Differential relationships between physical activity and pain phenotypes in individuals with spinal cord injury

Abstract

Background

Chronic pain affects 70% of individuals with spinal cord injury (SCI) and leads to declines in health and quality of life. Neuropathic and nociceptive pain are phenotypes derived from different mechanisms that contribute to pain perception. The objective of this research was to investigate differential pain responses to moderate-to-vigorous physical activity (MVPA) in two chronic pain phenotypes: neuropathic and nociceptive pain.

Methods

Community-based physical activity levels were collected for one week in 17 individuals with SCI using a wrist-worn accelerometer, and daily pain ratings were assessed and categorized by phenotype. Physical activity levels were summarized to calculate minutes of MVPA. Correlational analyses were conducted to compare relationships between pain intensity and MVPA across individual participants and between pain phenotype groups.

Results

The neuropathic pain group revealed significant negative correlation between MVPA and pain intensity. In the nociceptive pain group, there was no significant correlation between MVPA and pain intensity. Further analysis revealed two subgroups of positive (N = 4) and negative (N = 3) correlations between MVPA and pain intensity. Pain location differed between the subgroups of nociceptive pain. Individuals with negative correlation experienced neck and upper back pain, whereas individuals with positive correlation experienced unilateral upper extremity pain.

Conclusion

Differential relationships exist between pain phenotypes and MVPA in individuals with SCI. Pain location differed between the subgroups of nociceptive pain, which we presume may indicate the presence of nociplastic pain in some individuals. These results may contribute to the advancement of personalized pain management by targeting non-pharmacological interventions for specific pain phenotypes.

Trial registration: ClinicalTrials.gov identifier: NCT05236933..

Introduction

Chronic pain affects 70% of individuals with spinal cord injury (SCI), and leads to declines in health, function, and quality of life (1–4). Chronic pain is defined as pain that persists for a duration of three months of longer, in the absence or cessation of tissue damage (5). Based off clinical presentation and symptom manifestation, chronic pain in SCI is categorized into phenotypes. Phenotypes reflect the presumed underlying mechanisms of pain processing (6), which encompasses combinations of other mechanisms that contribute to overall pain (6). While neuropathic pain arises due to nervous system dysfunction, nociceptive pain originates from subcutaneous nociceptors and can develop from improper healing from injuries or repetitive microtrauma (7,8). In both of these phenotypes, pain intensity does not directly reflect tissue damage. Chronic nociceptive pain conditions, such as osteoarthritis and tendinitis, are often associated with inflammation. The inflammation of tissues leads to continuous activation of nociceptors and has a substantial role in the development and persistence of chronic pain (9). The biopsychosocial model of chronic pain describes the interaction between physiological, psychological, and social factors that contribute to chronic pain. Psychosocial factors dynamically influence both the development and persistence of pain (10). Therefore, effective treatment approaches are challenging due to the complex nature of chronic pain in SCI and there is no universal solution.

Moderate-to-vigorous intensity physical activity (MVPA) is a promising chronic pain intervention for individuals with SCI as increased durations of MVPA are strongly associated with decreases in pain (11). However, while several studies have shown MVPA improves chronic pain, some research has revealed that pain levels increase or remain unchanged (12). Classification of pain phenotypes may assist in determining appropriate pain management techniques. In individuals with SCI, chronic nociceptive pain often presents as aching or soreness associated with overuse injuries, which are encountered frequently by individuals with higher-level injuries due to a combination of increased physical demand of the upper extremities and strength deficits associated with SCI (13). In contrast, neuropathic pain presents as a combination of various sensations including burning, shooting, and paresthesia, for individuals with SCI at all levels of injury due to the diversity of underlying pathophysiological changes (6).

While physical activity is beneficial for chronic pain in individuals with SCI, the efficacy of MVPA as an intervention would be maximized by individually scaling the intensity, type, and duration of activity based on pain phenotype. Personalization of MVPA interventions would account for nuanced differences in response to chronic pain, which could improve consistency of results (14). In order to inform this need, the objective of this study was to determine if differential relationships exist between MVPA and pain phenotypes in individuals with SCI. We hypothesize that neuropathic pain will improve as MVPA levels increase, and that nociceptive pain will worsen as MVPA levels increase.

