If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Division of Cardiovascular Surgery, Hospital of the University of Pennsylvania, Philadelphia, PaDepartment of Cardiovascular Surgery, Heart Center University Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany
The optimal method for arterial cannulation in acute aortic dissection type A (ADA) remains controversial. The aim of this study was to compare central ascending aortic, axillary, and femoral cannulation in patients who underwent surgery for acute ADA.
Between 2006 and 2017, 584 patients were operated on for acute ADA. Of those, 355 (61%) underwent ascending aortic, 101 (17%) right axillary, and 128 (22%) femoral cannulation for arterial inflow. Clinical features and outcomes were compared after inverse probability weighting.
After inverse probability weighting there were no statistical differences in preoperative characteristics. Operative details differed significantly among the 3 groups: hemiarch replacement was performed more often in the central aortic and the femoral group (P < .001), whereas total arch replacement was performed more often in the axillary group (P < .001). Cardiopulmonary bypass (P = .022) and aortic cross-clamp (P = .021) times were shortest in the aortic cannulation group and longest in the femoral cannulation group. Postoperative morbidities were similar; procedure-related stroke (P = .783) and the need for renal replacement therapy (P = .446). In-hospital mortality (P = .680) and long-term survival were similar (log rank, P = .704). Multilevel multivariate mixed effect logistic regression showed that the cannulation strategy was not associated with in-hospital mortality.
Central ascending aortic cannulation in patients with ADA can be used as safely as axillary or femoral cannulation, providing another option for quick and easy establishment of cardiopulmonary bypass.
Arterial cannulation in ADA is challenging and the optimal method remains controversial. We compared ascending aortic, axillary, and femoral cannulation in patients who underwent surgery for acute ADA and concluded that ascending aortic cannulation can be used safely. It provides another option for quick and easy establishment of cardiopulmonary bypass.
Arterial cannulation for type A aortic dissection (ADA) has traditionally been achieved through the common femoral artery or the right axillary artery, and both are the most commonly used cannulation strategies in current practice.
Also, there is only 1 recent study with a relatively small population that compared axillary and aortic cannulation, indicating comparable early results but poorer long-term outcome in the central aortic cannulation group.
Our group has used central aortic cannulation, axillary cannulation, and femoral cannulation for arterial cardiopulmonary bypass (CPB) inflow in ADA patients for more than 10 years. The aim of this study was to evaluate early and late outcome of the 3 different cannulation methods. To reduce the effect of selection bias and potential confounding, this investigation was performed after inverse probability weighting.
The study population consisted of 584 patients operated on for acute ADA between January 2006 and June 2017. Diagnosis was confirmed using computed tomography (CT) scan and intraoperative transesophageal echocardiography (TEE). Patients were divided into 3 groups according to the cannulation site: central ascending aortic cannulation, right axillary cannulation, and femoral cannulation. Patients with previous cardiac surgery were not included in the study. We compared clinical features and outcomes after surgery. In addition, a subgroup analysis was performed for the safety of axillary cannulation in patients with the dissection involving the innominate and right axillary artery. The institutional review committee approved this retrospective study and the need for informed consent was waived.
Outcomes included postoperative, 1-month, and late follow-up data obtained from our center's aortic database that includes all available data from the University of Pennsylvania Health System as well as information gathered from patients' general practitioners or the patients and their family members. All patients were routinely followed-up after 1 month, and patients were encouraged to present to the aortic clinic for late follow-up. In case follow-up data were older than 6 months in surviving patients the University of Pennsylvania Health System database was screened for any entry. In patients without an entry, social security death index was screened for additional information. Follow-up data were available in 516 of 518 hospital survivors (99.6%). Patients were followed-up for a total of 2100 patient-years, with a median follow-up among survivors of 4.1 ± 3.1 years. Organ malperfusion was diagnosed with clinical examination and confirmed using CT angiography and was defined as altered blood flow with clinical evidence of a lack of blood flow resulting in ischemia with organ dysfunction. Shock was defined as persistent hypotension (systolic blood pressure <80 mm Hg) with severe reduction in cardiac index. Dissection-related stroke was defined as postoperative stroke in patients with clear preoperative neurologic deficit in the absence of hemodynamic compromise, whereas procedure-related stroke was defined as postoperative stroke in patients without clear preoperative neurologic deficit in the absence of hemodynamic compromise. Overall stroke included all patients with either dissection-related or procedure-related stroke. Early mortality was defined as in-hospital death and length of in-hospital stay was censored if a patient died during the hospital stay. Data for clinical presentation at admission and initial extent of the dissection were gathered from preoperative documentation.
