Reporting Standards for Carotid Artery Angioplasty and Stent Placement

Randomized Trials of Symptomatic Patients 

Randomized Trials of Asymptomatic Patients 

Nonrandomized Studies 

Innumerable case series of carotid endarterectomy have been published. A systematic review of the risk of stroke and death caused by endarterectomy for symptomatic carotid stenosis was performed by Rothwell et al (40). The authors analyzed 51 studies since 1980, including the NASCET, and reported an overall perioperative risk of stroke and/or death of 5.64%. The overall death rate was 1.62%, with the risk of a fatal stroke (0.86%) slightly exceeding the risk of a non-stroke death (0.7%). The authors noted there was significant heterogeneity in the reported stroke and mortality rates between studies. Surgical series in which neurological outcome assessment was independently adjudicated by neurologists reported significantly higher risks of stroke and death (41).

Rothwell et al performed a similar systematic review comparing the risk of stroke and death caused by carotid endarterectomy, performed by the same surgeons in the same institutions, for symptomatic versus asymptomatic stenosis (42). The authors analyzed 25 studies since 1980 and reported an overall perioperative risk of stroke and/or death for asymptomatic lesions of 3.35%. This risk estimate was significantly lower than for symptomatic lesions (5.18%) and consistent across virtually all studies.

Results from randomized trials may not be generalizable to the results of treatment in clinical practice. Wennberg et al assessed the perioperative mortality among 113 300 Medicare patients undergoing carotid endarterectomy during 1992 to 1993 in “trial hospitals” (those participating in NASCET and ACAS, n=86), and “nontrial hospitals” (nonfederal institutions performing endarterectomy n=2613), looking at all patients treated rather than those selected for a trial (43). The perioperative mortality rate was 1.4% at trial hospitals compared with 2.5% at low-volume (≤6 procedures/year) nontrial hospitals.

High-risk Patients 

NASCET, ACAS, and the VA Cooperative studies excluded patients with significant comorbid illnesses such as lung, liver, or renal failure, unstable an-gina, and recent myocardial infarction. Therefore, the complication rates reported in these studies may be lower than the rates achievable when all patients are included. Wennberg et al (43), as noted, found that the perioperative mortality rate at NASCET trial hospitals was 1.4% when all patients were included, compared with 0.6% in NASCET patients. Similar findings were reported by Lepore et al (44), who found that NASCET/ACAS trial-eligible patients had a perioperative stroke and death rate of 1.5% compared with 3.6% for trial-ineligible patients (45). However, according to Dorros, these “high-risk” patients were excluded from the NASCET, not because the surgical outcomes may have been adversely influenced but primarily because attendant comorbidities could have potentially contaminated the outcome data (46). The ASA and carotid endarterectomy (ACE) trial was a randomized trial of high- versus low-dose aspirin at the time of endarterectomy (47). Patients were not excluded based on comorbidities; therefore, the data are likely to reflect the patient selection found in standard medical practice rather than selecting only the patients at lowest risk. Both symptomatic (n=1292) and asymptomatic (n=1512) patients were treated. Most patients had a carotid stenosis >70% diameter. Among symptomatic patients, there was a 6.4% rate of perioperative stroke (minor or major) or death but only 2.8% rate of perioperative major stroke or death (48). Among asymptomatic patients, the rate of any peri-operative stroke or death was 4.6%.

A retrospective analysis by Sundt, in which 3111 consecutive endarterectomy patients were stratified into 6 risk classes according to neurological status, comorbid conditions, and angiographic variables, revealed a very low major complication rate (permanent stroke, myocardial infarction, or death) for class I and II patients, whereas class IV patients had an 8.1% risk and 2.9% mortality (Table 1, Table 2) (49, 50). Factors correlating with an increased surgical morbidity and mortality included unstable neurological status; the presence of comorbid conditions; age older than 70; and contralateral internal carotid artery occlusion.

Table 1. Sundt's Classification System Based on Retrospective Analysis of 3,111 Consecutive Carotid Endarterectomy Patients

	Sundt's Class (49)	Criteria for Classification	Combined Surgical Morbidity (Mortality), %
	I	Neurologically stable, no major medical/angiographically defined risk, with unilateral/bilateral ulcerative-stenotic disease	0.9
	II	Neurologically stable, no major medical risks, with or without angiographic risks	1.7
	III	Neurologically stable, major medical risks, with or without angiographic risks	3.7 (1.3)
	IV	Neurologically unstable, with or without medical/angiographic risks	8.1 (2.9)
	V	Acute internal carotid artery occlusion, progressive neurological deficit within six h of examination	Not included in study
	VI	Recurrent symptomatic carotid stenosis	Not included in study

Table 2. Sundt's Definition of Medical, Neurological, and Angiographic Risk for Carotid Artery Surgery

	Risk	Definition (49)
	Medical	Coronary artery disease (angina, myocardial infarction <six mo, congestive heart failure), hypertension (>180/110 mm Hg), severe peripheral vascular disease, chronic obstructive pul-monary disease, age older than 70 years, severely obese
	Neurological	Neurological deficit within 24 h, general cerebral ischemia, recent cerebrovascular accident (<seven d), frequent transient ischemic attacks
	Angiographic	Contralateral internal carotid artery occlusion, siphon stenosis, plaque >three cm distally in internal carotid artery or >five cm proximal in common carotid artery, bifurcation at C2 vertebra, short thick neck, and soft thrombus extending from an ulcerative lesion

Estes et al followed a random sample of 22 165 Medicare beneficiaries, who underwent carotid endarterectomy between 1988 to 1990, until the year 1992. They identified patients with acute myocardial infarction, congestive heart failure, diabetes mellitus, and age older than 80 years as having diminished perioperative and long-term survival rates after carotid endarterectomy. Recurrent carotid stenosis has also been felt to indicate a higher risk of complication from endarterectomy. Meyer et al reported a perioperative morbidity and mortality rate of 10.8% in their series of 82 patients undergoing carotid endarterectomy for recurrent stenosis; that was 5 times the risk of a routine endarterectomy at their institution (51). Cranial and cervical nerve injuries may occur in 21% of patients, although there are few permanent injuries (52).

Radiation-induced carotid stenosis is more difficult to treat surgically because of long lesion length, periarterial scarring with ill-defined planes of resection, and a higher rate of wound complications (53, 54, 55).