Methods

Participants

A convenience sample of 17 individuals with SCI who experience chronic pain were recruited from the greater Philadelphia area and screened using the following criteria: (1) age 18–75 years, (2) at least six months post-SCI, (3) persistent pain for a duration of at least three months, (4) functional ability to self-propel a manual wheelchair, and (5) medically stable. Individuals were excluded from enrollment if physical activity was medically contraindicated. This study was approved by the Temple University Institutional Review Board, and all participants provided written informed consent. This study was registered on ClinicalTrials.gov (NCT05236933) prior to participant enrollment.

Study design and protocol

A prospective repeated measures design was employed to examine differential responses of pain phenotypes to MVPA in individuals with SCI in a community-based setting. All measures collected are listed in Table 1. Participants were offered the option of meeting with a member of the research team at their home or at Temple University. After completing the consent process, participants provided information about demographics, level and completeness of injury, and history of depression. The International Spinal Cord Injury Pain Basic Data Set Version 2.0 (15) is a concise yet comprehensive measure consisting of eight items that provide information regarding the severity, location, and quality of pain, which was used in conjunction with clinical assessment to facilitate the categorization of each pain into neuropathic and nociceptive phenotypes. This tool was also used to assess primary and secondary pain in individuals with concurrent neuropathic and nociceptive pain. Participants described their three worst pain problems. The worst pain was designated as their primary pain. If the second worst pain was of a different phenotype, then it was designated as their secondary pain. If both the first and second pain problems were of the same phenotype, then the third worst pain was assigned as the secondary pain if it was of a different phenotype. Secondary pain was used to evaluate differences between pain phenotypes in individuals with concurrent neuropathic and nociceptive pain. The 11-point numeric pain rating scale embedded within The International Spinal Cord Injury Pain Basic Data Set Version 2.0 for pain intensity rating was readministered daily throughout the study via an electronic survey to obtain subsequent primary and secondary pain intensity ratings.

Table 1.

Measures collected and timepoints of collection.

Measure	Time of collection
	Day 1	Days 2–7
MVPA (ActiGraph GT9X Link)		X
Pain intensity and phenotype (International SCI Pain Basic Data Set Version 2.0)	X	X
Demographics questionnaire	X
Depression (Patient Health Questionnaire)	X
Pain interference (PROMIS Pain Interference Short Form 8a V1.0)	X
Quality of life (Subjective Quality of Life Scale)	X
Pain catastrophizing (Pain Catastrophizing Scale)	X

At the initial interview, participants also completed PROMIS Pain Interference Short Form 8a V1.0 (16), Subjective Quality of Life Scale (17), and Pain Catastrophizing Scale (18) to assess consequences of pain and quality of life. Participants were supplied with an ActiGraph GT9X Link and provided brief instruction on proper use of the device. The Borg Rating of Perceived Exertion Scale was reviewed with participants while they were asked to perform a three-minute bout of activity consisting of upper extremity exercises at a self-selected speed. This session was used to instruct participants in proper use of the RPE scale. Due to space constraints in the laboratory and home settings, not all participants were able to perform activity at an intensity level that they reported was moderate-to-vigorous. Participants were instructed to perform at least one bout of activity in the community during which they reported a rating of perceived exertion (RPE) of 12–14, which coincides with moderate-intensity physical activity (19). The time of this activity was recorded and used to confirm the moderate-intensity threshold for minutes of MVPA. Participants wore the device every day and maintained their usual activities for one-week while collecting real-world data in the community. Because we implemented the study in a community-based setting, the data obtained captures fluctuations in activity levels across participants and corresponding pain levels. Any day with less than ten hours of data collected were considered incomplete and excluded from analysis. At least six days of complete data were obtained from all participants.

Data analysis

All accelerometer data were collected at a sampling rate of 60 Hz. Raw data were translated into summary movement data by utilizing the Monitor Independent Motion Summary (MIMS unit) (20), which quantifies accelerometer values as a measure of physical activity volume. Selection of the moderate-intensity threshold was based off real-world data collection. Instead of using one standard threshold value for all participants, we selected a personalized MVPA threshold for each participant to account for the variability in function and movement patterns across individuals with SCI. Visual analysis of MIMS unit plots was used in conjunction with participant-reported moderate-intensity values to select appropriate thresholds and calculate minutes of MVPA. RPE was used to confirm that the moderate-intensity threshold was adjusted appropriately for each individual participant. The frequency of each MIMS unit value over five days was assessed and compared with visual inspection of plotted data. A summary of the process for determining minutes of MVPA is described in Fig. 1.

