Myelin Content in Mild Traumatic Brain Injury Patients with Post-Concussion Syndrome: Quantitative Assessment with a Multidynamic Multiecho Sequence

Post-concussion syndrome (PCS), characterized by a variety of somatic, cognitive, and behavioral deficits, may occur following mild traumatic brain injury (mTBI), which constitutes the majority of TBIs [1, 2, 3]. Although the reported incidence of PCS after mTBI varies from 40% to 80% during the first several weeks, symptoms have been reported to persist up to 3 months after injury in as many as 50% of patients and for more than a year in 10% to 15% of patients, imposing a considerable socioeconomic burden worldwide [4, 5, 6].

In the clinical setting, MR imaging performed to exclude any structural abnormality in PCS patients often causes patient frustration because conventional MR imaging does not reveal any abnormalities that can explain the obvious somatic, cognitive, or behavioral deficits experienced by patients. The limited role of conventional MR imaging in revealing microstructural changes in the brain of mTBI patients with PCS has led to increased efforts to detect these changes using advanced neuroimaging techniques. Over the past several decades, diffusion tensor imaging (DTI) has been used as the key imaging modality to assess axonal and myelin integrity in mTBI patients with PCS; however, studies have yielded mixed results [7, 8, 9, 10].

More recently, several studies have demonstrated that myelin volume can be estimated in terms of the myelin volume fraction (MVF) using a multidynamic multiecho (MDME) sequence [11, 12, 13, 14, 15]. In a study comparing various quantitative MR parameters in patients with multiple sclerosis, myelin partial volume and excess parenchymal water partial volume were demonstrated to be more sensitive to the disease process than R1, R2, and proton density [14, 16]. A recent study that evaluated the performance of quantitative values of various MDME sequences from different vendors reported that intrascanner repeatability and interscanner reproducibility of MVF are high across different scanners [12]. Moreover, MVF based on an MDME sequence has been shown to correlate well not only with other myelin estimation methods [13] but also with histological measures in postmortem human brain [15]. However, to the best of our knowledge, no previous studies have used MVF from an MDME sequence to quantify potential myelin loss in patients with mTBI. This study aimed to explore the myelin volume change in mTBI patients with PCS using an MDME sequence and automatic whole-brain segmentation.

This retrospective study was approved by the Institutional Review Board of Seoul National University Hospital, and the requirement for informed consent was waived due to the retrospective nature of the study (IRB No. 1905-167-1035).

Patient Selection

Our radiology report database was searched for 45 consecutive mTBI patients with PCS who underwent MR imaging at our institution between October 2016 and April 2018. The specific inclusion criteria were as follows: 1) clinical diagnosis of mTBI according to the American Congress of Rehabilitation Medicine [17], 2) clinical diagnosis of PCS based on the International Classification of Diseases, 10th Revision [18], 3) presence of MR imaging including an MDME sequence, 4) no other clinically significant neurological or psychiatric disorders, and 5) no alcohol or other substance dependence. Four patients were excluded for the following reasons: 1) overt traumatic hemorrhage (n = 2), 2) failure of automatic brain segmentation (n = 1), or 3) poor image quality of the MDME sequence (n = 1).

For comparison, 35 control subjects were selected from our radiology report database between November 2016 and March 2018. The inclusion criteria were 1) presence of MR images including MDME sequence for various neurologic symptoms (headache [n = 25], mild memory impairment [n = 2], visual disturbance [n = 1], and dizziness [n = 1]), 2) normal findings on structural MRI other than nonspecific T2 hyperintensities not exceeding the age threshold [19], and 3) no previous trauma history. Six subjects were excluded because of failure of automatic brain segmentation. The MDME sequence was included as part of our routine protocols for mild TBI patients and other patients with suspected subtle blood-brain barrier (BBB) disruption to detect subtle gadolinium contrast leakage due to mildly elevated BBB permeability.

Finally, 41 consecutive patients with mTBI (13 male and 28 female; mean age, 45 years; age range, 19–60 years) and 29 controls (7 male and 22 female; mean age, 49 years; age range, 24–60 years) were included (Fig. 1). For the mTBI patients, results of neuropsychological tests (Rivermead post-concussion symptoms questionnaire [RPQ] [20] and computerized neurocognitive function tests [CNTs] [21]) performed at the concussion clinic of the Department of Rehabilitation were recorded (details on the neuropsychological tests are shown in Supplementary Methods).

