Extracorporeal life support (ECLS) is a rescue therapy for reversible cardiac or respiratory failure when standard treatments are inadequate. Although ECLS is life saving, there are associated risks, including central nervous system (CNS) injury.1 There are multiple causes of CNS injury, including ischemic injury from circuit emboli and anticoagulation-related CNS hemorrhage. However, the location of the arterial cannula is a potentially modifiable risk factor for CNS injury in venoarterial (VA) ECLS, with both the carotid and femoral arteries available as potential sites. There is concern for alteration in cerebral blood flow with carotid cannulation during the ECLS run, which could increase the risk of ipsilateral ischemic stroke. Additionally, circuit-related thromboembolism of contralateral cerebral arteries may further increase the risk of ischemic injury in patients cannulated via the carotid artery.
There is some evidence in children that carotid cannulation is associated with increased risk of CNS injury when compared with alternative arterial cannulation strategies. Yet this difference may not be uniformly observed across all pediatric age groups.2,3 However, no studies have directly examined this relationship in adult patients who receive VA ECLS, nor has there been a comparison between adults and children. Cannulation of the carotid may have a lower risk of stroke in children compared with adults because of increased prevalence of carotid and cerebral atherosclerosis, impaired collateral cerebral circulation, and progressive vascular endothelial dysfunction-related cerebral hypoperfusion associated with aging.4–9 As a reference point, the prevalence of stroke in unilateral occlusion of the carotid artery in adults with intracranial aneurysm or head/neck cancer historically is 17–30%.10 It is unclear whether this is a similar situation for adult ECLS patients with carotid cannulation.
Given the concern for potential increased risk of stroke, cannulation of the femoral artery rather than carotid has been informally recommended for adults and older children requiring VA ECLS, yet without consensus in the ECLS community, particularly with regard to children and adolescents.3,11 However, there are also risks associated with femoral cannulation, including limb-threatening ischemia while on ECLS and even subsequent development of late vascular complications after the ECLS run.11–15 There is also potential risk of global cerebral hypoxia with femoral cannulation as the oxygenated blood delivered by the ECLS circuit must flow retrograde up the aorta to the brain. Thus, in a patient with pulmonary venous desaturation and good heart function, global cerebral oxygen delivery could be low. This theoretically could potentially lead to worse neurologic outcomes than carotid cannulation.16,17
The goal of this study was to evaluate the association between site of arterial cannulation and stroke in VA ECLS and to understand whether this relationship is modified by age from childhood to adulthood.
Study Design and Data Source
This is a retrospective cohort study. We obtained data from the Extracorporeal Life Support Organization (ELSO) Registry. There are more than 200 ECLS centers contributing data to the ELSO Registry with a purpose of research and quality improvement.18 Member ECLS centers submit data voluntarily with approval from their local institutional review boards. The Seattle Children’s Hospital Institutional Review Board approved this study (#14564).
We included all patients between 6 and 40 years of age cannulated via the carotid or femoral artery for VA ECLS during 2000–2012. We chose 6–40 years of age such that in each age group there would be at least 5% of the patients with the alternate cannulation site and to exclude younger children as femoral artery cannulation in this age group is technically challenging and frequently not feasible secondary to the diminutive size of the artery.
We excluded patients with more than one ECLS run, patients with multiple arterial cannulation sites (except those with bilateral femoral artery cannulation), patients with mechanical circulatory support before ECLS, and those whose indication was extracorporeal cardiopulmonary resuscitation (ECPR).
Arterial cannula location was our exposure of interest. Our primary outcome was CNS infarct. We also evaluated CNS hemorrhage as an outcome in case of hemorrhagic conversion of an ischemic stroke.
