The natural history of asymptomatic carotid artery disease

Transcription

1 The natural history of asymptomatic carotid artery disease Richard W. Bock, MD, Anthony C. Gray-Weale, FRACS, Philip A. Mock, M App Stats, David A. Robinson, MB, BS, Les Irwig, MB, BCh, PhD, FCCM, and Robert J. Lusby, MD, FRCS, FRACS, Sydney, Australia Purpose: The purpose of this article is to determine the natural history of carotid artery disease among asymptomatic patients with cervical bruits or other risk factors for stroke and to study the value of duplex ultrasonography in predicting future neurologic events. Methods: Two hundred forty-two asymptomatic, unoperated patients, referred for evaluation of asymptomatic carotid artery disease, were followed prospectively with duplex ultrasonography. Results: Fifteen ischemic strokes (6.2%) and 20 transient ischemic attacks (TIA) (8.3%) occurred in 34 patients during a mean follow-up of 27.4 months. Annual stroke, TIA, and combined event rates were 2.7%, 3.6%, and 6.2%, respectively. Although patients with 80% to 99 /6 lesions had a 20.6% annual event rate, most events occurred contralateral to these lesions; the vessel-specific annual event rate for 80% to 99% disease was 5.1%. Only one of 15 strokes occurred ipsilateral to an 80% to 99% stenosis. Echolucent plaques were associated with TIA and stroke (5.7% annual vessel event rate vs 2.4% for echogenic plaques, p = 0.03). Disease progression was highly correlated with TIA and stroke (to < ), but it usually occurred in association with rather than before ischemic events, thus proving more useful in explaining pathogenesis than in predicting future events. There was no association between aspirin use and TIA, but patients taking aspirin had a threefold higher annual stroke rate (1.6% vs 4.8%,p = 0.027). Conclusions: This study, while confirming significant risk for asymptomatic patients with critical stenosis or echolucent plaque, demonstrates the importance of contralateral disease and the absence of orderly progression from minimal disease through high-grade stenosis to symptomatic cerebral ischemia. TIA and stroke commonly occur in association with abrupt, unpredictable, quantum changes in carotid artery disease. (J VAse SURG 1993;17: ) Extracranial cerebrovascular disease is estimated to cause up to 70% of ischemic strokes, ls Recent prospective trials in patients with carotid artery atherosclerosis have clearly demonstrated the superiority of carotid endarterectomy over medical therapy in preventing stroke among patients with severe stenosis and transient ischemic attack (TIA).46 However, stroke is preceded by TIA in only about 10% of patients in community-based studies 7,s and in 15 % to 35% of patients in hospital-based surveys. 9,~ Thus From the Department of Surgery, University of Sydney. Supported by grants from W.L. Gore & Associates, Flagstaff, Ariz., and the (Australian) National Health and Medical Research Council, and by the Australian Department of Veteran's Affairs (BOMRAC). Presented at the Forty-sixth Annual Meeting of the Society for Vascular Surgery, Chicago, Ill., June 8-9, Reprint requests: Richard W. Bock, MD, Department of Surgery, School of Medicine, University of Louisville, 550 S. ~ackson St., Louisville, KY /6/ the overall impact on the public health resulting from successful prevention of stroke among patients with TIAs will likely be relatively limited. To achieve a significant decrease in the overall morbidity and mortality of stroke, further effort must be directed toward the prevention of stroke among patients with asymptomatic cerebrovascular atherosclerosis. Intense controversy persists regarding appropriate management of asymptomatic carotid artery disease. Aspirin has never been shown to be effective in preventing stroke among neurologically asymptomatic individuals and may introduce distinct risks, whereas the use of carotid endarterectomy among this group has been criticized as unproven and unsafe. ~ To determine the best preventative therapy for the asymptomatic population, the natural history of stroke in these patients must first be elucidated. Most natural history studies have used indirect testing methods, 12a* which have limited accuracy in /93/$

2 Volume 17 Number 1 January ],993 Natural history of asymptomatic carotid artery disease 161 quanff[ying carotid artery disease. Duplex ultrasonography, a relatively new means of evaluating cerebrovascular disease, is noninvasive, accurate, and reproducible Its widespread acceptance, coupled with the aging of the population, has generated a large and growing number of referrals for detection and quantification of asymptomafic disease. But the utility of duplex data in asymptomatic patients remains uncertain, because few natural history studies have used this method. To determine the implications of duplex ultrasound test results in asymptomatic individuals who are at risk for stroke, and m provide a groundwork for furore efforts in designing rational strategies of stroke prophylaxis among asymptomatic populations, we prospectively studied the natural history of asymptomatic carotid artery disease in a group of patients referred for evaluation of cervical bruits or other risk factors for stroke. PATIENTS AND METHODS From March 1985 through March 1991, 2350 patients underwent duplex evaluation of the carotid arteries at the Vascular Diagnostic Laboratory of the University of Sydney. Most (2108 patients, 89.7%) had symptoms and were referred after developing TIA, non-hemispheric symptoms, or stroke. Two hundred forty-two neurologically asymptomatic patients (10.3%), all referred for evaluation of carotid bruit or other risk factor for stroke, were!followed prospectively with annual carotid duplex: examinations. Only asymptomatic patients, defined as those with no recent history of TIA, amaurosis fugax, or stroke, were entered into the study. Patients with a history of carotid endarterectomy on either side were excluded, as were patients with vertebrobasilar or "non-hemispheric" cerebrovascular symptoms. Patients with a remote history of minor stroke or TLA were entered into the study if they were ambulatory, living independently, and without any new neurologic symptoms for 2 years before entry, because such patients would likely be considered asymptomatic, rather than symptomatic, for any purposes of therapeutic decision-making. At the initial visit each patient's asymptomatic status was confirmed with a standard, detailed, oral