Abstract
The default mode network (DMN), a set of brain regions, has been shown to be affected post-concussion.
This cross-sectional study aims to elucidate if children and adolescents with multiple concussions demonstrate long-term alterations in DMN functional connectivity (FC).
Participants (N = 57, 27 girls and 30 boys; 8-19 years old, M age = 14.7, SD = 2.8) were divided into three groups (orthopedic injury [OI] n = 20; one concussion n = 16; multiple concussions n = 21, M = 3.2 concussions, SD = 1.7) and seen on average 31.6 months post-injury (range 4.3-130.7 months; SD = 19.4). They underwent a resting-state functional magnetic resonance imaging scan. Parents completed the ADHD rating scale-5 for children and adolescents. Children and parents completed the post-concussion symptom inventory (PCSI).
Anterior and posterior DMN components were extracted from the fMRI data for each participant using FSL’s MELODIC and dual regression. We tested for pairwise group differences within each DMN component in FSL’s Randomize (5000 permutations) using threshold-free cluster enhancement to estimate cluster activation, controlling for age, sex, and symptoms of inattention. FC of the anterior DMN was significantly reduced in the group with multiple concussions compared to the two other groups, whereas there were no significant group differences on FC of the posterior DMN. There were no significant associations between DMN FC and PCSI scores.
These results suggest reduced FC in the anterior DMN in youth with multiple concussions, but no linear association with post-concussive symptoms.
Introduction
Over the past decades, awareness of concussion and its prevalence and recovery in children and adolescents has grown significantly, and the number of children diagnosed with concussion has increased accordingly (Coronado et al., 2015; Rosenthal et al., 2014). Clinical recovery from concussion is usually achieved within a month for most children and adolescents (Davis et al., 2017) with a minority experiencing a slower recovery (Barlow et al., 2015; Henry et al., 2016). Despite the positive outcome for most children, crucial questions remain unanswered, particularly whether concussions are associated with long-term neurobiological sequelae in youth and whether a dose-response relationship exists between the number of concussions and brain functioning (i.e., are more concussions associated with greater long-term neurobiological consequences?).
Diverse non-invasive neuroimaging techniques have been used to study brain functioning post-concussion, yielding a wide range of results and limited conclusions about specific brain networks in children and adolescents (Mayer et al., 2018; Schmidt et al., 2018). One particular brain network of interest measured using resting-state functional magnetic resonance imaging (rsfMRI) is the default mode network (DMN), a set of brain regions that are task negative, meaning they are more active in the absence of a particular task (Sharp et al., 2014). The DMN involves regions in the medial prefrontal cortex, posterior cingulate/precuneus (PCC), and lateral parietal cortex (Fox et al., 2005). The DMN is involved in self-referential monitoring processes, such as introspection and experiential memory (Buckner & Carroll, 2006; Raichle, 2015) and is known to facilitate task-switching and engagement of other functional networks (de Pasquale et al., 2012; Dunkley et al., 2018), such as those important for memory, attention, and executive functions more generally (Buckner et al., 2008; Rohr et al., 2019), all of which can be affected post-concussion (Baillargeon et al., 2012; van der Naalt et al., 1999). Given the multitude of possible symptoms following concussion, it stands to reason that investigating a well-connected “monitoring” network deep in the brain that is in constant touch with other brain networks could provide crucial insight into the occurrence of these symptoms. Research looking at the DMN and post-concussive recovery was thus first conducted to investigate if (1) its functional connectivity (FC) would be altered post-injury and reflect neuronal reorganization, and if (2) changes in DMN FC would be related to post-concussive functioning including emotional, cognitive, and behavioral domains.
The DMN was first studied in adults post-concussion, with reduced FC reported at 10 days post-concussion (Johnson et al., 2012) and from 3 weeks to 3–5 months post-injury (Mayer et al., 2011). Another study of adults (Zhou et al., 2012) found hypoconnectivity in the posterior DMN and hyperconnectivity in the anterior DMN at approximately 3 weeks post-injury. Additionally, both reduced FC (Mayer et al., 2011) and increased anterior DMN FC (Zhou et al., 2012) have been associated with higher post-concussive symptoms, comprised of somatic and sleep-related disturbances, cognitive issues, and especially psychological concerns. These results suggest a link between concussions, DMN FC, and post-concussive symptoms. However, this link cannot be generalized to children and adolescents given that the DMN undergoes maturation during development (Rohr et al., 2017, 2018; Sato et al., 2016).
