http://informahealthcare.com/bij ISSN: 0269-9052 (print), 1362-301X (electronic) Brain Inj, 2015; 29(1): 47–57 ! 2015 Informa UK Ltd. DOI: 10.3109/02699052.2014.947628

ORIGINAL ARTICLE

A longitudinal evaluation of diffusion kurtosis imaging in patients with mild traumatic brain injury Jesse A. Stokum1, Chandler Sours1,2, Jiachen Zhuo1, Robert Kane3, Kathirkamanthan Shanmuganathan1, & Rao P. Gullapalli1,2 1

Department of Diagnostic Radiology & Nuclear Medicine, 2Program in Neuroscience, University of Maryland School of Medicine, Baltimore, MD, USA, and 3Georgetown University School of Medicine, Washington, DC, USA

Abstract

Keywords

Primary objective: To investigate longitudinal diffusion tensor imaging (DTI) and diffusion kurtosis imaging (DKI) changes in white and grey matter in patients with mild traumatic brain injury (mTBI). Research design: A prospective case-control study. Methods and procedures: DKI data was obtained from 24 patients with mTBI along with cognitive assessments within 10 days, 1 month and 6 months post-injury and compared with age-matched control (n ¼ 24). Fractional anisotropy (FA), mean diffusivity (MD), radial diffusion (lr), mean kurtosis (MK) and radial kurtosis (Kr) were extracted from the thalamus, internal capsule and corpus callosum. Main outcomes and results: Results demonstrate reduced Kr and MK in the anterior internal capsule in patients with mTBI across the three visits, and reduced MK in the posterior internal capsule during the 10 day time point. Correlations were observed between the change in MK or Kr between 1–6 months and the improvements in cognition between the 1 and 6 month visits in the thalamus, internal capsule and corpus callosum. Conclusions: These data demonstrate that DKI may be sensitive in tracking pathophysiological changes associated with mTBI and may provide additional information to conventional DTI parameters in evaluating longitudinal changes following TBI.

ANAM, diffusion kurtosis imaging, longitudinal analysis, mild traumatic brain injury.

Introduction It is estimated that traumatic brain injury (TBI) was responsible for 1.4 million hospital emergency department visits each year [1]. In addition, TBI is a contributing factor in 30.5% of injury-related deaths in the US [2]. While TBI is empirically divided into mild, moderate and severe based on the Glasgow Coma Scale (GCS), a majority of these patients suffer mild traumatic brain injury (mTBI) (GCS 13–15). In many of these cases, CT and conventional MRI fail to demonstrate the diffuse axonal injury (DAI) that is often present among patients with mTBI [3]. Furthermore, many of these patients experience post-concussive symptoms [4, 5], stressing the importance of researching new methods to aid in the prediction of patient outcome, since the severity of detectable structural injury does not always correlate with degree of cognitive impairment [6]. Thus, there is much interest in developing imaging biomarkers that would demonstrate the extent of diffuse axonal abnormalities and have the ability to track these changes in vivo. Correspondence: Rao P. Gullapalli, PhD, Department of Diagnostic Radiology and Nuclear Medicine, University of Maryland School of Medicine, 22 South Greene Street, Baltimore, MD 21201, USA. Tel: 410-328-2099. Fax: 410-328-5937. E-mail: [email protected]

History Received 29 January 2014 Revised 26 June 2014 Accepted 20 July 2014 Published online 26 September 2014

Recent advances in neuroimaging techniques hold promise to be more sensitive and accurate in the detection and characterization of microstructural and biochemical changes that occur in the axon due to mTBI. In recent years, diffusion tensor imaging (DTI) has been found to be sensitive to mTBIinduced changes in both whole brain and regional white matter [6–8]. For example, mean diffusivity (MD) is reported to be increased in the corpus callosum and posterior internal capsule after mTBI [7], while decreased FA is noted after mTBI in white matter bundles such as uncinate fasciculus, corpus callosum, internal capsule and lobar white matter [6, 7]. More studies have indicated that DTI values are dynamically changing over time, due to the nature of TBI being a dynamically changing injury. For example, elevated FA, reduced MD and reduced lr are often observed at acute stages (within 1 week post-injury), indicating an inflammatory response such as axonal swelling or cytotoxic oedema [9–11]. Mayer et al. [12] reported such a process to persist to semi-acute phases (410 days post-injury). However, at chronic stages (6 months or longer), most studies reported a reduced FA and increased MD as a result of increased la, which is consistent with axonal damage [12]. Changes in FA are also shown to correlate with injury severity [13], cognitive deficits [6, 8] and functional outcome [14]. Increased lr, as an

48

J. A. Stokum et al.

indication of irreversible myelin damage, is also observed in persons with severe but not mild TBI [8]. Importantly, the DTI model used to estimate the diffusion tensor inherently assumes a mono-exponential decay of the water signal when diffusion gradients are applied. The concept of monoexponential decay is valid in most biological systems and for many routine clinical evaluations. Specifically, it is a suitable assumption where the b-values (a measure of the sensitivity to water diffusion) are limited to 1000 s mm2, which measures water diffusion distances of 5–10 mm over 50–100 milliseconds [15]. However, the concept of monoexponential decay has been shown to be invalid when higher b-values are used to probe water diffusion over even shorter molecular distances, making them increasingly sensitive to heterogeneous cellular structures [15, 16]. It has been shown that at high b-values (greater than 1500 s mm2) or high diffusion sensitization values, the signal attenuation deviates from linear behaviour and that the attenuation pattern is better captured by the diffusion kurtosis model. Using this model, diffusion kurtosis imaging (DKI) has shown to not only provide the diffusion tensor estimates such as FA, MD and the eigenvalues of the tensor, but also the kurtosis tensor which measures the complexities of the microstructure, especially across the axon in the case of white matter. In addition, DKI has been shown to be effective for monitoring microstructural changes even within the gray matter [17–19], changes associated with human ageing [20], Alzheimer’s disease [21, 22] and experimental models of TBI [23]. Human studies have demonstrated the value of DKI for tracking pathologies which involve changes in tissue microarchitecture. For instance, Raab et al. [24] demonstrated the value of DKI in characterizing gliomas, where higher grade gliomas that are highly vascularized and show increased cellular proliferation exhibit higher MK than lower grade tumours. DKI has also demonstrated changes in both grey and white matter, as Falangola et al. [20] found that grey and white matter MK undergoes changes over an individual’s lifetime, presumably due to changes in glial cell packing and myelination. MK in the anterior cingulum has also been identified to have better diagnostic performance than FA for Parkinson’s disease [25]. Finally, an animal study performed by this group using a controlled cortical impact experimental model of TBI demonstrated that DKI is capable of detecting white and gray matter changes in the acute and sub-acute stages of mTBI. One key finding in this study was that MK and not MD or FA was sensitive to reactive astrogliosis, emphasizing the non-overlapping roles for DKI and DTI [23]. While two human studies have previously explored DKI changes following mTBI, where DKI was found to be both sensitive to mTBI associated changes and related to neurocognitive performance [26, 27], additional research assessing longitudinal changes in DKI parameters is still warranted. As mTBI often involves DAI that is rarely detectable through conventional imaging techniques, it was hypothesized that DKI may be a powerful tool to study the progression of this disease as well as have prognostic value. This study evaluated DKI parameters on 24 patients with mTBI across three time points (within 10 days, 1 month and 6 months of injury). In addition, it evaluated the relationship between

