| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Department of Neuroradiology, University of Technology, Fetscherstrasse 74, D-01307 Dresden, Germany
Editor:
With great interest, I studied the State of the Art review by Dr Beauchamp and colleagues in the August 1999 issue of Radiology (1), in which they make the important point that the responsibilities of radiologists are increasing with regard to acute stroke management. I am disappointed, though, that the computed tomography (CT) section of this article contains what I consider to be errors and unproved concepts and ignores the specificity of CT for the detection of irreversible tissue damage in acute stroke.
1. The detection of intracerebral thrombosis with CT was not an exclusion criterion in the National Institute of Neurological Disorders and Stroke (NINDS) Recombinant Tissue Plasminogen Activator (rt-PA) Trial. The NINDS rt-PA Stroke Trial Group excluded patients with intracranial hemorrhage (2). In contrast to the European Cooperative Acute Stroke Study (ECASS) investigators, investigators in NINDS did not quantify the extent of hypoattenuation due to cerebral ischemia on the baseline CT scan (3,4).
2. To my knowledge, the concept of reperfusion hemorrhage is unproved as yet. There is no convincing evidence we know of—and no reference provided by the authors—that thrombolysis beyond 3 hours increases the likelihood of posttreatment hemorrhage. According to the ECASS findings, the risk of intracranial hemorrhage does not increase after 3 hours. Moreover, in a subgroup of patients who underwent repeated cerebral blood flow measurements, fatal hemorrhages occurred only in those patients who did not undergo reperfusion (5). It is important to mention that not every posttreatment hemorrhage is associated with clinical deterioration and that a net benefit is achieved despite an increased risk of hemorrhage.
3. To my knowledge, the ECASS group never disclosed the number of infarcts that were missed by the local investigators but identified by the panel that read the CT scans. Dr Beauchamp and colleagues mention 11% without reference. Where does this number come from?
4. Although I followed this concept more than 5 years ago (6), I am no further convinced that fluid redistribution as described for cytotoxic edema can cause a change in electron density detectable at CT. The change in electron density is better explained by ischemic edema. The net uptake of water starts immediately after arterial occlusion and accounts for the high sensitivity and specificity of CT for the detection of ischemic edema (7,8).
5. A State of the Art article should mention that nonionic contrast agents did not affect the volume of infarcts under experimental conditions (9). Kendall and Pullicino (10) used an ionic contrast agent and reported on unconvincing retrospective study findings without a balance between patients who had received contrast material and patients who had not.
6. Dr Beauchamp and colleagues agree with other authors that the sensitivity of CT is "relatively poor" in the 1st hours after ischemic insult. They refer to an article more than 19 years old that reported on 14 patients examined within 6 hours of stroke onset (11) and a review article that did not provide original data (12). The authors ignore all observations about the high incidence of positive findings (50%–90%, depending on patient selection) in the 1st few hours after symptom onset, as summarized in my book (13). Moreover, they miss a very important point: A normal CT scan in acute stroke does not imply insensitivity of the method. It represents the favorable situation in which ischemic edema has not yet developed and the chance to avoid irreversible damage is still good.
When the ischemic lesion on day 1 after a stroke is used as the standard, ischemic infarctions develop in about 33% of patients—selected according to ECASS II criteria (4)—after the first 6 hours (14). Dr Beauchamp and colleagues are incorrect in saying that a CT scan is typically normal within 3 hours of ictus and that the selection for thrombolysis should be based on the absence of infarct signs. When magnetic resonance (MR) imaging results were compared with CT results in identical patients within 3 hours of hemispheric stroke, the CT scan was positive in 19 (53%) patients, and the MR image was positive in 18 (50%) patients (15). We reported 17 (68%) positive CT scans obtained in a series of 25 patients with middle cerebral artery trunk occlusion during the first 2 hours (6). Grond et al (16) observed tissue hypoattenuation in 12 of 23 (52%) patients within the first 3 hours. The incidence of positive findings was similar in ECASS II (4). I am afraid that such unjustified statements in a State of the Art review may further prejudice radiologists and discourage them from carefully searching for signs of developing ischemic edema on the early CT scans of patients who have had a stroke.
