- Email: [email protected]
Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
ORIGINAL ARTICLE
Original Article
Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience Reece A. Davies, BAppSc (Phty) (Hons) b,d , Deborah Black, PhD b , Richmond W. Jeremy, PhD, FRACP, FCSANZ b,c , Paul G. Bannon, PhD, FRACS a,b,d , Matthew S. Bayfield, FRACS a,d , P. Nicholas Hendel, FRACS a,b,d , Clifford F. Hughes, AO, FRACS a,b,d , Michael K. Wilson, FRACS a,d and Michael P. Vallely, PhD, FRACS a,b,d,∗ a
Cardiothoracic Surgical Unit, Royal Prince Alfred Hospital, Sydney, Australia b Faculty of Medicine, The University of Sydney, Sydney, Australia c Department of Cardiology, Royal Prince Alfred Hospital, Sydney, Australia d The Baird Institute for Heart and Lung Surgical Research, Sydney, Australia
Background: Aortic arch replacement is a complicated and high risk procedure. There have been many advances over recent years. We review the changes in our unit’s techniques and outcomes over the past 22 years. Methods: Data were collated from databases and medical records for all patients who underwent aortic arch replacement surgery from January 1989 to December 2010. The patients were divided into two groups – Group A (1989–2005) and Group B (2006–2010). Data were analysed to compare early and late series patients’ outcomes. Logistic regression was used to identify variables that predicted mortality. Results: Seventy-five eligible patients (56 males; mean age: 57.5 years; Group A: 40, Group B 35) were identified. There were great changes in the technique and the methods of cerebral protection. The overall mortality rate was 30.7% – Group A: 50% and Group B: 8.6% (p < 0.001). Overall permanent neurological dysfunction was 23.7% – Group A: 40% and Group B: 11.8% (p = 0.012). Cardiovascular disease and circulatory arrest time were significant predictors of mortality. Conclusions: Increased experience and volume and advances in techniques over 22 years have resulted in major improvements in outcomes for patients having aortic arch replacement, allowing the procedure to be performed with greatly improved outcomes. (Heart, Lung and Circulation 2011;20:704–711) © 2011 Australasian Society of Cardiac and Thoracic Surgeons and the Cardiac Society of Australia and New Zealand. Published by Elsevier Inc. All rights reserved. Keywords. Thoracic aorta; Deep hypothermia induced circulatory arrest; Thoracic aortic aneurysm; Blood vessel prosthesis implantation; Aortic arch replacement
Introduction
A
ortic arch replacement is a complicated and relatively high risk operation primarily used in the treatment of aneurysms and/or dissections involving the aortic arch. The first successful arch replacement was reported by DeBakey et al. in 1957 after previously unsuccessful attempts [1]. Mortality in an early series was 42% [2], however with advancements in technique, especially in the field of cerebral protection, mortality and morbidity rates have decreased [3,4]. Deep hypothermic circulatory arrest (DHCA) was introduced and shown to be effective in proReceived 13 May 2011; accepted 22 July 2011; available online 26 August 2011 ∗
Corresponding author at: C/- Sydney Cardiothoracic Surgeons, 304/100 Carillon Avenue, Newtown, NSW 2042, Australia. Tel.: +61 295501933; fax: +61 295506669. E-mail address: [email protected] (M.P. Vallely).
tecting the brain [5]. However, further research showed the risk of stroke increased significantly after periods of arrest exceeding 40 minutes and mortality increased significantly after 65 minutes [6]. The two main techniques to overcome this have been retrograde cerebral perfusion (RCP) and antegrade cerebral perfusion (ACP) [7]. RCP was introduced in the 1980s with initially satisfactory results, including a mortality rate of 10–19.7% and a 4–11.8% rate of permanent neurological deficit [8,9]. However others have questioned its effectiveness in adequately perfusing the brain [10,11]. A prospective controlled trial has shown significantly higher rates of transient neurological dysfunction with RCP compared with ACP [12]. ACP, which was used in DeBakey’s original operation but subsequently fell from favour, has enjoyed a resurgence, having been shown to produce significant reductions in mortality and morbidity, especially neurological, compared to both RCP and DHCA alone [3]. ACP has increasingly become the primary form of cerebral protection used in complex
© 2011 Australasian Society of Cardiac and Thoracic Surgeons and the Cardiac Society of Australia and New Zealand. Published by Elsevier Inc. All rights reserved.
1443-9506/04/$36.00 doi:10.1016/j.hlc.2011.07.009
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
operations to replace the aortic arch and is seen by many surgeons as ‘best practice’ [3,4,12–16]. Many changes in operative technique have also occurred. The Carrel patch, with retention of the head and neck vessels on an ‘island’ of aorta, which is then anastomosed to the arch graft, was used in earlier series [17]. Criticism of this technique, including that it does not remove all the diseased aorta leading to possible re-dissection or emboli from athrosclerosis, led to the introduction of multibranched grafts [18,19]. Initially three branched, then after 1993 four branched to allow for a perfusion lumen [19]. Others, especially Spielvogel and colleagues, have advocated using trifurcated grafts, believing that it leads to less manipulation of the head and neck vessels, reduced circulatory arrest time to the brain and at least as satisfactory results in regards to morbidity and mortality compared to standard ACP [13,14,20–22]. The ‘Elephant Trunk’ technique, introduced by Borst et al. has been used to allow the distal anastomosis to be completed more proximally where there is greater vision and access as well providing a proximal attachment point for open or, increasingly, endovascular second stage operations for descending thoracic aortic lesions [23]. Improvements both in technique and cerebral protection have gradually lead to improving morbidity and mortality rates for patients undergoing aortic arch operations. Earlier series reported mortality rates of 42% in the 1960s [2] to 23–28% in the 1980s–90s [24,25]. However mortality rates in modern large series have been in the range of 6–16% [3,13] and permanent neurological morbidity has been in the range of 3–9% [4,26]. Our unit’s techniques have evolved over time, utilising most of the techniques mentioned, hence we decided to review our outcomes. Our aim in particular was determine if outcomes have improved with the ongoing modifications to our technique, as well as to determine predictors of mortality.
