The effect of unilateral internal carotid arterial occlusion upon contralateral duplex study: Criteria for accurate interpretation

Transcription

1 The effect of unilateral internal carotid arterial occlusion upon contralateral duplex study: Criteria for accurate interpretation Roy M. Fujitani, MD, Joseph L. MiLls, MD, Linda M. Wang, RID, and Spence M. Taylor, MD, Lackland AFB, San Antonio, Texas To determine the influence of unilateral internal carotid arterial occlusion (ICO) on Doppler frequency spectral analysis (DFSA) of the patent contralateral carotid artery, a retrospective review of 154 patients between July 1987 and December 1991 with angiographically confirmed ICO was performed, correlating duplex and arteriographic findings in a blinded fashion. Biplane arteriograrns and bilateral carotid artery duplex studies that used a 5.0 MHz Doppler probe with a 1.5 nun 3 sample volume at a 60 degree angle ofinsonation were performed on all patients. Each carotid artery was categorized by the severity ofstenosis as quantified by arteriography: 1% to 15% (n = 41); 16% to 49% (n = 48), 50% to 79% (n = 21), 80% to 99% (n = 34), and bilateral occlusion (n = 10). DFSA peak systolic frequencies were commonly exaggerated hi the presence of contralateral ICO and use of standard criteria for DFSA interpretation overestimated blfurcation stenoses in 43 of 89 lesions (48.3%) when determining nonhemodynamically significant lesions (<50% diameter reduction) with a sensitivity of only 57.3% and specificity of 96.9%. Conversely, prediction of hemodynamicauy significant lesions (> 50% diameter reduction) with standard criteria had 96.9% sensitivity but only 57.3% specificity. Modification of these criteria to account for the velocity increase or "jet effect" hi the ipsilateral carotid artery system increased the sensitivity and specificity to 97.8% in predicting nonhemodynamica/ly and hemodynamicauy significant stenoses respectively. A Doppler frequency spectrum with a peak systolic frequency (PSF) >4.0 khz and end-diastolic frequency (EDF) < 5 khz with an "open window" distinguished lesions with <50% diameter reduction. A PSF >4.5 khz with associated spectral broadening differentiated stenosis > 50%, whereas an increase in EDF > 5 khz further categorized those stenoses with 80% to 99% diameter reduction. Kappa statistics were sigulficandy different (p < 0.001) when validating these findings using the standard criteria (K = ) versus the modified criteria (K = ). We conclude that the presence of unilateral ICO influences the resultant flow velocity in the patent companion cerebral vessels, including the contralateral carotid artery and may lead to an overestimation of the actual degree of stenosis by standard DFSA criteria. A simple modification of the DFSA interpretation criteria permits more accurate differentiation between hemodynamically significant and nonhemodynamicalay significant lesions and is therefore important in the management of progressive extracranial cerebrovascular disease opposite an already occluded carotid artery system. (J VASC STJRG 1992;16: ) From the SGHSG/Secfion of Vascular Surgery, Wilford Ha~l USAF Medical Center, Lacldand AFB, Texas. The ~4ews expressed herein are those of the authors and do not reflect the offidal policy of the Department of Defense or other Departments of the U.S. Government. Presented at the Sixteenth Annual Meeting of the Southern Association for Vascular Surgery, St. Thomas, Virgin Islands, January 23-25, Reprint requests: Roy M. Fujitani, MD, SGHSG/Section of Vascular Surgery, Department of Surgery, Wilford Hall USAF Medical Center, Lacldand AFB, TX /6/40207 The ability m noninvasively screen for hemodynamically significant carotid bifurcation stenoses has been revolutionized by the technique of duplex scanning. The combination of B-mode real-time imaging and pulsed-wave Doppler spectral analysis has been established as an accurate, reproducible diagnostic test for detecting extracranial carotid artery disease. Clinical experience with the technique now allows reliable noninvasive classii%ation of 459

2 460 Fujitani et ai. Journal of VASCULAR SURGERY Table I. Standard Doppler frequency spectral criteria for dassification of carotid artery disease ~ Classification Arteriographic lesion Spectral criteria A B C D Normal 1%-15% Diameter reduction 16%-49% Diameter reduction 50%-79% Diameter reduction D+ 80%-99% Diameter reduction E Occlusion (100% diameter reduction) PSF <4 khz, (< 125 cm/sec) Minimal or no SB in deceleration phase of systole PSF < 4 khz, ( < 125 cm/sec) Minimal SB in deceleration phase of systole PSF < 4 khz, ( < 125 cm/sec) Increased SB throughout systole PSF > 4 khz, ( > 125 cm/sec) EDF <4.5 khz, (< 140 cm/sec) Marked SB is usually associated PSF >4 khz, ( > 125 cm/sec) EDF > 4.5 khz, ( > 140 cm/sec) tmarked SB No internal carotid flow signal Low or reversed diastolic component in common carotid artery