*3.2. General White Matter Changes/Injuries*

Differences in DTI-based parameters were observed in 29 white matter pathways in TBI patients compared to HCs (see Table 2). No rSRC athlete displayed significantly increased AD or RD or significantly decreased FA for any of the included white matter pathways (Table 2). On the other hand, six white matter pathways displayed significant difference in DTI parameters in rSRC compared with HC. The six structures displayed a completely different trend in DTI parameters with a higher FA and lower AD and RD compared with HC (Table 2). The median and interquartile range for all white matter pathways and DTI-based parameters are presented in electronic Supplementary Material Table S1 (ESM1).

**Table 2.** Analysis of global white matter damage.



**Table 2.** *Cont*.

Adjusted *p*-values are presented for each comparison between traumatic brain injury (TBI), healthy controls (HC), and repeated sport-related concussions (rSRC) and HC including, all DTI-based metrics and white matter pathways. Values in red emphasize results wherein the axial diffusivity (AD) or radial diffusivity (RD) values are increased in comparison to HC, defining the criteria for white matter pathway injury. Values in green emphasize results wherein AD or RD decreases, which is contradictory to the theory presented. Y: yes (injured), N: not injured, L: left, R: right; AC: anterior commissure; AF: arcuate fasciculus; FAT: frontal aslant tract; ATR: anterior thalamic radiation; Ci: cingulum; CS: cortico-spinal tract; FM: forceps major; Fo: fornix; CC: corpus callosum; hSLF: horizontal component of superior longitudinal fasciculus; IFOF: inferior fronto-occipital fasciculus; ILF: inferior longitudinal fasciculus; MLF: middle longitudinal fasciculus; OR: optic radiation; UF: uncinate fasciculus; VO: vertical occipital fasciculus; vSLF: vertical component of superior longitudinal fasciculus.

#### *3.3. Focal White Matter Changes/Injuries*

Three white matter structures displayed regional differences in DTI-based parameters in both groups compared to HCs. The left frontal aslant tract (FAT) displayed lower FA and higher RD in the fronto-opercular region in TBI patients. The same pathway displayed higher FA and lower RD in the supplementary motor area region in rSRC patients. The right anterior thalamic radiation (ATR) displayed lower FA and higher AD and RD in TBI patients localized mostly at the thalamic level. The rSRC group displayed higher FA and lower AD and RD mostly at the thalamic level. The genu of the corpus callosum (CC)

showed lower FA, higher AD and RD in the TBI group close to the midline. In the rSRC group, the focal analysis showed higher AD and lower RD close to the midline (Figure 1).

**Figure 1.** Focal analysis of three white matter pathways with white matter changes identified in both TBI patients and in rSRC patients compared with HC. False discovery rate (FDR) analysis was performed, as such significant differences are displayed as continuous lines on the x axis of each DTIbased parameter for each white matter pathway. The left frontal aslant tract (FAT) displayed lower FA and higher RD in the fronto-opercular region in TBI patients, while a higher FA and lower RD in the supplementary motor area region were found in the rSRC group. The right anterior thalamic radiation (ATR) displayed lower FA and higher AD and RD in the TBI group localized mostly at the thalamic level, while the rSRC group displayed higher FA and lower AD and RD mostly at the thalamic level. The genu of the corpus callosum (CC) showed lower FA and higher AD and RD in the TBI group close to the midline, while the rSRC group displayed a higher AD and lower RD close to the midline. TBI: traumatic brain injuries; rSRC: repeated sport related concussions; HC: healthy controls; AD: axial diffusivity; RD: radial diffusivity; FAT: frontal aslant tract; ATR: anterior thalamic radiation; CC: corpus callosum.

#### *3.4. Neuropsychology*

Both TBI patients and rSRC athletes were impaired on the RBANS global outcome analysis compared with HC (TBI: 75 ± 24; rSRC: 80 ± 17; HC: 105.5 ± 2; TBI to HC: *p* = 0.03 rSRC to HC: *p* = 0.006). RBANS Memory scores were lower in both TBI, and SRC when compared to HC (*p* = 0.048 and *p* = 0.04; respectively). The RBANS Verbal scores were lower in rSRC athletes compared to HC (*p* = 0.048) although not in TBI (*p* = 0.07). No significative difference was detected among the groups for the other RBANS functional domains (Figure 2).

