3.3.5.1 Anatomical location of lesions. Table 3-4 shows the number and percentage
of the 40 patients who were affected by brain contusions (visible on T1-weighted MRI) in each anatomical location and Table 3-5 shows the number and percentage of the 40 patients with lesions indicative of DAI (visible on T2*-weighted MRI) in each white matter region.
Table 3-4
Number/Percentage of Patients with T1 Brain Contusions
Contusions by anatomical location
Right hemisphere Left hemisphere Totals by region
Number Percentage Number Percentage Number Percentage
Frontal 11 27.5% 10 25% 13 32.5%
Temporal 7 17.5% 7 17.5% 10 25%
Parietal
(cortical and subcortical)
- - - - - -
Occipital 4 10% 1 2.5% 4 10%
One or more regions 15 37.5% 13 32.5% 17 43%
Table 3-5
Number/Percentage of Patients with T2* White Matter Lesions
Evidence of DAIa by
anatomical location Right hemisphere Left hemisphere Totals by WM region
Number Percentage Number Percentage Number Percentage
Infratentorial WM/Totals 1 2.5% 3 7.5% 4 10% Brainstem - - 2 5% 2 5% Cerebellum 1 2.5% 1 2.5% 2 5% Deep WM/Totals 6 15% 3 7.5% 7 17.5% Basal Ganglia - - 1 2.5% 1 2.5% Thalamus - - - - Internal Capsule 1 2.5% 1 2.5% 1 2.5% External Capsule 1 2.5% 1 2.5% 2 5% Corpus callosum 3 7.5% 1 2.5% 4 10% Deep/periventricular 2 5% - - 2 5% Lobar WM/Totals
(cortical and subcortical)
12 30% 14 35% 16 40% Frontal 12 30% 10 25% 15 37.5% Temporal 7 7.5% 5 12.5% 8 20% Parietal 5 12.5% 4 10% 8 20% Occipital 2 5% 1 2.5% 2 5% Insula 2 5% 1 2.5% 3 7.5%
One or more WM regions 12 30% 14 35% 16 40%
Notes: aTraumatic microbleeds or other white matter lesions observed on gradient-echo T2*-weighted MRI. Identification
of microbleeds was based on the Microbleed Anatomical Rating Scale (MARS; Gregoire et al., 2009). DAI = diffuse axonal injury. WM = white matter.
3.3.5.2 Anatomical location of lesions and cognitive function. Interrelationships between the anatomical location of brain contusions and scores on the six indices of verbal
learning and memory, executive function and information processing speed were investigated within the patient group. Three specific hypotheses (3a, 3b, and 3c) were tested. The fourth hypothesis (3d) relates to the anatomical location of white matter lesions and patients’ neuropsychological performance.
Hypothesis 3a: Medial temporal, prefrontal, and posterior/medial parietal contusions will be associated with worse verbal learning and memory. Voxelwise relationships between the
presence of brain contusions and the two indices of verbal learning and memory were tested using VLSM. For the People Test (immediate recall) the largest voxel-based t-statistic observed was 3.46 (one-tailed) in the right frontal pole region; this was not significant after FDR correction to adjust for the number of unique t-tests carried out. For Logical Memory I (first recall total) none of the voxel-based t-statistics survived FDR correction for multiple comparisons or approached the threshold value of t ≥ 4.78 at pFDR < .01. The largest t-statistic of 1.85 resulting
from the Logical Memory I first recall total/presence of brain contusions comparsions was observed in a cluster located in the left frontal pole region.
Hypothesis 3b: Frontal and lateral parietal contusions will be associated with worse executive function. Three separate VLSM analyses were carried out to test the voxelwise
relationships between brain contusions and each of the three indices of executive function. Negative t-statistics were expected for the first two measures (TMT-B − TMT-A completion time and the Color-Word Interference index) and positive t-statistics for the third (Verbal Fluency/letter fluency total correct), reflecting the convention of VLSM and the scoring directions for these measures. The only voxel-based t-statistic (one-tailed) to reach significance was that between brain contusions and the TMT alternating-switch cost index (t = −4.64, pFDR < .01). The
voxel (MNI −36, 23, −22) showing the peak t-statistic was part of a small cluster located in the left orbitofrontal cortex. The largest t-statistics for the Color-Word Interference index/brain contusions (t = −0.66, ns) and Verbal Fluency/letter fluency total correct/brain contusions (t = 1.67, ns) comparisons were observed in the left temporal pole region and in the right inferior frontal region, respectively.
pontocerebellar structures will be associated with slower information processing. The largest
voxel-based t-statistic between brain contusions and CRT median reaction time was −4.02 (one-tailed); this was marginally significant after pFDR < .01 adjustment for multiple comparisons.
This small cluster indicating a relationship between the presence of brain contusion and slowed information processing speed was centred at a voxel located in the left orbitofrontal/insular cortex (MNI −28, 13, −21).
Hypothesis 3d: Patients with deep and/or infratentorial white matter lesions will show worse neuropsychological performance than patients with lobar white matter lesions only. For
nine patients, lesions in deep/infratentorial white matter regions were observed on T2*-weighted MRI (some also had lobar lesions), whilst for seven patients only lobar white matter lesions were visible. Deep/infratentorial white matter lesions were only present in patients with moderate/severe TBI, whereas lesions restricted to lobar regions were found in four patients with moderate/severe and three patients with mild TBI. There were no significant differences between the deep/infratentorial vs. lobar white matter lesion subgroups in age, premorbid IQ (WTAR age-scaled scores), current intellectual ability (Similarities and Matrix Reasoning T- scores) or time since injury. Independent samples t-tests found no significant group differences in any of the six neuropsychological indices.
3.3.5.3 Lesion load. The median number of voxels affected by brain contusions
(individual intra-cranial volume-normalised) in the 17 patients with any contusions present was 5024; the range was 162 to 21,251. The median number of microbleeds in the 16 patients with white matter lesions was 9.5, ranging from 1 to 39. Thus, both volume of brain tissue affected by brain contusions and total number of microbleeds varied enormously, with over half the sample showing zero loads for each index.
3.3.5.4 Lesion load and cognitive function.
Hypothesis 4: Higher lesion loads, indexed by total brain contusion volume or the number of microbleeds, will be associated with more severe cognitive impairment in the domains of verbal learning and memory, executive function and information processing speed.
were calculated between normalised brain contusion load (number of voxels) and the six neuropsychological indices. Bonferroni adjustment of the p-value was applied variable-wise to account for the multiple neuropsychological measues (padjusted = .05 ÷ 6 = .008).
After controlling for the effects of severity of injury and time since injury, normalised contusion load was significantly correlated with the TMT alternating-switch cost index (TMT-B − TMT-A), but in the reverse direction from that expected, with greater lesion load associated with better mental flexibility (rhopartial = − .70, p = .008). No other correlations between contusion load
and the neuropsychological measures approached statistical significance.
Nonparametric correlations between the total number of microbleeds and neuropsychological performance were similarly explored, adjusting for the effects of confound variables where these were identified. None were significant.