Studies in younger populations have shown divergent rsfMRI results in contrast to those in adults. Murdaugh et al. (2018) demonstrated that acutely after concussion (within 7 days post-injury), adolescents showed hypoconnectivity of the anterior DMN and hyperconnectivity of the posterior DMN as compared to controls. This pattern of results was also found in adolescents tested sub-acutely (Borich et al., 2015; 2 months or less post-injury) and long after the injury (Orr et al., 2016; on average 40.6 months post-concussion). However, these results were not replicated by Murdaugh et al., (2018) when they retested the same adolescents on average 21 days post-injury; they found no significant differences between groups. In addition, at approximately 30 days post-injury, Newsome et al. (2016) found hyperconnectivity of the DMN in youth with concussion compared to controls, particularly between the PCC and the ventral lateral prefrontal cortex and between the right lateral parietal lobe and the lateral temporal cortex, but no significant differences in other DMN areas (i.e., medial prefrontal cortex and left lateral parietal lobe). Manning et al., (2017) have also reported only hyperconnectivity of the DMN in adolescents seen by 3 months post-concussion compared to controls. Thus, to-no-clear patterns have yet emerged regarding the DMN FC post-concussion.
No known studies have compared DMN FC of participants with a single versus multiple concussions, which is also consistent with the limited literature on psychological and behavioral effects of repetitive injuries (McAllister & McCrea, 2017). Additionally, adolescents and male participants were overrepresented in previous studies: age ranged from 13 to 23 years, with only one study extending down to age 11 years (Manning et al., 2017), and participants were only male or included 83% (Borich et al., 2015) and 92% (Newsome et al., 2016) male participants. Finally, only Orr et al. (2016) focused on the chronic period post-injury (more than 3 months post-concussion) and looked at correlations between the DMN FC and behavioral factors. They observed a significant positive correlation between FC of the posterior DMN and externalizing symptoms, but no significant correlations between DMN FC and post-concussive symptoms. Therefore, future studies are warranted to investigate the DMN FC and to compare youth participants with single or multiple concussions while controlling for sex, age, and behavioral variables.
Therefore, the main objective of this study was to investigative long-term DMN FC in children and adolescents who sustained either single or multiple prior concussions, compared to a control group with OI. To do so, we used a data-driven independent component analysis (ICA) approach, a voxelwise measure of network membership (Joel et al., 2011), to identify the DMN FC and focused solely on this network to conserve statistical power. We also examined associations between DMN FC and post-concussive symptoms (self- and parent reports). Finally, we wanted to consider inattentive symptoms when studying DMN FC, because lower FC of the anterior DMN has previously been reported in youth with attention deficit hyperactivity disorder (ADHD; inattentive type) compared to controls (Qiu et al., 2011). Therefore, we present results with and without controlling for the severity of inattentive symptoms.
Method
Participants
Children were recruited through previous concussion research projects in the emergency department or a previous research in a sports injury prevention program for children with orthopedic injuries. Of the 171 potentially eligible participants contacted, 61 participants were confirmed eligible and enrolled. They were invited to undergo an MRI session of about 1 h. Of the 61 enrolled, four did not complete imaging (three participants declined participation for the scanning portion of the study and one had incomplete and unusable data due to excessive motion). No significant differences were found between children who did and did not complete neuroimaging on age, time since injury, number of concussions, or parent and self-reported symptoms (Brooks et al., 2019).