**Figure 2.** Each domain of RBANS is shown for TBI, rSRC, and HC. Significant differences in RBANS Memory score were detected between HC and TBI patients, as well as rSRC athletes. For the other domains of the RBANS, there were no significant differences among the groups. TBI: traumatic brain injuries; rSRC: repeated sport-related concussions; HC: healthy controls; RBANS: repeatable battery for the assessment of neuropsychological status. \* Statistical significance with *p* value < 0.05.

#### **4. Discussion**

The most important finding in our study was that focal regional differences in white matter pathways were observed at the chronic stage post-injury in both TBI patients and rSRC athletes who had a similar impairment on memory testing in comparison to healthy, age-matched controls (HC). For the development of novel treatments, and for precision rehabilitation, an enhanced understanding of the structural basis of persistent symptoms is crucial [41].

White matter changes, demonstrated by differences in DTI-based parameters, have been previously identified in TBI patients [19,21,25,28,30]. Furthermore, differences in DTI-based parameters are also seen in patients with only minor cognitive impairment and in patients with normal conventional MRI, supporting the role of DTI in detecting subtle injuries missed by other modalities [8,20,25,30,42]. In our study, TBI patients displayed significant changes in DTI-based parameters compared with HC in 29 of the 37 analyzed white matter structures in the global white matter analysis. Using the same analysis interpreting the mean values for each single white matter in rSRC athletes, no white matter pathways displayed global axonal or myelin abnormalities compared with HC. However, a different trend in DTI parameters compared with HC was observed in six white matter pathways compared with HC. It is suggested that different injury mechanisms and levels of energy produced by trauma to cortical and subcortical structures may be explaining factors [15,26,28,30]. This variability leads to a lack of consensus on the interpretation of chronic DTI-based parameters after TBI or rSRC [16,21,23]. In fact, DTI has been demonstrated to be sensitive to a wide range of group differences, although no specific trends have been consistently identified [29,43]. For this reason, we performed post-hoc analyses to investigate specific and focal injuries to white matter pathways. A similar method was previously used to analyze regional white matter changes after radiation therapy in DTI-based parameters [44] and to investigate white matter anomalies in a patient with visual snow syndrome [45]. We found a different level of regional differences in the same white matter networks only partially revealed by the global analysis. Several white matter pathways displayed regionally decreased AD and RD in their midline segments in rSRC, in agreement with previous reports in chronic rSRC [16,46–48]. On the other hand, there are conflicting results showing increased AD and RD indicative of damage to axonal fibers or myelin, respectively [43,49]. Plausibly, focal and/or incomplete damage to myelin or axonal fibers may not affect the entire pathway in terms of DTI-based metrics at the chronic stage. Incomplete damage to both myelin and axons in rSRC may induce a myelin repair process with a change in the dominant cell type contributing to the signal, with axonal bundles being replaced by astrocytes and/or microglia [15]. Hence, significantly higher AD and RD and, potentially, lower FA in rSRC patients could be expected at the chronic stage [16,21,42,50]. Animal models showed that neuronal shrinkage can occur in the absence of cell death or perisomatic axotomy [16,51]. Decreased AD and RD post injury may be related to such shrinkage, leading to less surface area along axons for parallel diffusion [16,51].

These data suggest that the time course of physiological recovery extends longer than initially thought in rSRC athletes [17]. In addition, previous inconsistent results matching cognitive outcome and DTI may depend on the different DTI analysis methods used [22–24,43,51,52]. Changes in DTI-based parameters may be subtle and difficult to detect due to technical factors (such as the number of DTI directions, the algorithm used for the analysis, normalization process, the technique of tractography, the choice of global or focal indices' analysis among others) [52–55]. When the structural injury is focal and/or partially repaired, analysis of regional DTI-based metrics changes should be investigated to identify anatomical and possible links to functional information/dysfunction.

