Sensory integration processes characterize concussed athletes with balance deficits Open Access

The incidence of sport-related concussions (/mild traumatic brain injuries (mTBI)) is between 1.6 and 3.8 million annually in the USA (Langlois et al., 2006). SRC are associated to several symptoms such as headaches, dizziness, memory problems, and so forth (Didehbani et al., 2013; Echemendia et al., 2023; Lovell et al., 2006; Patricios et al., 2023) that can be long-lasting. In fact, repetitive head impacts have been associated to neurodegeneration in American football players (Daneshvar et al., 2021; Lehman et al., 2012; Nguyen et al., 2019) and soccer players (Mackay et al., 2019; Ueda et al., 2023).

There is a growing body of evidence that athletes with SRC have double the odds of sustaining a musculoskeletal injury after they return to sport compared to athletes without SRC (Brooks et al., 2016; Fino et al., 2019; Herman et al., 2017; Howell et al., 2018; Lynall et al., 2015). Howell et al. (2018) investigated dual-task gait patterns and symptoms in athletes post-concussion. Athletes who went on to sustain a subsequent time-loss injury after returning to sports demonstrated significant average walking speed dual-task cost worsening across time. In contrast, athletes who did not sustain an additional injury walked with consistent dual-task cost values across time (Howell et al., 2018). This indicates that SRC affect the motor control system.

Several studies showed long-term impacts of SRC onto the neuro-motor system (De Beaumont, Brisson, et al., 2007; De Beaumont et al., 2011, 2012; De Beaumont, Lassonde, et al., 2007; Howell, Buckley, et al., 2018; Howell, Lynall, et al., 2018; Powers et al., 2014). Previously concussed athletes showed altered postural control, and increased intracortical inhibition within the motor cortex that was related to the number of previous concussions (De Beaumont et al., 2011). It has, therefore, been argued that SRC particularly affect the sensorimotor system (Chmielewski et al., 2021), which results in an increased injury risk (McPherson et al., 2019), decreased neuromechanical responsiveness (Wilkerson et al., 2017), altered perception-action coupling (Eagle et al., 2020), and/or subtle neurocognitive and neuromuscular deficits (Howell, Lynall, et al., 2018).

Postural control dysfunction constitutes a core symptom of mTBI (Al-Husseini et al., 2022; Buckley et al., 2021; Cavanaugh et al., 2005; Felipe, 2021; Gera et al., 2018; Guskiewicz, 2011; Guskiewicz et al., 1997, 2001; Helmich et al., 2016, 2020; Jacob et al., 2022; Johnston et al., 2017; Kleffelgaard et al., 2012; Kumai et al., 2022; Lin et al., 2015; Manley et al., 2017; Martini et al., 2023; McCrory et al., 2017; Peterka, 2018; Purkayastha et al., 2019; Schmidt et al., 2018; Slobounov et al., 2005; Sosnoff et al., 2011; Valovich McLeod & Hale, 2015; Wood et al., 2019). Thus far, balance problems showed to be present 30% of the time following a concussive injury (Guskiewicz, 2011), which can remain for weeks or even years after the incident (Broglio & Puetz, 2008; Ingersoll & Armstrong, 1992; Sosnoff et al., 2011; Thompson et al., 2005; Valovich McLeod & Hale, 2015; Wood et al., 2019). Concussed individuals showed to be highly affected from postural deficits when sensory input is being altered such as during eyes closed conditions, and/or unstable surface conditions (Gera et al., 2018; Guskiewicz, 2011; Guskiewicz et al., 2001; Helmich et al., 2016, 2020; Lin et al., 2015; Martini et al., 2023). When controlling posture, sensory information must be integrated into the motor-sensory system (Horak, 2006; Peterka, 2002, 2018; Solis-Escalante et al., 2019; Van Der Kooij & Peterka, 2011). Decreased balance function has been attributed to deficits of sensory integration processes (Gera et al., 2018; Guskiewicz, 2011; Guskiewicz et al., 1997); however, this assumption has not been assessed via brain imaging tools yet. In fact, common neuroimaging methods such as Magnetic Resonance Imaging (MRI) and/or Positron Emission Tomography (PET) are restricted in the recording of real-life scenarios such as postural control tasks (Stangl et al., 2023).

