Carotid-Subclavian Bypass for Brachiocephalic Occlusive Disease
Choice of Conduit and Long-term Follow-up
Background and Purpose Atherosclerotic disease of the proximal brachiocephalic circulation may produce disabling symptoms referable to cerebral or upper extremity hypoperfusion and embolization. Bypass of occlusive lesions can provide durable relief of symptoms with minimal complications. The ideal conduit for carotid-to-subclavian and subclavian-to-carotid bypass remains controversial, and it is not clear whether the outflow vessel influences patency and survival.
Methods We performed a retrospective analysis of 60 consecutive carotid-to-subclavian and subclavian-to-carotid bypass procedures. Occlusive lesions were documented preoperatively by arteriography. Patency was determined during follow-up by ultrasound or duplex examination. Actuarial patency, symptom-free survival, and overall survival rates were calculated by the life-table method and analyzed by log-rank test.
Results Arterial transposition demonstrated the highest long-term patency rate (100.0±0.0%). Polytetrafluoroethylene grafts demonstrated the highest bypass graft patency rate (95.2±4.6%), followed by Dacron grafts (83.9±10.5%) and saphenous vein grafts (64.8±16.5%). Symptom-free survival paralleled patency rates, but these differences did not achieve statistical significance. While there were no differences in patency or symptom-free survival by outflow vessel, the overall survival of patients with common carotid lesions was significantly lower than that of patients with subclavian lesions (62.7±12.8% versus 100.0±0.0%; P<.05).
Conclusions The outflow vessel does not affect long-term patency in carotid and subclavian bypass procedures; however, patients with common carotid disease demonstrate significantly poorer long-term survival. Transposition results in superior long-term patency, with a trend toward lower results for synthetic grafts and relatively poor results for autogenous vein grafts.
Atherosclerotic disease of the proximal brachiocephalic circulation may produce disabling symptoms referable to cerebral or upper extremity hypoperfusion and embolization. Bypass of occlusive lesions can provide durable relief of symptoms with minimal complications. The ideal conduit for C→SC and SC→C bypass remains controversial, and it is not clear whether the outflow vessel influences patency and survival.
Subjects and Methods
From January 1973 to August 1992, 60 extrathoracic C→SC bypass operations were performed on 53 patients with symptomatic occlusive disease of one or more branches of the aortic arch. These 60 operations included both C→SC bypass for subclavian artery occlusive disease and SC→C bypass for innominate or carotid artery occlusive disease. The hospital and outpatient records of these patients were retrospectively reviewed to determine symptoms at presentation, the location and degree of occlusion as demonstrated by aortic arch angiography, details of the surgical procedure performed, immediate postoperative results, relief of symptoms, long-term patency, and survival. The interval of follow-up ranged from 0.9 to 168.4 months, with a mean of 44.0 months.
The sex distribution of this group of patients was essentially even, with 27 men and 26 women. Patient age at the time of operation ranged from 22 to 77 years, with a mean of 59 years. A history of cigarette smoking was by far the most prevalent risk factor for carotid or subclavian occlusive disease, present in 74% of patients. Other prevalent risk factors included a history of hypertension in 42%, coronary artery disease in 30%, and hypercholesterolemia in 17%. Diabetes was present in only 8% of patients.
Forty-nine of the 53 patients presented with neurological symptoms or symptoms of upper extremity arterial insufficiency, which initiated an evaluation of the cerebrovascular and upper extremity circulation by aortic arch, cerebral, and upper extremity angiography. The etiology of all stenotic, occlusive, and aneurysmal lesions identified by angiography was atherosclerotic disease.
Thirty-two patients were found to have subclavian artery stenosis, occlusion, or aneurysm and underwent a total of 37 C→SC bypass procedures. The nature of occlusive disease in both outflow groups is detailed in Table 1⇓. Eighteen procedures were performed for isolated subclavian artery stenosis. Two patients were found to have an associated internal carotid stenosis (1 ipsilateral, 1 contralateral) that required concomitant internal carotid endarterectomy. Thirteen procedures were performed for patients with complete occlusion of the subclavian artery. Occlusive disease was much more common in the left subclavian artery, with 16 stenoses and 11 occlusions, compared with 4 stenoses and 2 occlusions on the right. Additionally, there were 3 subclavian artery aneurysms and 1 symptomatic proximal anastomotic stenosis of a previously placed saphenous vein C→SC bypass graft.
