Fourty participants (21.3 $\pm$ 2.33 years of age) who attended Ontario Tech University (Oshawa, ON, Canada) between the ages of 18 and 30 were eligible to participate in this quasi-experimental (pre/post) study. Twenty SCNP participants and 20 non-SCNP participants performing a novel force matching tracking task (FMTT) with changes in brain plasticity measured before and after task performance using SEPs. All participants were right-hand dominant confirmed using The Edinburgh Handedness Inventory (EHI), scored $>$+40. Non-SCNP participants were to have a grade of 0 (pain-free) and SCNP participants were to have a grade of I (low disability-low intensity), II (low disability-high intensity) or III (high disability-moderately limiting) on the Von Korff Chronic Pain Grade Scale, as this has been used to discriminate pain severity, and evaluate pain intensity and degree of pain-related disability over the span of 6 months [41]. None of the participants had any neurological conditions that could have led to cognitive deficits or alterations in central processing, nor did they report having tension headaches or co-morbidities (e.g., fibromyalgia). Written and oral informed consent was obtained on the day of the collection. The Ontario Tech University’s Research Ethics Board approved this study (File #: 14686), and the study was conducted in accordance with their guidelines.

Bipolar surface electromyography (EMG) electrodes (Kendall™ Medi-Trace®, Mansfield, MA, USA) were placed on the skin overlying the median nerve, to administer an electrical stimulation at 2.47 Hz and 4.98 Hz. The anode EMG electrode was placed close to the distal crease of the right wrist while the cathode EMG electrode was placed adjacent to the anode electrode [27, 49]. Two different stimulating frequencies were used as 2.47 Hz results in clear visualization of the N30 SEP peak amplitude (without attenuating the signal), while a 4.98 Hz results in the identification of the N24 SEP peak (with attenuation of the N30 SEP peak) [50]. A stimulation intensity that consistently induced a twitch of the right thumb (~1 cm deviation from anatomical position) was administered [49].

Bipolar surface EMG electrodes were placed on non-cephalic sites, according to the International Federation of Clinical Neurophysiologists (IFCN) guidelines [49], in order to record the N9, N11 and N13 (peripheral SEP peaks). The N9 SEP peak was recorded over the ipsilateral Erb’s point, and referenced to the ipsilateral ear lobe [27, 49, 51]. The spinal SEP peaks (N11 and N13) were recorded over the spinous process of the fifth cervical vertebrae, with the reference being the anterior tracheal cartilage of the neck [27, 49]. These electrodes were referenced to the left clavicular bone, common ground electrode. Impendences of these electrodes were kept below 5 k$\mathrm{\Omega }$. Central SEP peaks (N18, N20, P25, N24, N30 and N60) were recorded using a Waveguard™ 64-electrode encephalography (EEG) cap (ANT Neuro, Hengelo, Netherlands), which was placed according to the 10–20 international system as per IFCN guidelines [49, 52]. The common average of the electrical signal was used for all electrodes on the EEG cap. The electrical impendences were kept below 10 k$\mathrm{\Omega }$ for the 64-channel EEG cap.

Throughout the SEPs collection, participants maintained an upright seated posture, backs supported by the chair, their feet on the floor, and their right arm rested in a supinated position on an adjustable-height table. To reduce the impact of movement artifacts on EEG recordings, participants were asked to minimize excessive movements or blinking.

The SEP signal acquired from non-cephalic electrodes was amplified by a gain 10,000$\times$ during preprocessing using Signal® software (Cambridge Electronic Design, Cambridge, UK). The SEP signal acquired from the cephalic electrodes was amplified by a gain of 40,000$\times$ amplification during the post-processing in Advanced Source Analysis (ASA) Software (ASA™; Hengelo, Netherlands). Artifacts (e.g., eye blinks, ocular movement, and muscular activity) were removed from the EEG data to extract the cortical SEPs by averaging epochs that were time locked to the median nerve stimulation. An average of 1000 right median nerve stimulus responses elicited SEPs, as per IFCN guidelines [49]. The SEP signal was filtered using a 0.2 Hz to 1000 Hz band-pass filter as it has been found to lead to reproducible SEP waveforms [33].

Participants attended two data collection sessions. Dual SEPs (2.47 Hz and 4.98 Hz) were collected prior and after acquiring the FMTT for each participant, during the first data collection session. It took 15 minutes to undergo each dual SEP measure, ~10 minutes for 2.47 Hz and ~5 minutes for 4.98 Hz. Retention of the performed task was measured 24 to 48 hours later, considered second session. The experimental sequence of this study can be seen in Fig. 1.