These categories of high risk for carotid endarterectomy are not uniformly accepted. In a recent study, age older than 80 was found not to be a significant risk (56). Hill et al and O'Hara et al found no increased morbidity or mortality for re-operation for carotid stenosis compared with primary carotid endarterectomy (57, 58). Combined carotid endarterectomy and coronary artery bypass grafting has a high risk of stroke, but this risk may come almost completely from the coronary artery surgery rather than from the carotid procedure (59). Gasparis et al stratified endarterectomy patients into low- versus high-risk based on age older than 80 years, New York Heart Association class III/IV angina, Canadian class III/IV heart failure, myocardial infarct within 6 months, steroid- or oxygen-dependent pulmonary disease, creatinine level of ≥3, contralateral carotid occlusion, high carotid bifurcation, re-operation, or neck irradiation (60). The risk of peri-operative stroke or death was 1.3% in the high-risk group and 1.1% in the low-risk group (not significant).

Carotid Balloon Angioplasty 

Percutaneous transluminal balloon angioplasty for carotid artery stenosis was first reported by Kerber et al in 1980 (64). In 1987, Theron et al published the first large series of internal carotid angioplasty in 48 patients with de novo atherosclerosis or postsurgical restenosis. Technical success was achieved in 94% of cases, with a 4.1% rate of serious morbidity (65). In Kachel's review of the literature through 1995, 523 carotid angioplasty procedures had subsequently been reported (66). The overall technical success rate was 96.2%, with a 2.1% rate of morbidity, 6.3% rate of transient minor complications, and no deaths. In one of the largest single-institution series, Gil-Peralta et al (1996) performed 85 balloon angioplasties in 82 patients with symptomatic carotid stenoses of >70% over a 4-year period (67). They reported a technical success rate of 92% (residual stenosis <50%), with 30-day mortality of 0% and a major morbidity rate of 4.9%, which compares very favorably to the ECST and NASCET results. The rate of recurrent stenosis (all of which were asymptomatic) was 6.7% at a mean follow-up of 18.7 months, with almost all of these occurring between months 3 and 6. Rates of restenosis (<2 years) reported in other large angioplasty series are between 0% and 16% (66, 68, 69, 70, 71); in carotid endarterectomy series, rate of restenosis was ≈10% in the first year (51, 72, 73, 74).

Despite these favorable results, balloon angioplasty has a number of potential limitations, including vessel wall recoil, angiographically evident intimal dissection, and plaque dislodgement with particulate embolization. Angioplasty of atherosclerotic lesions has been reported to generate emboli composed of atheroma, cholesterol crystals, thrombus, and platelet aggregates (75, 76, 77, 78). Embolization of microparticles has also been demonstrated during and after carotid endarterectomy and has been shown to correlate with complex plaque morphology (79) and with clinical postoperative cerebral ischemia (80, 81, 82, 83). However, studies examining the frequency of emboli during carotid balloon angioplasty using transcranial Doppler (TCD) have failed to show a clear-cut correlation between embolism frequency and neurological sequelae (76, 77, 78). Crawley et al, in an analysis of 14 patients undergoing carotid balloon angioplasty versus 14 endarterectomy patients with shunt placement, reported an average of 202 embolic signals during carotid balloon angioplasty compared with 52 during carotid endarterectomy. However, during the recovery period of 20 minutes, the average number of embolic signals was 5 for balloon angioplasty versus 19 for endarterectomy. There was no correlation between the number of TCD-detected emboli and the periprocedural stroke rate (77).

The risk of cerebral damage is thought to depend on the size and composition of the embolic material as well as the extent and location of brain involvement. Because it is difficult to accurately distinguish between air and particulate emboli using TCD (84), the lack of correlation between emboli and clinical sequelae has led to the suggestion that the majority of emboli detected during balloon angioplasty are either gaseous or small platelet aggregates <200 μm in diameter, both of which correlate with a more benign outcome (75, 76, 77). This underscores the importance of premedication with antiplatelet agents to prevent larger platelet aggregates (85).

Endovascular revascularization of carotid occlusive disease may result in cerebral hypoperfusion from luminal compromise by catheters and guide-wires crossing the stenotic lesion and/or during balloon inflation. This is of even greater relevance in the presence of contralateral carotid artery occlusion or stenosis. Eckert et al monitored 22 patients undergoing carotid balloon angioplasty using TCD, noting that a >50% reduction of middle cerebral artery mean blood flow velocity compared with baseline values represented a critical threshold for the development of ischemic symptoms in conscious patients even with short occlusion times of 10 to 40 seconds (86). However, Crawley et al reported hemodynamic ischemia was significantly greater with endarterectomy than balloon angioplasty and was not predictive of adverse neurological outcome (77). It is recommended to keep balloon inflation and occlusion times <30 seconds to avoid potential cerebral ischemia. If there is attendant compromise of the contralateral carotid artery, then establishing the adequacy of the cerebral collateral circulation becomes even more important and the procedure should be performed with as little sedation as possible to facilitate neurological monitoring. Tsurutani et al described the successful use of a continuous-perfusion dilatation catheter for high-risk patients with poor collateral circulation (87).

Carotid Stenting 

The impetus for carotid stenting has arisen principally from trials of stenting with balloon angioplasty versus simple balloon angioplasty in the coronary arteries, which have consistently demonstrated a persistent benefit in event-free survival at 1 year and a lower rate of repeat angioplasty (88, 89, 90). The purported advantages of stent placement over simple angioplasty include avoiding plaque dislodgment, intimal dissection, elastic vessel recoil, and late restenosis. Despite the fact that it is unknown whether these benefits apply to the carotid circulation, endovascular carotid revascularization is now most commonly performed with stents.

Since 1996, there have been 11 large carotid stent series published in which the total number of patients is 1311 (71, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101) (Table 3). Comparative analysis of these reports is made difficult by inconsistencies in the sample populations, lesion characteristics, endovascular techniques, and outcome data. However, the overall reported rate of technical success is >95%; procedure-related mortality rates (including cardiac deaths) are 0.6% to 4.5%; major stroke rates are 0% to 4.5%; minor stroke rates are 0% to 6.5%; and a 6-month restenosis rate is <5%. This excludes the other studies that represented very-high-risk cohorts (98, 99, 102, 103, 104, 105). Similar favorable results were reported by Al-Mubarak et al, who reported on 51 consecutive patients undergoing simultaneous or staged carotid artery stenting and percutaneous coronary intervention. Technical success was achieved in all carotid arteries, with a minor stroke rate of 4% and no major strokes, myocardial infarctions, or deaths (106). In the NASCET, the perioperative stroke and death rate was 5.8%, with 0.6% mortality rate mostly caused by myocardial infarction (23).

