Routine postendarterectomy duplex surveillance: Does it prevent late stroke? William C. Mackey, M D , Michael Belkin, M D , Rakesh Sindhi, M D , H a r o l d Welch, M D , and Thomas F. O'Donnell, Jr., M D , Boston, Mass. Our recent finding that less than 50% of late postendarterectomy strokes are related to recurrent carotid stenosis led us to question the utility of routine postendarterectomy duplex surveillance (RpCEADS) in the prevention of late stroke. To evaluate our RpCEADS program, we reviewed our postoperative duplex studies and correlated their results with clinical data. A total of 1053 postendarterectomy scans was carried out on 348 carotid arteries (258 patients) (3.0 -+ 0.1 studies/artery) during an average follow-up of 52.6 (+-2.3) months. Less than 50% of recurrent carotid stenosis was doctunented throughout follow-up in 292 (83.9%) of 348 arteries. Recurrent carotid stenosis of greater than 50% or occlusion of either the common or internal carotid artery was noted in the remaining 56 arteries (16.1%). Of the 56 duplex-detected recurrent stenoses, only two (3.6%) resulted directly in an unheralded stroke, whereas eight (14.3%) underwent prophylactic reoperation, eight (14.3%) resulted in transient ischemia requiring reoperation, eight (14.3%) occluded without causing stroke, and 29 (51.8%) remained asymptomatic and did not progress to occlusion. Assuming that each of our eight patients who underwent prophylactic reoperation would have had a stroke if operation had not been carried out and our two unheralded strokes could have been prevented with more rigorous follow-up, RpCEADS might have prevented late stroke related to 10 (2.9%) of 348 arteries in 10 (3.9%) of 258 patients after surgery. All other cases of duplex-detected recurrent carotid stenosis or occlusion were asymptomatic or manifest by transient cerebral ischemia. Therefore RpCEADS cannot be justified as a means of preventing late strokes related to recurrent stenosis. (J VAsc SURG 1992;16:934-40.)
In a recent study, our group determined that late stroke affected 35 (5.1%) of 688 patients after endarterectomy and could be attributed to recurrent carotid stenosis in only 11 (31.4%) of 35 cases. 1 The infrequent occurrence of late stroke after endarterectomy and the finding that late postendarterectomy stroke was most often not attributable to recurrent stenosis led us to question the utility of routine postendarterectomy duplex surveillance as a means of prevention of late stroke. To evaluate our program of postendarterectomy duplex follow-up, we reviewed all postoperative carotid duplex scans and correlated the duplex findings with the patients' clinical courses. From the Department of Surgery, Tufts University School of Medicine, and New England Medical Center, Boston. Presented at the Sixth Annual Meeting of the Eastern Vascular Society, New York, N.Y., April 30-May 3, 1992. Reprint requests: William C. Mackey, MD, Box 1035, New England Medical Center, 750 Washington St., Boston, MA 02111. 24/6/41705
934
METHODS In 1983 duplex ultrasonography became available for routine use in the New England Medical Center's noninvasive vascular laboratory. Since then all patients returning for follow-up after carotid endarterectomy have been subjected to routine duplex examinations to detect recurrent stenosis and monitor the opposite carotid artery. Our usual follow-up protocol includes duplex ultrasonography at 3 to 6 months, 12 months, and then yearly after endarterectomy. More frequent studies are obtained in patients whose routine studies reveal evidence for progressive recurrence, who have a prior history of recurrent stenoses, or who have neurologic symptoms. Most of the duplex studies in this report were carried out with Biosound or Biosound Phase II instruments (Biosound Inc., Indianapolis, Ind.). The most recent studies were carried out with an ATL Ultramark 9 instrument (Advanced Technologies Laboratories, Bothell, Wash.). The records at the New England Medical Center noninvasive vascular laboratory were reviewed for all
Volume 16 Number 6 December 1992