Fig. 1
Flowchart for the study patient selection.
ICD-10 = International Classification of Diseases, 10th Revision, MDME = multidynamic multiecho, mTBI = mild traumatic brain injury, PCS = post-concussion syndrome

• Click for larger image
• Download as PowerPoint slide

Regional Myelin Content Quantification Using Automated Segmentation

Automated segmentation of the whole brain based on 3D T1-weighted images was performed using FreeSurfer software (version 6.0; Laboratory for Computational Neuroimaging). Subsequently, Nordic ICE (version 4.1.2; NordicNeuroLab) was used to extract masks at four different regions of interest (ROIs) (i.e., bilateral cerebral white matter [WM], bilateral cerebral gray matter [GM], corpus callosum, and brainstem) and coregister the masks with MVF maps to obtain the average MVFs in the brain regions. Prior to the coregistration, the masks were first resampled based on myelin maps to minimize potential errors attributable to the differences in slice thickness, matrix size, and slice gap. Coregistration was automatically performed afterward by means of rigid body registration with the optimization of a mutual information metric to obtain the average MVFs at the ROIs. The total myelin volume was approximated by multiplying the average MVF by the ROI volume (Fig. 2).

Fig. 2
A schematic diagram for the regional myelin content quantification using automated segmentation.
Automated segmentation of the whole brain was performed based on precontrast 3D T1-weighted images. Masks at four different ROIs (bilateral cerebral WM, bilateral cerebral GM, corpus callosum, and BS) were extracted and coregistered with MVF maps to obtain the average MVFs at the brain regions. Total myelin volume was approximated by multiplying the average MVF by the ROI volume. BS = brainstem, D = dimensional, FSE = fast spin echo, GM = gray matter, MDME = multidynamic multiecho, MVF = myelin volume fraction, ROI = region of interest, WM = white matter

• Click for larger image
• Download as PowerPoint slide

The normality test showed that continuous clinical and imaging variables, including age and myelin volume parameters, did not follow a normal distribution (p < 0.05).

Clinical Characteristics of mTBI Patients and Controls

The age of mTBI patients (median age, 47.0 years; range, 19–60 years [interquartile range, 36.0–53.0 years]) did not significantly differ from that of the control group (52.0 years; 24–60 years [42.8–57.0 years]) (p = 0.075). No statistically significant difference in sex was observed between the two groups (p = 0.595) (Table 1).

Table 1
Clinical Characteristics of mTBI Patients with PCS and Controls

• Click for larger image
• Click for full table
• Download as Excel file

Myelin Content Quantification

Comparison of MVF between mTBI Patients and Controls

The average MVF at the bilateral cerebral WM was significantly lower in mTBI patients with PCS (25.2%; 18.2%–29.2% [22.6%–26.4%]) than that in controls (26.8%; range, 21.7%–29.4% [25.6%–27.8%]) (p = 0.004) (Figs. 3, 4, Table 2). The average MVF at the bilateral cerebral WM was not significantly correlated with the time interval between injury and MR imaging (r = -0.062, p = 0.699). In the subgroup analysis (time interval between injury and MR imaging < 3 months vs. ≥ 3 months), the average MVF at the bilateral cerebral WM was significantly lower in mTBI patients with PCS than that in controls in both subgroups (for < 3 months, 25.4%; 18.2%–29.2% [22.6%–26.4%] vs. 26.8%; 21.7%–29.4% [25.6%–27.8%], respectively; p = 0.011; for ≥ 3 months, 24.4%; 18.3%–28.5% [22.8%–26.7%] vs. 26.8%; 21.7%–29.4% [25.6%–27.8%], respectively; p = 0.024). Meanwhile, the average MVF at the bilateral cerebral GM (p = 0.524), corpus callosum (p = 0.151), and brainstem (p = 0.789) did not significantly differ between the two groups.