Indications for C→SC bypass (Table 2⇓) were symptoms of vertebrobasilar insufficiency (subclavian steal) in 21 patients, upper extremity claudication and/or weakness in 10, upper extremity ischemia in 3, subclavian aneurysm in 3, and stroke in 1 (some patients presented with more than one symptom type). Of the 2 patients with subclavian artery aneurysm, 1 reported dyspnea and stridor due to airway compression, and 1 reported upper extremity neurological symptoms presumed secondary to brachial plexus impingement and had a contralateral asymptomatic subclavian artery aneurysm.
Twenty-one patients were found on angiography to have innominate (n=4) or common carotid (n=17) artery stenosis or occlusion, and they underwent a total of 23 SC→C bypass procedures. The 4 patients with innominate artery occlusive disease underwent “crossover” bypass from the left subclavian to the right common carotid artery. There were 8 patients with an isolated stenosis of the innominate (n=2) or common carotid (n=6) artery; 1 of these patients had an associated ipsilateral internal carotid stenosis requiring concomitant endarterectomy, and 1 had an associated ipsilateral vertebral artery occlusion for which a SC→C graft to vertebral artery branch graft was constructed. Thirteen patients were found to have complete occlusion of the innominate (n=2) or common carotid (n=11) artery. Eleven patients had occlusive lesions of the left common carotid artery, 4 had lesions of the innominate, and 6 had lesions of the right common carotid artery. There was 1 patient with a symptomatic anastomotic stenosis of a previously placed saphenous vein bypass graft and 1 with an infection of a previously placed prosthetic SC→C bypass graft, for a total of 23 SC→C bypasses.
The indications for SC→C bypass (Table 2⇑) were TIAs of the anterior circulation in 13 patients, stroke in 3 patients, syncope in 1, and graft infection in 1. Four patients underwent SC→C bypass for asymptomatic high-grade common carotid artery stenosis with contralateral carotid artery occlusion.
The mean age at operation of patients undergoing C→SC bypass was 57 years compared with 61 years for patients undergoing SC→C bypass. Revascularization was accomplished with the use of the following conduits: PTFE grafts in 25 patients, Dacron grafts in 15, autogenous saphenous vein in 11, and arterial transposition in 9. The distribution of conduit types by outflow vessel is detailed in Table 3⇓. Each conduit was used in more subclavian outflow bypasses than carotid. The type of conduit used in each case was determined by the personal preference of the attending surgeon. All operating surgeons were vascular surgery faculty of the UCLA Department of Surgery.
Inpatient medical records were reviewed to determine the incidence and nature of complications. Patients were followed up by routine outpatient visits, in most cases every 6 months for the first year and then annually. Outpatient evaluation included historical information regarding recurrence of symptoms, physical examination, and duplex examination of the bypass to determine patency. No bypasses were considered patent unless unequivocally proven by ultrasound or duplex examination. Results were analyzed in terms of long-term patency, relief of symptoms, and survival and were calculated by the Kaplan-Meier life-table method. Results are presented graphically with the calculated standard error for each interval reported in parentheses. Statistical evaluation of life-table data was performed with the use of the log-rank test. Statistical significance was defined as P<.05.
The 5-year Kaplan-Meier patency rate for all 60 bypasses was 87.5±4.8%. The vessel of outflow had no effect on long-term patency, as the patency rate for SC→C bypass grafts was 88.2±8.0% compared with 86.7±6.2% for C→SC grafts (Fig 1⇓).
Of the four conduit types, arterial transposition demonstrated the best long-term results, with a 5-year patency rate of 100.0±0.0%. Only one occlusion of an arterial transposition was observed during the entire follow-up period, occurring at 168.4 months (>13 years) postoperatively. This patient was asymptomatic at the time the occlusion was detected by duplex examination. PTFE grafts demonstrated the highest patency rate of three bypass grafts used, with a 5-year patency of 95.2±4.6%, followed by Dacron grafts at 83.9±10.5% (Fig 2⇓). Saphenous vein grafts demonstrated 5-year actuarial patency of 64.8±16.5%. The differences between these four groups were not statistically significant (P=.20).