The final sample (n = 57) was divided into three groups: OI without prior concussions (n = 20), one concussion (n = 16), and multiple concussions (n = 21). To be eligible, participants in all groups had to be between 8 and 19 years old, and they and their parents had to understand and speak English to be able to complete questionnaires. Past hospitalization for a psychiatric condition; history of substance abuse or neurological disorder (e.g., seizures, hydrocephalus); visual, hearing, motor, or language deficits that prevented testing; and contraindications to MRI (e.g., braces, metal in body) were exclusion criteria for all three groups. Learning disabilities, attention problems, anxiety, or depression were not exclusion criteria because they are common in children of all groups.
For the two groups with concussions, participants were included if their last concussion was at least 6 months prior to enrolment, and if the concussion was diagnosed by a healthcare professional (Glasgow coma scale score between 13 and 15, duration of loss of consciousness less than 30 minutes, and/or duration of post-traumatic amnesia of less than 24 h at the time of injury; DeCuypere & Klimo, 2012), as confirmed by a family member. Participants in the OI group had no history of concussion but presented to a healthcare professional for management of an injury involving the thorax, upper extremity, or lower extremity.
Measures
MRI image acquisition. Neuroimaging was completed using a GE Discovery MR750w 3.0 T MR scanner (GE Healthcare, Milwaukee, WI) with a 32-channel head coil (MRI Instruments). The acquisition protocol included a whole-brain T1-weighted 3D anatomical image (time of repetition [TR] = 8.2 ms, time of echo [TE] = 3.2 ms, flip angle θ = 10°, field of view [FOV] = 240 mm, 230 slices, 0.8 mm thickness, scan time = 5.1 min) with 0.8 mm3 isotropic resolution. A rsfMRI sequence was run, during which participants were asked to fixate on a small black cross on a white screen and think of nothing in particular. Contiguous slices (64 3.6 mm-thick) were collected in this time frame (TE = 30 ms, TR = 2,000 ms, flip angle = 90°, FOV = 23.0 cm, matrix size = 64 × 64). The image was acquired sequentially top-down (prescribed parallel to the AC-PC line).
Clinical measures. To document pre-existing conditions in participants, parents answered the following questions (yes/no) on a questionnaire, based on their recall of their child pre-injuries: concerns about attention or attention disorder diagnosed; concerns about learning or learning disorder diagnosed; concerns about low mood, sadness, or depression; concerns about worrying and anxiety.
Parents filled out the ADHD Rating scale-5 for children and adolescents (DuPaul et al., 2016), using a 4-point Likert scale (0 = never or rarely—3 = very often) to rate 18 items evaluating inattention (nine symptoms) and hyperactivity/impulsivity (nine symptoms) over the past 6 months, based on criteria in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5; American Psychiatric Association, 2013). For the purpose of the study, separate inattention and hyperactivity/impulsivity percentile scores were reported.
Finally, parents and participants completed the post-concussion symptom inventory (PCSI: parent and self-report), a 26-item scale on which current post-concussive symptoms (based on yesterday and today) are rated on a 7-point Likert scale (0 = never – 6 = almost always), yielding a total symptom severity score ranging from 0 to 156. This questionnaire has adequate validity and reliability (Gioia et al., 2009; Sady et al., 2014).
Procedure
The study was conducted in a tertiary care children’s hospital (Alberta Children’s Hospital, Calgary, Alberta) and granted ethical clearance (REB13’-1199) by the University of Calgary Conjoint Health Research Ethics Board. Participants were recruited based on participation in previous research in the hospital’s emergency department or in a sports injury prevention program. Information about the number of participants contacted who refused to participate is provided elsewhere (Brooks et al., 2019). Research assistants explained the study to parents and their children at the beginning of the appointment and obtained informed consent and assent. No child was excluded based on an a priori criterion for motion of absolute maximum displacement > 3 mm for the imaging data. After the scan, participants completed a brief psychological assessment whereas parents completed questionnaires in a separate room. Gift certificates were offered to participants as compensation, and participants had their parking paid for at the hospital and received a brief summary of behavioral results.
Analyses
fMRI image pre-processing. Raw resting state images were pre-processed using the FMRIB Software Library (FSL 5.0.10; Woolrich et al., 2009). The pipeline in FSL FEAT (Smith et al., 2004) included minimal filtering (2,000 ms), MCFLIRT motion correction, bet brain extraction, interleaved slice timing correction, and spatial smoothing (4 mm Gaussian kernel full width at half maximum). Procedures to mitigate head motion were next performed and are described in more detail in the next section. Finally, each participant’s anatomical and functional images were co-registered to the standard space MNI152 T1 2mm brain template.
