Immediate Changes Following Manual Therapy in Resting State Functional Connectivity As Measured By Magnetic Resonance Imaging (fMRI) In Subjects With Induced Low Back Pain

Abstract

Objective

The purpose of this study was to use functional magnetic resonance imaging (fMRI) to investigate the immediate changes in functional connectivity (FC) between brain regions that process and modulate the pain experience following 3 different types of manual therapies (MT) and to identify reductions in experimentally induced myalgia and changes in local and remote pressure pain sensitivity.

Methods

Twenty-four participants (17 females, mean age ± SD = 21.6 ± 4.2 years), who completed an exercise-injury protocol to induce low back pain, were randomized into 3 groups: chiropractic spinal manipulation (n=6), spinal mobilization (n=8) or therapeutic touch (n=10). The primary outcome was the immediate change in FC as measured on fMRI between the following brain regions: somatosensory cortex, secondary somatosensory cortex, thalamus, anterior and posterior cingulate cortices, anterior and poster insula, and periaqueductal grey. Secondary outcomes were immediate changes in pain intensity measured with a 101-point numeric rating scale, and pain sensitivity, measured with a hand-held dynamometer. Repeated measures ANOVA models and correlation analyses were conducted to examine treatment effects and the relationship between within-person changes across outcome measures.

Results

Changes in FC were found between several brain regions that were common to all 3 manual therapy interventions. Treatment-dependent changes in FC were also observed between several brain regions. Improvement was seen in pain intensity following all interventions (p<0.05) with no difference between groups (p>0.05). There were no observed changes in pain sensitivity, or an association between primary and secondary outcome measures.

Conclusion

These results suggest that manual therapies (chiropractic spinal manipulation, spinal mobilization, and therapeutic touch) have an immediate effect on the FC between brain regions involved in processing and modulating the pain experience. This suggests that neurophysiological changes following MT may be an underlying mechanism of pain relief.

Introduction

Improvements in pain intensity and pain sensitivity are often reported following manual therapy (MT).[1–4] Research has demonstrated (1) neurophysiological changes are observed following MT and (2) reductions in pain intensity and pain sensitivity are associated with functional changes in central nervous system (CNS).[5–8] A current assumption is that neurophysiological changes following MT may underlie clinical improvement.

Functional magnetic resonance imaging (fMRI) research includes several different approaches to estimate cortical function. Several of these approaches have demonstrated functional changes associated with pain relief. One such measure is functional connectivity (FC). Functional connectivity has been defined as “the temporal correlation of a neurophysiological index measured in different brain areas” (see Figure 1). [9] Recently, Letzen et al (2013) used FC between the default mode network and brain regions associated with pain processing to investigate lidocaine induced analgesia.[6] While, Zyloney et al (2010) used FC between the Periaqueductal Gray (PAG) and cortical regions to investigate differential effects underlying analgesia from of genuine and sham electro-acupuncture.[10] With the evidence supporting efficacy of MT to reduce pain intensity and pain sensitivity, it is reasonable to assume that the underlying therapeutic effect of MT is likely to include a higher cortical component.[1, 4, 11, 12]

Figure 1.

Functional connectivity is defined as the temporal correlation of a neurophysiological index measured in different brain areas. This term has been applied to functional magnetic resonance imaging, where the changes in BOLD signal over time are compared between two regions of interest (ROI). The correlation between the time series equals the estimated functional connectivity.

Although models explaining the therapeutic effects of MT on pain and pain sensitivity include the potential for a higher cortical mechanism, he extent that MT exert effects on higher brain centers is not fully understood. [13–16] Thus far, only one study has used fMRI to assess changes in cortical function following MT.[17] Unlike the Letzen et al (2013) and Zyloney et al (2010) studies, the Sparks et al study used a different approach. They used peak Blood-oxygen-level-dependent (BOLD) contrast imaging to estimate of cortical function associated with a task. In their study, pain-free volunteers processed thermal stimuli applied to the hand before and after thoracic spinal manipulation (a form of MT). What they found was that following thoracic manipulation, several brain regions demonstrated a reduction in peak BOLD activity. Those regions included the cingulate, insular, motor, amygdala and somatosensory cortices and the PAG.

The purpose of this study was to investigate the changes in FC between brain regions that process and modulate the pain experience following MT. The primary outcome was to measure the immediate change in FC across brain regions involved in processing and modulating the pain experience and identify if there were reductions in experimentally induced myalgia and changes in local and remote pressure pain sensitivity.

Methods

Study Design

This study is made up of a subset of subjects who have completed a larger, ongoing, pre-clinical trial (NCT01406847). A randomized study design with blinded assessment was implemented with 3 groups, measured at 2 time points. Pain-free volunteers completed an exercise protocol to induce myalgia in the low back. Forty-eight hours following completion of the exercise protocol, subjects returned and underwent pre-intervention assessment. Pre-intervention assessment included collection of pain intensity, local and remote pressure pain measures and fMRI data by a blinded assessor. Subjects were then randomized to receive one of three manual therapy interventions. Sealed opaque envelopes were used to inform the treatment provider of assignment. Interventions were performed by either a licensed physical therapist or chiropractor. The randomization sequence was generated by an individual no responsible for determining study eligibility, outcome assessment, or intervention. Following the intervention subjects underwent the same assessment (post-intervention) performed by the same blinded-assessor.

Participants

Seventy five volunteers read and signed the informed consent form approved by the University of Florida Institutional Review Board. Enrolled subjects were recruited from the campuses of the University of Florida and UF Health Hospital and the local surrounding community. Subjects were eligible to participate in the study if they were between the ages of 18 and 44, and currently not experiencing back pain. Subjects were excluded from participating in the study if they met any of the following criteria: previous participation in a conditioning program specific to trunk extensors, any current back pain, any chronic medical conditions that may affect pain perception (e.g. diabetes, high blood pressure, fibromyalgia, headaches), kidney dysfunction, muscle damage, or major psychiatric disorder, history of previous injury including surgery to the lumbar spine, renal malfunction, cardiac condition, high blood pressure, osteoporosis, or liver dysfunction, performance of any intervention for symptoms induced by exercise and before the termination of their participation of the protocol. To be included in these analyses, subjects needed to undergone the exercise-injury protocol and have completed resting-state fMRI scans at both time points.

Exercise injury model to induce low back pain

Prior to exercise all subjects completed a 5-minute warm-up consisting of riding a stationary bicycle. Following the 5-minute warm up, subjects then performed an isometric test to establish a baseline measure of torque.[18] Subjects then performed repetitions of dynamic resisted exercise. Resistance was individualized to each subject using a weight load. Weight loads were equal to 90% of the peak torque measured during the baseline isometric test. Each repetition was performed through the full available range of motion (ROM). Subjects performed sets of 15 repetitions or until volitional fatigue. After each set, torque was re-estimated. Using a criterion of 50% drop in torque, subjects either completed another set if the torque was above 50% or ended if the re-estimated torque was 50% or less than the baseline. Following the exercise, subjects were instructed not to initiate any medication, or apply any intervention to the lumbar spine to reduce painful symptoms. This human model of acute endogenous low back pain has been previously described in greater detail elsewhere.[19, 20]

MT Interventions

Interventions were performed by health care professionals who were selected from a pool based on availability, and who were not involved with the assessment of the subject. The pool consisted of health care professionals who had an active license (either physical therapy or chiropractic), were currently or previously practiced, were comfortable providing manual therapy for musculoskeletal pain conditions, and underwent training (with MDB) on the specific techniques performed in this study. The amount of “hands-on” and personal contact was equivalent between interventions. Prior to the intervention, all subjects were given similar verbal instructions regarding the techniques performed. Each intervention met the criteria to be categorized as a ‘Mind and Body’ based therapy by the National Center of Complementary and Alternative Medicine. [21] In this study each intervention was considered a MT technique and is briefly described below.

