| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, Albert Einstein Medical Center, 5501 Old York Rd., Philadelphia PA 19141-3098
Received February 25, 1999;
accepted after revision May 21, 1999.
Presented at the annual meeting of the American Roentgen Ray Society, San
Francisco, April-May 1998.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. A review of all patients who underwent carotid artery sonography and angiography between 1993 and 1997 at our institution revealed 66 patients with complete sets of studies, yielding 132 examinations (right or left). Studies were not reinterpreted and angiography was considered to be the gold standard. Only stenoses of 60% or greater were included in our study. If the degree or location of stenosis differed on the two imaging studies, they were reviewed together to classify the type of sonographic error.
RESULTS. We found complete agreement of sonography and angiography in 115 cases (87%) and discrepancies in 17 (13%). Thirteen of 17 sonographic errors were false-positive interpretations and three were false-negative interpretations. One was an error in location. Retrospective review showed seven interpretive errors. In all these cases, the color Doppler image better revealed the degree of stenosis. Other complicating factors included inconsistencies between absolute velocities, velocity ratios, and waveforms obtained while a patient was being treated with an intraaortic balloon pump. In the other 10 discrepancies, the sonographic interpretation was accurate. Seven of these cases were false-positive interpretations in patients with contralateral occlusions or stenoses. The other three cases in this group showed long segments of stenosis, ulcerations, or tortuous vessels on angiography.
CONCLUSION. Our study suggests that increased accuracy can be achieved in the interpretation of carotid artery sonography by meticulous attention to the color image. When color Doppler sonography is technically limited by tortuosity or ulceration, or if significant contralateral disease is present, misinterpretation is more likely.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Duplex Doppler sonography is a noninvasive, relatively inexpensive study with a reported accuracy exceeding 90% [3]. However, more recently, the sensitivities and specificities of carotid duplex sonography have been questioned. A patient may have different results at different laboratories, or at the same laboratory on two different pieces of equipment [4, 5]. If asymptomatic patients are to be screened with carotid sonography, it becomes crucial to minimize the number of false-positive and false-negative studies to recognize the cost-benefit of screening such patients.
In our sonography laboratory, we routinely correlate carotid sonography with angiography, both to improve our interpretive skills and to monitor results for maintaining accreditation through the Intersocietal Commission for the Accreditation of Vascular Laboratories. We have been particularly interested in the small group of patients with discordant sonographic and angiographic results and set out to determine whether readings might be improved or if some discrepancies would always exist. By re-examining this subset of studies, we hoped to determine the types of errors, in particular those which were interpretive errors and those which arose from limitations of the technology. Our ultimate goal was to improve the accuracy of our carotid sonographic interpretation.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The sonograms were obtained on several units: HP (Acuson, Mountain View, CA), UltraMark 9, and 3000 (Advanced Technology Laboratories, Bothell, WA). Our protocol for carotid artery sonography includes a complete real-time evaluation in sagittal and transverse planes of the visible portions of the common carotid artery (CCA), internal carotid artery (ICA), and external carotid artery; and a sample of the vertebral arteries taken at the level of the carotid bifurcation. Both color and duplex imaging were used to identify vessels, image plaque, and evaluate stenoses. The pulsed Doppler angle was maintained at or less than 60°. Our standard set of recorded images includes sagittal and transverse images in black and white; sagittal in color; and sagittal with duplex imaging. The areas imaged were the CCA 2 cm proximal to the bifurcation, the bifurcation, the external carotid artery, and the proximal and distal ICA. The vertebral artery was imaged and sampled at the level of the carotid bifurcation. Power Doppler sonography has been available for the past 4.5 years and was used in all cases of suspected occlusion or near occlusion. Extra images were usually obtained in cases of suspected hemodynamically significant lesions.
Our major criteria for determining the degree of stenosis are the peak systolic and end diastolic velocities. For a critical stenosis (80-99%), peak systolic and end diastolic velocities must be greater than 250 cm/sec and 100 cm/sec, respectively. For a severe stenosis (60-79%), the corresponding values are greater than 200 cm/sec and 60 cm/sec, respectively. The peak systolic ICA to CCA ratios (ICA/CCA) are used for an internal control (severe stenosis: >1.8, critical stenosis: >3.7). These criteria, based on recommendations in the literature and our own experience, have been developed over time in our laboratory. Inconsistencies among the data often resulted in reimaging by the physician at the time of the study. All studies during the day were verified and interpreted at the time of the study by one of four staff radiologists who subspecialize in sonography. A few studies performed on weekends were interpreted by the general radiologists. The sonographers performing the studies were all certified or eligible by the Registry of Diagnostic Medical Sonographers and had 1-20 years of experience. The angiographic technique was a digital subtraction arterial study with arch and bilateral CCA injections in two projections. Stenoses were measured according to the North American Symptomatic Carotid Endarterectomy Trial criteria: (1-[narrowest diameter / diameter beyond visible plaque]) x 100% [1].
