| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Neuroradiology |
1 From the Department of Electrical Engineering, National Taiwan University, Taipei, R.O.C. (Y.J.L., H.W.C., I.J.H.); Department of Radiology, Tri-Service General Hospital and National Defense Medical Center, Number 325, Section 2, Cheng-Kung Rd, Neihu 114, Taipei, Taiwan, R.O.C. (Y.J.L., C.Y.C., H.W.C., C.S.L., S.C.C.); and Institute of Statistics Science, Academia Sinica, Taipei, Taiwan, R.O.C. (M.L.). Received May 31, 2002; revision requested July 29; final revision received April 11, 2003; accepted May 14. Supported in part by National Science Council grants NSC-89–2320-B-016–056-M08 and NSC-90–2213-E-002–003. Address correspondence to C.Y.C. (e-mail: sandy0928@seed.net.tw).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Twenty-five patients with ischemic injuries of middle cerebral arterial territories and receiving only conservative treatments were classified into territorial infarction (TI) (n = 16) and watershed infarction (WI) (n = 9) groups and were prospectively evaluated with longitudinal magnetic resonance (MR) examinations. Each patient underwent as many as five MR studies at various stroke stages following stroke symptom onset. Dynamic susceptibility-weighted contrast material–enhanced MR imaging was performed to yield the relative CBV (rCBV). Chemical shift imaging was used to measure the relative levels of N-acetylaspartate (NAA) and lactate of the ischemic brain tissue. Repeated-measures analysis of variance was used to examine the statistical significance in evolutional differences between TI and WI.
RESULTS: For patients with TI, rCBV followed a progressively increasing pattern, from initial low values (0.46 ± 0.28 [SD]) to peak high values (1.23 ± 0.34) at early chronic stage. Relative NAA level decreased to 0.40 ± 0.24 during acute stroke and was lost completely 4 days after ictus. Patients with WI showed consistently high rCBV throughout all stages, with residual relative NAA level (0.53 ± 0.25) even at 1 month after symptom onset. Relative lactate level of patients with TI was significantly higher than that of patients with WI at the acute stage (P < .01). Differences in the temporal changes of both rCBV and brain metabolites between TI and WI were significant (P < .01).
CONCLUSION: The different temporal patterns for stroke progression in TI and WI are associated with different evolutions of hemodynamics and neuronal injury.
© RSNA, 2003
Index terms: Brain, infarction, 13.4352, 13.781 • Brain, MR, 13.121413, 13.121414, 13.121416, 13.12143, 13.12144, 13.12145 • Magnetic resonance (MR), perfusion study, 13.12144
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
As a continuation of a previous investigation about ADC evolution in human subjects, this study aims at testing the hypothesis that the difference between the evolution of WI and that of TI is related to different time courses of hemodynamic changes and neuronal injury. Thus, the purpose of our study was to determine the temporal patterns of neuronal injury between infarction subtypes and to examine their possible association with changes in the cerebral blood volume (CBV).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
To observe the evolution of cerebral ischemic infarction, as many as five MR measurements were performed for each patient at the following six stages of stroke: hyperacute (6 or fewer hours after symptom onset), acute (more than 6 hours and 2 or fewer days after symptom onset), early subacute (3–4 days after symptom onset), late subacute (7–9 days after symptom onset), early chronic (10–15 days after symptom onset), and late chronic (30–31 days after symptom onset). There were a total of 92 MR examinations.
Classification of infarction subtypes was first determined by an experienced radiologist (C.Y.C.) from diffusion-weighted images by using existing atlas templates (3,11). TI was defined as the presence of an ischemic lesion that involved both the cortical and subcortical areas and was restricted to the unilateral middle cerebral arterial territory. WI was defined as the presence of an ischemic lesion in the subcortical white matter within the border zones between the superficial branches and the deep perforators of the same middle cerebral artery (7). The subtype classification of the patients was further confirmed with MR angiographic findings at their first MR examinations.