Brain Inj, 2015; 29(1): 47–57

changes in cognitive performance and changes in DKI parameters across the initial 6 months of recovery.

Materials and methods Subjects Twenty-four patients (aged 37.4 ± 14.1 years, 18 M:6F) admitted to the Adam Cowley Shock Trauma Center at the University of Maryland Medical Center were recruited into this study. Informed consent was obtained from each participant and the study was approved by the IRB of the University of Maryland Medical Center. All patients participated in MRI evaluations within 10 days of injury (6 ± 3 days; range ¼ 1–11), 1 month (33 ± 7 days; range ¼ 25–53) and 6 months (196 ± 33 days; range ¼ 137–277) following injury. One patient did not receive an MRI evaluation at the 6 month visit. During each visit, in addition to the MRI, the patients completed a computerized neuropsychological assessment using the Automated Neuropsychological Assessment Metrics (ANAM) [28]. All patients spoke fluent English and were 18 years of age or older. Criteria for inclusion in this study included an infield or admission GCS of 13–15, as well as a mechanism of injury indicative of blunt force trauma and either a positive head admission CT or altered mental status and/or loss of consciousness for less than 30 minutes. This patient population consisted of patients with mTBI sub-classified as complicated (positive admission CT) and uncomplicated (negative admission CT). Participants with history of prior neurological and psychiatric illness, stroke, brain tumours or seizure disorders or pregnant were excluded from this study. Twenty-four healthy control subjects (aged 33.9 ± 14.7 years, 13M:11F) with no history of TBI or any of the abovementioned patient exclusion criteria were also recruited into this study. The control population underwent one MRI evaluation and one ANAM assessment. Table I shows all the pertinent patient and control demographic data. Table II provides patient-specific information on an individual basis including in field and admission GCS, injury mechanism, gender, and CT results at final visit. Mechanisms of injury included fall (n ¼ 8), motor vehicle accidents (n ¼ 7), assault (n ¼ 5), sports-related injuries (n ¼ 2) and bicycle accidents (n ¼ 2). Six of the subjects had a positive CT for trauma-related injuries while two patients had non-trauma-related incidental findings (arachnoid cyst in cerebellum and left parietal lobe hemangioma). Since the incidental findings were not within the regions investigated within this analysis, this study included the patients and classified them as negative CT for trauma-related injuries.

Table I. Comparison of mTBI and healthy control demographic data. Variable Age (years ± SD) Gender Education length (years ± SD) Positive CT

mTBI

Healthy control

p value

37.4 ± 14.1 18 male/6 female 14.1 ± 3.2

33.9 ± 14.7 13 male/11 female 14.7 ± 2.1

0.415 0.137 0.458

NA

NA

6/24 (25%)

Longitudinal diffusion kurtosis imaging in mTBI

DOI: 10.3109/02699052.2014.947628

49

Table II. Patient demographic data. 10 day visit (days)

1 month visit (days)

6 month visit (days)

Male Male Male Male Male Male

2 6 6 1 10 5

28 33 34 29 33 32

24 53 26 39 27 19 47

Male Male Male Male Male Male Male

4 3 2 8 7 1 11

15 12

30 21

Female Male

15

15

40

15 12 15 15 10 15

15 15 13 15 15 15

14 15

15 15

Field GCS

Admit GCS

Age (years)

15 15 15 15 14 15

15 15 15 15 15 15

28 59 53 38 20 18

14 15 15 15 15 15 15

15 15 15 15 15 15 15

15 15

Injury mechanism

Positive head CT

188 215 214 202 158 163

MVA Fall Bicycle accident MVA Assault Sports accident

32 39 38 29 28 33 51

266 177 148 NA 218 216 218

MVA MVA MVA Assault Sports accident Fall Assault

6 6

30 53

215 188

Bicycle Accident Fall

Female

7

25

137

Fall

18 49 60 23 62 27

Male Male Female Female Female Male

9 1 10 9 10 2

36 28 27 33 30 32

197 178 188 177 277 205

Fall Assault MVA MVA Fall Fall

54 63

Female Male

10 5

31 30

184 173

Fall Assault

No No L frontal cortical contusion No No No (incidental arachnoid cyst in cerebellum) No No No No No No Epidural haematoma and left occipital lobe contusion No Epidural haematoma, scattered occipital SAH No, left parietal hemangioma (incidental) No No No No No Bilateral frontal lobe contusion and small anterior SAH SAH right frontal region SDH left posterior parietal lobe

Gender

MVA, motor vehicle accident, SAH, subarachnoid haemorrhage; SDH, subdural haemorrhage.

