Authors: Katherine L. Smulligan (1Division of Sports Medicine, Boston Children’s Hospital, Boston, MA, USA), Andrew C. Smith (2Department of Physical Medicine and Rehabilitation, University of Colorado School of Medicine, Aurora, CO, USA), Kenneth A. Weber, II (3Division of Pain Medicine, Stanford University School of Medicine, Palo Alto, CA, USA), Patrick Carry (4Department of Orthopedics, University of Colorado School of Medicine, Aurora, CO, USA; 5Sports Medicine Center, Children’s Hospital Colorado, Aurora, CO, USA), Carrie Esopenko (6Department of Rehabilitation and Human Performance, Icahn School of Medicine at Mount Sinai, New York, NY, USA), Christine M. Baugh (7Center for Bioethics and Humanities, Division of General Internal Medicine, University of Colorado School of Medicine, Aurora, CO, USA), James Elliott (8The Kolling Institute, The University of Sydney, Faculty of Medicine and Health, School of Health Sciences & The Northern Sydney Local Health District, St Leonards, NSW, Australia), Suzanne Snodgrass (9Discipline of Physiotherapy, College of Health, Medicine and Wellbeing, The University of Newcastle, Callaghan, NSW, Australia), Evert Onno Wesselink (3Division of Pain Medicine, Stanford University School of Medicine, Palo Alto, CA, USA; 10Faculty of Behavioural and Movement Sciences, Amsterdam Movement Sciences, Vrije Universiteit Amsterdam, The Netherlands), Julie C. Wilson (4Department of Orthopedics, University of Colorado School of Medicine, Aurora, CO, USA; 5Sports Medicine Center, Children’s Hospital Colorado, Aurora, CO, USA; 11Department of Pediatrics, University of Colorado School of Medicine, Aurora, CO), David R. Howell (4Department of Orthopedics, University of Colorado School of Medicine, Aurora, CO, USA; 5Sports Medicine Center, Children’s Hospital Colorado, Aurora, CO, USA)
Categories: Article, Neck, brain injury, imaging, multifidus, muscle fat infiltrate
Source: Journal of child neurology
Authors: Katherine L. Smulligan, Andrew C. Smith, Kenneth A. Weber, Patrick Carry, Carrie Esopenko, Christine M. Baugh, James Elliott, Suzanne Snodgrass, Evert Onno Wesselink, Julie C. Wilson, David R. Howell
We examined associations between MRI-obtained cervical spine flexor and extensor muscle fat infiltrate (MFI) and muscle volume with 1) return-to-play (RTP) clearance and 2) post-concussion injury incidence. Adolescents (N=30, age=15.8±1.3 years, 60% female) underwent cervical spine MRI initially post-concussion and were monitored until RTP. After RTP clearance, participants underwent follow-up MRI and completed monthly injury surveys for 12-months. Cox proportional hazards regression examined associations between MFI and muscle volume 1) initially post-concussion with RTP clearance and 2) at RTP with subsequent injury incidence. Adjusting for covariates, RTP clearance was not associated with cervical spine MFI (hazard ratio [HR]=0.99; 95% confidence interval [CI]: 0.96–1.02) or muscle volume (HR=0.98; 95%CI: 0.96–1.01). Adjusting for covariates, post-concussion injuries were not associated with MFI (HR=0.90; 95%CI: 0.78–1.05) or muscle volume (HR=1.01; 95%CI: 0.97–1.04). Cervical spine muscle characteristics may not be affected after concussion, or our MRIs may have occurred too early to capture post-concussion dysfunction.