Thump at the stump or origin of occlusion Modified from Zierler RE, Strandness DE Jr. In: Wood JH, ed. Cerebral blood flow physiologic and clinical aspects. New York: McGraw-Hill, 1987, Used with permission. PSF, peak systolic frequency; EDF, end-diastolic frequency; SB, spectral broadening. ~Frequency and velocity classification based on 5 MHz pulsed Doppler carrier frequency with a 1.5 mm a sample volume at a 60-degree angle of insonation. j-end-diastolic frequency and velocity values are only used as stenosis classification criteria for 80% to 99% diameter reduction lesions. Table II. Correlation of Doppler frequency spectral analysis and arteriography results with use of standard criteria with contralateral internal carotid artery occlusion Results from Doppler frequency spectral analysis, standard criteria ~ Arteriography results B C D D + E (% diameter reduction) No. pts. 1%-15%~ 16%-49% ~ 50%-79% ~ 80%-99% ~ lo0%p is-i5% 4i 34 (83) 5 (i2) 2 (5) C 16% -49% (25) 36 (75) D 50%-79% 21 1 (5) 1 (5) 16 (76) 3 (14) D + 80%-99% 34 3 (9) 31 (91) E 100% 10 1 (10) TOTAL (90) 9 K coefficient, Perfect agreement, 0.662; chance agreement, ~Predictive value (%) shown in parentheses. tpercentage of diameter reduction. arterial disease severity based on analysis of the frequency and amplitude of the Doppler flow signal, correlating with the degree of stenosis found by angiography. There have, however, been several anecdotal clinical reports of the diagnostic pitfalls created by severe contralateral carotid artery disease, often leading to an overestimation of the severity of ipsilateral stenosis. >s When such duplex studies are interpreted with use of preestablished Doppler frequency spectral analysis (DFSA) criteria, stenoses may be erroneously assigned to a higher category, resulting in a false-positive interpretation. These reports investigated the influence of a spectnun of contralateral carotid artery disease, ranging from moderate stenosis to total occlusion. 15 Only total contralateral occlusion uniformly affected DFSA by increasing the velocity in the ipsilateral carotid artery, whereas patent but severely stenotic contralateral arteries did not result in a consistent pattern of increased velocity. It has been postulated that contralateral carotid artery occlusion causes an increase in flow in the companion patent carotid and vertebral arteries, which function as collateral-vessels to the brain to maintain cerebral circulation through the circle of Willis. 1,2 This increased flow creates a velocity increase or "jet effect" in the ipsilateral carotid artery system that is disproportionately high

3 Volume 16 Number 3 September I992 Duplex crittha in contratateral carotid artery occlusion 461 Table III, Validation results of standard Doppler frequency spectral analysis with contralateral internal carotid artery occlusion Sensitivity,Specificity Accuracy Positive predictive Negative predictive Ipsilateral stenosis (%) (%) (%) value (%) value (%) 1%-15% %-49% %-79% %-99% % < 50% > 50% Table IV. Modified Doppler frequency spectral criteria for classification of carotid artery disease contralateral to an occluded internal carotid artery* Classification Arteriographic lesio'~e Modified spectral criteria BM DM D2M ~ Normal 1%-15% diameter reduction I6%-49% diameter reduction 50%-79% diameter reduction 80%-99% diameter redaction Occlusion (100% diameter reduction) PSF <4 khz, ( < 125 cm/sec) Minimal or no SB in deceleration phase of systole PSF <4 khz, (<125 cm/sec) Minimal SB in deceleration phase of systole PSF >4.0 khz, ( > 125 cm/sec) EDF < 5.0 ki-iz, ( < 155 cm/sec) Minimal SB in deceleration phase of systole (open window) PSF >4.5 khz, ( > 140 crn/sec) EDF < 5.0 khz, ( < 155 cm/sec) Marked SB is usually associated PSF >4.5 khz, ( > 140 cm/sec) EDF > 5.0 khz, ( > 155 cm/sec)t Marked SB No internal carotid flow signal Low or reversed diastolic component in common carotid artery Thump at the stump or origin of occlusion * Frequency and velocity classification based upon 5 MHz pulsed Doppler carrier frequency with a 1.5 mm a sample volume at a 60-degree angle of insonation. tend-diastolic frequency and velocity values are only used as stenosis classification criteria for 80% to 99% diameter reduction lesions. for the degree of stenosis actually present when interpreted by standard duplex Doppler sonographic criteria. 4 Only one report disputed tiffs finding, concluding that contralateral disease had no effect on ipsilateral hemodynamics or Doppler frequency. 