questionnaire administered by a physician, and data regarding coexisting disease and risk factors were recorded. At each annual follow-up visit, subjects were thoroughly questioned regarding new neurologic symptoms and changes in coexisting disease. Information regarding aspirin use was ob- tained by patient history alone, and all decisions regarding aspirin therapy were left to the referring physician. Duplex examination. Examination of the carotid arteries, utilizing a single duplex instrument combining high-resolution B-mode ultrasound with pulsed-wave Doppler and spectral analysis, was performed at each initial and annual follow-up visit. Color-flow imaging was added in Standard spectral criteria, first proposed by Roederer et al.,18 were used for all measurements of stenosis. Carotid artery stenosis was expressed as percentage of maximal decrease in diameter of the internal or common carotid artery and was classified into five categories: 0% to 15%, 16% to 49%, 50% to 79%, and 80% to 99% stenosis, and complete occlusion. Disease progression was defined as any increase in stenosis except between the 0% to 15% and 16% to 49% categories, because the accuracy of duplex ultrasonography in distinguishing between these two non-flow-significant categories is probably much less than in flow-significant (>50%) stenosis. 19 Carotid artery plaque morphology was evaluated with gray-scale B-mode ultrasonography alone, 2 and plaques were classified as either predominantly echogenic or predominantly echolucent. Vessels with echogenic vessel walls but without discernible plaque were considered echogenic. No attempt was made to evaluate plaque morphology in occluded vessels. Study endpoints. Absolute endpoints were TIA (including amaurosis fugax) and ischemic stroke. TIA was defined as an episode of unilateral motor deficit, with or without a sensory or aphasic component, lasting less than 24 hours. Amaurosis fugax was defined as transient monocular loss of vision. When any of these lateralizing symptoms lasted more than 24 hours, the event was classified as a stroke. Diagnoses of TIA and stroke were made clinically, by history and examination, and decisions regarding subsequent cardiac or other investigations were left to the primary physicians. Any non-hemispheric or non-lateralizing cerebrovascular symptoms (syncope, dizziness, bilateral sensorimotor symptoms, or bilateral blurred vision) that developed after entry into the study were excluded from tabulation Patients were excluded from further analysis if they underwent carotid endarterectomy. Except for one patient who suffered an immediately fatal stroke, all patients reaching study endpoints were evaluated after the event at our institution. Data analysis. Two fundamental decisions were made regarding proper tabulation of patients, duplex

3 162 Bock et al. Journal of VASCULAR SURGERY scans, and events. First, it was noted that the use of initial scans only in determining risk of subsequent TIA and stroke would ignore any contribution from asymptomatic progression of disease. This would skew study results by obscuring all event-free periods of observation that occurred after disease progression. (For example, one patient in the study progressed asymptomatically from an initial stenosis of 16% to 49% to a stenosis of 80% to 99% at the first follow-up scan and experienced 5 subsequent eventfree years without further change; analysis by initial scan only would disregard the benign course of his 80% to 99% lesion.) For this reason all data were tabulated in annual increments, counting each 1-year period equally in compiling results. Second, all ischemic events were charged to the duplex scan performed before the event, and all TIA and stroke rates were computed on this basis. Duplex data regarding the vessel status after an ischemic event were regarded as clinically irrelevant and were used only in a separate analysis of disease progression and the pathogenesis of stroke. Survival curves were calculated with the Kaplan- Meier product limit method 21 whereas comparisons (homogeneity and linear trend) of survival curves for the basefine stenosis categories were performed by use of log-rank tests with the statistical package BMDP1L (BMDP Statistical Software, Inc., Los Angeles, Calif.). 22 Likelihood ratio chi-squared tests were used to test linear trends for event rates among categories of stenosis. Pearson chi-squared tests were used to compare demographic and other risk factors with development of TIA and stroke. RESULTS Characteristics at entry. Two hundred fortytwo patients were monitored prospectively for up to 6 years (mean 27.4 months, SD 14.5). The average age at entry was 68.1 years (range 34 to 89), and 191 subjects (78%) were men. One hundred seventy-two patients (71%) were referred for evaluation of carotid artery bruits, whereas 15 (6.1%) were referred for preoperative evaluation before coronary artery bypass grafting (CABG) or other major vascular surgery. Forty (16%) underwent duplex studies as part of an evaluation for other, nonsurgical problems (peripheral vascular disease not requiring operation, hyperlipidemia, hypertension, etc.), and 15 (6.1%) were studied as part of a specific, institutional protocol for patients with a combination of diabetes and hypertension. Hypertension was present in 144 patients (59%), whereas 139 (57%) carried a diagnosis of ischemic heart disease. A history of symptomatic peripheral vascular (aortoiliofemoral) disease was present in 119 patients (49%), and 34 had undergone CABG before enrollment (14%). There were 56 patients with diabetes (23%). Forty-three patients in the study had never smoked cigarettes (18%), 114 had ceased smoking more than 1 year before entry (47%), and 85 (35%) had smoked within the previous year. Thirty-seven patients (15%), all ambulatory and living independently, had suffered a minor stroke more than 2 years previously (mean 5.9 years before entry into the study). At entry into the study, 65 patients (27%) were taking aspirin on a regular basis. The distribution of carotid artery stenosis at enrollment is shown in Fig. 1. Data were first analyzed for each patient, with each subject categorized according to his or her most highly stenosed vessel. Each vessel was tabulated separately in a second analysis. Plaque morphology, assessed by B-mode ultrasonography, was independently analyzed. At entry, carotid artery plaques in 83 (17%) of 484 vessels were predominantly