Thus far, only a few studies investigated neuronal processes during postural control tasks in individuals with mTBI by the application of functional near-infrared spectroscopy (fNIRS) and/or Electroencephalography (EEG; De Beaumont et al., 2011; Helmich et al., 2016, 2020; Jacob et al., 2022; Slobounov et al., 2005; Thompson et al., 2005; Urban et al., 2020). The investigation of movement-related cortical potentials (MRCP) measured by EEG in concussed athletes revealed a persistent reduction of MRCP amplitude in frontal and posterior brain regions prior to initiation of postural movement up to 30 days post-injury (Slobounov et al., 2005). This study also showed that brain activation patterns were altered although abnormal postural responses recovered within 10 days post-injury (Slobounov et al., 2005). The investigation of EEG power spectral density in delta and theta bands revealed that a symptomatic concussion group could be differentiated from a non-concussed group when applying a combined setup of balance and neuroimaging methods (Jacob et al., 2022). The authors concluded that concussed individuals may increase attention and cognitive effort when compared to individuals without a history of concussion (Jacob et al., 2022). Further studies indicated that athletes with SRC are characterized by hyperactivation in parietal cortices during postural control tasks (Urban et al., 2020). When balancing on an unstable (/foam) surface with open eyes, concussed individuals demonstrated hyperactivation in the left inferior parietal cortex (IPC) as well as in the left DLPFC but lower activation the right-hemispheric parietal cortex (Urban et al., 2020).

The parietal cortex may constitute a central interface in which visual, cognitive, and motor-related signals are integrated in order to influence task-dependent motor control (Freedman & Ibos, 2018). When individuals receive tactile stimuli on their feet and are asked to locate the input, sensory processes are coded within the parietal cortex (Klautke et al., 2023). Because concussed participants showed to respond more strongly to visual and vestibular stimuli during upright stance, it has been argued that SRC may increase the dependence on visual and vestibular feedback (Caccese et al., 2021). This indicates that concussed individuals may be particular depended on the increased integration of sensory information within somatosensory (/parietal) cortices during the control of posture. Furthermore, improvement in postural control following rehabilitation treatment in individuals’ acquired brain injuries is associated to reduced functional connectivity in sensorimotor cortices (Joubran et al., 2022). This further points out that SRC may increase sensory integration processes during the control of posture. With regard to previously observed parietal hyperactivation during postural control (Urban et al., 2020) and the argument that concussed individuals may increase their cognitive effort to control balance (Jacob et al., 2022), we hypothesized that athletes with persistent symptoms after SRC are characterized by neural upregulation in the postcentral (sensory/parietal) cortex during sensory integration processes, that is, when controlling posture during altered sensory balance conditions.

Sixty-six individuals (mean age: 23.4 ± 5.2 years (minimum [min]: 16, maximum [max]: 40; 50 men, 16 women) participated in this study. A priori power analysis by using G*Power 3.1.9.7 indicated that 36 participants were necessary for a statistical analysis between groups and repeated measures (effect size f = 0.25, Power = 0.95, calculated critical F = 2.696, calculated actual Power = 0.95). All participants were active athletes from various sports, including American football, soccer, and ice hockey. Individuals were categorized into three groups based on their post-concussion symptom scale score and their concussion history: 22 concussed and symptomatic athletes (PCSS > 10: symptomatic group (PCSS = 26.9 ± 14.3 (min: 12, max: 61)); mean age: 24.4 ± 5.7 years (min: 16, max: 39); number of experienced concussions: 2.8 ± 1.9 years (min: 1, max: 7)), 22 concussed and asymptomatic athletes (PCSS ≤ 10: asymptomatic group (PCSS = 3.3 ± 2.9 (min: 0, max: 10)); mean age: 24.05 ± 5.3 years (min: 16, max: 40); number of experienced concussions: 2.0 ± 1.6 years (min: 1, max: 8)), and 22 non-concussed and asymptomatic athletes (PCSS ≤ 10: control group (PCSS = 2.0 ± 2.9 (min: 0, max: 0); mean age: 21.8 ± 4.2 years (min: 16, max: 30), Table 1). Thus, groups differed significantly regarding their PCSS scores (PCSS, F(2, 63) = 58.780, p < 0.001)). The two concussed groups also significantly differed regarding their time post the concussive incident (t(42) = -2.392, p < 0.05). The time post injury was on average shorter in the symptomatic group (mean time post-concussion: 14.8 ± 5.5 months) when compared to the asymptomatic group (mean time post-concussion: 33.3 ± 5.5 months; Table 1). Furthermore, the groups differed regarding their time of participation in sports (F(2, 63) = 4.096, p < 0.05). The symptomatic showed to participate in sports significantly longer than the asymptomatic group (p < 0.05). The symptomatic and asymptomatic concussed groups did not significantly differ regarding the number of experienced concussions (χ²(df = 6, N = 44) = 7.974, p = 0.240). Asymptomatic and control athletes were matched to the symptomatic group for age and gender (i.e., no significant differences were found between groups (age, χ²(df = 38, N = 66) = 29.871, p = 0.824; gender, χ²(df = 2, N = 66) = 4.620, p = 0.099)).