Materials and Methods Ethical approval was obtained from the Sydney South West Area Health Service Human Research Ethics Committee (Protocol number X10-0164) which included a waiver for consent for the retrospective data presented in this paper. Patients having an aortic arch replacement at our institution between 1989 and 2010 inclusive were retrieved from a departmental database. Patients were included if they had open surgery that involved replacing a circumferential part of their aortic arch and involved the re-implantation of at least one of the three head and neck vessels. These patients’ records were reviewed and data was extracted with regard to their preoperative status, intraoperative variables and their postoperative hospital outcomes. The patients were divided into two groups: Group A (1989–2005) and Group B (2006–2010). This divide represents changes such as the settled use of the current technique as described below – including the routine use of 25 ◦ C hypothermia and the exclusive use of ACP – the use of preformed and clotted grafts, factor VIIa and an increase in the annual number of procedures performed.
705
Surgical Techniques A variety of surgical techniques have been used at our institution during the period included in this study. Early cases frequently used Carrel patches with deep hypothermic circulatory arrest at 13–18 ◦ C. RCP was introduced in the early 1990s, being used in combination with ACP from the mid 1990s. RCP was last used in 2002 with all cases since using a variation of ACP. Occasionally a modification of the trifurcated graft technique has been used, where a Carrel patch has been attached to an intermediate graft then onto the arch graft, rather than a direct anastomosis to the arch graft, frequently using this second graft as a method to supply ACP rather than directly cannulating the head and neck vessels. More recently trifurcated grafts have also been used via a similar method to provide ACP.
Current Technique The operation is usually conducted via a median sternotomy, with a clam-shell incision or left thoracotomy used on occasion. Intraoperative transoesophageal echocardiogram is performed routinely as is intrathoracic use of carbon dioxide. The placement of the arterial cannula varies at the discretion of the surgeon, usually utilising central, femoral or axillary sites. Axillary or femoral cannulation is preferred for use in reoperative procedures. Antegrade and retrograde cold blood cardioplegia solution is given every 20 minutes throughout the procedure. Moderate hypothermia of 25 ◦ C is used. During the period of cooling preparations are made for an aortic valve or aortic root replacement as required. When a nasopharyngeal temperature of 25 ◦ C is reached the systemic circulation is arrested and the aorta opened. ACP is commenced either via cannulation of the innominate and left common carotid arteries or via clamping of the innominate artery with axilliary perfusion and cannulation of the left common carotid artery. When using direct cannulation the vessel is slung with a vessel loop or nylon tape to prevent displacement. Cerebral arrest is generally limited to less than 5 minutes. If femoral cannulation is being used, the aorta is cross clamped distal to the arch and lower body perfusion is continued, otherwise lower body perfusion is arrested until the completion of the distal anastomosis. A 4 branched gelatine impregnated arch graft (Gelwave; Vascutek; Inchinnan, Scotland) is used. The distal anastomosis of the arch graft to the proximal descending aorta is performed first, with Teflon reinforcement and pledgets as necessary. The graft is then cross clamped proximally and circulation is then restored to the body via the perfusion branch of the graft. The arch vessels are then anastomosed to the branches of the graft in sequence, first the left subclavian artery, then left common carotid artery and then the innominate artery. Whilst each of the anastomoses is being performed, perfusion is maintained to the other vessels via the arch graft or ACP, as appropriate. After each of the head and neck vessels has been anastomosed any other concomitant procedures – such as aortic valve replacement, aortic root replacement or coronary artery bypass grafting – are completed whilst the patient is being rewarmed. Finally the proximal anastomosis of the arch
ORIGINAL ARTICLE
Heart, Lung and Circulation 2011;20:704–711
ORIGINAL ARTICLE
706
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
graft is performed to the distal ascending aorta or the distal end of the aortic root graft as applicable.
Analysis of Results Statistical analysis was performed using SPSS v19.0 (IBM; Chicago, Illinois, USA). All categorical variables were expressed as number of patients and the percentage of valid cases for that variable (excluding patients with missing data). Continuous variables were expressed as means and standard deviations or medians and interquartile range where appropriate. The Kolmogorov–Smirnov Z test was used to test normality. Data from the earlier half of the series (Group A: 1989–2005) were compared with the latter half (Group B: 2006–2010) using an independent sample t test or Mann–Whitney-U test for continuous variables and using a Pearson’s Chi Squared test for categorical variables, with p < 0.05 being deemed significant in all cases. All preoperative and perioperative variables were included in a univariate analysis looking for predictors of mortality. This was performed using the Chi squared statistic for categorical variables and an independent sample t test or Mann–Whitney-U test for continuous variables. All variables with a p < 0.2 were considered eligible for inclusion into a multivariate analysis to form a model to look for independent predictors of mortality. The multivariate model was performed using Binominal Logistic Regression. If more univariate predictors were found than could be included in the multivariate analysis due to a lack of power, then potential predictors were selected for inclusion on a combination of their univariate significance and their clinical significance, selecting approximately one variable for every ten patients. Cross clamp time was defined from the time the cross clamp was placed on the aorta until it was released, or when body circulation ceased and/or restarted if there was circulatory arrest and the cross clamps were not applied. Circulatory arrest time was defined as when the brain was not being perfused from any source. ACP and RCP time were defined from when separate perfusion of the brain from the rest of the body was commenced until it was ceased. Permanent neurological dysfunction (PND) was defined as central neurological injury with residual symptoms at discharge, or central neurological injury found using clinical, radiological or post mortem pathological examination where the patient did not survive, whilst transient neurological dysfunction (TND) was defined as temporary and resolving central neurological symptoms, including confusion, delirium or agitation. Blood products reported include all blood products used from entry to theatres for the aortic arch operation until discharge from hospital.