Imaging Imaging was performed using a 3 T Siemens Tim Trio Scanner (Siemens Medical Solutions; Erlangen, Germany). Diffusion weighted images were obtained with b ¼ 1000, 2000 s mm2 at 30 directions, together with four b0 images, in-plane resolution ¼ 2.7 mm2, TE/TR ¼ 101 ms/6000 ms at a slice thickness of 2.7 mm with two averages. High resolution T1-weighted-MPRAGE (TE ¼ 3.44 ms, TR ¼ 2250 ms, TI ¼ 900 ms, flip angle ¼ 9 , resolution ¼ 256  256  96, FOV ¼ 22 cm, slice thickness ¼ 1.5 mm) images covering the entire brain were acquired for anatomic reference. Diffusion weighted images were motion corrected using the co-registration option available in the SPM8 (Statistical Parametric Mapping, Wellcome Department of Imaging Neuroscience, University College London, UK). 3D Gaussian smoothing with FWHM ¼ 3.0 mm were applied to improve the signal SNR. DKI reconstruction was then carried out on each voxel using the in-house MATLAB program, as described by Zhuo et al. [23]. The 3D-MPRAGE images were co-registered and re-sampled to the b0 image and then followed by image segmentation to different tissue types (WM, GM and CSF) using SPM8. The theoretical DKI model is shown in equation (1) [30]. 1 ln½SðbÞ ¼ lnðS0 Þ  bDapp þ b2 D2app Kapp þ Oðb3 Þ 6

ð1Þ

where Dapp is the apparent diffusion coefficient, S0 is the diffusion weighted signal, b is the diffusion sensitization

factor and Kapp is the apparent kurtosis derived from the multiple directions of diffusion for a given voxel. The difference between equation (1) and the standard DTI form is that the standard DTI error term, which has a square dependence on ‘b’ is replaced by 16 b2 D2app Kapp þ Oðb3 Þ. Given that the error is now proportional to b3, higher b-values provide a measure of kurtosis. DKI reconstruction thus performed yields mean kurtosis (MK), radial kurtosis (Kr) or the average kurtosis perpendicular to the direction of maximal diffusion, axial kurtosis (Ka) or the kurtosis parallel to the direction of maximal diffusion, mean diffusivity (MD), fractional anisotropy (FA) and radial and axial diffusion (lr, la, respectively). Note that the diffusion tensor parameters MD, FA and lr can be obtained simultaneously with the kurtosis tensor using the DKI model in equation (1) and not from separate DTI model fitting. Furthermore, because the error term is smaller than that from a traditional DTI model, the MD, FA and lr estimates obtained from this approach are a more accurate representation of these parameters. Image post-processing Six regions of interest (ROI) selected based on previous DTI literature were defined and hand traced by a single investigator (JAS) on images from patients and healthy control subjects using FA and MD maps to ensure exclusion of CSF. Each ROI was visually inspected by a second investigator (CS) to insure reliability among ROIs. Figure 1 shows a FA

50

J. A. Stokum et al.

Brain Inj, 2015; 29(1): 47–57

Figure 1. Axial views of a T1-weighted MPRAGE image of a representative healthy control overlaid with the six of the ROIs. ROIs include the genu of corpus callosum, body of corpus callosum, splenium of corpus callosum, anterior and posterior limbs of the internal capsule, and thalamus.

map from a single healthy control subject as a representative sample with the ROIs. ROIs were bilateral and included the thalamus, anterior internal capsule, posterior internal capsule, genu, body and splenium of the corpus callosum. White matter regions were traced along the maximum points on white matter bundles on the FA map. MD maps were used to ensure that ROIs excluded the ventricular space. The thalamus was included to serve as a grey matter region. Each ROI was defined on exactly three axial slices. Custom software written in MATLAB was used to calculate the mean and standard deviation of MD, FA, MK, Kr, and lr values from each ROI. Neurocognitive testing During each patient and control subject visit, the ANAM was administered using a laptop computer. Due to their initial clinical condition, nine of the patients were unable to complete the ANAM during their first visit. This ANAM test battery consisted of seven sub-tests, including simple reaction time, procedural reaction time, code substitution, code substitution delayed, match to sample, mathematical processing and a repetition of the simple reaction time. Due to interest in more complex cognitive processing, it was opted to only analyse the code substitution, code substitution delayed, match to sample and mathematical processing individually. The selected sub-tests of the ANAM are shown in Table III along with the cognitive domains that they assess [28]. For each sub-test, accuracy and response time information was collected and a throughput score was computed. The throughput score is a ratio between the number of correct answers and overall reaction time and has been shown to have greater sensitivity and reduced variability compared to reaction time or percentage correct alone [31]. Specifically the ANAM assess the throughput score as the number of correct responses per the total time that the participant took to respond each time the stimulus appeared on the screen (number of correct responses/minute) [32]. A higher throughput score indicates a faster, more accurate performance. Finally, a weighted throughput (WT-TH) was computed (previously referred to as the Index of Cognitive

Table III. ANAM sub-tests and cognitive domains. Sub-test

Abbreviation

Cognitive domains

Code Substitution

CS

Math

MATH

Match to Sample

MTS

Code Substitution Delayed Weighted Throughput

CSD

Visual search, sustained attention, working memory Computational skills, concentration, working memory Spatial processing, visuospatial working memory Sustained attention, working memory Overall Cognitive Efficiency

WT

Efficiency), which is a weighted summary of the throughput scores from each of the sub-tests [33]. The weighting was done so that the throughput score from each sub-test contributed equally to the overall score. The precise equation used in this analysis is: 0 1 CS  4:35 þ CSD  5 þ MTS  6:63 @ þMATH  8:37 þ PRT  2:18 A þSRT þ SRT2 WT  TH ¼ : ð2Þ 7 Statistical analysis Differences in demographic variables between the patient and control group were compared using t-tests ( ¼ 0.05). Mean patient DKI values at each visit were examined for departure from control ranges by Analysis of Covariance (ANCOVA) ( ¼ 0.05) controlled for age. Patient performance on the ANAM battery was then compared to the performance of healthy controls. The mean patient throughput score for each ANAM sub-test and the weighted throughput score was compared to control levels with ANCOVA ( ¼ 0.05) controlled for age. All data shown are uncorrected.