6 It has been our experience in the vascular diagnostic laboratory that this phenomenon does indeed occur, most notably with contralateral carotid artery occlusion, and is responsible for frequent overestimation of ipsilateral extracranial carotid artery stenoses. We undertook this retrospective study to determine specifically the influence of contralateral internal carotid artery occlusion on the accuracy of duplex study interpretation and its effect on Doppler frequency spectral analysis. Because no DFSA criteria have been established in this clinical setting, we proposed to modify the existing criteria to optimize the sensitivity, specificity, and predictive value and thereby to study correlation and validation. MATERIAL AND METHODS The vascular laboratory records identifying 202 patients who had unilateral extracranial internal carotid artery occlusions by duplex scanning between July 1, 1987, and December 31, 1991, at Wilford Hall U.S. Air Force Medical Center were entered into this retrospective review. Only those patients who underwent angiographic studies confirming complete unilateral carotid artery occlusion (n = 154) within 6 weeks of the noninvasive tesmng were qualified for final inclusion. Indications for duplex examination included clinical suspicion for symptomatic carotid artery disease (n = 119), asymptomatic carotid artery bruits (n = 23), or preoperative

4 462 Fujitani et al,tournai of VASCULAR SURGERY Table V. Correlation of Doppler frequency spectral analysis and arteriography results with use of modified criteria with contralateral internal carotid artery occlusion Results from Doppler frequency spectral analysis, modified criteria ~ Arteriography results BM CM DM D~4 + E M (% diameter reduction) No. pts. I %-I 5 % ~ 16%-49%-/- 50%-79% t 80%-99%? 100% ~ B 1%-15% (83) 7 (17) C 16%-49% (96) 2 (4) D 50%-79 / (5) 1 (5) 19 (90) D + 80%-99% 34 3 (9) E 100% 10 TOTAL K coefficient, Perfect agreement, 0.903; Chance agreement, ~Predictive value (%) shown in parentheses. -~Percentage of diameter reduction. 31 (91) 1 (10) 35 9 (90) 9 Table VI. Validation results of modified Doppler frequency spectral analysis with contralateral internal carotid artery occlusion Ipsilateral Sensitivity Specificity Accuracy Positive predictive Negative predictive stenosis (%) (%) (%) value (%) value (%) 1%-15% %-49% %-79% %-99% % < 50% > 50% screening for patients with significant coronary or peripheral vascular disease (n = 12). Duplex examinations of bilateral extracranial carotid arteries were performed by an experienced vascular laboratory technician using either an ATL Ultramark 8 or an ATL Ultramark 9 (Advanced Technology Laboratories, BotheR, Wash.). Each machine provides simultaneous, high-resolution, real-time imaging (7.5 MHz transducer) and pulsed Doppler (5.0 MHz) digital fast Fourier spectrum analysis. Standard duplex techniques used real-time B-mode imaging in the sagittal and transverse planes with procurement of multiple correlative pulsed Doppler frequency measurements by use of a 1.5 mm 3 sample volume in the midstream of flow in each vessel. The carotid bifurcation was used as the anatomic reference point, with this site identified as the area of flow separation, allowing localization of both the internal and external carotid artery signals. The Doppler angle of incidence was assigned parallel to the direction of blood flow in the portion of the vessel being insonated and was made as close to 60 degrees as possible. Permanent printed copies of peak systolic and end-diastolic frequency data were recorded from the common carotid, internal carotid, and external carotid arteries. Further recordings were taken proximal to stenoses, at the site of stenoses, and after stenoses in the internal carotid artery as determined by real-time B-mode imaging. Special notation of any ulcerations and irregularities of the arterial luminal surface was also made. The duplex examination was always performed before the angiographic study, with a mean time interval of 12 days between the two studies (range 0 to 42 days). The criteria to determine contralateral internal carotid artery occlusion by duplex ultrasonographic data included the following: (1) failure to obtain a pulsed-wave Doppler signal from the internal carotid artery seen on real-time B-mode imaging despite increasing sample volume; (2) a low or reversed diastolic component in the common carotid artery, usually associated with a pronounced triphasic characteristic from primary external carotid artery outflow; and (3) an audible "thump" at the stump or origin of occlusion. 79 The degree ofipsilateral carotid artery stenosis, as