Results On review of patient databases and records 75 patients were found to be eligible for the study and their files were analysed. There were 40 in Group A and 35 in Group B. The overall preoperative characteristics of the patients and
Heart, Lung and Circulation 2011;20:704–711
a comparison between Groups A and B (early and late series respectively) is presented in Table 1. Of note there were significantly more emergency cases (66.7% vs 34.3%), acute dissections (42.5% vs 14.3%) and history of myocardial infarctions (13.3% vs 0%) and stroke (18.2% vs 0%) in Group A, representing an overall higher risk cohort. Overall 23 (32.9%) patients had had previous cardiothoracic surgery. There was no significant difference between the groups, with Group A having 11 (31.4%) patients and Group B 12 (34.3%) patients. The overall intraoperative statistics and a comparison between Groups A and B are presented in Table 2. There were significant changes in the way the operation was performed earlier in comparison to later patients. Variations of techniques utilising Carrel patches were much more common in Group A (A: 66.6% vs B: 20.0%), whilst branched grafts were much more common in Group B (A: 15.4% vs B: 57.1%). Also methods of cerebral protection varied between Groups A and B. DHCA in isolation was used exclusively in the earliest of Group A patients, and techniques involving RCP were also only used in Group A patients, whilst forms of ACP, including axillary perfusion were much more common in Group B patients. Lowest nasopharyngeal temperatures were significantly lower in Group A patients (16.65 ◦ C vs 21.88 ◦ C) and all the intraoperative deaths were in Group A patients as well. The overall postoperative outcomes and a comparison between Groups A and B is presented in Table 3. The main outcome measure was 30 day mortality. There were 23 (30.7%) deaths in that period. There were six deaths out of 34 elective cases (17.6%). The causes of death for these patients were: eight (36.4%) haemorrhage, five (22.7%) cerebrovascular accident (CVA), five (22.7%) multiorgan failure (MOF), two (8.7%) were unable to be weaned from bypass, one (4.3%) pulmonary embolism (PE) and one (4.3%) from a bradycardic arrhythmia leading to cardiac arrest. There was a significantly greater mortality amongst patients in Group A (20–50%) than Group B (3–8.6%). The causes of death in Group A patients were: eight (42.2%) haemorrhage, four (21.1%) CVA, three (15.8%) MOF, two (10.0%) were unable to be weaned from bypass, one (5.0%) PE and one (5.0%) from a bradycardic arrhythmia leading to cardiac arrest. There were five deaths amongst 12 elective cases in Group A (41.6%). The causes of death in the Group B patients were: two (66.7%) MOF and one (33.3%) CVA. Of the three Group B mortalities, only one was in an elective case, giving a late series elective mortality of 1/22 (4.5%). Neurological outcomes were the secondary outcome measure. There were 14 (23.7%) patients who sustained a PND. Of these 14 patients seven (50%) died within 30 days. There were significantly more PND in Group A (10–40%) than Group B (4–11.8%). In Group A five (50%) died within 30 days whilst in Group B two (50%) died within 30 days. There were 17 (28.8%) patients who suffered from TND whose symptoms resolved by discharge. There were relatively similar numbers between Group A (7–28%) and Group B (10–29.4%) patients. There were 10 (16.4%) patients who sustained recurrent laryngeal nerve
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
707
Table 1. Preoperative Characteristics. Variables Agea Sex (Males) BMIa <20 >30 Elective Operation Aetiologyc Aneurysm Acute Dissection Chronic Dissection Previous CTx Surgery Pre-existing Arrhythmia Hypertension Cardiac Failure Ejection Fractiona Diabetes Mellitus (Total) Type 1 Type 2 Smoker (Ever) Current Former Hypercholesterolaemia Previous MI Pre-Operative Creatinineb >200 Family History Aortic/CT Disease Ischaemic Heart Disease Previous Carotid/CVA CVD PVD COPD Marfan’s Disease
Overall % (n) Mean/Median (SD/IQR)
Group A (1989–2005) % (n) Mean/Median (SD/IQR)
57.52 (±14.52) 74.7% (56) 27.74 (±4.58) 4.5% (3) 25.4% (17) 49.3% (35)
59.75 (±13.92) 70.0% (28) 26.46 (±3.87) 9.4% (3) 15.6% (5) 33.3% (12)
Group B (2005–2010) % (n) Mean/Median (SD/IQR)
p
54.97 (±14.97) 80.0% (28) 28.91 (±4.92) 0% (0) 34.3% (12) 65.7% (23)
n/s n/s 0.03 n/s n/s 0.01
72.0% (54) 29.3% (22) 17.3% (13) 32.9% (23) 11.8% (8) 76.7% (56) 4.5% (3) 57.91 (±11.08) 10.6% (7) 1.5% (1) 9.1% (6) 66.2% (43) 16.9% (11) 49.2% (32) 50% (30) 6.2% (4) 91.00 (82.00–113.00) 6.2% (4)
67.5% (27) 42.5% (17) 17.5% (7) 31.4% (11) 12.1% (4) 81.6% (31) 9.4% (3) 61.93 (±18.24) 12.9% (4) 0% (0) 12.9% (4) 76.7% (23) 26.7% (8) 50.0% (15) 65.4% (17) 13.3% (4) 96.00 (81.75–135.50) 10% (3)
77.1% (27) 14.3% (5) 17.1% (6) 34.3% (12) 11.4% (4) 71.4% (25) 0% (0) 56.21 (±5.59) 8.6% (3) 2.9% (1) 5.7% (2) 57.1% (20) 8.6% (3) 48.6% (17) 37.1% (13) 0% (0) 90.00 (87.0–100.0) 2.9% (1)
n/s 0.01 n/s n/s n/s n/s n/s n/s n/s n/s n/s n/s n/s n/s
11.7% (7) 8.3% (5) 8.8% (6) 24.6% (17) 17.2% (11) 7.6% (5) 4.5% (3)
