Clinicotopographical Correlation of Corticospinal Tract Stroke
A Color-Coded Diffusion Tensor Imaging Study
Background— Small capsular strokes are difficult to assess with regard to the precise location and the extent of pyramidal tract damage with conventional brain imaging. Color-coded diffusion tensor imaging (CDTI) provides a means to visualize the course of the corticospinal tract within the white matter. In addition to T2-weighted MRI, diffusion-weighted MRI and CDTI were used to analyze the topographical patterns of small lacunar corticospinal tract strokes.
Methods— We examined 15 patients with pyramidal tract strokes in the subacute phase (days 3 to 7). Lesions were identified on diffusion-weighted MRI and superimposed on CDTI images. The anatomic location and pattern of the lesion were visualized on CDTI with regard to the corticospinal tract and subsequently compared with the clinical presentation. In addition, infarct areas were evaluated with quantitative parameters: mean diffusivity and lattice anisotropy index of lesions were determined.
Results— We identified 5 different patterns of corticospinal tract stroke falling into 2 clinical subgroups: (1) those with marked deficits and minor improvement (6/15) and (2) those with good recovery (9/15). Group 1 had long lesions centered in the pyramidal tract, involving the basal ganglia (anterior choroidal artery); group 2 lesions were very small and/or located anteriorly and medially (periventricular anterior choroidal artery territory; thalamogeniculate, tuberothalamic, and lateral striate branches). Lesions showed a significant increase of mean diffusivity and decrease of lattice anisotropy.
Conclusion— CDTI allows in vivo differentiation of distinct subcortical stroke subtypes. Improved anatomic definition of lesion localization using CDTI may help in better establishing the prognosis for patients after subcortical stroke.
Lacunar infarctions represent 20% to 25% of all strokes. They are caused by occlusion of the penetrating arteries with a diameter of 40 to 900 μm. The lesions are characteristically small, lack both extensive edema and mass effect, and are therefore usually not detected on early CT. Conventional MRI may not reliably identify the acute lacunar infarction related to the clinical symptoms since most patients with lacunar infarctions also show chronic white matter lesions whose signal characteristics do not differ from acute lesions. Diffusion weighted MRI (DWI) has been shown to detect the small acute lesions and to be valuable in the evaluation of patients with lacunar symptoms.1 However on DWI, as on T2-weighted images, it is difficult to assess the precise location of the lesion with respect to main motor or sensory pathways.
Diffusion tensor imaging (DTI) offers new parameters to assess the location and extent of focal brain damage, as it can visualize the main fiber bundles (eg, corticospinal tracts) and provides information on tissue integrity.2 In addition to isotropic scalar diffusion parameters like the mean diffusivity See Editorial Comment, page 92 (MD), calculation of the diffusion tensor provides quantitative information on directionality and the degree of anisotropy,3 both reflecting tissue integrity. Reductions of anisotropy (eg, on lattice anisotropy index [LAI] maps) and an increase of MD indicate edematous tissue change or loss of local tissue components (eg, myelin, axons).2 By using color coding with symmetrical RGB (red-green-blue) color axes, the direction of the main fiber bundles can be delineated by the different colors and the degree of anisotropy is encoded by color brightness (see Figure 1).4
See Editorial Comment, page 92
The goal of this study was to use CDTI for a more accurate visualization of the acute ischemic lesions with respect to the corticospinal tract and correlate imaging and clinical findings in patients with capsular or pericapsular strokes.
Subjects and Methods
Fifteen consecutive patients (10 men, 5 women; mean age, 69±7.8 years) admitted to our stroke unit with acute lacunar subcortical stroke in conventional MRI were included. Clinical neurological deficits were scored by 2 neurologists using the National Institutes of Health Stroke Scale (NIHSS) on admission and on discharge. Mean duration of inpatient treatment was 10 days (range, 5 to 21 days). Routine stroke diagnostic investigations included blood analysis, ECG monitoring, echocardiography, and Doppler ultrasound investigations. MRI was performed after informed written consent by each patient was given. The study was approved by the local ethics committee.
The DTI study was performed on day 3 to 7 (mean, day 5) after admission. As one part of the MRI protocol, standardized T2-weighted MRI and DTI were performed with 1.5 T Magnetom Vision and Sonata systems (Siemens). After a careful localizing procedure (transverse, coronal, and sagittal localizing sequences) to provide symmetrical anatomic landmarks and identical T2-weighted and DTI slices, transverse oblique contiguous images (slice thickness, 5 mm; field of view, 240 mm2) aligned with the inferior borders of the corpus callosum were acquired: (1) T2-weighted images (turbo spin echo, repetition time [TR], 2620 ms; echo time [TE], 85 ms; 192×256 matrix). (2) The DTI sequence scheme consisted of a single-shot DW fluid attenuated inversion recovery–spin echo–echo planar imaging acquisition (TR/TI/TE=6000/2000/110 ms; slice thickness, 5 mm; field of view, 240×240 mm2; matrix=128×128, interpolation, 256×256) containing gradient lobes for DW (b=0, 1030 s/mm2). Six diffusion directions—(1,1,0), (1,−1,0), (0,1,1), (0,1,−1), (1,0,1), and (1,0,−1) were acquired. Amplitude images were averaged from 8 measurements.