patients who had undergone previous carotid endarterectomy and at least one postendarterectomy duplex ultrasound examination. For each patient the interval between surgery and each postoperative duplex study, the total number of postoperative studies, and the duration of foUow-up (interval from surgery to most recent duplex study) were recorded. The results of each postendarterectomy duplex study, including degree of common carotid stenosis, degree of internal carotid stenosis, and plaque morphology in both the common and internal carotid arteries, were also recorded. The degree of stenosis in the common carotid and internal carotid arteries was categorized as follows: no data, normal (< 30%), 30% to 49%, 50% to 74%, 75% to 99%, and occlusion. Plaque morphology was recorded as intimal thickening, soft, smooth plaque, complex plaque with or without ulceration, or occlusion. Clinical follow-up data for correlation with the noninvasive data were obtained from our carotid follow-up registry, hospital records, and office records. Gender, age at surgery, risk factors for vascular disease at time of surgery, postendarterectomy neurologic events, and interval from surgery to any such neurologic events were recorded. For the purposes of this study, postendarterectomy stroke was defined as a permanent neurologic deficit of any purported cause referable to the hemisphere supplied by the postendarterectomy carotid artery. Also, for the purposes of this study, perioperative strokes were excluded as life-table end points. Patients who had duplex evidence for occlusion or recurrent stenosis of at least 50% of the common or internal carotid artery at any time during follow-up were considered to have significant recurrent stenosis. Life-table stroke-free rates for patients with and without recurrent stenosis by duplex scan were compared. All data were computer filed with dBASE III (Ashton-Tate, Los Angeles, Calif.). Statistical analyses were carried out with ABSTAT (Anderson-Bell, Carson City, Colo.). All means are expressed as the mean + SEM. Life tables were constructed to conform with the guidelines set forth by the committee for reporting standards of the Society for Vascular Surgery and International Society for Cardiovascular Surgery. 2 Life tables were compared with the log rank test. 3 Statistical significance was inferred for p < 0.05. RESULTS
Review of our noninvasive laboratory records revealed 1053 follow-up studies in 348 operated
Routine postendarterectomyduplex surveillance 935
Table I. Demographics and risk factor prevalence Sex Male Female Age (yr)
Risk factors Hypertension Cigarettes Angina Hyperlipidemia Diabetes Myocardialinfarction Unknown
No./total
%
154/258 104/258 64.7 (+ 0.43 SEM)
59.7 40.3
151/258 116/258 64/258 47/258 43/258 37/258 25/258
58.5 45.0 24.8 18.2 16.7 14.3 9.7
carotid arteries in 258 patients. There were 3.0 --- 0.1 postendarterectomy studies per artery (range I to 9). The average duration of postoperative follow-up (interval from surgery to most recent duplex study) was 52.6 _+ 2.3 months (range 4 to 240 months). Demographic and risk factor data for our patients are shown in Table I. As noted in our other carotid studies, risk factors for peripheral vascular disease, especially hypertension and cigarette use, were common in these patients. ~ Comparison of demographic a n d risk factor data in patients with (n = 50) and without (n = 208) recurrent stenosis are noted in Table II. The statistically significantly higher prevalence of hyperlipidemia in the group with recurrent stenosis has been documented by other investigators. 5 The follow-up duplex findings are summarized in Table III. The data in Table III are based on the maximal restenosis noted during follow-up in either the internal or common carotid arteries. Based on these data, 292 (83.9%) of 348 operated carotid arteries did not develop significant recurrent stenosis (___50%) during the period of follow-up, and 56 (16.1%) of 348 developed significant restenosis or occlusion. The mean lengths of follow-up were 84.3 _+ 6.9 months in the group with recurrent stenosis or occluson and 46.5 ___2.2 months in the group without significant recurrent stenosis < 0.0001 by nonpaired t test). The mean interval from surgery to maximal recurrence in the group with recurrent stenosis was 58.8--_ 6.4 months (p = 0.03 by nonpaired t test compared with length of follow-up for patients without significant recurrent stenosis (46.5 ___2.2 months). The interval from surgery to the diagnosis of the maximal recurrent stenosis for the 56 arteries that developed 50% or greater restenosis is illustrated in Fig. 1. Twenty-six recurrences (46.4%) were diag-
936
Journalof VASCULAR SURGERY
Mackey et al.
Table II. Demographics and risk factors in patients with and without recurrent stenosis Recurrent stenosis
No rec.u_rrentstenosis
(n = 5o)
(n - 208)
No./total
Age (yr) Sex Female Male Risk factors Hypertension Cigarettes Angina Hyperlipidemia Diabetes Cardiac infarction Unknown
%
65.2 _+ 0.55
No./total
%
74.4 _+ 12.1
p Value
0.12"
21/50 29/50
42
82/208 126/208
39.4
0.80t
32/50 28/50 10/50 19/50 8/50 7/50 5/50
64 56 20 38 16 14 10
119/208 88/208 54/208 28/208 35/208 30/208 20/208
57.2 42.3 26.0 13.5 16.8 14.4 9°6
0.57t 0.20t 0.45~ 0.0003t 0.90t 0.94t 0.94t
*By unpaired t test. tBy x2 analysis.