DOI: 10.3109/02699052.2014.947628

Longitudinal diffusion kurtosis imaging in mTBI

51

Figure 2. Time series plots of mean ± standard error of patient ANAM throughput scores at visit 1 (10 days), visit 2 (1 month) and visit 3 (6 months). In each plot, the mean and spread of healthy control values are represented by three dotted horizontal lines denoting the mean ± standard error of the given ANAM throughput score. Significant differences assessed by ANCOVAs controlling for age between each visit and the control population (*p50.05, #p50.10).

In addition, repeated measures Analysis of Covariance (ANCOVA) ( ¼ 0.05) controlled for age were performed to assess longitudinal changes within the mTBI group in both mean DKI values and mean ANAM throughput scores. Results are uncorrected for multiple comparisons. Associations between DKI parameters and neurocognitive performance during recovery following trauma were investigated. Pearson partial correlations between the weighted throughput score and DKI parameters were computed controlling for years of education and age. To further assess whether the recovery of cognition between two time points was associated with changes in DKI parameters in any given ROI, this study computed Pearson partial correlations between changes in weighted throughput scores and changes in DKI parameters. Correlations were performed for all visit pairs.

Results Patient demographics There was no significant difference between patients and healthy volunteers in age (p ¼ 0.415) or education (p ¼ 0.458). In addition, there was no significant difference in the gender mix between the healthy control subjects and the patient population (p ¼ 0.137). Neurocognitive assessment The throughput scores on each ANAM sub-test for the patients with mTBI at each time point were compared against the healthy control subjects. There was a non-significant trend in reduced performance on the code substitution delayed subtest during the 10-day time point (p ¼ 0.051). The patients’ performance was comparable to the control subjects for each of the other sub-tests across the three visits (Figure 2) (all p values40.05). In addition, based on longitudinal analysis, the results fail to provide evidence for longitudinal changes in ANAM performance across the three visits (all p values40.05). DKI assessment No significant differences in any of the ROIs observed between the patients with mTBI and control subjects were found for MD, FA or lr for the 10 day, 1 month or 6 month time points (Figure 3).

Greater differences were detected in kurtosis parameters, as shown in Figure 4. Kr was decreased in the anterior internal capsule at all three visits (10 days: Kr p ¼ 0.058; 1 month: Kr p ¼ 0.050; 6 month: Kr p ¼ 0.041), although it did not reach statistical significance at the 10 day time point. MK was reduced in the anterior internal capsule at all three visits (10 days: MK p ¼ 0.072; 1 month: MK p ¼ 0.046; 6 month: MK p ¼ 0.027), although it did not reach statistical significance at the 10 day time point. MK was reduced in the posterior internal capsule at the 10 day visit (p ¼ 0.042), but not the 1 month or 6 month time points. There was a trend in reduced Kr in the posterior internal capsule at the 10 day time point only (0.058). No differences in MK or Kr were detected in the thalamus or corpus callosum. Based on this longitudinal analysis of DKI parameters, the results fail to provide evidence to longitudinal changes within the mTBI group in any of the parameters (MD, FA, MK, Kr or lr) in any of the ROIs that were examined (all p values40.05). Associations between neurocognitive performance and DKI parameters To investigate the possibility that DKI values are associated with neurocognitive performance in patients, partial correlations corrected for age and years of education were computed between patient DKI parameters and ANAM weighted throughput scores at individual time points. No significant correlations of DKI parameters with ANAM throughput scores were observed within individual visits. Similar analysis on the healthy control group also failed to reveal a significant correlation between DKI parameters and cognitive performance. Finally, changes in MK and Kr between each visit were compared to changes in ANAM weighted throughput score. This analysis was done between visits 1 and 2, visits 2 and 3 and visits 1 and 3. While significant correlations were absent between visits 1 and 2 and visits 1 and 3, significant positive correlations for both MK and Kr were found between visits 2 and 3 in multiple regions (Figure 5). The changes in the ANAM weighted throughput score was positively correlated with changes in MK values between visits 2 and 3 (Figure 5a) in the thalamus (r ¼ 0.62, p ¼ 0.0016), anterior internal capsule (r ¼ 0.53, p ¼ 0.013), posterior internal capsule (r ¼ 0.60, p ¼ 0.0029) and splenium of the corpus callosum

J. A. Stokum et al.

Figure 3. Time series plots of mean ± standard error of patient DTI ROI values at visit 1 (10 days), visit 2 (1 month) and visit 3 (6 months). In each plot, the mean and spread of healthy control values are represented by three dotted horizontal lines denoting the mean ± standard error of the given DTI parameter. Significant differences assessed by ANCOVAs controlling for age between each visit and the control population (*p50.05, #p50.10).

52 Brain Inj, 2015; 29(1): 47–57

Longitudinal diffusion kurtosis imaging in mTBI Figure 4. Time series plots of mean ± standard error of patient DKI ROI values at visit 1 (10 days), visit 2 (1 month) and visit 3 (6 months). In each plot, the mean and spread of healthy control values are represented by three dotted horizontal lines denoting the mean ± standard error of the given DKI parameter. Significant differences assessed by ANCOVAs controlling for age between each visit and the control population (*p50.05, #p50.10).

DOI: 10.3109/02699052.2014.947628

53

(r ¼ 0.55, r ¼ 0.078). A similar trend was noted with the ANAM weighted throughput score being positively correlated with the changes in Kr values between visits 2 and 3 in the thalamus (r ¼ 0.59, p ¼ 0.0033), anterior internal capsule (r ¼ 0.64, p50.001), posterior internal capsule (r ¼ 0.60, p50.001) and splenium of corpus callosum (r ¼ 0.55, p ¼ 0.0079) (Figure 5(b)).