Exclusion criteria included a history of neurological, psychiatric, equilibrium, or hearing disorders that could potentially interfere with their ability to perform balance and cognitive tasks. Prior to participation, all participants provided informed consent after being instructed about the experimental procedures. The study was approved by the local Ethics Committee.

Individuals were instructed to control their posture in a standardized standing position (distance between feet: 2 cm, hands on hips, view straight ahead on a marked cross; Fig. 1). Four postural control conditions were carried out according to the Clinical Test of Sensory Interaction in Balance (Shumway-Cook & Horak, 1986), which examines combinations of visual and tactile manipulations during balance control. However, we modified conditions in order to make balance conditions compatible to brain activation measures applying fNIRS. Thus, postural control tasks were either performed with open and/or closed eyes and/or either on a firm (/stable) surface or on an unstable surface: condition 1 (c1): eyes open and stable surface; condition 2 (c2): eyes closed and stable surface; condition 3 (c3): eyes open and unstable surface; condition 4 (c4): eyes closed and unstable surface. The unstable surface was created using a piece of six cm thick foam pad (“AIREX Balance-Pad”). We created a block design with two blocks of each balance condition. Each block included three trials (10 seconds per trial; 15 seconds between conditions/instruction; 5 seconds between trials), resulting in a total of six trials per condition (Fig. 1). Between conditions, the participants were asked to stand still with their hands on their hips and open eyes. We applied a force plate system (“ZEBRIS platform, type FDM-S”, measure frequency 240 Hz) to register center of mass displacement (/postural sway) by measuring ground reaction forces. This system provides information about the ability to keep postural control by the registration of the deviations from the center of pressure (COP) by the mean length of the movement path (/track) per time (millimeters/second [mm/s]); COP track length is defined as the absolute length of the COP path movements throughout the testing period (10 seconds). The second parameter COP area (measured in square millimeters [mm2]) provides information about the area that is used to control posture during the 10 seconds of sway. The means of COP track length and COP area were exported for each subject and each condition for statistical analyses.

Functional near-infrared spectroscopy (fNIRS) maps human brain function by measuring and imaging local changes in hemoglobin concentrations in the brain that arise from the modulation of cerebral blood flow and oxygen metabolism by neural activity (Yücel et al., 2017). Similar to functional magnetic resonance imaging (fMRI), fNIRS detects neurovascular coupling to infer changes in neuronal activity. Neurovascular coupling refers to changes in local neural activity, which are correlated to changes in local cerebral blood flow (CBF; Tachtsidis & Scholkmann, 2016). Cerebral oxygenation changes were recorded using a portable continuous wave fNIRS system (NIRSport 2, NIRx, Medical Technologies LLC, Berlin, Germany; wavelengths of 760 and 850 nm; sampling rate 10.2 Hz). The NIRSport 2 system contains eight light sources and eight light detectors, and a short-distance detector bundle. The optodes were placed according to the 10–20-system (Jasper, 1958) above pre- and postcentral gyri of each hemisphere (Fig. 1; Table 2) using a standardized cap (EasyCap GmbH, Herrsching, Germany) and registrated via the fNIRS Optodes’ Location Decider (fOLD) toolbox (Zimeo Morais et al., 2018). The optodes covered pre- and postcentral cortices, including the primary motor cortex, primary somatosensory cortex, and parietal sensory association cortex in both the right and left hemisphere of the brain (Fig. 1; Table 2). Data were recorded from 18 (long-distance) channels of measurement. Eight short-distance (8 mm) channels were additionally applied to account for changes in extra-cerebral blood flow. The mean source-detector distance (long-distance channels) was 38.1 ± 2.7 mm.

