of the pathophysiological mechanisms of NP has undermined efforts to effectively develop treatment strategies that target the underlying mechanism or cause(s). Thus, we examined the effectiveness of a non-pharmacological treatment, transcranial magnetic stimulation (TMS), on NP, and variables linked to the aforementioned mechanisms.

TMS has received much attention over the last two decades as a potential treatment for NP partially due to its non-invasive nature, excellent safety profile, and tolerability [70,80,86]. TMS is a form of brain stimulation that uses electromagnetic induction to excite or inhibit neural activity in a small volume of the brain [12,94]. High frequency repetitive TMS (HF-rTMS) to the primary motor cortex (M1) has been utilized to modulate corticospinal excitability, enhance intracortical inhibition [38,110] and induce analgesic effects in patients with NP, experimental pain, and chronic pain [30,58,96]. In addition, HF-rTMS at the dorsolateral prefrontal cortex (DLPFC) has been approved by the Food and Drug Administration for the treatment of neurological conditions and neuropsychiatric disorders including depression [28,58,62], schizophrenia, and obsessive-compulsive disorder [21]. DLPFC stimulation has also exhibited efficacy in reducing experimentally induced pain [101] and in various chronic pain conditions including fibromyalgia [20,51,82,108] and spinal cord injury related pain [83]. Theta burst stimulation (TBS), a newer rTMS paradigm, provides repetitive bursts of magnetic stimuli at a higher frequency, uses relatively lower intensity and the treatment protocol is much shorter (within 1-3 minutes) than HF-rTMS [46,106]. Furthermore, the effects of TBS seem to persist longer, and the experience is less averse, suggesting that TBS may be a more effective, efficient, tolerable therapeutic intervention compared to HF-rTMS [19,106].

Although the mechanisms that explain the analgesic effects of HF-rTMS and pcTBS are unclear, previous studies have demonstrated that stimulating M1 and DLPFC separately modulates nociceptive pain processing via activation of descending pain modulation systems [25,56,78,108] and alterations in intracortical excitability [23,38,72]. The descending systems are inhibitory and composed of communications between the cortico-limbic structures and brain stem nuclei [31,88]. Lack of efficiency in these systems has been shown to contribute to chronic pain and NP [90,113–115]. The conditioned pain modulation paradigm is a psychophysical pain protocol used to assess the activity of the descending systems [64,90,113]. The ascending pain modulatory systems are also critical in processing pain stimuli [87,105]. Temporal summation of pain is used to examine the activity of the ascending pain systems and is quantified with an increase in pain ratings after application of a repeated brief noxious stimuli [17]. In chronic pain patients with NP, separate studies have demonstrated an impaired conditioned pain modulation response (lack of efficiency in the descending pain systems), and an enhanced temporal summation of pain response (diminished efficiency in the ascending pain systems) compared to asymptomatic controls [7,17,32,64,90]. Furthermore, in patients with chronic pain, two studies have examined these mechanisms at the same time and demonstrated impaired conditioned pain modulation and facilitated temporal summation of pain in patients with chronic pain [40,85].

Neurophysiological mechanisms of pcTBS have been studied using the motor evoked potential (MEP) amplitude as a marker of motor corticospinal excitability [34,38,110]. MEP is the biphasic response that occurs when TMS is administered at M1 over the cortical representation of a specific muscle [50]. Stimulation generates action potentials which induce descending volleys in the pyramidal tract projecting on the spinal motoneurons [12,50,94]. Patients with chronic pain have demonstrated reductions in MEP amplitude at baseline followed by an increase in MEP amplitude following HF-rTMS [23,89]. Furthermore, in healthy participants following pcTBS, an increase in MEP amplitude has been observed [49,68,79]. Thus, chronic pain exerts an inhibitory modulation on motor corticospinal excitability, reducing MEP amplitude that can be increased using HF-rTMS, and potentially pcTBS.

Intracortical inhibition is an indirect marker for Gamma-aminobutyric acid (GABA), an important neurotransmitter involved in pain transmission and perception [3,66]. GABA is a modulator of neural excitability mainly at the level of the dorsal horn in the spinal cord, via the activity of the GABA-A receptors and GABA-B receptors [54,66,73]. Their inhibitory activity modulates chloride influx and high voltage-gated Ca 2+ channels influx which plays an important role in spinal excitation, pain transmission and central sensitization [53,66]. A decrease in intracortical inhibition depicts a decrease in the inhibitory neural activity resulting in an increase in pain transmission and pain perception. Figure 1 summarizes the neural mechanisms discussed above that modulate pain perception in patients with NP.

Figure 1.

Neural mechanisms that modulate pain perception in patients with NP. (a) the descending pain systems, the ascending pain systems, (b) pathway of motor corticospinal excitability, and (c) intracortical inhibition linked to GABAergic activity (SICI= Short Intracortical Inhibition, LICI= Short Intracortical Inhibition, CS= Conditioning Stimulus, TS: Test stimulus, ISI: Interstimulus Intervals). The psychophysical mechanisms are composed of the descending (inhibitory) and the ascending (faciliatory) pain mechanisms. In patients with chronic pain, impaired pain perception is caused by lack of efficiency in the descending pain modulatory systems coupled with facilitation of the ascending pain modulatory systems. The neurophysiological mechanisms evaluated include measurement of motor corticospinal excitability, which is quantified using the motor evoked potential. This is generated when TMS administered over the cortical representation of a specific muscle at M1 generates an action potential which evokes a biphasic response and results in a twitch in a contralateral muscle measured using electromyography. Intracortical inhibition, measured using paired pulse TMS is a marker of GABA mediated activity and in patients with NP there is a decrease in GABA-A and GABA-B receptor activity, which results decrease in SICI and LICI. NP= Neuropathic pain. Figure created with Biorender.com and adapted from Rogasch et al. [93], Christiansen et al. [18]

