Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 May 17:11:e70450.
doi: 10.7554/eLife.70450.

Multi-tract multi-symptom relationships in pediatric concussion

Guido I Guberman  1 , Sonja Stojanovski  2   3 , Eman Nishat  2   3 , Alain Ptito  1 , Danilo Bzdok  4   5   6 , Anne L Wheeler  2   3 , Maxime Descoteaux  7   8
Affiliations

Multi-tract multi-symptom relationships in pediatric concussion

Guido I Guberman et al. Elife. .

Abstract

Background: The heterogeneity of white matter damage and symptoms in concussion has been identified as a major obstacle to therapeutic innovation. In contrast, most diffusion MRI (dMRI) studies on concussion have traditionally relied on group-comparison approaches that average out heterogeneity. To leverage, rather than average out, concussion heterogeneity, we combined dMRI and multivariate statistics to characterize multi-tract multi-symptom relationships.

Methods: Using cross-sectional data from 306 previously concussed children aged 9-10 from the Adolescent Brain Cognitive Development Study, we built connectomes weighted by classical and emerging diffusion measures. These measures were combined into two informative indices, the first representing microstructural complexity, the second representing axonal density. We deployed pattern-learning algorithms to jointly decompose these connectivity features and 19 symptom measures.

Results: Early multi-tract multi-symptom pairs explained the most covariance and represented broad symptom categories, such as a general problems pair, or a pair representing all cognitive symptoms, and implicated more distributed networks of white matter tracts. Further pairs represented more specific symptom combinations, such as a pair representing attention problems exclusively, and were associated with more localized white matter abnormalities. Symptom representation was not systematically related to tract representation across pairs. Sleep problems were implicated across most pairs, but were related to different connections across these pairs. Expression of multi-tract features was not driven by sociodemographic and injury-related variables, as well as by clinical subgroups defined by the presence of ADHD. Analyses performed on a replication dataset showed consistent results.

Conclusions: Using a double-multivariate approach, we identified clinically-informative, cross-demographic multi-tract multi-symptom relationships. These results suggest that rather than clear one-to-one symptom-connectivity disturbances, concussions may be characterized by subtypes of symptom/connectivity relationships. The symptom/connectivity relationships identified in multi-tract multi-symptom pairs were not apparent in single-tract/single-symptom analyses. Future studies aiming to better understand connectivity/symptom relationships should take into account multi-tract multi-symptom heterogeneity.

Funding: Financial support for this work came from a Vanier Canada Graduate Scholarship from the Canadian Institutes of Health Research (G.I.G.), an Ontario Graduate Scholarship (S.S.), a Restracomp Research Fellowship provided by the Hospital for Sick Children (S.S.), an Institutional Research Chair in Neuroinformatics (M.D.), as well as a Natural Sciences and Engineering Research Council CREATE grant (M.D.).

Keywords: diffusion MRI; human; medicine; multivariate statistics; neuroscience; pediatric concussions.

Plain language summary

Concussions can damage networks of connections in the brain. Scientists have spent decades and millions of dollars studying concussions and potential treatments. Yet, no new treatments are available or in the pipeline. A major reason for this stagnation is that no two concussions are exactly alike. People affected by concussions may have different genetic or socioeconomic backgrounds. The nature of the injury or how its effects change over time may also vary among people with concussions. One central question facing scientists is whether there are multiple types of concussions. If so, what distinguishes them and what characteristics do they share. Some studies have looked at differences among subgroups of patients with concussions. But questions remain about whether – beyond differences between the patients – the brain injury itself differs and what impact that has on symptoms or patient trajectory. To better characterize different types of concussion, Guberman et al. analyzed diffusion magnetic resonance imaging scans from 306 nine or ten-year-old children with a previous concussion. The children were participants in the Adolescent Brain Cognitive Development Study. Using specialized statistical techniques, the researchers outlined subgroups of concussions in terms of connections and symptoms and studied how many of these subgroups each patient had. Some types of injury were linked with a category of symptoms like cognitive, mood, or physical symptoms. Some types of damage were linked with specific symptoms. Guberman et al. also found that one symptom, sleep problems, was part of many different injury subtypes. Sleep problems may occur in different patients for different reasons. For example, one patient with sleep difficulties may have experienced damage in brain regions controlling sleep and wakefulness. Another person with sleep problems may have injured parts of the brain responsible for mood and may have depression, which causes excessive sleepiness and difficulties waking up. Guberman et al. suggest a new way of thinking about concussions. If more studies confirm these concussion subgroups, scientists might use them to explore which types of therapies might be beneficial for patients with specific subgroups. Developing subgroup-targeted treatments may help scientists overcome the challenges of trying to develop therapies that work across a range of injuries. Similar disease subgrouping strategies may also help researchers study other brain diseases that may vary from patient to patient.

