Carotid artery occlusion: Positive predictive value of duplex sonography compared with arteriography

Transcription

1 Carotid artery occlusion: Positive predictive value of duplex sonography compared with arteriography Jonathan D. Kirsch, MD, Louis R. Wagner, MD, E. Meredith James, MD, J. William Charboneau, MD, Douglas A. Nichols, MD, Fredric B. Meyer, MD, John W. Hallett, MD, Rnchester J Minn. Purpose: Duplex ultrasonography is an accurate, noninvasive method for diagnosing, characterizing, and classifying atherosclerotic stenoses of the extracranial carotid artery system. To date, however, no large series has studied the predictive value of duplex sonography in the diagnosis of carotid artery occlusion, an important consideration, given the marked difference in clinical treatment between patients with high-grade stenosis (surgical therapy) and those with occlusive disease (nonsurgical therapy). Methods: We retrospectively reviewed 158 patients with 174 occluded carotid artery segments (examined over a 6V2-year period) to determine the predictive value of duplex sonography in differentiating carotid artery occlusion from high-grade stenosis. Results: All patients had arteriographic correlation. Duplex ultrasonography had a positive predictive value of 92.5% (7.5% false-positive rate; 95% confidence interval, 3.6% to 11.4%) in establishing a diagnosis of carotid artery occlusion. Further analysis revealed no significant improvement in the false-positive rate with the addition of color Doppler flow imaging to high-resolution B-mode scanning and pulsed Doppler spectral analysis. Predictive value increased to 96.7% (95% confidence interval, 90.7% to 99.3%) over the last 2 years of the study, a statistically significant improvement. Conclusions: We believe that duplex ultrasonography is an acceptably accurate method for diagnosing carotid arterial occlusion in most patients. Arteriography should be reserved for patients with symptoms who are surgical candidates to identify those who may still have a surgically correctable high-grade stenosis. (J VASe SURG 1994;19:642-9.) Duplex ultrasonography is an effective noninvasive method for the detection and characterization of atherosclerotic disease involving the extracranial carotid artery system. Criteria for the classification of stenotic, nonocclusive lesions and the accuracy of those criteria have been well documented in the literature. 1. S When compared with angiography, duplex ultrasonography is highly accurate in the diagnosis and classification of stenoses greater than 50%. To date, however, no large study of the positive predictive value of duplex sonography in the detection of carotid artery occlusion has been reported. From the Department of Diagnostic Radiology, the Department of Neurologic Surgery (Dr. Meyer), and the Division of Vascular Surgery (Dr. Hallett), Mayo Clinic and Mayo Foundation, Rochester. Reprint requests: E. Meredith James, MD, Mayo Clinic, 200 First St., SW, Rochester, MN Copyright 1994 by The Society for Vascular Surgery and International Society for Cardiovascular Surgery, Notth American Chapter /94/$ /1/ Data previously reported in relatively small series show that the positive predictive value of an ultrasound diagnosis of occlusion ranges widely, from 58% to 100%.1-8 The need for an accurate assessment of duplex sonography to diagnose occlusion is important because patients with high-grade nonocclusive disease are treated with surgery, but those with occlusive disease are treated without surgery. Also, because some authors conclude that carotid artery duplex sonography can substitute for angiography in selected patients before surgery, knowing with what degree of confidence one can rely on the sonographic diagnosis of occlusion and thus elect not to operate is pararnount In addition, noninvasive techniques for the diagnosis of carotid artery occlusion avoid the inherent neurologic or systemic morbidity and potential mortality associated with arteriography and injection of contrast material. Duplex sonography is also less costly than arteriography and has virtually no contraindications for use. To provide better data on the ability of carotid