We included demographic data, pre-ECLS support, characteristics of the ECLS run, and ECLS-associated complications. We categorized age as 6–10, 11–20, and 21–40. Treatment era was categorized as 2000–2003, 2004–2007, and 2008–2012. We included dobutamine, dopamine, epinephrine, inamrinone, milrinone, norepinephrine, or other vasoactive/inotropic drugs as “pre-ECLS inotrope or vasopressor use.” “Pre-ECLS vasodilator use” includes nitroprusside or other vasodilator drugs. Sodium bicarbonate and tromethamine (THAM) were combined. Cardiac tamponade by air, serous fluid, or blood was all considered as “tamponade.” “Circuit thrombosis” was defined as clots anywhere in the ECLS circuit (bladder, bridge, hemofilter, oxygenator, or other sites). “Circuit rupture” includes raceway or tubing rupture. “Dialysis on ECLS” is defined as hemofiltration, dialysis, or continuous arteriovenous hemodialysis.
For our descriptive statistics, we used median and interquartile range (IQR) for continuous variables and frequencies and proportions for categorical variables. For bivariate comparisons by cannulation site, we used the Wilcoxon rank sum test for continuous variables and the χ2 test for categorical variables.
We estimated the relationship between arterial cannula location and our outcomes of interest using multivariable logistic regression with two models: 1) infarct and 2) hemorrhage. For covariate selection to adjust for in these models, we used backward stepwise logistic regression with p value for removal from the full model at greater than 0.15 and p value for reentry into the model at less than 0.1 (variables evaluated for inclusion [see Table, Supplemental Digital Content, http://links.lww.com/ASAIO/A360]). Age category was included as a dummy variable in addition to an interaction term between the age categories and arterial cannula location to evaluate for effect modification by age. The significance of the interaction term was tested with the likelihood ratio test. Seizure was not evaluated as a potential covariate as this may lie in the causal pathway. Using this method, the covariates that we included in the CNS infarct model are age category; year of treatment; pre-ECLS arrest; the following pre-ECLS therapies: continuous renal replacement therapy, high-frequency ventilation, and vasodilators; and the following ECLS complications: need for inotropes on ECLS, tamponade, bleeding, cannula problems, oxygenator failure, and pH < 7.2. The CNS hemorrhage model was adjusted for age category; race; year of treatment; indication for ECLS (cardiac or pulmonary); the following pre-ECLS therapies: nitric oxide and vasodilators; and the following ECLS complications: hypertension requiring treatment, disseminated intravascular coagulation, hemolysis, bleeding, clots or air in the circuit, and dialysis on ECLS. We used Stata software, version 14 (College Station, TX).
We identified 1,518 unique patients meeting our study criteria. Prevalence of femoral cannulation was the highest in the oldest age group with a shift to carotid with decreasing age. Notably though, for all ages, over time there has been increasing use of the femoral artery as opposed to the carotid (Figure 1). Those with carotid cannulation had a greater prevalence of having white race and pulmonary etiology rather than cardiac. Additionally, their median duration of ECLS support was longer (Table 1). The femoral group had lower ventilator settings and less frequent use of adjunctive respiratory support with better blood gas results. Femoral patients received narcotics and neuromuscular blocking agents less frequently than those with carotid cannulation. Femoral patients were also more likely to undergo cardiopulmonary bypass and receive continuous renal replacement therapy before ECLS. Notably, prevalence of pre-ECLS cardiac arrest was not statistically different between the two groups (p = 0.478) (Table 2). The carotid group endured hypertension, circuit rupture, seizures, and pneumothorax more commonly. Additionally, carotid patients required dialysis on ECLS more often. Finally, those with femoral cannulation experienced oxygenator failure and elevated pH more frequently (Table 3).
CNS infarct and CNS hemorrhage were generally mutually exclusive diagnoses, with 4.4% of the cohort having CNS infarct, 4.7% with CNS hemorrhage, and only 0.5% with both (Figure 2). Although the proportion of patients with infarct or hemorrhage was not statistically significantly different by era of treatment, the absolute number of patients with infarct or hemorrhage was the greatest in the most recent era (Table 4). Overall, 47 of 640 carotid patients (7.3%) experienced a CNS infarct as compared with 27 of 878 (3.1%) femoral patients. In our regression analysis, we found a fourfold increased odds of CNS infarct with carotid cannula location (adjusted odds ratio [OR]: 4.03; 95% CI: 2.17–7.46; p < 0.001) as compared with femoral with adjustment for confounding variables. Forty-two of 640 carotid patients (6.6%) and 36 of 878 femoral patients (4.1%) had CNS hemorrhage (adjusted OR: 1.12; 95% CI: 0.64–1.98; p = 0.692). We did not find a statistically significant difference across age groups in the association between cannula location and CNS infarct and hemorrhage (interaction p = 0.610 and 0.786, respectively) (Table 5 and Figure 3, A and B).