After sport-related concussion, immediate removal from competition and evaluation by a healthcare professional is recommended to avoid further injury.^1,2^ In most cases, athletes with a concussion experience a resolution of symptoms and return to sports participation within 4 weeks of injury.^2^ Unfortunately, once athletes receive clearance to return to play (RTP), they appear to be predisposed to increased risk of sustaining subsequent injuries in future sports participation.^3^ Meta-analyses suggest that athletes with concussion have more than twice the risk of sustaining a subsequent sports-related Injury (both musculoskeletal injuries and concussions) after RTP compared to athletes without a recent sport-related concussion.^4,5^ However, few objective measures (e.g., biomarker or imaging approaches) exist to inform clinical recovery timelines or identify individuals at risk of subsequent injuries,^2^ posing a challenge for healthcare providers assessing and managing patients at risk of adverse outcomes in the short- or long-term. In the absence of an agreed upon objective marker of clinical recovery, a multimodal assessment is recommended to determine concussion recovery and facilitate decisions regarding return to sports participation after concussion.^2,6^ Investigating whether imaging characteristics are associated with RTP clearance time and subsequent injury risk may inform assessment, treatment, and ultimately RTP decisions after concussion.
The forces acting upon the head, neck, and body that cause concussion may also cause concomitant cervical spine injuries such as whiplash.^7,8^ The injuries underlying concussions and whiplash share mechanics and often similar symptom profiles including headache, neck pain, dizziness, and fatigue,^7^ and functional deficits including cervical spine proprioception impairments.^9^ The overlap between injury mechanisms, symptoms, and functional deficits makes it difficult to differentiate between the injuries,^7,10,11^ presenting a unique challenge for guiding treatment. After whiplash, cervical spine muscle characteristics obtained via magnetic resonance imaging (MRI) have been observed shortly after the traumatic event with changes persisting in individuals with prolonged symptom duration,^12–16^ which may also be useful in informing the clinical course of concussion. Specifically, individuals who experience persistent symptoms after whiplash have been observed to have higher levels of deep cervical spine muscle fat infiltrate (MFI) and greater cervical spine multifidus muscle volume (including the MFI component) compared to those who no longer experience symptoms at 1 year post-injury.^12–16^ Given the overlap between injury mechanisms, symptoms, and functional deficits, it is possible that similar muscle characteristic changes exist after concussion and may identify individuals at risk for poor recovery outcomes, though this has yet to be investigated after concussion.
Although the mechanisms for how the observed deep cervical spine muscle changes negatively influence whiplash recovery are not fully understood, several theories exist. For example, MFI in the cervical spine muscles could be a result of an inflammatory process (i.e., increased expression of pro-inflammatory cytokines), muscle disuse or denervation, changes in sympathetic nervous system activation, and/or general inactivity post-injury.^13^ Elevated MFI may, in turn, result in larger cervical spine muscle volume among individuals with persistent whiplash symptoms compared to those no longer experiencing symptoms, potentially resulting in negative consequences for muscle function.^15^ More MFI in the cervical spine multifidus is associated with postural instability among individuals with other cervical spine pathologies (i.e., cervical spondylotic radiculopathy).^17^
Though not yet studied, it is possible that similar changes in cervical spine muscle characteristics exist after concussion given the similar biomechanical injury mechanisms. It is therefore possible that concomitant cervical spine injuries could lead to similar cervical spine muscle changes after concussion as observed after whiplash. These changes may contribute to functional deficits such as postural instability commonly observed after whiplash and concussion,^18–20^ and may ultimately predispose individuals to subsequent injuries, as has been observed among athletes upon concussion RTP.^4,5^ It is also possible that cervical spine muscle characteristics (i.e., MFI and muscle volume) may predispose individuals to sustaining a concussion. Investigating cervical spine muscle characteristics (i.e., MFI and muscle volume) after concussion represents a novel and innovative method to identify a potential factor underlying the development of prolonged RTP time and subsequent injury risk after concussion. Importantly, a 10-week cervical spine strengthening program for individuals experiencing persistent symptoms after whiplash was observed to reduce MFI in the cervical spine multifidus muscles.^21^ Thus, in addition to representing a potential mechanism of poor concussion outcomes, cervical spine muscle characteristics may also present a potential therapeutic target.