5 Volume 16 Number 3 September 1992 Duplex criteria in contralateral carotid e~ery occlusion 463 determined by duplex study, was quantified and assigned to one of five categories- 1% to 15%, 16% to 49%, 50% to 79%, 80% to 99%, and 100% (total occlusion) - according to the DFSA criteria estabfished by the University of Washington 1 (Table I) and logged in the vascular laboratory patient records. Conventional biplane arch and bilateral carotid artery magnification cut-film arteriography was performed in 138 patients and selective intra-arterial digital subtraction arteriography in 16 patients. Vertebral studies were obtained when dictated by the initial symptoms of the patient. Angiographic interpretation was performed independently by two of the coauthors (R.M.F. and L.M.W.) in a blinded fashion without reference to the findings of the noninvasive study. The maximum stenosis in the ipsilateral internal carotid artery was calculated by comparing the diameter at the point of minimal residual lumen (MRL) with the normal lumen (NL) in the internal carotid artery distal to the natural poststenotic dilation and all evident disease by caliper measurement? The degree of arteriographically confirmed stenosis was therefore determined by the formula: % Stenosis = (1 -MRL/NL) x 100 The degree of stenosis (diameter reduction) was estimated to the closest whole percentage. Each internal carotid artery was then categorized into one of five categories by the severity of stenosis as quantitated by arteriography: 1% to I5%, 16% to 49%, 50% to 79%, 80% to 99%, mad bilateral occlusion. When there was a discrepan T in the two individual angiographic interpretations in disease categorization, a third independent reading was performed (LL.M.) for deciding final categorization. Correlation between the noninvasive results and angiography, which served as the diagnostic standard, was determined by K (kappa) statistics, that is, the coefficient of agreement for nominal scales? ~ Perfect correlation woum yield a K value of + 1.0, whereas a perfect negative correlation would have a K value of A K value of 0.0 is consistent with a "perfecff' random distribution and therefore indicates a total lack of correlation. Therefore, K values :may vary from -1.0 through 0.0 to + 1.0, with correlation improving between the noninvasive test and the diagnostic standard (angiography) as it approaches Receiver-operating characteristic (ROC) curves were constructed to determine the DFSA criteria that would optimize test validity at various thresholds, n Validation data were expressed in sensitivity, specificity, accuracy, and positive and negative predictive values, which were expressed as percentages. RESULTS Of the 202 patients identified with at least unilateral total internal carotid artery occlusion by duplex examination, 154 patients had a corresponding bilateral carotid al~ceriogram within 6 weeks of the study confirming this finding. Only patients in this latter group were included in this retrospective review. The population included 115 men and 39 women with mean ages of 62.6 _+_ 8.4 and 61.4 _+ 5.7 years respectively. Unilateral internal carotid artery occlusions were found in 144 patients (68 right internal carotid artery-, 76 left internal carotid artery) and bilateral carotid artery- occlusions were confirmed in the remaining 10 patients. Table Ii compares the results of Doppler frequency spectral criteria interpretation by use of the standard criteria 1 (Table I) with the diameter reduction as measured on arteriography of 154 vessels contralateral to a completely occluded internal carotid artery. Perfect agreement was observed in 102 of the studies (66.2%) with a statistical chance agreement of 20A%. Fortysix (29.9%) of the duplex study results that used the standard criteria overestimated the degree of artery' stenosis and were allocated into a stenosis category' greater than that measured by arteriography. Of greatest interest are the 89 internal carotid arteries with nonhemodynamically significant lesions ( < 50% diameter reduction). DFSA interpretation overread 43 of 89 (48.3%) of the stenoses, accounting for the majority" (93.4%) of all false-positive studies. Seven of the 41 (17.1%) arterial stenoses in the 1% to 15% diameter stenosis category, 36 of 48 (75.0%) in the 16% to 49% category, and 3 of 21 (14.3%) of the 50% to 79% category" were erroneously assigned to a higher category. All of these false-positive readings were caused by an increased spectral frequency attributable to increased velocity in the ipsilateral internal carotid artery opposite flae occlusion. The K coefficient + SE was _ 0.113, indicating only a moderate level of correlation for reliable predictive value. The accuracy of standard DFSA in predicting associated ipsilateral carotid artery disease is summarized in Table III. Although the sensitivity of detecting > 50% diameter reduction was 96.9%, specificity was only 57.3%, with a positive predictive value of 62.4%. The high incidence of false-positive interpretations was evident by the 28.1% positive predictive value in the 50% to 79% category, indicating the misclassification of lesser diseased vessels into this category by standard