No previous study has examined pcTBS mechanisms in chronic pain patients. A few studies have investigated mechanisms utilizing HF-rTMS, however with various TMS procotols, participant characteristics, assessment techniques, and mechanisms of interest. Dall’Agnol et al. [23] demonstrated that 10 sessions of HF-rTMS targeted at M1 was associated with a significant reduction in daily pain scores and use of analgesic medication in patients with chronic myofascial pain syndrome. This effect was mediated by an increase in motor corticospinal excitability (increase in MEP amplitude). Furthermore, improvement in pain scores was mediated by the activity of the descending pain modulatory systems measured via quantitative sensory testing and conditioned pain modulation [23]; pcTBS was not utilized in this study. In another study in patients with myofascial pain syndrome that did not incorporate HF-rTMS or pcTBS, Botelho et al. [8] demonstrated at baseline that impaired conditioned pain modulation scores were correlated with greater MEP amplitude, decreased short intracortical inhibition (SICI) and lower heat pain thresholds. In a study that compared the analgesic effects of pcTBS and HF-rTMS (10hz rTMS) targeted at DLPFC, De Martino et al. [68] demonstrated that 24 hours post stimulation both pcTBS and HF-rTMS revealed a similar increase in pain threshold, decrease in pain sensitivity, and were correlated with changes in conditioned pain modulation. Furthermore, this study conducted with healthy adults, revealed an increase in MEP and no change in SICI for both treatments [68]. Contrary to those results, Moisset et al [79] and Klirova et al. [49] examining pcTBS at M1 in healthy subjects observed an increase in motor corticospinal excitability, but there were no changes in heat thresholds or conditioned pain modulation levels.

Addressing inconsistencies in protocol and analyzing neural mechanisms would help to explain the analgesic effects and clinical benefits of pcTBS in treating NP. Therefore, the purpose of this preliminary study was to examine psychophysical mechanisms of pain modulation, neurophysiological measures of motor corticospinal excitability, and GABA activity mediated intracortical inhibition in patients with NP, in response to pcTBS stimulation on the M1 and DLPFC regions of the brain. We hypothesized that following pcTBS at M1 and DLPFC, patients with NP would demonstrate an increase in the efficiency of the descending pain systems (measured using conditioned pain modulation), a decrease in the activity of the ascending pain systems (measured using temporal summation of pain), increases in motor corticospinal excitability (quantified with MEP), and increases in intracortical inhibition measures linked to GABA activity (quantified using SICI and LICI). To provide a more homogeneous population, participants diagnosed with type 2 diabetes and painful neuropathy were selected to participate in the study.

Methods

Participants

All participants provided written informed consent prior to participating in this study. This study was approved by the Virginia Commonwealth University (VCU) Institutional Review Board and was also registered on the ClinicalTrials.gov website (NCT04988321). The inclusion criteria were (a) female or males aged over 18 and under 75 years, (b)Type 2 diabetes diagnosis, (c) pain for at least 6 months, (d) current pain of at least 4/10 on the visual analog scale, (d) fulfilling the criteria for >90% likelihood of NP with a score of >19 on the painDETECT Questionnaire, [29,97] and (e) if being treated pharmacologically, then stable pharmacological treatment for pain at least 1 month before inclusion. Patients were excluded if they had any active contraindications to rTMS (previous severe head trauma, head surgery or concussion, past or current epilepsy, active brain tumor, intracranial hypertension, implanted ferromagnetic devices; e.g. cardiac pacemaker, neurostimulator or cochlear implants, surgical clips or medical pumps) or any other form of NP. Participants were also excluded if they were unable to read or interpret instructions due to any language barriers and all females of childbearing age if they were pregnant or looking to be pregnant. Eligible patients were recruited from the VCU Health Hospital System.

Randomization

Upon enrollment, participants were randomized into two groups (M1 or DLPFC) in a 1:1 ratio. Participants in the M1 group received sham pcTBS at M1 followed by active pcTBS at MI and participants in the DLPFC group received sham pcTBS at DLPFC followed by active pcTBS at DLPFC. Sham stimulation was always presented first with a 35-45 minute gap between sham and active pcTBS stimulation to avoid any carry over effects that could occur if active pcTBS stimulation was provided first [75–77]. The study participants were blinded to the pcTBS sequence. Figure 2 describes the data collection protocol for the two groups. Each session began with the participant completing of the Brief Pain Inventory for patients with diabetic neuropathy (BPI-DN) and identification of cortical hotspots for M1 or DLPFC. Then baseline measures of motor corticospinal excitability and intracortical inhibition were collected, followed by sham pcTBS stimulation at M1 or DLPFC. The motor corticospinal excitability measures were recorded again and following this, active pcTBS at M1 or DLPFC was performed. Lastly, motor corticospinal excitability and intracortical inhibition were collected and responses to the BPI-DN were gathered. Each individual study session took 120-150 minutes to be completed. Motor corticospinal excitability measures were recorded three times for every session whereas BPI-DN and intracortical inhibition were measured twice for every session.

Figure 2:

Data collection protocol for the two groups. Prior to collecting data, the cortical hotspots for M1 and DLPFC were identified. Primary motor cortex (M1), Dorsolateral prefrontal cortex (DLPFC).