Spinal Manipulative Therapy (SMT)

Subjects randomized to the SMT group received a high-velocity, low-amplitude (HVLA) thrust, (grade V), technique previously described in the literature and commonly utilized for the treatment of low back pain by several health care professions.[22–28] This particular technique was selected because it was used in LBP-clinical prediction rule studies involving patients with acute and sub-acute pain[23] and was the same technique used by our group which reported immediate pain sensitivity changes in healthy and LBP subjects.[24, 25]

Spinal Mobilization (MOB)

Subjects randomized to MOB received a “Grade III” mobilizing force applied and released to the lumbar spine, directed posterior to anterior at a rate of 1 Hz for 2 minutes, followed by a 1 minute rest, and then the mobilizing force again for 2 minutes. This particular technique was selected because it was used in a recent clinical trial involving patients with acute or sub-acute pain,[29] and commonly utilized for the treatment of low back pain.

Therapeutic Touch Control (TT)

Subjects randomized to TT lay prone. The therapist placed both hands in contact with the subjects’ pelvis across the top of the posterior aspect of the sacrum and ilia. Light pressure was applied for 5 minutes.

Primary Outcome (Functional Imaging)

Acquisition

Functional magnetic resonance imaging (fMRI) was performed using a research dedicated, Phillips Achieva 3 Tesla MRI Scanner, fitted with a 32-channel head coil. Each resting-state fMRI scan was 5 minute and 42 second long. One resting-state scan was taken before and after the assigned intervention. Functional data was collected in the trans-axial orientation using an EPI sequence, (XYZ dimension = 80*80*38; FOV [RL, AP, FH - mm] =240, 240,114; Slice thickness [mm] =3; Gap thickness = 0; Voxel dimension [mm] = 3*3*3; Repetition time [ms] =2000), which is consistent with recommended resting functional scanning parameters[30]. Structural data consisted of a high-resolution 3D structural T1 collected in sagittal orientation (XYZ dimension = 256*256*180; FOV [ap,fh,rl - mm] =240, 240, 180; Slice thickness [mm] =1; Gap thickness = 0; Voxel dimension [mm]= 1*1*1;TR/TE [ms] =8.1/3.7). During the functional scan, physiological measures (e.g. pulse oximeter and respirations) were recorded simultaneously using built-in recording equipment that is part of the Philips system. During scanning, subjects remained in the supine position with their heads cushioned to reduce motion. Subjects were instructed to remain awake, with their eyes open and fixated on a cross hair, “not to think about anything in particular”, and to remain as still as possible. Subjects wore earplugs throughout the experiment to attenuate MRI noise.

Processing

Functional data were preprocessed in SPM12 (Wellcome Department of Imaging Neuroscience, London, UK; http://www.fil.ion.ucl.ac.uk/spm). Images were (1) slice-time corrected, (2) realigned and resliced into 3mm isotropic voxels, (4) co-registered to the high resolution 3D anatomic volume, (5) warped into MNI standard space using the deformations used to realign the 3D anatomical data into MNI space and (6) spatially smoothed using a 6-mm full-width half-maximum (FWHM) Gaussian kernel. Data were spike-corrected to reduce the impact of artifacts using the post-processing Artifact Detection Tool (ART) toolbox for fMRI data (http://www.nitrc.org/projects/artifact_detect). Time points, where the mean global signal changed by above 3 standard deviations, translation movement exceeded 0.5mm or rotational movement exceeded 0.01°, were identified and later removed during first level general linear model. The final processing steps were then carried out using the functional connectivity toolbox Conn (http://www.nitrc.org/projects/conn) that implements the component-based noise correction method (CompCor) strategy for physiological and other noise source reduction, which included: (1)Temporal (band-pass) filtering set between 0.01 and 0.1 Hz and (2) removal of several sources of nonspecific variance by regression of nuisance variables. Those nuisance variables included: (1) the signal averaged over the lateral ventricles, (2) the signal averaged over the deep cerebral white matter, (3) the six parameters obtained by motion correction and (4) the outlier data points identified with the ART toolbox.

Regions of Interest (ROI)

Within the brain, the pain experience is sub served by an extended network of brain regions including the thalamus, primary and secondary somatosensory, cingulate and insular cortices.[31–33] Collectively, these regions are referred to as the pain processing network (PPN), and encode the sensory discriminate and cognitive and emotional components of the pain experience.[33, 34] Perception of pain is not merely dependent on the neural activity within the PPN, but also on the flexible interactions of this network with other functional systems, including the descending pain modulatory system.[35, 36] The descending pain modulatory system includes sub-cortical regions such as the PAG.[37, 38] Because of this, we choose to investigate 8 brain regions bilaterally (16 total). Those regions are: Anterior Cingulate Cortex (ACC), Posterior Cingulate Cortex (PCC), Anterior Insular Cortex (aINS) and Posterior Insular Cortex (pINS), Thalamus (THA), Primary and Secondary Somato-sensory Cortex (SI, SII), and the Periaqueductal Gray (PAG). We used previously published coordinates to center our ROIs. Those coordinates come from pain studies that included patients with back pain [39–41] and healthy volunteers.[42, 43] With the coordinates as the center, 9mm spheres were created for each ROI, except for the THA and PAG, which were 6mm each. Each ROI sphere was overlaid with a grey-matter mask to include only grey matter voxels. The ROI time series was then estimated by the spatial average of the BOLD time series over all voxels within the generated spherical region and grey-matter mask.

Functional Connectivity

Functional Connectivity was estimated using the “Conn Toolbox” (www.nitrc.org/projects/conn) [44] and represents the bivariate correlation between two ROIs’ time-series. See Figure 1 for a schematic of FC. The FC between each possible ROI-to-ROI connection was estimated twice (120 total connections); once before the intervention and again immediately following the intervention. Bivariate correlations were converted to normally distributed z-scores using the Fisher transformation. The difference (post FC – pre FC) in FC was estimated and used as the change in FC between each ROI-to-ROI connection.

Secondary Outcomes (Behavioral Measures)

Pain Intensity

Subjects used the 101-point numerical rating scale to provide a measure of the current intensity of their lower back pain.[45] The NRS scale is anchored with 0 = ‘no pain’ and 100 = ‘worst pain imaginable’. The therapist who performed the intervention collected subjects’ ratings. The therapist asked for the rating immediately prior to and after the subjects received their assigned intervention. The immediate effect on pain intensity was the difference between ratings given at the two time points (ie post – pre).