We additionally performed a preliminary subgroup analysis to gain insight into the association of carotid repair at decannulation with CNS injury. To do this, we evaluated those patients with carotid cannulation who survived the ECLS run (n = 376). Fifty-one of these patients were missing data on carotid repair (13.6%). Of the remaining patients in this subgroup, 128 of 325 (39.4%) had carotid repair. There was a significant association of carotid repair with lower prevalence of infarct (infarct in 2.3% with repair and 8.6% without repair; p = 0.021 by χ2). There was a trend of higher prevalence of carotid repair with younger age (42.5%, age 6–10; 39.5%, age 11–20; 18.2%, age 21–40; p = 0.094 by χ2).
In this study, we observed that femoral artery cannulation was utilized predominantly in older patients with a transition to using the carotid more commonly with decreasing age. We found that cannulation of the carotid artery was associated with a fourfold increased odds of CNS infarction compared with femoral. We had hypothesized that this association would differ by age, with increased risk of stroke in adults with carotid cannulation compared with children. Yet, in our analysis, there was not a statistically significant difference. This may be related to the relatively young age of the adults in this study. With cerebral atherosclerosis prevalence and severity increasing with age, there could be a greater difference at older ages.4,19 Alternatively, there may have been bias with only those adults with low cardiovascular risk profiles being cannulated via the carotid artery, masking a difference between adult and pediatric patients. Additionally, we were unable to account for differences in carotid cannulation strategy (complete interruption versus preservation of some degree of antegrade flow) nor the role of carotid repair at decannulation and prevalence and timing of infarct in our primary regression models. These issues warrant further investigation. There may be different practices between pediatric and adult providers, which could contribute to these results.
This is the first study to evaluate risk of CNS injury by location of the arterial cannula in adults on VA ECLS and to compare this with children and adolescents. There have been two prior studies evaluating the association of CNS injury and location of arterial cannulation in children. Using the ELSO Registry, Teele et al.2 analyzed 2,977 children on VA ECLS for any indication (respiratory, cardiac, or ECPR) from 2007 to 2008. They observed that cannulation of the carotid artery was associated with a 30% increased odds of seizure, CNS infarction, or CNS hemorrhage compared with cannulation of the femoral artery or aorta (OR: 1.30; 95% CI: 1.01–1.69). This association was not statistically significantly modified by age. However, only 19% of the patients were more than 1 year old. Thus, the bulk of the study population was of an age that would not typically be considered for femoral cannulation given the diminutive size of the femoral artery at that age. Also, the femoral artery and aorta were combined as the alternative to carotid for the regression analysis. Most patients were cannulated via the carotid (64%), with aortic (32%) and femoral (4%) combined as the comparative group in the multivariate model, making this more of a carotid versus aortic comparison than carotid versus femoral. Also using the ELSO Registry, Rollins et al.3 evaluated 1,632 children (1 month to 17 years old) on VA ECLS for a respiratory indication from 1993 to 2007. They observed that CNS hemorrhage was experienced more frequently in those with carotid artery cannulation compared with femoral cannulation (7% vs. 2%). Central nervous system infarction was not statistically significantly different overall between these two groups (6% vs. 5%). After stratifying by age, they did, however, find that older children (more than 10 years old) with carotid cannulation experienced CNS infarcts more often than those cannulated via the femoral artery. In their multivariable analysis, all sites of arterial cannulation were combined and compared with venovenous cannulation, without an adjusted estimate of carotid versus femoral. Our study is complementary to these evaluations with a larger number of patients, timeframe inclusive of more recent treatment years, and gives a frame of reference with a young adult population.