echolucent, whereas 378 (78%) were echogenic. Plaque morphology was not assessed in the 23 (5%) internal carotid arteries found to be occluded. Endpoints. Over the course of the study, 15 patients suffered stroke (6.2%). Twenty (8.3%) experienced TIAs; a contralateral stroke developed in one of the patients 2 years later. Thus the overall event rate (TIA plus stroke) was 14.0% (34/242) over the 27.4-month follow-up period. Annual event rates were 2.7% for stroke, 3.6% for TIA, and 6.2% for all ischemic events. There were no hemorrhagic strokes and no brainstem or lacunar infarcts. No stroke was heralded by ipsilateral TIA, although 11 patients underwent carotid endarterectomy after TIA. Fifteen other patients (6.2%) underwent endarterectomy for asymptomatic disease after entering the study, including five operations performed concurrently with CABG-one before abdominal aneurysm repair, one for loud pulsatile tinnitus, and three in patients who developed symptoms of vertebrobasilar ischemia (which was not considered symptomatic disease by the study criteria). These 15 endarterectomies were performed an average of 7 months after entry into the study; carotid artery stenosis was 50% to 79% in seven patients and 80% to 99% in eight. Kaplan-Meier survival curves for stroke and all cerebral ischemic events are shown in Fig. 2. Annual incidence of TIA and stroke was first analyzed by patient; in this method of analysis, each

4 Volume 17 Number 1 January 1993 Natural history of asymptomatic carotid artery disease 163 % a ~;~ N i:{i~ 0-15 ag{ OCCL PERCENTAGE STENOSIS --1 By patient (analyzed by most ~ By vessel highly stenosed vessel) ~ N [] 484 N [] 242 Fig. 1. Distribution of carotid artery stenosis on entry into the study. patient was classified by his or her most highly stenosed side (Table I). Incidence of TLA did not rise with increasing degree ofstenosis (p = 0.9), because most TLAs occurred in patients with maximal 20% to 49% disease. Both stroke and overall event rates did show an upward trend with increasing stenosis (p = andp = 0.04, respectively). Contralateral events and disease. To control for any contribution from events occurring contralateral to patients' most highly stenosed vessel, the data were reanalyzed, with ischemic events tabulated per vessel rather than per patient (Table II). Although a good correlation between carotid artery, stenosis and development of TIA and stroke had been demonstrated in the overall patient analysis, there was a surprising lack of correlation between these neurologic events and stenosis of the ipsilateral vessel. The ipsilateral vessel event rate rose only moderately with increasing degrees of stenosis; a trend was barely detectable (p = 0.05). The event rate for vessels with 80% to 99% stenosis was one fourth that observed in the patient-based analysis, whereas vessels with "stable," old occlusions had the highest rate. To resolve the difference between patient- and vessel-based event rates, the annual rate of develop- ment of contralateral symptoms was determined (Table HI). The likelihood of an event occurring contralateral to a vessel with 80% to 99% stenosis was 15.4%, three times the ipsilateral event rate. This finding was not accounted for by bilateral high-grade disease, because five of six patients with events occurring contralateral to an 80% to 99% lesion had less disease on the other (symptomatic) side. A graphic summary of data from Tables I, II, and III is shown in Fig. 3. The status of the contralateral vessel associated with each TIA and stroke is depicted in Table IV. Among 35 ischemic events overall, 10 (29%) occurred distal to the patient's most highly stenosed vessel, whereas 12 (34%) were contralateral to the side of greatest disease. Thirteen events (37%) occurred in patients with the same degree of stenosis on each side. More than half of all strokes occurred in vessel territories with 0% to 15% or 16% to 49% stenosis, yet only four patients who suffered a stroke had bilateral stenosis of less than 50%. Plaque morphology. The relationship between annual vessel event rates and carotid artery plaque morphology is shown in Table V. Vessels with echolucent carotid artery plaques were associated

5 164 Bock et al. Journal of VASCULAR SURGERY _J > > 0.8- tr --~ [49 L~49 Stroke All ischemic events o I I I I I I I MONTHS Fig. 2. Life-table curves for all events (TIA plus stroke). Number of patients at risk is shown at intervals along each curve. Table I. Annual TIA, stroke, and overall event rates analyzed per patient Overalls TIA * Stroke~ (TIA + stroke) Most highly Annual Annual Annual stenosed rate rate rate vessel (%) n Person-years (%) n Person-years (%) n Person-years (%) Occluded Total I[ Patients are classified by the most highly stenosed vessel side, so events occurring on the side of a less-diseased, contralateral vessel escape independent analysis. *p = 0.07; p (trend) = 0.9. tp = 0.03;p (trend) = :~p = 0.03;p (trend) = Person-years for stroke were slightly greater because observation after TIA was continued for stroke analysis only. [[One patient who experienced a TIA followed by a contralateral stroke 2 years later is counted once in overall event analysis. with a 5.7% annual rate of TIA and stroke, significantly greater than the 2.4% rate found among vessels with echogenic plaque (p = 0.03). Eight (24%) of 34 patients who experienced events had a change in plaque morphology diagnosed at the post-event scan (five from echogenlc to echolucent, three from echolucent to echogenic). Disease progression and natural history of stroke. Fig. 4 depicts serial duplex scans for each of the 15 patients who suffered stroke (as diagnosed by clinical examination) after entry into the study. Significant disease progression occurred in two (13%) of 15 patients before the stroke. Post-stroke scans (open circles) were available for all but one patient; seven (50%) of these 14 scans showed progression on at least one side. Of 20 patients experiencing TIA, progression occurred before the event in three (15%), but 11 (55%) showed progression on the post-stroke scan. Overall, 164 vessel scans demonstrated progression, of which 18 (11%) were associated with an ischemic event. Among 940 progression-free scans, 16 (1.7%) events occurred. Disease progression was highly correlated with TIA and stroke (p < ), but because it was diagnosed before the event in only five (15%) of 34 patients, it was not a useful predictor of furore cerebral ischemic events. Risk factors for TIA and stroke. Twenty-three patients who had 80% to 99% carotid artery stenosis for at least 1 year during the study supplied 39