Four patients required an interventional procedure or operation during the follow-up interval to relieve symptoms due to bypass stenosis. One C→SC transposition patient underwent angioplasty at 4 and 8 months postoperatively for an anastomotic stenosis and was patent and asymptomatic at last follow-up (16.9 months). One C→SC saphenous vein bypass patient underwent angioplasty 16 months postoperatively for a proximal anastomotic stenosis at the time of an ipsilateral internal carotid endarterectomy and was patent and asymptomatic at last follow-up (89.9 months). One SC→C PTFE bypass patient underwent operative revision of the distal anastamosis at 75 months for recurrence of symptoms and was also patent and asymptomatic at last follow-up (136.3 months). One C→SC saphenous vein bypass patient required surgical graft thrombectomy immediately after operation and was found to be symptomatic with an occluded graft at 12.9 months of follow-up. The patency data reflect primary, unassisted patency for all other bypasses.
Symptom-Free Survival Rates
Symptom-free survival rates closely paralleled patency rates. The 5-year symptom-free survival rate for all patients was 85.5±5.1%. The outflow vessel had no significant effect on symptom-free survival rates, as 86.6±6.3% of C→SC bypass patients and 83.2±8.9% of SC→C bypass patients reported no return of symptoms referable to subclavian or innominate/carotid occlusive disease at last follow-up (Fig 3⇓). Five-year symptom-free survival rates by type of conduit paralleled the observed patency rates: 100.0±0.0% for transpositions, 90.5±6.4% for PTFE grafts, 83.9±10.4% for Dacron grafts, and 63.6±17.7% for saphenous vein grafts (Fig 4⇓). As for patency rates, the symptom-free survival rates demonstrated the best results for arterial transposition, with slightly poorer performance for synthetic grafts and the worst performance for vein grafts. These differences, however, did not reach statistical significance (P=.19).
One patient with a left carotid occlusion and left hemispheric TIAs who underwent left SC→C bypass reported persistent symptoms at last follow-up despite a patent bypass graft. This patient was lost to follow-up at 1.8 months, and an explanation for the persistence of symptoms was not determined.
Overall Survival Rates
Five-year survival for the 60 patients as a whole was 83.5±6.6%. The anatomic location of the bypassed occlusive disease markedly influenced long-term survival (Fig 5⇓). There were no deaths during the follow-up period among patients with subclavian stenosis undergoing C→SC bypass, for a 5-year survival of 100.0±0.0%. This contrasts with a 5-year survival of 62.7±12.8% for those patients with innominate or carotid disease undergoing SC→C bypass (P<.05). There were no differences in mortality between the four conduit types.
Seven of the 23 SC→C bypass patients are known to have died during the follow-up period. One perioperative death due to myocardial infarction occurred 3 weeks after surgery. Other atherosclerosis-related deaths included the following: 1 respiratory death due to complications of a perioperative stroke at 13 months, 1 fatal myocardial infarction at 39 months, and 1 death due to heart failure at 79 months. There were two cancer-related deaths: 1 due to laryngeal cancer at 17 months and 1 due to gastrointestinal cancer at 45 months. There was 1 death at 15 months due to respiratory complications of a perioperative respiratory arrest.
Twelve operative complications occurred in 11 patients, for an overall complication rate of 18.3%. These complications included 2 strokes (stroke rate, 3.3%), 3 thoracic duct injuries, 2 nerve injuries, 2 hematomas, 1 myocardial infarction, 1 respiratory arrest with hypoxic brain injury, and 1 graft thrombosis. There was 1 death due to myocardial infarction within 30 days of surgery, for an operative mortality rate of 1.7%. Three patients underwent reoperation during the same admission: 1 for thoracic duct ligation, 1 for evacuation of a hematoma, and 1 for Fogarty catheter thrombectomy of a graft thrombosis. This represents an early reoperation rate of 5.0%.
The operative complication rate was greater for SC→C bypasses, at 35% (8/23), compared with a rate of 11% (4/37) for C→SC bypasses (P<.05). Both myocardial infarctions and nerve injuries occurred in the SC→C bypass group. The nerve injuries involved transient unilateral vocal cord paralysis presumed secondary to traction or crush injury of the recurrent laryngeal nerve. There was 1 stroke and 1 wound hematoma in each outflow group. Two of 3 thoracic duct injuries occurred in the C→SC bypass group. Two of these injuries were managed nonoperatively by temporary thoracostomy tube drainage of a chylous pleural effusion and bowel rest with parenteral nutrition; 1 patient underwent reoperation with thoracic duct ligation.