Table 3. Carotid Stent Series Data

	Reference	Patient Data			Technical Success Rates (%)	Complications				Restenosis Rate (%)		Remarks
		N of Patients	N of Stents	Asymptomatic Stenosis (%)		30-day Morbidity/Mortality (%)	No. of Deaths/Mortality Rate (%)	Major Stroke Rate (%)	Minor Stroke Rate (%)	6-month (%)	12-month (%)
	Roubin (101)	528	604	48	98	7.4	8(1.6)	1.0	4.8	16(3.0)
	Theron (71)	69	69	NS	100	NS 2 embolic complications (3) no deaths				4 (time not specified)
	Diethrich (92)	110	129	72	99.1	7.3	2(1.8)	2	4.5	NS/ID		One death, cardiac-related
	Yadav (95)	107	189	36	100	7.9	1(0.9)	1.9	6.5	4.9	NS
	Vozzi (94)	22	19	55	96	NS	1(4.5)	4.5	4.5	NS/ID
	Criado (93)	33	NS	27	100	NS	0	0	0	3 (mean eight-month follow-up)
	Wholey (96)	108	NS	44	95	NS	2(1.9)	1.8	1.8	1 (mean six-month follow-up)		One death, cardiac-related, day 20
	Henry (144)	163	178	35	99.4	NS	0	1.8	1.2	2.3%	ID	Distal cerebral protection used in 32 cases
	Teitelbaum (98)	22	31	32	96.2	27.3	1⁎(4.5)	18.2	13.6	14.3%	NS	High-risk cohort
	Waigand (99)	50	56	72	100	2	1(2)	2	2	8.7% (mean eight-month follow-up)		High-risk cohort
	Bergeron (100)	99	99	42	97	7.1	0	0	1	4.2% (mean 13-month follow-up)

NS indicates not specified; ID, insufficient data.

⁎ Cardiac-related death.

Vitek et al presented their experience in treating 404 patients with carotid angioplasty and stenting (107). They report a technical success rate of 98%, 30-day morbidity/mortality rate of 1.9%, major stroke rate of 0.7%, minor stroke rate of 5.8%, and a 5% rate of restenosis (>50% narrowing). The authors report a lower complication rate in the last 122 patients (with increased procedure-related clinical experience): minor stroke rate of 2.5% with no major strokes or deaths. Wholey et al have published the results of a worldwide stent survey in which 3047 endovascular cervical carotid artery stent procedures are reported from 24 centers in North America, Europe, Asia, and South America (108). They reported a 30-day procedure-related mortality rate of 0.98%, major stroke rate of 1.35%, minor stroke rate of 2.53%, and a restenosis rate at 6 and 12 months of 2.23% and 2.48%, respectively. Centers having worked with <50 cases had a 6.4% rate of major stroke and death compared with 2.3% for centers having worked with between 50 and 100 cases. The authors thus suggest a 50-case learning curve for carotid stenting.

Many of these case series also analyzed outcomes for patients considered at higher risk for surgery (ie, NASCET-ineligible). In the NASCET, the perioperative stroke and death rate was 5.8%, with 0.6% mortality rate mostly caused by myocardial infarction (23). NASCET criteria were applied to the patient cohorts in 5 of these carotid stenting series, demonstrating that 79% of these 574 patients would have failed eligibility because of comorbidities (91, 95, 96, 99). Despite this, morbidity rates of 2.0% to 7.9% and mortality rates of 0.6% to 2.0% for these early carotid stenting reports compare favorably to those in the NASCET and the ECST. However, 41% (n=233) of these patients had asymptomatic carotid stenosis for which lower endarterectomy morbidity and mortality rates would be expected. Similarly, Yadav et al reported on 22 patients with angioplasty and stenting for postendarterectomy carotid stenosis with no major and only 1 minor stroke (4.5%) (109). Lanzino et al also reported favorable results in 18 patients undergoing angioplasty and stenting for recurrent carotid stenosis after endarterectomy, with no periprocedural strokes and 1 transient ischemic attack (110). Mericle et al also reported on 23 patients with high-grade carotid stenosis and a contralateral carotid occlusion undergoing elective carotid stenting (111). The 30-day perioperative stroke and death rate was 0%. This is in contrast to the NASCET, which reported a perioperative stroke/death rate for contralateral occlusions of 14.3% (23).

Mathur et al retrospectively analyzed the risk factors for stroke in 231 patients undergoing elective stenting of the extracranial carotid arteries; 14% of which were NASCET-eligible (112). The overall 30-day stroke rate was 6.9%; however, by stratifying for the NASCET-eligible group, the stroke rate decreased to 2.7%. Advanced age (older than 80 years) and long or multiple stenoses were found to be independent predictors of periprocedural strokes. However, contralateral carotid occlusion, previous carotid endarterectomy, and combined carotid and coronary procedures, all of which are associated with a higher incidence of complications in carotid endarterectomy, were not found to have an increased risk of adverse outcome in stented patients (51, 112, 113, 114, 115). Increasing age was also identified as a predictor of increased risk for carotid stenting procedures by Chastain et al. The authors stratified 182 patients undergoing elective carotid artery stenting into 3 age groups: older than 80 years, 75 to 79 years, 74 years or younger. The overall rate of major stroke and death (0.5%) was 1.6%, with a 0.5% rate of myocardial infarction. Neurological complications (mostly minor strokes) were significantly more frequent in patients older than age 80 years than for those 74 years or younger (25% versus 8.6%, respectively) (116).

There are few published studies directly comparing carotid endarterectomy with carotid stenting. Jordan et al retrospectively compared 107 endarterectomy patients with 166 prospectively followed-up patients undergoing carotid stenting and reported a higher early minor stroke rate in the stent group (6.6% versus 0.6%) but a higher rate of major stroke/death in the surgical cohort (4.2% versus 2.8%) (117). Jordan also analyzed outcomes from a subset of 109 patients undergoing endarterectomy under regional anesthesia and compared these results to 268 patients undergoing carotid stenting in the same institution. The total early stroke and death rates for stent versus endarterectomy were 9.7% versus 0.9%. Most of these strokes were minor; the major stroke rates for stent versus endarterectomy were 1.5% versus 0%. There is one report of an early prospective randomized study in 17 patients treated for symptomatic ste-noses of >70% (118). The authors reported no complications in 10 patients undergoing endarterectomy; however, 5 of 7 stent patients had strokes, 3 of which were disabling at 30 days. This prompted early stoppage of the trial in favor of endarterectomy. However, this trial had methodological problems that may explain these poor results.

In reviewing the current literature on carotid artery stenting, the technical success rate, procedure-related morbidity and mortality rates, and restenosis rates seem to compare favorably to carotid endarterectomy (119, 120, 121). However, inconsistencies in the trial design, reporting criteria, and follow-up render comparison of endarterectomy with endovascular revascularization most challenging (Table 4). The validity of comparisons between such disparate clinical data is unsound. In the case of carotid endarterectomy, there exists class I evidence of efficacy, whereas the bulk of the endovascular data are derived from non-randomized uncontrolled studies, usually from single institutions (122, 123). Class I evidence is that which is proven by a prospective randomized trial with a small chance of false-negative or false-positive values. Therefore, 5 prospective, randomized, multicenter trials comparing carotid endarterectomy to carotid angioplasty or stenting have been undertaken, the results of which are included in Table 4.