6 Volume 17 Number 1 January 1993 Natural history of asymptomatic carotid artery disease 165 Table II. Annual incidence of events (TIA plus stroke) analyzed per vessel Vessel Ipsilateral stenosis event rate (%) Events Vessel-years (%) Occluded Total p = 0.09;p (trend) = Table III. Annual incidence of events (TIA plus stroke) contralateral to the index vessel Vessel Contralateral stenosis event rate (%) Events Vessel-years (%) ~ Occluded Total ~p = 0.006; p (trend) = vessel-years for analysis of critical stenosis. In this vessel-based analysis, 80% to 99% stenosis was associated with an annual ipsilateral event rate of 5.1% (one TIA and one stroke per 39 vessel-years of observation). This was not significantly greater than the 2.8% rate found with lesser degrees of stenosis (17 TIAs and 11 strokes per 992 vessel-years, p = 0.32). When patients, rather than vessels, with 80% to 99% stenosis were compared to those with lesser degrees of disease (a mode of analysis that includes events occurring contralateral to 80% to 99% lesions), there was a significantly greater annual event rate with critical stenosis (7 events per 34 patient-years, 20.6%, vs 22 events per 456 patientyears, 4.8%, respectively;p = 0.002). Occluded (and therefore essentially inoperable) vessels were grouped separately; the annual event rate ipsilateral to carotid artery occlusion was 5.5% (one TIA and three strokes per 73 vessel-years). Age, sex, hypertension, peripheral vascular disease, and obesity were not independently predictive of TIA or stroke within the study population. A history of cigarette smoking was present in 82% of patients, but continued smoking posed no additional risk; all but two of the 12 patients who suffered a stroke and who had a history of smoking had quit at least 1 year before suffering stroke. A remote history of minor stroke was related to overall new ischemic events (p = 0.01) but was not related to development of new stroke (p = 0.2). All four new strokes that occurred in patients with a remote history of minor stroke occurred on the side contralateral to the old event. Rates of development of TIA and stroke in patients with cervical bruit (24/172, 13.9%) were almost identical to those seen in patients referred for other reasons (10/70, 14.3%). There was no association between the use of aspirin and development of TIA; the annual rate of TIA for patients who took aspirin was 3.2% (six TIAs per 186 patient-years) versus 3.8% for patients not taking aspirin (14 TIAs per 366 patient-years, p = 0.72). There was, however, a significant association between aspirin use and stroke: among patients not taking aspirin the annual stroke rate was 1.6% (six strokes per 366 patient-years), whereas the annual stroke rate for patients who took aspirin was 4.8% (nine strokes per 186 patient-years, p = 0.027). The rate of aspirin use rose only slightly with increasing stenosis (a 10% difference between highest- and lowest-use groups), but the relatively low number of strokes precluded formal statistical control for this variable. DISCUSSION Only one prospective study using serial duplex examination of asymptomic, unoperated patients has heretofore been reported. Roederer et al.ls monitored 162 such patients who presented with asymptomatic cervical bruit. Data regarding disease progression seemed to implicate gradual, stepwise increases in stenosis, which progressed to critical (80% to 99%) disease, followed later by occlusion, cerebral ischemic symptoms, or both. But several factors limited the study's use in defining this natural history. Only four strokes and six TIAs occurred during the study. Serial examinations were available only in the 103 patients (61%) who returned for the first follow-up duplex study. Twenty percent of patients underwent carotid endarterectomy for asymptomatic disease and were kept in the study after operation. Mean follow-up was not reported. TIA and stroke were combined with asymptomatic carotid artery occlusion to lend statistical support to the conclusion that patients with 80% to 99% stenosis should undergo prophylactic carotid endarterectomy. A retrospective report from the same institution, 23 reviewing results of operative versus medical treatment of a subpopulation of patients with 80% to 99% carotid artery stenosis, also implicated critical stenosis as a harbinger of stroke, but inclusion of

7 166 Bock et al. Journal of VASCULAR SURGERY 20 Event rate per patient (classified by most highly stenosed vessel) A - - Event rate pervessel... Contralateral event rate per vessel LU I.-,< 15 to" Z,1110 > LM I-- \,\ I._1 z z < 5 n : s S ml.wat" I"gmlm ~ S m...,," " "' - ~, I I I I Occluded PERCENTAGE STENOSIS Fig. 3. Annual event rates (TIA plus stroke). Solid line depicts rate per patient, whereas dashed line represents rate per vessel. Difference between two is explained by dotted line, which is contralateral event rate per vessel. patients with non-hemispheric symptoms and patients who had undergone previous contralateral endarterectomy limited the applicability of this review to the natural history of asymptomatic, unoperated carotid artery disease. In our prospective study, 15 strokes and 20 TIAs occurred in 34 of 242 patients monitored with serial duplex ultrasonography. Patients with nonhemispheric symptoms or a history of endarterectomy were excluded, and silent (asymptomatic) carotid artery occlusion was not counted as an independent endpoint. Carotid endarterectomy for asymptomatic disease was performed on 6% of patients after entry, all of whom were removed from further analysis. Patients with 80% to 99% carotid artery stenosis had a 21% annual rate of risk of TIA and stroke, significantly higher than the 4.8% rate found among patients with lesser degrees of disease. However, correlation between the ischemic event and the side of greatest stenosis was poor, due to a significant contribution from events occurring contrallateral to the side of greatest stenosis. The key role of the contralateral circulation was confirmed by analysis of the 34 patients suffering ischemic events: about one third had ischemic symptoms on the side of greatest stenosis, one third had the same degree of disease on each side, and one third had the greatest degree of carotid artery disease on the contralateral, asymptomatic side.

