Abstract
Background
The optimal level of hypothermia and safe time of unilateral antegrade cerebral perfusion (uACP) in acute type A aortic dissection (ATAAD) repair remain controversial.
Objectives
To analyze the association of uACP time and circulatory arrest temperature with surgical outcomes of ATAAD.
Methods
We retrospectively analyzed 263 patients who had undergone ATAAD repair between 2006 and 2020 using uACP. The patients were stratified by three chronologically equivalent periods (period 1, 2006 to 2010; period 2, 2011 to 2015; period 3, 2016 to 2020) to demonstrate the decade-long evolution of surgical strategy and outcomes.
Results
The mean age of the patients was 59.4 ± 12.5 years, and 68.8% were male. The hospital mortality rates were 15.1%, 12.9%, and 11.0% from period 1 to 3 (p = 0.740). The median circulatory arrest temperatures were 20, 23, and 25 °C (p < 0.001), respectively, and the median uACP times were 72, 59, and 41 minutes (p < 0.001). The incidence rates of postoperative permanent neurologic deficits were 13.2%, 10.9%, and 18.3% (p = 0.312), and those of transient neurologic deficits were 9.4%, 10.9%, and 11.9% (p = 0.936), respectively. Multivariate logistic regression analysis showed that uACP time ≥ 60 minutes was an independent predictor of hospital mortality rather than postoperative stroke. ROC curve analysis estimated an optimal cutoff value of 52 minutes of uACP time when the circulatory arrest temperature was ≥ 25 °C to predict hospital mortality (area under the curve: 0.72).
Conclusions
Unilateral antegrade cerebral perfusion time was associated with hospital mortality after ATAAD surgery. A safe threshold of 50 to 60 minutes of uACP should be considered.
Keywords: Aortic dissection, Cerebral perfusion
INTRODUCTION
Antegrade cerebral perfusion (ACP) is generally considered to be preferable to deep hypothermic circulatory arrest and retrograde cerebral perfusion in acute type A aortic dissection (ATAAD).1-5 However, the optimal hypothermic level during ACP, unilateral (uACP) or bilateral (bACP), and superior cannulation strategy are still under debate.6-8 uACP is favorable due to its simplicity, less manipulation of supra-aortic vessels, and effectiveness in ATAAD repair.6,9,10 The evolving trend of extensive arch reconstruction requires a longer circulatory arrest time. This raises concerns over the safe time limit of uACP. Most studies with good uACP results have reported needing less than 40 minutes of circulatory arrest and cerebral perfusion time.11 Whether more extended uACP can provide acceptable clinical results during ATAAD repair is uncertain. uACP based on right axillary cannulation has been a primary strategy for ATAAD at our institution since 2006. This study aimed to analyze our 15-year experience of uACP, evaluate the evolution of our surgical approach and outcomes, and focus on the association of uACP time and circulatory arrest temperature with hospital mortality and postoperative neurological complications.
METHODS
Patients
The Ethics Committee and Institutional Review Board of our hospital approved this retrospective study. We have performed uACP via the right axillary artery as the first-choice strategy for ATAAD since January 2006. Up to December 2020, 295 consecutive patients underwent ATAAD surgery. Patients who underwent deep hypothermic circulatory arrest or retrograde cerebral perfusion (n = 6), whose uACP was preceded by pure deep hypothermic circulatory arrest for more than 10 minutes (n = 12), and those who underwent bACP because of surgeons’ judgment (n = 14) were excluded. Finally 263 patients were included, and they were further divided into chronologically equivalent periods as period 1 (2006 to 2010, n = 53), period 2 (2011 to 2015, n = 101), and period 3 (2016 to 2020, n = 109) (Figure 1) to evaluate the evolution of our surgical techniques and outcomes over 15 years.
Figure 1.
Flowchart of patient inclusion and exclusion. ACP, antegrade cerebral perfusion; DHCA, deep hypothermic circulatory arrest; RCP, retrograde cerebral perfusion.
Preoperative evaluation
Computed tomography was routinely used for preoperative diagnoses. Preoperative transthoracic echocardiography or intraoperative transesophageal echocardiography was used to evaluate hemopericardium and valve regurgitation. We defined preoperative organ malperfusion as image evidence of compromised arterial flow plus clinical or laboratory evidence of organ ischemia. However, the diagnosis of renal malperfusion was sufficient with image findings alone.
Operative procedure
Since 2006, we have used right axillary artery cannulation via an 8-mm or 10-mm side-arm graft for the cardiopulmonary bypass and uACP for brain protection. Traditionally, the femoral artery was cannulated if refractory cardiogenic shock or cardiopulmonary resuscitation occurred before incision. In the more recent period, additional left axillary or femoral artery cannulation or alternative innominate artery cannulation was considered if the flow was limited by the size or dissection of the right axillary artery or early perfusion of the ischemic limb was deemed to be necessary. Then access of later uACP was set via either a right axillary artery side-arm graft, innominate graft, or left direct carotid cannulation with a balloon-tipped catheter during distal open anastomosis.
