Predictors of Outcomes after Correction of Acute Type A Aortic Dissection under Moderate Hypothermic Circulatory Arrest and Antegrade Cerebral Perfusion

Introduction Hypothermic circulatory arrest is widely used for correction of acute type A aortic dissection pathology. We present our experience of 45 consecutive patients operated in our unit with bilateral antegrade cerebral perfusion and moderate hypothermic circulatory arrest. Methods Between January 2011 and April 2015, 45 consecutive patients were admitted for acute type A aortic dissection and operated emergently under moderate hypothermic circulatory arrest and bilateral antegrade cerebral perfusion. Results Mean age was 58±11.4 years old. Median circulatory arrest time was 41.5 (30-54) minutes while the 30-day mortality and postoperative permanent neurological deficits rates were 6.7% and 13.3%, respectively. Unadjusted analysis revealed that the factors associated with 30-day mortality were: preoperative hemodynamic instability (OR: 14.8, 95% CI: 2.41, 90.6, P=0.004); and postoperative requirement for open sternum management (OR: 5.0, 95% CI: 1.041, 24.02, P=0.044) while preoperative hemodynamic instability (OR: 8.8, 95% CI: 1.41, 54.9, P=0.02) and postoperative sepsis or multiple organ dysfunction (OR: 13.6, 95% CI: 2.1, 89.9, P=0.007) were correlated with neurological dysfunction. By multivariable logistic regression analysis, postoperative sepsis and multiple organ dysfunction independently predicted (OR: 15.9, 95% CI: 1.05, 96.4, P=0.045) the incidence of severe postoperative neurological complication. During median follow-up of 6 (2-12) months, the survival rate was 86.7%. Conclusion Bilateral antegrade cerebral perfusion and direct carotid perfusion for cardiopulmonary bypass, in the surgical treatment for correction of acute aortic dissection type A, is a valuable technique with low 30-day mortality rate. However, postoperative severe neurological dysfunctions remain an issue that warrants further research.


Degree of Postoperative Neurological Complications
Major postoperative neurological dysfunctions were evaluated by a neurologist and serial brain CT. In addition, neurological dysfunctions were divided in the four sub-groups: 1) no neurological dysfunction; 2) temporary neurological dysfunctions (TND): TIA, delirium and disorientation (<24 hours after extubation); 3) permanent neurological dysfunctions (PND): hemiplegia or paraplegia (>48 hours) that persisted after discharge at home; and 4) heavy neurological deficits (diffuse and irreversible brain damage or coma). Patients discharged were followed-up in regular intervals in outpatients' clinic with echocardiography and, if required, chest and brain CT scan.

Statistical Analysis
Continuous variables are presented as means ± SD and categorical variables are expressed as percentages. Not normally distributed continuous variables are presented as medians (interquartile range). Univariate binary logistic regression analysis was performed to identify factors associated with postoperative mortality and/or neurological complications within 30 days from surgery. For 30-day mortality, the hospital length of stay and the presence of postoperative neurological complications were not evaluated as potential predictors to avoid selection bias (all subjects who died experienced postoperative neurological complications and presented less time of hospitalization). For postoperative neurological complications, where adequate number of events was yielded (n=9), significant univariate predictors were further incorporated into the final multivariable logistic regression model. To address the small sample size of our study, we implemented exact logistic regression and resampling techniques. Exact logistic regression produces more accurate inference in small samples because it does not depend on asymptotic results and conditional maximum probability estimates were sequentially calculated for each predictor in multivariable logistic models by temporarily conditioning out the other independent variables. Bootstrapping with 1000 replications was conducted to replicate bias-corrected confidence intervals of the significant determinants of the outcomes on interest in univariate and multivariable regression models. However, due to the limited number of observations, the reported effect sizes for certain variables are still characterized by wide confidence intervals.
Statistical analysis was conducted using STATA package, version 11.1 (StataCorp, College Station, Texas, USA). We deemed statistical significance at P<0.05.

RESULTS
The cohort consisted of 34 male and 11 female patients. Mean age was 58±11.4 years. In 7 patients, haemodynamic instability and in 2 patients minor neurological deficits were noted preoperatively. Preoperative data and baseline demographic characteristics are shown in Table 1.
CPB was instituted with: right or left carotid (connecting through a synthetic 10-mm graft)/bicaval cannulation in 42 patients and femoral artery/bicaval cannulation in 3 patients. Combined procedures were performed in 12 patients. Median

INTRODUCTION
Hypothermic circulatory arrest (CA) is widely used for correction of acute type A aortic dissection (AAD). In an attempt to establish a bloodless surgical field and ameliorate brain protection, which is more vulnerable in this case, a variety of techniques for cardiopulmonary bypass (CPB) and cerebral perfusion during CA have been suggested over the years [1][2][3] .
We present our experience of 45 consecutive patients who underwent AAD correction with moderate hypothermic CA and bilateral antegrade cerebral perfusion (MHCA/BACP) via the common carotid arteries.