Pressure Pain Threshold

Pressure pain thresholds (PPT) were assessed by a blinded assessor using a Microfet 2 hand-held dynamometer (Hoggan Health Industries, Inc, West Jordan, UT). The tip of the dynamometer is equipped with a rubber foot-plate of 1-cm diameter. During testing, subjects were positioned prone and pressure was slowly applied until the subject reported that the sensation changed from pressure to pain. At that point when subjects reported pain, the applied force was recorded. Threshold measures were evoked in the paraspinal muscles bilaterally 2.5cm from the spinous processes of L1, L5 and S2 and over the dorsum of the hand and foot, in the web-space between the first and second toe/finger. Local pressure pain thresholds (local PPT) were a composite measure of the average threshold over the 3 paraspinal muscle locations assessed bilaterally. Remote pressure pain thresholds (remote PPT) were a composite measure of the average threshold assessed on the dorsum of the foot and hand.

Statistics

For all measures, means and standard deviations were created at each time point. Pearson’s correlation coefficients were estimated to assess the relationships across variables at each time point (ie pre-intervention, post-intervention). To establish within and between group changes over time, we used repeated measures analysis of variance (RM-ANOVA). Separate RM-ANOVAs were used to assess for the main effect of time and for group by time interactions for each ROI-to-ROI connection, pain intensity, and pressure pain sensitivity. In each RM-ANOVA the dependent variable was the post randomization value, the baseline value was a covariate and intervention-assignment was the between subject factor. We corrected for the number of separate RM-ANOVAs conducted for across the 120 ROI-to-ROI pairs by using a p-value of less than 0.01 as significant. Lastly, residual change scores were used to put all outcome measures into a standard metric of change. Pearson’s correlation coefficients were used to estimate the relationship of within person change between the different outcomes measured (ie primary and secondary outcomes). All data analyses were performed using SPSS v21.

Results

Of the 75 volunteers, 24 (mean age = 21.6, SD = ± 4.2 years, 17 female) met inclusion criteria and were included in these analyses. Figure 2 shows the derivation of the included sample. The imaging subgroup did no differ in age, percent female or other behavioral measures from the remainder of the non-imaged study population see Table 1 for more details.

Figure 2.

The data used for this study was obtained from a subgroup of volunteers derived from the pre-clinical trial.

Table 1.

Comparison of characteristics of the subgroup included to those not included in current analysis

Variable	Imaging Subgroup (N=24)	No Imaging Subgroup(N=51)	Mean Difference between Groups	p vaue
Age years (SD)	21.6, (4.2)	22.7, (4.0)	1.1	0.27
Gender	71%, (17 females)	73% (37 females)	2%	0.84
Pain Intensity (Pre)	10.5, (15.2)	12.1, (13.4)	1.6	0.64
Pain Intensity (Post)	6.6, (11.9)	9.0, (9.7)	2.5	0.35
Δ Pain Intensity	−3.9, (6.9)	−3.5, (9.5)	0.3	0.88
Local PPT (Pre)	14.9, (6.4)	18.6, (9.3)	3.7	0.08
Local PPT (Post)	14.9, (6.6)	18.0, (9.2)	3.1	0.14
Δ Local PPT	−0.0 (3.7)	−0.4, (2.9)	−0.3	0.68
Remote PPT (Pre)	13.6, (5.6)	16.0, (9.5)	2.5	0.24
Remote PPT (Post)	13.4, (5.2)	14.9, (8.3)	1.5	0.41
Δ Remote PPT	−0.2	−0.9	−0.7	0.36

(Pre) = prior to MBB therapy; (Post) = following MBB therapy; Pain Intensity values = mean rating using 101-point numerical rating system (standard deviation); (PPT) = pressure pain thresholds, values = mean force kg/cm², (standard deviation); (Δ) = change score (post minus pre).

Randomization of the imaging subgroup resulted in 6 subjects receiving SMT, 8 subjects receiving MOB, and 10 subjects receiving TT. Characteristics of the imaging subgroup, separated by randomization assignment, are summarized in Table 2. Pre-intervention, post-intervention and raw change scores are presented for FC measures between each ROI-to-ROI connection, as well as the correlation to behavioral measures at the same time point, in Tables 3, 4, and 5 for right hemisphere connections, left hemisphere connections, and cross hemisphere connections, respectively.

Table 2.

Characteristics of imaging subgroup separated by MBB therapy

Variable / Group	SMT	MOB	TT	Total
Number of Subjects	6	8	10	24
Age mean years (SD)	20.7 (1.8)	21.1 (3.2)	22.5 (5.9)	21.6 (4.2)
# Female (%)	5 (83%)	7 (88%)	5 (50%)	17 (71%)
Pain Intensity (Pre)	11.8 (13.7)	14.5 (22.1)	6.4 (8.5)	10.5 (15.2)
Pain Intensity (Post)	5.5 (8.1)	10.1 (18.2)	4.4 (7.2)	6.6 (11.9)
Δ Pain Intensity	−6.3 (8.8)	−4.4 (8.5)	−2.0 (3.5)	−3.9 (6.9)
Local PPT (Pre)	12.9 (4.8)	12.4 (4.3)	18.2 (7.5)	14.9 (6.4)
Local PPT (Post)	11.8 (4.6)	11.2 (2.9)	19.7 (7.0)	14.9 (6.6)
Δ Local PPT	−1.1 (0.8)	−1.2 (3.8)	1.5 (4.2)	−0.02 (3.7)
Remote PPT (Pre)	10.1 (3.4)	12.7 (3.7)	16.4 (6.8)	13.6 (5.6)
Remote PPT (Post)	10.2 (3.7)	11.6 (3.1)	19.7 (7.0)	13.4 (5.2)
Δ Remote PPT	0.1 (1.0)	−1.1 (3.9)	0.4 (3.0)	−0.2 (3.0)

(Pre) = prior to MBB therapy; (Post) = following MBB therapy; Pain Intensity values = mean rating using 101-point numerical rating system (standard deviation); (PPT) = pressure pain thresholds, values = mean force kg/cm², (standard deviation); (Δ) = change score (post minus pre).

Table 3.