Table HI. Postendarterectomy duplex findings
Normal 30%-49% Maximal restenosis 50%-74% Maximal restenosis 75%-99% Maxmal restenosis Occlusion
No./total
%
205/348 87/348 26/348 19/348 11/348
58.9 25.0 7.5 5.5 3.2
nosed within 3 years of surgery. The remaining 30 recurrences (53.6%) were first noted after 3 years from the time of surgery, and the majority of these (23/30) were first diagnosed 5 years from the time of surgery. The clinical follow-up data for 55 of the 56 arteries with recurrent stenosis or occlusion are shown in Table IV. One artery was omitted from this analysis because of immediate postendarterectomy thrombosis and stroke. Patency could not be maintained after immediate reoperation, and this perioperative occlusion was documented on follow-up duplex studies. Of the remaining 55 arteries with significant recurrence, 29 (52.8%) remained asymptomatic and patent throughout the follow-up period and did not undergo reoperation. Eight arteries (14.5%) were found to have occluded but caused only transient cerebral ischemia or no symptoms. Sixteen arteries (29.1%) tmderwent reoperation, eight (14.5%) because of recurrent symptoms and eight (14.5%) because of an asymptomatic highgrade ( > 80%) restenosis. Three arteries with recurrent stenosis (5.5%) caused strokes. One of these strokes was related to an early high-grade recurrent stenosis, one was related to a carotid occlusion that
occurred with a cardiac arrest 40 months after surgery, and one occurred with early reocclusion after one of the eight reoperations for asymptomatic high-grade recurrent stenosis diagnosed 8 years after the initial operation. If only the 19 arteries with 75% to 99% restenosis are considered, 10 (52.6%) underwent reoperation (six [31.6%] for symptoms of transient cerebral ischemia and four [21.1%] for asymptomatic restenosis), seven (36.8%) remained asymptomatic and did not undergo reoperation, and two (10.5%) resulted in strokes. Overall, symptoms related to recurrent stenosis or occlusion occurred in 19 (34.5%) of 55 cases with 50% or greater restenosis. In comparison, symptoms of stroke or transient ischemic attack occurred after surgery in 12 (4.1%) of 292 cases with less than 50% recurrence (p < 0.0001 by ×2). Most of the late symptoms in both groups were transient ischemic attacks. Artery-specific stroke incidences were three (5.5%) of 55 and five (1.7%) of 292 in those with and without recurrent stenosis, respectively (p = 0.12 by Fishers exact test). Patient-specific life-table stroke-free rates for patients with (n = 49) and without (n = 208) recurrent stenosis are shown in Table V. There is no statistically significant difference between stroke-free rates in the two groups (p --0.64 by the log rank test). DISCUSSION Postendarterectomy duplex ultrasound surveillance detected significant ( ___50%) recurrent stenosis in 16.1% of arteries examined. The statistically significant difference (p = 0.03) between time to maximal restenosis in the group with recurrent
Volume
16
Number 6 December 1992
Routine postendarterectomy duplex surveillance
+I
937
of arterles
21
18
1S-
96:30
I
In a recent study, our group determined that late stroke affected 35 (5.1%) of 688 patients after endarterectomy and could be attributed to recurrent carotid stenosis in only 11 (31.4%) of 35 cases. 1 The infrequent occurrence of late stroke after endarterectomy and the finding that late postendarterectomy stroke was most often not attributable to recurrent stenosis led us to question the utility of routine postendarterectomy duplex surveillance as a means of prevention of late stroke. To evaluate our program of postendarterectomy duplex follow-up, we reviewed all postoperative carotid duplex scans and correlated the duplex findings with the patients' clinical courses. From the Department of Surgery, Tufts University School of Medicine, and New England Medical Center, Boston. Presented at the Sixth Annual Meeting of the Eastern Vascular Society, New York, N.Y., April 30-May 3, 1992. Reprint requests: William C. Mackey, MD, Box 1035, New England Medical Center, 750 Washington St., Boston, MA 02111. 24/6/41705
934
METHODS In 1983 duplex ultrasonography became available for routine use in the New England Medical Center's noninvasive vascular laboratory. Since then all patients returning for follow-up after carotid endarterectomy have been subjected to routine duplex examinations to detect recurrent stenosis and monitor the opposite carotid artery. Our usual follow-up protocol includes duplex ultrasonography at 3 to 6 months, 12 months, and then yearly after endarterectomy. More frequent studies are obtained in patients whose routine studies reveal evidence for progressive recurrence, who have a prior history of recurrent stenoses, or who have neurologic symptoms. Most of the duplex studies in this report were carried out with Biosound or Biosound Phase II instruments (Biosound Inc., Indianapolis, Ind.). The most recent studies were carried out with an ATL Ultramark 9 instrument (Advanced Technologies Laboratories, Bothell, Wash.). The records at the New England Medical Center noninvasive vascular laboratory were reviewed for all