Distortions due to residual eddy currents were corrected afterward.5 The diffusion tensor was determined according to Basser and Pierpaoli.6 After tensor diagonalization and calculating the eigenvectors, maps of isotropic DWI (b=1030 s/mm2), the trace (D) (with trace(D)/3=MD), and the LAI were calculated. On directional DT maps, the direction of the eigenvector associated with the largest eigenvalue was color-coded using an RGB color model with symmetrical color axes. The color-coordinated system was adapted to the individual orientation of patient heads in space (color directions: green=anterior-posterior [parallel to interhemispheric line], red=left-right, blue=head-feet). The degree of anisotropy (LAI) was encoded by color brightness.
Segmentation of Infarct Lesions and Display on CDTI Maps
To correlate infarct location as assessed on isotropic DWI with the exact anatomic location on CDTI, an in-house–developed software tool based on the data processing and visualization system Khoros 2.0 (University of New Mexico) was used. This allowed (1) segmentation of the infarcted areas on isotropic DWI maps and (2) automatic display of lesions identified on isotropic DWI maps on corresponding CDTI maps (Figure 1). There were no problems as they may arise from misregistration, as CDTI and isotropic DWI were not registered but originated from the same data set.
Evaluation of Quantitative Parameters
Mean values of quantitative parameters (LAI, MD) were determined for the acute ischemic lesion volumes in each patient by region-of-interest (ROI) analysis. Lesions were mirrored onto the healthy contralateral hemisphere, while mirror axes were manually adapted for asymmetries in 3 patients. Calculation of mean values of MD and LAI in stroke lesions and healthy tissue of the homologous brain structures was performed. MD and LAI of healthy tissue and lesions were statistically compared by Student’s t test (Figure 1).
The highly defined delineation of subcortical anatomy on CDTI maps allowed us to specify the lesion location in each individual, accurately relating the lesion to adjacent anatomic structures (eg, thalamus, corticospinal tract). After the different lesion topographies were compared visually, 5 different infarct configurations were distinguished with regard to lesion topography and extent (patterns A through E). Two cases could not be assigned to either type and therefore were considered individually.
No a priori hypothesis was generated as to the total number of different stroke patterns. In a post hoc comparison of the infarct topographies with presumed maps of arterial supply, infarct patterns were found to have a high consistency with arterial territories of the deep penetrating arteries as described by Bogousslavsky et al.7
Clinical scoring was performed without imaging data evaluation. When correlating clinical outcome parameters (NIHSS at admission and on discharge) with the infarct patterns on CDTI maps, patients with stroke pattern A were found to differ markedly in outcome parameters compared with the remainder of patients. Therefore, 2 main clinical groups were distinguished: group 1 consisted of patients with marked deficits and relatively minor improvement (6/15); group 2 included patients with good recovery or minor motor dysfunction (9/15).
Group 1 patients were characterized by long lesions centered in the pyramidal tract involving the basal ganglia (consistent with stroke pattern A).
Infarcts (anterior choroidal artery territory) extended into the craniocaudal axis, involving the medial dorsal half of the pyramidal tract, passing through the middle third of the posterior limb of the internal capsule (IC) and finally reaching the lateral border in the caudal planes. Stroke lesions affected the thalamus at lower levels and the dorsal globus pallidus at higher levels.
Periventricular area of anterior choroidal artery territory: lesions lay within the dorsal half or at the medial border of the pyramidal tract.
Thalamogeniculate branches: those lesions were located medial to the dorsal third of the posterior limb of the IC. Lesions were small and affected the lateral thalamus.
Tuberothalamic branches: lesions were located in the anterior thalamus with involvement of the genu and the medial border of the 2 anterior thirds of the posterior limb of the IC.
Lateral striate artery territory: lesions affected the anterior limb of the internal capsule, with infarction of the head of the caudate and the putamen.
Clinical manifestation in pattern A (anterior choroidal artery territory, 6 patients) often consisted of dysarthria combined with a brachiofacial pure motor hemiparesis (see Table 1). Motor impairment was marked (mean NIHSS 5), sensory disturbance was encountered in only 1 patient. Patients recovered only partially over the observation period (mean NIHSS 4).
Pattern B (anterior choroidal artery territory, 3 patients) was clinically characterized by a transient brachiofacial hemiparesis with complete recovery (NIHSS 0). Sensory deficits affected the leg transiently in 2 patients.