Given the similar biomechanical mechanisms and symptom profiles between concussion and whiplash injuries,^7,8^ it is possible that similar muscle morphometry characteristics exist after concussion and may be associated with recovery outcomes (i.e., RTP clearance time, subsequent injury risk). Therefore, our primary objective was to investigate whether deep cervical spine flexor and extensor MFI or muscle volume were associated with RTP clearance time after sport-related concussion. Our secondary objective was to investigate whether deep cervical spine flexor and extensor MFI or muscle volume were associated with post-concussion injury risk (i.e., subsequent concussions and musculoskeletal injuries) in the year after RTP after concussion. We hypothesized that higher MFI levels and larger overall muscle volume (including the MFI) would be associated with longer RTP clearance time and increased risk of subsequent injury after concussion RTP.
We recruited and enrolled adolescent athletes participating in organized sports (i.e., with a coach and structured practices and competitions) for this prospective, longitudinal study between February 2021 and April 2024. Study inclusion criteria consisted of being 13–18 years of age, diagnosed with a concussion by a sports medicine physician or athletic trainer, undergoing initial study assessment within 3 weeks of sustaining a concussion, planning to return to sports once cleared by their physician, and having concussion symptoms upon enrollment defined as scoring ≥9 points on the Post-Concussion Symptom Inventory (PCSI).^22^ Exclusion criteria were sustaining a concussion via a high velocity mechanism such as a motor vehicle collision and findings of a structural brain injury on neuroimaging if performed as part of routine care for concussion evaluation. Participants and guardians provided written informed assent/consent. Prior to study commencement, the local institutional review board reviewed and approved the study protocol.
During the initial study assessment, participants underwent cervical spine MRI. We then monitored participants until they were cleared by their physician to begin the RTP protocol in line with the most recent consensus statement on concussion in sport at the time of their study participation^2,23^ (Figure 1). To determine associations with RTP clearance time, our outcome was the number of days from concussion until clearance for RTP by their physician. Participants returned for a follow-up assessment upon receiving RTP clearance during which they underwent another cervical spine MRI. We then monitored participants for subsequent sport-related injuries (i.e., musculoskeletal, concussion) in the following year using monthly electronic injury surveys. The outcome for determining associations with subsequent injury risk was the number of days between RTP clearance and sustaining a subsequent injury.
At each assessment, participants underwent MRI of their deep cervical spine flexor (longus colli, longus capitis) and extensor (multifidus, semispinalis cervicis) muscles from C3-C6 (Figure 2).^24^ MRI was performed using a 3.0 Tesla Siemens (Erlangen, Germany) Skyra scanner with a 64-channel head/neck coil with a dual-echo gradient-echo FLASH Dixon sequence (acquisition time = 4 min 57 sec, TR = 7.05 ms, TE1 = 2.46 ms, TE2 = 3.69 ms, flip angle = 12°, field of view = 320 × 320 mm^2^, slice oversampling = 20% with 40 slices per slab, slice thickness = 3.0 mm, resolution = 0.7 × 0.7 mm^2^, number of averages = 6). We used a previously developed and validated multi-muscle convolutional neural network (CNN) model for automated segmentation of cervical spine muscles from Dixon fat-water images.^24,25^ The automated segmentation was performed using a U-net CNN model to quantify MFI and muscle volume using the MONAI framework for deep learning in healthcare imaging.^26,27^ Using the fat and water signal intensities from each muscle segmentation, MFI was calculated using the following fat signal / ( fat signal + water signal) ×100%.^13,24^
The MRI at our facility underwent a software upgrade from VE11C to XA30A in September 2022. Due to slight differences in the MRI sequence following this upgrade, the values for MFI and muscle volume were significantly different between participants scanned before and after the upgrade. Thus, we adjusted for the software upgrade in our statistical analysis by performing a meta-analysis of pre- and post-upgrade data to allow for use of all available MRI data.^28^
After the RTP assessments, participants completed monthly electronic surveys for 12 months about subsequent sport-related injuries and athletic exposures.^29^ Each month, we sent surveys to participants using REDCap.^30^ Participants reported any sport-related injury they sustained that resulted in missing ≥ 1 practice or competition, and the number of practices and competitions they participated in for each sport during that month. If any injuries were sustained, participants answered additional questions about each injury. Questions included the date of injury, area of the body that was injured, the type of injury sustained, the sport during which the injury was sustained, the type of medical provider that diagnosed the injury, the amount of time-lost from their sport due to the injury, and a narrative description of each injury sustained.