Discussion Significance Reliable clinical detection of mTBI is currently restricted to cases with overt lesions or haemorrhage. The data suggest that DKI might complement existing neuroimaging methods used to characterize mTBI. Only one group has examined DKI changes following mTBI in a human population [26, 27] and there are limited studies examining longitudinal changes in DTI following mTBI [34]. This study expands on the findings of Grossman et al. [26, 27], by having a tighter control of time-since-injury for each visit and following patients across three visits within a 6-month period. Grossman et al. examined patients at an acute to 6-months post-injury time frame and found decreased MK in the thalamus and the internal capsule among seven patients with mTBI. In addition to these regions, the centrum semiovale and the splenium of the corpus callosum also exhibited lower MK among patients observed beyond 1 year of injury in 15 patients with mTBI (ranging from 1.33–9.58 years post-injury) [26]. Similar changes were found in another set of patients observed between 1 month and 1 year post-injury in the thalamus, internal capsule, external capsule, corpus callosum and total white matter. In this set of patients, Grossman et al. [27] also found similar trends as in their previous work with reduced MK in the thalamus, various other white matter regions and the whole brain within 1 month of injury. Furthermore, cognitive impairment was associated with MK in a few regions including the thalamus and the internal capsule. Although this patient group did not exhibit decreased MK in the thalamus during the chronic stage, it did detect decreased Kr and MK in the anterior limb of the internal capsule across the three visits (Figure 4), which is in agreement with the findings of Grossman et al. [26]. In addition, reduced MK was found within the posterior internal capsule in the initial 10 day visit. While this analysis fails to detect strong group difference in diffusion tensor parameters (MD, FA or lr) among patients with mTBI, the greatest contribution this analysis adds to the current literature is that between 1–6 months, changes in DKI parameters are associated with improvements in cognitive performance. These findings suggest a unique period of recovery of both biophysical changes and behavioural improvements. The differences between these findings and those of Grossman et al. [26] during the chronic stage are most likely due to the difference in time since injury. The patients in this study were assessed between 5–7 months, while Grossman et al. [26] assessed patients with mTBI during a much wider time frame, up to 9 years post-injury. Taken together, the findings of this study and that of Grossman et al. likely indicate the variability in the time course of changes towards recovery. The underlying pathophysiologic changes among

J. A. Stokum et al.

Figure 5. Significant correlations found between the change observed in ANAM weighted throughput and the change observed in ROI mean MK (A) and mean Kr (B). Both differences were calculated such that the second visit (1 month) value was subtracted from the third visit (6 month) value. Correlation coefficients and p values were computed by Pearson partial correlation controlled for age and years of education.

54 Brain Inj, 2015; 29(1): 47–57

DOI: 10.3109/02699052.2014.947628

patients with mTBI may continue to evolve well past the 6 month time frame, which suggests that longer longitudinal evaluations are needed to fully characterize the recovery from injury [26, 27]. Furthermore, given that axonal injury is likely both spatially and temporally dependent, differences in results may stem from differences in mechanical impact following injury in addition to the difference in post-injury time points examined [25]. Imaging findings MD, FA and lr demonstrated very little change in this patient group at each of the observed time points. While other studies with larger sample sizes demonstrate increased sensitivity of DTI in assessing traumatic white matter injury, the apparent lack of significance in these parameters in the data may in part be due to the relatively smaller sample size and the heterogeneous nature of the injury [12, 35]. However, in this study changes in kurtosis measures suggest abnormalities specifically in the anterior limb of the internal capsule across the three time points and in the posterior limb during the 10 day time point. mTBI-related processes in this region appear to be weighted towards those that produce an alteration of water diffusion heterogeneity, as most sensitively depicted by the diffusion kurtosis parameter [36], as opposed to changes in anisotropy which would have been noted by conventional diffusion parameters such as MD and FA. However, pathophysiological changes resulting in an altered diffusion profile among these parameters are dominant in this data only within the internal capsule. Following injury, white matter follows a complicated chain of events involving both apoptotic and necrotic pathways, leading to degradation of microtubules, cell swelling, axonal varicosities, demylination, secondary axotomy, oedema and cell death [37]. Given the complexity of neural response to injury it is unlikely that a single process is driving these changes. More reasonably, the detected DKI changes reflect an integration of multiple events, each of which may lead to tissue complexity changes of different signs and amplitudes. Many events that occur after trauma involve changes in tissue architecture and, thus, can potentially alter DKI parameters. Examples of such events include but are not limited to demyelination [38], vascular remodelling [39], axonal changes such as axonal varicosities, cell swelling and secondary axotomy [40], cerebral oedema, inflammation or reactive astrogliosis [23]. Unfortunately, while the aforementioned processes are likely detected by DKI, only reactive astrogliosis has been directly linked experimentally to changes in DKI [23]. The findings of reduced Kr suggest an overall decrease in diffusional heterogeneity stemming from changes in axonal and myelin density in the internal capsule in a manner that does not affect the standard DTI parameters of FA and MD. It should be noted that, although both MD and FA are reduced in these regions, this difference did not make significance among these group of patients. The data show that the response of white matter to trauma as measured by DKI varies between regions. For example, while the anterior limb of the internal capsule exhibits changes across all three time points, the corpus callosum is relatively unchanged from healthy ranges in the mTBI population. Given the varying nature of the injury, these