6 464 Fujitani et al. lournal of VASCULAR SURGERY DFSA criteria. The primary "problem" therefore appears to involve erroneous classification within the 16% to 49% and 50% to 79% categories because sensitivities and specificities for all other categories exceeded 90.0% with accuracy more than 95%. ROC curves were calculated with use of the existing and modified DFSA criteria to optimize the sensitivities and specificities for these varying degrees of stenosis. These modifications resulted in new thresholds as summarized in Table IV. Graphic illustrations of the Doppler-frequency spectra and corresponding arteriograms of the modified categories are shown in Figs. 1 to 3. The findings of the comparison between the Doppler frequency spectral interpretation using the modified criteria with the diameter reduction as measured on arteriography of these same 154 vessels contralateral to a completely occluded internal carotid artery are summarized in Table V. Perfect agreement was observed in 139 of the studies (90.3%) with a statistical chance agreement of 24.1%. Only nine (5.8%) of the duplex study findings overestimated the degree of artery stenosis and were allocated into a stenosis category greater than that measured by arteriography. All of these false-positive results fell in the nonhemodynamically significant (0% to 15% and 16% to 49% stenoses) categories. The K coefficient _+ SE was , consistent with a better correlation for reliable predictive value. The accuracy of the modified DFSA in predicting associated ipsilateral carotid artery disease with contralateral carotid artery occlusion is summarized in Table VI. Both sensitivities and specificities for each of the disease categories improved consistently to exceed 90%, including the ability to differentiate between hemodynamically significant and nonhemodynamically significant lesions, that is, < 50% and > 50% ipsilateral carotid artery stenoses. Overall, the accuracy improved from 74.0% to 97.4% in detecting these nonhemodynamically significant lesions by appropriate categorization of noninvasive Doppler frequency spectral examination results. DISCUSSION Duplex scan interrogation of the carotid bifurcation allows the use of quantitative physiologic parameters, including PSF, EDF, and spectral broadening (SB) obtained by pulsed-wave Doppler coupled with real-time B-mode imaging to add qualitative morphologic information. The established DSFA criteria to noninvasively predict carotid bifurcation disease have been well established ra and are generally reliable and reproducible. In the clinical situation where there is occlusion of the contralateral internal carotid artery, these standard criteria have not been reliable. This study supports the contention that a contralateral internal carotid artery occlusion influences the flow characteristics of the collateral vessels and notably the ipsilateral carotid artery, thereby decreasing the accuracy of pulsed-wave Doppler frequency spectral interpretation. This is most striking in the threshold distinguishing between hemodynamically significant lesions (50% to 79% diameter reduction; > 75% cross-sectional area reduction) and nonhemodynamicauy significant lesions (16% to 49% diameter reduction; <75% crosssectional area reduction), inasmuch as the majority of false positive noninvasive studies fall into the former category. Our findings are consistent with the reports by Cato et al. 1 and Spadone et al., 5 both suggesting that with more severe contralateral carotid artery disease, there is probable need to overread the Doppler frequency shifts by one category compared with patients with lesser degrees of disease in the opposite carotid artery. All previous studies that have examined the effect of contralateral carotid artery disease have compared the influence of a spectrum of contralateral carotid artery stenoses on the interpretation of ipsilateral carotid artery DFSA. ls Hayes et al. a used a continuous-wave system in a multicenter study and observed that, although all severe categories of contralateral stenosis proportionally affected the mean peak frequency of the ipsilateral carotid artery signal, this influence was most notable when the contralateral carotid artery was occluded. Standard DFSA criteria could be used to categorize those lesions at either end of the disease spectrum (i.e., 0% to 15% and 80% to 99%; 100% ipsilateral carotid artery diameter reduction) without significant compromise of either sensitivity or specificity. Comparison of the positive and negative predictive values of these categories shows this correlation (Tables III and VI). The most significant inconsistencies occurred in the 16% to 49% and 50% to 79% disease categories, in which 36 of 48 (75.0%) in the 16% to 49% category and 3 of 21 (14.3%) of the 50% to 79% were erroneously assigned to a higher category by standard DFSA criteria (Table II). In other words, 75% of the 16% to 49% arteriographically confirmed lesions and 14.3% of the 50% to 79% arteriographically confirmed lesions were interpreted as belonging to the 50% to 79% and 80% to 99% categories respectively by standard DFSA criteria. This finding is consistent with other reports that show similar distribution of false-positive results. 1,3'5 Simply downgrading one disease category was not optimal,