Fig. 1.

Experimental flow. SEP, somatosensory evoked potential; FMTT, Force Matching Tracking Task.

Motor Task: Force Matching Tracking Task (FMTT)

A series of force traces that were calibrated to the individual’s right thenar muscles (based on average of three maximum voluntary contractions (MVCs)) were presented on a screen. Participants were to push or flex and abduct their right thumb against a 50 kg load cell force transducer (Unbranded, Shenzhen, Guangdong, China) to match the force traces, see Fig. 2 for participant setup of FMTT. As described in our recent publication, red bars were positioned 0.05% above and below the white-dotted force traces, used as cues for participants to replicate the trace.These traces were presented on a monitor as one trial comprising of two 10-second-long force traces. The tracking performance and presentation of traces were managed via a custom-made LABVIEW software program (National Instruments, Austin, TX, USA). Throughout the task, participants received augmented feedback in the form of a yellow line that depicted their force exertion against the transducer while the target trace remained stationary [38].

Fig. 2.

Experimental set-up for the completion of the FMTT. Red arrow shows that the direction in which the thumb is pushing against the transducer. FMTT, force matching tracking task.

To acquaint participants with the task prior to initiating pre-acquisition and subsequent phases, they completed two practice trials using their right thumb. Participants were to match four blocks of force traces, where there was a variation in force amplitude (2% to 12% of averaged MVC), and isometric hold durations (1 to 2.75 seconds). Force amplitude, hold duration, and traces were randomized within each block, and between the four blocks. A rest interval with a work-to-rest ratio of 0.50 was embedded within each trial, to prevent and reduce likelihood of developing muscle fatigue [53]. For the acquisition phase, each block of force traces was performed three times, and post-acquisition measures consisted of the same four blocks, which took 20 minutes to complete. The same four traces were completed at retention, measured 24 to 48 hours later.

All statistical tests were performed using SPSS version 26 (IBM Corp., Armonk, NY, USA) [55] and statistical significance was set at p $\le$ 0.05. SPSS uses Tukey’s method to identify outliers and extreme outliers [55]. Outliers are determined on whether or not the value is outside of 1st or 3rd quartile of a boxplot, with a $\pm$ 1.5 * interquartile range for outliers and $\pm$ 3 * interquartile range for extreme outliers [55]. Percent of data rejected due to outliers of the SEPs data were: 10% for N11 (4 SCNP), 5% for N13 (2 SCNP) and N24 (1 SCNP and 1 non-SCNP), 7.5% for N20 (3 SCNP) and N30 (2 non-SCNP and 1 SCNP), and 12.5% for P25 (3 non-SCNP and 2 SCNP). 5% of the absolute motor performance data had to be removed (2 non-SCNP) and 10% for the relative motor performance data (2 SCNP and 2 non-SCNP). The distribution of the datasets was assessed using the Shapiro Wilk test following the removal of outliers.

Twenty SCNP participants (10 female (F) and 10 male (M); 21.4 $\pm$ 2.36 years of age) and 20 non-SCNP participants (10 F and 10 M; 21.2 $\pm$ 2.21 years of age) participated in this quasi-experimental pre/post study.

According to the Von Korff Chronic Pain Grade Scale, fifteen SCNP participants were classified as Grade I, four were classified as Grade II and the remaining one was classified as grade III. All 20 non-SCNP participants were classified as Grade 0. The NDI score on the day of collection was 7.25 $\pm$ 4.10 and 1.00 $\pm$ 1.52 for the SCNP and non-SCNP group, respectively. The pain VAS score on the day of collection was 1.15 $\pm$ 0.85 and 0.035 $\pm$ 0.11 for the SCNP and non-SCNP group, respectively.

The peripheral N9 SEP peak was not significantly different between groups, where the SCNP (0.99 $\pm$ 0.11) and non-SCNP group (1.02 $\pm$ 0.88) had similar amplitudes (t${}_{(38)}$ = 0.947, p = 0.175, d = 0.30), meaning changes in central SEP peaks in both groups were the result of changes in neural activity rather than an altered afferent volley.