Identification of the Cortical Hotspot for Left M1 and DLPFC

Participants were seated in a comfortable reclining chair with a tightly fitted hair dye cap placed over the head. They were instructed to keep their hands as relaxed as possible. Throughout each session participants’ attention was redirected to a monitor presenting nature videos to limit focus on the stimulator. In addition, participants were offered the use of earplugs to limit the noise during the study session. The optimal site for evoking motor responses for the cortical hotspot for M1 was the right abductor pollicis brevis. A surface electromyography (EMG) electrode was placed on the abductor pollicis brevis muscle. A grounding electrode was placed on the ulnar styloid. EMG signals were amplified (x1000) and bandpass-filtered using an AMT-8 amplifier (Bortec Biomedical) prior to A/D conversion which was sampled at 2 kHz. Single-pulse TMS was delivered to the contralateral M1 brain region (Left M1) using a Magstim 200 stimulator and a 90 mm figure-of-eight coil. The coil was held tangentially on the scalp with the coil center rotated to induce a posterior-to-anterior current across the central sulcus. Identification of the cortical “hotspot” and site of subsequent stimulation was the location evoking the largest peak-to-peak motor evoked potential amplitude in the abductor pollicis brevis at the lowest stimulation intensity. The cortical hotspot for left DLFPC was measured using the Beam F3 location system where F3 stands for hotspot location for DLPFC [4,74]. It utilizes three measurements: head circumference, nasion-inion distance, and left tragus-right tragus distance and uses an online calculator which then provides a polar-coordinate approximation of the F3 site with respect to the scalp vertex [4]. This method accounts for head size and shape and has a higher level of precision and reproducibility compared to other methods [74,112].

pcTBS Protocol

pcTBS was performed using a Magstim Super Rapid² Plus¹ stimulator and a 70 mm double air film coil and the pcTBS protocol consisted of three pulses at 50 Hz (i.e. 60 ms) repeated 400 times at intervals of 200 ms (total of 1200 pulses in 1 min and 44 s) [46,68,79]. For the sham condition, a sham coil (Magstim 70mm double air film sham coil) (P/N: 3950-00, Magstim, Whitland, UK), looking identical to the active coil and making a similar noise but without delivering any active stimulation, was applied to the hotspot. pcTBS stimulation intensity was set to 80% of the resting motor threshold, which was determined as the lowest stimulus intensity that induces motor evoked potential with an amplitude of ≥ 50 μV in at least 5 of 10 consecutive stimuli with the muscle fully relaxed [50,94]. Stimulus intensity was determined using an adaptive parameter estimation by sequential testing software [6]. This software triangulates to a threshold with fewer stimulations to prevent overstimulation before pcTBS.

Blinding Questionnaire

All the participants also completed a demographic questionnaire at baseline and blinding questionnaire at the end of the study session. More information about the blinding questionnaire has been included in the Appendix. Briefly, the participants were asked the following questions as part of the blinding questionnaire:

“You received two forms of pcTBS treatment during this study, active and inactive, which one do you think you received first?

1. Active pcTBS

2. Inactive pcTBS

The participants were also asked a question about how certain they were with their ability to correctly guess the treatment using a visual analog scale based on the guidelines of Broadbent et al. [11] with 0 being active pcTBS and 100 corresponding to inactive pcTBS.

Brief Pain Inventory for Patients with Diabetic Neuropathy

All study participants completed the Brief Pain Inventory (BPI) for patients with diabetic neuropathy (BPI-DN) questionnaire electronically on an iPad using REDCap electronic data capture tools [41,42] hosted at Virginia Commonwealth University. Data was collected at baseline and post-active pcTBS.

Psychophysical Pain Protocols

Testing for conditioned pain modulation and temporal summation of pain was performed on body locations that were without painful sensations: the forearm, trapezius, palm of the hand/wrist or shin area. Baseline pressure pain thresholds were assessed using a handheld digital pressure algometer. Mechanical force was applied using a .5-cm² probe covered with polypropylene pressure-transducing material. Pressure was increased at a steady rate until the subject indicated that the pressure was “first perceived as painful.” For conditioned pain modulation, the cold pressor test was used as the conditional stimuli and participants immersed their contralateral (left) hand up to the wrist in a cold-water bath maintained at 4°C; a water temperature used as conditioning stimulus in previous conditioned pain modulation studies [61,64,67]. Twenty to 30 seconds after hand immersion, pressure equivalent to the pressure pain threshold using the pressure algometer was applied on the contralateral hand (right hand, not in water) and participants again indicated when the increasing pressure stimulation first became painful. The pressure at this point represented their conditioned pressure pain threshold. For each of the conditioned pain modulation trials, a conditioned pain modulation index was derived by calculating the percent ratio of pressure pain threshold during cold pressor test to pressure pain threshold before cold pressor test. Scores from the two conditioned pain modulation trials were averaged, and higher conditioned pain modulation scores (an increase in threshold) represents greater pain-inhibitory capacity [113,116].

To assess temporal summation of pain, using the same pressure algometer device, ten identical pressure stimuli equivalent to a pressure at the individual’s pressure pain threshold level, with 1 s duration and 1 s inter-stimulus interval, participants were asked to rate their pain intensity for each of the 10 pressure stimuli using a 10-point visual analog scale (VAS). For analysis of temporal summation of pain, the mean VAS score was calculated in the interval from the first to the end of the fourth stimulus (VAS-I) and in the interval from the eighth to the end of the 10th stimulus (VAS-II). Temporal summation of pain was defined as the difference between VAS-I and VAS-II (i.e. VAS-II minus VAS-I). This protocol has been used in previous studies that have assessed temporal summation of pain in healthy subjects and in patients with chronic pain [7,44,81].