8 Volume 17 Number 1 January 1993 Natural history of asymptomatic carotid artery disease 167 Table IV. Status of the contralateral vessel associated with TIA and stroke Stenosis of Ischemic ipsilateral events (ffmptomatic) vessel (%) TIA CVA Stenosis of contralateral vessel (%) TIA CVA TIA CVA TIA CVA Occluded TIA CVA TIA CVA Occluded 1 3 Total CVA, Cerebrovascular accident These findings echo those of the Framingham project a4 and the Evans County, Georgia study, 2s two population-based studies of asymptomatic patients with carotid bruit. Fewer than half those suffering stroke in each study experienced the event on the same side as the bruit. A similar result was obtained by Meissner et al.,14 who studied 292 patients with flow-significant carotid artery lesions diagnosed by oculopneumoplethysmography and found that only 45% of strokes were ipsilateral to the diseased vessel. Other studies that categorized patients for analysis solely by the most highly stenosed side n,26 failed to address the laterality of ischemic events. The mechanism of stroke contralateral to critical carotid artery stenosis is unknown, but could relate to hemodynamic changes in the less-diseased vessel. With preservation of total cerebral perfusion, carotid artery flow would increase contralateral to the critical stenosis. A concurrent increase in blood velocity and vessel wall shear stress would result, perhaps causing acute changes in the atherosclerotic plaque lining the vessel wall and leading to thrombogenesis, embolism, or plaque fracture. The significant effect of contralateral disease confounds the question of whether critical carotid artery stenosis is a risk factor for cerebral ischemia, and it complicates the issue of treatment for asymptomatic 80% to 99% disease. The 21% annual rate of TIA and stroke in our patient-based analysis confirmed a significantly greater risk among patients with critical stenosis. But because contralateral events were so frequent, analysis by carotid vessel rather than by patient reduced this risk to 5.1%, which was not significantly different from the 2.8% event rate for less-diseased vessels (although a trend toward significance was present). Patient-based data are epidemiologic in nature, and may be suitable for considering proposed benefits of nonsurgical therapy for stroke Table V. Relationship between plaque morphology and vessel event rates Plaque Event rate structure Events Vessel-years (%) Echolucent ~ Echogenic Occluded Total Statistical analysis compares echolucent with echogenic plaques, because plaque type was not evaluated in occluded vessels. ~p = prevention. But carotid endarterectomy is a unilateral procedure, and its direct benefits and attendant risks relate to operation on a single artery. Therefore vessel-based rates are probably more appropriate for consideration of surgical prophylaxis of stroke. An annual per-vessel event rate of 5.1% is nonetheless relatively high, and may of itself justify carotid endarterectomy for asymptomatic critical stenosis. It is also possible that endarterectomy in this setting provides an additional, indirect benefit by increasing collateral vessel flow to the contralateral, less-diseased side, which we have demonstrated to be at high risk as well. Furthermore, by restoring normal hemodynamics bilaterally, endarterectomy would relieve the contralateral artery from increased blood velocity and shear stress. Finally, operation for critical stenosis would substantially reduce the risk of asymptomatic carotid artery occlusion. Though this is, by definition, a benign event, our data demonstrate that "stable," occluded carotid arteries have the highest subsequent event rate among all categories of carotid artery disease (5.5% annual event rate). It therefore seems logical that endarterectomy may provide further protection through prevention of asymptomatic occlusion and subsequent cerebral ischemia. Echolucent carotid artery plaque, which contains

9 168 Bock et al. Journal of VASCULAR SURGERY OCCLUDED 80-99% O o. P i, 0 STROKE (N=15) CONTRALATERAL PROGRESSION i/o, 50-79% o. 0//0 ~= % e ~ (..) f()../ : % Fig. 4. Annual duplex results in patients suffering stroke. X denotes single patient suffering immediately fatal stroke, and asterisk indicates significant disease progression contralateral to depicted vessel. All open circles represent post-stroke scans. either intraplaque hemorrhage 27 or collections of liquid, macroscopic lipid, 28 was a significant risk factor in the development of TIA and stroke in our series. A recent meta-analysis of the literature has confirmed the sensitivity of ultrasonography in detecting intraplaque hemorrhage, 29 which in turn has a well-documented association with cerebral ischemia However, like critical carotid artery stenosis, echolucent plaque is uncommon among asymptompatic patients, and most patients suffering stroke have neither risk factor. Our finding of a statistically significant, threefold higher stroke rate among patients taking aspirin was unexpected. It should be interpreted with caution, because referring physicians in our uncontrolled study made all treatment decisions, and although no obvious variations in aspirin use among patients in different risk groups was detected, such bias could not be excluded by statistical means. For these reasons, and because compliance with aspirin and no-aspirin treatment regimens was not biochemically confirmed, our aspirin data should be regarded as speculative. Nevertheless, because there is no published evidence that aspirin use among asymptomatic patients is of any benefit in prevent- ing TIA or stroke, our findings serve as a reminder that a properly controlled, randomized study of this widespread but unproven practice is urgently needed. Progression of carotid artery disease among our study patients was highly correlated with TIA and stroke, but because progression was rarely diagnosed before the post-event scan, it was of little clinical use in predicting future stroke. However, the sixfold higher rate of TIA and stroke found with disease progression, and the frequent finding of progression in post-event duplex scans, may help explain the pathogenesis of stroke. It seems likely that stroke and TIA are the result of abrupt changes in plaque structure rather than a slow, smooth progression of stenosis. Fig. 4 provides further insight: quantum leaps and unpredictable strokes are the rule, and only one of 15 strokes was observed after progression to 80% to 99% stenosis. Most neurologic events observed in our study occurred in the absence of 80% to 99% stenosis. Those events that did occur in association with critical stenosis were usually contralateral to the high-grade lesion. Thus, while our data do confirm a higher risk for the relatively small subgroup of

