Symptomatic patients had a lower long-term stroke-free survival.
Symptomatic patients had a lower long-term overall survival.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Cooper M, Arhuidese IJ, Obeid T, Hicks CW, Canner J, Malas MB. Perioperative and Long-term Outcomes After Carotid Endarterectomy in Hemodialysis Patients. JAMA Surg. 2016;151(10):947–952. doi:10.1001/jamasurg.2016.1504
Is carotid endarterectomy (CEA) safe and effective in dialysis-dependent patients?
In this review of 5142 dialysis-dependent patients who underwent CEA, complications were high. The 30-day stroke rate, myocardial infarction, and mortality for the asymptomatic and symptomatic groups were 2.7% vs 5.2%, 4.6% vs 5.0%, and 2.6% vs 2.9%, respectively; 5-year stroke-free and overall survival were low in both the symptomatic and asymptomatic cohorts.
The risks outweigh the benefits of CEA in asymptomatic patients and CEA should only be offered for a carefully selected cohort of symptomatic patients.
Early landmark trials excluding dialysis patients showed carotid endarterectomy (CEA) decreased stroke risk compared with medical management. Dialysis dependence has been associated with poor outcomes after CEA in small studies, but, to our knowledge, there are no large studies evaluating outcomes of CEA in this patient group.
To delineate perioperative and long-term outcomes after CEA in dialysis-dependent patients in a large national database.
Design, Setting, and Participants
A retrospective review of all patients who underwent CEA in the US Renal Disease System–Medicare-matched database between January 1, 2006, and December 31, 2011, was performed in June 2015. The median follow-up time was 2.5 years. Logistic and Cox regression analyses were used to evaluate perioperative and long-term outcomes.
Main Outcomes and Measures
The primary outcomes of interest were perioperative stroke, myocardial infarction and mortality, and long-term stroke and mortality.
A total of 5142 patients were studied; 83% of whom were asymptomatic. The mean (SD) age was 68.9 (9.6) years for asymptomatic patients and 70.0 (9.1) years for symptomatic patients. The 30-day stroke rate, myocardial infarction, and mortality for the asymptomatic and symptomatic groups were 2.7% vs 5.2% (P = .001), 4.6% vs 5.0% (P = .69), and 2.6% vs 2.9% (P = .61), respectively. Predictors of perioperative stroke were symptomatic status (odds ratio [OR], 2.01; 95% CI, 1.18-3.42; P = .01), black race (OR, 2.30; 95% CI, 1.24-4.25; P = .008), and Hispanic ethnicity (OR, 2.28; 95% CI, 1.17-4.42; P = .02). Freedom from stroke and overall survival were lower in symptomatic compared with asymptomatic patients at 1, 2, 3, 4, and 5 years (in asymptomatic vs symptomatic patients, freedom from stroke rates were 92% vs 87% at 1 year, 88% vs 83% at 2 years, 84% vs 78% at 3 years, 80% vs 73% at 4 years, and 79% vs 69% at 5 years, respectively, and overall survival rates were 78% vs 73% at 1 year, 60% vs 57% at 2 years, 46% vs 42% at 3 years, 37% vs 32% at 4 years, and 33% vs 29% at 5 years; P < .05). Predictors of long-term stroke were preoperative symptoms (hazard ratio, 1.67; 95% CI, 1.24-2.24; P < .001), female sex (hazard ratio, 1.34; 95% CI, 1.03-1.73; P = .04), and inability to ambulate (hazard ratio, 1.81; 95% CI, 1.25-2.62; P = .002). Predictors of long-term mortality were increasing age (OR, 1.02; 95% CI, 1.01-1.03; P < .01), active smoking (OR, 1.22; 95% CI, 1.00-1.48; P = .045), history of congestive heart failure (OR, 1.25; 95% CI, 1.12-1.39; P < .001), and chronic obstructive pulmonary disease (OR, 1.26; 95% CI, 1.09-1.45; P = .002).
Conclusions and Relevance
To our knowledge, this is the largest study to date of dialysis patients who have undergone CEA. We have shown that the risks of CEA in asymptomatic patients is high and may outweigh the benefits. The risk of CEA in symptomatic patients is also high, and it should only be offered to a small carefully selected cohort of symptomatic patients.