Study Population
Between January 2011 and April 2015, 45 consecutive patients underwent an emergency operation for type A AAD (with or without aortic root and valve involvement) with MHCA/BACP. The preoperative diagnosis was established with computed tomography (CT) of thoracic and abdominal aorta, including aortic arch branches. All preoperative, perioperative and postoperative data were recorded.
Two patients were operated with preoperative minor neurological dysfunction [transient ischemic attack (TIA)].
This study was carried out according to the principles outlined in the Declaration of Helsinki. It is a retrospective analysis approved by the hospital's institutional ethics committee (546/30-04-2015) and all patients gave their informed consent prior to the operation.

Surgical Technique
All patients underwent an emergency surgical correction of AAD within 24 hours of admission. The briefly applied surgical technique consisted of standard median sternotomy and CPB via: 1) right or left carotid artery (connecting through a 10-mm synthetic graft)/ bicaval cannulation in 42 patients and 2) femoral artery/bicaval cannulation in 3 patients. Gradual cooling of the body (1°C/5 min) and alpha-stat (pH management) was utilized. Myocardial arrest and protection was achieved with retrograde delivery of histidinetryptophan-ketoglutarate (HTK) crystalloid solution (Custodiol ®).
Once CPB was instituted, the aortic cross-clamp was applied, the cooling initiated and, with the heart stopped, correction of the proximal ascending aorta ± aortic root was undertaken, including any other required correction.
When bladder temperature reached ≈ 23°C, the head was packed in ice and the circulation, except for the carotids, was interrupted. CPB and antegrade cerebral perfusion (ACP) were achieved using synthetic graft (10 mm) in the right or left common carotid artery (CCA) and selective cannulation of the contralateral CCA. Where CBP was established with femoral artery/bicaval cannulation. ACP was performed with selective cannulation of both CCA. The ACP blood flow was 10 mL/kg/min (perfusion pressure around 50-70 mmHg, blood temperature ≈20°C). The distal anastomosis in the proximal or middle aortic arch was performed. After heart de-airing and gradual rewarming, the patients were weaned from CPB.

Arterial Cannulation Sites
The optimal arterial cannulation site during proximal aortic arch surgery has been widely discussed over the years. Direct cannulation of the ascending aorta in patients who have an aortic aneurysm is an accepted technique and supported by many authors [8] . On the other hand, in cases with acute and chronic aortic dissection, the approach of arterial cannulation is controversial. Many techniques have been suggested in these cases: ascending aortic cannulation, right axillary or subclavian artery, right or left carotid artery, right or left common femoral artery and even transapical aortic cannulation [9][10][11][12][13][14] . These studies have demonstrated a relatively similar 30-day mortality regardless of the cannulation technique, while the incidence of stroke varied widely between 3.8 and 21%. In 2010, Tiwari et al. [15] , after analysis of several studies, concluded that ascending aortic cannulation has promising results with a lower mortality rate, P=0.045) with the incidence of severe postoperative neurological complications (Table 6) independently of other significant univariate predictors (i.e. bladder temperature during circulatory arrest and presence of preoperative hemodynamic instability).

DISCUSSION
Publications of series on operations in the ascending aorta and hemiarch continue to discuss advantages and disadvantages amongst different techniques particularly in AAD pathology [1][2][3][4][5][6][7] . However, there is unanimity in that the most significant concern during this type of operations is the protection of the central nervous system (CNS) during the interval of hypothermic CA. The main issues that need to be addressed in these complex procedures are route of CPB establishment (site of arterial cannulation), core temperature management during operation and type of cerebral perfusion during CA with or without aortic arch repair. Postoperative in-hospital stay (days), median (IQR) 12 (8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18) Follow-up (months), median (IQR) 6 (2-12) Overall mortality 6 (13.3) 30-days mortality 3 (6.7) All cause deaths during median follow-up 6 (2-12) months 3 (6.7) Data are presented as mean ± SD or median (IQR) for continuous variables and as number and percentage for categorical variables. Permanent neurological dysfunction (PND)=hemiplegia or paraplegia >48 hours after discharge at home. Heavy neurological deficits=diffuse and irreversible brain damage or coma. ICU=intensive care unit; IQR=interquartile range NT-proBNP=N-terminal prohormone of brain natriuretic peptide; ICU=intensive care unit but a higher stroke rate in type A aortic dissections. In our study, the incidence of PND and 30-day mortality was 13.3% and 6.7%, respectively. Although the numbers of patients in our study are too small to draw safe conclusions, the survival rate seems to be very good, with good survival in the medium term. Neurological complications, however, according to the above remain an issue.