Extracorporeal life support providers have long been interested in understanding the risk of CNS injury with cannulation of the carotid artery. Most of what we have learned in this regard is in the neonatal population. Although there have been numerous studies in neonates on VA ECLS with carotid cannulation, the results are mixed; some researchers observed increased ipsilateral brain injury,20–24 whereas others did not discover lateralized injury.25–31 Interestingly, multiple research groups have shown relative preservation of cerebral blood flow to the ipsilateral hemisphere as arterial cannulation in neonatal/infant patients cannulated via the right carotid artery.32–38 Research on cerebral blood flow in older age groups has been sparse, leaving this an area open for further research.16,39
Based on our analyses, we would suggest consideration of femoral artery cannulation in older children, adolescents, and adults, understanding that risk of CNS injury is only one of many factors that must be considered when determining site of arterial cannulation. Selection of cannulation site must also weigh age-related and non–age-related anatomic features, feasibility, comorbidities, and risk of central and peripheral vascular insufficiency, in addition to clinical urgency and technical proficiency of the provider performing the procedure. In the case of femoral artery cannulation, providers must assure adequate distal perfusion of the involved leg (e.g., distal perfusion catheter or femoral artery graft) given the risk of limb-threatening ischemia.11–15 Also, attention must be paid to adequate oxygen delivery to the upper body given that oxygenated blood returning from the ECLS circuit flows retrograde up the aorta potentially leading to the brain being perfused with deoxygenated blood if heart function is good and lung function is poor potentially causing global CNS injury.16,17
The current study has several limitations. An important limitation of this study is that a uniform approach to identifying CNS injury was not utilized. Thus, there could be underascertainment of the true prevalence of CNS infarct and hemorrhage. We cannot be certain this is nondifferential based on location of cannulation, and potential practice variation related to evaluating CNS injury in pediatric versus adult patients cannot be ascertained. Additionally, given the voluntary nature of the database, there could be underreporting of some variables. Importantly, we were unable to account for anticoagulation or use of antifibrinolytics as this information was not available, which could be important as it relates to risk of CNS hemorrhage and thromboembolic events. As a surrogate for anticoagulation, we did include “bleeding complications” and “clots in the circuit” in our analyses. Although we did not identify a significant difference in prevalence in bivariate analysis by cannulation site, “bleeding complications” was retained in the models for both infarct and hemorrhage and “clots in the circuit” was retained in the hemorrhage model, highlighting this issue. During the study period, the registry did not collect information about use of distal perfusion catheters or other limb sparing adjuncts in patients with femoral artery cannulation. Additionally, the incidence of ischemic limb injury is unknown in these patients. The observed age-related bimodal distribution of cannulation location limits the ability to determine the impact of cannulation site on CNS injury in extremely young and old patients. The ECLS center where patients were treated was not able to be incorporated in our statistical analyses as these data were not available from the registry, nor was another surrogate such as center ECLS volume. Accounting for ECLS center could be important as most centers have site-specific protocols that differ across centers, such as anticoagulation management and cannulation site preferences. Also, all ECLS centers do not share similar outcomes. Finally, the location or laterality of the CNS injury was not available in the dataset.
In this study, we found an increased likelihood of stroke with cannulation of the carotid artery in patients 6–40 years old in comparison with femoral cannulation during VA ECLS. This relationship was not statistically significantly different between the included age groups. However, risk of stroke is only one of many factors that must be considered when determining site of arterial cannulation.
The initial version of this study was submitted as a Master’s Thesis in Epidemiology by Dr. Di Gennaro at the University of Washington. It has since undergone multiple revisions. It is electronically archived in the “ResearchWorks Archive” at the University of Washington (https://digital.lib.washington.edu/researchworks/handle/1773/27162).
1. Dalton HJ, Menon S. Fuhrman BP, Zimmerman JJ. Extracorporeal life support
, in Pediatric Critical Care, 2011, pp. 4th ed. Philadelphia, PA, Elsevier, 717–737.
2. Teele SA, Salvin JW, Barrett CS, et al. The association of carotid artery
cannulation and neurologic injury in pediatric patients supported with venoarterial extracorporeal membrane oxygenation
*. Pediatr Crit Care Med 2014.15: 355–361.