PubMed Disclaimer

Conflict of interest statement

GG, SS, EN, AP, DB, AW No competing interests declared, MD works as Chief Scientific Officer for IMEKA. He holds the following patents: DETERMINATION OF WHITE-MATTER NEURODEGENERATIVE DISEASE BIOMARKERS (Patent Application No.: 63/222,914), PROCESSING OF TRACTOGRAPHY RESULTS USING AN AUTOENCODER (Patent Application No.: 17/337,413)

Figures

Figure 1.
Figure 1.. Flowchart describing the participant selection procedure.
Figure 2.
Figure 2.. Illustration of the study’s post-processing pipeline.
(A). We applied the DKT parcellation onto each tractogram, thus building a binary connectivity matrix that displayed for all 306 subjects in the full dataset (rows), whether (black) or not (white) a streamline existed between each pair of labels (columns). (B). We thresholded connectomes using the full dataset, only keeping connections that existed across 90% of participants (a threshold of 100% is illustrated here for simplicity). On these connections, we also filtered streamlines by computing COMMIT weights. This technique assigns weights to streamlines depending on how well they explain the diffusion signal. We identified connections as spurious if all their streamlines had a COMMIT weight of 0. We only retained connections that were found to be non-spurious across 90% of participants in the full dataset. We then split the dataset into a discovery set (n = 214) and a replication set (n = 92). Using the discovery set, we then constructed connectomes of 6 scalar diffusion measures (Fractional Anisotropy (FA), Axial Diffusivity (AD), Mean Diffusivity (MD), Radial Diffusivity (RD), Apparent Fiber Density along fixels (AFDf), and Number of Fiber Orientations (NuFO)), by computing the average measure across each connection. (C). We stacked all columns from each connectivity matrix, creating vectors of every pair of subject and connection, and then joined together these vectors. We then performed principal component analysis (PCA) on these matrices. Principal component (PC) scores were calculated for each subject/connection combination, thus reconstructing connectomes weighted by PC scores. (D). From each these new connectomes, we selected 200 connections based on Pearson correlations with symptom-oriented measures. We then performed partial least squares correlation on each of these PC-weighted features and symptom measures, which allowed us to obtain pairs of multi-tract connectivity features (‘MCF’) and multi-symptom features (‘MSF’). Each multivariate feature is composed of linear combinations (weighted sums, illustrated by the black arrows called ‘weights’) of variables from its corresponding feature set.
Figure 3.
Figure 3.. Illustration of multi-tract multi-symptom pairs 1 and 7 obtained from the microstructural complexity PLSc (A and C respectively), and pair 3 from the axonal density PLSc (B).
Left: Polar plots displaying the weights of all 19 symptom measures for each multi-symptom feature. Bars pointing away from the center illustrate positive weights, bars pointing towards the center represent negative weights. White stars illustrate symptoms that significantly contributed to the pair. Bar graphs underneath the polar plots illustrate the % covariance explained by each pair, with the currently-shown pair highlighted. Right: Scatter plots showing the expression of multi-tract features (x-axis) and multi-symptom features (y-axis). In each scatter plot, the same 6 participants are labeled (1 through 6). Small bar graphs illustrate the scaled symptom measures (i.e.: not the expression of multi-symptom features) for two participants, one expressing low levels of a pair, the other expressing high levels. For each illustrated participant, positive bars illustrate symptoms that are higher than the sample average, negative bars represent symptoms that are lower. The black dashed line illustrates 1 standard deviation above the group mean. Participants with ADHD diagnoses are illustrated in black. Correlation coefficients inset in each scatter plot represent Pearson correlations between expression of multi-tract features (near x-axis), or multi-symptom features (near y-axis) and a binary variable indexing whether or not a participant had a diagnosis of ADHD.
Figure 4.
Figure 4.. Line plot showing the percent overlap between univariate analyses and each multi-tract connectivity feature.
Highest overlap occurred for the first multi-tract connectivity feature from both PLSc analyses. Brain renderings shown above graph illustrate which connections were found to be significant for univariate comparisons of microstructural complexity (red), univariate comparisons of axonal density (blue), multi-tract connectivity feature 1 from the microstructural complexity PLSc (violet), multi-tract connectivity feature 3 from the axonal density PLSc (turquoise), and multi-tract connectivity feature 7 from the microstructural complexity PLSc (green). The percent overlap score for each of the three illustrated multi-tract connectivity features are identified in the line plot with a circle of the corresponding color. Univariate brain graphs show connections significant at p < 0.01 for illustrative purposes. Multivariate brain graphs show connections significant at p < 0.05. Brain renderings were visualized with the BrainNet Viewer (Xia et al., 2013).
Figure 5.
Figure 5.. Patterns of connection/symptom correspondences across multi-tract multi-symptom pairs.
(A). Adjacency matrix, illustrating the number of multi-tract multi-symptom pairs from the microstructural complexity PLSc where a given significant connection corresponded to a given significant symptom (based on bootstrap analyses). Darker colors illustrate more consistent correspondences. Symptom categories are illustrated in colors (green: somatic, black: sleep problems, purple: mood problems, yellow: cognitive problems). Orange rectangles highlight three connections (right lateral occipital – right precuneus; left putamen – left rostral anterior cingulate; left putamen – left lingual) that were only present in one multi-tract multi-symptom pair (pair 3), which also represented broadly all cognitive problems. This pair is illustrated in B, where cognitive problems are illustrated in color and all other symptoms are illustrated in black, and the highlighted connections are circled in orange. Although only three connections are highlighted, several such ‘broad cognitive problems’ connections can be observed. The blue rectangle highlights two connections (right pars opercularis – right post-central; right pars opercularis – right supramarginal) that were present in 4 multi-tract multi-symptom pairs, all of which also implicated attention problems. These pairs are illustrated in panel C, where attention problems are illustrated in color, all other symptoms are illustrated in black, and the highlighted connections are circled in blue.
Appendix 1—figure 1.
Appendix 1—figure 1.. Illustration of the effects of regressing out scanner.