7 Volume!6 Number 3 September 1992 Duplex criteria in contralateral carotid artery occlusion 465 Fig. 1. Pulsed-wave Doppler frequency spectrum correlating with "CM" lesion by modified criteria (16% to 49% diameter reduction) shown with associated 40% internal carotid artery stenosis contralateral to confirmed carotid artery occlusion. This would have been overclassified as "D" lesion by standard criteria (50% to 79% diameter reduction). because it resulted in much overlap, and this led us to develop modified criteria in interpreting ipsilateral carotid artery stenosis in the face of a contralateral internal carotid artery occlusion (Table IV). The correlation of angiography and DFSA vdth use of these modified criteria established a K coefficient of (Table V). Characteristically, the Doppler frequency spectral waveform corresponding to a 16% to 49% diameter reduction lesion opposite an occluded internal carotid artery will have an increased PSF >4.0 khz, EDF < 5.0 khz, with an "open window" (minimal or no SB) under the systolic peak (Fig. 1). This Doppler frequency spectral waveform is sometimes called an "open window D" when adopting the standard nomenclature, noting its association with a lower category of disease (16% to 49% s~enosis). A stenosis of 50% to 79% contralateral to an occluded internal carotid artery can be accurately classified if the Doppler frequency spectra shows a PSF >4.5 khz and an EDF <5.0 khz, which is usually associated with marked SB or "closing of the vdndow" (Fig. 2). A stenosis of 80% to 99% can be differentiated by noting an increase of EDF > 5.0 khz and is associated with marked SB (Fig. 3). Although it has been theorized that the changes in the Doppler frequency or velocity spectra are based on increased velocity in the ipsilateral carotid artery resulting from occlusion of the contralateral carotid artery, ~,a,4 it is important to realize that there should also be an accompanying increase in flow velocity" in the vertebrobasilar system. Unquestionably, there are multiple factors that influence the altered hemodynamics after the occlusion of a primary extracranial arterial conduit, including the condition of the posterior circulation. The overall result appears to be a compensatory increase in flow velocity to maintain a homeostatic circulatory state. Finally, it is important to realize that there are several technical pitfalls that may contribute to erroneous procurement or interpretation of DFSA data. 4'12 A review of the equation used to calculate frequency change (/if) helps to understand these factors: AF = 2VFocos 0/C. The magnitude of the change in frequency is dependent on (1) the speed of sound in tissue (C), (2) the velocity of the flowing blood (V), (3) the frequency of the transducer (Fo), and (4) the angle of incidence of the Doppler beam relative to the direction of blood flow (0). 12 Accurate vessel identification, carefia/ positioning of the pulsed-wave Doppler sample volume of appropriate size in the center of the flow stream, and attention to recording the Doppler frequency signal at a beam angle of 60 degrees are all very important components of duplex examination to avoid erroneous interpretation of study findings. Assignment of an inaccurate angle of insonation wider than the standard 60 degrees will commonly

8 466 Fujitani et al. Journal of VASCULAR SURGERY Fig. 2. Pulsed-wave Doppler frequency spectrum correlating with "DM" lesion by modified criteria (50% to 79% diameter reduction) shown with associated 75% internal carotid artery stenosis. Increased spectral broadening ("closing of the window") with a peak systolic frequency > 4.5 khz and an end-diastolic frequency < 5.0 khz differentiates this from C M lesions. Fig. 3. Pulsed-wave Doppler frequency spectrum correlating with "Dlvl +" lesion by modified criteria (80% to 99% diameter reduction) shown with associated preocclusive internal carotid artery stenosis. Spectral broadening is marked with a peak systolic frequency > 4.5 khz and an end-diastolic frequency > 5.0 khz. cause falsely elevated frequency or velocity changes. This usually occurs when encountering tortuous vessels in which it is difficult to set the angular correction to reflect the true direction of blood flow throughout a sharp curvature in the vessel. All proposed DFSA criteria in this study are based on a 5 MHz pulsed Doppler carrier frequency with a 1.5 mm 3 samplc volume at a 60 degree angle of insonation. The duplex scanner offers the advantage of allowing the concomitant use of real-time B-mode imaging to help identify vessel tormousity and morphologic condition, thereby minimizing erroneous calculations of Doppler frequency spectra. Additional reliance upon B-mode ultrasonography images to determine the percent of luminal stenosis may help with appropriate categorization of these lesions;