3. Rollins MD, Hubbard A, Zabrocki L, Barnhart DC, Bratton SL. Extracorporeal membrane oxygenation
cannulation trends for pediatric respiratory failure and central nervous system injury. J Pediatr Surg 2012.47: 68–75.
4. Flora GC, Baker AB, Loewenson RB, Klassen AC. A comparative study of cerebral atherosclerosis in males and females. Circulation 1968.38: 859–869.
5. Berenson GS, Srinivasan SR, Bao W, Newman WP 3rd, Tracy RE, Wattigney WA. Association between multiple cardiovascular risk factors and atherosclerosis in children and young adults. The Bogalusa Heart Study. N Engl J Med 1998.338: 1650–1656.
6. Wang JC, Bennett M. Aging and atherosclerosis: Mechanisms, functional consequences, and potential therapeutics for cellular senescence. Circ Res 2012.111: 245–259.
7. Epstein SE, Lassance-Soares RM, Faber JE, Burnett MS. Effects of aging on the collateral circulation, and therapeutic implications. Circulation 2012.125: 3211–3219.
8. de Weerd M, Greving JP, de Jong AW, Buskens E, Bots ML. Prevalence of asymptomatic carotid artery
stenosis according to age and sex: Systematic review and metaregression analysis. Stroke
9. Toda N. Age-related changes in endothelial function and blood flow regulation. Pharmacol Ther 2012.133:159–176.
10. American Society of Interventional and Therapeutic Neuroradiology: Carotid artery
balloon test occlusion. AJNR Am J Neuroradiol 2001.22(8 suppl): S8–S9.
11. Haley MJ, Fisher JC, Ruiz-Elizalde AR, Stolar CJ, Morrissey NJ, Middlesworth W. Percutaneous distal perfusion of the lower extremity after femoral cannulation for venoarterial extracorporeal membrane oxygenation
in a small child. J Pediatr Surg 2009.44: 437–440.
12. Gander JW, Fisher JC, Reichstein AR, et al. Limb ischemia after common femoral artery cannulation for venoarterial extracorporeal membrane oxygenation
: An unresolved problem. J Pediatr Surg 2010.45: 2136–2140.
13. Zimpfer D, Heinisch B, Czerny M, et al. Late vascular complications after extracorporeal membrane oxygenation
support. Ann Thorac Surg 2006.81: 892–895.
14. Foley PJ, Morris RJ, Woo EY, et al. Limb ischemia during femoral cannulation for cardiopulmonary support. J Vasc Surg 2010.52: 850–853.
15. Bisdas T, Beutel G, Warnecke G, et al. Vascular complications in patients undergoing femoral cannulation for extracorporeal membrane oxygenation
support. Ann Thorac Surg 2011.92: 626–631.
16. Wong JK, Smith TN, Pitcher HT, Hirose H, Cavarocchi NC. Cerebral and lower limb near-infrared spectroscopy in adults on extracorporeal membrane oxygenation
. Artif Organs 2012.36: 659–667.
17. Cove ME. Disrupting differential hypoxia in peripheral veno-arterial extracorporeal membrane oxygenation
. Crit Care 2015.19: 280.
18. Paden ML, Rycus PT, Thiagarajan RR; ELSO Registry: Update and outcomes in extracorporeal life support
. Semin Perinatol 2014.38: 65–70.
19. Gutierrez J, Elkind MS, Virmani R, et al. A pathological perspective on the natural history of cerebral atherosclerosis. Int J Stroke
20. Campbell LR, Bunyapen C, Holmes GL, Howell CG Jr, Kanto WP Jr. Right common carotid artery
ligation in extracorporeal membrane oxygenation
. J Pediatr 1988.113(1 pt 1): 110–113.
21. Schumacher RE, Barks JD, Johnston MV, et al. Right-sided brain lesions in infants following extracorporeal membrane oxygenation
. Pediatrics 1988.82: 155–161.
22. Lott IT, McPherson D, Towne B, Johnson D, Starr A. Long-term neurophysiologic outcome after neonatal extracorporeal membrane oxygenation
. J Pediatr 1990.116: 343–349.