Head-motion mitigation procedure. We used a three-step process to address motion confounds in the data. First, we used motion estimates derived from the preprocessing in order to exclude participants with excessive head motion; scans were excluded if they exhibited > 3 mm maximum absolute displacement. Second, we used AROMA, an ICA-based cleaning method, which has recently been shown to be most effective in mitigating the impact of head motion, and allows for the retention of the remaining "true" neural signal within an affected volume. AROMA is an automated procedure that uses a small but robust set of theoretically motivated temporal and spatial features (timeseries and power spectrum) to distinguish between “real” neural signals and motion artifacts. We chose a conservative threshold (“aggressive”) in order to decrease the chance of false positives. Noise components identified by AROMA were used to clean the data. Third, images were de-noised by regressing out the six motion parameters, as well as signal from white matter, cerebral spinal fluid and the global signal, as well their first-order derivatives following recent recommendations (Parkes, Fulcher, Yücel, & Fornito, 2018). While no gold standard (Murphy & Fox, 2017) exists guiding the removal of the global signal, we chose to remove it based on recent evidence that it relates strongly to respiratory and other motion-induced signals, which persist through common denoising approaches including ICA and models that attempt to approximate respiratory variance (Power et al., 2018).
fMRI image analysis. The pre-processed and cleaned fMRI data were subjected to an ICA in FSL’s MELODIC, using a multivariate exploratory linear decomposition in a temporal concatenation approach. Thirty group-level ICA components were extracted, among which anterior and posterior DMN components were identified. ICA components were subsequently used to generate participant-specific versions of the spatial maps and associated timeseries using FSL’s dual regression approach (Filippini et al., 2009). Dual regression was used to identify, within each participant’s fMRI data, spatial maps and associated time courses corresponding to the extracted ICA components. Dual regression was chosen as an analytic approach because it allows the generation of individual-subject maps of network membership, which can then be contrasted between groups, and also because it has been shown to have high test–retest reliability relative to seed-based techniques (Chen et al., 2015; Zuo et al., 2010; Zuo & Xing, 2014). The dual-regression procedure was carried out as follows. First, for each participant, the average group spatial maps of the ICA components were regressed into the participant’s four-dimensional (4D; space/time) dataset simultaneously. This resulted in a set of participant-specific time series, one per network. Then, these timeseries were regressed into the same 4D dataset, resulting a set of participant-specific spatial maps, one per network. This provided pairs of estimates, which form a dual space and jointly best approximate the original group ICA maps.
We then tested for pairwise group differences (controlling for age and sex) using the respective DMN component as a mask in FSL’s Randomize (5,000 permutations; Winkler et al., 2014); thus in this study, we specifically examined intra-network FC. Threshold-free cluster enhancement (TFCE) (Smith & Nichols, 2009)—the default standard in Randomize—was used to estimate cluster activation with p < 0.05 corrected for multiple comparisons (2 ICNs). TFCE is a new method for finding clusters in the data without having to define clusters in a binary way, meaning it avoids the long-standing, problematic need to define an initial cluster-forming threshold (see e.g., Friston et al., 1996; Friston et al., 1994). TFCE can find smaller areas of difference than conventional thresholding, because it takes a raw statistic image and produces an output image in which the voxelwise values represent the amount of cluster-like local spatial support. In other words, cluster-like structures are enhanced but the image remains fundamentally voxelwise (Smith & Nichols, 2009).
We also tested for pairwise group differences (controlling for age, sex in a first time and for age, sex, and inattention symptoms in a second time), to control for group differences on inattention symptoms that could have affected the DMN FC. Finally, partial correlations (age, sex, and inattention as covariates) were computed between post-concussive symptoms and the participant’s average neural signal (e.g., average time course for DMN areas) for areas in which group differences were significant.