Stroke is the fifth leading cause of death in the United States and is a leading cause of major long-term disability in adults.1,2 Extracranial carotid artery stenosis is the cause of ischemic stroke in 10% to 20% of patients.3 Early landmark trials showed that in select patient populations, carotid endarterectomy (CEA) decreased stroke risk compared with medical management alone.4-6 However, dialysis patients were excluded from these large randomized clinical trials as they were considered high risk. Furthermore, since those trials, medical management has improved with the use of statins and antiplatelet agents.7 With the improvement in medical management, the role of CEA is being reevaluated even in low-risk patients.8
Chronic renal insufficiency has in multiple studies been associated with poor outcomes after CEA including higher incidence of cardiac events and pulmonary complications as well as higher perioperative and long-term mortality. In these studies, the risks were highest in patients with more advanced renal disease.7,9-13 Furthermore, per guidelines from the Society for Vascular Surgery,14 the perioperative risk of stroke and death in an asymptomatic patient should be less than 3% to ensure benefit for the patient. Additionally, asymptomatic patients at high risk for intervention or with less than 3 years’ life expectancy should be considered for medical management as the first-line therapy. Previous studies have shown complication rates higher than these metrics in dialysis-dependent patients, bringing into question whether CEA should be performed in the dialysis population.7,9-13
However, the ability to generalize the results from previous studies to all hemodialysis (HD) patients is limited by the heterogeneity of patients in the studies, the small number of HD patients in each study, and by many of the studies being single surgeon or single institution. The objective of this study is to delineate patterns of presentation and perioperative and long-term outcomes after CEA in an exclusive cohort of dialysis-dependent patients drawn from a national database.
Retrospective analysis of the US Renal Data System was performed in June 2015 for all patients with end-stage renal disease who underwent CEA between January 1, 2006, and December 31, 2011. Created and maintained by the National Institute of Diabetes and Digestive and Kidney Diseases, the US Renal Data System is a prospective database that includes every HD patient in the United States. It integrates patient-specific data from the Centers for Medicare and Medicaid Services, Centers for Disease Control and Prevention, United Network for Organ Sharing, and end-stage renal disease networks.15 The use of the US Renal Data System data for this study was approved by the Johns Hopkins institutional review board; consent was not required per institutional policy, as this was a retrospective quality-control study.
Patients were identified using the procedural terminology code for CEA (35301). Demographic and medical history data were collected from the Centers for Medicare and Medicaid Services form 2728, “End-Stage Renal Disease Medical Evidence Report,” and Medicare claim files. Demographic data collected included age, sex, and race/ethnicity. Medical history collected included history of transient ischemic attack or stroke, hypertension, diabetes, coronary artery disease, peripheral vascular disease, congestive heart failure, chronic obstructive pulmonary disease, smoking history, malignancy, and ability to ambulate. Symptomatic status was defined as transient ischemic attack or stroke occurring within 6 months prior to CEA. The primary outcomes of interest were perioperative stroke, myocardial infarction (MI) and mortality, and long-term stroke and mortality. Mortality data were collected from the Centers for Medicare and Medicaid Services form 2746, “ESRD Death Notification Form.” Follow-up data were obtained from the Medicare institutional claims database. Stroke (codes 433.11, 434.11, 434.91, and 997.02) and MI (410 and 997.1) were assessed using International Classification of Diseases, Ninth Revision diagnosis codes. To maximize alignment of ipsilateral symptoms, treatment, and outcomes, we restricted our analyses to patients with only 1 CEA in the study period. Patients with previous carotid interventions (carotid artery stenting or CEA) were excluded.
Descriptive analyses of the study groups were performed using χ2, analysis of variance, and t tests. Univariable and multivariable logistic regression analyses were used to evaluate and identify the predictors of the perioperative outcomes (stroke, MI, and death). Long-term outcomes (stroke and death) were analyzed using Kaplan-Meier and univariable and multivariable Cox regression methods, and their predictors were identified. Applying techniques of survival analyses, the incidence of stroke in the long-term was computed at every time point based on patients with stroke as the numerator and patients at risk for stroke as the denominator. Subsequently, freedom from stroke was computed from the difference between 100% and the cumulative incidence of stroke at the various time points. In a similar manner, the numerator for the incidence of mortality in the long-term was patients who died at a given time, while the denominator was patients at risk for death. Patient survival was then computed as the difference between 100% and the cumulative incidence of mortality at the respective time points. Patients who received a kidney transplant were censored on the transplant date. Statistical models were built based on predictive variables from univariable analyses, prior literature, guidance of likelihood ratio tests, and Akaike information indices. All analyses were performed using Stata version 12.1 statistical software (StataCorp), and statistical significance was accepted at P < .05.