2 JOURNAL OF VASCULAR SURGERY Volume 19, Number 4 Kirsch et al. 643 artery ultrasonography to correctly diagnose carotid artery occlusion, we retrospectively reviewed all diagnoses of carotid artery occlusion made by duplex sonography at the Mayo Clinic over the past 6 1 /2- years. We correlated these findings with arteriography to determine the predictive value of our ultrasound diagnosis. The data also were analyzed for time within the 61/2-year period at which the examination was performed and for addition of color-flow imaging to determine whether these variables had any effect on our ability to diagnose occlusion. Finally, the clinical significance of our false-positive diagnoses of occlusion by ultrasonography was examined. PATIENTS AND METHODS The records of all patients with an ultrasound diagnosis of carotid artery occlusion made between January 1,1985, and July 1,1991, were analyzed. In that time, 14,728 carotid artery ultrasound examinations were performed at the Mayo Clinic. An ultrasound diagnosis of occlusion was made in 1049 patients. Of the 1049, 323 patients also underwent cerebral angiography. Certain exclusions were made from this group of 323 patients to obtain a subset of comparable ultrasound examination results with angiographic correlation. These exclusions consisted of the following: (1) Delay of more than 30 days between ultrasound examination and angiography (121 patients); exclusion avoided the potential error of a high-grade stenosis becoming an occlusion over time. (2) Angiograms obtained by intravenous digital subtraction technique only (37 patients); this technique was believed not to be sufficiently sensitive at differentiating high-grade stenosis from occlusion in comparison with selective intraarterial angiography.12,13 (3) Stroke between ultrasound examination and angiography (seven patients). After these patients were excluded, a subset of 158 patients with comparable ultrasound and angiographic examination results remained. All examinations were performed by sonographers with special training in duplex scanning. They used Acuson 128 and X/P 128 Computed Sonography systems (Acuson, Inc., Mountain View, Calif.) or ATL Ultramark 8 and 9 systems (Advanced Technology Laboratories, Bothell, Wash.) with 5 MHz or 7.5 MHz transducers. One hundred twenty patients underwent duplex examination with color Doppler flow imaging (CDFI) on Acuson equipment. Thirty-eight patients underwent duplex examination with spectral Doppler imaging only, with either Acuson or ATL equipment, before the introduction of CDFI in our department. Staff radiologists interpreted the studies before angiography was performed. The diagnosis of carotid artery occlusion was based solely on the finding of absent flow by spectral and (when available) color Doppler imaging (Fig. 1). When occlusion was suspected by the examiner, the instrument was adjusted for maximum flow sensitivity and minimum wall filter, and the Doppler sample volume was enlarged to encompass the entire diameter of the vessel. To help in the positive identification of the external carotid artery, the examiner routinely "tapped" on the ipsilateral temporal artery and looked for the resulting transmitted oscillations in the spectral tracing from the external carotid artery. Several additional findings, which may be seen with either high-grade stenosis or occlusion of the internal carotid artery, were also encountered. These were the identification on B-mode imaging of echogenic plaque or thrombus filling the lumen, dampened high-resistance waveform in the ipsilateral common carotid artery (CCA) (Fig. 2 A), "internalization" of the waveform of the ipsilateral external carotid artery because of collateral vessel supply to the intracranial circulation from the external carotid artery by way of periorbital branches (Fig. 2 B), and higher velocity in the contralateral common carotid artery than in the CCA on the highly stenosed or occluded side because of higher flow volume on the widely patent side (Fig. 2 C). The importance of these secondary findings, when encountered, was simply to alert the examiner to a significant obstructing lesion, either high-grade stenosis or occlusion. The relative frequency with which these secondary findings were encountered was not tabulated. Arteriography was performed by staff radiologists using selective arterial catheterization. Contrast agent injection and film sequence were dictated by the performing radiologist. Cut film with photographic subtraction was used. A diagnosis of occlusion was made when the contrast column terminated at the site of occlusion (Fig. 3). RESULTS The 158 patients (125 men and 33 women; average age, 68.2 years) with comparable duplex ultrasound examination and angiography results had 174 occluded carotid artery segments (154 internal carotid arteries, 11 CCA, and 9 external carotid arteries). Of these occlusions 161 diagnosed by ultrasonography were confirmed by angiography, so that the positive predictive value for duplex examination was 92.5% (95% confidence interval,

3 644 Kirsch et at. JOURNAL OF VASCULAR SURGERY April 1994 Fig. 1. A, High-resolution gray-scale image of occluded right internal carotid artery (ICA). Echogenic material (white arruws) fills lumen. Spectral tracing reveals no evidence of flow. There is small baseline artifact caused by pulsatile movement of occluded vessel. B, Color-flow Doppler image of same vessel reveals termination of flow within lumen of occluded vessel (curved arruw) with small area of flow reversal at point of occlusion (arruw).]v, Internal jugular vein.

4 JOURNAL OF VASCULAR SURGERY Volume 19, Number 4 Kirsch et al. 645 Fig. 2. Spectral tracings in patient with occluded right internal carotid artery. A, Highresistance waveform in ipsilateral CCA with diminished systolic velocity (0.64 m/sec) and virtually absent diastolic flow. B, "Internalization" of ipsilateral external carotid artery (ECA) with increased diastolic flow. C, Increased systolic velocity (0.83 m/sec) in contralateral CCA.