Fig. 3
Representative myelin maps in a 43-year-old female control (A) and a 43-year-old female mTBI patient with PCS (B).
A, B. Myelin maps displaying the MVF for each voxel demonstrate that MVF at bilateral cerebral white matter is lower in the mTBI patient with PCS (B) than that in the control (A). mTBI = mild traumatic brain injury, MVF = myelin volume fraction, PCS = post-concussion syndrome

• Click for larger image
• Download as PowerPoint slide

Fig. 4
Postprocessed myelin maps in a 43-year-old female control (A) and a 43-year-old female mTBI patient with PCS (B) (same patients as above).
A-C. Myelin maps were postprocessed to display only the voxels with the MVF of 10% or higher on a normative atlas of the brain. When the two maps were overlaid (C), it became apparent that there were fewer voxels with MVFs of 10% or higher in the mTBI patient (B) than that in the control (A). mTBI = mild traumatic brain injury, MVF = myelin volume fraction, PCS = post-concussion syndrome

• Click for larger image
• Download as PowerPoint slide

Table 2
Comparison of Average MVF and Total Myelin Volume between mTBI Patients with PCS and Controls

• Click for larger image
• Click for full table
• Download as Excel file

Comparison of ROI Volume and Total Myelin Volume between mTBI Patients and Controls

The ROI volume was significantly lower in mTBI patients with PCS than that in controls at the corpus callosum (1.87 cm³; 1.28–2.72 cm³ [1.70–2.05 cm³] vs. 2.21 cm³; 1.52–5.06 cm³ [1.86–3.46 cm³], respectively; p = 0.003) and brainstem (9.98 cm³; 7.80–13.36 cm³ [9.45–11.00 cm³] vs. 11.05 cm³; 8.96–13.43 cm³ [10.10–11.53 cm³], respectively; p = 0.015), but not at the bilateral cerebral WM (225.15 cm³; 182.32–278.49 cm³ [208.67–249.85 cm³] vs. 220.44 cm³; 169.43–292.29 cm³ [211.38–235.82 cm³], respectively; p = 0.302) and GM (223.33 cm³; 188.81–266.17 cm³ [212.74–230.96 cm³] vs. 224.05 cm³; 172.72–268.37 cm³ [207.16–234.02 cm³], respectively; p = 0.938).

The total myelin volume was significantly lower in mTBI patients with PCS than that in controls at the corpus callosum (p = 0.004) and brainstem (p = 0.042), but not at the bilateral cerebral WM (p = 0.165) and GM (p = 0.508) (Table 2). The total myelin volumes at the corpus callosum and brainstem were not significantly correlated with the time interval between injury and MR imaging (r = -0.004, p = 0.979; r = -0.267, p = 0.091, respectively). In the subgroup analysis, the total myelin volume at the corpus callosum was significantly lower in mTBI patients with PCS than that in controls in both subgroups (for < 3 months, 0.45 cm³; 0.27–0.66 cm³ [0.37–0.49 cm³] vs. 0.48 cm³; 0.33–0.79 cm³ [0.45–0.54 cm³], respectively; p = 0.015; for ≥ 3 months, 0.44 cm³; 0.26–0.56 cm³ [0.39–0.47 cm³] vs. 0.48 cm³; 0.33–0.79 cm³ [0.45–0.54 cm³], respectively; p = 0.018). The total myelin volume at the brainstem was significantly lower in mTBI patients with PCS than that in controls in the subgroup with the interval ≥ 3 months (1.37 cm³; 1.17–1.69 cm³ [1.22–1.46 cm³] vs. 1.54 cm³; 1.17–1.87 cm³ [1.42–1.67 cm³], respectively; p = 0.005), but not in the subgroup with the interval < 3 months (1.48 cm³; 0.88–1.93 cm³ [1.35–1.61 cm³] vs. 1.54 cm³; 1.17–1.87 cm³ [1.42–1.67 cm³], respectively; p = 0.328). Meanwhile, the total myelin volume at the bilateral cerebral WM tended to be lower in patients (56.56 cm³; 36.93–76.15 cm³ [49.85–60.55 cm³]) than that in controls (60.34 cm³; 44.03–75.03 cm³ [53.73–62.67 cm³]) (p = 0.165) (Table 2).