Figure 1.

Process for determining personalized thresholds to calculate minutes of MVPA.

A Kolmogorov–Smirnov test determined that the data were not normally distributed, therefore nonparametric analyses were implemented. Descriptive statistics of participant demographics and psychosocial factors were calculated as median and interquartile range (IQR). Separate Spearman’s rank-order correlation analyses were performed to analyze the relationships between MVPA levels and pain intensity for each individual participant, as well as for each phenotype. Phenotype correlations were further evaluated using repeated measures correlation, as previous studies have identified that variation within subjects must be accounted for in addition to variation between subjects. Several methods have been proposed to avoid violating the assumption of independent observations (21), including repeated measures correlation (rmcorr). Rmcorr is a technique similar to multilevel modeling, but instead of simultaneously analyzing different sources of variance using fixed and random effects, rmcorr estimates a common regression slope with varying intercepts (22). By utilizing a less complex analysis, rmcorr can be performed with a smaller sample size requirement (22). Rmcorr can be computed using an open-source package available for R statistical software (22). Average secondary pain was calculated and the relationship between secondary pain and MVPA was analyzed using Spearman’s rank-order correlation for each phenotype and for each individual participant. All data analyses were performed using R 4.2.2 software with statistical significance determined by an alpha level of 0.05.

Results

Participant demographics and psychosocial factors are outlined in Table 2, and average pain intensity for primary and secondary pains is displayed in Table 3.

Table 2.

Participant demographics and factors. All values are reported as median (interquartile range) or number (percent within group).

	Neuropathic (n = 10)	Nociceptive (n = 7)
Age in years	40.5 (21)	58.0 (16)
Sex: Male	5.0 (50%)	4.0 (57%)
Female	5.0 (50%)	3.0 (43%)
Race: White	7.0 (70%)	6.0 (86%)
Black or African American	2.0 (20%)	0.0 (0%)
Asian	1.0 (10%)	1.0 (14%)
Level of injury: Quadriplegia	4.0 (40%)	4.0 (57%)
Paraplegia	6.0 (60%)	3.0 (43%)
Completeness of injury: Complete	4.0 (40%)	3.0 (43%)
Incomplete	6.0 (60%)	4.0 (57%)
Years since injury onset	10.5 (11.0)	29.0 (32.0)
Depression	4.0 (6)	2.0 (3)
Subjective quality of life	6.5 (2)	7.0 (0)
Pain interference	19.0 (15)	11.0 (5)
Pain catastrophizing	22.5 (37)	12.0 (13)

Table 3.

Average pain intensity and minutes of MVPA in each phenotype for primary and secondary pains, reported as median (interquartile range).

	Neuropathic (primary)	Nociceptive (primary)	Neuropathic (secondary)	Nociceptive (secondary)
Number of participants (overall percentage for primary or secondary pains)	10.0 (59%)	7.0 (41%)	7.0 (78%)	2.0 (22%)
Pain intensity	4.5 (3.5)	4.0 (2.8)	4.0 (2.8)	3.0 (2.0)
Minutes of MVPA	47.0 (52.5)	34.0 (71.8)	39.5 (28.8)	40.0 (52.5)

Neuropathic pain and minutes of MVPA were negatively correlated (ρ = −0.75, P < 0.001) when evaluated with Spearman’s correlation (Fig. 2) as well as rmcorr techniques (r_rm = −0.73, P < 0.001) (Fig. 3(a)), and each individual participant with neuropathic pain demonstrated negative correlation between pain intensity and number of minutes of MVPA each day (Fig. 4).

Figure 2.

Spearman’s Rank Order correlation between pain intensity and number of minutes of MVPA in neuropathic (ρ =  – 0.75, P < 0.001) and nociceptive (ρ = −0.17, P = 0.23) pain phenotype groups.

Figure 3.