16% (4) 20% (5) 18.2% (6) 32.4% (11) 24.1% (7) 6.5% (2) 3.1% (1)
8.6% (3) 0% (0) 0% (0) 17.1% (6) 11.4% (4) 8.6% (3) 5.7% (2)
n/s 0.01 0.01 n/s n/s n/s n/s
0.03 n/s n/s
Continuous Variable – expressed as mean ± Standard Deviation (SD). Skewed Continuous variable expressed as median and Interquartile Range (IQR). All other data is categorical – expressed as number (n) and percentage of valid cases (%). n/s = difference between early and late series not statistically significant (p > 0.05). c Patients with both an aneurysm and dissection will be represented in both categories as appropriate. BMI = Body Mass Index; CTx = Cardiothoracic; MI = Myocardial Infarction; CT Disease = Connective Tissue Disease; CVA = Cerebrovascular Accident; CVD = Cardiovascular Disease; PVD = Peripheral Vascular Disease; COPD = Chronic Obstructive Pulmonary Disease. a
b
injury intraoperatively. There was a larger number of patients in Group B (8–22.9%) who suffered this injury than in Group A (2–7.7%) however the difference was not significant. In respect to other outcomes, length of stay and ICU time were similar between groups. Inotropes were frequently used in both groups post operatively and there was a significantly longer ventilator time (38.0 hours vs 18.9 hours) and a greater chest drainage blood loss (2880.0 ml vs 680.0 ml) in Group A as opposed to Group B. Post operative atrial arrhythmias, especially atrial fibrillation, were very common in both Group A and B patients. Both groups had similar rates of respiratory complications and red blood cell and plasma transfusions, however significantly more platelets were transfused in Group A in comparison to Group B. The preoperative and perioperative variables were entered into a univariate analysis as described earlier. Statistically significant univariate predictors of mortality are presented in Table 4. Amongst these predictors increasing age, increasing circulatory arrest time and decreasing
lowest core body temperature were all found to be significant predictors of mortality. Aneurysm (as opposed to dissection) as an indication and ACP only (as opposed to other methods of cerebral protection including combination with RCP) were found to have significant protective effects. Due to the large number of significant predictors (16), not all could be included in the multivariate analysis. Therefore seven predictors were chosen based on their univariate statistical significance and clinical importance (Table 5). Due to circulatory arrest time and lowest nasopharyngeal temperature being very highly correlated (Pearson Correlation = −0.506; p < 0.001), it was decided to only include circulatory arrest time as it is more indicative of the clinical difficulty of the case and hence more clinically relevant. The multivariate model (Table 5) found that a history of cardiovascular disease and an increasing circulatory arrest time were both independent predictors of mortality for patients undergoing aortic arch replacement in our series.
ORIGINAL ARTICLE
Heart, Lung and Circulation 2011;20:704–711
ORIGINAL ARTICLE
708
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
Heart, Lung and Circulation 2011;20:704–711
Table 2. Perioperative Data. Variables Vessel Attachment Method Carrel Patch Trifurcated Graft Branched Graft Carrel Patch to Graft to Arch Other CPB Time (min)a Cross Clamp Time (min)a Cerebral Protection Method DHCA Only RCP RCP and ACP Combined Axillary ACP Direct ACP Circ. Arrest Time (min)b ACP Time (min)a RCP Time (min)a Lowest NP Temp. (◦ C)a Concomitant Procedures AVR Root Replacement Ascending Replacement CABG Intraoperative Mortality
Overall % (n) Mean/Median (SD/IQR)
Group A (1989–2005) % (n) Mean/Median (SD/IQR)
37.8% (28) 5.4% (4) 35.1% (26) 6.8% (5) 14.9% (11) 276.15 (±72.01) 199.44 (±74.59)
53.8% (21) 2.6% (1) 15.4% (6) 12.8% (5) 15.4% (6) 274.33 (±88.91) 183.35 (±83.17)
8.0% (6) 13.3% (10) 8.0% (6) 24.0% (18) 46.7% (35) 9.50 (2.75–38.00) 92.04 (±45.31) 41.30 (±23.63) 19.16 (±3.77) 42.7% (32) 46.7% (35) 64.0% (48) 12.0% (9) 13.3% (10)
15.0% (6) 25.0% (10) 15.0% (6) 12.5% (5) 32.5% (13) 36.00 (4.00–51.00) 88.77 (±54.67) 41.30 (±23.63) 16.65 (±2.83) 35.0% (14) 40.0% (16) 62.5% (25) 17.5% (7) 25% (10)
Group B (2005–2010) % (n) Mean/Median (SD/IQR)
p
20.0% (7) 8.6% (3) 57.1% (20) 0% (0) 14.3% (5) 227.71 (±54.89) 175.97 (±67.14)
<0.01 n/s <0.01 0.03 n/s n/s n/s
0% (0) 0% (0) 0% (0) 37.1% (13) 62.9% (22) 6.00 (1.00–13.00) 94.21 (±38.61) n/a 21.88 (±2.59)
0.02 <0.01 0.02 0.01 0.01 <0.01 n/s n/a <0.01
51.4% (18) 54.3% (19) 65.7% (23) 5.7% (2) 0% (0)
n/s n/s n/s n/s <0.01
Continuous Variable – expressed as mean ± Standard Deviation (SD). Skewed Continuous variable expressed as median and Interquartile Range (IQR). All other data is categorical – expressed as number (n) and percentage of valid cases (%). n/s = difference between early and late series not statistically significant (p > 0.05). CPB = Cardiopulmonary Bypass; DHCA = Deep Hypothermic Circulatory Arrest; RCP = Retrograde Cerebral Perfusion; ACP = Antegrade Cerebral Perfusion; NP = Nasopharyngeal; AVR = Aortic Valve Replacement; CABG = Coronary Artery Bypass Grafts.