Correlations between Myelin Volume Parameters and Neuropsychological Tests

The scores of neuropsychological tests and specific PCS symptoms of mTBI patients are summarized in Supplementary Tables 2 and 3. Of the 41 mTBI patients, 22 patients had an interval of 2 weeks or less between the CNTs and MR scanning, whereas 18 had an interval of 2 weeks or less between the RPQs and MR scanning. In the subgroup analysis, the average MVF at the bilateral cerebral WM did not have significant correlations with any neuropsychological tests (all p > 0.05). However, the total myelin volume in the bilateral cerebral WM had a significant positive correlation with VLT (delayed recall) (r = 0.425; p = 0.048) and a borderline positive correlation with VLT (delayed recognition) (r = 0.391; p = 0.072) (Table 3). No significant correlations were noted between the total myelin volume at the bilateral cerebral WM and other neuropsychological tests (p > 0.05). The total myelin volumes at the corpus callosum and brainstem did not show significant correlations with any neuropsychological tests (all p > 0.05) (Supplementary Table 4).

Table 3
Correlations between Myelin Volume Parameters at Bilateral Cerebral WM and Neuropsychological Tests

• Click for larger image
• Click for full table
• Download as Excel file

In this study, the utility of the 2D FSE MDME sequence was investigated to quantify the myelin volume change in mTBI patients with PCS. The key findings of this study were as follows: 1) the average MVF at the bilateral cerebral WM was lower in mTBI patients with PCS than that in controls, 2) the total myelin volumes at the corpus callosum and brainstem were lower in mTBI patients with PCS than that in controls, and 3) no significant correlation was observed between myelin volume parameters and neuropsychological test scores, except that between the total myelin volume at the bilateral cerebral WM and verbal learning test (delayed recall).

Traditionally, considerable attention has been devoted to the pathological alterations in neuronal cells within the GM in prior studies on brain injury [22]. However, recent studies have underscored that the loss of WM integrity is also implicated in the long-term prognosis of TBI patients [23, 24]. In the analysis of WM integrity in mTBI, myelin is an attractive target because it mainly consists of bundles of myelin-ensheathed axons [22]. Several preclinical studies have reported that myelin loss can occur even after mTBI and that it can persist chronically, particularly after repeated injury [25, 26]. Moreover, it has been shown that myelin loss can subsequently lead to reciprocal damage of the underlying axon in WM injury because demyelination in WM injury accompanies the death of myelin-producing oligodendrocytes, which actively sense the axonal energy needs and thus play an important role in maintaining axonal integrity [27, 28].

Over the past several years, DTI has been shown to have an advantage over conventional structural MR imaging in detecting subtle changes in WM integrity [29]. In DTI, the direction and magnitude of diffusion in a given voxel are represented by various anisotropy indices, including fractional anisotropy, mean diffusivity, radial diffusivity, and axial diffusivity [30]. Fractional anisotropy, the most widely used DTI measure, has been reported to be decreased in several WM tracts (e.g., anterior corona radiata, cingulum bundle, uncinate fasciculus, and superior longitudinal fasciculus), representing less restricted (more isotropic) diffusion in the injured WM [7, 8, 9]. However, the decrease in fractional anisotropy value is not specific to myelin loss and may be attributable to several other causes, including disturbances in axonal membranes, neurofilaments, or microtubules [31]. In comparison, radial diffusivity is known to be more specific to myelin injury because it measures the diffusion perpendicular to the predominant orientation of axonal fibers [32, 33]. However, previous studies have shown mixed results with reports on both increased and decreased radial diffusivity values in mTBI as compared with controls [9, 34, 35]. Moreover, factors such as inflammation, axonal properties, and the presence of crossing fibers have been shown to have a significant influence on radial diffusivity [36, 37, 38].