Table 4. Carotid Stent Series Data

	Study	Stenosis, %	Symptomatic	N Patients (CEA)	High-risk Excluded	Perioperative				Annual
						Minor Stroke, %	Major Stroke	MI	Death	Ipsilateral Stroke, %	Ipsilateral Disabling Stroke, %	Any Stroke, %	Death, %	Duration
	NASCET (23, 25, 26)	>70	Yes, <120 days	328	Yes	3.7c	1.5	0.9	0.6	4.5	1.3	6.3	1.6	2
	NASCET (23, 25, 26)	50–70	Yes, <180 days	430	Yes	4.7	0.8	NA	1.2	3.1	0.6	4.8	1.9	5
	NASCET (23, 25, 26)	<50	Yes, <180 days	678	Yes	Combined	Combined	NA	Combined	3	0.9	5.1	2.1	5
	ECST (37)	All	Yes, <180 days	1,745	No	3.5d	3.2h		1	2.3	1.6	3.2	2.9	3
	VA-CSP-309 (36)	>50	Yes, <120 days	91	Yes	2.2e	1.1i		3.3	3			NA	3
	ACE (47, 48)	>70a	Yes, <180 days	1,292	No		6.4j
	ACAS (34)	>60	No	825	Yes		2.1k	NA	0.2n	2.2o	1.2l	3.5	3.9	5
	VA-CSP-167 (36)	>50	No	211	Yes	0.9f	2.8e	1.9m	1.9	1.2p		2	8.2	4
	ACE (47, 48)	>70 (most)	No	1,512	No		4.6l
	CAVATAS (125)	>30b	90	253 CEA	Yes	4g	4	NA	2	4.1	2	NA	2.7s	3
				251 Stent		4	4	NA	2	3.2	1.1	NA	3.7	3
	Wallstent (127)	>60	Yes	112 CEA	NA		4.5l				0.9q	3.6		1
				107 Stent			12.1				3.7	12.1r		1
	Brooks et al (128)	>70	Yes	51 CEA	No	0	2	1	1
				53 Stent		2	0	0	0
	SAPPHIRE (129)	≥50%t	Yes, >48 h	151 CEA	No	3.3	2.0	7.3	2
				156 Stent		3.2	0.6	2.6	0.6

Notes on unique definitions and exceptions used by the various studies:

a Most lesions >70% stenosis. b Minimum stenosis not specified. c Defined as modified Rankin Scale <three at 90 days after treatment. d Nondisabling stroke >seven days after procedure. e Type of stroke not defined. f Transient neurological deficit, ie, TIA. g Defined as modified Rankin Scale <three at 30 days. h Defined as fatal or disabling stroke. i Defined as stroke before endarterectomy. j Defined as any stroke or death, reported as 2.2% disabling stroke or death in CAVATAS commentary. k Major stroke of 1.2% after endarterectomy, 0.6% after angiography, and 0.2% before angiography. l Defined as any stroke or death. m Excluded fatal myocardial infarction. n 0.1% death from endarterectomy. o Includes any perioperative stroke or death. p Includes 1.2% transient neurological deficits. q Includes any major stroke. r Includes any major stroke, vascular death, or procedural death. s Estimated from figures presented for death and disabling stroke. t ≥50% symptomatic stenosis, ≥80% asymptomatic stenosis.

The Carotid and Vertebral Transluminal Angioplasty Study (CAVATAS), a large, prospective, randomized, multicenter trial comparing carotid endarterectomy to carotid angioplasty was recently completed. In the CAVATAS, patients with symptomatic stenoses (at least 30% luminal diameter reduction) suitable for surgery were randomized to either angioplasty or surgery (124). Patients unsuitable for endarterectomy were randomized to percutaneous transluminal techniques or medical treatment alone. The CAVATAS randomized 504 patients to surgery and angioplasty over 5 years. There was no significant difference in the risk of stroke or death related to the procedure between carotid endarterectomy and angioplasty. The rate of any stroke lasting >7 days or death within 30 days of first treatment was ≈10% in the surgery or angioplasty groups. The rate of disabling stroke or death within 30 days of first treatment was 6% in both groups. Preliminary analysis of long-term survival showed no difference in the rate of ipsilateral stroke or any disabling stroke in patients up to 3 years after randomization. The rates of stroke or death within 30 days in CAVATAS in both groups are higher than those of many previous reports but not significantly different to the ECST rate of 7%. Cranial nerve injury and myocar-dial ischemia were only reported at the time of treatment in the surgical group. Long-term follow-up is not yet available (125).

The Wallstent Trial was an industry-supported, prospective, randomized trial comparing endarterectomy and carotid stenting for symptomatic stenosis ≥60% (126). The results have been published in abstract (127). In this study, 219 patients with symptomatic carotid stenosis of 60% to 90% diameter were randomized to endarterectomy or stenting. The risk of any perioperative stroke or death was 4.5% for surgery and 12.1% for stenting. At 1 year, the risk of a major stroke was 0.9% for surgery versus 3.7% for stenting. This trial was stopped early because of poor results from stenting.

A single-center community hospital study randomized 104 symptomatic patients to either carotid endarterectomy or stenting without distal protection (128). Perioperative stroke or death rate was 2% for surgery and 0% for stenting. Other complications for the surgery group totaled 16% and included hematoma (requiring treatment), cranial/cervical nerve injury, and hypotension (requiring treatment). Other complications for the stent group totaled 45% and included transient cerebral ischemia, leg amputation, retroperitoneal hemorrhage, bradycardia (requiring temporary pacing), and hypotension (requiring treatment).

The Stenting and Angioplasty with Protection in Patients at High Risk for Endarterectomy (SAPPHIRE) trial randomized 307 high-risk patients to endarterectomy or carotid stenting using a distal protection device. Perioperative (30-day) results have been presented in abstract (129). High risk was defined as congestive heart failure (class III/IV) or ejection fraction <30%, open heart surgery needed within 6 weeks, recent myocardial infarction (24 hours to 4 weeks), or unstable angina (CCS class III/IV). Perioperative stroke and death rates were 7.3% for surgery versus 4.4% for stenting. Rates of myocardial infarction were 7.3% for surgery versus 2.6% for stenting.