Analysis of Motor Corticospinal Excitability

Quantification and analysis of motor corticospinal excitability and intracortical inhibition has been previously described [75–77]. Purpose-written code in MATLAB (MATLAB v 9.7.0.1190202) was used to calculate peak-to-peak MEP amplitudes from the motor target EMG data of each session. The root mean square (RMS) amplitude was calculated for the evoked response over a 50 ms window (12-62 ms post TMS pulse), and for a 50 ms window prior to the TMS pulse (pre-stimulus). For instances where the pre-stimulus RMS was greater than the evoked response amplitude or where voluntary activation was observed, the evoked response RMS was excluded [24]. During each time interval in which MEPs was recorded, no more than 15 stimulations were delivered. MEPs were normalized and presented as a percentage of the recorded MVC value. Normalized MEPs (nMEPs) served as the measure of motor corticospinal excitability.

Measurement of Intracortical Inhibition

Using paired pulse TMS, paired pulses were delivered randomly at interstimulus intervals of 2 ms and 4 ms for measuring SICI [23,60,79] and at interstimulus intervals of 100 ms and155 ms for LICI, [98,102] with the intensity of the first stimulus set at 80% of the resting motor threshold and the intensity of the second stimulus at 120% of the resting motor threshold [59,68,79]. For each measurement, the results of at least four trials were averaged, and the changes in test MEP induced by conditioned stimuli (paired pulses) were expressed as a percentage of the control MEP amplitude at 120% [59]. SICI and LICI were expressed as the amount of inhibition (intracortical inhibition=100%-pp(paired-pulses)/ control MEP%) [59].

Statistical Analysis

Previous studies have reported standardized effect sizes of 0.45-0.47 for changes in MEP [8,15]. Using an effect size of 0.50 with power of 0.80 at the 0.05 significance level, the appropriate sample size was calculated to be 42. To account for attrition, 90 patients were recruited, and 47 participants were enrolled. Twenty-one participants in each group completed the study. Demographic characteristics are presented, and normality tests were performed using visual inspection and the Shapiro–Wilk test. Statistical Package for Social Science (SPSS^®software, v. 28.1, IBM Corporation) was used for all statistical analysis with significance set at p < 0.05. Independent sample t-tests were conducted to look for differences in demographic characteristics for the two groups and results were interpreted using Levene’s Test for Equality of Variances.

For variables that had a normal distribution, the middle 95% of the data was identified and outliers were eliminated if the data was outside the plus or minus 2 standard deviations of the mean. If the data was not normal, quartiles (Q1, lowest 25% of the data; Median, lowest 50% of data; Q3, lowest 75% of the data) were computed, and if any observation was less than Q1-1.5 X Interquartile range (equal to Q3 – Q1) or if it was greater than Q3+1.5 X Interquartile range, they were eliminated.

The dependent variables for the statistical analysis were BPI-DN, conditioned pain modulation, temporal summation of pain, MEP, SICI at 2ms, SICI at 4ms, LICI at 100ms, and LICI at 155ms. The two independent variables were the two brain regions, M1 and DLPFC, and the three time points of measurement: baseline, post-sham pcTBS and post-active pcTBS. A two-way mixed model repeated measures analysis of variance (RMANOVA; 2 brain regions by 2 time points) was conducted to evaluate the effects of pcTBS stimulation at M1 and DLPFC on BPI-DN. Likewise, a two-way mixed model RMANOVA (2 brain regions by 3 time points) was performed to evaluate the effects of pcTBS stimulation at M1 and DLPFC for MEP. Furthermore, a two-way mixed model RMANOVA (2 brain regions by 2 time points) was also conducted to evaluate the effects of pcTBS stimulation at M1 and DLPFC for each of the following variables: conditioned pain modulation, temporal summation of pain, SICI at 2ms, SICI at 4ms, LICI at 100ms, and LICI at 155ms. The Greenhouse–Geisser approach was used to correct for violations of sphericity if the estimated epsilon (ε) was less than 0.75. Huynh-Feldt correction was used if ε was greater than 0.75. Effect sizes (partial eta-squared [η²]) are reported for significant effects. Where appropriate, post hoc analyses were performed using a Bonferroni multiple comparison correction.

Results

Demographics

Participant demographics for the two groups are reported in Table 1. With 47 participants enrolled in the study and randomized to receive either pcTBS at M1 or pcTBS at DLPFC (see Figure 2), one participant was excluded from testing because they could not perceive any pressure anywhere in the hand/forearm/ thighs/shin with the pressure algometer. Another participant did not complete the conditioned pain modulation procedure because they had Raynaud’s phenomenon, and their symptoms could be exacerbated with exposure to cold. Two participants refused to participate in the post-pcTBS testing of conditioned pain modulation and temporal summation of pain because of pain. In addition, two participants did not complete the study session due to an inability to locate the cortical hotspot for the M1 brain region.