More recently, multicomponent relaxometry, a quantitative MR technique that provides a more specific measure of the myelin–myelin water fraction (MWF) (i.e., a surrogate marker of myelin volume)–as compared with DTI, has been developed. Using a 32 echo T2 scan sequence to quantify MWF, Wright et al. [39] demonstrated decreased MWF in several regions of concussed athletes (hockey players), including the corpus callosum, corona radiata, internal capsule, posterior thalamic radiation, and superior longitudinal fasciculus at 2 weeks post-injury as compared to the baseline. In addition, a previous study on veterans with mTBI used multicomponent-driven equilibrium single-pulse observation of T1 and T2 to derive MWF and revealed that MWF was significantly lower in a small number of veterans with a history of mTBI than in those without a history [40]. In this study, we used an MDME sequence to quantify MVF and showed that the average MVF at the bilateral cerebral WM was lower in mTBI patients with PCS than that in controls, in keeping with those of previous studies [39, 40]. However, Wright et al. [39] found that MWF values recovered at 2 months post-injury. Moreover, in a prospective study of MWF changes after mTBI in football and rugby players [41], increased myelination was observed in contact sports players at the time of injury than that in noncontact sports players, and even larger increases were found at 3 months afterward. In contrast, decreased MVF even in the mTBI subgroup with the time interval between injury and imaging longer than 3 months in our study may suggest that myelin loss following mTBI may persist chronically. The possibility remains that our mTBI patients may have had partial remyelination, but a future study with longitudinally collected data is warranted to elucidate the exact natural course of myelin loss in mTBI.

Despite the lack of significant difference in MVF, we found that the total myelin volumes at the corpus callosum and brainstem were lower in mTBI patients with PCS than that in controls, which was attributed to the decrease in ROI volume in mTBI patients with PCS. This finding is in line with a previous study that reported that the corpus callosum and brainstem are sites commonly involved in focal atrophy following TBI, along with the thalamus, hippocampus, cerebellar GM, and corona radiata [42].

In this study, clinical correlations between the myelin volume parameters and various neuropsychological tests were also performed. Unlike the previous study, which reported the association between lower MWF across several ROIs and worse performance on a speeded attention task [40], we did not find significant correlations between MVF and any of the neuropsychological tests. Furthermore, no correlation was noted between the total myelin volume and neuropsychological test scores, except for the total myelin volume at the bilateral cerebral WM and verbal learning test (delayed recall). Although the preliminary result may imply that the total myelin volume at the bilateral cerebral WM was more important than the average MVF in terms of verbal learning, the possibility remains that the finding is opportunistic. Therefore, a future study including a larger number of patients is warranted to explore the relationship between myelin volume and clinical parameters.

The clinical implications of the findings of this study are as follows. At present, the diagnoses of mTBI and PCS are made clinically based on subjective questionnaires and neuropsychological tests. Nonetheless, there is considerable overlap in the symptoms and cognitive test performance between PCS and psychiatric disorders such as depression or posttraumatic stress disorder [29], hindering the clinical diagnosis. In this regard, MVF from the 2D FSE MDME sequence may provide more objective and quantitative data for the microstructural changes in myelin, thereby helping differentiate PCS from psychiatric disorders.

Our study has several limitations. First, owing to the retrospective nature of the study, a correlational analysis between myelin volume parameters and neuropsychological tests was performed in a subgroup of patients who had a relatively short time interval between the two tests. A future prospective study with a larger sample size is needed to validate the correlation. Second, the comparison was made with the controls without TBI who were not explicitly matched for age and sex. Although there was no significant difference in age and sex between the two groups, the group difference could have been higher if the age and sex were matched between mTBI patients and controls for comparison. Third, the study lacked longitudinal data; thus, the natural course of myelin loss or the prognostic value of MVF could not be elucidated from the present study. Fourth, although the MR imaging of the controls included in this study revealed no structural abnormalities except nonspecific T2 hyperintensities that did not exceed the age threshold, they presented with various neurologic symptoms prior to MRI. The possibility remains that the detected difference in myelin volume could have been higher if we included individuals without any neurologic symptoms as controls.

In conclusion, the MDME sequence can readily depict myelin loss in patients with mTBI with PCS. In particular, MVF quantified from the MDME sequence was decreased in the bilateral cerebral WM in mTBI patients with PCS. Moreover, the total myelin volumes at the corpus callosum and brainstem were decreased in mTBI patients with PCS due to atrophic changes.

The Supplement is available with this article at https://doi.org/10.3348/kjr.2021.0253.

Click here to view.^{(48K, pdf)}