In our patients, the ascending aorta was cross-clamped and opened. Antegrade blood cardioplegia was infused directly into the coronary artery orifice for myocardial protection every 20-30 minutes. The proximal aorta was trimmed until only non-fragile tissue was left, and it was reinforced using the sandwich technique with Teflon felt. If root reconstruction was necessary, coronary buttons were created at this stage. Systemic circulatory arrest was initiated after the patient’s body had been cooled to the target temperature (ranging from 18 °C to 28 °C across the study periods). The innominate artery was clamped to start uACP at a 10 ml/kg/min flow to maintain a right radial arterial pressure of 50-70 mmHg. The left carotid and left subclavian arteries were clamped after backflow had been confirmed. Transcutaneous cerebral oximetry (INVOS 3100-SD; Somanetics, Troy, MI) was used, and a value < 75% of baseline was considered to be the threshold of brain ischemia.12 The ascending aorta and aortic arch were examined and trimmed until the entry site was excluded, if possible. It was reinforced in the same manner used in the proximal aorta, followed by an open distal aortic anastomosis. If the entry site was located in the ascending aorta only, an ascending aorta or hemi-arch replacement was done. If the entry site was found in the aortic arch or when the arch vessels were compromised, subtotal or total arch replacement was done with a custom-made bifurcated or trifurcated graft in the early period, but mostly with a quadrifurcated graft in the recent period. The indication for a frozen elephant trunk (Medtronic Valiant® thoracic stent graft, exclusively in period 3) was mostly due to distal true lumen compromise and malperfusion, but also selectively for young patients or those with a distal aortic aneurysm. The frozen elephant trunk was mainly combined with total arch replacement, but was also sometimes followed by hemi-arch replacement without reconstruction of supra-aortic vessels. Proximal reconstruction was traditionally done last when whole-body perfusion was resumed and systemic rewarming began. If postoperative extracorporeal membrane oxygenation (ECMO) was required, it was mainly established via central cannulation with delayed sternal closure.13
Outcome definitions
Surgical mortality was defined as mortality within 30 days after surgery. A permanent neurological deficit (PND) was defined as any neurological signs or stroke, plus image evidence of a compatible cerebral insult that did not improve before discharge. A transient neurological deficit (TND) was defined as any neurological signs or reversible neurocognitive dysfunction recognized without image evidence of cerebral damage. Brain magnetic resonance imaging (MRI) was arranged if there were no definite findings on computed tomography and the implanted prosthetic valve or frozen elephant trunk graft was MRI-compatible. Patients with complete recovery of neurological symptoms but with image findings showing new brain insults were categorized as having PND.
Statistical analysis
Continuous variables are expressed as median [interquartile range] and were tested with the nonparametric Kruskal-Wallis test. Categorical variables are presented as numbers (percentages), and Pearson’s χ 2 or Fisher’s exact test was used where appropriate. Risk factors for hospital mortality and neurological deficits were examined with binary logistic regression. Variables with p-values < 0.05 in the univariate logistic regression were included in the multivariate model. Variables with a variance inflation factor of more than five were considered to be multicollinear and were removed from the multivariate model. The chronological factor (period 2 or 3 versus 1) was arbitrarily included in the multivariate model to adjust for potential uncoded discrepancies in clinical protocols and surgical techniques among the three periods. The significance was set at p < 0.05. Receiver operating characteristic (ROC) curves were generated for significant continuous variables in the multivariate regression model to determine the optimal cutoff threshold of each variable. The performance of the ROC curve was assessed according to the area under the curve (AUC). Statistical analyses were performed using R, version 4.1.0 (R Foundation for Statistical Computing, Vienna, Austria).
RESULTS
Patient demographics and presentation
The mean age of the patients was 59.4 ± 12.5 years, and 68.8% were male. The distributions of age, male-to-female ratio, and most underlying comorbidities were comparable among the three periods, except that the patients in period 2 had a higher incidence of old stroke before aortic dissection, and the patients in period 3 had a lower incidence of underlying liver disease and the lowest serum level of the liver enzyme alanine transaminase before surgery (Table 1).
Table 1. Baseline characteristic and preoperative presentations.