Longitudinal diffusion kurtosis imaging in mTBI

55

differences may be attributed to the extent to which each of the regions is affected, which may have an influence on the rate at which some of the inflammatory processes are upregulated. For example, it has been argued that small caliber unmyelinated neurons show more diameter changes, density changes and secondary axotomy following TBI than larger caliber myelinated neurons [41]. However, factors other than myelination likely play a role in trauma-related tissue changes including the orientation of fibres to the shearing stress associated with the DAI as well as the proximity of tissue to compromised micro-vessels [42]. Future studies that focus on long-term sequelae following TBI would benefit from histopathological correlation for accurate inferences of the changes in DKI parameters to the underlying changes in pathology. Neurocognitive findings This battery of neurocognitive tests was selected from a subset of the ANAM traumatic brain injury battery [43]. Specifically, these sub-tests assess sustained attention, working memory, spatial processing and computation skills as well as a weighted throughput score, which acts as an overall summary of accuracy and reaction time across the seven subtests [28]. The only sub-test that demonstrates a cognitive deficit in the mTBI population is the code substitution delayed sub-test at the 10 day point. The code substitution delayed sub-test assesses memory as well as sustained attention, which is consistent with previously reported deficits in sustained attention [44] and working memory [45–47]. However, in this patient population, the ANAM may not be sensitive enough to detect the subtle cognitive impairments associated with mTBI. This may be due to the small sample size included in this analysis. For example, using the ANAM in a sample of 41 patients, this group has shown that patients with mTBI at the 10 day and 1 month time point exhibit reduced performance on the weighted throughput score and code substitution sub-test compared to a control group [48]. Interestingly, neither patients at any individual time point nor healthy controls demonstrated any association between DKI parameters and cognitive performance. This is in contrast to previous findings of an association between MK and performance in different neuropsychological domains and cognitive impairment at the more acute stage, which was nonexistent at 1 year following injury [26, 27]. It is possible that person-to-person variability and the small sample size in both baseline DKI parameter values and cognitive ability might have obscured an effect. It should be noted that patients in this study were more acutely observed than those in the other two studies, which might explain the variability in the data at the acute time point. However, as shown in Figure 5, when the changes in DKI parameters were compared with the increases in the cognitive performance of the patients with mTBI, positive associations between regional DKI parameters and changes in cognitive performance were detected. In other words, patients who showed the greatest increase in DKI parameters between the 1 month and 6 month time points demonstrated the greatest improvements in overall cognitive functioning, as assessed by the weighted throughput score of the ANAM. Furthermore, this is true across ROIs including those that showed no group differences compared to the

56

J. A. Stokum et al.

control group (thalamus and splenium of the corpus callosum). The association of neurophysiological changes to the global grey and white matter DKI parameters indicates that de-arrangement of water diffusion environment is rampant and that certain regions are more affected than others. These associations were particularly stronger in the internal capsule regions and suggest that changes in DKI parameters may be sensitive to the changes in functional recovery from injury. Similar regions have been targeted for tissue sparing and functional recovery in animal models [49]. While the data supports a correlation, causality is still unknown. However, the regions in which changes were observed are consistent with previous research on mTBI. For example, the anterior internal capsule contains frontopontine, thalamocortical and corticothalamic bundles whose injury may have partially resulted in cognitive impairment and, hence, any repair in these regions might aid cognitive improvements [50]. Internal capsule injuries have been previously associated with working memory deficits following mTBI [51]. Overall this study suggests the importance of using DKI on patients with mTBI to understand the underlying pathophysiological changes as they complement the well-studied DTI parameters of MD, FA, radial and axial diffusivities and may have an association with the clinical outcomes of patients. This study has a few limitations. First, one potential weakness of this study is the limited number of patients observed and the relatively short duration of followup. Second, while participant age was accounted for in this analysis, this study was not able to account for other factors which might alter white matter structure such as smoking [52]. Third, the healthy control group was imaged at a single time point and does not account for natural change in DKI over time. Finally, other neuropsychological tests which may have been more sensitive to post-traumatic cognitive changes were not performed. Future studies on larger patient populations, combined with more rigorous domain-specific neurocognitive assessments and functional neuroimaging over longer periods of follow-up, would further substantiate the findings in this study.

Conclusions This study demonstrates that mTBI-related brain changes can be observed through 6 months post-injury using DKI. Furthermore, changes in DKI parameters have a strong correlation with improvements in cognitive ability during the recovery from mTBI. The pathological changes detected using traditional diffusion parameters such as MD, FA, radial and axial diffusivity may under-estimate the extent of injury. Therefore, incorporating of DKI methods to also compute MK and Kr may complement water diffusion measurements, thus providing a more comprehensive assessment of the changes in tissue microstructure. Ultimately, this has the potential to lead to a better understanding of the pathophysiological changes associated with the sequelae following mTBI.

Acknowledgements The authors would like to thank Joshua Betz, Jacqueline Janowich and Teodora Stoica for their help with participant

Brain Inj, 2015; 29(1): 47–57

recruitment and George Makris and Steven Roys for their help with acquiring and processing the data.

Declaration of interest The authors report no conflicts of interest. This work was partly supported by the grants W81XWH-08-1-0725 and W81XWH-12-1-0098 from the US Army.