(A) Weights of each diffusion measure for each principal component obtained after running a principal component analysis on data that was processed without regressing out scanner. (B) Barplot illustrating the expression of multi-tract connectivity feature 1 averaged across all participants for each scanner. The blue bar illustrates the scanner with the lowest multi-tract connectivity feature 1 expression, and the red bar illustrates the scanner with the second highest multi-tract connectivity feature 1 expression (the scanner with the highest expression only had one participant, so it was not chosen for illustrative purposes). (C) Scatter plot illustrating expression of multi-tract multi-symptom pair 1 from the microstructural complexity PLSc using data that was processed without regressing out scanner. The blue dots illustrate participants from the scanner with the lowest average multi-tract connectivity feature 1 expression, the red dots illustrate participants from the scanner with the second-highest feature 1 expression. These two groups are distinguishable in their multi-tract connectivity feature 1 expression. (D) Scatter plot illustrating expression of multi-tract multi-symptom pair 1 from the microstructural complexity PLSc using data where scanner had been regressed out. The same participants identified in scatter plot C are illustrated in scatter plot D. After regressing out scanner, these two groups are not distinguishable in their multi-tract connectivity feature 1 expression.
Appendix 1—figure 2.
Appendix 1—figure 2.. Scatter plots illustrating the expression of three multi-tract multi-symptom pairs (first column: pair 1 from the microstructural complexity PLSc, second column: pair 3 from the axonal density PLSc, third column: pair 7 from the microstructural complexity PLSc), color-coded by total family income (first row), race/ethnicity (second row), and sex (third row).
The upper and bottom rows illustrate the correlation coefficient for the expression of multi-tract features (over x-axis) and multi-symptom features (over y-axis) and variables representing family income (top), and sex (bottom). For race/ethnicity, correlations were performed between multivariate feature expression and dummy-coded variables representing each specific race/ethnicity. The correlation coefficients are presented in the bar graphs following the same order as listed in the color code.
Appendix 1—figure 3.
Appendix 1—figure 3.. Plots illustrating the weights of each diffusion measure for each principal component for different connectome thresholds (85%, 90%, 95%, 100%).
The interpretation of the first two principal components are consistent across thresholds.
Appendix 1—figure 4.
Appendix 1—figure 4.. Polar plots illustrating the weights of each symptom measure for every retained multi-symptom feature obtained from the microstructural complexity PLSc performed using all 19 symptom measures as well as connectivity features selected from connectomes thresholded at T = 90%.
Black stars indicate symptoms that significantly contributed to the multi-tract multi-symptom pair based on bootstrapping analyses.
Appendix 1—figure 5.
Appendix 1—figure 5.. Brain graphs illustrating the connections that were found to be significant (P < 0.05 based on bootstrap analysis) for each of the retained multi-tract features from the microstructural complexity PLSc.
Brain renderings were visualized with the BrainNet Viewer (Xia et al., 2013).
Appendix 1—figure 6.
Appendix 1—figure 6.. Bar graph illustrating the expression of multi-tract connectivity feature 2 from the microstructural complexity PLSc, averaged according to subgroups of participants defined by Injury Mechanism.
1: Fall/hit by object; 2: Fight/shaken; 3: Motor vehicle collision; 4: Multiple; 5: Unknown.
Appendix 1—figure 7.
Appendix 1—figure 7.. Bar graph illustrating the expression of multi-tract connectivity feature 15 from the microstructural complexity PLSc, averaged according to subgroups of participants defined by Injury Mechanism.
1: Fall/hit by object; 2: Fight/shaken; 3: Motor vehicle collision; 4: Multiple; 5: Unknown.
Appendix 1—figure 8.
Appendix 1—figure 8.. Bar graph illustrating the expression of multi-tract connectivity feature 15 from the microstructural complexity PLSc, averaged according to subgroups of participants defined by Total TBIs.
0: Unknown. Other numbers represent the total number of TBIs.
Appendix 1—figure 9.
Appendix 1—figure 9.. Plot illustrating the weights of each diffusion measure for each principal component for the PCA performed on the replication set data using a threshold of 90% during processing.
Appendix 1—figure 10.
Appendix 1—figure 10.. Polar plots illustrating the weights of each symptom measure for every retained multi-symptom feature obtained from the microstructural complexity PLSc performed using all 19 symptom measures as well as connectivity features selected from connectomes thresholded at T = 90% using the replication dataset.
Black stars indicate symptoms that significantly contributed to the multi-tract multi-symptom pair based on bootstrapping analyses.
Appendix 1—figure 11.
Appendix 1—figure 11.. Percentage of covariance explained by the first multi-tract multi-symptom pair from the microstructural complexity PLSc as a function of the number of features selected from the univariate feature selection step.
The connectivity features are selected based on decreasing strength of correlation with any symptom. The number of features tested ranged from 19 to 214.
Appendix 1—figure 12.
Appendix 1—figure 12.. Polar plots illustrating the weights of each symptom measure for multi-symptom features that were obtained from the microstructural complexity PLSc performed using all 19 symptom measures as well as connectivity features selected from connectomes thresholded at T = 100%.
Only the multi-symptom features that were found to be significant in the corresponding PLSc performed at a threshold of T = 90% are shown here, for comparison with those multi-symptom features (Appendix 1—figure 4). All Black stars indicate symptoms that significantly contributed to the multi-tract multi-symptom pair based on bootstrapping analyses.
Appendix 1—figure 13.
Appendix 1—figure 13.. Matrix illustrating correlation coefficients between the expression of every pair of multi-tract connectivity features obtained from the microstructural complexity PLSc.
The matrix illustrates the correlation between features obtained from the PLSc analysis performed on connectivity features obtained from the 90% and 100% thresholds. Given that this matrix is symmetrical, only the bottom triangular is shown. The main diagonals illustrate autocorrelations. These matrices illustrate how corresponding multi-tract connectivity features between thresholds (e.g.: multi-tract connectivity feature 1 from T = 90%, multi-tract connectivity feature 1 from T = 100%) are highly correlated.
Author response image 1.
Author response image 1.
See this image and copyright information in PMC