Both TBI patients and rSRC athletes performed significantly worse in the memory domain. This result agrees with other studies demonstrating impaired neuropsychological outcome in TBI patients and rSRC athletes during the subacute/chronic stage [2,4,5,7,20,23,39,56]. In patients with evidence of structural and functional abnormalities after TBI, effective

cognitive rehabilitation interventions initiated post TBI enhance the recovery process and minimize the functional disability [57]. In rSRC patients, the persistence of symptoms long beyond the generally accepted time frame for recovery may reflect the development of postconcussion syndrome (PCS) [58,59]. There is, however, no consensus regarding the clinical neuroradiological criteria for PCS and, increasingly, the term persistent post-concussive symptoms (PPCS) is used [60]. Despite persisting symptoms, many of these athletes have normal MRI investigations [59]. The possible evidence of white matter alterations may represent an important factor to consider to plan early education [61], cognitive behavioral therapy [62], and/or aerobic exercise therapy [63], which have been shown to be effective in certain patients with post-concussion syndrome [59]. Moreover, since memory seems the common most affected domain in our two populations, the specific use of external memory aids and computer-assisted strategies may be indicated, since they have also been shown to improve attention, memory, and executive skills after TBI and therefore may be of help in SRC patients with memory impairment [10,64–66].

The indication for tailored cognitive rehabilitation based on the affected white matter networks and functional impairment would therefore be necessary to either restore or compensate for memory deficits or other functional impairments commonly debilitating to rSRC patients, as well as for TBI patients [67,68].

#### *4.1. Functional Correlates of Regional White Matter Injuries*

We found FAT, ATR, and CC to be among the abnormal white matter pathways due to significant variations in DTI-based parameters, at the regional analysis in both TBI and rSRC athletes. The two groups displayed similar functional outcome and a similar mid-line location for the DTI-based parameters changes at the regional analyses for the ATR and CC. FAT is a key component of a cortico-basal ganglia- thalamic-cerebellar circuit involved in action control [69,70]. In both hemispheres, the FAT plays a role in selecting among competing representations for actions that require the same motor resources (mainly the articulatory apparatus on the left hemisphere and the oculomotor and manual/limb action systems on the right hemisphere) [69,71,72]. Its damage has been related to impairment in speech and language functions, as well as executive functions, visual–motor activities, inhibitory control, working memory, and attention [69,71]. The damage/changes to segments of ATR can also be observed in the midline fluid percussion brain injury, a rodent model of diffuse TBI wherein memory deficits are observed [73]. Electrophysiological evidence suggests a crucial role for ATR connectivity in human memory formation, connecting the anterior and midline thalamic nuclear groups to the frontal lobe [74–79]. Commissural white matter pathways such as the genu and splenium of the CC are more vulnerable in their segment close to the midline due to a close relationship with fibrotic structures such as the falx cerebri [28,80,81]. Lesions or regional changes of the CC are known to disable the interhemispheric communication of multiple memory systems and disturb memory function [82–84], in particular in long-term verbal memory performance [85,86].

Taken together, these findings suggest that regional differences in the frontal and fronto-thalamic networks (FAT, ATR) may be key contributors to the poor memory performance observed in our study. The additional involvement of the genu of CC, may have resulted in a lack of interhemispheric modulation and resilience leading to a less compensable functional impairment [82–87]. Knowledge of the presence of such white matter changes is of importance to individualize treatment strategies [41].

#### *4.2. Limitations*

Our study has some limitations. The sample size of the three groups is small, especially the TBI group. Although the groups are thoroughly characterized and matched (age and gender) the injury mechanisms resulting in the TBI or rSRC were heterogeneous. Hence, our results should, therefore, be interpreted with caution. The limited sample size influenced our analysis and some of the variables such as age, gender, and hospitalization time were not included as confounding factors in this article. Since the TBI group is too small and other studies from our group did not identify significant differences among subgroups of rSRC athletes [88], we did not include the other variables in our analysis, which was beyond the aim of this article. We aimed to investigate differences in white matter changes with two different methods and to match the neuropsychological outcome with diffusion changes in the two groups. Further studies with larger cohorts are needed to clarify the role of the confounding factors in TBI and rSRC subjects at the chronic phase. We also believe that only longitudinal studies with repeated investigations would minimize the risk of hidden differences between the groups displaying trends for possible recovery. Another limitation is that we cannot exclude that possible comorbidities, such as post-traumatic stress disorder (PTSD), which has been described in TBI patients, might have contributed to the worse performance on memory domain by the TBI group [89]. On the other hand, the similar performance in rSRC athlete, the different background of the TBI patients included in this study and the absence of similar white matter alterations as previously reported may indicate a different reason for the results presented in this article.