a
b
Discussion This study has shown much improved outcomes from aortic arch replacement at our institution over time. Mortality has substantially improved from 50% in the earlier patients to 8.6% in all cases and 4.5% in elective cases in recent years. These recent results are in keeping with much of the recent international literature of around 6–16% [3,4,13,26]. Rates of PND have also significantly improved from 40% in the Group A patients to 11.8% in the Group B patients, which is comparable to results in the recent literature [3,4]. There have been many changes in the technical aspects of the procedure and patient selection which are believed to have contributed to this improvement in outcomes. In the very earliest patients only DHCA was used for cerebral protection, then RCP was introduced, sometimes combined with ACP and finally since 2003 ACP, which has become the ‘gold standard’, has been used exclusively. This improvement in cerebral protection, with concomitant decreases in circulatory arrest time, has allowed for less hypothermia. Hypothermia and circulatory arrest have been associated with alterations in coagulation leading to diffuse intravascular coagulation (DIC), consumption and dysfunction of platelets and coagulation factors, end organ damage and increased haemorrhage [27]. The significantly shorter periods of circulatory arrest in the later series of patients may well be contributing to the improved morbidity and mortality rates, as suggested by the univariate analysis, due to the known association
between prolonged circulatory arrest and poorer outcomes [6]. In the Group A patients the predominant method used was a Carrel patch, whilst in the Group B patients the predominant method used was branched grafts with a combination of other techniques, such as trifurcated grafts, used in both periods as well. The benefits of branched grafts include the complete excision of the diseased aorta as well as short circulatory arrest times. In this current study, neither Carrel patches nor branched grafts were significant in the univariate analysis of mortality. Except for one non-randomised study indicating that trifurcated grafts may be better than a modified Carrel patch technique, there is little research comparing various techniques of aortic arch replacement – such as branched grafts, Carrel patches or trifurcated grafts – to determine their relative benefits [14]. The current grafts are supplied preconstructed and are made of preclotted and more recently gelatin impregnated Dacron. These advances reduce the potential for graft leaking and haemorrhage, leading to reduced blood loss, as well as reduced operation time due to the prefabricated grafts not having to be constructed. In more recent years, in the Group B patients, there has been a dramatic increase in the rate of which aortic arch replacements have been performed at our institution. In fact almost as many procedures have been performed in the five years of Group B (35) as in the 17 years of Group A (40). This has given an average of seven operations per year for the last five years; compared to the average of just
709
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
Table 3. Postoperative Outcomes. Variables 30 Day Mortality Length of Stay (days)b Return to Theatre Haemorrhage Other Arch Related Other Non Cardiac ICU Time (hr)b Ventilator Time (hr)b Inotropes >4 h Post Op Drain Losses (ml)b Post Op Creatinineb Required Dialysis PND Post Operatively TND Post Operatively RLN Injury Post Op Arrhythmia Atrial Ventricular Resp. Complications LRTI/Pneumonia Pleural Effusion APO Pneumothorax Other Blood Products PRBC (units)b FFP (units)a Platelets (units)b
Overall % (n) Mean/Median (SD/IQR)
Group A (1989–2005) % (n) Mean/Median (SD/IQR)
30.7% (23) 11.00 (7.0–19.0)
50.0% (20) 9.00 (2.25–21.75)
12.5% (9) 8.3% (6) 19.4% (14) 88.5 (47.0–150.5) 27.5 (15.0–111.3) 69.5% (41) 1215.0 (492.5–2900.0) 106.0 (89.5–165.0) 11.1% (7) 23.7% (14) 28.8% (17) 16.4% (10)
10.8% (4) 10.8% (4) 16.2% (6) 96.00 (62.5–172.0) 38.00 (20.0–145.0) 62.5% (15) 2880.0 (1400.0–5635.0) 109.0 (95.25–221.00) 10.7% (3) 40.0% (10) 28.0% (7) 7.7% (2)
Group B (2005–2010) % (n) Mean/Median (SD/IQR) 8.6% (3) 12.00 (8.0–18.0) 14.3% (5) 5.7% (2) 22.9% (8) 81.00 (37.0–155.0) 18.00 (12.0–81.0) 22.9% (8) 680.0 (310.0–1400.0) 104.0 (87.0–143.0) 11.4% (4) 11.8% (4) 29.4% (10) 22.9% (8)
p <0.01 n/s n/s n/s n/s n/s 0.02 n/s <0.01 n/s n/s 0.01 n/s n/s
55.9% (33) 15.4% (10)
54.2% (13) 20% (6)
57.1% (20) 11.4% (4)
n/s n/s
19.1% (13) 16.2% (11) 5.9% (4) 4.4% (3) 8.8% (6)
12.1% (4) 21.2% (7) 0% (0) 3.0% (1) 15.2% (5)
25.7% (9) 11.4% (4) 11.4% (4) 5.7% (2) 2.9% (1)
n/s n/s 0.04 n/s n/s
5.0 (4.0–11.0) 8.75 (±8.42) 2.0 (1.0–4.0)
5.00 (4.00–14.00) 9.66 (±10.49) 3.00 (0.00–10.50)
5.00 (2.75–10.25) 7.97 (±6.12) 2.00 (1.00–3.25)
n/s n/s 0.02
Continuous Variable – expressed as mean ± Standard Deviation (SD). Skewed Continuous variable expressed as median and Interquartile Range (IQR). All other data is categorical – expressed as number (n) and percentage of valid cases (%). n/s = difference between early and late series not statistically significant (p > 0.05). ICU = Intensive Care Unit; PND = Permanent Neurological Dysfunction; TND = Transient Neurological Dysfunction; RLN = Recurrent Laryngeal Nerve; PRBC = Packed Red Blood Cells; FFP = Fresh Frozen Plasma; LRTI = Lower Respiratory Tract Infection; APO = Acute Pulmonary Oedema.