The fNIRS data were analyzed using the Satori (v.1.8) toolbox (Lührs & Goebel, 2017). The raw intensity data were converted to optical density, which was converted via the modified Beer–Lambert law (MBLL) into concentration changes of oxygenated hemoglobin (∆HbO₂) and deoxygenated hemoglobin (∆HbR). Triggers were set manually for each trial of postural control tasks. Movement artifacts were corrected by applying the (I) spike removal motion correction function (settings: iterations: 10, lag: 5 seconds, threshold: 3.5, influence: 0.5, interpolation: monotonic) and (II) the temporal derivative distribution repair (TDDR) function according to Fishburn et al. (2019). Because the use of short-separation detector measurements as a regressor in the general linear model (GLM) has been previously shown to statistically improve HRF estimation (Gagnon et al., 2011; Tachtsidis & Scholkmann, 2016), we used short-distance signals to regress out signals from extra-cerebral layers from the long-distance channels. To account for cardiac oscillations and Mayer-waves, we used a 0.4 Hz low-pass filter, a high-pass filter (butterworth) of 0.01 Hz, and the linear detrending function of Satori. The data were normalized by z-transformation, and the betas of the hemodynamic response were estimated applying the GLM. Betas of each channel and condition were exported for further statistical analyses.

Comparisons of the mean(s) (repeated (rmANOVA) analyses of variance) were performed for postural control (COP area, COP track length) and brain oxygenation parameters (∆HbO₂, ∆HbR) using IBM SPSS statistics (Version 28). Each parameter was additionally tested by univariate analyses (uniANOVA). For the betas of ∆HbO₂ and ∆HbR as well as for COP area and COP track length, we calculated the repeated within-subjects factors vision (eyes opened/closed) and surface (stable/unstable surface). For the analysis of brain activity, we additionally integrated the within-subjects factor regions of interest (ROI): (I) precentral, (II) postcentral, and (III) parietal cortices. The three ROIs were calculated by the mean oxygenation of the channels covering the corresponding brain regions: precentral: ch1 and ch11; postcentral: ch2–ch4, and ch12–ch13; parietal: ch5–ch10, and ch14–ch18 (Table 2). The between-subject factor group was calculated between symptomatic, asymptomatic and control individuals. Significant results are reported from p < 0.05. Multiple post hoc pairwise comparisons were corrected with Bonferroni corrections. If the requirements of the ANOVA (i.e., sphericity) were violated, we used the Greenhouse-Geisser correction. Furthermore, t-contrasts of brain activation during postural control were calculated applying the GLM (with the threshold of p < 0.05). The following contrasts were statistically calculated for each group separately: (I) eyes closed versus eyes open condition overall surface conditions; (II) unstable surface versus stable surface condition overall vision conditions; and (III) overall conditions versus baseline. The baseline refers to any period not defined as a task, that is, here this concerned standing with hands resting on the hips and eyes opened. Visual depiction of t-contrasts (∆HbO₂) was performed for each channel by applying the Satori toolbox. Furthermore, in order to understand the individual variation of concussed and symptomatic athletes from both other groups, we added visual variation maps (∆HbO₂) for each condition to the Appendix Figures A1 and A2. According to best practice suggestions (Yücel et al., 2021), we report significant effects of both chromophores (∆HbO₂, ∆HbR) in the results section. However, in line with previous studies (e.g., Helmich et al., 2020; Hocke et al., 2018; Urban et al., 2020), we focus on ∆HbO₂ in the discussion.

The rmANOVA showed (over both balance parameters) a significant effect of group (F(4, 126) = 2.888, p < 0.05, partial eta squared [η²] = 0.084; Table 3). The univariate ANOVA (uniANOVA) for the factor COP area revealed a statistically significant effect for the factor group (F(2, 63) = 3.654, p < 0.05, η² = 0.104). Post-hoc pairwise comparisons showed a significantly increased COP area in the symptomatic group when compared to the control group (p < 0.05; Fig. 2B).

The rmANOVA showed over both parameters a significant effect of group × vision (F(4, 126) = 3.305, p < 0.05, η² = 0.095). The uniANOVA for the factor COP area demonstrated a statistically significant interaction effect for group × vision (F(2, 63) = 3.178, p < 0.05, η² = 0.092; Table 3). Post-hoc comparisons revealed a significantly increased COP area in the symptomatic group as compared to the control group during the closed eyes condition (p < 0.05).