Temperature Management during Hypothermic Circulatory Arrest
The goals of hypothermia during thoracic aortic surgery are reduction of brain metabolism and attenuation of CNS damage during CA. In 1975, Griepp et al. [16] described four patients with successful outcome who underwent aortic arch replacement with prosthetic graft under deep hypothermic circulatory arrest (DHCA) with lowest esophageal temperature at 14°C and lowest rectal temperature at 18°C. Decreased mortality and neurological complications were achieved with cerebral perfusion during CA. Deep hypothermic circulatory arrest with retrograde cerebral perfusion (DHCA/RCP) (18°C) allowed longer CA times and improved brain protection [17] . In 1991, Bachet et al. [18] published the technique with MHCA/ACP during transverse aortic arch repair (core temperature 25-28°C and brain perfused with blood cooled at 6-12°C). This method created a favorable circumstance to perform more complex operations in the aortic arch, with simultaneous improvement of neurological and overall outcomes.
In 2011, the analysis of 1558 patients (GERAADA study) by Krüger et al. [19] noted that relative increase in mortality was observed in patients with hypothermic CA (<15°C) alone, even more so when CA arrest time exceeded 30 min. Estrera et al. [20] used hypothermic CA (nasopharyngeal temperature 15-20°C) with or without retrograde cerebral perfusion (RCP) and stroke rate and 30-day mortality were 2.3% and 10.4%, respectively. On the other hand, Urbanski et al. [21] analyzed 347 patients who underwent nonemergent arch surgery under MHCA with rectal temperature (28-34°C) utilizing ACP. Their 30-day mortality and overall postoperative neurological dysfunction were 0.9% and 3.2%, respectively. Perreas et al. [22] , in a retrospective analysis of 208 patients operated on with DHCA/RCP (temperature range 11-24°C), concluded that the core temperature within the specific range was not a risk factor for 30day mortality and severe neurological events. In another recent study, Zierer et al. [4] noted that increase of the core temperature (28-30°C) during CA with implementation ACP allowed more time (CA>90 min), more complex corrections of ascending aorta and aortic arch pathology with low incidence of postoperative complications (new postoperative neurological deficits were 7%). Leshnower et al. [23] analyzed 500 patients who underwent a hemiarch replacement under mild (28.6°C) vs. moderate (24.3°C) hypothermic CA with unilateral ACP with operative mortality of 4.2% vs. 4.8% (P=0.80) and no differences in TND between the two groups. However, the incidence of PND was reduced in mild vs. moderate hypothermia (2.5% vs.7.2%) (P=0.01). In hemiarch with or without total aortic arch replacement, with or without stent grafting, the question remains: which method of brain protection is most effective and safe during these complex operations, particularly in patients with AAD?

Type of the cerebral protection and perfusion
Usui et al. [1] analyzed 2792 patients and found no differences between the ACP and RCP groups in 30-day mortality (3.4% vs. 2.4%) and stroke rate (5% vs. 3%), but in the subgroup with RCP higher incidence of transient neurological dysfunction (5.8%) was observed. Misfeld et al. [2] divided 636 patients who underwent aortic arch surgery in four groups: UACP, bilateral ACP (BACP), DHCA/RCP and DHCA only. The study showed that early mortality and five-year survival were not different between the surgical groups, but stroke rate was different in patients who did not receive ACP (P=0.035). Urbanski et al. [21] presented results of non-emergent aortic arch surgery using mild to moderate hypothermic CA (31.5±1.6°C) with ACP (blood temperature 28°C) in 347 patients. The results show that 30-day mortality was 0.9%, and PND and TND observed in 0.9% and 2.3%, respectively. On the other hand, Estrera et al. [20] reported the study with 1107 patients who operated under DHCA/RCP in 82% of cases, and the results were 30-day mortality and stroke rate occurred 10.4% and 2.8%, respectively. After comparative analysis of 1558 patients with AAD, Krüger et al. [19] concluded that 30-day mortality was higher in the DHCA group (19.4%) than in the BACP (15.9%) and UACP (13.9%) groups (P<0.05). The same study noted that PND were: DHCA-14.9%, BACP-14.1% and UACP-12.6%. The most recent results from the Japanese AAD database (2016) repair with ACP vs. RCP did not show significant differences regarding mortality and postoperative neurological dysfunctions rates (11.2% vs. 9.7%) [3] . Nowadays, many studies discuss the incidence of stroke and mortality rate after AAD repair with hemiarch versus hemiarch plus total arch replacement (with or without antegrade stent grafting) [3,[24][25][26] . After analysis, the results show that the rates of mortality and stroke are similar between the two groups. However, it is possibly relevant regarding false lumen thrombosis rate and reduction of late reoperation rate in aortic arch and descending aorta. In 2015, Di Bartolomeo et al. [26] , in a review of AAD type A repair with frozen elephant trunk technique, noted that mortality and postoperative stroke range were 0-27.7% and 0-12%, respectively. In addition, in the same study, the authors revealed that the spinal cord injury rate ranged from 0 to 13.8%. It is evident that postoperative neurological complications range in the literature from 0 to 39.7%.

CONCLUSION
In aortic arch surgery for AAD, the total CA can be avoided by maintaining cerebral perfusion. Thus, continuous cerebral perfusion via UACP or BACP resulted in a reduction of postoperative mortality. However, the postoperative severe neurological complications remain at high frequency, requiring further refinement of this technique. Reduction of these major postoperative complications should be explored in multicenter studies.

Limitations
This is clearly a retrospective analysis with a small cohort, albeit with consecutive patients from a single unit. Two patients with preoperative TIA were included in our study with possible impact on severe postoperative neurological complications. Furthermore, most patients with AAD were operated emergently without accurate preoperative neurological assessment and without preoperative brain CT. On the other hand, more extensive follow-up is required to demonstrate potential improvements in the long-term outcomes.