Right Hemisphere ROI to ROI Connections

ROI-to- ROI Pair	(Pre) FC	ρ (Pre) Pain Intensity	ρ (Pre) Local PPT	ρ (Pre) Remote PPT	(Post) FC	ρ (Post) Pain Intensity	ρ (Post) Local PPT	ρ (Post) Remote PPT	Δ FC	ρ Δ Pain Intensity	ρ Δ Local PPT	ρ Δ Remote PPT
R_ACC- to-R_PCC	0.26 (0.28)	.053	.282	.456*	0.17 (0.24)	.271	.062	.204	−0.08 (0.30)	.008	−.129	.118
R_ACC- to-R_aINS	0.26 (0.24)	.204	.302	.389	0.30 (0.22)	−.424*	.185	.115	0.04 (0.30)	.127	.142	−.117
R_ACC- to-R_pINS	0.29 (0.22)	.214	−.116	.169	0.22 (0.27)	−.370	−.019	−.041	−0.07 (0.30)	.141	.220	−.187
R_ACC- to-R_SI	0.30 (0.25)	.033	.185	.289	0.25 (0.24)	−.214	.124	.043	−0.05 (0.28)	−.093	.153	.053
R_ACC- to-R_SII	0.31 (0.22)	−.011	.360	.604**	0.29 (0.24)	−.397	−.007	−.102	−0.02 (0.33)	−.105	.052	.008
R_ACC- to-R_THA	0.09 (0.19)	.062	.151	.129	0.13 (0.19)	.339	−.291	−.031	0.04 (0.27)	.039	−.018	−.255
R_ACC- to-R_PAG	0.18 (0.22)	.088	.014	.248	0.20 (0.17)	−.049	−.031	.070	0.10 (0.31)	−.143	−.139	.216
R_PCC- to-R_aINS	−0.00 (0.17)	.275	.145	.432*	0.04 (0.14)	−.178	.164	.074	0.04 (0.18)	.062	−.208	−.218
R_PCC- to-R_pINS	0.16 (0.22)	.282	.089	.432*	0.09 (0.19)	−.333	−.051	−.104	−0.07 (0.29)	.249	−.070	.064
R_PCC- to-R_SI	0.16 (0.30)	−.080	.267	.498*	0.08 (0.26)	−.085	.021	.177	−0.08 (0.35)	−.073	−.305	.467*
R_PCC- to-R_SII	0.16 (0.31)	.041	.322	.587**	0.11 (0.23)	−.266	.020	.116	−0.05 (0.32)	.070	−.271	.125
R_PCC- to-R_THA	0.18 (0.25)	.180	.006	.137	0.16 (0.18)	−.093	.110	.173	−0.02 (0.26)	.569**	−.018	−.277
R_PCC- to-R_PAG	0.22 (0.25)	.259	.067	.239	0.23 (0.18)	−.140	.293	.244	0.01 (0.28)	.237	−.200	−.240
R_aINS- to-R_pINS	0.50 (0.31)	−.139	−.039	.103	0.52 (0.30)	−.262	−.052	.036	0.02 (0.28)	.560**	.009	−.030
R_aINS- to-R_SI	0.13 (0.16)	−.016	.016	.276	0.13 (0.22)	−.239	.004	−.317	−0.00 (0.28)	.093	−.208	−.300
R_aINS- to-R_SII	0.34 (0.21)	.058	−.098	.055	0.35 (0.29)	−.338	−.094	−.126	0.01 (0.27)	.174	.287	−.204
R_aINS- to-R_THA	0.12 (0.16)	−.232	−.028	.134	0.04 (0.14)	−.210	.026	.147	−0.08 (0.20)	.192	−.078	.164
R_aINS- to-R_PAG	0.09 (0.16)	−.196	.042	.225	0.11 (0.20)	−.449*	.121	.168	0.01 (0.16)	.226	.204	−.089
R_pINS- to-R_SI	0.34 (0.25)	−.097	−.314	.102	0.28 (0.21)	.105	−.101	−.351	−0.05 (0.27)	−.300	−.417*	−.047
R_pINS- to-R_SII	0.60 (0.26)	−.076	.012	.313	0.61 (0.25)	−.015	.170	.009	0.01 (0.25)	−.031	−.115	−.219
R_pINS- to-R_THA	0.23 (0.16)	.214	.026	−.008	0.25 (0.18)	−.383	.104	.111	0.02 (0.24)	−.070	.000	−.092
R_pINS- to-R_PAG	0.12 (0.20)	.006	−.093	.215	0.15 (0.22)	−.221	−.150	−.267	0.03 (0.22)	.099	−.124	.134
R_SI-to- R_SII	0.45 (0.22)	.114	.234	.254	0.48 (0.25)	.067	−.056	−.257	0.03 (0.27)	−.046	−.312	.090
R_SI-to- R_THA	0.06 (0.18)	.069	.216	.177	0.04 (0.27)	.366	−.314	−.059	−0.02 (0.30)	−.178	.172	.156
R_SI-to- R_PAG	0.09 (0.22)	.138	.289	.359	0.09 (0.14)	.199	−.531**	−.390	0.00 (0.25)	.001	−.093	−.120
R_SII-to- R_THA	0.17 (0.21)	.271	−.020	.069	0.13 (0.13)	.199	−.531	−.390	−0.04 (0.24)	−.015	−.159	.076
R_SII-to- R_PAG	0.14 (0.24)	.153	.060	.317	0.12 (0.18)	.070	−.218	−.112	−0.02 (0.25)	.055	−.420*	−.009
R_THA- to-R_PAG	0.16 (0.19)	−.004	.134	.339	0.15 (0.16)	−.214	.029	.060	−0.00 (0.25)	.049	−.148	.204

^*- p < 0.05;

^**- p < 0.01; (ROI) = region on interest; (Pre) = prioR_to MBB therapy; (Post) = following MBB therapy; (FC) = functional connectivity, values = mean fisher transformed zero-order estimate (standard deviation); (ρ) = Pearson product-moment correlation coefficient; (PPT) = pressure pain thresholds; (Δ) = change score (post minus pre); (L) = left; (R) = right; (ACC) = anterior cingulate cortex; (PCC) = posterior cingulate cortex; (aINS) = anterior insular cortex; (pINS) = posterior cingulate cortex; (SI) = primary somatosensory cortex; (SII) = secondary somatosensory cortex; (THA) = thalamus; (PAG) = periaqueductal grey.

Table 4.