In this study we compared the results of carotid artery sonography and angiography to obtain a subset with discrepant readings. Sonographic studies were not reinterpreted, and angiographic results were considered the gold standard. We defined a discrepant reading as a difference in stenosis of an entire category. For example, if sonography showed a critical stenosis and angiography a severe stenosis, this finding was tabulated as a false-positive interpretation. A sonographic stenosis at least one category less than an angiographic stenosis was a false-negative interpretation. Discrepancies among normal, mild, and moderate lesions were not tallied. Inaccurate location of severe or critical stenoses also constituted a discrepancy. A discrepancy between the two studies prompted a review of the angiographic images by another neuroradiologist. If the angiographic results were confirmed, the sonographic study was considered to be incorrect. For each discrepant case, two of the authors together reviewed the sonographic study in an unblinded fashion to determine if, in retrospect, it was interpreted correctly. An interpretive error occurred when the review of the sonographic images disclosed information, such as an inconsistency, an overlooked piece of data, or misinterpreted data, that would have permitted the same diagnosis that was obtained by angiography. If the sonographic interpretation remained the same despite knowledge of the angiographic results, the error was ascribed to a limitation of sonographic technology. Imaging results from the external carotid and vertebral arteries were not considered in this study. For statistical purposes, the results of each side (right or left) were considered independent data.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
In only one case did an initial discrepancy between sonography and angiography eventually result in a reinterpretation of the angiographic study. In this case, the sonographic reading was a critical ICA stenosis, and the angiographic reading was interpreted as an occlusion. Because of the discrepancy at the time of the studies, CT angiography confirming the sonographic results was performed. All cases of occluded vessels revealed on sonography were confirmed by angiography. Three occlusions of the CCA and eleven of the ICA were found.
Our retrospective correlation showed that seven of 17 sonograms were interpreted incorrectly. In all seven cases, the color Doppler image gave a more accurate depiction of the degree of stenosis than the pulsed Doppler data. Two studies were false-negative interpretations in which hemodynamically significant lesions were missed. In the first of these cases a severe CCA stenosis was completely overlooked, presumably because of its unusual location proximal to the carotid bifurcation. The stenosis was recorded on the color image, but no Doppler signal was obtained from that location. The second of these cases was a critical, flow-limiting ICA stenosis interpreted as a moderate lesion. This error would have been avoided by using the color image and the ICA/CCA ratios. The absolute ICA velocities were only in the moderate range because the stenosis was so critical that flow was compromised and the peak systolic velocities had fallen. The ICA/CCA ratio of 4:1 and the color image were appropriate for a critical stenosis.
Five of the interpretative errors were false-positive cases. Use of ICA/CCA ratios with the color image would have resulted in a correct interpretation of bilateral mild lesions instead of severe stenoses. In the first case, the patient was 34 years old and hypertensive, factors that are associated with high velocities in all the sampled vessels [6].
Another interpretive error was made by recording the duplex data while a patient was undergoing treatment with an intraaortic balloon pump (Fig. 2), (Fig. 2), (Fig. 2). An area of narrowing that did not look significant on the color image was thus interpreted as critical. Pump-assisted waveforms have two systolic peaks for each pulse. The first peak is the intrinsic left ventricular contraction. The second peak in early diastole is caused by inflation of the balloon. No true measurable peak systole and end diastole occur. Several subsequent patients have also shown artificially high systolic-type velocities from the balloon pump waveforms. Having learned from this error, we have the balloon pump turned off for several beats while recording the duplex data through the stenosis if we notice any area of narrowing in color.
|
|
|
The final two cases with interpretive errors would have been avoided by making better use of the color image. In the one case of a correctly graded stenosis in an incorrect location (CCA instead of ICA), the correct location was obvious on the color image. In the last case, the peak systolic velocity was in the severe range, but the color image was not consistent with a severe lesion.