Table 1:

Demographic Data for all Participants

	Total (N=47)	pcTBS at M1 (n=23)	pcTBS at DLPFC (n=24)	p-value
Sex, n (%)				0.16
Male	19 (40.42)	11 (47.7)	8 (33.33)
Female	28 (59.38)	12 (52.2)	16 (66.67)
Race, n (%)				0.13
• Non-Hispanic Black	24 (51.10)	12 (52.17)	12 (50.00)
• Non-Hispanic White	18 (38.30)	8 (34.78)	10 (41.66)
• Asian	1 (2.10)	0 (0.00)	1 (4.23)
• Hispanic/Latino/Spanish	1 (2.10)	1 (4.30)	0
• Mixed	2 (4.30)	1 (4.30)	0
• Prefer not to say	1 (2.10)	1 (4.30)	1 (4.23)
Age (years)	58.65 ± 8.82	59.65 ± 10.23	57.71 ± 7.33	0.46
Duration of pain (months)	67.07 ± 6.51	67.65 ± 58.47	66.50 ± 72.97	0.48
PD-Q score (−1 and 38 range)	22.15 ± 65.55	21.78 ± 2.58	22.50 ± 3.36	0.21
Current pain on VAS (0-10 range)	5.87 ± 1.88	5.91 ± 1.90	5.83 ± 1.90	0.44
BMI, kg/m2	31.87 ± 6.51	33.26 ± 6.57	30.54 ± 6.30	0.08
Pre RMT (%MSO)	55.74 ± 8.87	56.17 ± 9.49	55.33 ± 8.44	0.37
Post active pcTBS RMT (%MSO)	54.39 ± 9.68	53.13 ± 11.02	55.60 ± 8.27	0.38
MVC, mv	37.51 ± 13.09	40.00 ± 14.28	35.03 ± 11.56	0.20

BMI: body mass index, VAS: visual analog scale, PD-Q: painDETECT score, RMT: Resting motor threshold, MVC: Maximum voluntary contraction.

Changes in BPI-DN Scores

The effects of pcTBS stimulation at M1 and DLPFC on the BPI-DN pain severity subscale and on the BPI-DN pain interference subscale revealed no significant interaction effects, F (1,40) = 0.002, p = .963 and F (1,40) = 2.843, p = .100, respectively. In addition, there was no main effect for time on the BPI-DN pain severity subscale and the BPI-DN pain interference subscale. To examine acute pain intensity pre and post pc TBS stimulation, an analysis of the participants response to “Please rate your pain due to your diabetes by sliding to the one number that tells how much pain you have right now” was conducted. Our complete study sample of 42 subjects reported an average improvement of at least 13.53% from pre-pcTBS to post-pcTBS (p=0.03) on this question about their current pain.

Changes in Psychophysical Pain Protocol

Figure 3 depicts the changes in psychophysical pain protocol for the two brain regions post-pcTBS. The effects of pcTBS stimulation at M1 and DLPFC on the conditioned pain modulation scores, revealed no significant interaction effect for brain region and time, F (1,36) = 0.044, p = .834. Similarly, no effect was observed for pcTBS at M1 and DLPFC on the temporal summation of pain scores, F (1,36) = 2.060, p = .160. Furthermore, there was no main effect of time or brain region on the conditioned pain modulation and temporal summation of pain scores.

Figure 3:

Effects of pcTBS at M1 and DLPFC on psychophysical pain protocol across the two time points for the M1 and DLPFC group brain regions. There was no interaction effect for brain region and time and no simple main effect for brain region or time.

Changes in Motor Corticospinal Excitability

Figure 4 depicts the changes in MEP amplitude for the two brain regions from baseline to post-sham pcTBS, post-sham pcTBS to post-active pcTBS and baseline to post-active pcTBS. In contrast to conditioned pain modulation and temporal summation of pain, the RMANOVA revealed that although there was no significant interaction effect for brain region and time, F (2,76) = 2.198, p = .118, there was a statistically significant main effect for time (F [2,76] = 16.144, p = <.001, partial η2= .298). Post hoc Bonferroni analyses revealed that for the M1 group, there was a significant decrease in MEP from baseline to post-sham pcTBS. There was a significant increase in MEP from post-sham pcTBS to post-active pcTBS. With regards to the DLPFC brain region, post hoc analyses revealed that there was a significant increase in MEP from baseline to post-active pcTBS and from post-sham pcTBS to post-active pcTBS.

Figure 4:

Effects of pcTBS at M1 and DLPFC on MEP across the three time points for the M1 group (# indicates a significant decrease from baseline to post-sham pcTBS; ### indicates significant increase from post-sham pcTBS to post-active pcTBS) and for the DLPFC group ( * indicates a significant decrease from post-sham pcTBS to post-active pcTBS; ** indicates a significant increase from baseline to post-active pcTBS). There was no interaction effect for brain region and time but there was a statistically significant effect of time. Post hoc Bonferroni analyses revealed a significant increase for both M1 and DLPFC for MEP from increase in MEP from post-sham pcTBS to post-active pcTBS. For the M1 brain region, there was a significant decrease in MEP from baseline to post-sham pcTBS. For the DLPFC brain region, there was a significant increase in MEP from baseline to post-active pcTBS.

Changes in Intracortical Inhibition

Figure 5 depicts the changes in intracortical inhibition (SICI at 2ms, SICI at 4ms, LICI at 100ms) for the two brain regions post-pcTBS. The effects of pcTBS stimulation on SICI at 2ms revealed no significant interaction effect, F (1,31) = 2.594, p = .117. There was no main effect for time or brain region on pcTBS stimulation at M1 and DLPFC on SICI at 2ms. In addition, pcTBS stimulation at M1 and DLPFC on SICI at 4ms revealed no significant interaction effect for brain region and time, F (1,38) = 1.472, p = .233. However, SICI at 4ms did result in a statistically significant effect for time (F [1,38] =17.713, p = <.001, partial η2= .318) and for brain region (F [1,38] = 5.564, p = 0.024, partial η2= .128). Post hoc analyses revealed that for the M1 group there was a significant increase in SICI at 4ms from baseline to post-active pcTBS. Similarly, for the DLPFC group, there was a significant increase in SICI at 4ms from baseline to post-active pcTBS.