9 Volume 16 Number 3 September 1992 Duplex criteria in contralateral carotid artery occlusion 467 however, this is not always possible and DFSA is an important component in these noninvasivc studies. Our study found that unilateral internal carotid artery occlusion influences the resultant flow velocity in the patent collateral extracranial cerebral circulation including the contralateral carotid artery and leads to an overestimation of the actual degree of stenosis by standard DFSA criteria. This is of particular importance in serial examinations of patients for disease progression opposite an already occluded carotid artery system. Before these proposed modified criteria are adopted, a prospective study will be undertaken to further refine test sensitivity, specificity, and overall accuracy. REFERENCES 1. Cat{} FR, Bandyk DF, Livigni D, et al. Carotid collateral circulation decreases the diagnostic accuracy of duplex scanning. Bruit 1986;10: Forconi S, Johnston KW. Effect of contralateral internal carotid stenosis on the accuracy of continuous wave Doppler spectral analysis results. J Cardiovasc Surg 1987;28: Hayes AC, Johnston KW, Baker WH, et al. The effect of contralateral disease on carotid Doppler frequency. Surgery 1988;103: Beckett WW, Davis PC, Hoffman IC Jr. Duplex Doppler sonography of the carotid artery: false-positive results in an artery contralateral to an artery with marked stenosis. AJNR 1990;11: Spadone DP, Barkmeier LD, Hodgson KJ, eta/. Contralateral internal carotid artet3~ stenosis or occlusion: pitfall of correct ipsilateral classification - a study performed with color-flow imaging. J VAse SURG 1990;11: Fisher M, Alexander K. Influence ofcontralateral obstructions on Doppler-frequency spectral analysis of ipsilateral stenoses of the carotid arteries. Stroke 1985;16: Bandyk DF, Levine AW, Pohl L, et al. Classification of carotid bifurcation disease using quantitative Doppler spectrum analysis. Arch Surg 1985;120: Bornstein NM, Beloev ZG, Norris JW. 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: Zierler RE, Strandness DE Jr. Noninvasive dynamic and real-time assessment of extracranial cerebrovasculature. In: Wood JI-I, ed. Cerebral blood flow: physiologic and clinical aspects. New York: McGraw Hill, 1987; Sumner DS. Evaluation of noninvasive procedures: data analysis and interpretation. In: Bernstein EF, ed. Noninvasive techniques in vascular surgery. St Louis: CV Mosby, 1985: Nelson TR, Pretorius DH. The Doppler signal: where does it come from and what does it mean? AIR 1988;151: Submitted Jan. 29, 1992; accepted June 11, DISCUSSION Dr. William M. Blackshear (Tampa, Fla.). Thank you, Dr. Edwards. I would like to congratulate Dr. Fujitani and his colleages for an in-depth study of a vexing clinical diagnostic problem, and I appreciate the opportunity to review their article. This study was undertaken because of the empiric observation noted by others that, in the presence of a unilateral internal carotid artery occlusion, flow velocities in the contralateral internal carotid artery sometimes seem to be disproportionately elevated by duplex scanning, presumably because of an increase in flow velocity to compensate for loss of contralateral perfusion to the brain. The authors suggest that application of one set of standard diagnostic criteria to these vessels may result in overestimation of the degree of internal carotid artery stenosis. In their study, use of the standard criteria resulted in the misclassification of 75% of category C (< 50%) stenoses into high grade category D ( > 50%) lesions. This is an error of considerable clinical importance because the results of duplex scanning are often used to guide patient management: and to select patients for angiography and, occasionally, for operation. We have also noted this phenomenon of increased flow velocity on occasion and have generally found it to be most evident when significant cross-perfusion through the anterior communicating artery can be shown angiographically. It is not a universal observation, however. We briefly reviewed our 20 most recent duplex scans in patients with unilateral internal carotid artery occlusions, and we noted only one patient with a contralateral stenosis, which was classified as high grade and this was angiographically confirmed. Therefore our data are somewhat at variance with Dr. Fujitani's and suggest that the problem may not be quite as widespread as he noted. One explanation for the difference in our findings may be the degree of intracranial couateralization. Many patients with ICA occlusion perfuse the affected hemisphere principally through ipsilateral periorbital or posterior circulation collaterals. In these patients, who are not uncommon, flow velocity in the contralateral internal carotid artery should be affected very little, if at all. In view of the authors' results, which show a dramatic improvement in the ability to correctly classify category C and D lesions by increasing the systolic frequency criteria to 4.5 khz, I wonder how the authors can explain their success in virtually all patients, many of whom