5 646 Kirsch et at. JOURNAL OF VASCULAR SURGERY April 1994 revealed no significant difference in the false-positive rates between the first and the second periods of the study. However, when the false-positive rate in the third period (3.3%) was compared with the collective rate in the first two periods, there was a statistically significant decrease (P = ). Clinical outcome in the 13 patients with falsepositive diagnoses was examined (Table II). Seven patients underwent endarterectomy on the segment with erroneous diagnosis, two had endarterectomy on the contralateral carotid artery, and four were treated by follow-up examinations. Fig. 3. Lateral subtraction view of left carotid artery injection reveals occlusion of internal carotid artery (arrow). 88.6% to 96.4%). There were 13 false-positive diagnoses of occlusion, a rate of7.5%. All ultrasound errors involved internal carotid arteries. Angiographic descriptions of the 13 false-positive results were very high-grade stenosis with slow flow or "slim sign" (seven patients) (Fig. 4), greater than or equal to 95% stenosis (five patients), and minimal stenosis (one patient). The data were analyzed to determine whether the addition of CDFI improved our accuracy in differentiating high-grade stenosis from occlusion. Occlusion in 44 segments was diagnosed early in the study by spectral analysis only. There were four falsepositive results, for a rate of 9%, or a positive predictive value of91 %. Occlusion of 130 segments was diagnosed with spectral analysis and CDFI. Confirmation of 121 by angiography yielded a 6.9% false-positive rate, or a positive predictive value of 93.1 %. Statistical analysis (Fisher'S exact test; P = 0.740) revealed no significant difference between these two groups. The data also were analyzed by when an examination occurred in the course of the study. The study was divided into three equal periods of 26 months each (Table I). Statistical analysis (Fisher'S exact test) DISCUSSION Duplex sonography has been shown to be highly accurate relative to angiography in the diagnosis and classification of carotid artery stenoses greater than 50%.1-5 However, a review of the literature reveals no large study addressing the predictive value of duplex sonography in diagnosing carotid artery occlusion. Small subsets of data on occlusion extracted from larger studies on stenosis exist. If one calculates from these data the positive predictive value of ultrasound imaging in making a diagnosis of occlusion, the values range widely-from 58% to 100%.1-8 Making the distinction between occlusion and high-grade stenosis, however, is clinically important. The North American Symptomatic Carotid Endarterectomy Trial (NASCET) has shown that carotid endarterectomy for high-grade symptomatic carotid artery stenosis (70% to 99%) is highly effective in preventing ipsilateral stroke and can be safe. 14 Operating on occluded arteries, however, is rarely worthwhile. 6 Therefore it is important to know with what degree of confidence one can rely on a diagnosis of occlusion made by duplex sonography, especially if the clinical decision will be one of electing not to operate. What should be the role of angiography in further evaluating these patients? Some authors believe that in selected patients, performance of ultrasonography on carotid arteries is accurate enough to preclude arteriography before the decision to perform endarterectomy.9-11 These studies, however, did not specifically address occlusion. Other investigators believe that patients who have appropriate clinical indications for carotid endarterectomy should undergo angiography even when noninvasive tests such as duplex sonography indicate total occlusion. 6,12 In fact, the NASCET concluded that ultrasound findings were unsatisfactory for detecting high-grade carotid artery stenosis. 14,15 The authors pointed out, however, that ultrasound examinations from various