Figure 5:

Effects of pcTBS at M1 and DLPFC on ICI across the two time points for the M1 group (# indicates a significant increase from baseline to post-active pcTBS) and for the DLPFC group (* indicates a significant increase from baseline to post-active pcTBS). Panel A: SICI at 2ms. Panel B: SICI at 4ms. Panel C: LICI at 100ms. Panel D: LICI at 155ms. There was no interaction effect for all the four ICI measures but there was a statistically significant simple main effect of brain region and time for SICI at 4ms and LICI at 100ms. There was a statistically significant simple main effect of time for LICI at 155ms. Post hoc Bonferroni analyses revealed a significant increase for both M1 group brain region and DLPFC group brain region for SICI at 4ms from baseline to post-active pcTBS. There was also a significant increase in ICI measured using LICI at 100 ms for the M1 group brain region and LICI at 155ms for the DLPFC group brain region.

The effects of pcTBS stimulation at M1 and DLPFC on LICI at 100ms, revealed no significant interaction effect for brain region and time, F (1,32) = 1.595, p = .216. Although, there was a statistically significant effect for time (F [1,32] =13.586, p = <.001, partial η2= .298) and brain region (F [1,32] = 6.485, p = 0.024, partial η2= .169) on LICI at 100ms. Post hoc analyses conducted at both time points (baseline and post active pcTBS) revealed significant differences for LICI at 100ms for the M1 and DLPFC groups. Post hoc analyses for the M1 group also demonstrated a significant increase in LICI at 100ms from baseline to post-active pcTBS. There were no significant differences across time for the DLPFC group. With regards to the effects of pcTBS stimulation at M1 and DLPFC on LICI at 155ms, there was no significant interaction effect for brain region and time, F (1,31) = 1.013, p = .322. There was a statistically significant effect for time (F [1,31] =12.48, p = .001, partial η2= .287) on LICI at 155ms. Post hoc analyses revealed that for the DLPFC group, there was a significant increase in LICI at 155ms from baseline to post-active pcTBS.

Relationship between Motor Corticospinal Excitability and Inhibition on BPI-DN

To further examine the statistically significant increases in motor corticospinal excitability (MEP) and intracortical inhibition (SICI and LICI) for the two brain regions post pcTBS, a stepwise regression was utilized to investigate if these neural mechanisms, would predict scores on the BPI-DN at baseline for the entire study sample. The dependent variables were the BPI-DN pain severity, and BPI-DN pain interference subscale scores and the independent variables were MEP, SICI at 2ms, SICI at 4ms, LICI at 100ms, and LICI at 155ms. The regression analysis revealed that, without including any demographic variables from Table 1, MEP, SICI and LICI did not predict any of the subscales at baseline. Similarly, when including the demographic of age, gender, body mass index, race, PDQ score, duration of symptoms and current pain, as covariates, MEP, SICI, and LICI did not predict the BPI-DN subscales of pain severity and pain interference. At baseline, only current pain significantly predicted scores on the BPI-DN pain severity subscale with B= 0.406, standard error= 0.162, t = 2.502 and p= 0.017.

Blinding

Forty participants completed the blinding questionnaire. Fifteen participants (37.5%) reported that they received inactive pcTBS (sham pcTBS) first with 55% certainty whereas 25 participants (63.5%) reported receiving active pcTBS first with 44.44% certainty.

Discussion

With this preliminary study, we examined the neural mechanisms associated with pcTBS stimulation at the M1 and DLPFC brain regions in patients with NP. We hypothesized that pcTBS targeting these two brain regions would cause alterations in the activity of the endogenous pain modulatory systems, coupled with an increase in neurophysiological mechanisms of motor corticospinal excitability and neurochemical mechanisms linked to GABA activity. Our results demonstrated that a single session of pcTBS at M1 for one group and at DLPFC for a second group did not influence the activity of the ascending and descending pain modulatory systems as measured by conditioned pain modulation and temporal summation of pain using a psychophysical pain protocol, respectively. In addition, for both groups, pcTBS was related to an increase in MEP amplitude, depicting an increase in motor corticospinal excitability. Enhancement of GABA-A receptor activity measured using SICI at 4ms was observed following pcTBS at M1 and DLPFC. With regards to GABA-B receptor activity measured using LICI, an increase was only observed for the M1 brain region with an interstimulus interval of 100 ms and for the DLPFC brain region at 155 ms. Lastly, there was an approximately 13% improvement in acute pain intensity scores post pcTBS targeted at M1 and DLPFC compared to baseline.

Contrary to the results of our study regarding the descending and ascending pain systems, Giannoni-Luza et al. [33] determined from a meta-analysis that non-invasive brain stimulation techniques had a significant effect on conditioned pain modulation compared to sham, both in healthy subjects and patient populations. Although, it should be noted that this meta-analysis included only three studies that utilized rTMS targeted at the M1 brain region and the remaining studies assessed temporal summation of pain. Results from the present study were not consistent with the meta-analysis in that no differences in the activity of the descending pain systems and the ascending pain systems, measured using conditioned pain modulation and temporal summation of pain, respectively, were observed following a single session of pcTBS targeted at either M1 or DLPFC in patients with painful diabetic neuropathy.

There are at least two possible explanations for the lack of changes in the endogenous pain systems in our study; pcTBS targeted at M1 and DLPFC does not influence the endogenous pain systems and/or patients with painful diabetic neuropathy have an efficient endogenous pain system at baseline. Regarding the first explanation, De Martino et al. [68] in a study in healthy participants utilizing pcTBS targeted at DLPFC, after three sessions did not find any increase in the efficiency of the conditioned pain modulation. Similarly, Moisset et al. [79] targeted the M1 brain region and after one session of pcTBS in healthy participants, there was no change in conditioned pain modulation efficiency. The present study is unique in that conditioned pain modulation was assessed in a clinical population of NP patients and temporal summation of pain was used to examine the ascending pain systems, whereas past studies have only assessed conditioned pain modulation in healthy subjects, and no study has evaluated changes in temporal summation of pain at baseline and following pcTBS targeted at M1 and DLPFC.