To investigate the association between RTP time and cervical spine muscle health characteristics, we used separate univariable Cox proportional hazards regression models for each muscle variable (i.e., MFI and muscle volume). For both models, time to RTP clearance was the outcome. We used average MFI of the deep cervical spine flexors and extensors as the predictor variable in one model, and muscle volume of the same flexor and extensor muscles as the predictor in the other model. Due to the software upgrade, we analyzed pre-upgrade and post-upgrade participants separately. We then performed a fixed effects meta-analysis to combine the hazard ratios for RTP clearance for participants pre- and post-upgrade. To maximize our sample, we only inferred data using the meta-analysis (combined) approach.
We then repeated these steps using a multivariable Cox proportional hazards regression model to investigate the relationship between RTP time and MFI or muscle volume while adjusting for initial concussion symptom severity (initial PCSI score) as that is the most consistent predictor of recovery time in previous research.^31^ We again used a fixed effects meta-analysis to combine the hazard ratio for RTP clearance from participants pre- and post-upgrade, adjusting for initial symptom severity.
To investigate the association between time to subsequent injury and cervical spine MFI and muscle volume assessed at the RTP clearance visit, we used univariable Cox proportional hazards regression models with time to subsequent injury as the outcome. We used separate models for MFI and muscle volume of the deep cervical spine flexors and extensors. Due to the software upgrade, we analyzed pre-upgrade and post-upgrade participants separately. We then performed a fixed effects meta-analysis to combine the hazard of subsequent injury from participants pre- and post-upgrade.^28^ Due to model convergence issues for post-upgrade participants, we were unable to adjust for additional variables in the meta-analysis and thus performed a univariable analysis only. We censored participants who did not sustain an injury during the follow-up period or did not complete the 12-month follow-up period, and we used their available data for the analysis.
To understand the association between subsequent injury and cervical spine MFI and muscle volume, we calculated the change between initial and RTP measures of cervical spine MFI and muscle volume (RTP minus initial). Thus, a positive value indicates an increase in MFI or muscle volume over time, while a negative value indicates a decrease. We used the within-person MFI and muscle volume change in multivariable Cox proportional hazards regression models, adjusting for RTP clearance time and prior concussion and/or musculoskeletal injury.
All tests were 2-sided and statistical significance was set a priori at p<0.05. Statistical analysis was performed using R studio (version 4.2.2, R Core Team 2022, Vienna Austria) and the Metafor package (version 4.6–0).^32^
We enrolled and initially assessed 30 participants who we monitored through RTP clearance (Table 1). The participants who underwent MRI scans pre- and post- software upgrade were similar in demographics and clinical characteristics including age (15.8±1.3 vs. 15.7±1.5 years, p=0.82), proportion of female participants (58% vs 64%, p>0.99), prior concussion history (37% vs. 55%, p=0.45), musculoskeletal injury history (63% vs 36%, p=0.14), height (171±11.3 vs. 166±7.3 cm, p=0.25), and weight (66.7±18.9 vs. 62.6±11.7 kg, p=0.51). The meta-analysis results for the univariable Cox proportional hazards regression models indicated there was not a significant association between RTP clearance time and cervical spine MFI or muscle volume of the deep cervical spine flexor (longus colli, longus capitis) and extensor (multifidus, semispinalis cervicis) muscles (Table 2). Similarly, after adjusting for initial symptom severity, the multivariable meta-analyses results indicated there were not significant associations between RTP clearance time and cervical spine MFI or muscle volume (Table 3).
Twenty-six participants returned for follow-up MRI scans upon receiving RTP clearance (age=15.7±1.4 years; 58% female; RTP assessment=36.6±16.4 days post-concussion) and four were lost to follow-up (age=15.9±0.6; 75% female). The 26 participants were monitored for subsequent injuries via monthly injury surveys. During the injury monitoring period, 10 participants (38% of the sample) sustained a total of 15 injuries (Table 4). The meta-analysis results for the univariable Cox proportional hazards regression models indicated there was not a significant association between hazard of subsequent injury and cervical spine MFI or muscle volume in the deep cervical spine flexors and extensors (Table 5). When using the within participant change from initial to RTP assessments for cervical spine MFI and muscle volume, we did not observe significant associations with hazard of subsequent injury after adjusting for RTP clearance time and prior concussion and/or musculoskeletal injury in the multivariable model (Table 6).