The Carotid Revascularization Endarterectomy versus Stent Trial (CREST), a North American, multi-center, randomized, controlled trial comparing the efficacy of surgical endarterectomy with carotid stenting, is currently in progress (130, 131). In the CREST, patients with symptomatic extracranial carotid stenosis of ≥50% are randomized between stent-supported angioplasty and endarterectomy. To demonstrate a clinically significant difference between the 2 treatments, CREST will require at least 2500 patients, excluding comparison between various clinical subgroups (eg, recurrent stenosis), for which an even greater number will be required.

Adjuncts to Stenting 

Potential areas of innovation include methods of providing cerebral protection using intravascular filters or balloons; smaller-diameter, more flexible, and hence less traumatic delivery systems; lowering rates of restenosis secondary to intimal hyperplasia by using local catheter brachytherapy; radiation-emitting stents (132, 133, 134); or biologically active coatings (135) and improved adjuvant pharmacological regimes using antiplatelet agents such as the glycoprotein IIb/IIIa inhibitors, which could potentially reduce the incidence of acute thromboembolism and improve long-term patency (136). This genuine potential for future improvement has prompted some authors to suggest a direct comparison with carotid endarterectomy may be premature (46, 137, 138, 139).

It remains unclear whether carotid artery stent placement helps reduce the number of particulate emboli by trapping them beneath the metallic meshwork. Current stent designs may trap larger fragments but not efficiently prevent microemboli because the interstices size is too large (140). A recent prospective study, in which patients undergoing carotid artery stenting were examined preprocedurally and postprocedurally with MRI, failed to show evidence of brain signal abnormality referable to emboli. However, a similar prospective study examining 17 patients with diffusion-weighted MRI showed new (clinically silent) lesions in 3 cases (141, 142).

The problem of distal embolization during balloon angioplasty and carotid stenting has increased interest in providing methods of cerebral protection. A triple-coaxial catheter system, designed to provide cerebral protection, has been described by Theron et al (143). Theron et al reported distal embolic complications in 3 of 38 patients (8%) undergoing internal carotid artery balloon angioplasty without distal protection versus 0 of 43 patients (0%) undergoing the same procedure using distal balloon protection (71, 140). Henry et al used Theron's distal occlusion balloon technique in 32 of 163 carotid stenting cases (144). However, 2 of the 3 major strokes that occurred in this series were in conjunction with this device. The authors cited prolonged procedure time and increased embolism risk when traversing ulcerated lesions as potential problems associated with its use.

A number of commercial devices aimed at reducing the microembolic burden associated with carotid angioplasty and/or stenting, using filters or guide wire attached balloons, are currently undergoing development. These include a low-profile embolic filter deployed and retrieved on a 0.014-inch or 0.018-inch shaft that serves as the guidewire for balloon and stent delivery catheters. Studies in ex vivo models have shown >90% capture of particles >200 μm and 100% capture of particles >500 μm (145). Alternatively, a 0.014-inch guidewire with a protection balloon incorporated into the tip has been used. After angioplasty and stenting with the protection balloon inflated, an over-the-wire aspiration catheter is passed through the dilated area to clear debris (146). This balloon-protection method has the disadvantage of temporary occlusion of carotid flow, whereas filter devices allow constant cerebral perfusion.

Patient Selection 

Discussion 

Patients can be selected according to multiple factors such as degree of stenosis, presence of symptoms, and presence of factors that place the patient at high risk for endarterectomy. Degree of stenosis is first measured with noninvasive studies such as carotid duplex sonography, MRA, or CTA. Various endarterectomy and stent trials (Table 3, Table 4) have defined symptomatic patients as those with symptoms within 90 to 180 days of trial entry. For purposes of consistency, it is recommended that symptomatic patients be defined as those with neurological symptoms or neurological changes within 6 months (180 days) of treatment, consistent with NASCET entry criteria. Symptoms may be before stroke or transient neurologic event. Transient events may be hemispheric or retinal (amaurosis). The patient population can be classified as either “low-risk cohort” or “high-risk cohort” of patients with respect to endarterectomy. It is unknown if such patients are also at higher risk if treated with carotid stenting. Although it is generally accepted that some patients can be categorized as high risk, as noted earlier this is not uniformly accepted (60). The low-risk cohort is those patients with similar risk factors who were enrolled into the NASCET and ACAS study. The definition of the high-risk cohort is more controversial. Suggestions for categorizing high-risk patients are listed in Table 5, based on criteria used in previous stent trials and from the criteria of Sundt (Table 2). The high-risk cohort can be defined as having ≥1 anatomical risk factors or ≥2 group A comorbidity risk factors or ≥1 group B comorbidity risk factors as defined in Table 5.

Table 5. Definition of Risk Factors for Carotid Endarterectomy

	Anatomical Risk Factors (≥1)	Comorbidity Risk Factors Group A (≥2)	Comorbidity Risk Factors Group B (≥1)
	Previous carotid endarterectomy Contralateral carotid artery occlusion Previous radical neck dissection or radiation therapy to neck region Target lesion above C2 (level of jaw) Low cervical carotid lesions Dissection Inability to extend neck (ie, cervical osteoarthritis, mobility limitations) Contralateral laryngeal palsy At risk for wound infection Tracheostomy	Patients older than age 80 years Myocardial infarction within previous six mo Unstable angina (defined as resting pain with ECG changes) Requires concurrent CABG (coronary artery bypass grafting), AAA (abdominal aortic aneurysm) repair or peripheral vascular surgery	NYHA (New York Heart Association) class III or IV heart failure COPD (chronic obstructive pulmo-nary disease) with FEV1 (forced expiratory ventilation) <30% predicted History of liver failure with elevated prothrombin time

To avoid confounding the effects of other procedures with the effect of elective carotid revascularization, the selected patients should not undergo any interventional carotid, coronary, or peripheral treatment within 30 days before enrollment or after procedure unless this is part of the study protocol.11

Define Outcome of Therapy 

Pretreatment Evaluation 

Treatment Description 

The results of carotid revascularization will vary depending on operator and institutional experience (43, 56, 147, 148, 149, 150, 151). Previous operator experience should, therefore, be reported.

Medical therapy may need to be adjusted for arteriography. Patients using long-term anticoagulation may need to have their treatment converted to heparin. Patients with preexisting renal disease may be admitted 1 day early for intravenous hydration or monitored vasodilator therapy (152). Oral enteric-coated aspirin and clopidogrel (Plavix) have been recommended at least 3 days before the procedure to reduce periprocedural platelet emboli. Experimental data suggest that aspirin and clopidogrel have a synergistic effect on platelet anti-aggregation, antithrombotic activity and in preventing myointimal proliferation (restenosis) (153, 154). After the procedure, oral antiplatelet agents are usually continued.