| Period 1 | Period 2 | Period 3 | p value | |
| 2006-2010 (N = 53) | 2011-2015 (N = 101) | 2016-2020 (N = 109) | ||
| Age | 38 (71.7) | 70 (69.3) | 73 (67.0) | 0.829 |
| Male | 56.4 [50.8, 67.8] | 60.6 [51.9, 69.4] | 61.2 [50.0, 68.0] | 0.654 |
| DeBakey type 2 | 4 (7.5) | 10 (9.9) | 10 (9.2) | 0.961 |
| Hypertension | 44 (83.0) | 91 (90.1) | 101 (92.7) | 0.181 |
| Diabetes | 3 (5.8) | 5 (5.0) | 11 (10.1) | 0.368 |
| Coronary artery disease | 5 (9.6) | 11 (10.9) | 4 (3.7) | 0.100 |
| Old stroke | 2 (3.8) | 11 (10.9) | 1 (0.9) | 0.004 |
| Chronic liver disease* | 5 (9.6) | 11 (10.9) | 2 (1.8) | 0.014 |
| ALT (mg/dl) | 29.0 [20.0, 48.8] | 29.0 [19.0, 52.0] | 20.5 [15.2, 37.8] | 0.004 |
| Chronic kidney disease# | 8 (15.4) | 8 (7.9) | 8 (7.3) | 0.231 |
| Creatinine (mg/dl) | 1.2 [0.9, 1.7] | 1.1 [0.9, 1.4] | 1.1 [0.9, 1.4] | 0.157 |
| End stage renal disease | 1 (1.9) | 1 (1.0) | 5 (4.6) | 0.328 |
| Marfan syndrome | 2 (3.8) | 4 (4.0) | 0 (0.0) | 0.065 |
| Previous type B dissection | 4 (7.5) | 6 (5.9) | 3 (2.8) | 0.297 |
| Previous aortic aneurysm | 6 (11.3) | 9 (8.9) | 7 (6.4) | 0.569 |
| ≥ Moderate aortic regurgitation | 22 (41.5) | 34 (33.7) | 30 (27.5) | 0.201 |
| Number of organ malperfusion | 0.0 [0.0, 1.0] | 0.0 [0.0, 1.0] | 1.0 [0.0, 2.0] | 0.001 |
| Brain | 1 (1.9) | 10 (9.9) | 11 (10.1) | 0.169 |
| Spinal cord | 0 (0.0) | 2 (2.0) | 2 (1.8) | 0.839 |
| Myocardium | 7 (13.5) | 12 (11.9) | 14 (12.8) | 0.970 |
| Mesentery | 2 (3.8) | 4 (4.0) | 10 (9.2) | 0.257 |
| Kidney | 8 (15.4) | 9 (8.9) | 31 (28.4) | 0.001 |
| Extremity | 8 (15.4) | 21 (20.8) | 39 (35.8) | 0.008 |
| Cardiac tamponade | 13 (24.5) | 23 (22.8) | 22 (20.2) | 0.777 |
| Hemopericaridum | 22 (41.5) | 44 (43.6) | 36 (33.0) | 0.279 |
* Chronic hepatitis or cirrhosis. # Baseline creatinine > 1.5 mg/dl.
ALT, alanine transaminase.
The incidence of aortic regurgitation or cardiac tamponade was similar between groups. The cumulative incidence of organ malperfusion was highest in period 3 (median one organ malperfusion per patient, p = 0.001), among which kidney (28.4%) and limb (35.8%) malperfusion were significantly more common compared with the other periods (Table 1).
Procedures and intraoperative findings
There was a significant evolution of our surgical strategy from period 1 to period 3 (Table 2). In periods 1 and 2, nearly all of the cannulation strategies were restrictively based on the axillary artery. In period 3, the cannulation strategy was more individualized, such as innominate cannulation (8.2%) or adjunctive femoral artery cannulation (30.3%), which reflected the higher incidence of preoperative limb malperfusion in this period.
Table 2. Intraoperative findings.
| Period 1 | Period 2 | Period 3 | p value | |
| 2006-2010 (N = 53) | 2011-2015 (N = 101) | 2016-2020 (N = 109) | ||
| Arterial cannulation site | < 0.001 | |||
| Axillary | 48 (90.6) | 95 (94.1) | 59 (54.1) | |
| Axillary + Femoral | 5 (9.4) | 5 (5.0) | 33 (30.3) | |
| Femoral | 0 (0.0) | 0 (0.0) | 8 (7.3) | |
| Innominate | 0 (0.0) | 1 (1.0) | 9 (8.2) | |
| Circulatory arrest temperature (°C) | 20.0 [20.0, 20.0] | 23.0 [20.0, 25.0] | 25.0 [25.0, 25.0] | < 0.001 |
| uACP time (minutes) | 72.0 [50.0, 104.0] | 59.0 [48.0, 78.0] | 41.0 [31.0, 59.0] | < 0.001 |
| Cross-clamp time (minutes) | 193.0 [145.2, 230.8] | 170.0 [136.0, 213.0] | 117.0 [90.0, 166.0] | < 0.001 |
| CPB time (minutes) | 275.0 [233.0, 340.0] | 259.0 [221.0, 326.0] | 220.0 [187.0, 290.0] | 0.001 |
| Proximal procedure | 0.099 | |||
| Valve resuspension | 3 (5.7) | 24 (23.8) | 20 (18.3) | |
| Aortic valve replacement | 0 (0.0) | 1 (1.0) | 1 (0.9) | |
| Bentall’s procedure | 17 (32.1) | 23 (22.8) | 21 (19.3) | |
| David’s procedure | 1 (1.9) | 1 (1.0) | 3 (2.8) | |
| Distal procedure | ||||
| Hemiarch | 11 (20.8) | 26 (25.7) | 17 (15.6) | |
| Subtotal arch | 22 (41.5) | 22 (21.8) | 41 (37.6) | |
| Total arch | 6 (11.3) | 11 (10.9) | 17 (15.6) | |
| Elephant trunk | < 0.001 | |||
| Traditional | 3 (5.7) | 10 (9.9) | 4 (3.7) | |
| Frozen | 0 (0.0) | 0 (0.0) | 16 (14.6) | |
| Associate procedure | ||||
| Coronary artery bypass | 6 (11.3) | 9 (8.9) | 12 (11.0) | 0.866 |
| Femoral femoral bypass | 3 (5.7) | 3 (3.0) | 5 (4.6) | 0.674 |
| Delayed sternal closure | 6 (11.5) | 11 (10.9) | 19 (17.4) | 0.402 |
| PostOP ECMO | 5 (9.6) | 11 (10.9) | 11 (10.1) | 1.000 |
CPB, cardiopulmonary bypass; PostOP ECMO, postoperative extra-corporeal membrane oxygenation; uACP, unilateral antegrade cerebral perfusion.