For all patients a previously described standardized integrated surgical management was applied. This approach includes fast-track direct admission to the operating room and immediate surgical repair.
The proximal aortic repair included the replacement of the entire ascending aorta either with resuspension of the aortic valve (supracoronary ascending replacement, sinus of Valsalva repair with a Teflon felt neo-media and valve resuspension in patients with native aortic valve), aortic root replacement with a valved conduit, a Wheat procedure (aortic valve replacement, sinus of Valsalva repair with a Teflon felt neo-media and supracoronary ascending replacement), or a valve-sparing aortic root replacement as described previously.
In short, the entire residual ascending aorta and most of the lesser curvature of the aortic arch were routinely replaced in an aggressive, extended transverse hemiarch fashion and dissected layers were repaired with Teflon felt neo-media. Isolated ascending replacement was performed only in rare instances (7 patients in the study period). In case of an entry tear at the larger curvature of the aortic arch, complete arch replacement was performed. Aortic arch surgery was performed with the patient in hypothermic circulatory arrest (HCA) with selective cerebral perfusion and open distal anastomosis in all but 5 patients in the study. In those 5 patients, the aortic cross-clamp was placed distally enough for an isolated ascending replacement with complete removal of dissected tissue. Cerebral protection was achieved by either antegrade cerebral perfusion or retrograde cerebral perfusion (RCP). In those requiring more extensive arch reconstruction necessitating more than 40 minutes of circulatory arrest and RCP, selective unilateral or bilateral antegrade cerebral perfusion was typically done using balloon-tipped cannulas placed in the innominate and left common carotid arteries. In patients with right axillary cannulation, selective antegrade cerebral perfusion was established through the axillary artery by clamping the innominate artery and, when needed, a second balloon-tipped cannula placed in the left common carotid artery. In case of femoral or central aortic cannulation, an arterial cannula was inserted in the graft and antegrade CPB was restarted after HCA.
Specific technical details on ascending aortic cannulation have been previously described by our group (Video 1).
Briefly, the relationship of the true and false lumen in the ascending aorta and aortic arch were evaluated according to CT angiography, intraoperative TEE, and during visual examination of the ascending aorta. The true lumen was cannulated using the Seldinger technique (Figures 1 and 2). Careful collaboration with the echocardiographer is of paramount importance, and the wire has to be well down in the true lumen within the descending aorta before dilatation. In patients with large arch tears, we are especially careful, because it might be difficult to keep the wire in the true lumen when advancing the wire distally. Femoral cannulation was performed in a similar fashion with true lumen cannulation using the Seldinger technique and TEE confirmation of correct wire placement. Axillary cannulation was performed through a Dacron graft anastomosed to the axillary artery. The cannulation sites were selected primarily according to the surgeons' preference and the location of the primary entry tear.
We used inverse probability weighting to limit confounding by indication (see the Appendix E1 for details). Weights were calculated using the procedures implemented in the twang package for R.
Because no significant outliers were observed in weights among cannulation groups, they were used for the regressions. Balance between cannulation groups was assessed with the use of standardized mean differences. A standardized difference of ≤10% was considered an ideal balance, and a standardized difference of ≤30% was considered an acceptable balance. Bias reduction for variables with >20% standardized difference are reported in Table E1 as well.
Data are presented as mean ± standard deviation for normally distributed variables and as median (first quartile-third quartile) for nonparametrically distributed variables. Analysis of variance or the Kruskal–Wallis test were used for comparisons of continuous variables. Categorical variables were compared using the χ2 test with the calculation of exact values and are shown as number (percentage). In case of small group sizes (n < 5), the Fisher exact test was used.
A weighted 2-level multivariate mixed effect logistic regression with a robust variance estimator was used to determine the marginal effect of different cannulation groups on in-hospital mortality as well as the influences of selected variables (sex, chronic obstructive pulmonary disease, chronic renal failure, age >80 years, shock, tamponade, any organ malperfusion, and cannulation type), with surgeon-specific factors included as a random effect. Weighted Cox regression with a robust variance estimator was used to estimate long-term mortality differences between cannulation groups. As sensitivity analyses, stabilized weights were also generated and applied to the aforementioned logistic and Cox models (see the Appendix E1 for details).
All tests were performed with an α threshold of .05. The statistical analysis was performed using STATA version 15.1 (STATA Corp, College Station, Tex) and R version 3.5.1 (R Foundation for Statistical Computing, Vienna, Austria).