Clinical measures analysis. IBM Statistical Package for the Social Sciences, version 25.0 for Macintosh was used for statistical analyses. To control for multiple analyses, the p level was set at 0.05. PCSI scores (self- and parent-report) were square-root transformed because of their positively skewed distributions and used for all analyses.
Chi-square (χ2) likelihood ratio analyses were conducted for group differences of categorical data, using Phi (rφ) as an effect size indicator, interpreted as small (rφ = 0.1), medium (rφ = 0.3), or large (rφ = 0.5). Analyses of variance (ANOVAs) were used for group comparisons of continuous data, using partial eta squared (η2p) to measure effect size, interpreted as small (η2p = 0.01), medium (η2p = 0.06), or large (η2p = 0.14) (Richardson, 2011). Tukey’s honest significant difference (HSD) post-hoc tests were conducted if ANOVAs were significant, with effect sizes (Cohen’ s d) interpreted as small (d = 0.20), medium (d = 0.50), or large (d = 0.80) (Cohen, 1992).
Results
Demographics are reported in Table 1. The average number of concussions in the multiple concussions group was 3.2 (SD = 1.7, median = 3.0, range = 2–8). Chi-square tests revealed no significant differences between groups on sex and ethnicity (p > 0.05). The groups did not differ significantly on pre-existing low mood, sadness, and depression (reported in 13.0% of participants), worrying and anxiety (reported in 20.4% of participants), pre-existing attention concerns or diagnosis of attention-deficit/hyperactivity disorder (reported in 21.8% of participants). There were significant group differences on pre-injury learning difficulties or disorder (χ2(2) =9.15, p = 0.01, rφ = 0.34, medium effect size; 0% in the group with OI; 18.8% in the group with a single concussion; and 30% in the group with multiple concussions).
Sample characteristics
| Participants | Total | Groups | ||
|---|---|---|---|---|
| N | Orthopedic injury (OI) n | Single concussion n | Multiple concussion n | |
| Sex (% boys) | 52.6 | 50 | 50 | 57.1 |
| 57 | 20 | 16 | 21 | |
| Ethnicity (% Caucasian) | 86 | 85 | 75 | 95.2 |
| 57 | 20 | 16 | 21 | |
| Age M (SD), range | 14.7(2.8) 8.5–19.1 | 14.4(3.1) | 14.2(2.7) | 15.2(2.7) |
| 57 | 8.5–19.1 | 8.5–18.5 | 8.8–18.9 | |
| — | 20 | 16 | 21 | |
| Months from last injury to testing M (SD), range | 31.6(19.4) | 30.9(29.4) | 35.9(13.0) | 28.8(11.6) |
| 4.3–130.7 | 4.3–130.7 | 13.8–54.6 | 6.5–45.6 | |
| 55 | 18 | 16 | 21 |
Notes: M = mean; SD = standard deviation.
The three groups did not differ significantly on age at time of injury, time since last injury, PCSI self-report total score, and hyperactivity/impulsivity score (p > 0.05). However, the groups did differ significantly on PCSI parent-report total score, F(2,53) = 5.05, p = 0.01, and on inattention symptoms of the ADHD Rating scale-5, F(2,53) = 5.21, p = 0.01. Tukey’s HSD post-hoc analyses showed that PCSI parent-report (p = 0.01; d = 1.05, large effect size) and inattention symptoms (p = 0.01; d = 1.09, large effect size) scores were significantly higher in the multiple concussion group compared to the OI group (see Table 2).
Descriptive statistics on rating scales
| Participants | Total | Groups | ||
|---|---|---|---|---|
| N | Orthopedic injury (OI) n | Single concussion n | Multiple concussion n | |
| Post-concussion symptoma | 15.77 (23.42) | 6.53 (9.58) | 18.56 (22.48) | 22.00 (30.26) |
| Inventory–parent M (SD) | 56 | 19 | 16 | 21 |
| Post-concussion symptoma | 24.63 (25.87) | 15.85 (17.32) | 22.44 (24.48) | 34.67 (30.87) |
| Inventory–self M (SD) | 57 | 20 | 16 | 21 |
| Inattention symptoms– | 47.81 (31.53) | 32.11 (26.58) | 47.69 (33.19) | 62.12 (28.69) |
| ADHD rating scale-5 M (SD) | 56 | 19 | 16 | 21 |
| Hyperactivity/impulsivity symptoms–ADHD rating | 39.49 (27.12) | 31.84 (23.69) | 41.28 (31.93) | 45.05 (25.75) |
| Scale-5 M (SD) | 56 | 19 | 16 | 21 |
Notes. M = Mean; SD = Standard deviation.