To our knowledge, this is the first study applying focal injury analysis to white matter pathways in both TBI patients and rSRC athletes. Hence, another limitation may be linked to our methods with the creation of several small indices for each DTI-based parameter together with the intrinsic sensitivity to noise. We defined regional differences as a minimum of 10 consecutive indices reaching the pre-determined significant levels in comparison to controls at group level in both direction (increased or decreased when compared with healthy controls). We believe the FDR analysis may represent a reliable and solid way to discriminate the effects of confounds through the analysis of voxel's spatial neighborhood. Another limitation is the possible role of cerebral contusions and CMB on our DTI results. Post-traumatic lesions such as contusions and CMB can be associated with lasting changes in perilesional white matter properties, even remotely from the lesion, and midsagittal CMB have been associated with cognitive decline after TBI [90–92]. Since we did not anatomically normalize CMB, an effect on the DTI results cannot be excluded. However, at the time of DTI, neither acquisition nor an expansive effect of the contusions was present, nor was a significant deformation of anatomical structures detected. The quality of the normalization process was carefully assessed before the analysis. In view of the differences in acquisition period and the different trend in AD and RD between TBI and rSRC, a major role for CMB is unlikely. Further studies with more advanced models of white matter investigations are necessary to minimize the effects of confounds and to, in detail, assess network vulnerability in TBI and rSRC.

#### **5. Conclusions**

In our study, TBI patients and rSRC athletes displayed different morphological images and different DTI abnormalities at the first level analysis of white matter but shared a similarly impaired memory performance at the chronic stage after injury. DTI analysis detected white matter pathway changes especially in TBI patients but underestimated focal or incomplete abnormalities when the white matter pathway was considered as one entity. At the regional analysis, similar regions of the left FAT, the genu of the CC, and the right ATR displayed different focal changes in both rSRC and TBI patients, reflecting possible differences in trauma and recovery mechanisms. The concomitant presence of white matter findings and the functional impairment observed in both TBI patients and rSRC athletes may suggest a long-term chronic impairment in some subgroups of patients despite the normal standard MRI images. This information seems crucial to better interpret the functional outcome of athletes with rSRC and to tailor individualized rehabilitative plans.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jcm11020358/s1, Table S1: Median with interquartile range for all white matter pathways and DTI-based parametric values.

**Author Contributions:** Author contributions included conception and study design, F.L., M.F. and N.M.; data collection or acquisition, F.V., S.S., E.-M.L., M.L., Y.T., J.J., A.W., G.A. and N.M.; statistical analysis, M.F.; interpretation of results, F.L., M.F., E.-M.L., S.H. and N.M.; drafting the manuscript work, F.L., M.F. and N.M.; and approval of the final version to be published and agreement to be accountable for the integrity and accuracy of all aspects of the work, All authors. All authors have read and agreed to the published version of the manuscript.

**Funding:** Swedish Brain Foundation FO2020-0147, FO2019-0190, FO2018-0166, Swedish Research Council (SRC) 2018-02500\_VR, SRC under the framework of EU-ERA-NET NEURON CnsAFlame, Swedish Research Council for Sport Science P2019-0133, P2020-0116, P2021-0105, Bissen Brain Walk and Selander Foundation—all to N.M.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the institutional ethics review board (2015/210/2).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patients to publish this paper.

**Data Availability Statement:** The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy and ethical restrictions.

**Conflicts of Interest:** N.M. and Y.T. are scientific advisors for PolarCool Inc. The other authors have no competing/conflict of interests to declare.