Across the conditioned pain modulation literature, heterogeneity in the methodology of evoking conditioned pain modulation has contributed to discrepancies in results [27,33,35,91]. Pud et al. [91] summarized the methodological differences and proposed that discrepancies were linked to the type of test stimulus (utilizing a tonic, suprathreshold, pain threshold or pain tolerance approach), modality of conditioning and test stimuli (thermal, cold, pressure, ischemic), and the site of testing the conditional and test stimuli (affected vs unaffected body areas. In a review by Fernandes and colleagues to examine the concurrent validity of conditioned pain modulation in patients with chronic pain, they reported that stimulation site was a critical factor influencing conditioned pain modulation results [27]. More specifically, testing on painful areas seems to alter results due to the level of sensitivity. In the present study, prior to conditioned pain modulation testing, participants were asked to identify non-painful areas. However, the influence of chronic pain, along with the presence of sensory loss, tingling, and numbness cannot be ruled out as possibly altering the measurement of the pressure pain threshold. This raises an important methodological issue to consider when assessing conditioned pain modulation in patients with peripheral neuropathic sensory changes.

The possible effect of the pathology of painful diabetic neuropathy on the assessment of conditioned pain modulation has received attention in the literature. Granovsky et al., in two separate studies, [36,37] examined endogenous pain modulation using conditioned pain modulation in painful diabetic neuropathy patients and concluded that these patients have a more efficient conditioned pain modulation compared to patients with nonpainful diabetic neuropathy. They also observed that longer pain duration (more than 2 years) in patients with painful diabetic neuropathy, was related to a more efficient conditioned pain modulation response and enhanced temporal summation of pain response [36]. In the present study, the average pain duration was 5.5 years. Thus, it is possible that the duration and chronicity of NP could bring about alterations in the pain modulation systems, such that conditioned pain modulation and temporal summation of pain no longer indicate changes in the descending and ascending pain system patients with painful diabetic neuropathy [36,37]. A correlation analysis of the present data for duration of pain and conditioned pain modulation scores for the entire study sample (N=44) was significant at r = .373 (p= 0.013). There was no relationship between temporal summation of pain and duration of pain. Lastly, previous studies have also suggested that conditioned pain modulation is more efficient in younger populations [40,92]. There were no significant correlations between Age and conditioned pain modulation and Age and temporal summation of pain in the present study, and participants were middle age and older. Future studies should evaluate conditioned pain modulation and temporal summation of pain in body areas without any alterations in sensory function and account for the chronic pain duration in painful diabetic neuropathy patients.

Another psychophysical paradigm utilized to assess the descending pain systems is offset analgesia, defined as the disproportionate decrease in pain perception followed by a slight decrease in noxious stimulation [39,117]. This assessment can be complementary to conditioned pain modulation and is associated with a temporal filtering mechanism, while conditioned pain modulation utilizes a spatial filtering mechanism. Another possible advantage of using offset analgesia is that it can only be assessed using thermal stimuli, thus providing a more standardized protocol compared to conditioned pain modulation. Additional studies are needed to evaluate the utility of offset analgesia as a marker of the endogenous descending pain systems. Further investigation into the effectiveness of temporal summation of pain and conditioned pain modulation following single and multiple sessions of pcTBS targeted at M1 or DLPFC in NP patients and other chronic pain populations is warranted.

The increase in motor corticospinal excitability we found after a single session of pcTBS in patients with NP was similar to the increase in motor corticospinal excitability reported in healthy subjects after a single session of pcTBS targeted at M1 [49,79] and after three sessions of pcTBS targeted at DLPFC [68]. Although this work is a preliminary investigation, our results highlight the role of facilitation of motor corticospinal excitability (increase in MEP) as a neurophysiological mechanism to help explain the effects (alleviation of pain) of pcTBS targeted at M1 and DLPFC in patients with NP [38,55].

In chronic pain patients, motor cortex reorganization contributes to the intensity of chronic pain; and pain relief has been correlated with reversal of the cortical reorganization [89]. The increase in motor corticospinal excitability causes increased excitability of the motor cortex and the corticospinal tract due to changes that potentially contribute to the reversal of cortical reorganization [57,107,108]. Furthermore, this increased excitability exerts inhibitory effects on the top-down endogenous system via activation of cortical and subcortical structures. These structures involved with pain processing and perception, include the limbic region, anterior cingulate cortex, orbitofrontal cortex, periaqueductal gray, thalamus and the subcortical brain regions which are also involved in pain processing and perception [5,25,55,71].

Among previous studies that have demonstrated changes in motor corticospinal excitability following pcTBS at M1 [79] and DLPFC [68] no changes in SICI at interstimulus intervals (2ms, 3ms and 4ms) were revealed. Our study results concur with no changes in SICI at 2ms, but with regard to SICI at 4ms, there was a significant increase, suggesting GABAergic inhibition following pcTBS at DLPFC. A decrease in SICI [13] has been observed at baseline in cross-sectional studies in patients with chronic NP [72,100], musculoskeletal pain [13] and experimental pain [99]. Similarly, following a single session of HF-rTMS targeted at the M1 brain region [60] an increase in SICI at different interstimulus intervals (2ms, 3ms and 4ms) from baseline has been demonstrated [59,60]. Only two studies [98,102] have assessed LICI in chronic pain conditions. With a cross-sectional analysis, Salerno et al. examined LICI and found a significant reduction in LICI at 155ms when comparing patients with fibromyalgia and rheumatoid arthritis to healthy controls [98].