The median systemic temperature during uACP increased from 20 °C in period 1 to 23 °C in period 2, and finally 25 °C in period 3 (p < 0.001, Table 2). This temperature change was associated with a reduction in uACP time from 72 minutes in period 1 to 41 minutes in period 3 (p < 0.001, Table 2). The aortic cross-clamp time and cardiopulmonary bypass time were significantly reduced throughout the three periods.
The choices of proximal root or valve procedures and the extent of distal arch procedures were not statistically different among the three periods. The patients in period 3 underwent significantly more traditional or frozen elephant trunk procedures (exclusively in period 3) compared to the other periods (18.3% versus 5.7% and 9.9% in periods 1 and 2, respectively, p < 0.001) (Table 2). However, the incidence of postoperative ECMO support was not different between groups.
Hospital mortality and postoperative complications
There was a trend of improvement in mortality rate over the 15 years (11.0% hospital mortality in period 3, compared to 15.1% in period 1 and 12.9% in period 2, Table 3). There was no significant difference in median postoperative alanine transaminase level between groups. However, a decreasing trend of postoperative creatinine across period 1 to period 3 was noted, parallel with the decrease in ACP time. There were no significant differences in the incidence of postoperative re-exploration for bleeding, hemodialysis, infection, or tracheostomy among the three groups, except that for the patients who underwent surgery in period 3, the incidence of postoperative new-onset PND (14.7%) was numerically more than double that of period 1 and 2 (5.7% and 6.9%, respectively, p = 0.109) (Table 2). The average occurrence of TND was around 10% in all three periods.
Table 3. Postoperative results.
| Period 1 | Period 2 | Period 3 | p value | |
| 2006-2010 (N = 53) | 2011-2015 (N = 101) | 2016-2020 (N = 109) | ||
| Check bleeding | 4 (7.7) | 14 (13.9) | 15 (13.8) | 0.545 |
| Creatinine (mg/dl)* | 2.3 [1.5, 3.3] | 1.7 [1.3, 2.8] | 1.6 [1.1, 3.5] | 0.063 |
| ALT (mg/dl)* | 61.0 [32.0, 99.5] | 72.5 [40.0, 142.8] | 64.0 [33.0, 165.0] | 0.201 |
| Neurologic deficit | ||||
| TND | 5 (9.4) | 11 (10.9) | 13 (11.9) | 0.936 |
| PND | 7 (13.2) | 11 (10.9) | 20 (18.3) | 0.312 |
| PND de novo# | 3 (5.7) | 7 (6.9) | 16 (14.7) | 0.109 |
| Respiratory failure† | 13 (24.5) | 19 (18.8) | 23 (21.1) | 0.695 |
| Hemodialysis | 7 (13.5) | 16 (15.8) | 22 (20.2) | 0.553 |
| Ischemic bowel | 1 (1.9) | 2 (2.0) | 2 (1.8) | 1.000 |
| Limb ischemia | 0 (0.0) | 2 (2.0) | 3 (2.8) | 0.733 |
| Wound infection | 4 (7.7) | 2 (2.0) | 7 (6.4) | 0.177 |
| Systemic infection | 10 (19.2) | 10 (9.9) | 12 (11.0) | 0.251 |
| ICU days | 4.2 [2.5, 10.0] | 5.0 [3.0, 8.3] | 5.0 [3.0, 11.0] | 0.245 |
| Hospital days | 14.9 [10.3, 22.8] | 15.0 [10.0, 26.0] | 16.0 [11.0, 25.0] | 0.817 |
| 24hr mortality | 3 (5.7) | 1 (1.0) | 0 (0.0) | 0.014 |
| Surgical mortality | 7 (13.2) | 9 (8.9) | 8 (7.3) | 0.485 |
| Hospital mortality | 8 (15.1) | 13 (12.9) | 12 (11.0) | 0.740 |
* Highest level within postoperative one week. # Postoperatively new onset stroke. † Tracheostomy or prolonged intubation > one week.
ALT, alanine transaminase; ICU, intensive care unit; PND, permanent neurologic deficit; TND, transient neurologic deficit.
Predictors of hospital mortality and neurological complications
Independent risk factors for hospital mortality are listed in Table 4. After adjusting for chronological factors, uACP time ≥ 60 minutes [odds ratio (OR): 2.88; p = 0.048], age (OR: 1.08; p = 0.001), total arch replacement (OR: 3.75; p = 0.021), and cross-clamp time (OR: 1.01; p = 0.011) predicted hospital mortality. Cannulation strategy, the temperature during circulatory arrest, and preoperative malperfusion were not significant, even in the univariate regression model.