A total of 5142 HD patients underwent CEA between January 1, 2006, and December 31, 2011. Patient characteristics for the symptomatic and asymptomatic groups can be seen in Table 1. Patients in the symptomatic group were older (70.0 years vs 68.9 years; P = .01), had a lower body mass index (calculated as weight in kilograms divided by height in meters squared; 27.6 vs 28.5; P = .005), and were more often active smokers (10.8% vs 7.6%; P = .008). Patients in the symptomatic group also had a higher incidence of coronary artery disease (49.1% vs 33.6%; P < .001), peripheral artery disease (45.6% vs 25.3%; P < .001), congestive heart failure (48.0% vs 38.8%; P < .001), chronic obstructive pulmonary disease (19.4% vs 13.2%; P < .001), and inability to ambulate (8.1% vs 4.1%; P < .001) (Table 1). The median follow-up time was 2.5 years (interquartile range, 1.28-3.75 years; maximum, 5.0 years).
The 30-day stroke rate was 2.7% in asymptomatic and 5.2% in symptomatic patients (P = .001). The 30-day MI and mortality rates were not different between the asymptomatic and symptomatic cohorts (MI: 4.6% vs 5.0%, P = .69; mortality: 2.6% vs 2.9%; P = .61). Based on a multivariable logistic regression model, predictors of perioperative stroke were preoperative symptoms (odds ratio [OR], 2.01; 95% CI, 1.18-3.42; P = .01) and black race (OR, 2.30; 95% CI, 1.24-4.25; P = .008) or Hispanic ethnicity (OR, 2.28; 95% CI, 1.17-4.42; P = .02) compared with white race (Table 2).
Freedom from stroke in asymptomatic patients was 92% at 1 year, 88% at 2 years, 84% at 3 years, 80% at 4 years, and 79% at 5 years. In symptomatic patients, freedom from stroke was lower at 87% at 1 year, 83% at 2 years, 78% at 3 years, 73% at 4 years, and 69% at 5 years (P < .001) (Figure 1). In a multivariable Cox regression analysis, predictors of long-term stroke were preoperative symptoms (hazard ratio, 1.67; 95% CI, 1.24-2.24; P < .001), female sex (hazard ratio, 1.34; 95% CI, 1.03-1.73; P = .04), and inability to ambulate (hazard ratio, 1.81; 95% CI, 1.25-2.62; P = .002). Overall survival in asymptomatic patients was 78% at 1 year, 60% at 2 years, 46% at 3 years, 37% at 4 years, and 33% at 5 years. In symptomatic patients, overall survival was lower at 73% at 1 year, 57% at 2 years, 42% at 3 years, 32% at 4 years, and 29% at 5 years (P = .02) (Figure 2). In multivariable Cox regression analysis, predictors of long-term mortality were increasing age (OR, 1.02; 95% CI, 1.01-1.03; P < .01), active smoking (OR, 1.22; 95% CI, 1.00-1.48; P = .045), history of congestive heart failure (OR, 1.25; 95% CI, 1.12-1.39; P < .001), and chronic obstructive pulmonary disease (OR, 1.26; 95% CI, 1.09-1.45; P = .002).
Multiple previous studies have shown that chronic renal insufficiency is associated with poor outcomes after CEA, with the highest-risk patients being those with the most advanced renal disease.7,9-13 Per the Society for Vascular Surgery guidelines,14 for asymptomatic patients, the perioperative risk for stroke and death should be less than 3% for patients to undergo a CEA. In previous large studies of the general population, including a large study of the Medicare population between 2004 and 2006 that evaluated outcomes of CEA for 9365 patients,16 the risk for perioperative stroke and death were far below this metric. In the asymptomatic cohort in the Medicare study,16 perioperative stroke and mortality were 1.2% and 0.5%, respectively. In the Medicare symptomatic cohort, perioperative stroke and mortality rates were 3.5% and 1.2%, respectively.16 Additionally, in CREST (Carotid Revascularization Endarterectomy versus Stenting Trial), a randomized clinical trial evaluating the outcomes of carotid stenting vs CEA in 2272 patients, the perioperative stroke rates were 1.1% and 2.5% in the asymptomatic and symptomatic CEA cohorts, respectively.17 In our study, the rate of perioperative stroke and mortality were 2.7% and 2.6%, respectively, in the asymptomatic group and 5.2% and 2.9%, respectively, in the symptomatic group. Our findings support those of previous smaller studies that perioperative risk for CEA in HD patients is high, even in the asymptomatic group.
In addition to increased perioperative risk in HD patients, long-term outcomes are also poor. Per the Society for Vascular Surgery guidelines,14 asymptomatic patients with less than 3 years’ life expectancy should be considered for medical management as the first-line therapy. In the asymptomatic cohort in our study, freedom from stroke at 1 year was 92%, the 3-year stroke-free survival was 84%, and the 3-year overall survival was 46%. In the symptomatic cohort in our study, freedom from stroke at 1 year was 87%, the 3-year stroke-free survival was 78%, and the 3-year overall survival was 42%. In the Medicare study,16 freedom from stroke at 1 year was 96.8% in asymptomatic patients and 89.7% in symptomatic patients, which is significantly higher than in either cohort in our study.