We did not observe significant associations between MRI-obtained cervical spine MFI or muscle volume of the deep cervical spine extensors and flexors and RTP clearance time or subsequent injury risk, contrary to our hypotheses. After whiplash, more cervical spine MFI and larger multifidus muscle volume have been observed among individuals who experience persistent symptoms.^12–16^ This prior evidence suggests that MRI-obtained muscle health measures, specifically MFI and muscle volume, may represent mechanisms of poor recovery after not just whiplash injuries, but other neuromuscular conditions (e.g., charcot-marie tooth, sarcopenia) and diseases (e.g., cancer, diabetes).^12–16,33–36^ Given that the cervical spine is often injured concomitantly with concussion,^7,8^ it is important to consider how cervical spine dysfunction after concussion may present to better inform assessment, treatment, and prognosis. Understanding factors that have associations with subsequent recovery timelines (i.e., RTP clearance, subsequent injury), could be helpful for healthcare providers in determining concussion prognosis and setting recovery expectations. Our results, however, suggest that cervical spine MRI muscle health measures obtained within 3 weeks of sustaining a sport-related concussion and again upon receiving RTP clearance (36.6 ± 16.4 days post-concussion) are not associated with RTP clearance time or subsequent injury risk among adolescents recruited from a sports medicine clinic.
Due to the similar injury mechanisms between concussion and whiplash,^7,8^ we hypothesized that higher cervical spine MFI and larger muscle volume in the deep flexors and extensors would be similarly associated with prolonged recovery after concussion (i.e., longer RTP clearance time) as it is after whiplash.^12–16,37^ However, we did not observe significant associations between cervical spine muscle morphometry and RTP clearance time in our univariable or multivariable models which adjusted for initial symptom severity. Similarly, both our unadjusted and adjusted results suggest that cervical spine MFI and muscle volume after adolescent concussion are not associated with subsequent injury risk. The lack of associations we observed between cervical spine muscle measures and concussion outcomes may be partially due to the timing of the MRI scans. The initial scan occurred an average of 10 days after concussion, and the follow-up scan occurred approximately 5.5 weeks after concussion. After whiplash, no differences in cervical spine multifidus MFI between those who recovered fully and those who developed persistent symptoms were observed at 4 weeks.^12^ However, at 3 months post-injury, MFI was higher among those with ongoing whiplash symptoms compared to those who recovered.^12^ Similarly, increased cervical spine multifidus volume has been observed between 3 months and 3 years after whiplash among individuals with persistent whiplash symptoms,^15^ though other studies have reported no associations between persistent whiplash symptoms and deep cervical spine extensor and flexor muscle volume.^38,39^ It may be that if changes in cervical spine muscle health exist after concussion, they would be observed at a later time post-concussion than our study evaluated. Thus, our MRI scans may have occurred too early during the post-injury period to adequately capture the chronic dysfunction necessary to result in cervical spine muscle volume and MFI changes on MRI.
In addition to the relatively early timing of our MRI scans after concussion, the young age of the participants in our study (13–18 years) may be another reason that our results were not significant. Previous whiplash studies reporting elevated cervical spine MFI after whiplash were performed among an adult population, and cervical spine muscle changes are more pronounced among adults compared to adolescents.^12–15^ Further, many of the previous MRI studies performed among the whiplash population included participants who sustained their injury in a motor vehicle collision.^12–16,37^ Given that we specifically excluded individuals who sustained their injury via a high velocity mechanism, such as a motor vehicle collision, the forces sustained during sport-related concussions may not have elicited alterations in cervical spine muscle health due to a lower velocity injury mechanism. Accordingly, future investigation of cervical spine muscle health after concussion among individuals with neck pain/neck-related dysfunction, an adult population, those with more chronic injuries and/or experiencing persistent concussion symptoms, and those who sustained concussions from high velocity mechanisms may be worthwhile to further the understanding of cervical spine involvement throughout concussion recovery.