a
b
over two per year in the earlier patients of this series and the 10 [3] to 25 [22] per year of other contemporary series. This increase in volume has led to an increase in practical experience, allowing greater fine tuning and consistency in the procedures for looking after these highly complex
patients, both in the operating theatre and post operatively. The fact that some of the best reported outcomes come from studies involving large numbers of patients in relatively short periods of time [15] indicates that high volumes are associated with good results. This reinforces
Table 4. Positive Univariate Predictors. Variable
Odds Ratio
Agea Emergency Aneurysm Dissection CCF Hypertension Myocardial Infarct Family History of IHD Pre Op Creatinine >200 Previous Carotid/CVA CVD PVD Circ Arrest Timea ACP and RCP Combined ACP Only Lowest NP Temp. (◦ C)a
1.056 3.867 0.349 5.352 n/c n/c 10.071 6.750 n/c 6.267 6.000 8.556 1.041 14.167 0.293 0.754
a
95% CI 1.010–1.104 1.290–11.589 0.121–1.006 1.796–15.951 n/c n/c 0.970–104.609 0.994–45.854 n/c 1.042–37.689 1.831–19.660 2.063–35.481 1.014–1.069 1.549–129.560 0.102–0.840 0.637–0.894
p 0.003 0.013 0.047 0.002 0.005 0.002 0.022 0.03 0.001 0.027 0.002 0.001 0.004 0.004 0.019 <0.001
Continuous Variable, odds ratio expressed as per unit increase in variable. n/c = Odds Ratio incalculable due to zero patients in one or more categories. CCF = Congestive Cardiac Failure; IHD = Ischaemic Heart Disease; CVA = Cerebrovascular Accident; CVD = Cardiovascular Disease; PVD = Peripheral Vascular Disease; NP = Nasophayngeal; ACP = Antrograde Cerebral Perfusion; RCP = Retrograde Cerebral Perfusion.
ORIGINAL ARTICLE
Heart, Lung and Circulation 2011;20:704–711
ORIGINAL ARTICLE
710
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
Heart, Lung and Circulation 2011;20:704–711
Table 5. Multivariate Predictors. Variable Significant Predictors CVD Circ Arrest Time (min)a Other Predictors in Final Equation Hypertension Pre Op creatinine >200 Other Predictors Entered in Model Agea Emergency Dissection
Odds Ratio
95% CI
11.106 1.066
p
1.437–85.816 1.015–1.120
0.021 0.011
– –
– –
0.998 0.999
– – –
– – –
– – –
a
Continuous Variable, odds ratio expressed as per unit increase in variable. CVD = Cardiovascular Disease.
the idea that arch replacement should be conducted in specialised centres performing a relatively high volume of cases. Patients in Group A were also of much higher risk than the patients in Group B. There were significantly more elective operations and significantly less acute dissections in the later patients. The later patients also happened to show a trend towards lower prevelance for most of the comorbidities. The fact that there are generally fewer cases of the type typically regarded as especially high risk would also help explain the improved results, as well as the poorer results seen in earlier patients. The increase in numbers, especially in the number of elective cases, supports the idea that patients who were not previously offered aortic arch replacements are now being offered surgery. Contributing factors for our unit’s more aggressive approach to surgery include the improving outcomes of surgery, since operations are only offered if the risk of the procedure is outweighed by the risks of the underlying disease. Hence with reduced risks of an operation, more patients become eligible for arch replacement, giving rise to greater number of elective, as opposed to the more emergent and therefore higher risk cases than in previous eras. There are limitations to this current study which impact on how strongly conclusions can be drawn. It is a retrospective study based partially on prospectively collected data therefore it shares the limitations of all similar studies. Part of this includes missing data that could lead to potential biases in comparing earlier and later patients. Our study also includes the experience of multiple surgeons, using many changing techniques over a relatively long period of time. This results in great heterogeneity between the included cases. Whilst this is an advantage in allowing the overall effect of the improved techniques over time to be shown in a ‘real world’ manner, it makes it difficult to ascertain the exact causes of changing results. To overcome these limitations a prospective randomised trial between specific techniques is required, although this would be very difficult to practically accomplish at the rates, and in the circumstances, that aortic arch replacements are typically done. Areas of especial interest for further research include the relative merits of branched versus trifurcated grafts, the benefits of axillary perfusion vs other ACP strategies and determining the optimal level of hypothermia for use with modern techniques.