Of the 584 unweighted patients, central aortic cannulation was performed in 355 (61%) patients, right axillary cannulation was performed in 101 (17%) patients, and femoral cannulation was performed in 128 (22%) patients. After inverse probability weighting, there were no difference in baseline patient characteristics (Table E2). Characteristics before inverse probability weighting are summarized in Table 1.
Table 1Baseline characteristics of the study population before inverse probability weighting
Aortic (n = 355)
Axillary (n = 101)
Femoral (n = 128)
60 ± 14
58 ± 14
60 ± 14
Diabetes mellitus type 2
Chronic renal failure
Coronary artery disease
Bicuspid aortic valve
Values are n (%) or mean ± standard deviation. Bold indicates significant results (P < .05). COPD, Chronic obstructive pulmonary disease.
Before inverse probability weighting, significantly more patients presented in shock in the aortic and the femoral cannulation group and significantly more patients in the axillary and femoral group had complete aortic dissection compared with the central aortic cannulation group (Table 2). There was no difference after inverse probability weighting (Table E3).
Table 2Clinical characteristics of the study population before inverse probability weighting
Aortic (n = 355)
Axillary (n = 101)
Femoral (n = 128)
Moderate to severe AR
Any organ malperfusion
Limited, proximal dissection
Supra-aortic vessel dissection
Right axillary artery
One iliac artery
Both iliac arteries
Values are n (%) except where otherwise noted. Bold indicates significant results (P < .05). AR, Aortic regurgitation.
Of the 584 surgeries for ADA, 548 surgeries were performed by 1 of the 10 aortic surgeons working at the University of Pennsylvania Health System during the study period. Thirty-six surgeries were performed by other cardiothoracic surgeons performing <20 ADA surgeries a year. Operative details and times are summarized in Table 3. Although there was no difference in the proximal aortic repair (aortic valve resuspension was the most frequent proximal repair in all groups), the distal aortic repair differed significantly among the 3 groups: hemiarch replacement was performed more often in the aortic and the femoral cannulation groups (more than 90% of all patients), whereas total arch replacement was performed significantly more often in the axillary group. Operative times were shortest in the aortic cannulation group. RCP was the preferred method of cerebral perfusion in the aortic and the femoral cannulation group, whereas antegrade cerebral perfusion was significantly more common in the axillary cannulation group.
Table 3Operative details of the study population after inverse probability weighting
The distribution of perioperative complications was similar among all groups (Table 4): there was no statistical difference in the tracheotomy rate, the rate for re-exploration for bleeding, procedure-related and/or dissection-related stroke, temporary neurological dysfunction, renal failure, or the need for renal replacement therapy. The length of stay in the intensive care unit as well as the time to discharge were similar among the 3 groups. In-hospital mortality (aortic: 10% vs axillary: 11% vs femoral: 13%; P = .680) was comparable among the groups and was lowest in the aortic cannulation group. Long-term survival did not differ (Figure 3; log rank, P = .704).
Table 4Outcome characteristics of the study population after inverse probability weighting
Stroke in patients without clear preoperative neurologic deficit in the absence of hemodynamic compromise.
Temporary neurological dysfunction
Renal replacement therapy
ICU stay, d
In-hospital stay, d
Values are n (%) or median (first quartile-third quartile) except where otherwise noted. Bold indicates significant results (P < .05). ESS, Effective sample size; PRBC, packed red blood cells; FFP, fresh frozen plasma; ICU, intensive care unit.
∗ Stroke in patients with clear preoperative neurologic deficit in the absence of hemodynamic compromise.
† Stroke in patients without clear preoperative neurologic deficit in the absence of hemodynamic compromise.
The multilevel multivariate mixed effect logistic regression showed that the cannulation strategy was not associated with in-hospital mortality, after adjusting for surgeon-specific factors (Table 5). However, age >80 years (odds ratio, 5.000; P = .018) and any organ malperfusion (odds ratio, 4.065; P < .001) were predictors of in hospital mortality.
Of the 584 patients with ADA, the dissection extended into the innominate artery in 208 patients in this study population. Of those, axillary cannulation was performed in 40 patients compared with 168 patients with nonaxillary cannulation. Similar rates of dissection-related stroke (axillary: n = 3 [5%]; nonaxillary: n = 23 [7%]; P = .425), procedure-related stroke (axillary: n = 2 [5%], nonaxillary: n = 11 (7%); P = 1.000), and overall stroke (axillary: n = 5 [13%], nonaxillary: n = 34 [20%]; P = .367) were observed. The aortic dissection extended into the axillary artery in 6 patients in the axillary group and no stroke was diagnosed in those patients. In-hospital mortality was comparable between the 2 groups (axillary: n = 4 [10%], nonaxillary: n = 18 [11%]; P = 1.000).