N18 SEP Peak: There was a statistically significant difference between groups (t${}_{(29.058)}$ = 2.031, p = 0.026, d = 0.64), where the SEP peak amplitude increased by 3.06 $\pm$ 0.45% in the SCNP group (n = 20) and by 102.08 $\pm$ 1.55% in the non-SCNP group (n = 20). N20 SEP Peak: There was a statistically significant difference between groups (t${}_{(35)}$ = –5.460, p 0.001, d =1.80), with a 14.06 $\pm$ 0.12% increase in SCNP (n = 17) and a 6.27 $\pm$ 0.11% decrease in non-SCNP (n = 20) participants. P25 SEP Peak: There was a statistically significant difference between groups (t${}_{(33)}$ = –2.857, p = 0.004, d = 0.97), with a 10.86 $\pm$ 0.89% increase in the SCNP group (n = 18) and a 2.65 $\pm$ 0.81% increase in the non-SCNP group (n = 17). See Fig. 3 for a representative dataset from an individual in each group for the aforementioned cortical SEP peaks. See Fig. 4 for mean proportional change of the SEP peaks that demonstrated statistical significance.

Fig. 3.

Representative dataset of cortical SEP peaks (N18, N20, and P25) with group differences from an individual in the SCNP and non-SCNP group, recorded at 2.47 Hz. Black solid represents pre-acquisition, and red dotted line represents post-acquisition. SCNP, subclinical neck pain; SEP, somatosensory evoked potential.

Fig. 4.

Box and whisker plot of mean proportional change (relative to baseline) in (A) N18 (n = 40), (B) N20 (n = 37), (C) P25 (n = 35) SEP peak amplitude following acquisition of FMTT. Red dotted line represents baseline. Orange represents non-SCNP group and blue represents SCNP group. The whiskers reflect minimum and maximum values. The boxes reflect the quartiles, and x symbol represents median. Dashed bars and asterisk(s) denote significant group differences. *p 0.05, **p 0.01, ***p 0.001.

The N11 did not have significant group differences (t${}_{(34)}$= 0.368, p = 0.357, d = 0.123), with a 1.46 $\pm$ 0.30% decrease in the SCNP group (n = 18) and a 3.14 $\pm$ 0.44% increase in the non-SCNP group (n = 18). The N13 did not have significant group differences (t${}_{(36)}$ = 0.117, p = 0.454, d = 0.038), with a 0.34 $\pm$ 0.22% decrease in the SCNP group (n = 20) and a 0.81 $\pm$ 0.36% increase in the non-SCNP group (n = 18). The N24 did not have significant group differences (t${}_{(36)}$ = –4.95 , p = 0.312, d = 0.161), with a 4.31 $\pm$ 0.25% increase in the SCNP group (n = 19) and a 0.87 $\pm$ 0.17% increase in the non-SCNP group (n = 19). The N30 did not have significant group differences (t${}_{(35)}$ = –1.439, p = 0.079, d = 0.47), with a 10.26 $\pm$ 0.15% increase in the SCNP group (n = 19) and a 2.03 $\pm$ 0.20% increase in the non-SCNP group (n = 18). The N60 did not have significant group differences (t${}_{(38)}$ = –0.726, p = 0.236, d = 0.230), with a 11.96 $\pm$ 0.32% increase in the SCNP group (n = 20) and a 5.15 $\pm$ 0.27% increase in the non-SCNP group (n = 20).

It is speculated that reorganization of the various series and parallel circuits of synchronized flow of information from the medulla, subcortical and cortical levels of the central nervous system is impacted during SMI in those with altered cortical and/or cerebellar processing due to a musculoskeletal dysfunction [5, 17, 63]. It is the neural substrates that contribute to somatosensory adaptations of a repetitive task and the feedforward and feedback loops during SMI that are of interest, specifically the cerebellum. The modulation and coordination of movement via the cerebellum enables sensorimotor learning and/or adaptation [11, 12, 64, 65]. The execution of the motor response via the motor loop of the cerebral cortex is dependent on the sensory information received by the primary somatosensory cortex [66]. These processes have been demonstrated to be impacted in those with SCNP [31, 32, 33], and they also appear to be impacted following the acquisition of the novel FMTT.