(a) Repeated measures correlation between pain intensity and number of minutes of MVPA (r_rm = −0.73, P < 0.001) in neuropathic pain phenotype group. Each colored line denotes an individual participant. (b) Repeated measures correlation between pain intensity and number of minutes of MVPA (r_rm = −0.06, P = 0.72) in nociceptive pain phenotype group. Each colored line denotes an individual participant.

Figure 4.

Correlation between neuropathic pain intensity and number of minutes of MVPA in individual participants.

Nociceptive pain and minutes of MVPA were not significantly correlated using either statistical technique (Figs. 2 and 3(b)). Participants with nociceptive pain demonstrated distinct positive and negative trends between pain and minutes of MVPA (Fig. 5).

Figure 5.

Correlation between nociceptive pain intensity and number of minutes of MVPA in individual participants. Individual participants PMA01, PMA03, and PMA07 demonstrate a negative trend between nociceptive pain and minutes of MVPA, and individual participants PMA06, PMA08, PMA11, and PMA17 demonstrate a positive trend between nociceptive pain and minutes of MVPA.

All individual correlation values and pain locations are presented in Table 4. Three individuals with nociceptive pain, PMA01, PMA03, and PMA07, demonstrated negative trends between pain intensity and number of minutes of MVPA, whereas four participants, PMA06, PMA08, PMA11, and PMA17, demonstrated positive trends. We found that rmcorr, similar to Spearman’s Rank correlation, yielded a significant, negative correlation between neuropathic pain and MVPA. In both methods, no significant relationship was revealed between nociceptive pain and MVPA. The combination of correlation analyses reinforce the results.

Table 4.

Individual participant pain and MVPA correlation values and location of pain for primary and secondary pain. Significant P values are represented in bold font.

Participant	Primary or secondary	Phenotype	Location of Pain	Correlation Coefficient	P value
PMA01	Primary	Nociceptive	Neck	−0.59	0.160
PMA01	Secondary	Neuropathic	Bilateral hips	−0.93	0.003
PMA02	Primary	Neuropathic	Lower back	−0.83	0.039
PMA02	Secondary	Nociceptive	Left (nondominant) shoulder	0.97	0.001
PMA03	Primary	Nociceptive	Neck	−0.93	0.001
PMA04	Primary	Neuropathic	Mid back	−0.97	0.001
PMA04	Secondary	Nociceptive	Bilateral medial border of scapulae	0.20	0.700
PMA05	Primary	Neuropathic	Lower back/ right (dominant) hip	−0.78	0.023
PMA05	Secondary	Nociceptive	Bilateral wrists	0.72	0.045
PMA06	Primary	Nociceptive	Left (nondominant) shoulder	0.51	0.240
PMA06	Secondary	Neuropathic	Mid back	−0.60	0.150
PMA07	Primary	Nociceptive	Upper back/ bilateral scapulae	−0.88	0.009
PMA08	Primary	Nociceptive	Right (dominant) shoulder/ scapula	0.77	0.044
PMA09	Primary	Neuropathic	Lower back	−0.83	0.042
PMA09	Secondary	Nociceptive	Bilateral wrists/ hands	0.83	0.042
PMA10	Primary	Neuropathic	Lower back	−0.95	0.001
PMA11	Primary	Nociceptive	Right (dominant) shoulder	0.96	0.001
PMA12	Primary	Neuropathic	Right (dominant) leg	−0.85	0.016
PMA12	Secondary	Nociceptive	Right (dominant) shoulder/ upper trapezius	0.88	0.009
PMA13	Primary	Neuropathic	Bilateral feet	−0.93	0.003
PMA13	Secondary	Nociceptive	Left (nondominant) shoulder	0.78	0.040
PMA14	Primary	Neuropathic	Lower back	−0.90	0.006
PMA14	Secondary	Nociceptive	Bilateral upper trapezius	−0.92	0.004
PMA15	Primary	Neuropathic	Bilateral lower legs/ feet	−0.94	0.001
PMA16	Primary	Neuropathic	Left (nondominant) hip/ lower back	−0.88	0.001
PMA17	Primary	Nociceptive	Right (dominant) shoulder	0.35	0.450

Discussion

Neuropathic pain was found to decrease as MVPA levels increased, as hypothesized. However, our findings of nociceptive pain in relation to MVPA were less clearly defined. Some individuals with nociceptive pain demonstrated worsening pain as MVPA levels increased, whereas other individuals demonstrated improvements in pain with increasing levels of MVPA. The following sections will discuss these findings and their clinical implications in further detail.