10 Volume 17 Number 1 lanuary 1993 Natural history of asymptomatic carotid artery disease 169 patients with 80% to 99% stenosis, it seems reasonable to conclude that factors other than ipsilateral critical carotid artery stenosis may account for most TIAs and ischemic strokes. Slow, steady progression in carotid artery stenosis may be less important than acute changes in carotid artery plaque caused by increased shear forces, plaque fracture, acute endothelial injury, and intraplaque hemorrhage (possibly exacerbated by aspirin). Thrornbogenesis, embolism, and acute vessel occlusion, all commonly identified causes of stroke, may follow these sudden events.' Together, these abrupt, quantum, unpredictable events may best explain most strokes caused by cerebrovascular disease. We thank J. David Richardson, MD, and Thomas M. Bergamini, MD, for their valuable advice in preparation of the manuscript. REFERENCES 1. Haas WK, Fields WS, North P_K, et al Joint study of extracranial arterial occlusion, II: arteriography, techniques, sites, complications. JAMA 1968;203: Kannel MB, Wolf PA, Verter J, et al. Epidemiologic assessment of the role of blood pressure and stroke: the Framingham study, JAMA 1970;214: Whisnant JP, Fitzgibbon JP, Kurland LT, et al. Natural history of stroke in Rochester, Minnesota Stroke 1971;2: NASCET Collaborators. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade stenoses. N Engl J Med 1991;325: Mayberg MR, Wilson SE, Yatsu F, et al. Carotid endarterectomy and prevention of cerebral ischemia. JAMA 1991;266: European Carotid Surgery Trialists' Collaborative Group. M.R.C. European carotid surgery trial: interim results for symptomatic patients with severe (70-99%) or mild (0-29%) carotid stenosis. Lancet 1991;337: Friedman GD, Wilson WS, Mosier JM, et al. Transient ischemic attacks in a community. JAMA 1969;210: Whisnant JP, Matsumoto N, Elveback LR. Transient ischemic attacks in a community: Rochester, Minnesota, Mayo Clin Proc 1973;48: Herman B, Leyten ACM, van Luijk JH, et al. An evaluation of risk factors for stroke in a Dutch community. Stroke 1982;13: Mohr JP, Caplan LR, Melski JW, et al. The Harvard Cooperative Stroke Registry: a prospective registry. Neurology 1978;28: Chambers BR, Norris JW. The case against surgery for asymptomatic carotid stenosis. Stroke 1984;15: Chambers BR, Norris IW. Outcome in patients with asymptomatic neck bruits. N Engl J Med 1986;315: Bogousslavsky J, Despland PA, Regli F. Asymptomatic tight stenosis of the internal carotid artery: long-term prognosis. Neurology 1986;36: Meissner I, Wiebers DO, Whisnant JP, O'Fallon WM. Thenatural history of asymptomatic carotid artery occlusive lesions. JAMA 1987;258: Roederer GO, Langlois YE, Jager KA, et al. The natural history of carotid arterial disease in asymptomatic patients with cervical bruits. Stroke 1984;15: Fowl RI, Marsch JG, Love M, et al. Prevalence of hemodynamically significan t stenosis of the carotid artery in an asymptomatic veteran population. Surg Gynecol Obstet 1991;172: Zierler RE, Kohler TR, Strandness DE Jr. Duplex scanning of normal or minimally diseased carotid arteries: correlation with arteriography and clinical outcome. J VASC SURG 1990;12: Roederer GO, Langlois YE, Jager KA, et al. A simple spectral parameter for accurate classification of severe carotid disease. Bruit 1984;8: Taylor DC, Strandness DE Jr. Carotid artery duplex scanning. J Clin Ultrasound 1987;15: Steffen CM, Gray-Weale AC, Byrne IcE, Lusby RJ. Carotid artery appearance in symptomatic and asymptomadc vessels. Aust N Z J Surg 1989;59: Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53: Dixon WJ, ed. BMDP statistical software manual, vol 2. Los Angeles: University of California Press, Moneta GL, Taylor DC, Nicholls SC, et al. Operative versus nonoperative management of asymptomatic high-grade internal carotid artery stenosis: improved results with endarterectomy. Stroke 1987;18: Wolf PA, Kannel WB, Sorlie P, McNamara P. Asymptomatic carotid bruit and risk ofsttoke: the Framingham study. JAMA 1981;245: Heyman A, Wilkinson WEt Heyden S. Risk of stroke in asymptomaric persons with cervical arterial bruits: a population study in Evans County, Georgia. N Engl I Med 1980;302: Humphties AW, Young JR, Santilli PH, Beven EG, dewolfe VG. Unoperated, asymptomatic significant carotid artery stenosis: a review of 182 instances. Surgery 1976;80: Reilly LM, Lusby RJ, Hughes L, Ferrell LD, Stoney RJ, Ehrenfeld WK. Carotid plaque histology using real-time ultrasonography. Am J Surg 1983;146: Bock RW, Lusby RJ. Carotid plaque morphology and interpretation of the echolucent lesion. In: Labs KH, lager K.A, Fitzgerald DE, Woodcock JP, Neuerburg-Heusler D, eds. Diagnostic vascular ultrasound. London: Edward Arnold, Ltd, 1992: Reilly LM. Carotid intraplaque hemorrhage: noninvasive detection and clinical significance. In: Bernstein EF, ed. Noninvasive diagnostic techniques in vascular disease. St Louis: CV Mosby, 1990: Lusby RJ, Ferrell LD, Ehrenfeld WK, Stoney RJ, Wylie EJ. Carotid plaque hemorrhage: its role in production of cerebral ischemia. Arch Surg 1982; 117:147% Langsfeld M, Gray-Weale AC, Lusby RJ. The role of plaque morphology and diameter reduction in the development of new symptoms in asymptomatic carotid arteries. J VAsc SURG 1989;9: Imparato AM, Riles TS, Mintzer R, Baumann FG. The importance of hemorrhage in the relationship between gross morphologic characteristics and cerebral symptoms in 376 carotid plaques. Ann Surg 1983;197: Submitted 1UlY 28, 1992; accepted Oct. 5, 1992.