MR Imaging Protocols
All MR images were acquired at 1.5 T (Magnetom Vision Plus system; Siemens, Erlangen, Germany). In addition to diffusion-weighted echo-planar imaging as a routine stroke screening protocol (4,700/120 [repetition time msec/echo time msec], 256 x 256 matrix, 230-mm field of view, 5-mm section thickness, 1.5-mm gap between sections, 20 sections in one acquisition, and diffusion-weighted factors b of 0 and 1,000 sec/mm2 at three orthogonal gradient axes), each patient underwent dynamic susceptibility-weighted contrast material–enhanced perfusion-weighted imaging at all examination periods. For this purpose, after a manual bolus injection of 0.2 mmol per kilogram of body weight gadopentetate dimeglumine (Magnevist; Schering AG, Berlin, Germany) at about 4–5 mL/sec through a 20-gauge venous catheter inserted into the antecubital vein, echo-planar imaging was performed to acquire a series of perfusion-weighted images (44-msec echo time, 230-mm field of view, 128 x 128 matrix, 5-mm section thickness, three to six sections to cover the ischemic regions, 60–75 frames per section at 1-second interval). Images were transferred to a personal computer for pixel-by-pixel analysis of the CBV. T2-weighted images with cerebrospinal fluid suppression were also obtained by using a fluid-attenuated inversion-recovery technique (9,000/110; inversion time, 2,500 msec) to assess the final infarct size.
Information about the temporal evolution of neuronal injury was obtained by using hydrogen 1 chemical shift imaging performed at predefined intervals, as previously described. A point-resolved spectroscopic technique was used (1,500/270, 16-cm field of view, 16 x 16 phase encodings, 10-mm section thickness, 1-kHz spectral width). The acquisition time for the chemical shift imaging sequence was about 13 minutes, with two signals acquired. Shimming was performed before imaging to ensure field homogeneity with full width at half maximum of the spectral peaks below 10 Hz. A volume of interest encompassing both lesion and normal areas was selected first by using three orthogonal localization images and was properly positioned to match closely the diffusion-weighted image that covered the greatest extent of the lesion. This position was confirmed by one of the radiologists (C.Y.C. or S.C.C.) before execution. Depending on lesion size, the volume of interest contained about 8 x 8 voxels, each of which was about 10 x 10 x 10 mm3 in volume and zero filled to 5 x 5 x 10 mm3 for subsequent spectral analysis. The MR protocols for this ischemic stroke study were approved by the hospital review board.
Data Processing
The relative CBV (rCBV) maps were derived from the dynamically acquired perfusion-weighted images with analysis software (Matlab; MathWorks, Natick, Mass). The signal-time curve of each pixel was first smoothed with a third-order low-pass Butterworth filter at a cutoff frequency of 0.15 Hz to reduce interframe noise interference. The concentration of the contrast agent was then computed by taking the negative logarithm of the time signal after subtraction of the precontrast baseline intensity (ie, concentration proportional to changes in the apparent transverse relaxation rate 1/T2*) (12). Thus, one concentration-time curve was obtained for each pixel. Subsequently, gamma-variate fitting was applied to the concentration-time curve to remove recirculation effects. Pixel rCBV was then calculated by integrating the area under the concentration-time curve. A region of interest was drawn manually (Y.J.L.) and confirmed by a radiologist (C.Y.C.) to include the infarction region to represent the ischemic lesion. Determination of the region of interest for rCBV analysis was based on diffusion-weighted images with a b of 1,000 sec/mm2 at the acute stages because of the conspicuous contrast between lesion and normal areas. At later stages of stroke, regions of interest were encircled from diffusion-weighted images with a b of 0 sec/mm2 (intrinsically T2 weighted). In our experience, this procedure provides robust measurements that substantially reduce the variance of the intra- and interoperator definition of the infarct core that results from the intrinsic heterogeneity of signal intensities in ischemic lesions (7). Another same-sized region of interest homologous to the lesion was drawn to obtain the rCBV of the contralateral normal tissue. Lesion-to-rCBV ratio, defined as the ratio of the mean rCBV of lesions to the mean rCBV of normal areas, was used as an indicator of possible hemodynamic abnormality. Since the main target in our study was the regulatory vasodilation following the onset of stroke symptoms, only rCBV data were reported in this study, even though other perfusion parameters, such as time to peak gadolinium concentration, were routinely used in diagnosis.
Postprocessing of multivoxel MR spectra, including phase and baseline corrections, was performed by using the software installed on the operating console (Numaris; Siemens, Erlangen, Germany). The spectral peaks corresponding to N-acetylaspartate (NAA) and lactate were specifically examined to represent the extent of neuronal injury and oxidative metabolism, respectively. Relative concentrations of NAA were obtained by calculating the peak area of NAA after Gaussian curve fitting of the spectral peak at 2.02 ppm. For this purpose, one to three contiguous voxels in the core region of infarction were selected (Y.J.L.) according to lesion size and confirmed by a radiologist (C.Y.C.), and peak areas of NAA were recorded. NAA of normal brain tissue was also obtained from three contiguous voxels at the contralateral normal hemisphere. The lesion-to-relative NAA ratio was then determined as the mean NAA peak areas in the infarction core divided by the contralateral normal NAA. In a similar manner, the relative concentration of lactate was obtained from the same voxels in the lesion as the peak area of the spectral doublet centered at about 1.33 ppm. The lesion-to-relative lactate ratio was determined as the mean lactate peak area in the infarction core divided by the contralateral normal NAA. Note that the NAA area was used as the denominator because of the absence of lactate in the normal brain parenchyma.