Precision-based approach for pain modulation using MVPA

Physical activity interventions have been efficacious for chronic pain management in many populations (23–25). However, due to the intricate nature of chronic pain and the multidimensional factors that contribute to the experience of pain, a precision-based treatment approach would be beneficial. One factor that can guide chronic pain management recommendations is pain phenotype. Our findings have unveiled that neuropathic pain responds more favorably to MVPA than nociceptive pain. All participants in this study revealed negative correlations between MVPA and neuropathic pain intensity, meaning increasing this type of physical activity can have a positive effect on this type of pain.

For individuals with nociceptive pain as primary, three participants showed negative correlations between MVPA and pain intensity, while the other four participants showed positive correlations between MVPA and pain intensity. Pain location is a characteristic that differs between the subgroups of nociceptive pain subgroup. Individuals with negative correlation experienced neck and upper back pain, whereas individuals with positive correlation all experienced unilateral upper extremity pain. The location of pain suggests that individuals with positive correlations between pain and MVPA may be experiencing pain due to overuse injuries, which can lead to tendinopathies and other microtraumas that are exacerbated by repetitive movement (26,27). For individuals with negative correlations between pain and MVPA, pain may be mechanical and related to posture, which can improve from generalized movement (28).

When visualizing the neuropathic data collectively, it appears that lower amounts of MVPA may be related to greater differences in pain reduction than higher amounts of MVPA, which appear to be related to relatively smaller decreases in pain. This points to a potentially nonlinear relationship between pain and amount of activity. In a previous study, we found that the relationship between physical activity and pain was curvilinear and was best represented by a polynomial regression of second order (29). In that study pain was assessed using The Chronic Pain Grade Scale (30), which does not identify phenotypes of pain, so we expect that we measured an overlap of both neuropathic and nociceptive pain. In individuals with multiple pain problems, an intermediate amount of PA may be optimal for balancing the therapeutic benefits of MVPA. These findings suggest that changes in pain reduction are relative to pain intensity, and that MVPA is most efficacious for modulating pain when pain intensity is greater, and when a mid-range volume of MVPA is achieved.

The type of MVPA is an important factor in prescribing activity interventions. Neuropathic pain may benefit from all types of MVPA, as they assist in metabolic adaptations of contractile tissues and structural remodeling of muscles (31). However nociceptive pain may benefit most from strength-specific resistance training designed to compensate for decreased innervation in overused muscles (13,32). Therefore, overuse likely contributes to increased nociceptive pain in individuals with SCI. Personalized thresholds may be beneficial in determining appropriate dosage of MVPA to maximize benefits for pain modulation while balancing detrimental effects of overuse.

Personalized thresholds for MVPA for individuals with SCI

Physical activity can be classified by intensity: sedentary, light, moderate, vigorous, and high. These intensities coincide with specified ranges of measures including heart rate, RPE, and accelerometry. Heart rate is not an accurate representation of exercise intensity for individuals with cervical and high thoracic injury because the T1-T5 segments of the spinal cord control heart rate, and if these segments of the spinal cord are affected by SCI, cardiovascular regulation will be altered (33).

Accelerometry has been explored to quantify PA for individuals with SCI (34), but there is limited evidence supporting standard values to define MVPA in accelerometer data (35). Accelerometers are more effective at capturing repetition-based activities than quantifying resistance of physical activity. This study used RPE as a secondary measure to confirm that what was identified in the data as MVPA for each participant aligned with what each participant perceived was MVPA. RPE has been validated to predict exercise intensity in individuals with SCI (36), however it is highly variable across different levels of injury and function (19). Combining accelerometry and RPE has been suggested as a more reliable method to measure the intensity of real-world activities (37,38), which is the fundamental direction of personalized thresholds. Leveraging self-reported RPE with accelerometer-based quantification to develop a personalized metric presents a more accurate estimation of MVPA. Individually calibrating each participant’s threshold for moderate intensity evaluates MVPA more comprehensively than either method individually, and attempts to reduce the limitations of each. We employed a method of developing personalized intensity thresholds using a RPE comparison for a more accurate measurement of MVPA, which may be useful in future protocols using real-world data. Incorporating personalized thresholds into real-world data can also further assist in recognizing overuse and deciphering pain attributed to overuse versus other factors (39). Recent best practice guidelines for chronic pain management have recommended personalized, multimodal treatment approaches (40), supporting further research regarding combinations of chronic pain treatments. As all fields of medicine move towards precise and targeted interventions, personalized thresholds to measure physical activity intensity invite the opportunity to support this progression in rehabilitation science. Personalized intensity thresholds will facilitate targeted interventions that account for nuanced differences in response to chronic pain interventions (14).