23. Mendoza JC, Shearer LL, Cook LN. Lateralization of brain lesions following extracorporeal membrane oxygenation
. Pediatrics 1991.88: 1004–1009.
24. Hahn JS, Vaucher Y, Bejar R, Coen RW. Electroencephalographic and neuroimaging findings in neonates undergoing extracorporeal membrane oxygenation
. Neuropediatrics 1993.24: 19–24.
25. Taylor GA, Short BL, Fitz CR. Imaging of cerebrovascular injury in infants treated with extracorporeal membrane oxygenation
. J Pediatr 1989.114(4 pt 1): 635–639.
26. Lohrer RM, Bejar RF, Simko AJ, Moulton SL, Cornish JD. Internal carotid artery
blood flow velocities before, during, and after extracorporeal membrane oxygenation
. Am J Dis Child 1992.146: 201–207.
27. Lazar EL, Abramson SJ, Weinstein S, Stolar CJ. Neuroimaging of brain injury in neonates treated with extracorporeal membrane oxygenation
: Lessons learned from serial examinations. J Pediatr Surg 1994; 29: 186–190, discussion 190.
28. Bulas DI, Glass P, O’Donnell RM, Taylor GA, Short BL, Vezina GL. Neonates treated with ECMO: Predictive value of early CT and US neuroimaging findings on short-term neurodevelopmental outcome. Radiology 1995.195: 407–412.
29. Lago P, Rebsamen S, Clancy RR, et al. MRI, MRA, and neurodevelopmental outcome following neonatal ECMO. Pediatr Neurol 1995.12: 294–304.
30. Bennett CC, Johnson A, Field DJ, Elbourne D; UK Collaborative ECMO Trial Group: UK collaborative randomised trial of neonatal extracorporeal membrane oxygenation
: Follow-up to age 4 years. Lancet 2001.357: 1094–1096.
31. Roelants-van Rijn AM, van der Grond J, de Vries LS, Groenendaal F. Cerebral proton magnetic resonance spectroscopy of neonates after extracorporeal membrane oxygenation
. Acta Paediatr 2001.90: 1288–1291.
32. Mitchell DG, Merton D, Desai H, et al. Neonatal brain: Color Doppler imaging. Part II. Altered flow patterns from extracorporeal membrane oxygenation
. Radiology 1988.167: 307–310.
33. Taylor GA, Short BL, Glass P, Ichord R. Cerebral hemodynamics in infants undergoing extracorporeal membrane oxygenation
: Further observations. Radiology 1988.168: 163–167.
34. Matsumoto JS, Babcock DS, Brody AS, Weiss RG, Ryckman FG, Hiyama D. Right common carotid artery
ligation for extracorporeal membrane oxygenation
: cerebral blood flow velocity measurement with Doppler duplex US. Radiology 1990.175: 757–760.
35. Liem KD, Hopman JC, Oeseburg B, de Haan AF, Festen C, Kollée LA. Cerebral oxygenation and hemodynamics during induction of extracorporeal membrane oxygenation
as investigated by near infrared spectrophotometry. Pediatrics 1995.95: 555–561.
36. Fukuda S, Aoyama M, Yamada Y, et al. Comparison of venoarterial versus venovenous access in the cerebral circulation of newborns undergoing extracorporeal membrane oxygenation
. Pediatr Surg Int 1999.15: 78–84.
37. Van Heijst A, Liem D, Hopman J, Van Der Staak F, Sengers R. Oxygenation and hemodynamics in left and right cerebral hemispheres during induction of veno-arterial extracorporeal membrane oxygenation
. J Pediatr 2004.144: 223–228.
38. Ejike JC, Schenkman KA, Seidel K, Ramamoorthy C, Roberts JS. Cerebral oxygenation in neonatal and pediatric patients during veno-arterial extracorporeal life support
. Pediatr Crit Care Med 2006.7: 154–158.
39. O’Brien NF, Hall MW. Extracorporeal membrane oxygenation
and cerebral blood flow velocity in children. Pediatr Crit Care Med 2013.14: e126–e134.