Post-concussion symptom inventory scores are raw scores where higher values suggest more or worse symptoms. Raw scores are reported in the Tables but analyses were computed using square root transformed (SQRT) scores. ADHD Rating Scale 5 scores are percentiles (age and sex adjusted) where higher percentiles suggest more or worse symptoms.
FC in the anterior DMN was significantly lower in the group with multiple concussions compared to the group with one concussion and the OI group when controlling for age and sex only (Table 3). In contrast, the posterior DMN FC showed no significant group differences. The results remained largely unchanged after controlling for age, sex, and inattention symptoms, except that 1 out of 4 clusters of the anterior DMN was no longer significant (Table 4). Figure 1 (parts A to C) therefore present DMN FC differences between groups when controlling for age, sex, and inattention symptoms. Supplementary analyses (results not reported) compared the three groups on the differences between the anterior and posterior DMN average time courses, obtaining no significant differences.
Significant differences between groups in DMN FC when controlling for age and sex
| Contrasts | Network | Cluster # | Cluster size (voxels) | Z max | x | y | z |
|---|---|---|---|---|---|---|---|
| Contrast 1 (g1 control > g3 multiple) | Anterior DMN | 2 | 6 | 4.53 | 6 | 52 | 2 |
| 1 | 3 | 4.22 | –2 | 48 | 2 | ||
| Contrast 2 (g2 single > g3 multiple) | Anterior DMN | 2 | 10 | 3.84 | 4 | 54 | 24 |
Notes: DMN = default mode network; g1 control = group with OI; g2 single = group with single concussion; g3 multiple = group with multiple concussions. Z max = Z statistic of local maxima of the estimated difference between groups. Coordinates are in MNI space and indicate location of local maxima.
Significant differences between groups in DMN FC when controlling for age, sex, and inattention symptoms
| Contrasts | Network | Cluster # | Cluster size (voxels) | Z max | x | y | z |
|---|---|---|---|---|---|---|---|
| Contrast 1 (g1 control > g3 multiple) | Anterior DMN | 1 | 20 | 4.3 | 0 | 48 | 2 |
| Contrast 2 (g2 single > g3 multiple) | Anterior DMN | 2 | 63 | 3.8 | 4 | 54 | 24 |
| 1 | 57 | 3.9 | 6 | 48 | 40 |
Notes: DMN = default mode network; g1 control = group with OI; g2 single = group with single concussion; g3 multiple = group with multiple concussions. Z max = Z statistic of local maxima of the estimated difference between groups. Coordinates are in MNI space and indicate location of local maxima.
Significant differences between groups in DMN FC when controlling for age, sex, and inattention symptoms. The yellow underlay depicts the anterior DMN mask. Part A: Lower anterior DMN (contrast 1, cluster 1), group with OI > group with multiple concussion. Part B: Upper anterior DMN (contrast 2, cluster 1), indicated with a red arrow, group with single concussion > group with multiple concussion. Part C: Upper anterior DMN (contrast 2, cluster 1), indicated with a red arrow, group with single concussion > group with multiple concussion.
Finally, partial correlations between the anterior DMN average time course and PCSI scores across the sample (parent report: r = 0.06; self-report: r = −0.02) were not significant. Separate partial correlations for the OI group (parent report: r = 0.26, small effect size; self-report: r = −0.17), the single concussion group (parent report: r = 0.45, small effect size; self-report: r = −0.16), and the multiple concussions group (parent report: r = −0.27, small effect size; self-report: r = 0.10) were also not significant (p > 0.05).