The procedure may be performed under general anesthesia but is more commonly performed under local anesthesia at the puncture site along with conscious sedation, which allows continuous monitoring of the patient's neurological status (155). Although patients may have been included in the study based on previous noninvasive imaging or arteriography, the final determination of stenosis severity is made from the arteriogram at the time of intervention (Figure). Because the carotid stenosis may be eccentric, the optimal projection to demonstrate the stenosis should be found, which may require multiple projections or rotational angiography (156). In the case of presumed ICA occlusion, prolonged filming is necessary, because otherwise delayed, faint, antegrade opacification of the cervical ICA may be missed (“string sign” of critical ICA stenosis) (157). Standard AP and lateral intracranial views should be obtained in all cases to establish the adequacy of the intracranial collateral circulation via the external carotid and anterior communicating arteries and also to document any intracranial stenotic lesions. Many experienced operators also advocate routine vertebral angiography to assess collateral flow via the posterior communicating arteries and document any extracranial or intracranial vertebrobasilar stenoses. Certainly, in cases of contralateral carotid occlusion, bilateral severe stenosis, or deficient anterior collateral pathways, evaluation of the adequacy of collateral flow via the posterior circulation becomes even more important.

View Large Image Download to PowerPoint

Figure. Measurement of carotid stenosis using North American Symptomatic Carotid Endarterectomy Trial (NASCET), European Carotid Surgery Trial (ECST), and Common Carotid (CC) methods. All three methods demonstrate a high degree of reproducibility overall. The NASCET method used most frequently in the United States is reliable but tends to underestimate the degree of stenosis. The NASCET ratio should not be applied if there is near-occlusion with reduction in the diameter of the cervical internal carotid artery beyond the stenosis. Such a reduction in the diameter of the internal carotid artery beyond the site of stenosis would consequently underestimate the severity of stenosis. CC and ECST methods grade the stenosis similarly and generally are in agreement. For atherosclerotic disease that narrows the carotid bulb, the percentage difference between NASCET and ECST increases. Carotid stenosis measured at ultrasound tends to correlate better with ECST and CC methods (23, 26, 29, 186, 187, 188).

The use of systemic heparin is standard. Glycoprotein IIb/IIIa inhibitors such as abciximab have been shown to decrease mortality and morbidity in a number of coronary stent studies and also potentially improve long-term patency rates, but their role remains undefined in carotid artery stenting (158, 159, 160). These agents may have a potential role in the uncommon event of acute stent thrombosis because platelet aggregation (white thrombus) represents the primary mechanism (161).

The decision to predilate the lesion using balloon angioplasty depends on the type and size of stent being used, the narrowest lumen diameter, and the morphological configuration of the stenotic segment. Many operators perform routine predilatation to ≈4.0 mm diameter. However, others suggest that the risk of embolism is high during this part of the procedure, and if atraumatic crossing of the lesion with the stent-delivery catheter is possible without predilatation, then it may be attempted.

The currently used stents are either balloon-expandable or self-expandable, with the self-expandable stents made of either stainless steel or Nitinol. Early reports of carotid stenting mainly used balloon-expandable stents such as the Palmaz stent (Johnson & Johnson Interventional Systems), which can usually be positioned with greater precision than self-expandable types (91, 92, 93, 94, 96, 162). However, reports of a 2% to 16% rate of Palmaz stent collapse on follow-up imaging has led to increased use of self-expandable types (91, 108, 163). Mathur et al reported Palmaz stent collapse in 11 of 70 patients (16%) at 6-month angiographic follow-up, attributing this to probable external compression. Furthermore, carotid Doppler ultrasound performed retrospectively in 7 patients with stent compression was only 29% sensitive (163). Compression of balloon-expandable stents has been found to be a significant cause of restenosis in the superficial femoral arteries and hemodialysis grafts (164). The 1999 global survey of 3047 carotid stenting procedures by Wholey et al reported 3033 endovascular carotid stents placed. Balloon-expandable Palmaz stents were used in 47% of cases; self-expandable Wall stents in 46%; and other stents including Strecker (Medi-tech), Integra (Medi-tech), Symphony (Medi-tech), and SMART stents (Cordis) in the remaining 7% of cases (108). In this series, 28 stent deformations occurred exclusively with the Palmaz stent (2%). However, Bergeron et al reported no instances of compression at a mean 13-month follow-up among 96 patients with carotid stenosis, all treated with Palmaz stents (100). Nevertheless, the balloon-expandable stent may not be an ideal choice for superficially exposed arteries such as the internal carotid. However in non-exposed locations such as the vertebral artery or great vessel origins, the superior positional accuracy of deployment may be advantageous. Self-expanding Nitinol stents offer the purported advantage of crush recoverability or “spring-like” behavior. If an external force compresses or deforms a Nitinol stent, then it will reassume its expanded shape on removal of the external stress.

In terms of stent sizing, based on current techniques, the stent margins should optimally extend 1 cm beyond the proximal and distal margins of the stenotic plaque. This may necessitate crossing the external carotid artery origin, which does not usually result in significant clinical sequelae. The stent diameter should be 1 to 2 mm larger than the largest vessel diameter that the stent will need to appose. Stent oversizing leads to a greater metallic coverage of the lesion per unit area, which is theoretically advantageous in preventing distal embolism, reducing tissue prolapse, and is not necessarily associated with a higher rate of restenosis (165, 166).

Postdeployment angioplasty using a high-pressure (12- to 20-atm), semi-compliant balloon may then be performed to closely appose the stent and intima and, moreover, expand regions of residual stent narrowing. This may require the use of varying diameter balloons, particularly if the stent extends into the common carotid artery. Obvious gaps between the stent circumference and endoluminal surface potentially increases the risk of acute or delayed thromboembolism (140). Angioplasty beyond the stent margins is not recommended because it may result in acute vessel dissection, symptomatic vasospasm, or subsequent restenosis. Some operators do not advocate routine postdeployment angioplasty of the stent other than for obvious gaps between the stent and vessel wall and regions of residual stenosis, suggesting that this reduces the risk of embolic complications and improves restenosis rates because of reduced intimal injury.

There is insufficient information to define technical success scientifically. For extremity and renal angioplasty, technical success requires <30% diameter residual stenosis by angiography and may require improvement in trans-stenotic pressure gradient (167, 168). In the coronary literature, technical success for balloon angioplasty and stenting had originally been defined as at least 20% relative improvement with a decrease in stenosis to <50% but has recently been revised to a decrease in stenosis to <20% diameter (169, 170). However, unlike extremity, renal, or coronary stenoses, carotid stenoses are very rarely symptomatic because of hemodynamic compromise. Rather, symptoms arise because of thromboembolism formation on the carotid plaque. It is unknown what degree of correction of carotid stenosis is necessary to reduce the risk of embolization, but removal of the embolic source is fundamental. It is possible that in the attempt to more completely eliminate residual stenosis by full balloon dilation, additional emboli may be produced during the procedure that could cause a higher risk of procedural complications. Alternatively, leaving a higher degree of residual stenosis may lead to a higher rate of late restenosis, which at this time is of uncertain clinical significance. Some carotid stenting trials have defined technical success as residual stenosis <30% (171). Others have used a definition of residual stenosis <50% (172). In the absence of definitive scientific evidence, technical success in this document is arbitrarily defined as stent placement resulting in improvement of the stenosis by >20%, with a final residual stenosis <50% using NASCET measurement criteria. Some practices may prefer to use a lesser degree of residual stenosis as their desired endpoint for technical success.