Volume 16 Number 6 December 1992
patients who had undergone previous carotid endarterectomy and at least one postendarterectomy duplex ultrasound examination. For each patient the interval between surgery and each postoperative duplex study, the total number of postoperative studies, and the duration of foUow-up (interval from surgery to most recent duplex study) were recorded. The results of each postendarterectomy duplex study, including degree of common carotid stenosis, degree of internal carotid stenosis, and plaque morphology in both the common and internal carotid arteries, were also recorded. The degree of stenosis in the common carotid and internal carotid arteries was categorized as follows: no data, normal (< 30%), 30% to 49%, 50% to 74%, 75% to 99%, and occlusion. Plaque morphology was recorded as intimal thickening, soft, smooth plaque, complex plaque with or without ulceration, or occlusion. Clinical follow-up data for correlation with the noninvasive data were obtained from our carotid follow-up registry, hospital records, and office records. Gender, age at surgery, risk factors for vascular disease at time of surgery, postendarterectomy neurologic events, and interval from surgery to any such neurologic events were recorded. For the purposes of this study, postendarterectomy stroke was defined as a permanent neurologic deficit of any purported cause referable to the hemisphere supplied by the postendarterectomy carotid artery. Also, for the purposes of this study, perioperative strokes were excluded as life-table end points. Patients who had duplex evidence for occlusion or recurrent stenosis of at least 50% of the common or internal carotid artery at any time during follow-up were considered to have significant recurrent stenosis. Life-table stroke-free rates for patients with and without recurrent stenosis by duplex scan were compared. All data were computer filed with dBASE III (Ashton-Tate, Los Angeles, Calif.). Statistical analyses were carried out with ABSTAT (Anderson-Bell, Carson City, Colo.). All means are expressed as the mean + SEM. Life tables were constructed to conform with the guidelines set forth by the committee for reporting standards of the Society for Vascular Surgery and International Society for Cardiovascular Surgery. 2 Life tables were compared with the log rank test. 3 Statistical significance was inferred for p < 0.05. RESULTS
Review of our noninvasive laboratory records revealed 1053 follow-up studies in 348 operated
Routine postendarterectomyduplex surveillance 935
Table I. Demographics and risk factor prevalence Sex Male Female Age (yr)
Risk factors Hypertension Cigarettes Angina Hyperlipidemia Diabetes Myocardialinfarction Unknown
No./total
%
154/258 104/258 64.7 (+ 0.43 SEM)
59.7 40.3
151/258 116/258 64/258 47/258 43/258 37/258 25/258
58.5 45.0 24.8 18.2 16.7 14.3 9.7
carotid arteries in 258 patients. There were 3.0 --- 0.1 postendarterectomy studies per artery (range I to 9). The average duration of postoperative follow-up (interval from surgery to most recent duplex study) was 52.6 _+ 2.3 months (range 4 to 240 months). Demographic and risk factor data for our patients are shown in Table I. As noted in our other carotid studies, risk factors for peripheral vascular disease, especially hypertension and cigarette use, were common in these patients. ~ Comparison of demographic a n d risk factor data in patients with (n = 50) and without (n = 208) recurrent stenosis are noted in Table II. The statistically significantly higher prevalence of hyperlipidemia in the group with recurrent stenosis has been documented by other investigators. 5 The follow-up duplex findings are summarized in Table III. The data in Table III are based on the maximal restenosis noted during follow-up in either the internal or common carotid arteries. Based on these data, 292 (83.9%) of 348 operated carotid arteries did not develop significant recurrent stenosis (___50%) during the period of follow-up, and 56 (16.1%) of 348 developed significant restenosis or occlusion. The mean lengths of follow-up were 84.3 _+ 6.9 months in the group with recurrent stenosis or occluson and 46.5 ___2.2 months in the group without significant recurrent stenosis < 0.0001 by nonpaired t test). The mean interval from surgery to maximal recurrence in the group with recurrent stenosis was 58.8--_ 6.4 months (p = 0.03 by nonpaired t test compared with length of follow-up for patients without significant recurrent stenosis (46.5 ___2.2 months). The interval from surgery to the diagnosis of the maximal recurrent stenosis for the 56 arteries that developed 50% or greater restenosis is illustrated in Fig. 1. Twenty-six recurrences (46.4%) were diag-
936
Journalof VASCULAR SURGERY
Mackey et al.