6 JOURNAL OF VASCULAR SURGERY Volume 19, Number 4 Kirsch et at. 647 trial centers were not standardized. This retrospective review of cases of carotid artery occlusion diagnosed by duplex ultrasonography and correlated with angiography in a single experienced center was performed to provide data to help resolve this problem. The study design permitted the calculation of only the positive predictive value of an ultrasound diagnosis of occlusion. Sensitivity, specificity, and negative predictive value could not be assessed. Positive predictive value, however, is an important variable to measure in this clinical situation, because the key question being asked is "Given an ultrasound diagnosis of occlusion, what are the chances that the vessel is, in fact, patent?" By failing to detect a patent residual channel, the false-positive result of the ultrasound examination results in the continued exposure of the affected patient to the risk of stroke and denies the patient the potentially beneficial effects of carotid endarterectomy. Our data indicate that the positive predictive value of duplex ultrasonography in the diagnosis of carotid artery occlusion is 92.5%. The reader should be aware that because positive predictive value depends on the prevalence of the disease in question in a given population, considerably lower values could be encountered in clinical settings where the prevalence of occlusion is lower than in a tertiary care center such as the Mayo Clinic. Our false-positive diagnoses occurred almost exclusively with high-grade stenoses described with angiography as being greater than 95%. Possible factors contributing to diagnostic error were technologic factors - for example, sensitivity of the equipment to detect extremely slow flow through and beyond a high-grade stenosis; patient factors, such as vessel calcification or tortuosity and physical features (for example, neck thickness obscuring visualization of the lumen); and operator experience and thoroughness of the examination. Although several studies have stated that the addition of CDFI should improve the accuracy of differentiating high-grade stenosis from occlusion,1,16 our data do not support this contention. False-positive rates of9% and 6.9% were found with spectral analysis only and CDFI with spectral analysis, respectively. Statistically, there was no significant difference between these two groups. Examination of the data chronologically revealed that examinations performed during the last third of the study had a significantly lower false-positive rate than those performed during the first and second thirds of the study. Because the diagnostic criteria and methods used to diagnose occlusion did not Fig. 4. Subtraction arteriogram in patient with falsepositive ultrasound diagnosis of internal carotid artery occlusion. Lateral view of right carotid artery injection reveals very high grade stenosis ("slim sign") in proximal internal carotid artery (arrow). change during the course of the study, this change in the false-positive rate presumably was due to examiner or equipment factors. Possible factors that contributed to this decline in the false-positive rate were a learning curve for radiologists and sonographers as they gained experience with the examination over the years and the continual upgrading of the equipment software and hardware in the department, resulting in increased sensitivity to the detection of small amounts of slow flow. The method of examination and interpretive criteria were not unique or specialized in any way compared with methods used at other institutions, and our results, therefore, should be achievable by other laboratories. Review of the clinical outcome in our patients with false-positive ultrasound diagnoses revealed that only slightly more than half (seven of 13) of them had an endarterectomy on the segment with an erroneous

7 648 Kirsch et at. JOURNAL OF VASCULAR SURGERY April 1994 Table 1. False-positive rates for the ultrasonographic diagnosis of carotid artery occlusion during three periods Portion of Date of False-positive Correlated False-positive study examination results, no. examinations, no. rate (%) First third 1/85-2/ Middle third 3/87-4/ Last third 5/89-7/ Total Table II. Clinical outcome in false-positive occlusions of the carotid artery.age Indication for Angiographic Clinical Patient (yr) Sex evaluation description management 1 66 M (R)CVA 99% "slim sign" (R)CEA 2 72 M (R) amaurosis fugax 95% stenosis (R)CEA 3 83 M (L)CVA 99% with distal (L) CEA thrombus 4 81 M TIAs 99% with slow flow (R) CEA 5 73 F Bilateral amaurosis fugax > 95% stenosis (L)CEA 6 75 M (R) TlA 99% stenosis with slow (R) CEA flow 7 75 M (R) amaurosis fugax 99% stenosis with slow (R)CEA flow 8 67 F (R)CVA 98% (L) stenosis* (R)CEA 85% (R) stenosis 9 46 F (L)CVA 99% (L) stenosis with (R)CEA slow flow* 90% (R) stenosis F Dizzy spells Minimal (R) stenosis Follow M CVA 99% stenosis Follow F Follow-up of (L) CEA 99.9% "slim sign" Follow M Amaurosis fugax 99% "slim sign" Follow CEA, Carotid endarterectomy; CVA, cerebrovascular accident (stroke); (L), left; (R), right; TIA, transient ischemic attack. *Ultrasound error segment. The "slim sign" is diffuse narrowing of the internal carotid artery with slow flow to base of skull. diagnosis (Table II). Clinical reasons for not operating on the false-positive segment (n = 6) could be categorized into three groups: unfavorable anatomic features, nonspecific or no cerebrovascular symptoms, and patient refusal of any operation. The unfavorable anatomic features included disease extending into the very distal internal carotid artery or carotid siphon and the "slim sign," in which the diseased internal carotid artery provided minimal flow to the ipsilateral hemisphere and collateral flow already seemed to be excellent. In two cases, the contralateral carotid artery stenosis was also severe and appeared technically more favorable for a safe endarterectomy. In four patients medical management was preferred by the patient or physician. Three of these four patients had a "slim sign" pattern that was not favorable for endarterectomy. In summary the predictive value of duplex ultrasonography for the diagnosis of carotid artery occlusion is high (92.5% over the 6112 years of the study, 96.7% over the final 2 years). The improved accuracy over the last 2 years, we believe, is due both to our gaining greater experience with the examination over time and to the continual improvements in equipment software and hardware. CDFI did not significantly improve the diagnostic accuracy of duplex sonography in differentiating high-grade stenosis from occlusion. We conclude that duplex sonography is an acceptably accurate method for diagnosing carotid artery occlusion in most patients. We believe that symptom-free patients with an ultrasound diagnosis of carotid artery occlusion do not require arteriographic confirmation. Recent findings of the NASCET, however, suggest that any patient with symptoms who has a false ultrasonographic diagnosis of internal carotid artery occlusion and who does not undergo operation has an increased risk of subsequent ipsilateral stroke and is denied the possible beneficial effects of endarterectomy in reducing this