The main findings of this study can be summarized as follows: (1) central ascending aortic cannulation in patients with ADA can be used as safely as axillary or femoral cannulation; (2) axillary cannulation is more commonly used for patients who require total arch repair, and does not reflect in a higher incidence of postoperative morbidity or mortality; and (3) cannulation of right axillary arteries in patients with ADA with involvement of the innominate artery does not result in worse clinical outcome and can safely be performed when needed.
The ideal cannulation strategy for ADA is quick to establish, minimizes any organ malperfusion, prevents progression of the dissection flap or aortic rupture, and allows optimal brain protection when HCA is needed.
and shock requiring prompt CPB commencement was more common in patients with those 2 cannulation methods in this study although the difference did not reach statistical significance in the inverse probability weighted analysis. Nevertheless, operative times were shortest in the aortic cannulation group, whereas operative times were significantly longer in the axillary cannulation group. This might be attributed to longer hemostasis times after total arch replacement, but it seems reasonable that the longer time to dissect the axillary artery contributed to the longer operative times as well.
Antegrade perfusion is considered to be advantageous in preventing perioperative malperfusion
To avoid this serious complication, it is essential to carefully evaluate the relationship of the true and false lumen in the ascending aorta and aortic arch according to preoperative CT angiography and intraoperative TEE. We strongly recommend using the Seldinger technique and controlling the wire and later the position of the aortic cannula with TEE, which considerably reduces the risk of inserting the cannula into the false lumen. In this study, we did not observe a difference in the cannulation strategy in patients with any preoperative organ malperfusion or with involvement of the supra-aortic vessels. Also, the different perfusion strategies did not result in different rates of perioperative malperfusion as indicated by the similar rates of postoperative stroke, postoperative renal failure, or the need for renal replacement therapy. However, a considerably higher incidence of temporary neurological deficit was observed in the femoral cannulation group (P = .051), and the incidence of procedure-related stroke was the highest in this group too (not significant). Embolization has been reported to be a problem with femoral cannulation.
Our data do not prove that this was a perioperative problem, but the possibility of microembolization cannot be ruled out.
Only few centers routinely use central aortic cannulation for the surgical treatment of ADA and the acceptance for central aortic cannulation is limited by concerns regarding its safety. In fact, some authors have reported aortic rupture after central aortic cannulation,
but we have not observed one case of aortic rupture after central aortic cannulation in this large case series. We routinely use the Seldinger technique in all cases of central aortic cannulation, which allows controlled insertion of the cannula and minimizes the stress applied to the dissected aortic wall.
The operative details in this study reveal the significant difference among the 3 cannulation groups: axillary cannulation is more commonly used for patients who require total arch repair. This highlights the substantial influence of the location of the dissection tear on the choice of cannulation strategy. In our center, we routinely perform aggressive hemiarch replacement with resection of the entire lesser curvature and most of the dissected aortic arch wall in ADA patients. Total aortic arch replacement is performed when the intimal tear is along the greater curvature of the aortic arch.
Therefore, axillary cannulation is the preferred cannulation method for more complex surgery in patients with more complex aortic dissections. In fact, in the axillary cannulation group HCA times were longest and the incidence of complete aortic dissection was highest. However, central cannulation seems to be the preferred method in patients with proximal ADA and operative times were shortest in this group. Of note, the cannulation method depended on the surgeon's personal preference in this study, and the cannulation method was individualized for each patient. At this point, there is no institutional cannulation policy. Generally, axillary cannulation seems to be the preferred cannulation method when the need for complete arch replacement is anticipated, particularly in stable patients. In unstable patients, central aortic cannulation (especially in patients with tamponade or when complete arch replacement is likely) or femoral cannulation are the preferred strategies.
Despite the significant differences in operative strategy and the invasiveness of the surgery, we observed similar postoperative morbidity among the study groups. In addition, in-hospital mortality was similar among all 3 cannulation groups with a numerically, but not statistically, higher mortality rate in the femoral cannulation group. Late survival analysis revealed no significant statistical difference among the groups (Figure 3), whereas others have reported significantly worse long-term survival in patients after central aortic cannulation in smaller collectives.
