Evaluation of risk factors for transient neurological dysfunction and adverse outcome after repair of acute type A aortic dissection in 122 consecutive patients Free

Abstract

OBJECTIVES

The aim of this retrospective study was to assess pre- and intraoperative factors leading to neurological complications and early death following repair of acute type A aortic dissection (ATAAD).

METHODS

There were 122 patients (85 male, age: 58.6 ± 12.5 years) with ATAAD, treated consecutively from August 2003 to August 2010. Pre- and intraoperative variables were analysed using a logistic regression model in order to identify risk factors for temporary neurological dysfunction (TND) and adverse outcome (AO), defined as stroke and 30-day mortality.

RESULTS

The 30-day mortality rate was 16.4%. Forty-one patients (33.6%) suffered transient neurological dysfunction and 20 (16.4%) had a postoperative stroke. Mean hypothermic circulatory arrest (HCA) temperature was 24 ± 4°C. Selective cerebral perfusion (SCP) was performed in 99 (82%) patients, with a mean SCP flow rate of 10.3 ml/kg/min. The duration of lower body ischaemia (LBI) was 36 ± 27 min, of HCA 8.7 ± 15.5 min and of SCP 34 ± 28 min, respectively. Male gender [odds ratio (OR): 3.30, 95% confidence interval (CI): 1.15–9.47], diabetes (OR: 3.95, 95% CI: 1.18–13.24), compromised consciousness (OR: 6.65, 95% CI: 1.41–31.48) and manifest arterial atherosclerosis (OR: 6.68, 95% CI: 1.31–34.09) were detected as risk factors for TND, whereas a high body mass index (OR: 1.14, 95% CI: 1.01–1.3), a preoperative malperfusion syndrome (OR: 2.28, 95% CI: 0.84–6.18) and left ventricular ejection fraction <50% (OR: 3.84, 95% CI: 1.41–10.43) were detected as independent predictors for an AO. A dissection entry localized in the aortic arch or the descending aorta independently increased the risk for a postoperative stroke. A prolonged LBI increased the risk for AO (OR: 1.02, 95% CI: 1.00–1.04), whereas femoral cannulation showed a trend to an increased stroke incidence (OR: 4.2, 95% CI: 0.8–21.3).

CONCLUSIONS

Regardless of standardized neuroprotective techniques, treatment of ATAAD remains a high-risk operation. Preoperatively, the presence of a reduced ejection fraction, a malperfusion syndrome or a high body mass index may increase the perioperative risk for an adverse outcome. A dissection ‘entry’ localized in the aortic arch or the descending aorta may increase the risk for postoperative stroke. Intraoperatively, cannulation of the femoral artery and extension of the LBI time over 45 min should be avoided. Especially in patients with manifest preoperative cerebral and/or end-organ malperfusion, the cannulation modality as well as the entire neuroprotective management should be chosen individually, respecting its limitations.

INTRODUCTION

Although the postoperative outcome after surgical treatment of acute type A aortic dissection (ATAAD) has been dramatically improved by neuroprotective strategies such as hypothermic circulatory arrest (HCA), selective cerebral perfusion (SCP) and individual arterial cannulation modalities, neurological complications in terms of transient neurological dysfunctions (TND) or permanent stroke, as well as high perioperative mortality represent the major complications of these procedures [1, 2]. Often the outcome is related to the preoperative condition of the patient. The difficulty—especially in an emergency situation—represents the adequate risk stratification in the preoperative algorithm, the choice of the cannulation modality and the individual neuroprotective management. The aim of the present study was to assess pre- and intraoperative factors that lead to neurological complications and early death following ATAAD. In addition, a main focus of this investigation was to identify the ideal cannulation modality and to quantify the limitations of the neuroprotective methods in use.

MATERIALS AND METHODS

One hundred and twenty-two consecutive patients with ATAAD underwent emergency surgery in the Department of Cardiothoracic Surgery at the University of Cologne between August 2003 and August 2010. All patients were treated surgically following a standardized procedure. After arterial and venous cannulation, cooling was started on cardiopulmonary bypass (CPB) until the desired hypothermic temperature was reached and HCA and/or SCP were introduced. After aortic replacement, systemic rewarming and weaning from CPB, the patients were transported for surveillance to the intensive care unit (ICU). Surviving patients were followed for at least 30 days postoperatively. All documentation was done by a third party, blinded to the statistical analyses, using the clinical database (QIMS).