Discussion
The present study investigated DMN FC in children and adolescents long after sustaining one or multiple concussions, or an OI. The first hypothesis was partially supported. On average, 32 months post-injury, the group with multiple concussions demonstrated significantly lower FC in the anterior DMN compared to the other two groups, but no significant FC differences in the posterior DMN. In addition, no significant correlations were found between anterior DMN FC and post-concussive symptoms for all groups.
The finding of significantly lower connectivity in the anterior DMN replicates some of the results previously obtained in youth (Borich et al., 2015; Murdaugh et al., 2018; Orr et al., 2016) and adults (Mayer et al., 2011). Our results suggest that altered brain connectivity of the DMN may still be present a long time after multiple concussions in children and adolescents. This could reflect an undergoing cerebral recovery process from the injuries (Borich et al., 2015), providing preliminary evidence for a potential dose-response relationship between the number of concussions in children and adolescents and long-term DMN FC.
These results could, however, also reflect other variables pre-injury or post-injury that could have differed between the group with multiple concussions and the other groups and could have affected the DMN FC. For instance, these include task-switching or executive processes, variables in which the DMN is involved (Buckner & Carroll, 2006) but not measured in the current study. Interestingly, the FC results were largely unchanged when controlling for age, sex, and severity of inattention symptoms. This is a relevant finding, because (1) previous studies of post-concussion FC have primarily focused on adolescent men and (2) previous studies of youth with ADHD (inattentive type) have also reported hypoconnectivity of the anterior DMN when compared to controls (Qiu et al., 2011). Therefore, the present results suggest that reduced FC in the anterior DMN is not accounted for by age, sex, or the more severe symptoms of inattention reported in the group with multiple concussions.
In contrast, the failure to detect group differences in the posterior DMN differs from previous studies in youth showing hyperconnectivity in comparison with controls, from 7 days to 2 months (Borich et al., 2015; Murdaugh et al., 2018), and up to 40 months post-injury (Orr et al., 2016). Our results do not support previous studies that have suggested that increased connectivity in the posterior DMN represents reallocation or compensatory mechanisms to the reduced connectivity obtained in the anterior DMN (Borich et al., 2015; Murdaugh et al., 2018; Orr et al., 2016).
Finally, the present study documented hypoconnectivity of the anterior DMN in the group with multiple concussions, as well as higher levels of post-concussive symptoms reported by parents in that group compared to the group with OI. This converges with prior research showing DMN FC differences in patients who were still symptomatic after concussion compared to controls (Borich et al., 2015). However, the higher levels of post-concussive symptoms by parents compared to controls might also reflect the very low mean score in the control group, as much as elevated symptoms in the multiple concussion group (see raw scores in Table 2).
Moreover, the current study reported no significant differences between the three groups in levels of self-reported post-concussive symptoms and no significant relationships between the anterior DMN FC and post-concussive symptoms (self-and parent ratings) for all groups. These results align with previous studies showing DMN FC differences compared to controls in participants asymptomatic from the concussion (Manning et al., 2017; Orr et al., 2016). This suggests that in some cases, neurobiological differences may persist despite clinical recovery. The lack of significant relationships between anterior DMN FC and post-concussive symptoms might also reflect reduced statistical power when conducting correlations separately for the three groups, which had 16–21 participants.
In addition to the small sample size and limited statistical power, we acknowledge other limitations in the present study. First, participants agreed to volunteer to this study long after sustaining their injury and we did not obtain information related to the mechanisms/causes of injury neither on puberty, which might have affected the imaging results and which limit the representativeness of the sample. The study design was cross sectional, so that no baseline data were available and all pre-injury data were collected in a retrospective manner (including the number of concussions reported by parents), raising the possibility of recall bias.
Future studies with larger number of participants that compare OI, single concussion, and multiple concussions are warranted to increase statistical power and replicate the dose-response association observed regarding the lower FC of the anterior DMN a long time after multiple concussions. These next neuroimaging studies investigating the potential long-term impact of multiple concussions should follow children longitudinally and control for baseline differences that might affect the DMN FC. A multi-methods and multi-respondents approach is also recommended to assess post-concussive recovery outcomes and adaptive functioning.