In 10 of 17 cases we found no obvious cause for the discrepancy. The duplex data were correctly interpreted and internally consistent (peak systolic and end diastolic velocities and ICA/CCA ratios) and the color images seemed appropriate. In seven of these ten cases, hemodynamically significant contralateral lesions (five of seven occlusions) resulted in a false-positive sonographic interpretation on the ipsilateral side (Fig. 3), (Fig. 3). In the other three cases, the angiographic images showed long segments of stenosis, large ulcerations, and tortuous vessels (Fig. 4), (Fig. 4), (Fig. 4), (Fig. 4). Two of these were false-positive and one was a false-negative interpretation.
|
|
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although the purpose of this study was to investigate the sources of errors in the sonographic examination, the data show acceptable levels of accuracy (87%). Nonetheless, in seven cases, errors were caused by inaccurate interpretation. The common factor in this group was an inconsistency between the color Doppler images and the duplex Doppler findings. The next most common error was ignoring the ICA/CCA ratio. This ratio helps detect flow-limiting critical stenoses and prevents the misdiagnosis of a stenosis in a patient with diffusely elevated velocities.
The types of cases in which the sonographic interpretation was accurate but discrepant fell into two categories. Most of these cases (70%) had a significant contralateral lesion. The phenomenon of overestimation of a stenosis contralateral to an occlusion or high-grade stenosis is well known [10]. The explanation is thought to be compensatory increased flow through a fixed diameter, which results in an elevated velocity contralateral to a high-grade stenosis or occlusion. This idea is based on a model of only two vessels, the carotid arteries, supplying the brain. To test this hypothesis, Busuttil et al. [10] repeated the duplex Doppler studies of overestimated stenoses after an endarterectomy on the contralateral high-grade stenosis. In 44 (58%) of 76 cases, the degree of stenosis decreased by at least one duplex category. This decrease supported the theory of compensatory increased flow in at least half the cases. However, this theory has not been proved by angiographic correlation. Angiography should show intracranial cross-filling via the circle of Willis from the side of the overestimated lesion toward the side of the occlusion or high-grade stenosis. In a study by Beckett et al. [11], such cross-filling was equally common in patients with accurate duplex Doppler studies as in those with an overestimation. The difficulty in proving this theory stems in part from the simplification of only two major vessels supplying the brain. In reality, some patients show increased flow through one or both vertebral and external carotid arteries.
The overestimation phenomenon is most apparent with a contralateral occlusion, as in our study, but can sometimes be seen with hemodynamically significant contralateral stenoses [12, 13]. This phenomenon occurs between 27% and 48% of the time [10,12]. Both Fujitani et al. [12] and AbuRahma et al. [13] reported revised criteria for grading a stenosis contralateral to an occlusion or high-grade stenosis. In both of these studies, use of a higher peak systolic frequency decreased the frequency of overestimation. Polak [6] suggests that the ICA/CCA velocity ratio will not be affected by the contralateral lesion and can be used to correctly grade the stenosis. In our experience, this ratio did not allow a downgrading of the stenosis. Furthermore, the color images seemed to support the overcalled stenoses (Fig. 3A).
The final three cases of accurate but discrepant sonographic interpretation had no contralateral lesion, and all had internally consistent duplex and color Doppler data. However, the corresponding angiographic studies had the common finding of ICAs with large ulcerated plaques or tortuous vessels or both. Tortuous vessels are known to cause difficulties with accurate duplex sonography [7]. It may be impossible to obtain an accurate angle of insonation and, conversely, what appears to be an adequate angle on the sonographic images is not in reality. Accurate detection of ulcerated plaques has had variable success. Small ulcerations are routinely over-looked and even large ones may not be appreciated. In our study, one case had an ulceration that was so large that it was misinterpreted as part of the normal lumen of the vessel. This misinterpretation resulted in an underestimation of the degree of stenosis (Fig. 4).
This study shows that increased accuracy can be achieved in the interpretation of carotid artery sonography. The key to interpretive improvement is the color Doppler image; it must be consistent with the duplex data. In all the cases of interpretive errors, the color image portrayed the angiographic degree and location of stenosis. If the color image and duplex Doppler data are concordant, a stenosis can be interpreted with confidence. Careful attention to the color image should help prevent the failure to diagnose a flow-limiting, critical stenosis and conversely the misdiagnosis of a severe lesion in a patient with diffusely high velocities. Errors will occur in cases with heavily calcified plaques and large ulcerations because in these situations the color image will be limited in its ability to display the vessel lumen. Errors will also occur in patients with tortuous vessels because the duplex Doppler data will be less accurate. However, careful attention to the color image should alert the sonographer to the degree of tortuosity. In such cases, interpretation of stenosis should be qualified with some degree of uncertainty. Similarly, one should interpret a stenosis contralateral to an occlusion or high-grade stenosis with caution. Velocity data obtained with a patient being treated by an intraaortic balloon pump may be misleading, and any suspicious area found with color Doppler sonography should be insonated for several beats with the pump turned off.