Left Hemisphere ROI to ROI connections

ROI-to- ROI Pair	(Pre) FC	ρ (Pre) Pain Intensity	ρ (Pre) Local PPT	ρ (Pre) Remote PPT	(Post) FC	ρ (Post) Pain Intensity	ρ (Post) Local PPT	ρ (Post) Remote PPT	Δ FC	ρ Δ Pain Intensity	ρ Δ Local PPT	ρ Δ Remote PPT
L_ ACC- to- L_ PCC	0.18 (0.26)	−0.09	0.32	0.59	0.17 (0.22)	0.13	0.20	0.17	−0.01 (0.27)	−.020	−.232	.229
L_ ACC- to- L_ aINS	0.30 (0.21)	0.02	0.39	0.38	0.33 (0.24)	0.09	0.05	−0.09	0.03 (0.24)	.097	−.149	−.278
L_ ACC- to- L__pINS	0.29 (0.29)	0.06	0.02	0.34	0.23 (0.21)	−0.29	0.42	0.28	−0.06 (0.30)	−.041	.123	−.082
L_ ACC- to-L_ SI	0.25 (0.26)	0.08	0.08	0.21	0.18 (0.22)	−0.19	0.10	−0.04	−0.07 (0.26)	.052	.007	.064
L_ ACC- to-L_ SII	0.34 (0.19)	−0.18	0.18	0.36	0.37 (0.24)	−0.10	0.01	−0.15	0.03 (0.29)	−.224	.193	.078
L_ ACC- to- L_ THA	0.14 (0.24)	0.05	−0.01	0.09	0.25 (0.16)	0.48	−0.37	−0.18	0.11 (0.25)	−.028	−.030	.100
L_ ACC- to- L_ PAG	0.12 (0.22)	−0.05	0.02	0.33	0.22 (0.22)	0.17	0.02	0.09	0.09 (0.30)	−.174	−.135	.223
L_ PCC- to- L_ aINS	−0.02 (0.17)	0.14	0.13	0.30	0.15 (0.21)	0.05	−0.01	−0.01	0.17 (0.23)	0.08	−0.17	0.07
L_ PCC- to- L__pINS	0.11 (0.22)	−0.16	0.24	0.61	0.16 (0.17)	−0.12	−0.24	−0.17	0.05 (0.27)	−0.13	−0.23	0.38
L_ PCC- to-L_ SI	0.15 (0.27)	−.231	.131	.357	0.12 (0.25)	−.131	−.246	−.140	−0.04 (0.34)	−0.07	−0.28	0.37
L_ PCC- to-L_ SII	0.12 (0.31)	−0.15	0.32	0.57	0.11 (0.21)	0.11	−0.06	−0.08	−0.01 (0.33)	−0.23	−0.36	0.38
L_ PCC- to- L_ THA	0.12 (0.27)	0.20	−0.01	0.11	0.23 (0.20)	0.17	0.03	−0.06	0.11 (0.29)	0.29	0.02	−0.11
L_ PCC- to- L_ PAG	0.23 (0.24)	0.30	0.17	0.19	0.26 (0.17)	0.29	0.14	−0.01	0.04 (0.27)	0.24	−0.04	−0.18
L_ aINS- to- L__pINS	0.62 (0.29)	−.244	.007	.034	0.54 (0.31)	−.305	.234	.240	−0.08 (0.28)	−.167	.033	−.268
L_ aINS- to-L_ SI	0.18 (0.21)	−.167	.192	.133	0.18 (0.26)	.021	−.060	−.191	−0.00 (0.32)	−.209	−.285	−.201
L_ aINS- to-L_ SII	0.39 (0.23)	.251	−.056	−.154	0.29 (0.29)	−.114	.005	.035	−0.10 (0.35)	.066	.370	−.063
L_ aINS- to- L_ THA	0.04 (0.18)	.221	−.010	.119	0.07 (0.21)	.065	.011	−.051	0.03 (0.25)	.146	−.007	−.293
L_ aINS- to- L_ PAG	0.05 (0.17)	.115	.215	.409*	0.17 (0.20)	−.065	.094	.006	0.12 (0.26)	−.081	−.390	−.205
L__pINS- to-L_ SI	0.35 (0.28)	−.029	.214	.273	0.35 (0.27)	.047	.255	−.003	−0.01 (0.31)	.207	−.189	−.065
L__pINS- to-L_ SII	0.61 (0.26)	−.186	.439*	.324	0.49 (0.30)	.036	.413*	.182	−0.12 (0.25)	.008	.208	.097
L__pINS- to- L_ THA	0.04 (0.26)	.129	−.089	−.001	0.14 (0.24)	.373	−.206	.000	0.10 (0.29)	−.047	.101	.058
L__pINS- to- L_ PAG	0.03 (0.21)	.106	−.053	.345	0.17 (0.21)	−.093	.060	.203	0.13 (0.21)	.052	−.244	−.210
L_ SI-to- L_ SII	0.48 (0.31)	−.077	.299	.400	0.37 (0.31)	−.051	.220	.120	−0.12 (0.33)	.332	−.196	−.038
L_ SI-to- L_ THA	−0.06 (0.24)	.230	−.006	.127	0.01 (0.20)	.489*	−.225	−.045	0.07 (0.30)	.294	−.042	.220
L_ SI-to- L_ PAG	0.02 (0.21)	.079	.025	.306	0.10 (0.16)	.057	−.242	−.128	0.08 (0.22)	−.065	−.388	−.109
L_ SII-to- L_ THA	0.08 (0.24)	.045	−.006	−.078	0.15 (0.20)	.687**	−.348	−.098	0.06 (0.28)	−.283	.164	.222
L_ SII-to- L_ PAG	0.03 (0.22)	.046	−.027	.298	0.11 (0.19)	.173	.055	.092	0.08 (0.23)	−.201	−.241	.246
L_ THA- to- L_ PAG	0.32 (0.24)	.064	.080	.177	0.32 (0.16)	.152	−.117	−.033	0.00 (0.29)	.425*	−.184	.202

^*- p < 0.05;

^**- p < 0.01; (ROI) = region on interest; (Pre) = prior to MBB therapy; (Post) = following MBB therapy; (FC) = functional connectivity, values = mean fisher transformed zero-order estimate (standard deviation); (ρ) = Pearson product-moment correlation coefficient; (PPT) = pressure pain thresholds; (Δ) = change score (post minus pre); (L) = left; (R) = right; (ACC) = anterior cingulate cortex; (PCC) = posterior cingulate cortex; (aINS) = anterior insular cortex; (pINS) = posterior cingulate cortex; (SI) = primary somatosensory cortex; (SII) = secondary somatosensory cortex; (THA) = thalamus; (PAG) = periaqueductal grey.

Table 5.