References 1. Centers for Disease Control and Prevention. Traumatic brain injury in the United States: emergency department visits, hospitalizations, and deaths, 2002–2006. Atlanta, GA: CDC. 2010. 2. Centers for Disease Control and Prevention (CDC), National center for injury prevention and control. Report to congress on mild traumatic brain injury in the United States: Steps to prevent a serious public health problem. Atlanta, GA: CDC. 2003. 3. Li J, Li XY, Feng DF, Pan DC. Biomarkers associated with diffuse traumatic axonal injury: Exploring pathogenesis, early diagnosis, and prognosis. The Journal of Trauma 2010;69:1610–1618. 4. Iverson GL, Lovell MR, Smith S, Franzen MD. Prevalence of abnormal CT-scans following mild head injury. Brain Injury 2000; 14:1057–1061. 5. Hughes DG, Jackson A, Mason DL, Berry E, Hollis S, Yates DW. Abnormalities on magnetic resonance imaging seen acutely following mild traumatic brain injury: Correlation with neuropsychological tests and delayed recovery. Neuroradiology 2004;46: 550–558. 6. Niogi SN, Mukherjee P, Ghajar J, Johnson C, Kolster RA, Sarkar R, Lee H, Meeker M, Zimmerman RD, Manley GT, McCandliss BD. Extent of microstructural white matter injury in postconcussive syndrome correlates with impaired cognitive reaction time: A 3T diffusion tensor imaging study of mild traumatic brain injury. American Journal of Neuroradiology 2008;29:967–973. 7. Inglese M, Makani S, Johnson G, Cohen BA, Silver JA, Gonen O, Grossman RI. Diffuse axonal injury in mild traumatic brain injury: A diffusion tensor imaging study. Journal of Neurosurgery 2005; 103:298–303. 8. Kraus MF, Susmaras T, Caughlin BP, Walker CJ, Sweeney JA, Little DM. White matter integrity and cognition in chronic traumatic brain injury: A diffusion tensor imaging study. Brain: A Journal of Neurology 2007;130:2508–2519. 9. Bazarian JJ, Zhong J, Blyth B, Zhu T, Kavcic V, Peterson D. Diffusion tensor imaging detects clinically important axonal damage after mild traumatic brain injury: A pilot study. Journal of Neurotrauma 2007;24:1447–1459. 10. Wilde EA, McCauley SR, Hunter JV, Bigler ED, Chu Z, Wang ZJ, Hanten GR, Troyanskaya M, Yallampalli R, Li X, Chia J, Levin HS. Diffusion tensor imaging of acute mild traumatic brain injury in adolescents. Neurology 2008;70:948–955. 11. Chu Z, Wilde EA, Hunter JV, McCauley SR, Bigler ED, Troyanskaya M, Yallampalli R, Chia JM, Levin HS. Voxel-based analysis of diffusion tensor imaging in mild traumatic brain injury in adolescents. American Journal of Neuroradiology 2010;31: 340–346. 12. Mayer AR, Ling J, Mannell MV, Gasparovic C, Phillips JP, Doezema D, Reichard R, Yeo RA. A prospective diffusion tensor imaging study in mild traumatic brain injury. Neurology 2010;74: 643–650. 13. Benson RR, Meda SA, Vasudevan S, Kou Z, Govindarajan KA, Hanks RA, Millis SR, Makki M, Latif Z, Coplin W, Meythaler J, Haacke EM. Global white matter analysis of diffusion tensor images is predictive of injury severity in traumatic brain injury. Journal of Neurotrauma 2007;24:446–459. 14. Wilde EA, Chu Z, Bigler ED, Hunter JV, Fearing MA, Hanten G, Newsome MR, Scheibel RS, Li X, Levin HS. Diffusion tensor imaging in the corpus callosum in children after moderate to severe traumatic brain injury. Journal of Neurotrauma 2006;23: 1412–1426. 15. Assaf Y, Cohen Y. Non-mono-exponential attenuation of water and N-acetyl aspartate signals due to diffusion in brain tissue. Journal of Magnetic Resonance 1998;131:69–85.

DOI: 10.3109/02699052.2014.947628

16. Niendorf T, Dijkhuizen RM, Norris DG, van Lookeren Campagne M, Nicolay K. Biexponential diffusion attenuation in various states of brain tissue: Implications for diffusion-weighted imaging. Magnetic Resonance in Medicine: Official Journal of the Society of Magnetic Resonance in Medicine/Society of Magnetic Resonance in Medicine 1996;36:847–857. 17. Cheung MM, Hui ES, Chan KC, Helpern JA, Qi L, Wu EX. Does diffusion kurtosis imaging lead to better neural tissue characterization? A rodent brain maturation study. NeuroImage 2009;45: 386–392. 18. Hui ES, Cheung MM, Qi L, Wu EX. Towards better MR characterization of neural tissues using directional diffusion kurtosis analysis. NeuroImage 2008;42:122–134. 19. Wu EX, Cheung MM. MR diffusion kurtosis imaging for neural tissue characterization. NMR in Biomedicine 2010;23:836–848. 20. Falangola MF, Jensen JH, Babb JS, Hu C, Castellanos FX, Di Martino A, Ferris SH, Helpern JA. Age-related non-gaussian diffusion patterns in the prefrontal brain. Journal of Magnetic Resonance Imaging 2008;28:1345–1350. 21. Falangola MF, Jensen JH, Tabesh A, Hu C, Deardorff RL, Babb JS, Ferris S, Helpern JA. Non-gaussian diffusion MRI assessment of brain microstructure in mild cognitive impairment and alzheimer’s disease. Magnetic Resonance Imaging 2013;31:840–846. 22. Gong NJ, Wong CS, Chan CC, Leung LM, Chu YC. Correlations between microstructural alterations and severity of cognitive deficiency in alzheimer’s disease and mild cognitive impairment: A diffusional kurtosis imaging study. Magnetic Resonance Imaging 2013;31:688–694. 23. Zhuo J, Xu S, Proctor JL, Mullins RJ, Simon JZ, Fiskum G, Gullapalli RP. Diffusion kurtosis as an in vivo imaging marker for reactive astrogliosis in traumatic brain injury. NeuroImage 2012;59: 467–477. 24. Raab P, Hattingen E, Franz K, Zanella FE, Lanfermann H. Cerebral gliomas: Diffusional kurtosis imaging analysis of microstructural differences. Radiology 2010;254:876–881. 25. Kamagata K, Tomiyama H, Motoi Y, Kano M, Abe O, Ito K, Shimoji K, Suzuki M, Hori M, Nakanishi A, et al. Diffusional kurtosis imaging of cingulate fibers in parkinson disease: Comparison with conventional diffusion tensor imaging. Magnetic Resonance Imaging 2013;31:1501–1506. 26. Grossman EJ, Ge Y, Jensen JH, Babb JS, Miles L, Reaume J, Silver JM, Grossman RI, Inglese M. Thalamus and cognitive impairment in mild traumatic brain injury: A diffusional kurtosis imaging study. Journal of Neurotrauma 2012;29:2318–2327. 27. Grossman EJ, Jensen JH, Babb JS, Chen Q, Tabesh A, Fieremans E, Xia D, Inglese M, Grossman RI. Cognitive impairment in mild traumatic brain injury: A longitudinal diffusional kurtosis and perfusion imaging study. American Journal of Neuroradiology 2013;34:951, 957, S1–3. 28. Kane RL, Roebuck-Spencer T, Short P, Kabat M, Wilken J. Identifying and monitoring cognitive deficits in clinical populations using automated neuropsychological assessment metrics (ANAM) tests. Archives of Clinical Neuropsychology: The Official Journal of the National Academy of Neuropsychologists 2007;22(Suppl 1): S115–S126. 29. Teasdale GM, Pettigrew LE, Wilson JT, Murray G, Jennett B. Analyzing outcome of treatment of severe head injury: A review and update on advancing the use of the glasgow outcome scale. Journal of Neurotrauma 1998;15:587–597. 30. Jensen JH, Helpern JA, Ramani A, Lu H, Kaczynski K. Diffusional kurtosis imaging: The quantification of non-gaussian water diffusion by means of magnetic resonance imaging. Magnetic Resonance in Medicine: Official Journal of the Society of Magnetic Resonance in Medicine/Society of Magnetic Resonance in Medicine 2005;53:1432–1440. 31. Thorne DR. Throughput: A simple performance index with desirable characteristics. Behavioural Brain Methods 2006;38:569–573. 32. Ivins BJ, Kane R, Schwab KA. Performance on the automated neuropsychological assessment metrics in a nonclinical sample of soldiers screened for mild TBI after returning from iraq and afghanistan: A descriptive analysis. The Journal of Head Trauma Rehabilitation 2009;24:24–31. 33. Reich S, Short P, Kane R, Weiner W, Shulman L, Anderson K. Validation of the ANAM test battery in parkinson’s disease. Ft. Belvoir, VA: Defense Technical Information Center; 2005.