Statistical Analysis
The repeated-measures analysis of variance (ANOVA) was used to examine the differences of neuronal loss and CBV between patients with TI and those with WI, as well as among MR examinations performed at various stages of stroke. Post hoc comparisons between patients with TI and those with WI across evolutionary stages of MR examinations were performed by using the Tukey wholly significant difference tests (13), which are known to give more accurate experiment-wise type I error rates (ie, the probability that a set of multiple tests contains at least one type I error). Findings of empirical studies also showed that the wholly significant difference test is robust to the homogeneity of variance assumption and becomes slightly conservative, with unequal sample sizes having positive correlations of sample size and variance (14,15). P < .01 was considered to represent a significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Our longitudinal investigation about both the hemodynamics and the metabolic status in which human subjects with natural stroke progression were studied provides a consolidating evidence for the subtype-dependent temporal evolutions as stated previously. In addition to ADC alterations demonstrated previously (7), the results of our study further indicate that CBV and neuronal injury are also evolutionally dependent on infarction subtypes. In particular, the neuronal injury, as represented by changes in relative NAA, is slower in WI than in TI, which is similar to and consistent with the behavior of ADC evolution (7).
The temporal profiles of rCBV for the two distinct subtypes of infarctions revealed important messages regarding the mechanisms of stroke evolution. For patients with TI, rCBV follows a progressively increasing pattern, from lower values at the hyperacute stage to higher than normal values at the early chronic stage. The reduced rCBV in the hyperacute stage is obviously the consequence of a severe vascular compromise from the abrupt artery occlusion. Reperfusion starts after the acute stage, with maximum rCBV higher than that of the normal brain tissue, which is reached at about the late subacute stage. The early reperfusion is accompanied by abnormal vasodilatation (increased rCBV) and decreased oxygen extraction fraction (nonnutritional 
luxury perfusion
), which was suggested as postischemic rebound of cellular metabolism in structurally preserved tissue (16,17).
In contrast to TI, the temporal profile of rCBV in patients with WI shows a sustained high value in the ischemic tissue throughout the six stages investigated in this study. Since WI has classically been thought to arise from a critical reduction in cerebral blood supply (ie, a hemodynamic mechanism) due to previous existence of vessel stenosis (1,8), an increase in rCBV is anticipated to reflect autoregulatory vasodilation in order to reduce vascular resistance in response to decreased cerebral perfusion pressure (18). Furthermore, the relatively stable rCBV values seen in patients with TI suggested that maximum compensatory vasodilation may have already existed in the ischemic tissues at the early stages of symptom onset. This preexisting vasodilation could additionally reduce the vascular response to extrinsic stimuli such as acidosis (19), resulting in the absence of significant temporal changes in rCBV.
The distinct temporal changes in brain metabolites in TI and WI may also imply different pathogenetic mechanisms between these two infarction subtypes. The lower level of relative lactate indicates that oxidative metabolism might have failed only partially in patients with WI at the acute stage. This would suggest unexhausted oxygen supplied by increased oxygen extraction along with an increased rCBV that maintains oxidative metabolism and brain function when cerebral perfusion pressure is reduced (20). Patients with TI, on the other hand, exhibit a substantially higher relative lactate level than do those with WI during the acute stage, which reflects anaerobic glycolysis as the main source of energy metabolism. This metabolic difference might account in part for the differing ADC evolutions between TI and WI reported in a previous article (7). The decrease of relative lactate in patients with TI at subsequent stages may be attributed to increased perfusion, which facilitates lactate removal from the damaged tissue into the bloodstream through passive diffusion in the presence of blood-brain barrier disruption (21,22). In other words, evolution of relative lactate for the two subtypes of infarction possibly reflects a consequence of different underlying hemodynamics. Alternatively, the reduced demand of cerebral metabolism with complete loss of living neurons at about 4 days after symptom onset could be another reason for the reduced relative lactate level in patients with TI.