Both the perioperative and long-term outcomes in our study confirm the finding of several smaller studies that patients with severe renal insufficiency are at higher risk for complications.7,9,13,18,19 Although perioperative outcomes for asymptomatic patients are acceptable in our study, the long-term outcomes were very poor, with less than 50% of patients surviving to 3 years in both the asymptomatic and symptomatic cohorts. Based on these findings, it is clear that asymptomatic HD patients do not derive long-term benefits from CEA over best medical therapy.13,18-20 Given that most of the patients (83%) in our study who underwent CEA were asymptomatic, most CEAs currently performed in HD patients could be avoided with optimal medical management.13,16,19 Furthermore, previous large population-based studies have identified that strict compliance with dialysis treatment, optimizing nutritional status, and sustainable access to a nephrologist improve survival in HD patients.21-24 Using a multidisciplinary approach, we can fully optimize care for HD patients with carotid stenosis.
In symptomatic HD patients, although CEA outcomes in our study were poorer than in the asymptomatic cohort, the stroke risk with medical management alone was higher as well. In a systematic review,25 at 2 weeks after onset of carotid-related symptoms in the general population, stroke risk was 20.6% to 32.5%. This is higher than the perioperative stroke rate and is similar to the 1-year stroke rate in symptomatic patients in our study, indicating that there is a subpopulation of symptomatic patients that would benefit from CEA. Additionally, HD patients who have strokes do poorly. In the CHOICE (Choices for Healthy Outcomes in Caring for End-Stage Renal Disease) Study, a prospective study of 1041 dialysis patients, of patients who had strokes, 35% were fatal and only 56% of patients were able to be discharged to home or acute rehabilitation.26 Using additional imaging, including magnetic resonance imaging, computed tomography, positron-emission tomography, and ultrasonography for evaluation of plaque characteristics, we could have a more selective approach to determine those patients where the benefit of CEA would outweigh the risk. In a review of clinical and imaging features associated with increased risk for early and late stroke in patients with symptomatic carotid artery disease, predictors of stroke included male sex, age older than 75 years, hemispheric symptoms, irregular stenosis, contralateral occlusion, tandem intracranial disease, and the presence of acute cerebral ischemia on computed tomography/magnetic resonance imaging, and fluorodeoxyglucose uptake in the carotid plaque on positron-emission tomography.27 Another study evaluating the correlation between plaque composition and ipsilateral stroke risk found that intraplaque thrombosis and increased fibrous content of the plaque were associated with increased stroke risk.28
This study had the inherent limitations of any retrospective analysis. All patients in the study had undergone CEA and there was no control group that underwent medical management. Additionally, owing to database limitations, we were not able to include information known to affect outcomes after CEA such as degree of stenosis, contralateral disease, medication use, or the duration of dialysis treatment prior to CEA. However, to our knowledge, this is the largest study to date evaluating HD patients who have undergone CEA and it allowed good comparison between symptomatic and asymptomatic patient cohorts.
To our knowledge, this is the largest study to date of dialysis patients who have undergone CEA, and we have shown that outcomes are relatively poor and that survival is limited in the long-term, consistent with previous smaller studies. In asymptomatic patients, we recommend optimizing medical management and avoiding CEA. In symptomatic patients, we recommend selective CEA for patients at very high risk for a second transient ischemic attack or stroke.
Accepted for Publication: May 4, 2016.
Corresponding Author: Mahmoud B. Malas, MD, MHS, The Center for Research Excellence and Surgical Trials, School of Medicine, Johns Hopkins University, 4940 Eastern Ave, Baltimore, MD 21224 (email@example.com).
Published Online: June 29, 2016. doi:10.1001/jamasurg.2016.1504
Author Contributions: Dr Cooper had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Cooper, Arhuidese, Obeid, Malas.
Acquisition, analysis, or interpretation of data: Arhuidese, Obeid, Hicks, Canner, Malas.
Drafting of the manuscript: Cooper, Arhuidese.
Critical revision of the manuscript for important intellectual content: Cooper, Obeid, Hicks, Canner, Malas.
Statistical analysis: Cooper, Arhuidese.
Administrative, technical, or material support: Hicks, Canner, Malas.
Study supervision: Malas.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This study was presented at the American College of Surgeons Clinical Congress; October 8, 2015; Chicago, Illinois.