Cross Hemisphere ROI to ROI connections

ROI-to- ROI Pair	(Pre) FC	ρ (Pre) Pain Intensity	ρ (Pre) Local PPT	ρ (Pre) Remote PPT	(Post) FC	ρ (Post) Pain Intensity	ρ (Post) Local PPT	ρ (Post) Remote PPT	Δ FC	ρ Δ Pain Intensity	ρ Δ Local PPT	ρ Δ Remote PPT
R_ACC-to- L_ACC	2.58 (0.25)	.065	.286	.110	2.60 (0.34)	.171	−.115	−.123	0.02 (0.35)	−.083	−.121	.030
R_ACC-to- L_PCC	0.18 (0.26)	−.074	.309	.551**	0.18 (0.24)	.140	.197	.185	−0.00 (0.28)	−.002	−.183	.174
R_ACC-to- L_aINS	0.31 (0.22)	.018	.387	.367	0.33 (0.25)	.083	.050	−.048	0.03 (0.24)	.070	−.158	−.279
R_ACC-to- L_pINS	0.29 (0.29)	.039	.026	.309	0.24 (0.21)	−.336	.423*	.258	−0.05 (0.31)	−.050	.111	−.091
R_ACC-to- L_SI	0.26 (0.26)	.059	.103	.222	0.19 (0.22)	−.207	.103	−.040	−0.07 (0.26)	.043	−.011	.060
R_ACC-to- L_SII	0.34 (0.20)	−.171	.169	.337	0.38 (0.25)	−.124	−.008	−.162	0.04 (0.29)	−.204	.199	.028
R_ACC-to- L_THA	0.13 (0.24)	−.011	.035	.135	25 (0.16)	.449*	−.348	−.179	0.12 (0.26)	−.051	−.065	.114
R_ACC-to- L_PAG	0.12 (0.23)	−.022	.018	.354	0.22 (0.23)	.145	.038	.138	0.10 (0.31)	−.143	−.139	.216
R_PCC-to- L_ACC	0.26 (0.28)	.033	.302	.506*	0.17 (0.24)	.277	.073	.197	−0.09 (0.30)	−.014	−.154	.168
R_PCC-to- L_PCC	1.32 (0.20)	−.027	.067	.019	1.28 (0.25)	−.041	−.358	−.441*	−0.04 (0.26)	.003	.043	−.153
R_PCC-to- L_aINS	−0.02 (0.20)	.132	.185	.408*	0.09 (0.20)	.148	.017	.055	0.11 (0.22)	−.048	−.132	.072
R_PCC-to- L_pINS	0.06 (0.24)	−.031	.191	.555**	0.12 (0.17)	−.126	−.136	−.117	0.06 (0.24)	.027	−.104	.273
R_PCC-to- L_SI	0.12 (0.30)	−.186	.187	.243	0.12 (0.21)	−.141	−.185	−.087	0.00 (0.34)	−.104	−.232	.325
R_PCC-to- L_SII	0.10 (031)	−.078	.282	.490*	0.09 (0.20)	.101	−.124	−.080	−0.01 (0.32)	−.225	−.274	.332
R_PCC-to- L_THA	0.12 (0.30)	.275	−.112	.065	0.23 (0.25)	.174	.137	.165	0.10 (0.27)	.265	.025	−.138
R_PCC-to- L_PAG	0.24 (0.25)	.301	.020	.087	0.26 (0.16)	.145	.139	.213	0.02 (0.23)	.199	.029	−.228
R_aINS-to- L_ACC	0.25 (0.23)	.206	.291	.374	0.28 (0.22)	−.414*	.145	.066	0.03 (0.29)	.138	.120	−.114
R_aINS-to- L_PCC	−0.04 (0.14)	.205	.181	.386	−0.00 (0.17)	−.159	.039	−.058	0.04 (0.18)	.019	−.099	−.100
R_aINS-to- L_aINS	0.54 (0.27)	.176	.225	.357	0.48 (0.30)	.004	.251	.258	−0.06 (0.30)	−.106	−.270	.088
R_aINS-to- L_pINS	0.26 (0.30)	−.018	.103	.149	0.30 (0.21)	−.391	.243	.282	0.03 (0.27)	.338	−.079	−.253
R_aINS-to- L_SI	0.13 (0.19)	−.043	.308	.328	0.07 (0.20)	−.202	.064	−.241	−0.05 (0.28)	−.160	−.145	−.373
R_aINS-to- L_SII	0.21 (0.23)	.213	.028	.126	0.17 (0.25)	−.476*	.199	−.025	−0.04 (0.34)	.098	.263	−.263
R_aINS-to- L_THA	0.02 (0.21)	.167	−.242	−.101	−0.03 (0.18)	−.463*	.141	−.045	−0.06 (0.27)	.221	.222	−.005
R_aINS-to- L_PAG	0.06 (0.19)	−.015	.100	.320	0.09 (0.20)	−.631**	.266	.181	0.03 (0.18)	.316	−.011	−.145
R_pINS-to- L_ACC	0.28 (0.22)	.237	−.110	.180	0.21 (0.27)	−.386	.007	−.021	−0.07 (0.30)	.170	.231	−.201
R_pINS-to- L_PCC	0.15 (0.21)	.044	.159	.459*	0.08 (0.22)	−.168	−.169	−.168	−0.07 (0.30)	−.036	−.154	.123
R_pINS-to- L_aINS	0.29 (0.19)	.114	.170	.200	0.18 (0.27)	−.252	.028	.002	−0.11 (0.30)	.390	.076	−.304
R_pINS-to- L_pINS	0.54 (0.31)	.018	.365	.292	0.50 (0.29)	.284	.028	.043	−0.04 (0.36)	.125	.065	.016
R_pINS-to- L_SI	0.36 (0.26)	−.197	.089	.089	0.20 (0.26)	.040	.171	−.056	−0.16 (0.26)	.004	−.188	−.296
R_pINS-to- L_SII	0.43 (0.34)	−.147	.200	.127	0.39 (0.27)	.034	.194	.093	−0.05 (0.28)	−.341	−.048	−.308
R_pINS-to- L_THA	0.01 (0.20)	.197	−.098	−.174	0.03 (0.20)	.016	−.199	−.197	0.02 (0.27)	.058	.212	.118
R_pINS-to- L_PAG	0.10 (0.21)	.301	−.258	.139	0.13 (0.22)	−.220	−.053	−.147	0.03 (0.24)	.272	−.118	−.034
R_SI-to- L_ACC	0.28 (0.26)	.064	.167	.287	0.24 (0.23)	−.212	.126	.036	−0.04 (0.27)	−.061	.129	.041
R_SI-to- L_PCC	0.16 (0.28)	−.058	.274	.619**	0.06 (0.26)	.048	−.001	.169	−0.09 (0.35)	−.015	−.339	.478*
R_SI-to- L_aINS	0.15 (0.20)	−.148	.170	.290	0.12 (0.21)	−.152	.108	−.034	−0.03 (0.29)	−.090	−.130	−.137
R-SI-to-	0.29 (0.17)	.012	.000	.288	0.30 (0.25)	−.075	.269	−.115	0.00	.073	.001	−.348
toL_pINS									(0.28)
R_SI-to- L_SI	0.65 (0.29)	−.029	.254	.321	0.66 (0.30)	−.230	.286	.092	0.01 (0.29)	.241	−.035	−.168
R_SI-to- L_ SII	0.42 (0.24)	−.183	.301	.333	0.44 (0.28)	−.017	.175	.118	0.01 (0.26)	−.157	−.259	.109
R_SI-to- L_THA	−0.03 (0.21)	.188	−.026	.012	0.03 (0.21)	.312	−.102	.239	0.05 (0.25)	.000	.258	.224
R_SI-to- L_ PAG	0.03 (0.21)	.011	.310	.340	0.13 (0.15)	.144	−.073	.071	0.1 (0.23)	−.067	−.004	−.042
R_SII-to- L_ACC	0.31 (0.22)	.011	.365	.633**	0.30 (0.23)	−.392	.010	−.078	−0.02 (0.33)	−.095	.049	.021
R_SII-to- L_PCC	0.16 (0.30)	−.047	.383	.570**	0.09 (0.24)	−.115	.046	.086	−0.06 (0.33)	−.001	−.253	.213
R_SII-to- L_aINS	0.26 (0.19)	.141	.095	.106	0.15 (0.25)	−.387	.090	−.044	−0.11 (0.28)	.044	.146	−.163
R_SII-to- L__pINS	0.40 (0.20)	.094	.333	.374	0.36 (0.24)	−.205	.195	−.104	−0.04 (.025)	.271	.041	−.234
R_SII-to- L_SI	0.38 (0.19)	−.202	.265	.131	0.32 (0.22)	−.429*	.265	−.148	−0.07 (0.28)	.027	−.035	−.335
R_SII-to- L_SII	0.63 (0.29)	.157	.412*	.189	0.54 (0.31)	−.193	.400	.138	−0.09 (0.34)	−.273	.127	−.349
R_SII-to- L_THA	0.07 (0.22)	.287	−.078	−.062	0.01 (0.20)	.055	−.338	−.198	0.05 (0.26)	−.040	.110	.039
R_SII-to- L_PAG	0.12 (0.23)	.204	−.033	.129	0.11 (0.18)	−.025	.089	.173	−0.01 (0.25)	.090	−.189	−.083
R_THA-to- L_ACC	0.08 (0.19)	.117	.115	.089	0.13 (0.18)	.313	−.326	−.048	0.05 (0.26)	.052	−.039	−.305
R_THA-to- L_PCC	0.14 (0.24)	.082	.058	.170	0.12 (0.19)	−.222	.217	.135	−0.01 (030)	.426*	.046	−.261
R_THA-to- L_aINS	0.04 (0.18)	.078	.216	.158	−0.01 (0.15)	−.119	.129	.116	−0.05 (0.26)	.221	.254	−.319
R_THA-to- L_pINS	0.05 (0.19)	.368	.035	.060	0.07 (0.19)	−.058	.080	.163	0.03 (0.22)	.225	.461*	−.375
R_THA-to- L_SI	0.02 (0.18)	.244	.398	.344	−0.01 (0.15)	.324	−.124	.003	−0.03 (0.23)	.298	.335	.036
R_THA-to- L_SII	0.03 (0.20)	.194	.184	.159	0.07 (0.20)	.180	.043	.096	0.04 (0.28)	.087	.189	−.188
R_THA-to- L_THA	0.47 (0.22)	−.146	.027	.224	0.48 (0.20)	−.411*	.252	−.009	0.01 (0.25)	.265	.269	−.259
R_THA-to- L_PAG	0.16 (0.21)	.100	−.095	.407*	0.16 (0.20)	−.006	.097	.249	0.00 (0.27)	.080	−.198	.279
R_PAG-to- L_ACC	0.18 (0.22)	.065	.028	.269	0.19 (0.16)	−.067	−.007	.030	0.02 (0.27)	.164	−.077	.049
R_PAG-to- L_PCC	0.19 (0.26)	.327	.134	.341	0.22 (0.16)	−.011	.216	−.134	0.03 (0.30)	.265	−.237	−.210
R_PAG-to- L_aINS	0.07 (0.20)	−.124	.250	.279	0.15 (0.20)	−.190	−.043	−.095	0.08 (0.26)	−.058	−.182	−.267
R_PAG-to- L_pINS	0.04 (0.20)	−.145	.123	.481*	0.14 (0.25)	−.033	−.246	−.113	0.10 (0.26)	−.160	−.204	.123
R_PAG-to- L_SI	0.06 (0.25)	.132	.076	.327	0.08 (0.18)	−.017	−.449*	−.395	0.01 (0.27)	−.003	−.329	−.120
R_PAG-to- L_SII	0.09 (0.22)	−.053	.006	.356	0.08 (0.15)	.038	−.313	−.296	−0.01 (0.23)	−.240	−.369	.208
R_PAG-to- L_THA	0.28 (0.21)	−.060	.065	.132	0.27 (0.17)	−.050	−.058	.011	−0.01 (0.26)	.297	−.155	.187
R_PAG-to- L_PAG	1.00 (0.28)	.034	−.103	.050	1.10 (0.26)	−.051	−.307	−.139	0.08 (0.34)	−.167	−.168	−.105