8 JOURNAL OF VASCULAR SURGERY Volume 19, Number 4 Kirsch et al. 649 risk. Therefore despite the high positive predictive value of duplex ultrasonography, we believe that arteriography is still warranted for patients who are deemed candidates for surgery, who have an ultrasound diagnosis of carotid artery occlusion, and who continue to have neurologic symptoms. The rationale for arteriography in such patients is twofold: to be certain that the symptomatic internal carotid artery is truly occluded and to identify surgically correctable contralateral carotid artery disease that may improve collateral circulation. REFERENCES 1. Erickson SJ, Mewissen MW, Foley WD, et al. Stenosis of the internal carotid artery: assessment using color Doppler imaging compared with angiography. Am J Roentgenol 1989;152: Steinke W, Kloetzsch C, Hennerici M. Carotid artery disease assessed by color Doppler flow imaging: correlation with standard Doppler sonography and angiography. Am J RoentgenoI1990;154: Polak JF, Dobkin GR, O'Leary DH, Wang A-M, Cutler SS. Internal carotid artery stenosis: accuracy and reproducibility of color-doppler-assisted duplex imaging. Radiology 1989; 173: Zwiebel WJ, Austin CW, Sackett JF, Strother CM. Correlation of high-resolution, B-mode and continuous-wave Doppler sonography with arteriography on the diagnosis of carotid stenosis. Radiology 1983;149: Jacobs NM, Grant EG, Schellinger D, Byrd MC, Richardson JD, Cohan SL. Duplex carotid sonography: criteria for stenosis, accuracy, and pitfalls. Radiology 1985;154: Bornstein NM, Beloev ZG, Norris ]W. The limitations of diagnosis of carotid occlusion by Doppler ultrasound. Ann Surg 1988;207: Bridgers SL. Clinical correlates of Doppler/ultrasound errors in the detection of internal carotid artery occlusion. Stroke 1989;20: Sullivan ED, Cranley J]. Noninvasive diagnosis of internal carotid artery occlusion. Am J Surg 1983;146: Thomas GI, Jones TW, Stavney LS, Manhas DR, Spencer MP. Carotid endarterectomy after Doppler ultrasonographic examination without angiography. Am J Surg 1986;151: Crew JR, Dean M, Johnson JM, et al. Carotid surgery without angiography. Am J Surg 1984;148: Wagner WH, Treiman RL, Cossman DV, Foran RF, Levin PM, Cohen JL. The diminishing role of diagnostic arteriography in carotid artery disease: duplex scanning as definitive preoperative study. Ann Vasc Surg 1991;5: O'Leary DH, Gibbons GW, Pinel DF. Limitations of noninvasive testing in assessing the "occluded" carotid artery. Am J Neuroradiol 1983;4: Pexman JHW, Wriedt CHR, Richard TC. The significance of internal carotid artery occlusion shown by IV digital subtraction angiography. Am J Neuroradiol 1987;8: North American Symptomatic Carotid Endarterectomy Trial Collaborators. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. N Engl J Med 1991;325: Haynes B, Thorpe K, Taylor W, et al. Poor performance of ulttasound in detecting high-grade carotid stenosis [Abstract J. Can J Surg 1992;35: Middleton WD, Foley WD, Lawson TL. Color-flow Doppler imaging of carotid artery abnormalities. Am J Roentgenol 1988; 150: Submitted Nov. 6, 1992; accepted June 11, 1993.