Another limitation to the study is the relatively small numbers of patients. This reduced the statistical power of the study, especially in regards to constructing the multivariate model. This meant that not all the univariate predictors could be included in this model, thereby reducing its validity in producing truly independent predictors of mortality. Instead the seven most clinically and statistically significant univariate predictors were chosen for inclusion. Since circulatory arrest time and lowest nasopharyngeal temperature were highly correlated, as mentioned in the results it was decided to only include circulatory arrest time in the model. Therefore the model is unable to determine what relative effect each of those two variables has in predicting mortality. However it is consistent with the great improvement in results over time with changed techniques, which have allowed operations which minimise both the period of circulatory arrest and the degree of cooling required. The circulatory arrest time is also biased due to the more complex nature of, and pathology in, some of the procedures with long circulatory arrest times. These procedures would be independently associated with a more adverse outcome due to their nature as well as the clinical effects of a long circulatory arrest. Due to the limitations in its construction, which are intrinsic to this study, the model should be regarded as an attempt try to explain some of the predictors of mortality whilst accounting for some, but not all, of the confounding factors.
Conclusion Our experience has shown that the improvements in technique over the last two decades, as well as increased experience, have led to greatly improved results from aortic arch surgery. Key amongst these improvements have been antegrade cerebral perfusion, higher core body temperatures, use of factor VIIa and prefabricated aortic arch grafts, especially multibranched grafts. In combination these techniques have allowed complete replacement of the diseased aortic arch whilst minimising the time the brain is without perfusion. These improvements have also led to aortic arch replacement being a viable option in the more aggressive management of elective surgical patients with aortic aneurysms affecting the aortic arch. We believe that future research is vital, especially in determining the relative benefits of the different forms of antegrade
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
cerebral perfusion and the different graft designs, as well as determining the outcomes of patients both in the long term and in different at risk groups.
Acknowledgements The authors would like to thank Dr. John Wynter and Prof. Richard Walsh for their assistance in collating perfusion chart data and Lisa Turner for her assistance in management of the departmental database and ethics application. There is no external financial support or conflicts of interest for any of the authors.
References [1] De Bakey M, Crawford E, Cooley D, Morris Jr G. Successful resection of fusiform aneurysm of aortic arch with replacement by homograft. Surg Gynecol Obstet 1957;105(6): 657. [2] De Bakey ME, Henly WS, Cooley DA, Crawford ES, Morris Jr GC, Beall Jr AC. Aneurysms of the aortic arch: factors influencing operative risk. Surg Clin North Am 1962;42:1543–54. [3] Sundt 3rd TM, Orszulak TA, Cook DJ, Schaff HV. Improving results of open arch replacement. Ann Thorac Surg 2008;86(3):787–96. [4] Kazui T, Yamashita K, Washiyama N, Terada H, Bashar AHM, Suzuki K, et al. Aortic arch replacement using selective cerebral perfusion. Ann Thorac Surg 2007;83(2):S796–8. [5] Griepp R, Stinson E, Hollingsworth J, Buehler D. Prosthetic replacement of the aortic arch. J Thorac Cardiovasc Surg 1975;70(6):1051. [6] Svensson L, Crawford E, Hess K, Coselli J, Raskin S, Shenaq S, et al. Deep hypothermia with circulatory arrest. Determinants of stroke and early mortality in 656 patients. J Thorac Cardiovasc Surg 1993;106(1):19–28. [7] Milewski RK, Pacini D, Moser GW, Moeller P, Cowie D, Szeto WY, et al. Retrograde and antegrade cerebral perfusion: results in short elective arch reconstructive times. Ann Thorac Surg 2010;89(5):1448–57. [8] Ueda Y, Okita Y, Aomi S, Koyanagi H, Takamoto S. Retrograde cerebral perfusion for aortic arch surgery: analysis of risk factors. Ann Thorac Surg 1999;67(6):1879–82, discussion 91–4. [9] Usui A, Abe T, Murase M. Early clinical results of retrograde cerebral perfusion for aortic arch operations in Japan. Ann Thorac Surg 1996;62(1):94–103. [10] Ehrlich MP, Hagl C, McCullough JN, Zhang N, Shiang H, Bodian C, et al. Retrograde cerebral perfusion provides negligible flow through brain capillaries in the pig. J Thorac Cardiovasc Surg 2001;122(2):331–8. [11] Moon MR, Sundt 3rd TM. Influence of retrograde cerebral perfusion during aortic arch procedures. Ann Thorac Surg 2002;74(2):426–31. [12] Okita Y, Minatoya K, Tagusari O, Ando M, Nagatsuka K, Kitamura S. Prospective comparative study of brain protection in total aortic arch replacement: deep hypothermic circulatory arrest with retrograde cerebral perfusion or selective antegrade cerebral perfusion. Ann Thorac Surg 2001;72(1):72–9.