10 468 Fujitani et al loumal of VASCULAR SURGERY presumably rely on vertebral collaterals for cerebral perfusion. I suggest that there may be an alternative explanation for their observations that has to do with the diagnostic criteria. Duplex scanning has taught us that we should continually question our "gold standard" which initially was arteriogtaphy, and the same applies to duplex scanning. Results from our laboratory, which we reported at this meeting 3 years ago, suggested that a Doppler frequency of 5.0 khz, not 4.0, was the best for discriminating between stenoses greater or less than 50%. Exclusive of peak frequency criteria, the primary discriminant between high- or low-grade stenoses in their report was a subjective evaluation of the degree of spectral broadening. Therefore merely raising the systolic frequency criteria to 5 khz would, in and of itself, eliminate many of the false-positive results they noted independent of the presence of contralateral disease. I would be interested in the author's comments on this issue because the great majority of their diagnostic errors were in discriminating between stenoses greater or less than 50%, tn addition, have you validated the so-called standard criteria in your laboratory population who do not have contralateral internal carotid artery occlusion? This is a complex issue and it is also possible that the method of grading stenoses by angiography may play a role in the categorization problems. The method selected by the authors, which is widely accepted, is conservative and it does, on occasion, underestimate the severity of plaque accumulation in the normally bulbous proximal internal carotid artery. The authors conclude their article by indicating that they are undertaking a prospective study of this problem. I suggest that they prospectively evaluate their diagnostic criteria for all degrees of carotid artery disease, not only those with unilateral internal carotid artery occlusion. Nevertheless, in spite of my concerns, this article is a refreshing look at a very difficult and important clinical problem. Dr. Roy Fujitani. Dr, Blackshear has outlined very insightful and important observations by use of the duplex scanner in identifying carotid bifurcation disease. The observation of increased Doppler spectral frequency and velocity seems to be related to the compensatory increased flow within the carotid artery contralateral to the occlusion. Because the intracranial circulation is normally supplied by four major extracraniai vessels communicating through the circle of Willis, it allows for collateral flow in the case of proximal arterial occlusion. This hemodynamic influence on the collateral arteries, including the vertebral arteries, should therefore be reflected in the Doppler frequency and velocity pattern obtained by duplex scanning. This appears to be the primary cause for the misclassification errors found in this group of patients identified with unilateral carotid artery occlusion. Our method of estimating the degree of contralateral internal carotid artery stenosis on biplanar contrast arteriography was determined by caliper measurement of the most severe area of lumen narrowing on a single plane film, with stenosis expressed as a percentage of diameter reduction relative to the distal normal vessel. This technique estimates carotid bifurcation stenosis and has been validated previously. We found this method most quantitatively reproducible, even when taking into account interobserver and intraobserver variability in interpretation. Although measurement of the plaque at the level of most severe involvement would be' most precise, its accuracy may still be challenged because the actual diameter of the artery at this level would, at best, still have to be estimated. We have adapted the standard criteria for DFSA in our vascular diagnostic laboratory, as initially determined by the University of Washington group, using the duplex scanner "with very good correlation in predicting hemodynamically significant carotid artery stenoses. This study was done because of the frequent observation of increases in flow and Doppler-derived frequency and velocity that were disproportionately high for the degree of stenosis actually present in the contralateral carotid artery. The use of any diagnostic technique requires recognition of its limitations and potential pitfalls and should therefore prompt each vascular laboratory to perform validation studies of diagnostic accuracy. Therefore we have proposed that a prospective study be undertaken to further refine the sensitivity, specificity, and overall accuracy of the proposed criteria. It will also allow for the possible determination of the various factors that may contribute to these hemodynamic changes found with unilateral extracranial carotid arterial occlusion.