The analysed cohort included 85 men and 37 women, with a mean age of 58.6 ± 12.5 years (range: 29–84 years) and a mean body mass index (BMI) of 26.2 ± 3.9 kg/m² (range: 18–40 kg/m²). More than 60% of the patients had a preserved left ventricular ejection fraction (LVEF) and adequate renal function. Preoperatively assessed risk evaluation showed a mean standard EuroSCORE of 10 ± 3.3 (range of 8–20). Fifteen preoperative characteristics, including demographic data and preoperative morbidities, were considered for the risk factor analysis (Table 1).

Operative procedures

All cases were performed as urgent or emergency approaches. Intraoperatively, in 78.7% of the cases a dissection entry was seen in the ascending aorta, in 11.5% the entry was located in the proximal aortic arch, in 4.9% in the distal arch and in 6.5% the entry was not found but presumed to be in the descending aorta. The chosen cannulation sites, the extent of the aortic replacement and the performed aortic valve surgery procedures are presented in Table 2.

The mean values for the operative periods/duration are shown in Table 3. Lower body ischaemia (LBI) time was defined as the entire period during which the aortic arch and the descending aorta were not perfused. This could be the interval of isolated HCA, the combination of a short interval of HCA and SCP, or the period of primary unilateral SCP (perfusion through the clamped right brachiocephalic trunk) followed by a bilateral SCP (completed bi-hemispheric perfusion).

Operations and techniques

Three different modalities for arterial cannulation were used, as follows: femoral or axillary cannulation with an 18–20 French arterial cannula (Fem-Flex II; Edwards Lifescience, Irvine, CA, USA) or direct aortic cannulation. During the axillary approach, if a pronounced atherosclerosis or otherwise damaged axillar artery was seen intraoperatively, cannulation was done indirectly, through a previously attached prosthetic Dacron graft. Venous drainage was completed with a 34–39 French double-stage cannula (Maquet Cardiopulmonary AG, Rastatt, Germany) inserted via the right atrium. After implementation of the CPB, systemic perfusion with a pump flow index of 2400 ml/min/m² body surface area was started in order to maintain a perfusion pressure of 50–75 mmHg. Acid–base balance was preserved with the alpha-stat strategy.

Systemic cooling was performed until the desired hypothermic temperature was achieved. The coldest temperature, targeted for HCA as a ‘stand-alone’ neuroprotective strategy, was 18°C. If SCP was performed, cooling was limited to moderately hypothermic levels (25–28°C).

For HCA without SCP, the patient was set in the Trendelenburg position, the systemic perfusion stopped and the aortic arch opened. During the opened aortic arch procedure, CO₂ insufflation into the surgical field was carried out in order to prevent air embolization. If a short HCA interval was followed by a more extended SCP period, two 15 French cannulae with a self-inflating cuff were inserted under direct vision into the innominate and left common carotid arteries. SCP was started with a pump flow rate of 10 ml/kg/min, and adjusted to maintain a mean arterial pressure of 40–60 mmHg at the tip of the self-inflating cuff.

Neuroprotection strategy and cannulation sites

In the early period (August 2003 until May 2004), a deep hypothermic circulatory arrest (at 18°C) was used as a ‘stand-alone’ neuroprotective strategy, without SCP. After April 2004, a short period of HCA was primarily used for inspection of the aortic arch, which was followed by an extended period of SCP at more moderate hypothermic temperatures (20–25°C) for the open distal anastomosis. From October 2006 onwards, the cannulation policy switched from direct aortic or femoral cannulation to an axillary cannulation. In order to accomplish this change, the switch from total CPB to SCP was possible without an intermediate HCA period. Between 2007 and 2010, most operations were done following this new concept of axillary cannulation, moderate hypothermia (26–28°C), clamping of the brachiocephalic trunk and starting of a uni-hemispheric SCP via the right carotid artery. Once the aortic arch was opened, cerebral perfusion was completed to a bi-hemispheric SCP by introducing a second arterial cannula into the left common carotid artery.