7. Dr Beauchamp and colleagues compared the sensitivity of CT with that of MR imaging but did not refer to the study in which both modalities were compared in the same patients in a short interval (15). They cited an observation for which CT was performed 4 hours earlier than MR imaging (17).
8. I very much missed a discussion about the specificity of CT findings in acute stroke. The data are available (14). Dr Beauchamp and colleagues just mentioned a low specificity of CT without providing a reference for that important statement. The authors are incorrect: Brain parenchymal hypoattenuation in patients with acute stroke as detected with CT has an 88% specificity (95% CI: 86%, 90%) for irreversible tissue damage (14).
It is unfortunate that the article by Dr Beauchamp and colleagues remains far behind the state of the art in describing the potential of CT in the diagnosis of acute stroke and ignores the role of CT in thrombolytic therapy, as described in an article and editorial in Radiology (18,19). In contrast to CT, the prospective value of early (<6 hours of stroke onset) MR imaging is not established as yet. Moreover, for various reasons, access to MR imaging is limited for many patients who have had a stroke. Many physicians who treat stroke must rely on CT for the foreseeable future. We should support them with evidence-based science to gather all the information that CT can provide in the diagnosis of acute stroke.
REFERENCES
Russell H. Morgan Department of Radiology, Johns Hopkins Medical Institutions, 601 North Caroline Street, Room 4223, Baltimore, MD 21287
In response to the letter by Dr von Kummer, we want to point out that it was not our intention to downplay the important contributions of CT in the management of acute stroke (1,2). The main focus of the review was to highlight the central role of the two major modalities in acute stroke imaging, CT and MR imaging. As such, we divided the discussion equally between these modalities, with an emphasis on the unique contributions each provides in patient evaluation. In so doing, we mentioned that notable strengths inherent in CT evaluation include greater availability, lower cost, and proved value in clinical trials. In addition, we emphasized the exciting potential of MR imaging to delineate not only areas of infarction but also tissue at risk (ie, the ischemic penumbra). Several interesting points are raised by Dr von Kummer, which we address in detail.
1. Dr von Kummer is correct in pointing out that the exclusion criteria in the NINDS trial did not include the CT detection of thrombosis. A comparison of the criteria in the NINDS trial with those in the ECASS can be found in a related article on this subject (3).
2. Dr von Kummer states that the concept of reperfusion hemorrhage is unproved. He correctly states that in the ECASS, the risk of intracranial hemorrhage did not increase after 3 hours. However, one reason that a statistically significant difference did not appear in any single thrombolytic trial may be related to the statistical power of the study to detect this difference. Specifically, to my knowledge, no study to date has been designed with time to treatment or hemorrhage as the primary end point. Compelling evidence is seen when two studies with similar design are compared. For example, the risk of symptomatic hemorrhage was 11.5% in the ECASS and 6.4% in the NINDS trial. This represents a factor of three increase associated with the 6-hour window of treatment reported by the ECASS group, as compared with the 3-hour treatment window in the NINDS trial. We consider these data adequately compelling to raise concern about the risk of reperfusion with increasing time to treatment.
3. Dr von Kummer states that data on the number of infarcts missed by the local investigators in the ECASS group are not available. These data can be obtained from the syllabus of Dr von Kummer’s 1998 presentation to the Symposium Neuroradiologicum for the focused topic session on cerebrovascular disease (4). This presentation reviewed the CT component of the ECASS (5), the findings of which indicated that 52 patients with extended ischemic edema "were overlooked by the local investigator as identified by a central CT reading panel following randomization."