^*- p < 0.05;

^**- p < 0.01; (ROI) = region on interest; (Pre) = prior to MBB therapy; (Post) = following MBB therapy; (FC) = functional connectivity, values = mean fisher transformed zero-order estimate (standard deviation); (ρ) = Pearson product-moment correlation coefficient; (PPT) = pressure pain thresholds; (Δ) = change score (post minus pre); (L) = left; (R) = right; (ACC) = anterior cingulate cortex; (PCC) = posterior cingulate cortex; (aINS) = anterior insular cortex; (pINS) = posterior cingulate cortex; (SI) = primary somatosensory cortex; (SII) = secondary somatosensory cortex; (THA) = thalamus; (PAG) = periaqueductal grey.

Primary Outcome Measure

Common FC changes for all MT groups

The RM-ANOVAs found a main effect for time in two left-hemisphere connections and one cross-hemisphere connection, See Figure 3. In the left-hemisphere the connections between the PCC and aINS (p = 0.001) and pINS and PAG (p = 0.005) significantly changed over time. Prior to MT, the left PCC and left aINS showed a weak inverse (negative) relationship (FC = −0.02, SD=0.17). Following MT, the relationship flipped; showing a weak positive relationship (FC = 0.15, SD = 0.21). The relationship between the left pINS and left PAG showed an overall increase over time going from 0.03 (SD = 0.21) to 0.17 (0.21). The cross-hemisphere connection between the left SI and right pINS decreased over time (p = 0.005). These regions shared a moderately strong positive relationship prior to intervention (FC = 0.36, SD = 0.26) that became weaker overtime (FC = 0.20, SD = 0.26).

Figure 3.

A. Following manual therapy (MT), a decrease in functional connectivity (FC) between the left somatosensory cortex (SI) and the right posterior insula (pINS) was observed. B. The left anterior insula (aINS) and left posterior cingulate cortex (PCC) showed increased FC over time. C. Following MT, the left posterior insula (pINS) and left periaqueductal grey (PAG) showed increased FC

Treatment-dependent FC changes

Two right-hemisphere and one cross-hemisphere ROI-to-ROI connections demonstrated significant (p < 0.01) treatment-dependent differences in the rate of FC change, see Figure 4. In the right-hemisphere, the strength of the positive connection between SI and aINS increased in the SMT group (ΔFC=0.28), while deceased in the MOB (ΔFC = −0.04), and TT (ΔFC = −0.14), groups. Also in the right-hemisphere, the strength between SI and PAG increased in the SMT (ΔFC = 0.04) and MOB (ΔFC = 0.19) groups, while decreasing in the TT group (ΔFC = −0.17). The cross-hemisphere connection between the right aINS and left PCC increased in the SMT (ΔFC = 0.21) and MOB (ΔFC = 0.05) groups, while decreasing the TT group (ΔFC = −0.06).

Figure 4.

A. The change in functional connectivity (FC)between the right anterior insula (aINS) and right somatosensory cortex (SI) differed across the three MT groups. The spinal manipulative therapy (SMT) group showed an increase (Δ = 0.28), while the spinal mobilization group (MOB) (Δ = −0.04) and the therapeutic touch group (TT) (Δ = −0.14) showed decreases in FC. B. The FC changes between the right anterior insula (aINS) and left posterior cingulate cortex (PCC) differed across the three MT groups. The SMT group (Δ = 0.21) and the MOB (Δ = 0.05) showed an increases, while the TT group (Δ = −0.06) showed a decrease. C. The FC changes between the right somatosensory cortex (SI) and right periaqueductal grey (PAG) increased in the SMT (Δ = 0.04) and MOB (Δ = 0.19) groups, while FC decreased in the TT group (Δ = −0.17).

Secondary Outcome Measure

On average, the pain intensity in our sample significantly decreased overtime, mean pain reduction = 3.88, 95% CI = (0.98,6.77), (F_1,21 = 8.60, p < 0.01), regardless of the particular intervention received. No significant differences were observed in change of pain intensity across the MT groups as evidenced by group by time interaction. (F_2,21 = 0.77, p = 0.50). No significant change was observed in local or remote pressure pain sensitivity over time or for particular MT treatments [mean local PPT change = −0.02, 95% CI = (−1.57,1.52); mean remote PPT change = −0.17, 95% CI = (−1.41,1.08), (p > 0.05)].