Lastly, we were able to confirm the safety and efficacy of cannulating the right axillary artery in the setting of a dissected innominate artery for ADA repair. To date, a scarcity of data exist regarding this question and our findings support the previously reported safe usage of the axillary artery cannulation site although the innominate artery is involved in the dissection
when needed. Our cannulation strategy was not influenced by involvement of the innominate artery. However, in cases in which the axillary artery was involved, axillary cannulation was probably not the first choice for arterial inflow, although our data suggest it can be done when no other cannulation method is optimal.
Limitations and Strengths
Although this was a retrospective single-center study, it is—to the best of our knowledge—the first to compare all 3 commonly used cannulation sites. Also, it is the largest study to compare central aortic cannulation with axillary and femoral cannulation. However, we cannot exclude bias in terms of patient selection and treatment because the cannulation method and the choice of cerebral perfusion remained the individual surgeon's preference. Also, our retrospective database analyses might under-report events and risk factors and clinical outcome analysis focused on postoperative mortality and morbidity, and we were not able to provide information on late complications, quality of life, and cause of death after discharge.
Central cannulation in patients with acute ADA can be used as safely as axillary or femoral cannulation, providing another option for quick and easy establishment of CPB. In addition, cannulation of right axillary arteries in patients with ADA with involvement of the innominate artery does not result in worse clinical outcome and can be performed when needed.
Conflict of Interest Statement
Authors have nothing to disclose with regard to commercial support.
Appendix E1. Methods
Propensity score pair-matching is widely adopted for causal inference in observational studies via creating matched pairs with similar probability of receiving the same treatment, yet received different treatments. However, patients might be excluded from the study because of lack of appropriate match(es), which will not only decrease the sample size, but also limit the generalizability of the analysis. We aimed to investigate the effect of the cannulation method on all patients who received ADA repair (average treatment effect) with 584 observations. Therefore, inverse probability weighting was adopted to balance baseline characteristics (age, sex, hyperlipidemia, hypertension, diabetes, stroke, chronic renal failure, chronic obstructive pulmonary disease, coronary artery disease, bicuspid aortic valve disease, and Marfan syndrome, shock, tamponade, moderate to severe aortic regurgitation, organ malperfusion, dissection type [limited proximal, innominate, right axillary artery, left subclavian/left common carotid artery, 1 iliac artery, both iliac arteries]). We used the twang package in R 3.5.1 (R Institute for Statistical Computing) to perform matching between 3 cannulation groups with 3000 iterations. Default stopping rules were used (ie, using the absolute standardized bias metrics and summarizes across variables with the mean. The assigned iteration number was determined to be adequate and appropriate; Figure E1). The generated multinomial propensity scores met the overlap assumption, and are graphed in Figure E2. The weights obtained are graphed in Figure E3. After weighing, the standardized mean differences are presented in Love plots (Figures E4 and E5), whereas variables with >20% standardized difference are reported in Table E1.
Table E1Standardized mean difference decrease in variables with >20% standardized difference
Stabilized weights were calculated by trimming the weights at the 90th percentile (ie, replacing weights over the 90th percentile as the 90th percentile weights). The distributions of stabilized weights are graphed in Figure E5. The results of logistic regression assessing differences of in-hospital mortality with stabilized weights were similar to results from using raw weights (Table E4). The results of Cox regression assessing differences in long-term survival with stabilized weights were also similar to results from using raw weights (Figure E6).
Table E4Logistic regression, in-hospital mortality, using stabilized weight
Type A aortic dissection (TAAD) requires expedient surgical treatment. The first challenge encountered by cardiac surgeons is cannulation strategy for cardiopulmonary bypass. Cannulation requires thoughtful integration of the aortic anatomy, dissection extent, planned aortic repair strategy, prior surgical history, and hemodynamic status. Femoral cannulation has been a traditional mainstay in dissection repair. As techniques for cerebral perfusion have evolved, proximal arterial cannulation has increased to allow for antegrade flow and antegrade cerebral perfusion.
Cannulation techniques for surgical intervention on the thoracic aorta requiring cardiopulmonary bypass has experienced quite an evolution since Lellehei first reported the use of femoral cannulation.1 As a result of the early complications associated with femoral cannulation for cardiac surgery, mainly related to arterial dissection and retrograde cerebral and visceral emboli,2 this method was soon replaced by direct cannulation of the ascending aorta in the main surgical field and, subsequently, the axillary artery in the presence of ascending aortic pathology.