Table 4. Predictors for hospital mortality.
| Univariate | Multivariate | |||||
| OR | 95% CI | p value | OR | 95% CI | p value | |
| Age | 1.06 | 1.03-1.10 | 0.001 | 1.08 | 1.03-1.12 | 0.001 |
| Total arch replacement | 3.73 | 1.54-8.64 | 0.003 | 3.75 | 1.22-11.64 | 0.021 |
| uACP time ≥ 60 mins | 4.81 | 2.21-11.39 | < 0.001 | 2.88 | 1.02-8.54 | 0.048 |
| Cross-clamp time (mins) | 1.01 | 1.01-1.02 | < 0.001 | 1.01 | 1.00-1.02 | 0.011 |
| Cardiac tamponade | 2.29 | 1.02-4.93 | 0.038 | 1.14 | 0.39-3.13 | 0.803 |
| Root replacement | 3.92 | 1.85-8.40 | < 0.001 | 1.83 | 0.55-5.83 | 0.311 |
| Period 2* | 0.83 | 0.33-2.24 | 0.703 | 1.2 | 0.40-3.88 | 0.755 |
| Period 3* | 0.7 | 0.27-1.89 | 0.460 | 1.69 | 0.50-6.08 | 0.405 |
| Circulatory arrest temperature ≥ 25 °C | 0.55 | 0.25-1.14 | 0.111 | |||
| Femoral artery cannulation | 0.72 | 0.31-1.80 | 0.453 |
* Versus period 1.
CI, confidence interval; OR, odds ratio; uACP, unilateral antegrade cerebral perfusion.
A multivariate logistic regression model for postoperative PND (Table 5) showed that total arch replacement (OR: 6.70; p < 0.001), preoperative brain malperfusion (OR: 12.56; p < 0.001), and cardiac tamponade (OR: 2.98; p = 0.018) were significant clinical predictors after adjusting for the surgical periods, cannulation strategy, circulatory arrest temperature, and any perfusion time. None of the perioperative factors were associated with postoperative TND in univariate regression.
Table 5. Predictors for permanent neurologic deficit.
| Univariate | Multivariate | |||||
| OR | 95% CI | p value | OR | 95% CI | p value | |
| Brain malperfusion | 6.57 | 2.56-16.73 | < 0.001 | 12.56 | 4.13-40.18 | < 0.001 |
| Cardiac tamponade | 2.40 | 1.13-4.97 | 0.020 | 2.98 | 1.19-7.44 | 0.018 |
| Total arch replacement | 5.05 | 2.22-11.29 | < 0.001 | 6.70 | 2.41-19.3 | < 0.001 |
| uACP time ≥ 60 mins | 2.03 | 1.02-4.12 | 0.045 | 1.31 | 0.46-3.72 | 0.606 |
| Cross-clamp time | 1.01 | 1.00-1.01 | < 0.001 | 1.01 | 0.99-1.01 | 0.112 |
| Root replacement | 3.32 | 1.62-6.79 | 0.001 | 2.25 | 0.73-6.77 | 0.151 |
| Period 2* | 0.80 | 0.30-2.31 | 0.671 | 0.94 | 0.28-3.43 | 0.922 |
| Period 3* | 1.48 | 0.60-4.00 | 0.412 | 2.39 | 0.72-9.02 | 0.173 |
| Circulatory arrest temperature ≥ 25 °C | 1.29 | 0.65-2.63 | 0.470 | |||
| Femoral artery cannulation | 0.62 | 0.29-1.44 | 0.246 |
* Versus period 1.
CI, confidence interval; OR, odds ratio; uACP, unilateral antegrade cerebral perfusion.
Threshold of cerebral perfusion time to predict hospital mortality
ROC curves between uACP time, circulatory arrest temperature, and hospital mortality were generated to define the cutoff threshold (Table 6). The overall AUC was 0.70, with an optimal cutoff ACP time of 60 minutes. An uACP time below 60 minutes was associated with a hospital mortality rate of 5.7%, and the mortality rate increased to 22.6% if the ACP time was more than 60 minutes. If the ROC curve was stratified with circulatory arrest temperature, the performance was better when the temperature was ≥ 25 °C (AUC 0.72 if the cutoff value was 52 minutes) (Figure 2). Since uACP time was not an independent predictor of postoperative PND, the ROC curve between ACP time and PND failed to achieve satisfactory performance.
Table 6. Performance of ROC curve to predict hospital mortality.
| Predictor | Optimal cut point | Area under curve | Hospital mortality % | |
| Below cut point | Above cut point | |||
| Temperature | 25 | 0.57 | 16.0 | 9.4 |
| uACP time | ||||
| Overall | 60 | 0.70 | 5.7 | 22.6 |
| < 25 °C | 68 | 0.65 | 8.3 | 26.4 |
| ≥ 25 °C | 52 | 0.72 | 2.5 | 19.0 |
ROC, receiver operating characteristic; uACP, unilateral antegrade cerebral perfusion.
Figure 2.
ROC curve of unilateral ACP time to hospital mortality, stratified by circulatory arrest temperature. ACP, antegrade cerebral perfusion; AUC, area under the curve; ROC, receiver operating characteristic.
DISCUSSION
In this retrospective study, we analyzed the surgical results of ATAAD using uACP as a cerebral protection strategy over 15 years at a single center. Our trend of decreasing surgical mortality from 13.2% to 7.3% and hospital mortality from 15.6% to 11.0% is consistent with the worldwide improvement in surgical outcomes of ATAAD reported in international registries.14,15 It was associated with an elevation of circulatory arrest temperature and a reduction in uACP time. As in other studies including decade-long cohorts,6,16 variations in the patients’ demographics, and clinical and surgical strategies over the years may not have been sufficiently coded for analysis, and adjustments for the period of surgery were necessary to exclude clinical confounders. After chronological adjustments, we found that a uACP time ≥ 60 minutes, age, cross-clamp time, and total arch replacement were significant risk factors for hospital mortality. Postoperative PND was not predicted by uACP time or temperature, however it was associated with preoperative brain malperfusion, cardiac tamponade, and total arch replacement.