The rmANOVA showed a statistically significant effect for the factors vision (F(2, 62) = 316.896, p < 0.001, η² = 0.911), surface (F(2, 62) = 112.783, p < 0.001, η² = 0.784), and the interaction vision × surface (F(2, 62) = 92.729, p < 0.001, η² = 0.749; Table 3).

The uniANOVA for COP area showed a statistically significant effect for the factor vision (F(1, 63) = 67.008, p < 0.001, η² = 0.515), surface (F(1, 63) = 264.386, p < 0.001, η² = 0.808), the interaction vision × surface (F(2, 63) = 109.735, p < 0.001, η² = 0.635; Table 3). Post-hoc comparisons for the factor vision demonstrated an increased COP area during the eyes closed condition when compared to the eyes open condition (p < 0.001; Fig. 3). Post-hoc comparison for factor surface demonstrated an increased COP area during the unstable surface condition when compared to the stable surface condition (p < 0.001; Fig. 4). Post-hoc comparison for the interaction of vision × surface showed an increased COP area during the open eyes condition on the unstable surface when compared to the stable surface condition (p < 0.01); during the closed eyes condition on an unstable surface when compared to the stable surface condition (p < 0.001); during the stable surface condition with closed eyes when compared to the open eyes condition (p < 0.001); and during the unstable surface with closed eyes condition and when compared to the open eyes condition (p < 0.001; Table 3).

The uniANOVA for COP track length showed significant effects for the within-subjects factors vision (F(1, 63) = 214.528, p < 0.001, η² = 0.773) and surface (F(1, 63) = 639.551, p < 0.001, η² = 0.910), and for the interaction of vision × surface (F(1, 63) = 188.248, p < 0.001, η² = 0.749; Table 3). Post-hoc comparison for factor vision demonstrated an increased COP track length during the eyes closed condition when compared to the eyes open condition (p < 0.001; Fig. 3A). Post-hoc comparison for the factor surface demonstrated an increased COP track length during the unstable surface condition when compared to the stable surface condition (p < 0.001). Post-hoc comparison for the interaction of vision × surface showed in increased COP track length during the open eyes condition on unstable surface when compared to the stable surface condition (p < 0.001); during the closed eyes condition on an unstable surface when compared to the stable surface condition (p < 0.001); during the stable surface condition with closed eyes when compared to the open eyes condition (p < 0.001); and during the unstable surface with closed eyes condition and when compared to the open eyes condition (p < 0.001; Table 3).

The uniANOVA for ΔHbO₂ showed significant group effects in the parietal ROI (F(2, 63) = 3.212, p < 0.05, η² = 0.093); Table 4). Post-hoc pairwise comparison revealed marginally significant higher ΔHbO₂ in the symptomatic group when compared to the asymptomatic group in the parietal ROI (p = 0.052; Fig. 2C).

The uniANOVA for ∆HbR showed marginally significant interaction effects for group × vision in the precentral ROI (F(2, 63) = 3.124, p = 0.051, η² = 0.090; Table 4). Post-hoc pairwise comparison revealed a significantly reduced ΔHbR in the control group during the eyes closed condition when compared to the eyes opened condition (p < 0.05).

The rmANOVA for the factor vision showed significant effects for ΔHbO₂ (F(3, 61) = 3.976, p < 0.05, η² = 0.164; Table 4). The uniANOVA for the factor vision showed significant effects for ΔHbO₂ the postcentral (F(1, 63) = 10.231, p < 0.01, η² = 0.140) and parietal ROI (F(1, 63) = 9.620, p < 0.01, η² = 0.132). Post-hoc pairwise comparison revealed significantly increased ΔHbO₂ during the eyes closed condition when compared to the eyes open condition in the postcentral (p < 0.01) and the parietal ROI (p < 0.01; Fig. 3C).

The uniANOVA for the factor surface showed significant effects for ΔHbO₂ in the parietal ROI (F(1, 63) = 4.436, p < 0.05, η² = 0.066; Table 4). Post-hoc comparisons showed higher ΔHbO₂ during the unstable surface condition when compared to stable surface conditions (p < 0.05; Fig. 4).

The uniANOVA results revealed for ΔHbR significant effects for the interaction effect of vision × surface in the postcentral ROI (F(1, 63) = 5.210, p < 0.05, η² = 0.076). Post-hoc comparison revealed reduced ΔHbR during closed eyes when compared to open eyes conditions during the stable surface condition (p < 0.01).