Both of our outcomes of interest, RTP clearance time and post-concussion injury risk, are complex and highly variable. For example, RTP clearance requires that concussion symptoms have resolved.^2^ However, concussion symptoms are non-specific which may make it difficult to determine if symptoms are due to ongoing effects of the concussion, existed prior to the concussion, or were related to behavioral or biological function unrelated to the concussion.^40^ Further, while symptom resolution is necessary for RTP clearance, there are no established guidelines for determining physiologic recovery from concussion.^6^ Thus, athletes may be cleared for RTP based on symptom resolution while still having underlying physiological dysfunction that is not detected with current RTP assessments.^40^ There is also the subjective element of self-reported symptom resolution, which allows for potential over or under-reporting of symptoms, based on personal motivation to return to sports sooner or remain out of their sport longer.^41,42^ This collective evidence indicates that determining RTP readiness is complex and heterogenous, and objective markers to better determine readiness for RTP are needed.^43^ Earlier identification and management of concomitant cervical spine injuries after concussion has been suggested as a factor that may mitigate the risk of prolonged recovery associated with cervical spine dysfunction.^44^ However, our results suggest that cervical spine muscle health may not be useful for concussion recovery prognosis among adolescent athletes recruited from a specialized sports medicine clinic.
Similar to determining RTP readiness, identifying individuals at risk for subsequent post-concussion injuries is also a complex process. Although the increased risk of subsequent injuries after concussion RTP has been observed among male and female athletes across various sports and competition levels,^4,5^ the mechanisms of this increased risk are not fully understood. Many factors that have been investigated in the context of post-concussion injuries have not been associated with subsequent injuries and, therefore, do not help clinicians identify and intervene with at-risk individuals.^45,46^ Specifically, assessments of reaction time, standing balance, cognition, and mental health outcomes (i.e., anxiety and depression) were not associated with subsequent injuries.^45,46^ In contrast, previous research has identified changes in lumbar spine multifidi cross sectional area (i.e., smaller multifidi) among individuals with a concussion history.^47^ Smaller lumbar multifidi muscle size measured prior to a rugby season was associated with sustaining a concussion during the season^47^ which may have implications for other injuries, including subsequent injuries after RTP from concussion. Our results suggest that cervical spine muscle health measures obtained upon athletes receiving RTP clearance after concussion were not associated with subsequent injuries over the following year. It is possible that cervical spine muscle changes may develop later after concussion and may represent a potential mechanistic explanation for subsequent injuries, rather than a predictive factor as we evaluated in this study.
Our study limitations should be considered when interpreting the results. Our participant population of adolescent athletes who were recruited from a sports medicine clinic within a pediatric hospital in one geographical region may not generalize to all populations of athletes with concussion. The specialized nature of the sports medicine clinic may have also biased the study sample to participants with longer than average recovery times due to the complexity of their clinical presentation. Adults and individuals who sustain a concussion during a high velocity mechanism may have cervical spine muscle changes on MRI. Since our participant population consisted of adolescent athletes participating in a variety of sports, the variation across sports may have contributed to our non-significant findings. Collision/contact sports may result in higher velocity mechanisms of injury, which may yield different results when investigating cervical spine involvement after concussion. Our monthly injury surveys relied on accuracy of participant reporting, and the results could have been affected by recall bias. The timing of our initial and follow-up MRI scans may have occurred too early after concussion to capture chronic dysfunction that may lead to muscle changes on MRI. Thus, cervical spine muscle changes may occur later after concussion than our study investigated. The relatively small sample size may have also limited our power to detect associations with RTP time and injury risk, and future research with a larger sample may help further the understanding of factors associated with post-concussion recovery outcomes.
Cervical spine deep flexor and extensor MFI and muscle volume after concussion were not associated with RTP clearance time or subsequent injury risk among the adolescent athletes in our study. Although changes in cervical spine muscle health have been well documented among adults after whiplash, our results suggest that the same changes may not occur early after adolescent sport-related concussion.