Definition of end-points and statistical analysis

Primary end-points of our study were the appearance of TND or an adverse operative outcome. TND was defined as presence of prolonged postoperative delirium, disorientation and Parkinson-like movements and/or persistent loss of cognitive function, with no pathological correlate in CT scan and complete resolution before discharge. Adverse outcome (AO) was defined as permanent stroke and/or death during the first 30 days after surgery.

Statistical analysis was performed using SPSS 18 software (SPSS Inc., Chicago, IL, USA). Continuous data are presented as the mean ± standard deviation or as the median with the interquartile range; categorical data are presented as percentages unless otherwise specified. Statistical significance was determined with Fisher's exact test for categorical variables and the Mann–Whitney U-test for continuous variables. Significance was defined at a P-value < 0.05. Significant pre- and intraoperative factors were further analysed in a multivariate logistic regression model in order to find risk factors for TND and AO. A separate multivariate logistic regression model was used to identify isolated risk factors for stroke. Parameters that showed an effect on mortality in the univariate analysis were included in a multivariate model.

RESULTS

Operative outcome

Intensive care unit stay was 9.6 ± 12.3 days (median 5 days). The intubation period lasted 6.1 ± 11 days (median 2 days). Postoperative non-neurological complications were persistent haemorrhage (66%), systemic inflammatory syndrome (52%), low cardiac output (51%), prolonged respiratory weaning (32%), requirement for renal dialysis (30%) and transient or permanent/final visceral ischaemia (11%). In 33 cases (27%) a re-exploration for persistent bleeding was necessary. AO was documented in 31 patients (25%). The 30-day mortality rate, including operative deaths, was 16.4%. Forty-one patients (33.6%) suffered TND and 20 (16.4%) had a postoperative stroke.

Correlation between pre- and intraoperative characteristics with temporary neurological dysfunction

From the investigated preoperative characteristics, the univariate analysis revealed a correlation of two factors with postoperative temporary neurological dysfunction: the mental status at hospital admission (P = 0.022) and the prevalence of a manifest peripheral arterial disease (P = 0.05). The patient’s gender (P = 0.09) as well as manifest diabetes (P = 0.088) were also found suitable for inclusion into the multivariate model. None of the six investigated intraoperative risk factors (operation type, cannulation strategy, cooling temperature, entry localization, cross-clamp time and total LBI time) showed any correlation with the postoperative prevalence of TND. The multivariate analysis identified a poor preoperative mental status (P = 0.027), male gender (P = 0.027), diabetes (P = 0.026) and manifest peripheral arterial disease (P = 0.022) as preoperative risk factors for TND.

Correlation between pre- and intraoperative characteristics with adverse outcome

The univariate analysis of preoperative risk factors for AO revealed a correlation of BMI, preoperative mental status, malperfusion syndrome, left ventricular ejection fraction and EuroSCORE with postoperative stroke rate and 30-day mortality (Table 4).

The following multivariate analysis identified only the following three preoperative characteristics as independent predictors for an AO: BMI, presence of malperfusion and a reduced LVEF. After the addition of the only independent intraoperative risk factor, the LBI time (P = 0.03), into to the existing multivariate model, the effect of ‘malperfusion’ turned out not to be significant (P = 0.106). Nevertheless, given that it was found to be significant in the multivariate analyses of preoperative data, the characteristic ‘malperfusion’ was kept in the final logistic regression model (Table 5).

Documentation of isolated HCA, SCP and LBI times was 99.2% complete. Analysis of the duration of LBI time and the postoperative outcome revealed an increased risk [unadjusted odds ratio (OR) of 2.05; 95% confidence interval (CI): 0.82–5.13] for an AO in patients with an LBI time >45 min (10 patients with AO of 27 patients with a LBI time >45 min; 37%) when compared with those who underwent a shorter LBI (21 patients with AO of 94 patients with a LBI time ≤45 min; 22%).