The symptomatic group showed significantly increased t-contrasts (ΔHbO₂) overall balance conditions in ch2 ((t = 2.56, p < 0.05); Table 5 (I); Fig. 2D)). The asymptomatic group showed significantly decreased t-contrasts in ch5 (t = -2.65, p < 0.05), ch7 (t = -3.55, p < 0.05), ch9 (t = -4.98, p < 0.05), ch14 (t = -2.16, p < 0.05), ch16 (t = -2.67, p < 0.05), and ch17 (t = -2.62, p < 0.05) The control group showed significantly decreased t-contrasts in ch2 (t = -2.03, p < 0.05), ch3 (t = -2.17, p < 0.05), ch8 (t = -2.21, p < 0.05), ch9 (t = -3.33, p < 0.05), and ch10 (t = -2.18, p < 0.05).

The symptomatic group showed significantly increased t-contrasts (ΔHbO₂) on the unstable surface condition when contrasted to stable surface condition in ch1 (t = 2.20, p < 0.05), ch5 (t = 2.26, p < 0.05), ch9 (t = 2.04, p = 0.054), and ch18 (t = 2.61, p < 0.05; Table 5 (II)). The asymptomatic group showed significantly increased t-contrasts during the unstable surface condition when contrasted to stable surface condition in ch2 (t = 2.249, p < 0.05). The control group showed no significant contrast results.

The symptomatic group showed significantly increased t-contrasts (ΔHbO₂) during the eyes closed condition when contrasted to the eyes open condition in ch7 (t = 2.71, p < 0.05; Table 5 (III)). The asymptomatic group showed significantly increased in ch16 (t = 2.18, p < 0.05). The control group showed significantly increased t-contrasts in ch2 (t = 2.50, p < 0.05), ch5 (t = 2.11, p < 0.05), ch7 (t = 2.66, p < 0.05), ch8 (t = 3.51, p < 0.05), ch9 (t = 2.72, p < 0.05), and ch16 (t = 2.38, p < 0.05).

The contrast of eyes closed versus eyes open releveled overall participants significant t-contrasts (ΔHbO₂) in ch2 (t = 3.01, p < 0.05), ch7 (t = 3.74, p < 0.05), ch12 (t = 2.67, p < 0.05), and ch16 (t = 2.92, p < 0.05; Fig. 3D; Table 6). The contrast of unstable surface versus stable surface releveled overall participants’ significant t-contrasts in ch6 (t = 2.06, p < 0.05) and ch10 (t = 2.19, p < 0.05; Fig. 4D; Table 6).

We present functional neuroimaging evidence of altered brain activation in a sample of concussed (with and without ongoing post-concussion symptoms) and non-concussed athletes during postural control tasks. This study investigated postural sway and simultaneous brain oxygenation patterns in the motor-sensory cortex during balance tasks with closed and opened eyes as well as during stable and unstable surface conditions. The results showed that symptomatic concussed athletes were characterized by increased postural sway when compared to athletes without experienced concussions overall postural control conditions. Symptomatic concussed athletes showed increased brain activation within postcentral/parietal cortices overall balance conditions and when compared to asymptomatic athletes. In overall participants, increased postural sway was observed during postural control with closed eyes when compared to eyes open conditions as well as during unstable surface conditions when compared to stable surface conditions. Increased postural sway during eyes closed conditions was accompanied by increased brain oxygenation within postcentral and parietal brain regions and when compared to open eyes conditions. The contrast of unstable versus stable surface conditions revealed significant increased brain activation within parietal brain cortices.