The 2 patients who suffered a postoperative stroke underwent urgent extrathoracic revascularization for crescendo TIAs and limb-threatening ischemia, respectively. One of these patients had a known recent left carotid artery occlusion and developed left-hemispheric TIAs, for which a left SC→C bypass was performed. The other patient was found to have an occluded left subclavian artery on evaluation for severe left upper extremity ischemia with distal gangrenous changes and suffered a left middle cerebral artery distribution stroke after left C→SC bypass.
One graft infection was observed in this group of patients during the follow-up interval, for an infection rate of 1.7%. This occurred in a Dacron SC→C bypass at 17.6 months of follow-up. This complication required graft removal and revision; however, follow-up information for this procedure was not available.
Other late complications after SC→C bypass included 1 saphenous vein bypass patient with return of symptoms at 1.8 months despite a patent graft, 1 PTFE bypass patient with recurrent symptoms at 75 months requiring graft revision, and 1 Dacron bypass patient who suffered an ipsilateral stroke 12 months postoperatively. This patient was found to have a patent graft, with an ulcerated common carotid artery plaque proximal to the graft–carotid artery anastamosis. Ligation of the proximal common carotid artery was effective in preventing further neurological events. The incidence of late stroke was thus 4.3% for SC→C bypass patients.
Late complications among C→SC bypass patients included 2 patients with stroke, for a late stroke incidence of 5.4%, similar to that of SC→C bypass patients. One patient who underwent C→SC bypass for upper extremity ischemia experienced an ipsilateral stroke at 10.9 months and was found to have an occluded graft. The embolic source for this event could not be determined. The other stroke occurred at 14.0 months in a saphenous vein bypass patient who was found on angiography to have a critical ipsilateral internal carotid artery stenosis, as well as stenosis of the proximal graft anastamosis. This patient underwent internal carotid endarterectomy and graft angioplasty, with no further neurological events and documented graft patency at 89.9 months.
Other late complications among C→SC bypass patients included 1 saphenous vein graft thrombosis at 12.9 months with return of symptoms and 2 patients previously described who required angioplasty to maintain graft patency and relieve recurrent symptoms (1 patient at 4 and 8 months, the other at 16 months).
Choice of Conduit
The first 30 bypasses were performed from 1973 to 1983. The choice of conduit for these bypasses included 11 Dacron grafts, 10 saphenous vein grafts, 6 PTFE grafts, and 3 arterial transpositions (Fig 6⇓). Follow-up data on conduit performance from the earlier patients in the series1 clearly influenced subsequent choice of conduit, as the use of vein decreased in favor of transposition and synthetic grafts, and among synthetic grafts the use of Dacron decreased in favor of PTFE. Conduits used for the last 30 bypasses from 1984 to 1992 were as follows: 18 PTFE grafts, 6 arterial transpositions, 4 Dacron grafts, and 1 saphenous vein graft.
Proximal occlusive disease of the common carotid and subclavian arteries is encountered infrequently among patients with peripheral vascular disease, particularly in comparison with carotid bifurcation disease. In the Joint Study of Extracranial Arterial Occlusion, only 17% of 6534 patients were found to have a stenosis of 30% or greater in either the innominate or subclavian arteries.2 Accumulated experience with such patients suggests that in most cases these lesions remain asymptomatic and progress slowly.3 4 5 6 It is therefore difficult to make a convincing argument for prophylactic surgical intervention in asymptomatic patients with proximal subclavian or carotid artery disease.
Because of the very low risk of suffering an infarct, most surgeons agree that surgery is only indicated for those patients experiencing frequent and disabling symptoms.7 8 Extrathoracic revascularization by arterial transposition or bypass has emerged as the treatment of choice for symptomatic patients with proximal subclavian and carotid occlusive disease as a result of the high morbidity and mortality rates associated with transthoracic approaches.9 10 11 Patients with subclavian stenosis or occlusion require surgical intervention for symptoms of vertebrobasilar insufficiency due to subclavian steal or for symptoms of upper extremity ischemia such as claudication or distal embolization. Those with stenosis or occlusion of the common carotid artery most frequently require revascularization for TIAs, stroke, or syncopal episodes. While experience suggests that stroke is a relatively uncommon sequela in this patient population, particularly in those with subclavian disease, the functional disability is significant. Modern series that use the extrathoracic approach in symptomatic patients have demonstrated durable relief of symptoms, with mortality rates of 0% to 3% and perioperative stroke rates of 0% to 5%. The operative mortality rate of 1.7% and perioperative stroke rate of 3.3% observed in this report are consistent with past experience.