Separate analysis of risk factors for stroke

The multivariate logistic regression analysis identified in a univariate analysis the preoperative factor EuroSCORE (P = 0.028) as well as the intraoperative factors aortic cross-clamp time (P = 0.04) and LBI time (P = 0.012) as possible risk factors for postoperative stroke. The multivariate analysis identified the variables EuroSCORE (P = 0.042, OR: 1.158, 95% CI: 1.00–1.34) and aortic cross-clamp time (P = 0.029, OR: 1.014, 95% CI: 1.00–1.03) as risk factors for a postoperative stroke.

Cannulation strategy and stroke incidence

Forty per cent of patients in whom femoral cannulation was performed suffered a postoperative stroke, whereas a direct aortic cannulation resulted in a postoperative stroke in only 20%. The lowest stroke incidence was registered after direct axillary/subclavian cannulation (12%) or indirect axillary cannulation (through a Goretex prothesis, 0%). The univariate analysis revealed femoral cannulation as a risk factor for a stroke (P = 0.022). Included in a multiple logistic regression model, together with the parameters EuroSCORE and hypothermic temperature, the femoral cannulation had a 4.2 times higher stroke risk than axillary cannulation (P = 0.086, OR: 4.2, 95% CI: 0.8–21.3), although this trend did not achieve significance. The degree of hypothermia used had no influence on the stroke rate (Table 6).

Separate analysis of the 30-day mortality

The 30 day survival rate was 83.6%. Most of the patients who did not survive (n = 20) the postoperative 30 day period died during the first 7 days postoperatively; six patients died intraoperatively, whereas three died on day 24 and three on day 29. The causes of death were low cardiac output in 10 patients (50%), haemorrhage in three patients (15%), sepsis in four patients (20%), multiple organ failure in two patients (10%) and visceral ischaemia in one patient (5%). Of the investigated variables, the univariate Cox regression identified the logistic EuroSCORE (P < 0.001), the entry localization (P = 0.01) and the cannulation policy (P = 0.038) as having a possible influence on the 30-day mortality. The multivariate analysis revealed a dissection entry localized in an aortic segment other than the ascending part (P = 0.018, hazard ratio 3.25, 95% CI: 1.23–8.59) and a high EuroSCORE (P < 0.001, hazard ratio 1.056, 95% CI: 1.04–1.08) as covariate risk factors for mortality.

DISCUSSION

Neuroprotection with the use of HCA and SCP has gained primary importance in the surgical treatment of ATAAD over the last 20 years. Nevertheless, neurological complications still occur in terms of TND—with an incidence of 2.5–16% after SCP and 33% after isolated HCA—and permanent stroke, with a reported incidence of 4–24% [3–6].

As there are a multitude of pre- and intraoperative factors that may influence the postoperative outcome, we undertook this retrospective study on an unselected consecutive series of 122 patients with ATAAD in order to identify influenceable and non-influenceable risk factors for a poor outcome. As others did before [7], we combined permanent neurological injury (stroke) and death as adverse outcome to increase the sensitivity for identifying risk factors. As described by Ergin et al. [5], we defined TND as symptoms of prolonged postoperative obtundation, disorientation, persistent loss of cognitive function with no correlate on CT scan and complete resolution before discharge. Our 30-day mortality rate and incidence of TND were comparable to the results of others [3, 8, 9], but our stroke rate of 16.4% may be a matter of concern, even if the literature provides a large range of reported stroke rates between 4 and 24% [6].

As we suspected, there was a strong correlation between the non-influenceable preoperative morbidity and a poor postoperative outcome. We identified a poor preoperative mental status, characterized as unconsciousness or deep sedation at admission, persistent insulin-dependent diabetes and manifest peripheral arterial disease as preoperative risk factors for TND. The reason why the men have a 3.4 times higher risk than women of developing a postoperative TND is still unclear. It may need further clinical investigations to clarify this topic.