Iatrogenic vasospasm may occur during the procedure but usually resolves soon after removal of the guide-wire from the internal carotid artery. Some investigators describe the use of injectable nimodipine (200 μg diluted in 10 mL injected slowly asa2mLto3 mL bolus) (140) or nitroglycerin (100 μg) (96) into the carotid artery to treat mechanically induced vasospasm. Recalcitrant iatrogenic vasospasm usually responds well to low-pressure balloon angioplasty (<3 atm).

AP and lateral cerebral angiograms are obtained after stenting to exclude any embolic branch occlusion and to document new patterns of flow. Heparinization is generally allowed to taper physiologically rather than reverse using protamine sulfate. Alternatively, the femoral sheath may be removed immediately after the procedure without the need for heparin reversal using percutaneous closure devices. The patient is usually monitored in the intensive care unit for 12 to 24 hours after treatment. Antiplatelet agents are frequently administered after procedure, such as Clopidogrel for 4 to 6 weeks and enteric-coated aspirin indefinitely.

The patient should be carefully examined neurologically after the procedure. Neurological deficits may be caused by intracranial embolism, hemorrhage, or reperfusion injury (173, 174). After successful revascularization, lowering of the mean arterial pressure to 10% to 20% below baseline may be desirable to prevent cerebral reperfusion injury. Intravenous diltiazem may be used postprocedurally to control elevated blood pressure, particularly if associated with headache or neurological sequelae, because it has minimal cerebral vasodilatory effects (175).

Postprocedural prolonged brady-cardia or hypotension, or both may occur as a result of carotid sinus dysfunction, necessitating the use of intravenous vasopressors or inotropic agents (176).

Evaluation 

Discussion 

Stroke is a complication of carotid revascularization and an outcome measure of the effectiveness of the revascularization. The stroke may be ipsilateral or contralateral to the treated vessel. Neurological deficits as a complication of the procedure may be caused by intracranial embolism, hemorrhage, or reperfusion injury (174). The phenomenon of reperfusion hemorrhage after surgical endarterectomy is well described. Ouriel et al in a review of 1471 patients undergoing surgical endarterectomy reported a 0.75% incidence of intracerebral hemorrhage. Hemorrhage occurred at a median of 3 days postoperatively and accounted for 35% of the total perioperative neurological events. Death from massive hemorrhage and herniation occurred in 36% of cases. Factors correlating with an increased hemorrhage risk were hypertension, high-grade ipsilateral stenosis, high-grade contralateral stenosis or occlusion, and younger age (180). This complication has also been reported with carotid angioplasty. Schoser et al reported 2 patients with reperfusion hemorrhage, 1 intraparenchymal hemorrhage and 1 subarachnoid hemorrhage after balloon angioplasty for carotid and vertebral artery stenosis (174).

Minor, major, and disabling stroke have not been uniformly defined in previous studies of endarterectomy or carotid stenting (Table 4). The authors of this document have attempted to provide a uniform set of definitions for neurological events and a means of classification for these events by category (Table 6). It is felt that these definitions will allow a more specific and relevant categorization of stroke incidence and severity to be consistently described. These definitions are listed.

Table 6. Classification of Complications

	Complication	Class
	Abscess	Infectious/inflammatory
	Angina/coronary ischemia	Cardiac
	Idiosyncratic reaction	Medication-related
	Allergic/anaphylactoid reaction	Contrast-related
	Arterial occlusion/thrombosis, puncture site	Vascular
	Arterial occlusion/thrombosis, remote from puncture site	Vascular
	Arteriovenous fistula	Vascular
	Congestive heart failure	Cardiac
	Device malfunction with adverse effect	Device-related
	Death related to procedure	Death
	Death unrelated to procedure (30-day mortality)	Death
	Embolization, arterial	Vascular
	Fluid/electrolyte imbalance	General nonvascular
	Hematoma bleed, remote site	Vascular
	Hematoma bleed at needle, device path: nonvascular procedure	Vascular
	Hematoma bleed, puncture site: vascular procedure	Vascular
	Incorrect drug	Medication-related
	Incorrect dosage	Medication-related
	Intimal injury/dissection	Vascular
	Ischemia/infarction of tissue/organ	Vascular
	Incorrect site of administration	Medication-related
	Local infection	Infectious/inflammatory
	Liver failure	General nonvascular
	Migration	Device-related
	Myocardial infarction	Cardiac
	Malposition	Device-related
	Nausea/vomiting	General nonvascular
	Other (cardiac)	Cardiac
	Other (contrast-related)	Contrast-related
	Other (CNS complication)	Neurologic
	Other dose-dependent complication	Contrast-related
	Other (device related)	Device-related
	Other (gastrointestinal)	General nonvascular
	Other (general nonvascular)	General nonvascular
	Other (hematologic)	General nonvascular
	Other (infectious/inflammatory)	Infectious/inflammatory
	Other (medication related)	Medication-related
	Other (neurologic)	Neurologic
	Other (respiratory/pulmonary)	Respiratory/pulmonary
	Other (vascular)	Vascular
	Pancreatitis	Infectious/inflammatory
	Pulmonary embolism	Respiratory/pulmonary
	Pulmonary embolism	Vascular
	Peritonitis	Infectious/inflammatory
	Hypotension	Cardiac
	Hypoxia	Respiratory/pulmonary
	Pulmonary edema	Respiratory/pulmonary
	Peripheral nervous system complication	Neurologic
	Pneumothorax	Respiratory/pulmonary
	Pseudoaneurysm	Vascular
	Respiratory arrest	Respiratory/pulmonary
	Renal failure	Contrast-related
	Arrhythmia	Cardiac
	Septicemia/bacteremia	Infectious/inflammatory
	Seizure	Neurologic
	Septic shock	Infectious/inflammatory
	Stroke	Neurologic
	Tissue extravasation	Contrast-related
	Transient ischemic attack	Neurologic
	Unintended perforation of hollow viscus	General nonvascular
	Vascular perforation or rupture	Vascular
	Vagal reaction	Cardiac
	Vasospasm	Vascular
	Venous occlusion/thrombosis, puncture site	Vascular
	Venous occlusion/thrombosis, remote from puncture site	Vascular

Other major complications include myocardial infarction and death, which, in combination with stroke, are the major adverse events used to compare carotid stenting and endarterectomy. Some complications are unique to endovascular as compared with surgical revascularization. Carotid stenting may cause severe bradycardia and hypotension as a result of carotid sinus dysfunction requiring cardiac pacing, although the need for this has diminished with use of intravenous vasopressors or inotropic agents (176). Endovascular revascularization may also produce vascular injuries such as dissection, perforation, hematoma, pseudoaneurysm, and groin infection. Surgical revascularization may produce cranial nerve injuries and local wound hematomas and infections.