There have been many reports of comparable surgical results between uACP and bACP under various circulatory arrest temperatures.5,6,9,10 The German Registry for ATAAD (GERAADA) registry5 reported equivalent 30-day mortality and neurologic outcomes between uACP and bACP (13.9% and 10.0% versus 15.9% and 11.0%, respectively), after mean ACP times of 32.2 and 37.6 minutes, with a wide range of circulatory arrest temperature from < 15 °C to > 30 °C. A meta-analysis of 5100 patients undergoing arch surgery also showed similar rates of 30-day mortality (8.6% for uACP versus 9.2% for bACP), PND (6.1% versus 6.5%), and TND (7.1% versus 8.8%), after a mean circulatory arrest time of around 30 minutes at 23 °C.17 Extracranial collaterals via concomitant right vertebral artery perfusion may compensate for most deficiencies in the circle of Willis. Monitoring of contralateral cerebral perfusion, such as transcutaneous cerebral oximetry, is mandatory. Additional contralateral perfusion should be considered when inadequate perfusion is evidenced by decreased left side cerebral oximetry or left carotid backflow. However, this seems exceptional, with a reported conversion rate ranging from 0.2 to 2%.6,18
Some authors have suggested that direct cannulation of the left carotid artery for bACP can increase the risk of embolic stroke, and therefore that uACP is preferrable because of its simplicity and equal effectiveness. Zierer et al.14 found a higher stroke rate in patients treated with bACP (6% versus 2% for uACP; p = 0.06). Norton et al. also found a trend of more left-sided embolic stroke in the bACP compared to the uACP group (3% versus 0.7%).6 These findings, however, do not preclude worries about inadequate protection for the contralateral hemisphere after prolonged uACP for complex arch reconstruction. Krähenbühl et al. found that bACP led to a better mid-term quality of life than uACP when the circulatory arrest time was longer than 40 minutes.19 In a meta-analysis of 3548 patients, Malvindi et al. concluded that, given a neurological injury rate threshold of less than 5%, 30-50 minutes of uACP was acceptable.20
In this study, a longer uACP time was an independent risk factor for hospital mortality. We estimated a safe time limit of uACP of around 60 minutes based on an ROC curve between uACP time and hospital mortality. Despite the systemic hypothermia during ACP, organ ischemia during the ongoing circulatory arrest is inevitable. The relatively higher serum levels of postoperative creatinine and alanine transaminase and the longer uACP time in our periods 1 and 2 partly explained the correlation between circulatory arrest time and systemic organ ischemia. The prediction performance of uACP time for hospital mortality was even more robust when the circulatory arrest temperature was ≥ 25 °C. In this subgroup, the AUC was 0.72, and an estimated safe time threshold of uACP of around 52 minutes was comparable with the suggested safe time threshold from the aforementioned comparative studies.20 Therefore, the influence of uACP time on hospital mortality may vary under different hypothermic levels, because additional adverse effects of hypothermia, such as longer cardiopulmonary bypass times, coagulopathy, and systemic vasoconstriction, could interfere with the impact of ongoing uACP time alone. Circulatory arrest temperature was not an independent predictor of hospital mortality in our regression model. However, with a shift in circulatory arrest temperature from low to high through the three periods, the improvement in mortality was apparent. Tsai et al. reported that deep hypothermia < 20 °C was independently associated with hospital mortality regardless of the ACP time.21 Tong et al. reported that ACP circulatory arrest time was an independent risk factor for mortality after total arch replacement in ATAAD.9
We did not find a significant association between uACP time and circulatory arrest temperature with PND. The influence of ACP time and circulatory arrest temperature on neurologic complications, however, remains controversial. Preventza et al. reported that greater actual temperatures during the circulatory arrest were associated with less stroke in proximal and arch operations involving ACP for > 30 minutes.7 They also reported that circulatory arrest time > 30 minutes was an independent predictor of stroke after ATAAD surgery.10 On the other hand, Pacini et al. compared hypothermia above or below 25 ° C in 305 ACP cases, but found no difference in PND (3.1% versus 1.7%) or TND (7.9% versus 8.6%).22 Zierer et al. demonstrated that an ACP time of 60 minutes or longer was not correlated with an increased risk of mortality and neurologic deficits even if the ACP was conducted under mild systemic hypothermia (28-30 °C).23 The reported incidence of postoperative PND after ATAAD surgery ranges widely from 2.9 to 18.9%.24 This considerable variation suggests that differences in clinical definitions or diagnostic imaging modalities might exist, which could be a potential reason for the inconsistent conclusions in prognostic factors of postoperative neurologic complications. The incidence of PND in our period 3 group, in which the uACP time was the shortest, was surprisingly more than double that in the other two periods. The more liberal usage of MRI in the recent era has potentially increased the diagnostic rate of PND. However, the incidence of TND did not decrease in period 3. Our findings suggest that reducing uACP time may not be an effective solution to improve postoperative neurologic outcomes.