Finally, there is a need to better elucidate the neural and psychological mechanisms underlying this hypoconnectivity in youth with multiple concussions. This could be achieved by using other imaging techniques at-rest, such as magnetic resonance spectroscopy, and by combining multiple imaging techniques, as well as by investigating the cognitive processes happening during the rsfMRI sequence, such as executive functioning, the level of alertness, self-reflection, or the frequency and content of mind wandering (Raichle, 2015; Whitfield-Gabrieli & Ford, 2012).
Conclusion
In summary, our study demonstrated reduced FC in the anterior DMN in youth with multiple concussions when compared to youth with a single concussion and youth with orthopedic injuries. This reduced FC, however, was not significantly correlated to higher levels of post-concussive symptoms. Future longitudinal and population-based studies are necessary to help us better appreciate whether the hypoconnectivity of the anterior DMN is stable over time, persists despite clinical recovery, and correlates with daily functioning in youth who present repeated concussions.
Acknowledgements
The authors thank the members of the “NEURO-detect” study team (Karen M. Barlow MB.ChB. MRCPCH (UK), MSc, Helen Carlson PhD, Michael Esser MD PhD, Zeanna Jadavji, BSc, Catherine Lebel PhD, Ashley Harris PhD, Marc Lebel PhD, Frank P. MacMaster PhD, Kathryn Schneider PhD PT, and Trevor Low BSc). Thanks to Lonna Mitchell BA and Kalina Slepicka BA for scoring protocols and (alphabetically) Amy Bobyn, Dominique Bonneville BA (Hons), Shauna Bulman, Christianne Laliberté-Durish MSc, Shelby MacPhail BSc, Maya Sohn, and Cole Sugden BSc, for assistance with data entry/checking. Thanks to Brenda Turley BA (Hons), Carolyn Emery PhD PT, and Kathryn Schneider PhD PT for assisting with recruitment. Recruitment of some orthopedic control participants was done using the Healthy Infants and Children Clinical Research Program database (HICCUP; http://www.ucalgary.ca/paediatrics_hiccup). Thanks to Elodie Boudes MSc, Aneesh Khetani MSc, and the Child and Adolescent Imaging Research (CAIR) technicians for assistance with the MRI. Thank you to the families who participated and generously donated their time to research.
Funding
This study was funded by the Shaikh Family Research Award, an endowment held by the Alberta Children’s Hospital Foundation and granted to Brian Brooks by the Alberta Children’s Hospital Research Institute. Additional support was provided from the Ronald and Irene Ward Chair in Pediatric Brain Injury (awarded to Keith Yeates). Brian Brooks acknowledges salary funding from the Canadian Institutes for Health Research (CIHR) Embedded Clinician Researcher Salary Award. Vickie Plourde acknowledges fellowship funding from the Alberta Children’s Hospital Research Institute (University of Calgary), the Integrated Concussion Research Program (University of Calgary), and Alberta Innovates-Health Solutions. Keith Yeates acknowledges funding from the Ronald and Irene Ward Chair in Pediatric Brain Injury.
Conflict of interest
Brian Brooks receives royalties for the sales of the Pediatric Forensic Neuropsychology textbook (2012, Oxford University Press) and three pediatric neuropsychological tests [Child and Adolescent Memory Profile (ChAMP, Sherman and Brooks, 2015, PAR Inc.), Memory Validity Profile (MVP, Sherman and Brooks, 2015, PAR Inc.), and Multidimensional Everyday Memory Ratings for Youth (MEMRY, Sherman and Brooks, 2017, PAR Inc.)]. He has previously received in-kind support (free test credits) from the publisher of the computerized cognitive test (CNS Vital Signs, Chapel Hill, North Carolina) used in this study. Keith Yeates receives royalties for book sales from Guilford Press and Cambridge University Press, and occasionally serves as a paid expert in forensic cases. None of the authors have a financial interest in any measures used in the present study.
References
Author notes
Vickie Plourde and Christiane S Rohr are Equal co-first authors.