The present study showed that concussed individuals with high self-reported symptoms are characterized by increased postural sway independently of balance condition, that is, when balancing with closed and open eyes as well as when balancing on unstable and stable surfaces. Balance maintenance is achieved by integration of sensory information to the motor system (Horak, 2006; Peterka, 2002, 2018; Solis-Escalante et al., 2019; Van Der Kooij & Peterka, 2011). Because various sensory systems (somatosensory/proprioception, visual, vestibular) contribute to balance control, it has been argued that impaired postural control after mild traumatic brain injuries may be grounded in a deficit of integrating sensory information (Gera et al., 2018; Guskiewicz, 2011; Guskiewicz et al., 2001). Standing on an unstable surface alters the somatosensory and proprioceptive inputs that need to be processed and (neuronally) integrated (Gera et al., 2018). Here, postural control showed to be impaired in concussed and symptomatic athletes overall conditions as well as during the unstable surface condition. The unstable surface reduces the accuracy of the orientation information (Guskiewicz, 2003). Increased postural sway during this condition (i.e., when proprioception from the feet is eliminated) suggests that concussed individuals cannot compensate by other sensory systems for the loss of somatosensation. Gera et al. (2018) found that postural sway was higher in concussed individuals when compared to controls during several postural control conditions such as when balancing with closed and open eyes on a stable surface as well as on an unstable surface. However, whereas the latter authors investigated concussed individuals 2–3 days post-concussion, in the present study we investigated individuals that reported ongoing symptoms long-term post-concussion. In fact, balance impairments have been reported up to 4 years after a concussive injury (Geurts et al., 1996; Kleffelgaard et al., 2012). Because we also found an effect between groups overall balance conditions, we assume that individuals that suffer from post-concussion symptoms long-term must be affected from a general defect of integrating sensory information.

Central sensorimotor integration constitutes the ability to process and integrate sensory information in order to transform it into a motor output. This function constitutes a critical aspect of balance control (Gera et al., 2018). It is believed that communication between the different sensory systems is lost in pathologic conditions such as, for example, in moderate/severe traumatic brain injury, cerebellar atrophy, and ataxia, causing moderate to severe postural instability (Guskiewicz, 2003). A concussion may lead to dysfunction of neurotransmission (Solar et al., 2024) and therefore may impair the brain’s ability to efficiently integrate the sensory inputs, especially during challenging surface conditions (e.g., unstable surface conditions). A study applying magnetoencephalography in a sample of military members with a history of repetitive subconcussive impacts revealed disrupted neuronal activity, including neural slowing, and functional dysconnectivity (Solar et al., 2024).

Here, the application of fNIRS above the motor-sensory cortex showed over all postural control conditions that concussed athletes with ongoing symptoms are characterized by increased brain oxygenation, particularly within the parietal cortex. This is not the case in asymptomatic and/or control athletes. The balance control system involves a hierarchical network of neural centers such as the cerebral cortex, cerebellum, basal ganglia, brainstem, and spinal cord that are related by peripheral and central feedback mechanisms and controlling voluntary movements (Guskiewicz, 2003). Higher level (/hierarchical) systems of postural control include the basal ganglia, cerebral cortex, and cerebellum, whereas lower levels are represented by the brainstem and spinal cord (Lalonde & Strazielle, 2007). Whereas lower hierarchical regions trigger more automatic postural responses, the cerebral cortex is involved in postural control each time a person must consciously maintain balance, for example, while walking across a slippery floor (Kandel et al., 2013). The present data showed that individuals of all experimental groups increase their brain processes when balancing either with closed eyes (vs. open eyes) or when controlling posture during unstable (vs. stable) surface conditions. Increased brain activation overall balance conditions in symptomatic concussed athletes (but not in asymptomatic and control athletes), therefore, indicates that these individuals activate higher hierarchal neuronal resources to control for postural sway. It has been argued that concussed individuals may increase attention and cognitive effort when compared to individuals without a history of concussion (Jacob et al., 2022). Because the cerebral cortex has more control over anticipatory postural adjustments than automatic postural reactions (Kandel et al., 2013), the present findings indicate that postural control does not constitute an automatized process but rather relies on explicit cortical control in individuals that suffer from long-term post-concussion symptoms.

Multisensory integration showed to be impaired in individuals with lesions to the parietal cortex (Derouesne et al., 1984) and can increase body sway (Pérennou et al., 2000; Young et al., 2020). Here, increased brain oxygenation patterns were not only evident in concussed and symptomatic athletes within the parietal cortex but also overall individuals during eyes closed (vs. eyes open) conditions as well as during unstable (vs. stable) surface conditions. This not only evidences that the parietal cortex serves sensorimotor integration when receiving input from lower body movements, and movements coordinating multiple body parts (Sereno & Huang, 2014) but also points out that it may constitute a (I) characteristic brain region of sensory integration during postural control tasks as well as (II) a specific region of post-concussion balance deficits. Higher activation within the parietal cortex was also observed in a previous study applying fNIRS during postural control in a youth group with recent concussions (Urban et al., 2020). It has been suggested that the additional recruitment of certain brain regions following a concussion may reflect increased mental processing required to perform a task (Bryer et al., 2013; Plenger et al., 2016; Urban et al., 2020). Thus, the present findings indicate that concussed and symptomatic athletes must not only explicitly control posture but also the parietal cortex upregulates its activity for additional sensory integration processes post-concussion.