The progress of neuronal injury represented by a decline of relative NAA is slower in WI than in TI, which is consistent with previous findings regarding infarction subtype-dependent ADC evolution (7). Results of our study indicate complete neuronal loss in TI at about 4 days following the onset of symptoms. The significant presence of inefficient anaerobic glycolysis in TI during acute stages is possibly the major origin for an eventual energy failure, which leads to the complete loss of living neurons (23–25). In patients with WI, however, residual relative NAA of 0.53 ± 0.25 was found during the late chronic stage, which suggests neuron survival even after 1 month after ictus. With preexisting vascular autoregulation plus the elevated oxygen extraction fraction as mentioned earlier for WI, neuron survival is more likely to result (26). In addition, a clinical investigation regarding hyperacute ischemic stroke suggests that an increased rCBV likely corresponds to preexisting compensatory vasodilation caused by collateral circulation, which leads to generally better clinical outcomes (5). The slower relative NAA decline depicted for WI is in good agreement with this finding, further supporting the hypothesis that the subtype-dependent progression of neuronal injury is associated with the pathophysiologic hemodynamics.
An integration of the aforementioned information obtained in our study has allowed us to construct a model for the pathophysiologic time course of human ischemic stroke. As demonstrated in Figure 6, WI is likely to arise from a preexisting reduction in cerebral perfusion pressure due to arterial stenosis. The cerebral autoregulation fulfills metabolic demands of the living brain tissues with an increased rCBV through compensatory vasodilation (ie, an opening of the perfusion reserve). Possible elevated oxygen extraction fraction further protects the neurons from experiencing energy deficiency. Overall, the factors lead to a generally slower temporal stroke progression following the critical hypoperfusion that causes cerebral ischemia. On the other hand, the thromboembolic nature of TI causes abrupt cerebral ischemia without a chance to develop sufficient collateral arteries during the critical period. Subsequent alterations in rCBV are deemed ineffective. As a result, complete loss of living neurons occurs at as early as 4 days after ischemic attack. Our observations may implicate that the previous ineffective trials regarding the treatment of ischemic stroke with neuroprotective agents can be reevaluated in the WI patient group, considering the long evolution of stroke (27).
|
Partial volume effects could be a potential source of error for the spectroscopic results, especially considering that WI lesions are often smaller than TI lesions. In this study, partial volume effects were minimized with careful control of the anatomic location from which the spectra were acquired. Since all the infarct volumes reported in this study were larger than the voxel size (acquisition volume, 10 x 10 x 10 mm3) at chemical shift imaging, and zero filling was applied to further reduce the chemical shift imaging voxel volume to 5 x 5 x 10 mm3 for analysis, biasing of the results is anticipated to be minimal.
In conclusion, findings of our longitudinal study show clearly distinct temporal patterns in both the hemodynamics and the metabolism for two types of infarctions: TI and WI. We believe that the most important factor determining the speed of progression lies in the pathophysiologic hemodynamics, an understanding of which may have important implications in future patient treatments.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, C.Y.C.; study concepts, C.Y.C., H.W.C., C.S.L.; study design, C.Y.C., H.W.C.; literature research, Y.J.L., I.J.H.; clinical studies, C.Y.C., S.C.C., Y.J.L., H.W.C., C.S.L.; data acquisition, Y.J.L.,. H.W.C.; data analysis/interpretation, Y.J.L., H.W.C., C.Y.C., C.S.L.; statistical analysis, M.L.; manuscript preparation, Y.J.L., H.W.C., C.S.L.; manuscript definition of intellectual content and editing, H.W.C., C.Y.C.; manuscript revision/review, H.W.C., C.Y.C., C.S.L.; manuscript final version approval, C.Y.C.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  V. Cvoro, J. M. Wardlaw, I. Marshall, P. A. Armitage, C. S. Rivers, M. E. Bastin, T. K. Carpenter, K. Wartolowska, A. J. Farrall, and M. S. Dennis Associations Between Diffusion and Perfusion Parameters, N-Acetyl Aspartate, and Lactate in Acute Ischemic Stroke Stroke, March 1, 2009; 40(3): 767 - 772. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.M. Provenzale, K. Shah, U. Patel, and D.C. McCrory Systematic Review of CT and MR Perfusion Imaging for Assessment of Acute Cerebrovascular Disease AJNR Am. J. Neuroradiol., September 1, 2008; 29(8): 1476 - 1482. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. B. Singhal, E. Ratai, T. Benner, M. Vangel, V. Lee, W. J. Koroshetz, P. W. Schaefer, A. G. Sorensen, and R. G. Gonzalez Magnetic Resonance Spectroscopy Study of Oxygen Therapy in Ischemic Stroke Stroke, October 1, 2007; 38(10): 2851 - 2854. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