A consensus has not been reached regarding the appropriate management of a hemodynamically significant but asymptomatic proximal common carotid artery lesion. As a result, clinical practice in this setting has varied. This series includes four patients who underwent revascularization for asymptomatic high-grade proximal carotid artery stenoses. It should be noted, however, that each of these patients also had a contralateral carotid artery occlusion and was therefore at risk for a catastrophic neurological event upon occlusion of the patent carotid artery.
The selection of the ideal conduit for C→SC reconstruction has been controversial, and accumulating evidence suggests that the choice of conduit does have a significant effect on long-term patency. In this series arterial transposition demonstrated superior results, with 100% 5-year actuarial patency. Only one occlusion of an arterial transposition was observed during the entire follow-up interval, at more than 13 years after surgery. Our experience with transposition parallels that of other modern series, which have reported early (30-day) patency rates of 100% and long-term patency of 95% to 100%, with somewhat lower results for vein grafts or prosthetic bypass grafts.1 12 13 Because of these excellent results, some vascular surgeons have adopted arterial transposition as the revascularization procedure of choice. Weimann et al14 reported a series of 38 patients with symptomatic subclavian steal, all of whom underwent C→SC transposition, with 97% patency during an average follow-up interval of 13 months. Edwards et al15 recently reported the largest series of C→SC transposition, in which 178 patients followed up for a mean interval of almost 4 years had a patency rate of 99%.
A number of factors may be responsible for the superior performance of arterial transposition compared with prosthetic and venous bypass grafts. By virtue of the fact that only a single anastomosis is constructed, the theoretical risk of anastomotic complications is in essence reduced by half. Additionally, since a nonarterial conduit is not interposed between the carotid and subclavian arteries, the subsequent development of intimal hyperplasia related to compliance mismatch may be averted. Transposition also provides an advantage over prosthetic materials with respect to thrombogenicity and an advantage over autologous vein with respect to size match. Initial concerns regarding the possibility of ipsilateral hemispheric ischemic episodes from either the creation of a “carotid steal” or as a result of intraoperative common carotid artery cross-clamping have proven to be unfounded. We and others have observed that a carotid steal will only occur in the presence of a proximal common carotid occlusive lesion, and modern series have failed to report a single perioperative stroke attributable to the operative clamping of the common carotid artery during a transposition procedure.
Of the conduits used for C→SC bypass, PTFE demonstrated the best results with a 5-year actuarial patency of 95.2±4.6%, followed by Dacron grafts at 83.9±10.5%. Saphenous vein grafts had the poorest long-term patency rate, at 64.8±16.5%. While there was a trend toward poorer patency for saphenous vein grafts and PTFE grafts appeared to perform better than Dacron grafts, the differences between these four groups was not statistically significant (P=.20). As the overall 5-year patency rate is relatively high for all 60 bypasses/transpositions (87.5±4.8%), a greater number of patients would be required to demonstrate a statistically significant difference between each type of conduit. Saphenous vein grafts also more frequently required subsequent intervention, such as thrombectomy or angioplasty, to maintain graft patency.
A difference in performance between prosthetic grafts and autogenous vein grafts was first noted by the senior author in 1986.1 In this review of 31 bypass procedures, we reported a 5-year patency rate by life-table methods of 91% for prosthetic grafts (n=18) compared with only 57% for saphenous vein grafts (n=13). Our experience with prosthetic and saphenous vein bypass grafts parallels that reported in the literature during the past 30 years. In the cumulative published experience from 1962 to 1994, when we examine those series in which results were analyzed according to the choice of conduit, the patency rate for 389 prosthetic grafts is 97% compared with 83% for vein grafts (length of follow-up varied between series).1 8 11 12 16 17 18 19 Several factors may explain the superior performance of prosthetic conduits in C→SC bypass procedures. These include the relatively small size of autogenous saphenous vein grafts in comparison to prosthetic grafts and the potential for graft kinking due to neck motion in several axes. The inherent stiffness of prosthetic grafts may provide some advantage for grafts in this position. Some surgeons have used prosthetic grafts exclusively to avoid the potential patency problems associated with vein grafts.20 The theoretical advantage of a decreased risk of graft infection with vein grafts does not appear to be an important factor, since only one graft infection occurred in this series of 60 bypasses (1.7%). It is notable, however, that in one recent series including 15 C→SC bypass grafts constructed with autogenous saphenous vein, the long-term patency rate during a follow-up interval of 1 to 122 months was 100%.19 All of these grafts were SC→C bypasses performed for proximal common carotid lesions, and it is possible that this graft configuration, as well as the relative low resistance of the cerebral circulation compared with that of the upper extremity, contributes to the favorable outcome observed.