Pattern C patients (thalamogeniculate artery territory, 4 patients) had mainly sensory deficits with paraesthesias in the distal upper extremity. Discrete motor weakness was observed in 1 patient. In all patients the clinical outcome was favorable, with minor persisting sensory deficits (NIHSS 1).
Pattern D was encountered in 1 patient (lateral striate artery territory) with severe left-sided brachiofacial sensory motor hemiparesis with dysarthria and abulia. This patient was treated acutely with intravenous recombinant tissue plasminogen activator and speech and sensory motor recovery was complete (NIHSS 0), while the abulic syndrome remained.
Pattern E (tuberothalamic artery territory) was noted in 1 patient with central facial weakness. Recovery was complete (NIHSS 0) (Table 1).
After exclusion of an embolic source and severe atherosclerosis, the majority of strokes were diagnosed as probably the result of small-vessel disease (73%). Four patients (27%) had evidence consistent with cardiogenic embolization (2 persistent foramen ovale, 1 atrial fibrillation, 1 atrial septal aneurysm). Mild to extensive chronic T2 hyperintense lesions suggesting small-vessel disease were encountered in 14 of 15 patients (94%). One patient had no signs of chronic white matter lesions and probable cardioembolism was diagnosed on the basis of atrial fibrillation. Most of the patients showed a high vascular risk profile (73% arterial hypertension, 53% diabetes, 47% nicotine, 47% hyperlipidemia, 2% hyperhomocysteinemia (see Table 3), similar to previous epidemiological data7,8 (Table 2).
Evaluation of Quantitative Parameters
Infarcted regions were easily differentiated from healthy tissue on MD maps and LAI maps (Table 3). Impaired tissue integrity was indicated by an increase of MD (mean 1.58±0.23×10−3 mm2/s) and decrease of LAI (mean 0.22±0.030) in acute lesions (P<0.0001). Quantitatively, changes of MD were more pronounced (+76% in infarcted tissue as compared with homologous healthy areas) than changes of LAI (−38% reduction of LAI in infarcted brain regions as compared with homologous healthy areas).
This study demonstrates the potential of CDTI to delineate distinct subcortical stroke subtypes. It was encouraging that the anatomic description of lesions as identified on CDTI correlated with clinical features in these patients, in that the evaluation of clinical parameters, which was performed without the imaging data, paralleled the different infarct patterns identified on CDTI. This suggests that CDTI may be useful to estimate the prognosis in patients with subcortical stroke, although larger patient cohorts are needed to confirm and extend these findings. Furthermore, for future studies it might be useful to include automated analysis algorithms to correlate anatomic/vascular templates to CDTI maps of individual lesion configurations to avoid potential bias from reviewers. It is evident that CDTI improves the delineation of the anatomic involvement of subcortical structures, providing a new framework for future studies.
Although a number of well-defined clinical pictures have been described for lacunar stroke syndromes, visualization of exact lesion location has been limited with conventional neuroimaging.9 Autopsy studies have supported the classical view of a homunculus in anterior-posterior direction as a somatotopic principle of the functional organization of the IC: the posterior part of the posterior limb of the IC was considered to supply motor pathways for the lower limbs. Lesions of genu and anterior limb of the IC were assumed to cause partial hemiparesis with prominent facial involvement and lesser involvement of the lower extremities.10 This notion has been supported in part by CT findings, although a correlation of the clinical picture with the assumed anatomic organization was rather weak. The range of clinical syndromes was much wider than the assumed anatomic basis would have allowed it to be.11 This is not surprising in view of the frequently small size of lesions and the lower contrast of ischemic lesions on CT compared with DWI.12
Besides this limitation with regard to the lower lesion conspicuity, conventional imaging allows localization of the pyramidal tracts only indirectly with the help of few anatomic landmarks (eg, thalamus, basal ganglia, caudate). For this reason, previous studies have also been limited to assessment of the IC on 1 to 2 axial slices, which is another limitation that can be overcome with CDTI, which allowed assessment of the corticospinal tract along its craniocaudal course. Single focal lacunar small penetrator lesions were shown to occasionally affect the corticospinal tract at various levels along its craniocaudal course: in cranial parts a focal lesion may affect the medial border of the posterior third of the corona radiata, while the same lesion passes in lower planes right through the pyramidal tract, to finally reach in caudal planes the most lateral border of the IC (Figure 2). DTI allowed distinction between lacunar stroke of the corticospinal tract with or without involvement of adjacent structures. Affected adjacent anatomic structures could be precisely determined on CDTI maps, which may be important with regard to the clinical syndrome.