In another ATAAD study comparing 140 uACP and 167 bACP cases, Norton et al. also failed to show that ACP time was a significant factor associated with new-onset postoperative stroke in their multivariable logistic analysis.6 Most stroke lesions in their cohort were on both sides (44% in uACP and 53% in bACP), and nearly all were embolic (100% in uACP and 87% in bACP). They suggested that thromboembolization due to significant intra-arterial instrumentation and intramural thrombus, global hypoperfusion due to unstable hemodynamics, and compromise of the true lumen of the carotid arteries might be more significant causes of postoperative stroke.6 The results of our logistic regression model for PND are consistent with their conclusions: mobilization of emboli in supra-aortic vessels with a total arch procedure and global or regional cerebral hypoperfusion from cardiac tamponade or preoperative brain malperfusion, instead of uACP time, were more influential on postoperative PND. This correlation was more apparent in our period 3. The incidence rates of brain malperfusion (10.1%), total arch replacement (15.6%), and PND (18.3%) were all highest in period 3 compared to the earlier periods (Table 1-3). Attempting to minimize cerebral emboli and preoperative hypoperfusion, rather than reducing uACP time, could be more critical to improve postoperative neurologic complications.
This study has several limitations. During the retrospective 15-year study period, there were substantial technical and methodological advances in our strategy for ATAAD. Classification of neurologic injury and recognition of malperfusion were highly dependent on the quality of the imaging modalities, which were not consistent throughout the whole study period. bACP was less common in our practice and was not compared directly in this study. The upper limit of temperature during circulatory arrest was 28 °C in our cohort. Therefore, the correlation between the grade of hypothermia and mortality and the expected safe time threshold of uACP could not be extrapolated to a higher temperature.
CONCLUSIONS
Surgical results of ATAAD have improved, but the optimal cerebral perfusion strategy continues to evolve. Longer unilateral antegrade cerebral perfusion time was associated with hospital mortality. However, postoperative permanent neurologic deficits were more correlated with preoperative cerebral hypoperfusion and embolism from arch manipulation. Based on the prediction of hospital mortality, a safe threshold of 50 to 60 minutes of uACP should be considered.
Acknowledgments
We acknowledge Ms. Chao-Rung Shih and Yu-Chin Huang for data collection. This work was supported by grants from (the Ministry of Science and Technology, Executive Yuan, Taiwan (grant numbers: 110-2314-B-006-104-to JNR and grant number: 104-2314-B-006-098 to MTT).
DECLARATION OF CONFLICT OF INTEREST
All the authors declare no conflict of interest.
REFERENCES
- 1.Tian DH, Wan B, Bannon PG, et al. A meta-analysis of deep hypothermic circulatory arrest versus moderate hypothermic circulatory arrest with selective antegrade cerebral perfusion. Ann Cardiothorac Surg. 2013;2:148–158. doi: 10.3978/j.issn.2225-319X.2013.03.13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Spielvogel D, Kai M, Tang GH, et al. Selective cerebral perfusion: a review of the evidence. J Thorac Cardiovasc Surg. 2013;145:S59–S62. doi: 10.1016/j.jtcvs.2012.11.073. [DOI] [PubMed] [Google Scholar]
- 3.Usui A, Miyata H, Ueda Y, et al. Risk-adjusted and case-matched comparative study between antegrade and retrograde cerebral perfusion during aortic arch surgery: based on the Japan adult cardiovascular surgery database: The Japan Cardiovascular Surgery Database Organization. Gen Thorac Cardiovasc Surg. 2012;60:132–139. doi: 10.1007/s11748-011-0857-2. [DOI] [PubMed] [Google Scholar]
- 4.Misfeld M, Leontyev S, Borger MA, et al. What is the best strategy for brain protection in patients undergoing aortic arch surgery? A single center experience of 636 patients. Ann Thorac Surg. 2012;93:1502–1508. doi: 10.1016/j.athoracsur.2012.01.106. [DOI] [PubMed] [Google Scholar]
- 5.Kruger T, Weigang E, Hoffmann I, et al. Cerebral protection during surgery for acute aortic dissection type A: results of the german registry for acute aortic dissection type A (GERAADA). Circulation. 2011;124:434–443. doi: 10.1161/CIRCULATIONAHA.110.009282. [DOI] [PubMed] [Google Scholar]
- 6.Norton EL, Wu X, Kim KM, et al. Unilateral is comparable to bilateral antegrade cerebral perfusion in acute type A aortic dissection repair. J Thorac Cardiovasc Surg. 2020;160:617–625 e615. doi: 10.1016/j.jtcvs.2019.07.108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Preventza O, Coselli JS, Akvan S, et al. The impact of temperature in aortic arch surgery patients receiving antegrade cerebral perfusion for > 30 minutes: how relevant is it really. J Thorac Cardiovasc Surg. 2017;153:767–776. doi: 10.1016/j.jtcvs.2016.11.059. [DOI] [PubMed] [Google Scholar]