The neural correlates recorded over the N18 SEP peak reflect the inhibitory activity of the region between the lower medullar and mid-brain-pontine region [50, 67]. It is postulated that the dorsal column nuclei, collaterals from the dorsal column nuclei which contribute to the cuneo-cerebellar tract, the cerebellum, and the accessory inferior olives are the neural generators of this SEP peak [50, 68, 69, 70]. It could also reflect cerebellar SMI, as the inferior olive transmits proprioceptive input to the cerebellar cortex and deep cerebellar nuclei via the inferior cerebellar peduncle, where information is filtered prior to cortical processing for feedforward or feedback control [71]. The large increase in the N18 in the non-SCNP group was also observed in other studies that have employed the same task [38, 39, 72]. This could suggest that the non-SCNP participants have greater inhibitory activity along the olivary-cerebellar pathway, to accommodate for the need to accurately produce and modulate forces via the right thumb during acquisition. The minimal increase in the SCNP group does not correspond with past work that has assessed neuroplastic changes in response to the FMTT in those with attention-deficit/hyperactivity disorder [72] or healthy participants who had their neck vibrated for 10 minutes at 60 Hz [39], where they demonstrated a decrease in the N18 [39, 72]. This could suggest that this group is less able to accurately process proprioceptive input or rely more on vision (via augmented feedback provided during the task), due to the chronic altered neck sensory feedback impacting feedforward control. The group differences could suggest that there are differences in cerebellar SMI as past studies have also demonstrated that those with SCNP exhibit altered cerebellar processing following the acquisition of a motor task [31, 32, 33].

The N20 and P25 SEP peaks reflect the arrival of somatosensory input from the periphery at Brodmann’s area 3b [73, 74] and the processing of that somatosensory input in the primary somatosensory cortex, S1 [74, 75], respectively. The changes observed in the N20 SEP peak in the non-SCNP group have been observed in past work [38, 72], possibly suggesting that there is sensory gating of somatosensory/proprioceptive input. The slight increase in the P25 SEP peak in the non-SCNP group corresponds with the control group in a past study of individuals who did not have their neck musculature vibrated for 10 minutes at 60 Hz [39], which could suggest that accurate sensory feedback is needed to perform this task. The increases in the N20 and P25 SEP peaks are in line with work that employed the same task in those with attention-deficit/hyperactivity disorder [72], and individuals who had their neck muscle vibrated for 10 minutes at 60 Hz [39], suggesting that altered sensory input influences force production and modulation, whether it be neurodevelopmental, transient or chronic. The N20 and P25 SEP peaks are similar in the SCNP group, and the corresponding spinal SEP peaks (N11 and N13) are also similar, suggesting that there may not be any gating of somatosensory information at a spinal level and greater reliance on feedback processing of proprioceptive input to perform this motor skill, which is in line with the minimal increase in the N18 SEP peak. It is also possible that these cortical somatosensory processing changes are the result of greater reliance on visual feedback that was provided during the task (seen in visuomotor control), as opposed to proprioception alone. This could suggest that vision may have played a role in motor control of the right thumb while matching a series of static force traces. It is possible that the impaired feedforward control in response to altered chronic neck sensory input leads to alterations in the arrival of somatosensory input and somatosensory processing in the SCNP group when learning to produce and modulate forces, reflected by the group differences and subsequent increased reliance on feedback processing.

The cortico-basal ganglia-thalamo-cortical loop and the motor circuit (supplementary motor area, primary motor cortex and the somatosensory cortex) within the pre-frontal cortex have been identified as the neural source of the N30 SEP peak [76, 77, 78, 79, 80]. This peak reflects the neural basis of SMI (feedforward and feedback processing), which is reliant on the synchronized flow of somatosensory information from various levels of the central nervous system [76, 78, 79, 81]. The increase in the non-SCNP group corresponds with other studies that have utilized the same motor task [38, 39, 72], and the same goes for the increase in the SCNP group with altered central processing, such as attention-deficit/hyperactivity disorder [72] or neck muscle vibration [39]. The lack of group differences in N30 could be the fact the visuomotor tracking task was static in nature, with minimal force modulation perturbations and greater reliance on augmented feedback, barely impacting the motor circuits greatly, which are needed for motor output that relies on going changes in somatosensory information. Despite there not being a group difference, the visibly greater increase (10%) in the SCNP group compared to the non-SCNP group (2%) alongside the minimal increase in the N18 SEP peak could suggest that there is greater reliance on feedback loops to make rapid changes during motor training in response to acquisition of the FMTT.