The EuroSCORE, initially developed for risk calculation for coronary revascularization, demonstrated an adequate predictive capacity in our ATAAD patient cohort. A 1% increase of the logistic EuroSCORE raised the mortality risk by a factor of 1.056 and the stroke risk by a factor of 1.158. A body mass index over 28 kg/m², the presence of malperfusion and a LVEF < 50% were identified as risk factors for an AO. Similar risk factors were reported by Chiappini et al., who investigated a cohort of 487 patients with ATAAD [8]. Surprisingly, haemodynamic instability or preoperative shock, the integrity of the aortic valve and the coronary ostia, or the extension of the aortic dissection had no influence of the postoperative outcome. This is in contrast to the findings of Ehrlich et al. and Chiappini et al., who found in two separate studies that haemodynamic instability was an independent risk factor for an adverse outcome [7, 8]. In our model, the dissection localization (entry), and not the extension, increased the 30-day mortality risk, with a hazard ratio of 3.25.

Looking at the more influenceable intraoperative risk factors, we found a strong correlation between a prolonged LBI time and an AO. It is known from the literature that an HCA interval (without additional SCP) exceeding 25 min is—even at deeply hypothermic temperatures—detrimental for the postoperative neurological outcome. After 40 min of HCA the incidence of TND and stroke increases, especially in elderly patients [10]. SCP is considered to provide safer neuroprotection for more extended operations; because of the more physiological perfusion and less hypothermic conditions (up to 30°C), the cerebral vessels may retain part of their autoregulation and the brain is generally not ischaemic. Nevertheless, the Hannover group reported a higher risk of paraplegic outcome after a prolonged LBI period (>60 min) in a series of 76 patients with ATAAD and aortic arch replacement [11].

The mean LBI duration (including the HCA and/or SCP time) in our cohort was comparable with the LBI time reported by others [11, 12]. Even though our neuroprotective strategies differed over time, with the use deep hypothermic circulatory arrest in the earlier period and a more moderate hypothermia with SCP in the later period, the analysis revealed no correlation between the different strategies, cooling temperature and the postoperative outcome. Only a prolonged LBI period exceeding 45 min was identified as a risk factor for AO. An extension of the LBI time for 1 min increased the risk for an AO by a factor of OR = 1.02. An additional 10 min of LBI increased the risk for an AO by a factor of OR = 1.22.

The most influenceable variable for the neurological outcome may be the cannulation policy. Our analysis revealed a trend to an advantage of an axillary cannulation regarding the postoperative stroke rate. Similar findings were published earlier by Sabik et al. and more recently by Numata et al. and Strauch et al., who showed that direct axillary artery cannulation and additional selective cannulation of the left carotid artery, with clamping of the left subclavian artery, reduces the incidence of transient and permanent neurological dysfunction [13–15]. In our study, the femoral cannulation resulted in a 4.2 times higher stroke risk, although the difference did not achieve the level of significance. In a larger cohort of patients with ATAAD, the differences between the different cannulation strategies could be more pronounced, because it is known that femoral cannulation may lead to retrograde embolization, especially because of the non-physiological retrograde blood flow and the often calcified or atherosclerotic diseased vessels. In aortic dissections reaching the abdominal aorta, the iliac and femoral arteries, there is in addition a risk of false lumen perfusion and postoperative ischaemia of the lower extremities.

CONCLUSION

Regardless of standardized perfusion and neuroprotective techniques, treatment of ATAAD remains a high-risk procedure. Although early death and permanent stroke represent the worst complications of the surgical treatment, in our opinion, the occurrence of TND has also to be considered critically. In the investigated cohort, a compromised consciousness at admission, insulin-dependent diabetes, manifest arterial occlusive disease and the male gender correlated with TND, whereas a reduced left ventricular ejection fraction, the presence of a malperfusion syndrome and a high BMI increased the risk for an AO. A dissection entry with located other than in the ascending aorta (arch or descending aorta) had a higher risk for postoperative stroke. The statistical model revealed that an extension of the LBI time over 45 min may increase the risk for an AO, whereas a femoral cannulation policy could influence the stroke rate. These findings should encourage clinicians to look more critically at the neurological status and the presence of a clinically manifest malperfusion syndrome in patients with ATAAD. In these patients in particular, the cannulation modality as well as the entire neuroprotective management should be chosen individually, respecting its existing limitations.

Conflict of interest: none declared

REFERENCES