- 8.Zierer A, Risteski P, El-Sayed Ahmad A, et al. The impact of unilateral versus bilateral antegrade cerebral perfusion on surgical outcomes after aortic arch replacement: a propensity-matched analysis. J Thorac Cardiovasc Surg. 2014;147:1212–1217; discussion 1217-1218. doi: 10.1016/j.jtcvs.2013.12.022. [DOI] [PubMed] [Google Scholar]
- 9.Tong G, Zhang B, Zhou X, et al. Bilateral versus unilateral antegrade cerebral perfusion in total arch replacement for type A aortic dissection. J Thorac Cardiovasc Surg. 2017;154:767–775. doi: 10.1016/j.jtcvs.2017.02.053. [DOI] [PubMed] [Google Scholar]
- 10.Preventza O, Simpson KH, Cooley DA, et al. Unilateral versus bilateral cerebral perfusion for acute type A aortic dissection. Ann Thorac Surg. 2015;99:80–87. doi: 10.1016/j.athoracsur.2014.07.049. [DOI] [PubMed] [Google Scholar]
- 11.Comas GM, Leshnower BG, Halkos ME, et al. Acute type A dissection: impact of antegrade cerebral perfusion under moderate hypothermia. Ann Thorac Surg. 2013;96:2135–2141. doi: 10.1016/j.athoracsur.2013.06.085. [DOI] [PubMed] [Google Scholar]
- 12.El-Sayed Ahmad A, Papadopoulos N, Risteski P, et al. The standardized concept of moderate-to-mild (≥ 28 degrees c) systemic hypothermia during selective antegrade cerebral perfusion for all-comers in aortic arch surgery: single-center experience in 587 consecutive patients over a 15-year period. Ann Thorac Surg. 2017;104:49–55. doi: 10.1016/j.athoracsur.2016.10.024. [DOI] [PubMed] [Google Scholar]
- 13.Lin TW, Tsai MT, Hu YN, et al. Postoperative extracorporeal membrane oxygenation support for acute type A aortic dissection. Ann Thorac Surg. 2017;104:827–833. doi: 10.1016/j.athoracsur.2017.01.059. [DOI] [PubMed] [Google Scholar]
- 14.Evangelista A, Isselbacher EM, Bossone E, et al. Insights from the international registry of acute aortic dissection: a 20-year experience of collaborative clinical research. Circulation. 2018;137:1846–1860. doi: 10.1161/CIRCULATIONAHA.117.031264. [DOI] [PubMed] [Google Scholar]
- 15.Conzelmann LO, Weigang E, Mehlhorn U, et al. Mortality in patients with acute aortic dissection type A: analysis of pre- and intraoperative risk factors from the german registry for acute aortic dissection type A (GERAADA). Eur J Cardiothorac Surg. 2016;49:e44–e52. doi: 10.1093/ejcts/ezv356. [DOI] [PubMed] [Google Scholar]
- 16.Benedetto U, Sinha S, Dimagli A, et al. Decade-long trends in surgery for acute type a aortic dissection in England: a retrospective cohort study. Lancet Reg Health Eur. 2021;7:100131. doi: 10.1016/j.lanepe.2021.100131. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Angeloni E, Benedetto U, Takkenberg JJ, et al. Unilateral versus bilateral antegrade cerebral protection during circulatory arrest in aortic surgery: a meta-analysis of 5100 patients. J Thorac Cardiovasc Surg. 2014;147:60–67. doi: 10.1016/j.jtcvs.2012.10.029. [DOI] [PubMed] [Google Scholar]
- 18.Urbanski PP, Babin-Ebell J, Frohner S, Diegeler A. Insufficient unilateral cerebral perfusion during emergent aortic arch surgery. Interact Cardiovasc Thorac Surg. 2012;14:122–124. doi: 10.1093/icvts/ivr042. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Krahenbuhl ES, Clement M, Reineke D, et al. Antegrade cerebral protection in thoracic aortic surgery: lessons from the past decade. Eur J Cardiothorac Surg. 2010;38:46–51. doi: 10.1016/j.ejcts.2010.01.016. [DOI] [PubMed] [Google Scholar]
- 20.Malvindi PG, Scrascia G, Vitale N. Is unilateral antegrade cerebral perfusion equivalent to bilateral cerebral perfusion for patients undergoing aortic arch surgery? Interact Cardiovasc Thorac Surg. 2008;7:891–897. doi: 10.1510/icvts.2008.184184. [DOI] [PubMed] [Google Scholar]
- 21.Tsai JY, Pan W, Lemaire SA, et al. Moderate hypothermia during aortic arch surgery is associated with reduced risk of early mortality. J Thorac Cardiovasc Surg. 2013;146:662–667. doi: 10.1016/j.jtcvs.2013.03.004. [DOI] [PubMed] [Google Scholar]
- 22.Pacini D, Leone A, Di Marco L, et al. Antegrade selective cerebral perfusion in thoracic aorta surgery: safety of moderate hypothermia. Eur J Cardiothorac Surg. 2007;31:618–622. doi: 10.1016/j.ejcts.2006.12.032. [DOI] [PubMed] [Google Scholar]
- 23.Zierer A, El-Sayed Ahmad A, Papadopoulos N, et al. Selective antegrade cerebral perfusion and mild (28 degrees c-30 degrees c) systemic hypothermic circulatory arrest for aortic arch replacement: results from 1002 patients. J Thorac Cardiovasc Surg. 2012;144:1042–1049. doi: 10.1016/j.jtcvs.2012.07.063. [DOI] [PubMed] [Google Scholar]
- 24.Pacini D, Murana G, Di Marco L, et al. Cerebral perfusion issues in type A aortic dissection. J Vis Surg. 2018;4:77. doi: 10.21037/jovs.2018.03.20. [DOI] [PMC free article] [PubMed] [Google Scholar]