11 170 Bock et al. Journal of VASCULAR SURGERY DISCUSSION Dr. Leonard J. Perloff (Philadelphia, Pa.). Because of the North American Symptomatic Carotid Endarterectomy Trial and the European Carotid Surgery trial, we now know that patients with symptoms can derive extraordinary benefit from successfial surgery. Although they were well intended, early studies, such as the Framingham or the Evans County, Georgia study, were flawed for want of a clear description of the underlying carotid artery lesion in patients who do not have symptoms. Today, with the advent of sophisticated, noninvasive diagnostic techniques, we can perform much more detailed and precise analyses. Dr. Bock et al. have again confirmed the high risk of central nervous system events for patients without symptoms who have critical carotid artery stenoses or echolucent plaques, and they also stress the importance of contralateral disease. This study, which is similar to previous studies, such as those of Roederer et al. is and Chambers and Norris, 12 reveals an annual event rate of 6.2% with a 4:3 ratio of TIA to stroke. In addition, a stenosis of 80% to 99% was a harbinger of much higher morbidity rates. Of greater interest are the areas of difference between this and other reports. The authors emphasize that disease progression is highly correlated with TIA and stroke, but they suggest that these events occur "in association with rather than before an ischemic event" and that the disorderly, unpredictable quantum changes are therefore less usefial as a guide to the surgeon. In fact, only one of 15 strokes occurred downstream from an 80% to 99% lesion, and vessels ipsilateral to TIAs and strokes were "less diseased than their contralateral counterparts." In contrast, the Seattle group has shown that marked disease progression to greater than 80% was always observed before occlusion and that 89% of the patients showed progression before TIA or stroke. Furthermore, Perler et al. noted that 64% of patients with internal carotid artery occlusion had experienced symptoms from 2 to 120 months previously, and Nichols et al. observed progressions to total internal carotid artery occlusion with an ipsilateral event rate of 40% in 24 patients. So, although this article suggests that the degree of stenosis and the status of the collateral circaxlation influence neurologic outcomes, it does little to help us define the high-risk lesion from the standpoint of surgical intervention. This leads to several questions. Will we fail to heed the warning of a tight stenosis that may actually protect the patient from ipsilateral neurologic morbidity? Or does this simply represent a statistical trick that does not charge the event to the initial degree of stenosis? Were there any clues to vulnerability? That is, how detailed was the questioning regarding previous, possibly trivial central nervous system events and what studies of the brain-computed tomography (CT), magnetic resonance imaging, or others-were done in these patients? Should we be more attentive to the measurement of collateral vessel flow, and have we finally found a home for transcranial Doppler ultrasonography? Because the authors have cast doubt on the utility of monitoring the progression of stenotic lesions, is the technology on a level that will allow laboratories to communicate with each other regarding uniform descriptive criteria of atheromatons disease and plaque structure? In view of the recently reported, spectacular results from the University of California at Los Angeles, with 123 patients who had no symptoms, who had greater than 75% stenosis, and who underwent 141 carotid endarterectomies with no postoperative deaths and only two postoperative strokes, is the whole question rapidly becoming moot? Dr. Richard W. Boek. The report by Roederer et al.ls was certainly a landmark article. It represented the first use of duplex ultrasound to study the natural history of asymptomatic carotid disease, and it remains the sole prospective duplex study reported to date of asymptomatic unoperated patients. Unfortunately, the limited numbers of TIAs and strokes (six and four, respectively), the short follow-up and modest size of the study (103 patients followed serially), the high rate of endarterectomy (20%), and the continued inclusion in the study of operated patients all combine to limit the utility of that first report. Nonetheless, many researchers and clinicians have inferred from the Seattle data that cerebrovascular stroke is caused by a slow, steady, observable increase in carotid stenosis, progressing to 80% to 99% disease and only then culminating in cerebral ischemia, occlusion, or both. Our own study, which followed a larger group of patients for a longer time and which had more than three times as many ischemic endpoints, fails to support this theory. Indeed, the vast majority of strokes were wholly unpredictable. Should we ignore the relatively small number of patients with critical carotid stensosis? Certainly not. We demonstrated a 5.1% annual event rate in association with ipsilateral 80% to 99% stenosis and a 20.6% annual rate when the contralateral side was included in the analysis (e.g., when patients rather than vessels were tabulated). Our data confirm the long-held belief among surgeons and others that critical stenosis is dangerous, and we agree that patients with these lesions should be seriously considered for operation. From a larger view, however, our results are somewhat depressing: the overwhelming majority of patients with asyrnptomatic carotid disease who are destined to suffer stroke will not first develop 80% to 99% stenosis. These data, along with the fact that stroke is heralded by TIA in only about 10% of cases, mean that the