This work has shown that even in the early stages of pain development (presence of recurrent pain) there are neural changes in the processing of incoming sensory input following performance of an FMTT that is heavily reliant on proprioception, in those with SCNP. Individuals in both groups demonstrated SEP changes in response to the acquisition of the FMTT, with some group differences, in response to task acquisition, as well as small to large effect sizes even in non-significant SEP peaks. The altered SMI in those with recurrent neck pain, who were tested on a pain-free day, is likely to be the result of altered neck sensory input resulting in altered central processing of somatosensory input [28, 29, 30]. This altered neck sensory input likely arises due to a change in muscle spindle firing or tendon organs, or muscle afferents (group III and IV), which has been observed in fatigue of the neck musculature [82]. The processing of altered proprioceptive input arriving via the dorsal column medial lemniscus pathway and anterior spinocerebellar pathways would result in a misrepresentation of the internal body map which is dependent on feedforward processing, leading to a mismatch between the internal and egocentric or allocentric frame of reference, seen as impaired sensorimotor function due to the correction of an inaccurate reference frame using feedback processing between the motor circuits of the cortex [28, 29, 30, 83]. The SEP changes observed are reflective of the processing of altered sensory feedback at the cortical level on feedforward control, specifically the region between the lower medullar and mid-brain-pontine region, as well as the cortical somatosensory processing pathway.

The SCNP group has visibly worse performance at all time points, however, there were no group differences in their rate of learning, suggesting that the central nervous system of those with SCNP was able to compensate for the altered processing. Learning a sensorimotor skill refers to improvements in performance accuracy (either rapid improvement within a single motor training session or continual improvement across multiple motor training sessions), and efficiency of a movement following memory consolidation [84, 85]. This study appears to demonstrate rapid improvements within the single motor training session within each group, seen as improvements at post-acquisition and at retention, relative to pre-acquisition. These improvements in relation to baseline suggest that both groups learned the novel FMTT. The visible difference between the SCNP and non-SCNP group at each time point in absolute error, suggests that the SCNP participants exhibit a decreased ability to perform the FMTT. Both groups showed minimal continued improvement at retention in following post-acquisition improvements, which coincides with past studies that have employed the same motor task [38, 39, 72]. In contrast, a task reliant on visuomotor information demonstrated that SCNP participants had deficits in acquiring a pursuit movement task compared to healthy controls, who continued to improve following memory consolidation [33]. The lack of time by group interactions relative to retention could be due to the use of a discrete FMTT (e.g., lack of ballistic thumb movements to produce and modulate force of clearly defined plateaus) as there may have been insufficient contextual interference during the acquisition of the FMTT, since higher interference yields greater improvements at retention [86]. It may be attributed to the nature of the FMTT, greater reliance on visual feedback alongside lack of continuous force production and modulation perturbations in the task, which resulted in rapid improvement in visuomotor control in a single training session with no room for improvement thereafter [87] since it may rely more on memory recall once the pattern is recognized [88].

There are two cortical circuits with their own distinct subcortical and cortical interactions that change in response to motor skill learning, where the respective circuits change in accordance with the demands of the motor task and stage of skill acquisition [64, 89]. The acquisition of a sequential motor skill is suggested to increase activity in the deep cerebellar nuclei and cerebellar cortex followed by an increased reliance on the striatal-cortical circuit (striatum, supplementary motor area, inferior parietal, precuneus and ventrolateral pre-frontal cortex) and diminished reliance of cerebellar-cortical circuits with improvements in task proficiency [12]. Both groups in this study demonstrated improvements in the FMTT following acquisition, which corresponded with differential changes in the olivary-cerebellar pathway and cerebellar SMI, and cortical somatosensory processing pathways. These changes correspond with the literature pertaining to increased reliance on cerebellar networks in the early stages of skill acquisition [10, 11, 12, 89, 90]; however, this does not translate to a transition to reliance on striatal networks for long-term retention, reflected by an increase in the N24 SEP peak (cerebellar-somatosensory cortex processing) in the SCNP group, but no group differences in the N24 SEP peak, and between post-acquisition and retention. These cortical changes with a lack of drastic changes with respect to motor performance between groups correspond with literature regarding experimentally induced cutaneous pain on motor learning [25, 91, 92]. Early SEP changes can be used as a marker for altered neuroplasticity in the context of motor skill acquisition of a novel discrete FMTT in those with SCNP.

This study comprises a study population who have SCNP, which may not be reflective of the general public, nor are the results presented applicable for those outside of the ranges of 18 and 30 years of age. The version of the FMTT that we used may not have been sensitive enough to show the effects of altered proprioceptive input as the force traces were stationary and not as dynamic (e.g., they hit a plateau of 0% instead of constantly fluctuating between 2% and 12%). The participants may also have “learned” the task too easily, leaving minimal room for additional improvement in retention. Future work should employ a more difficult, unpredictable and/or dynamic force-matching task, to address these limitations.