Longitudinal diffusion kurtosis imaging in mTBI

57

34. Grossman EJ, Inglese M, Bammer R. Mild traumatic brain injury: Is diffusion imaging ready for primetime in forensic medicine? Topics in Magnetic Resonance Imaging 2010;21:379–386. 35. Arfanakis K, Haughton VM, Carew JD, Rogers BP, Dempsey RJ, Meyerand ME. Diffusion tensor MR imaging in diffuse axonal injury. American Journal of Neuroradiology 2002;23:794–802. 36. Jensen JH, Helpern JA. MRI quantification of non-gaussian water diffusion by kurtosis analysis. NMR in Biomedicine 2010;23: 698–710. 37. Johnson VE, Stewart W, Smith DH. Axonal pathology in traumatic brain injury. Experimental Neurology 2013;246:35–43. 38. Bramlett HM, Dietrich WD. Quantitative structural changes in white and gray matter 1 year following traumatic brain injury in rats. Acta Neuropathologica 2002;103:607–614. 39. Park E, Bell JD, Siddiq IP, Baker AJ. An analysis of regional microvascular loss and recovery following two grades of fluid percussion trauma: A role for hypoxia-inducible factors in traumatic brain injury. Journal of Cerebral Blood Flow and Metabolism: Official Journal of the International Society of Cerebral Blood Flow and Metabolism 2009;29:575–584. 40. Bigler ED, Maxwell WL. Neuropathology of mild traumatic brain injury: Relationship to neuroimaging findings. Brain Imaging and Behavior 2012;6:108–136. 41. Reeves TM, Smith TL, Williamson JC, Phillips LL. Unmyelinated axons show selective rostrocaudal pathology in the corpus callosum after traumatic brain injury. Journal of Neuropathology and Experimental Neurology 2012;71:198–210. 42. Cloots RJ, van Dommelen JA, Nyberg T, Kleiven S, Geers MG. Micromechanics of diffuse axonal injury: Influence of axonal orientation and anisotropy. Biomechanics and Modeling in Mechanobiology 2011;10:413–422. 43. Vincent AS, Roebuck-Spencer T, Gilliland K, Schlegel R. Automated neuropsychological assessment metrics (v4) traumatic brain injury battery: Military normative data. Military Medicine 2012;177:256–269. 44. Maruishi M, Miyatani M, Nakao T, Muranaka H. Compensatory cortical activation during performance of an attention task by patients with diffuse axonal injury: A functional magnetic resonance imaging study. Journal of Neurology, Neurosurgery & Psychiatry 2007;78:168–173. 45. McAllister TW, Flashman LA, McDonald BC, Saykin AJ. Mechanisms of working memory dysfunction after mild and moderate TBI: Evidence from functional MRI and neurogenetics. Journal of Neurotrauma 2006;23:1450–1467. 46. McDowell S, Whyte J, D’Esposito M. Working memory impairments in traumatic brain injury: Evidence from a dual-task paradigm. Neuropsychologia 1997;35:1341–1353. 47. Miotto EC, Cinalli FZ, Serrao VT, Benute GG, Lucia MC, Scaff M. Cognitive deficits in patients with mild to moderate traumatic brain injury. Arquivos De Neuro-Psiquiatria 2010;68:862–868. 48. Sours C, Rosenberg J, Kane R, Roys S, Zhuo J, Shanmuganathan K, Gullapalli RP. Associations between interhemispheric functional connectivity and the automated neuropsychological assessment metrics (ANAM) in civilian mild TBI. Brain Imaging and Behavior 2014; [Epub ahead of print]. 49. Thompson HJ, Marklund N, LeBold DG, Morales DM, Keck CA, Vinson M, Royo NC, Grundy R, McIntosh TK. Tissue sparing and functional recovery following experimental traumatic brain injury is provided by treatment with an anti-myelin-associated glycoprotein antibody. The European Journal of Neuroscience 2006;24: 3063–3072. 50. Axer H, Keyserlingk DG. Mapping of fiber orientation in human internal capsule by means of polarized light and confocal scanning laser microscopy. Journal of Neuroscience Methods 2000;94: 165–175. 51. Levin HS, Wilde E, Troyanskaya M, Petersen NJ, Scheibel R, Newsome M, Radaideh M, Wu T, Yallampalli R, Chu Z, Li X. Diffusion tensor imaging of mild to moderate blast-related traumatic brain injury and its sequelae. Journal of Neurotrauma 2010;27:683–694. 52. Umene-Nakano W, Yoshimura R, Kakeda S, Watanabe K, Hayashi K, Nishimura J, Takahashi H, Moriya J, Ide S, Ueda I, et al. Abnormal white matter integrity in the corpus callosum among smokers: Tractbased spatial statistics. PloS One 2014 Feb 7; 9(2):e87890.

Copyright of Brain Injury is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

A longitudinal evaluation of diffusion kurtosis imaging in patients with mild traumatic brain injury.

To investigate longitudinal diffusion tensor imaging (DTI) and diffusion kurtosis imaging (DKI) changes in white and grey matter in patients with mild...
1MB Sizes 0 Downloads 13 Views