Table II. Demographics and risk factors in patients with and without recurrent stenosis Recurrent stenosis
No rec.u_rrentstenosis
(n = 5o)
(n - 208)
No./total
Age (yr) Sex Female Male Risk factors Hypertension Cigarettes Angina Hyperlipidemia Diabetes Cardiac infarction Unknown
%
65.2 _+ 0.55
No./total
%
74.4 _+ 12.1
p Value
0.12"
21/50 29/50
42
82/208 126/208
39.4
0.80t
32/50 28/50 10/50 19/50 8/50 7/50 5/50
64 56 20 38 16 14 10
119/208 88/208 54/208 28/208 35/208 30/208 20/208
57.2 42.3 26.0 13.5 16.8 14.4 9°6
0.57t 0.20t 0.45~ 0.0003t 0.90t 0.94t 0.94t
*By unpaired t test. tBy x2 analysis.
Table HI. Postendarterectomy duplex findings
Normal 30%-49% Maximal restenosis 50%-74% Maximal restenosis 75%-99% Maxmal restenosis Occlusion
No./total
%
205/348 87/348 26/348 19/348 11/348
58.9 25.0 7.5 5.5 3.2
nosed within 3 years of surgery. The remaining 30 recurrences (53.6%) were first noted after 3 years from the time of surgery, and the majority of these (23/30) were first diagnosed 5 years from the time of surgery. The clinical follow-up data for 55 of the 56 arteries with recurrent stenosis or occlusion are shown in Table IV. One artery was omitted from this analysis because of immediate postendarterectomy thrombosis and stroke. Patency could not be maintained after immediate reoperation, and this perioperative occlusion was documented on follow-up duplex studies. Of the remaining 55 arteries with significant recurrence, 29 (52.8%) remained asymptomatic and patent throughout the follow-up period and did not undergo reoperation. Eight arteries (14.5%) were found to have occluded but caused only transient cerebral ischemia or no symptoms. Sixteen arteries (29.1%) tmderwent reoperation, eight (14.5%) because of recurrent symptoms and eight (14.5%) because of an asymptomatic highgrade ( > 80%) restenosis. Three arteries with recurrent stenosis (5.5%) caused strokes. One of these strokes was related to an early high-grade recurrent stenosis, one was related to a carotid occlusion that
occurred with a cardiac arrest 40 months after surgery, and one occurred with early reocclusion after one of the eight reoperations for asymptomatic high-grade recurrent stenosis diagnosed 8 years after the initial operation. If only the 19 arteries with 75% to 99% restenosis are considered, 10 (52.6%) underwent reoperation (six [31.6%] for symptoms of transient cerebral ischemia and four [21.1%] for asymptomatic restenosis), seven (36.8%) remained asymptomatic and did not undergo reoperation, and two (10.5%) resulted in strokes. Overall, symptoms related to recurrent stenosis or occlusion occurred in 19 (34.5%) of 55 cases with 50% or greater restenosis. In comparison, symptoms of stroke or transient ischemic attack occurred after surgery in 12 (4.1%) of 292 cases with less than 50% recurrence (p < 0.0001 by ×2). Most of the late symptoms in both groups were transient ischemic attacks. Artery-specific stroke incidences were three (5.5%) of 55 and five (1.7%) of 292 in those with and without recurrent stenosis, respectively (p = 0.12 by Fishers exact test). Patient-specific life-table stroke-free rates for patients with (n = 49) and without (n = 208) recurrent stenosis are shown in Table V. There is no statistically significant difference between stroke-free rates in the two groups (p --0.64 by the log rank test). DISCUSSION Postendarterectomy duplex ultrasound surveillance detected significant ( ___50%) recurrent stenosis in 16.1% of arteries examined. The statistically significant difference (p = 0.03) between time to maximal restenosis in the group with recurrent
Volume
16
Number 6 December 1992
Routine postendarterectomy duplex surveillance
+I
937
of arterles
21
18
1S-
96:30
I