More meticulous scanning by the technologists with specific attention to protocols will decrease the number of interpretive errors. Careful correlation of transverse and sagittal images should prevent the miss of a CCA stenosis or the error in locating a stenosis in the ICA instead of the bifurcation. Clearly, the sonographer's skill will affect the radiologist's interpretation. However, once the images are presented for review (as in this study), it is the color image that reveals the interpretive errors. Inconsistencies in the duplex and color Doppler data should be detected at the time of the original study and should have prompted a second look by the radiologist.
This study shows an acceptable level of accuracy of carotid artery sonography in our study population. Few hemodynamically significant lesions were missed because most discrepancies were false-positive sonographic interpretations. Nonetheless, these errors may result in unnecessary arteriography with its associated morbidity and cost. Although we believe that the protocol of our carotid examination is sufficient, the interpretation must include correlation of the duplex and color data and comparison of the two sides.
This study has several limitations. The most important issue, and one that is commonly addressed, is use of angiography as the gold standard. Definite variations in intraobserver and interobserver interpretations of angiographic ICA stenoses exist [14]. Nonetheless, we chose angiography as our standard because it is the most accurate technique available and correlates with symptoms and surgical outcome [15]. Furthermore, strict adherence to the angiographic criteria for stenosis used in the North American Symptomatic Carotid Endarterectomy Trial allows better quantification of the angiographic stenoses [1].
A second important limitation is the technical quality of the sonography. If the sonographic images do not adequately portray the abnormality, the number of seemingly accurate but discrepant sonographic interpretations will increase. Presumably, this potential problem was not the issue in the seven cases with significant contralateral disease. However, a technically better study would have avoided the other three discrepant sonographic interpretations.
Though the purpose of this paper is more descriptive than statistical, we did calculate the percentage of agreement between sonography and angiography. These calculations assume that the data from the right and left sides can be considered independent measurements. However, considering the seven discrepant cases caused by contralateral high-grade lesions, we should not always make this assumption. Nonetheless, we chose to calculate percentages in this manner because it is considered the standard for reporting vascular data.
The relative lack of experience of some technologists and staff radiologists undoubtedly contributed to our rate of errors. It would certainly be ideal to have only experienced sonographers and radiologists performing and interpreting these studies. However, our situation, with varying levels of skill, is more representative of the general radiology community.
Another potential limitation of this study is the cohort of patients who actually underwent both the sonographic and angiographic examinations. Although most patients underwent angiography because sonography detected a symptomatic hemodynamically significant lesion, other patients may have undergone angiography because the symptoms did not correlate with the sonographic findings. This phenomenon could increase the number of discrepancies.
Finally, because only a small number of patients who have carotid artery sonography also have carotid artery arteriography, a bias of ascertainment occurs. For example, if sonography results in many false-negative interpretations, and, as a result, the patients never proceed to angiography, we would have no way of knowing about these errors. The calculated accuracy in this study is only based on a small subset of patients and does not necessarily reflect the larger population of all patients having carotid sonography.
Carotid artery sonography will probably be performed ever more frequently in the future as cerebrovascular disease is treated more aggressively and as carotid artery sonography is used to screen for general atherosclerotic disease [16]. To achieve the greatest accuracy, errors must be avoided. This study suggests that meticulous attention to the color image is the best way to achieve that level of accuracy. However, we must recognize that some errors will always occur because of the inherent limitations of current sonographic technology.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  W. W. Lien, A. H. Lee, Y. Kono, G. C. Steinbach, and R. F. Mattrey Noninvasive Estimation of the Pressure Gradient Across Stenoses Using Sonographic Contrast: In Vitro Validation J. Ultrasound Med., May 1, 2004; 23(5): 683 - 691. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. S. Kriegshauser, M. D. Patel, and K. D. Nelson Carotid Pseudostring Sign From Vasa Vasorum Collaterals J. Ultrasound Med., September 1, 2003; 22(9): 959 - 963. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. A. Fellner, W. Lang, A. I. Qureshi, S. H. Kim, M. F. K. Suri, and L. N. Hopkins The Way Out * Response Stroke, August 1, 2002; 33(8): 1948 - 1949. [Full Text] [PDF]
|
|
|
|
 |
|
 F. A. Fellner, R. Wutke, W. Lang, O. E. H. Elgersma, B. C. Eikelboom, and W. P. T. M. Mali Imaging of Internal Carotid Arterial Stenosis: Is the New Standard Noninvasive? Dr Elgersma and colleagues respond: Radiology, June 1, 2001; 219(3): 858 - 859. [Full Text] [PDF]
|
|
|
|
 |
|
 How Accurate Is Carotid Ultrasound? Journal Watch (General), January 18, 2000; 2000(118): 3 - 3. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