Recommendations for Reporting 

Complications within 30 days of the procedure are considered perioperative and procedure related. Neurological deficits are reported using the definitions listed. A stroke may be described as reversible and minor versus major or permanent and minor versus major.

A stroke is any sudden development of neurological deficits attributable to cerebral ischemia and infarction.

Neurological complication—neurological deterioration evidenced by an increase in the NIHSS of ≥1 point.

Transient deficit—a neurological complication having complete resolution within 24 hours.

Reversible stroke—a neurological complication having a duration of >24 hours and ≤30 days.

Permanent stroke—a neurological complication having a duration of >30 days.

Minor deficit—neurological deterioration evidenced by an increase of the NIHSS of <4 points without the presence of aphasia or hemianopsia.

Major deficit—neurological deterioration evidenced by an increase of the NIHSS of ≥4 points or the presence of aphasia or hemianopsia.

Neurological deficits should be listed as ipsilateral or contralateral. Because deaths are frequently secondary to stroke or myocardial infarction, it is misleading to sum the individual totals of stroke, myocar-dial infarctions, and death. Rather, a combined rate of nonfatal stroke, nonfatal myocardial infarction, and death should also be reported, with cause of death described. Other complications should be listed and graded according to previously defined SIR categories of type and severity (Table 6, Table 7).

Table 7. Definitions of Complications

	Minor complications
	No therapy, no consequence
	Nominal therapy, no consequence; includes overnight admission for observation only
	Major complications
	Require therapy, minor hospitalization (<48 h)
	Require major therapy, unplanned increase in level of care, prolonged hospitalization (>48 h)
	Permanent adverse sequelae
	Death

Randomization and Blinding 

Case series may be useful in demonstrating safety and effectiveness of a technique but are unreliable in comparing different therapies. The major advantage of a trial over an observational study is the ability to demonstrate causality. In particular, randomly assigning the intervention can eliminate the influence of confounding variables, and blinding can eliminate the possibility that the observed effects of the intervention are caused by other treatments or by biased ascertainment (96). Because the success and complications of carotid revascularization are so heavily dependent on patient selection, comparisons of revascularization techniques should be made in randomized trials rather than by comparison with historical controls.

It may be impossible for observers to be blinded with respect to therapy (ie, surgery versus endovascular). However, neurological and functional assessments performed at defined time points after treatment should ideally be performed by examiners blinded to the subject's treatment. In addition, a core laboratory should assess the results of the follow-up imaging studies in a manner blinded to treatment assignment and clinical outcome.

Intention to Treat 

Randomized surgical trials of carotid endarterectomy have been reported based on intention to treat. Because of this, some events such as stroke, death, or myocardial infarction may be attributed to the treatment, even if the event occurs before revascularization. Case series will underreport these events, because events are usually reported only in those patients actually treated. The longer the delay between randomization and treatment, the greater the discrepancy between randomized trials and case series. The determination for the study entrance start time should be defined as the randomization contact date/ time. Therefore, analyses in randomized trials should be performed for 3 subject populations: randomized subjects (intent to treat), subjects treated as randomized, and evaluable subjects. Randomized subjects should be defined as all subjects randomized to treatment or control. The population of subjects treated as randomized should include all subjects randomized to treatment who were revascularized and all subjects randomized to control who were not revascularized, if a nonrevascularization arm is included in the study. More likely, studies will include patients treated with different methods of revascularization. The evaluable subject population should be defined as all subjects who did not have certain protocol deviations as specified by the project team. Analysis of this sub-population is helpful to distinguish treatment effect in the intended population from the effect of treating unintended patients included as protocol violations. A complete analysis for each of the 3 populations may be necessary, depending on study design.

Analysis 

Demographics should be summarized, including age, gender, and presence of symptoms. Outcomes should be stratified according to presence of symptoms, degree of stenosis, and risk of endarterectomy, including medical and anatomic comorbidities, as described in Table 5. Patients treated with combined carotid revascularization and coronary artery bypass grafting, if included in the study, should be stratified separately, because it is not possible to identify which of the 2 procedures is responsible for the complication. Groups should be compared for peri-operative (30 days) and long-term (6 months and preferably up to 2 to 3 years) events. Perioperative group comparison should include mortality, any stroke (ischemic versus hemorrhagic), ipsilateral stroke, major versus minor stroke (as defined previously), cardiac events (myocardial infarction, increased congestive heart failure), and medical and procedural complications. If the comparison group includes carotid endarterectomy, then procedural complications should include cranial nerve injuries and neck hematomas. Combined outcomes of stroke plus death, and stroke plus death plus myocardial infarction, should also be compared, because there will be some patients with multiple events.

Because the major stroke morbidity is in the perioperative period, providing annual risk of stroke over time can be misleading. For example, if a procedure has a 10% risk of peri-operative stroke but a 0% risk of stroke subsequently, over 5 years the annual risk of stroke is 2%. The same long-term risk is calculated from a procedure with a 0% risk of perioperative stroke but a 2% risk of stroke over each of the next 5 years. Furthermore, the European Carotid Stroke Trial found that the risk of stroke from an untreated carotid stenosis decreases to the risk of a revascularized carotid after ≈3 years. If long-term data are acquired for 10 years, then the clinical benefit in the initial years will be masked. The long-term benefit is more accurately perceived if the long-term risk is reported excluding the perioperative period (“subsequent annual event rate”). This is not consistent with the reporting used in previous surgical trials; therefore, it is recommended that long-term annual risk of stroke be reported including and excluding the perioperative period. The analysis of long-term data should include Kaplan-Meier or life-table curves indicating freedom from stroke (ipsilateral or any stroke), death, stroke plus death, and restenosis.

Recommendations for Reporting 

Randomized trials should be reported based on intention to treat. Other analyses (patients treated as randomized, assessable patients) may be helpful. Results are presented using life-table or Kaplan-Meier analyses. Results should be stratified according to symptoms, degree of stenosis, and risk factors. Results are reported for stroke (stratified as described), cardiac events, mortality, and restenosis. Annualized event rates should be reported including and excluding the perioperative events.