Discussion

This study assessed the relationship of brain activity between regions of the PPN before and after MT. Using this approach we found common and treatment-dependent changes in FC. These results provide further support to speculations that neurophysiological mechanisms may be involved in the clinical benefit following manual therapy. [15] Current speculation contends that the biomechanical aspect of MT induces neurophysiological effects that underlie clinical improvement. Our study is unique in our neurophysiological measure as we used resting-state fMRI in conjunction with functional connectivity analyses. Our results are in agreement with studies that have found immediate changes using other neurophysiological outcomes, such as Hoffman-reflex and motor-neuron excitability, electroencephalography with somatosensory evoked potentials, transcranial magnetic stimulation with motor evoked potentials, and task-based fMRI with peak BOLD response.[17, 46–49]

Directly comparing our results with that of the only other fMRI study is difficult because of the different estimates of cortical function. We used functional connectivity estimated at rest, while they used peak BOLD activity while processing a thermal-stimulus-task. Despite these differences, the results from these studies do show similarities. For example, they found reduced activity within the insular cortex, somatosensory cortices and periaqueductal grey. We found the interrelationship between some of these regions changed in the left hemisphere following MT. We also found treatment-dependent changes in the interrelationship between similar regions in the right hemisphere. The ROI-to-ROI FC changes that were common to all three interventions (ie main effect of time) may represent shared neurophysiological mechanisms. However, we cannot conclusively assume these common shifts are more than natural history or that they are specific to a MT intervention, as these potential confounds were not controlled for in the study design. The timing of these changes (ie less than 1 hour) could be viewed as associative, but more work is needed in this area.

The treatment-dependent changes in ROI-to-ROI FC may be reflective of differing biomechanical attributes of the treatments. The three manual therapy techniques used in our study differ in their force-duration profiles. Research has shown differences in the biomechanical and neurophysiological responses that are force-duration dependent. [50–54] Although the three MT interventions showed similar pain relief in our study, there remains conflicting evidence as to the superiority of clinical outcomes following MT techniques with differing forcer-duration attributes.[3, 29, 55, 56]

We found a significant reduction in exercised-induced pain intensity following MT. Our findings are consistent with a prior study that reported positive effects of MT on exercised induced myalgia in the extensor muscles of the wrist. [57] In that study, there was a improvement in pain intensity with stretch seen in the two MT treatments over a no-treatment group. There was no difference between the two MT treatments. A reduction in resting pain intensity was seen over time; however there was no difference between MT treatment groups and the no-treatment group. Since we did not include a no-treatment group to compare, caution is needed when attributing the pain relief in our study to the intervention.

We did not find immediate changes in remote or local pressure pain sensitivity. This is contrary to the previous study using MT on exercise induced myalgia in the wrist and recent systematic reviews looking at the immediate effect of MT on pressure pain sensitivity. [1, 4, 57] We suggest our sample size is one possible reason for our results conflicting with previous findings. Given our relatively small sample size, (n=24), small main effects (eta = 0.24) effects are likely to be obscured.

Limitations and Future Studies

Without a natural history group, attributing any change solely to the intervention is unsubstantiated at this point as all of our groups improved equally over time. Additional analyses are needed to disentangle treatment specific and natural improvement, as our exercise induced injury model has a favorable prognosis and resolution is expected naturally with time. Caution is needed when applying the findings in our model to chronic conditions because the acute pain and pain sensitivity increases in our model resolve quickly over time, where as in chronic pain, it does not.

In this study we did not control for the variability between providers, nor did we look for an interventionist effect within the analyses. However, we feel this also strengthens our study as this increases the ability to generalize the mechanism beyond a single provider and increases the likelihood of findings can be replicated.

Although we found changes in functional connectivity, the within person relationship across the primary and secondary outcomes were weak and inconsistent. Thus, interpretation of these findings to represent a mechanism of pain relief currently cannot be made. There is, however some evidence that suggests the functional connectivity between brain regions prior to a stimulus influences the perception of pain.[36, 58, 59] Those studies have shown that (1) the functional connectivity between the anterior insular cortex and the periaqueductal gray (PAG), measured prior to the evoked stimulus, was predictive of pain perception and (2) prior to the evoked stimulus, fluctuations in functional connectivity between regions in the classic ‘pain network’ and brain areas related to attention, emotion, and descending pain modulation sub serve contextual modulations of pain intensity. In this context, our findings that show functional connectivity changes between the insula and somatosensory cortex and periaqueductal grey may sub serve the subsequent reductions in peak BOLD while processing stimuli.

The involvement of the insular cortex in our study is consistent with the close association between activity in this region and the subjective experience of pain. This brain region is extensively interconnected with various other brain regions and integrates sensory and contextual information to generate a higher order representation of interoception.[60]

Our ROI-to-ROI approach has several limitations. On the positive side, this approach allows for a priori ROIs to be investigated. However, numerous brain regions were not included such as prefrontal, motor, reward and aversion regions. Since measuring change in FC is a novel approach, futures studies may consider a more exploratory approach where only a single seed region is used to identify other relevant brain regions. However, this approach is limited by the selection of the seed region. There are consequences to increasing the number of brain regions. With our limited ROI approach, 8 ROI bilateral, approximately 120 ROI-to-ROI comparisons we generated. We used a liberal correction, p<0.01, for the number of planned comparisons. This liberal correction does increase the likely hood that some of our findings may be false positives. Thus achieving the correct balance between the number of ROIs and correction for the number of comparisons needs to be taken into account.

Our sample size, (n=24), which may obscure moderate to small group by time interactions. With our estimated partial eta² (η²=0.068) for the 3 group by 2 time-point interaction we would need a sample of 133 subjects to be properly powered. We suggest that the sample size is a possible reason for our results that conflict with previous published findings that suggest there are differential MT effects. As well, we investigated only the immediate effects, so extrapolating our findings beyond this cannot be made.

This study has implications for future mechanistic research. In addition to disentangling common versus shared cortical pathways, future research can build upon our results by using more sophisticated network modeling approaches. We found changes between sensory discriminate and sensory affective ROIs, as well as between these ROIs and pain modulatory brain areas. Our findings show that changes, albeit modest, can be expected in the interrelationships between these brain regions. Network modeling approaches, such as functional network connectivity and dynamic causal modeling, may provide additional insight into the central modulatory effects of MT therapy on cortical function. Continued research in this area is needed to address the extent to which these changes underlie pain-relief.

Conclusion

This mechanistic study identified brain regions of the PPN where FC changed immediately following 3 different types of MT. Neurophysiological changes following MT may be an underlying mechanism of pain relief.

Practical Applications.

• Resting-State fMRI in conjunction with functional connectivity analyses demonstrate common and treatment-dependent changes following 3 different types of manual therapy.

• The intensity of experimentally induced myalgia is reduced following manual therapy

• Neurophysiological changes following MT may be an underlying mechanism of pain relief.