In another study, Siniatchkin et al. evaluated LICI at baseline using different interstimulus intervals (20, 60, 120 ms) in patients with migraine compared to healthy controls and found no differences [102]. The present study is the first to investigate LICI at 100ms and 155ms at baseline and following a single session of pcTBS targeted at the M1 and DLPFC brain regions. Results suggest an increase in GABA-A receptor activity measured using SICI at 4ms following pcTBS at DLPFC and an increase in GABA-B activity following pcTBS at M1 and DLPFC measured using LICI at 100ms and 155ms. This demonstrates a reversal of the dysfunction in GABAergic inhibition at baseline seen with chronic painful diabetic neuropathy, following a single session of pcTBS at M1 and DLPFC.

Proton magnetic resonance spectroscopy, a type of neuroimaging technique, can investigate the mechanistic role of GABA on the effects of TMS and pcTBS targeted at the M1 and the DLPFC brain regions. Previous studies examining GABA levels in the brain following TMS and rTMS targeted at M1 and DLPFC have observed a correlation between presynaptic GABA levels in different brain regions and intracortical inhibition markers (SICI and LICI) in healthy subjects [16] and in patients with neuropsychiatric disorders such as schizophrenia [84], depression [26,63]. These studies also found altered presynaptic levels of GABA in the DLPFC brain region [16,26,63,84]. The results in the present study provide inconsistent results with regards to SICI measured at 2ms and 4ms and LICI at 100ms and 155ms. The standardization of assessment of the interstimulus intervals and stimulus intensities used to assess intracortical inhibition must be considered to provide an important control for examining the consistency of results across studies.

Motor corticospinal excitability and intracortical inhibition are critical in maintaining a balance between the cortical excitability and inhibitory networks in the brain involved in the modulation of pain perception. pcTBS targeted at M1 and DLPFC seems to restore this balance by inducing synaptic neuroplasticity, increasing neurotransmitter levels, and activating other cortical structures and circuits involved in pain perception. Stagg et al. [104] revealed that continuous TBS over the primary motor cortex increased GABA levels without altering the levels of glutamate, suggesting that the increase in these presynaptic levels of GABA are driving the changes. Furthermore, Huang et al. [45,47,65,95] and Morales et al. [14] have highlighted the role of N-methyl-D-aspartate (NMDA) receptors [45], glutaminergic receptors and GABA receptors [47,65] in regulating intracellular calcium dynamics. These alterations can induce synaptic plasticity related changes in the mediation of long-term potentiation (quick and rapid influx) and long-term depression (slow and moderate) at the level of the post-synaptic terminal neuron [46,48]. More recently, Larson and Munkacsy [52] elucidated the role of GABAergic circuits using modeling studies and demonstrated that bursts repeated at theta rhythm induce maximum long-term potentiation by disabling feedforward inhibition that involves presynaptic GABA receptors leadings to GABAergic inhibition. Future studies in patients with chronic pain should consider utilizing neuroimaging techniques to examine different brain regions and incorporate multiple sessions of pcTBS to clarify the role of GABA as a mediator of synaptic plasticity.

Limitations

This preliminary study included participants that were homogeneous regarding type of chronic pain, (a methodological advantage), external validity may have been compromised. Thus, additional studies with a larger number of patients who are experiencing various types of chronic pain are needed to assess the potential benefits of pcTBS in various clinical settings with various clinical populations. Findings should be considered exploratory and need to be replicated in larger future randomized controlled trials. In addition, this was a single blind study and double-blind studies are necessary. Furthermore, the use of a neuronavigational system to optimize the identification of cortical hotspots for the M1 brain region and the DLPFC brain region would provide greater reliability. And, as previously mentioned, future studies should incorporate multiple sessions to evaluate possible changes in the three mechanisms over time that may explain the efficacy of pcTBS targeted at M1 and DLPFC. This was the first study to investigate the neural mechanisms that may explain the changes in pain perception using single session of pcTBS targeted at M1 and DLPFC in patients with NP. Finally, the sham stimulation was provided first to all the participants prior to active pcTBS for both the brain regions. This was done because a reversal of this order would have made it impractical to assess the effects of active pcTBS, however, an order effect cannot be ruled out. Thus, investigations that incorporate a sham session on a different day with greater time between the sham and treatment sessions are warranted. Of note, the results from the blinding questionnaire revealed that more than 60% (n=25/40, 63.5%) of the participants guessed the order of stimulation incorrectly, suggesting that patients were not aware of the study protocol. Future work should consider including a separate sham group as a control group.

Conclusions

This preliminary study highlights a link between the neurophysiological mechanisms of motor corticospinal excitability and GABA activity (utilizing intracortical inhibition measures) after one session of pcTBS targeted at M1 and DLPFC in patients with NP. Future studies should incorporate multiple sessions, different subject populations and neuroimaging methods to further elucidate the mechanisms that govern the efficacy of pcTBS targeted at M1 and DLPFC.

Appendix

Patient Blinding questionnaire

Please complete the survey below.

Thank you!

1)	“You received two forms of pcTBS treatment during this study, active and inactive, which one do you think you received first?	○ Active pcTBS ○ Inactive pcTBS
2)	How certain are you in your ability to correctly guess the treatment that you received first ?	Active	Inactive