Experience with these bypass materials early in the series clearly influenced the choice of conduit in later patients (see Fig 6⇑). During 1973 to 1983, Dacron and saphenous vein were most commonly used, with only six PTFE grafts and three transpositions performed. Because of the number of occlusions observed with vein grafts and the excellent results with PTFE grafts and transposition, during the second half of this series the use of saphenous vein was essentially abandoned. PTFE was used most frequently, and the number of arterial transpositions performed doubled.
It is notable that of the seven grafts that were observed to fail during the follow-up interval (a mean of 44.0 months), all failures occurred within the first 17 months. This observation suggests that technical and anatomic factors related to bypass graft construction significantly influenced later patency. Those grafts with adequate caliber, inflow, and outflow were thus likely to remain patent indefinitely, while grafts that were suboptimal with respect to any of these factors failed within a relatively short amount of time.
Revascularization provided relief of symptoms due to subclavian or carotid occlusive disease in all patients with patent bypasses. Five-year actuarial symptom-free survival rates paralleled patency rates when examined by both outflow vessel and type of conduit. There was no significant difference in symptom-free survival rates between bypasses with carotid outflow versus those with subclavian outflow. Transpositions demonstrated a 100% 5-year symptom-free rate, followed by PTFE at 90.5%, Dacron at 83.9%, and saphenous vein at 63.6%. Again, these differences fell short of statistical significance as a result of the small number of failed bypasses (P=.19).
While the anatomic location of the bypassed occlusive lesion had no effect on either graft patency or the overall symptom-free survival rate, location markedly influenced the overall survival rate. The actuarial 5-year survival of patients undergoing revascularization for proximal subclavian disease was 100%, compared with 62.7% for patients whose indication for operation was an innominate or common carotid lesion. This difference was statistically significant (P=.001). The reasons behind this observation are not clear. Since these widely divergent survival statistics occurred in the absence of any significant difference between the two groups in patency or relief of symptoms, graft durability or function cannot be implicated. One possible explanation may be that the natural history of atherosclerosis in patients with disease affecting the carotid circulation is more rapidly progressive than that of patients with occlusive lesions of the subclavian artery. Patients with carotid occlusive disease may be more likely to have associated coronary artery disease. The one perioperative death and two of the six late deaths in this group were attributable to myocardial ischemia.
In summary, our experience with C→SC bypass suggests that arterial transposition is the conduit of choice for revascularization of symptomatic brachiocephalic occlusive disease. Prosthetic grafts demonstrated slightly lower patency results, and the performance of vein grafts was relatively poor. Although these differences fell short of statistical significance because of the high overall patency rate, they concur with trends observed by other authors. Long-term patency was not influenced by whether the occlusive disease is in the subclavian versus common carotid/innominate location; however, patients with common carotid/innominate disease had significantly lower long-term survival. This finding may be due in part to more advanced systemic atherosclerotic disease. C→SC bypass can provide durable relief of disabling symptoms with minimal mortality and acceptable morbidity.
Selected Abbreviations and Acronyms
|TIA||=||transient ischemic attack|
This study was supported in part by a grant from the Joash Foundation.
Presented in part at the 19th International Joint Conference on Stroke and Cerebral Circulation, San Diego, Calif, February 17-19, 1994.
- Received January 27, 1995.
- Revision received May 8, 1995.
- Accepted June 8, 1995.
- Copyright © 1995 by American Heart Association
Ackermann H, Diener HC, Seboldt H, Huth C. Ultrasonographic follow-up of subclavian stenosis and occlusion: natural history and surgical treatment. Stroke. 1988;19:431-435.
Hennerici M, Klemm C, Rautenberg W. The subclavian steal phenomenon: a common vascular disorder with rare neurologic deficits. Neurology. 1988;38:669-673.
Criado FJ. Extrathoracic management of aortic arch syndrome. Br J Surg. 1982;69(suppl):S45-S51.