Previous studies that investigated skilled athletes showed that such individuals were characterized by decreased of cortical activity during visuo-motor performances (Del Percio et al., 2009). In fact, skilled athletes with high postural control showed to recruit brain cortices for additional cognitive tasks other than balance control (Chen et al., 2023). These findings are in line with the “neural efficiency hypothesis” indicating that highly trained skills become more automated (Debarnot et al., 2014) and therefore necessitate minimal energy consumption (Nakata & Yabe, 2001). The fact that individuals with post-concussion symptoms are characterized by increased brain activation within parietal cortices overall postural control conditions therefore indicates that these individuals upregulate sensory integration processing more than healthy individuals do in order to keep balance under control. However, as the sway data indicate, even increased parietal activity does not help to counteract existing balance deficits. Thus, the overactivation of such athletes may constitute an ineffective way to compensate for postural deficits as such processes may serve a compensatory function (Reuter-Lorenz & Cappell, 2008). We, therefore, conclude that balance control does not constitute an automatized process in concussed athletes that suffer from post-concussion symptoms but the upregulation of neural activation within the parietal cortex indicates an inefficient (/compensatory) neural mechanism to integrate additional sensory information for deficient balance control.

While we report new findings about neural and behavioral characteristics of athletes with long-term post-concussion symptoms, it is important to acknowledge the limitations of this study. Our sample consisted of unbalanced groups with regards to gender. Knowing that there may exist gender differences regarding clinical symptoms, alterations in brain structure and function, and recovery trajectories (Koerte et al., 2020), future studies must account for balanced groups to better understand potential outcomes between the sexes.

We also did not account for the heterogeneity of injury mechanisms, locations, and severity. Furthermore, both concussed groups reported their (multiple) concussive injuries long-term after the incident. Because previous studies applying balance protocols and fNIRS analyses focused on youth athletes only with experienced concussions within weeks of injury (e.g., Urban et al., 2020), different results may be grounded in the different samples. Thus, next steps must not only consider the investigation of athletes before and after concussions but also add investigations about short- and long-term post-concussion parameters within the same sample.

Regarding the method of analysis, fNIRS is an increasingly used method in neuroscience research. However, it is also limited in its application. When compared to fMRI, it lacks spatial sensitivity as it is restricted to superficial layers of the cortex. Furthermore, whereas a typical voxel size in human fMRI detects signals within the range of millimeters, fNIRS detects signals only within centimeters. Similar to fMRI, fNIRS relies on neurovascular coupling to infer changes in neuronal activity. However, there is the possibility that changes of oxygenation may not uniquely be caused by neurovascular coupling. Change in fNIRS signals can also be due to changes in intracerebral hemodynamics caused by task-related systemic activity and/or extracerebral hemodynamics (Tachtsidis & Scholkmann, 2016). By including short-separation channels (to regress out the superficial hemodynamics) and carefully designing the balance protocol (e.g., several blocks per condition; not comparing versus baselines but between conditions), we tried to minimize potential “false positives” within the present study.

This study provides new evidence about the neuronal and behavioral outcomes of long-term post-concussion symptoms in athletes. The combined application of postural control tasks and fNIRS above motor-sensory cortices showed that concussed athletes with postural deficits are characterized by increased brain oxygenation, particularly within the parietal cortex. Such neural upregulation indicates that postural control in concussed athletes constitutes a rather conscious than unconscious process. Individuals without post-concussion symptoms may control posture by a more automatic/unconscious strategy. Thus, concussed and symptomatic athletes increase their cognitive effort to control posture, which may increase energy demands. Because it has been argued that higher susceptibility to injuries after concussion may be grounded in impaired motor-sensory control mechanisms (Chmielewski et al., 2021), we conclude that upregulation of parietal activity during sensory integration processes constitutes a characteristic of concussed individuals. Whether this pattern is related to an increased risk of injury must be investigated in future studies.

The data will be shared upon request.

Bhagyashree Singh collected/analyzed the data and wrote the manuscript. Ingo Helmich planned the study, collected/analyzed the data, and wrote and revised the manuscript. All authors discussed the results and contributed to the final version of the manuscript.

This research was supported by the Federal Institute for Sports Science (ZMI4-070401/21-23).

The authors disclose no conflicts of interest.