Concurrent pain phenotypes

Though the subset of individuals with concurrent neuropathic and nociceptive pain was limited, different phenotypes of pain also responded differently to MVPA within individuals. Of the nine participants with a secondary concurrent pain, most was nociceptive (n = 7), with only two reporting neuropathic secondary pain. This indicates that, in general, neuropathic pain is rated as the worst pain in individuals who experience both neuropathic and nociceptive pain concurrently. Neuropathic pain is often perceived as more severe in intensity (41), and it has been characteristically described as burdensome and debilitating (42,43). The neuropathic pain group also had worse pain interference by eight points and worse pain catastrophizing by ten points, with minimally clinical important differences ranging from 3.5–5.5 (44) and 3.2–4.5 (45), respectively.

Neuropathic pain is negatively correlated with MVPA across primary and secondary pain. Two opposing trends are discernable between nociceptive pain and MVPA in individuals with primary nociceptive pain. However, individuals with a secondary nociceptive pain tend to only experience worsening pain with increased MVPA. This suggests that MVPA alleviates neuropathic pain at the expense of exacerbating overuse injuries, and it is possible that individuals with concurrent neuropathic and nociceptive pain prioritize reducing neuropathic pain over consequential nociceptive pain. Concurrent pain in individuals with SCI is common (42), therefore, a nuanced approach is vital to determine an optimal recommendation for MVPA levels to manage overall pain.

Determining phenotypes of chronic pain

The present study examined pain across two phenotypes. Our findings indicate that additional stratification by location or source of nociceptive pain may be beneficial in categorization of pain for the development of targeted MVPA interventions. The two phenotypes analyzed in the present study are commonly researched (46,47), however, a third phenotype of pain exists which may explain the divergence of our nociceptive pain findings. Nociplastic pain encompasses conditions such as fibromyalgia (48). In 2021, the International Association for the Study of Pain (IASP) updated the clinical criteria for nociplastic pain to reflect the underlying mechanism of central nervous system sensitization, and to include conditions such as nonspecific neck pain (49). Central sensitization is an alteration in sensory processing with disrupted resting state functional connectivity, resulting in hyperalgesia (50,51).

Deciphering between nociplastic and chronic nociceptive pain is complex, but one potential way to identify nociplastic pain is by evaluating onset of pain. Because there is no associated tissue damage, the onset of nociplastic pain is idiopathic. Chronic nociceptive pain may have insidious onset in overuse injuries, but in many cases the origin can still be identified. Additional studies should examine the effects of nociplastic pain to determine if this type of pain benefits from increased MVPA, or if caution is advised to avoid exacerbation.

Limitations

Several limitations exist in this study. This research was designed as a pilot study, and therefore the sample size required was small. Findings may not be generalizable due to the diversity in SCI level, completeness, and physical function of individuals with SCI. While our findings did not show any apparent patterns in the level or completeness of injury, a larger sample may reveal additional insight with regards to these factors. Additionally, further scrutiny has revealed that some individuals within the nociceptive pain phenotype potentially fit into the category of nociplastic pain, which should be carefully examined and considered in future studies.

Conclusion

Differential relationships appear to exist between pain phenotypes and MVPA in individuals with SCI. A thorough analysis indicated that pain location differed between the subgroups of nociceptive pain, which we presume to indicate the presence of nociplastic pain in some individuals. Individuals with a negative correlation experienced neck and upper back pain, whereas individuals with a positive correlation experienced unilateral upper extremity pain. These results may contribute to the advancement of personalized pain management by targeting non-pharmacological interventions for specific pain phenotypes.

Disclaimer statements

Contributors None.

Funding None.

Conflicts of interest Authors have no conflict of interests to declare.