Traditionally the motor system was considered to be based on hierarchical organizational principles, with the primary motor cortex as the executing instance, supported by preparative centers such as the supplementary motor area (SMA) and the premotor cortex. Fries et al13 showed in tractographic studies in macaque monkeys projections of the SMA and the premotor cortex to pass through different segments of the IC: axons of the premotor cortex (dorsolateral and postarcuate area 6) passed through the genu of the IC, and those of the SMA (medial area 6) through the anterior limb of IC. Axons of primary motor cortex were found to pass through the middle third of the posterior limb of the IC. With CDTI-based data we can provide in vivo support for the results of Fries at al,13 who described the fibers of the SMA (area 6) as passing almost horizontally in the anterior limb to join the primary motor cortex fibers at the genu of the IC: in CDTI maps, the anterior limb is always encoded green (see Figure 1), indicating fiber course in the anterior-posterior direction, which is in line with the observations of Fries et al. On the basis of these findings, they suggested that the motor system operated in parallel ways rather than in a strictly hierarchical way. In view of this anatomic basis, small localized lesions as found in lacunar strokes could indeed selectively cause disruption of only 1 of the output branches of the 3 different cortical motor areas. Therefore, the clinical observation of frequently excellent motor recovery from localized lesions may be hypothesized as a compensatory phenomenon effected by the 2 remaining, unaffected motor pathways. However, mechanisms of reorganization and functional regeneration are complex, and a favorable outcome of motor deficits after lacunar subcortical strokes may also be explained by other factors (eg, local effects of transient perifocal edema of small lesions in “eloquent” brain regions).
Diffusion changes have been shown to evolve dynamically in the early phase after stroke.14 In this series, MRI was performed in the subacute stage of stroke (days 3 to 7); MD was increased in all cases and no reductions of the apparent diffusion coefficient (signs of cytotoxic cell swelling) were noted, indicating that some vasogenic edema and/or structural damage had already taken place in the lesions. The reductions of the LAI values was in line with this. Given the small size of the lesions, prone to partial volume effects, we would hypothesize that a quantitative evaluation of MD and LAI as measures of the severity of tissue destruction may be limited with regard to clinical correlations.
Given the encouraging results of this study, further refinement of DTI resolution and tractography demonstrating different components of subcortical fiber bundles may well confirm in vivo the traditional concepts of somatotopic organization of the cerebral motor pathways, allowing us to gain insight into the pathophysiology of lacunar stroke syndromes and the contributing factors to recovery. Current technology of clinically used MR systems provides the means for sequences and data postprocessing for the approach used in this study and may therefore become quickly available as an additional diagnostic tool in selected patients.
- Received May 5, 2003.
- Revision received September 16, 2003.
- Accepted September 22, 2003.
Singer MB, Chong J, Lu D, Schonewille WJ, Tuhrim S, Atlas SW. Diffusion-weighted MRI in acute subcortical infarction. Stroke. 1998; 29: 133–136.
Horsfield MA, Jones DK. Applications of diffusion-weighted and diffusion tensor MRI to white matter diseases: a review. NMR Biomed. 2002; 14: 570–577.
Le Bihan D, van Zijl P. From the diffusion coefficient to the diffusion tensor. NMR Biomed. 2002; 15: 341–434.
Bogousslavsky J. Ischemic topographic subtypes. In: Ginsberg MD, Bogousslavsky J, eds. Cerebrovascular Disease II: Pathophysiology, Diagnosis, and Management. Malden, Mass: Blackwell Science; 1998: 924–925.
Mohr JP, Caplan LR, Melski JW, Goldstein RJ, Duncan GW, Kistler JP, Pessin, Bleich HL. The Harvard Cooperative Stroke Registry: a prospective registry. Neurology. 1978; 28: 754–762.
Fisher CM. Lacunar strokes and infarcts: a review. Neurology. 1982; 32: 871–876.
Mohr JP. Lacunes. Stroke. 1982; 13: 3–11.Review.
Chamorro A, Sacco RL, Mohr JP, Foulkes MA, Kase CS, Tatemichi TK, Wolf PA, Price TR, Hier DB. Clinical-computed tomographic correlations of lacunar infarction in the Stroke Data Bank. Stroke. 1991; 22: 175–181.
Fiebach JB, Schellinger PD, Jansen O, Meyer M, Wilde P, Bender J, Schramm P, Juttler E, Oehler J, Hartmann M, et al. CT and diffusion-weighted MR imaging in randomized order: diffusion-weighted imaging results in higher accuracy and lower interrater variability in the diagnosis of hyperacute ischemic stroke. Stroke. 2002; 33: 2206–2210.
Fries W, Danek A, Scheidtmann K, Hamburger C. Motor recovery following capsular stroke: role of descending pathways from multiple motor areas. Brain. 1993; 116 (pt 2): 369–382.
Schlaug G, Siewert B, Benfield A, Edelman RR, Warach S. Time course of the apparent diffusion coefficient (ADC) abnormality in human stroke. Neurology. 1997; 49: 113–119.