4. We agree with Dr von Kummer that it is less likely that fluid redistribution as experienced during cytotoxic edema will show measurable changes at CT, since the total water density does not change during this process. Only when ischemic edema (early blood brain barrier breakdown and cytolysis) is initiated should there be a measurable attenuation or signal intensity change observable at CT or T2-weighted MR imaging. The strength of MR imaging is its capability to directly depict cytotoxic edema through diffusional changes. On this basis, it is also not surprising that diffusion-weighted imaging has been demonstrated to be more sensitive than CT and T2-weighted MR imaging in the detection of hyperacute stroke (6). However, because the concept of ischemic edema is at present incompletely defined, in writing a State of the Art article we thought it necessary to present the commonly held belief on the basis of established and recent articles that attribute CT hypoattenuation to cytotoxic edema (7).
5. We thank Dr von Kummer for his statement that nonionic contrast material does not affect the volume of infarct under experimental conditions.
6. Dr von Kummer’s comments in his point 6 were partially addressed in point 4. In addition, we point out the following related to the sensitivity of CT in evaluating acute ischemic infarction: (a) Dr von Kummer is emphasizing the comparability of CT and conventional T2-weighted MR imaging for the detection of irreversible ischemia. This remains a point of debate with, as Dr von Kummer points out, reported sensitivities for CT ranging from 50% to 90%, depending on patient selection. (b) More important are the performance of the diagnostic methods in the clinical setting and their ability to help detect tissue at risk of infarction.
With respect to a, Dr von Kummer acknowledges ECASS II and the sensitivity limitations of CT, stating that "the study was encumbered by a large number of protocol violations, most often related to the inclusion of patients who were retrospectively found to have early CT findings of stroke. The ECASS investigators recognize that the identification of that subgroup of patients with early CT findings may indeed prove to be difficult" (7). With regard to b, it is now well established that MR imaging provides a single examination in which infarcted tissue, tissue at risk for infarction, and the perfusion status of relevant brain regions can be assessed. For example, Dr von Kummer states that a normal CT scan does not imply insensitivity of the method. However, this depends on how sensitivity is defined. Although a normal CT scan can be used to exclude infarction, unlike diffusion and perfusion MR imaging, it does not provide the ability to distinguish normal brain from brain ischemia, which can progress to infarction.
Dr von Kummer expresses concern that we discourage radiologists from carefully searching CT scans for early signs of acute stroke. In this regard, we direct attention to the section entitled "Newly Defined Role of CT" (1). The second sentence of the first paragraph states "it is now incumbent on all radiologists to be familiar with the CT evaluation of acute stroke." This is followed by a careful review of all important CT findings and a conclusion emphasizing that CT has a clearly defined role in the management of acute stroke, CT is appealing because of its availability, and the utility of CT has been proved in large-scale clinical trials. In addition to noting the important contributions of CT, however, our goal was also to highlight the new capabilities of MR imaging in the performance of a comprehensive and quantitative acute stroke examination.
7. We agree with Dr von Kummer that referencing the recent study by Barber et al (6), the findings of which show that diffusion MR imaging is more sensitive than CT when it is performed during the same time frame, would have strengthened our article.
8. The specificity of CT as compared with that of MR imaging is discussed in points 4 and 6.
In conclusion, our State of the Art review was intended to summarize the contributions of CT to improved diagnosis and treatment of acute stroke. We reiterate that the role of CT is essential and proved in clinical trials. However, conventional and advanced MR imaging techniques will also contribute greatly to future stroke diagnosis and treatment.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  A.K. Gilligan, R. Markus, S. Read, V. Srikanth, T. Hirano, G. Fitt, M. Arends, B.R. Chambers, S.M. Davis, and G.A. Donnan Baseline Blood Pressure but Not Early Computed Tomography Changes Predicts Major Hemorrhage After Streptokinase in Acute Ischemic Stroke Stroke, September 1, 2002; 33(9): 2236 - 2242. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Barthel, S. Hesse, C. Dannenberg, A. Rossler, D. Schneider, W. H. Knapp, J. Dietrich, and J. Berrouschot Prospective Value of Perfusion and X-Ray Attenuation Imaging With Single-Photon Emission and Transmission Computed Tomography in Acute Cerebral Ischemia Stroke, July 1, 2001; 32(7): 1588 - 1597. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