12 Volume 17 Number 1 January 1993 Natural history of asymptomatic carotid artery disease 171 overall benefit to the public health from a policy of endarterectomy for both symptomatic stenosis and asymptomatic 80% to 99% disease ~ be rather small. Regarding clues to vulnerability, our patients were carefully questioned regarding previous central nervous system events. This was a clinical study; routine CT and magnetic resonance imaging of all study entrants would have been prohibitively expensive and was not done. Most study patients suffering stroke did undergo confirmatory CT. Measurement of collateral cerebral flow remains almost more art than science. Our data regarding contralateral ischemic events and disease confirm that the collateral cerebral circulation may be a critical component of overall stroke risk and remind us that any simple model for prediction of stroke that treats the two internal carotid arteries and their ipsilateral cerebral hemispheres as distinct, unrelated entities is is probably grossly inadequate. If reproducible, precise mapping of the integrated cerebral vasculature ever becomes possible, whether through transcranial Doppler ultrasonography, magnetic resonance angiography, or some combination, then perhaps accurate prediction of cerebrovascaflar stroke will move closer to reality than it is today. And finally, the excellent study of Freischlag, Hanna, and Moore (Improved prognosis for asymptomatic carotid stenosis with prophylactic carotid endarterectomy. Stroke 1992;23:479-82) did indeed show a very low risk of endarterectomy for patients with asymptomatic disease. In my view, this is testimony to the high degree of surgical skill among one small group of surgeons. The fact that stroke and death rates are substantially higher for endarterectomy performed in the community at large means that we still need to understand the natural history of asymptomatic carotid disease. Professor Roger M. Greenhalgh (London, England). Would you agree that the low morbidity and mortality rates that can be achieved in the performance of carotid artery surgery for asymptomatic disease are irrelevant; and that what is relevant is to know whether the operation should be performed at all? Is there anything in your data that would indicate that carotid artery surgery should be performed in the patient who has no symptoms? In patients with carotid artery bruit, we (Ellis and Greenhalgh) failed to show that there was an increased risk of stroke without warning; therefore there was no need to operate for asymptomatic disease. At a meeting of the European Society for Vascular Surgery in 1991, we reported on 1198 patients with asymptomatic carotid artery bruit, whom we monitored for 5 years. Several hundred patients had a stenosis greater than 50%, some had a stenosis greater than 80%, and others had bruit. We were able to show that the risk of stroke without warning was not increased. The relative increased risk for stroke was 1.78%, but this was not significant. With a larger group we would have achieved statistical significance. I stress that it was necessary to monitor almost 1200 patients in one center for more than 5 years to achieve just that. The size of the problem is immense. Do you believe that there is anything ih your study that at the moment justifies intervention in the asymptomatic state, and do you believe that there is a high-risk group somewhere? Dr. Bock. Mr. Greenhalgh is of course correct that, no matter how low the morbidity and mortality of any operation, the primary question remains that of indications and benefits (and, perhaps, cost). Our study was not intended to determine whether carotid endarterectomy is indicated in asymptomatic patients-several large, randomized, multi-center studies of this question should be complete in a few more years-but rather to delineate the natural history of asymptomatic carotid artery disease. As you know, tens of thousands of carotid duplex studies are performed on asymptomatic patients each year. Tens of thonsands of carotid duplex results are reported to clinicians each year. How should these reports be interpreted? What does a given duplex result mean? What are the implications to the patient? We hope that, in addition to providing natural history data as a foundation for future efforts in prevention of stroke, our report provides a much-needed framework for interpretation of the many carotid duplex examinations currently performed. With these caveats regarding the purpose of our study aside, some conclusions regarding the risk of critical carotid stenosis are nonetheless inescapable. Patients presenting with or progressing to 80% to 99% stenosis are at high risk for both ipsilateral and contralateral cerebral ischemia; such risk exceeds that of operation before even 1 year has passed. This strongly suggests that endarterectomy should be considered in these patients. Again, the unfortunate fact that stroke does not characteristically result from slow, steady, observable progression of carotid stenosis means that most patients fated to suffer stroke will never have the oppornmity for prevention through endarterectomy of an 80% to 99% lesion. This problem, together with the fact that stroke is only occasionally heralded by TIA, and the possibility that aspirin may be useless or even detrimental in stroke prevention, means that though vascular surgeons may indeed remove their masks and walk in the sunshine again, we nevertheless have only begun to cast meaningful light on the problem of predicting, much less preventing, stroke in asymptomatic patients.