References

    1. Anderson ED, Giudice JS, Wu T, Panzer MB, Meaney DF. Predicting Concussion Outcome by Integrating Finite Element Modeling and Network Analysis. Frontiers in Bioengineering and Biotechnology. 2020;8:309. doi: 10.3389/fbioe.2020.00309. - DOI - PMC - PubMed
    1. Armstrong RC, Mierzwa AJ, Marion CM, Sullivan GM. White matter involvement after TBI: Clues to axon and myelin repair capacity. Experimental Neurology. 2016;275 Pt 3:328–333. doi: 10.1016/j.expneurol.2015.02.011. - DOI - PubMed
    1. Beaton D, Chin Fatt CR, Abdi H. An ExPosition of multivariate analysis with the singular value decomposition in R. Computational Statistics & Data Analysis. 2014;72:176–189. doi: 10.1016/j.csda.2013.11.006. - DOI
    1. Bigler ED, Maxwell WL. Neuropathology of mild traumatic brain injury: relationship to neuroimaging findings. Brain Imaging and Behavior. 2012;6:108–136. doi: 10.1007/s11682-011-9145-0. - DOI - PubMed
    1. Bodien YG, McCrea M, Dikmen S, Temkin N, Boase K, Machamer J, Taylor SR, Sherer M, Levin H, Kramer JH, Corrigan JD, McAllister TW, Whyte J, Manley GT, Giacino JT, TRACK-TBI Investigators Optimizing Outcome Assessment in Multicenter TBI Trials: Perspectives From TRACK-TBI and the TBI Endpoints Development Initiative. The Journal of Head Trauma Rehabilitation. 2018;33:147–157. doi: 10.1097/HTR.0000000000000367. - DOI - PMC - PubMed

Publication types