711
[13] Bischoff MS, Brenner RM, Scheumann J, Bodian CA, Griepp RB, Lansman SL, et al. Long-term outcome after aortic arch replacement with a trifurcated graft. J Thorac Cardiovasc Surg 2010;140(6 Suppl.):S71–6. [14] Strauch JT, Spielvogel D, Lauten A, Galla JD, Lansman SL, McMurtry K, et al. Technical advances in total aortic arch replacement. Ann Thorac Surg 2004;77(2):581–9. [15] Minatoya K, Ogino H, Matsuda H, Sasaki H, Tanaka H, Kobayashi J, et al. Evolving selective cerebral perfusion for aortic arch replacement: high flow rate with moderate hypothermic circulatory arrest. Ann Thorac Surg 2008;86(6):1827–31. [16] Touati GD, Marticho P, Farag M, Carmi D, Szymanski C, Barry M, et al. Totally normothermic aortic arch replacement without circulatory arrest. Eur J Cardiothorac Surg 2007;32(2):263–8. [17] Crawford ES, Saleh SA. Transverse aortic arch aneurysm: improved results of treatment employing new modifications of aortic reconstruction and hypothermic cerebral circulatory arrest. Ann Surg 1981;194(2):180–8. [18] Sakamoto S, Matsubara J, Nagayoshi Y, Nishizawa H, Takeuchi K, Nonaka T, et al. Clinical results of aortic arch replacement using a four branched prosthetic graft. J Cardiovasc Surg 2003;44(6):751–5. [19] Kazui T, Washiyama N, Muhammad B, Terada H, Yamashita K, Takinami M, et al. Total arch replacement using aortic arch branched grafts with the aid of antegrade selective cerebral perfusion. Ann Thorac Surg 2000;70(1):3. [20] Spielvogel D, Strauch JT, Minanov OP, Lansman SL, Griepp RB. Aortic arch replacement using a trifurcated graft and selective cerebral antegrade perfusion. Ann Thorac Surg 2002;74(5):S1810–4. [21] Spielvogel D, Halstead JC, Meier M, Kadir I, Lansman SL, Shahani R, et al. Aortic arch replacement using a trifurcated graft: simple, versatile, and safe. Ann Thorac Surg 2005;80(1):90–5. [22] Spielvogel D, Etz CD, Silovitz D, Lansman SL, Griepp RB. Aortic arch replacement with a trifurcated graft. Ann Thorac Surg 2007;83(2):S791–5. [23] Borst H, Frank G, Schaps D. Treatment of extensive aortic aneurysms by a new multiple-stage approach. J Thorac Cardiovasc Surg 1988;95(1):11. [24] Ergin MA, O’Connor J, Guinto R, Griepp RB. Experience with profound hypothermia and circulatory arrest in the treatment of aneurysms of the aortic arch. Aortic arch replacement for acute arch dissections. J Thorac Cardiovasc Surg 1982;84(5):649–55. [25] Ehrlich M, Grabenwoger M, Luckner D, Simon P, Laufer G, Wolner E, et al. Operative management of aortic arch aneurysm using profound hypothermia and circulatory arrest. J Cardiovasc Surg 1996;37(6 Suppl. 1):63–4. [26] Schwartz JP, Bakhos M, Patel A, Botkin S, Neragi-Miandoab S, Schwartz JP, et al. Repair of aortic arch and the impact of cross-clamping time, New York Heart Association stage, circulatory arrest time, and age on operative outcome. Interact Cardiovasc Thorac Surg 2008;7(3):425–9. [27] Wilde JT. Hematological consequences of profound hypothermic circulatory arrest and aortic dissection. J Card Surg 1997;12(2 Suppl.):201–6.
ORIGINAL ARTICLE
Heart, Lung and Circulation 2011;20:704–711
Original Article
Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience Reece A. Davies, BAppSc (Phty) (Hons) b,d , Deborah Black, PhD b , Richmond W. Jeremy, PhD, FRACP, FCSANZ b,c , Paul G. Bannon, PhD, FRACS a,b,d , Matthew S. Bayfield, FRACS a,d , P. Nicholas Hendel, FRACS a,b,d , Clifford F. Hughes, AO, FRACS a,b,d , Michael K. Wilson, FRACS a,d and Michael P. Vallely, PhD, FRACS a,b,d,∗ a
Cardiothoracic Surgical Unit, Royal Prince Alfred Hospital, Sydney, Australia b Faculty of Medicine, The University of Sydney, Sydney, Australia c Department of Cardiology, Royal Prince Alfred Hospital, Sydney, Australia d The Baird Institute for Heart and Lung Surgical Research, Sydney, Australia
Background: Aortic arch replacement is a complicated and high risk procedure. There have been many advances over recent years. We review the changes in our unit’s techniques and outcomes over the past 22 years. Methods: Data were collated from databases and medical records for all patients who underwent aortic arch replacement surgery from January 1989 to December 2010. The patients were divided into two groups – Group A (1989–2005) and Group B (2006–2010). Data were analysed to compare early and late series patients’ outcomes. Logistic regression was used to identify variables that predicted mortality. Results: Seventy-five eligible patients (56 males; mean age: 57.5 years; Group A: 40, Group B 35) were identified. There were great changes in the technique and the methods of cerebral protection. The overall mortality rate was 30.7% – Group A: 50% and Group B: 8.6% (p < 0.001). Overall permanent neurological dysfunction was 23.7% – Group A: 40% and Group B: 11.8% (p = 0.012). Cardiovascular disease and circulatory arrest time were significant predictors of mortality. Conclusions: Increased experience and volume and advances in techniques over 22 years have resulted in major improvements in outcomes for patients having aortic arch replacement, allowing the procedure to be performed with greatly improved outcomes. (Heart, Lung and Circulation 2011;20:704–711) © 2011 Australasian Society of Cardiac and Thoracic Surgeons and the Cardiac Society of Australia and New Zealand. Published by Elsevier Inc. All rights reserved. Keywords. Thoracic aorta; Deep hypothermia induced circulatory arrest; Thoracic aortic aneurysm; Blood vessel prosthesis implantation; Aortic arch replacement
Introduction
A
ortic arch replacement is a complicated and relatively high risk operation primarily used in the treatment of aneurysms and/or dissections involving the aortic arch. The first successful arch replacement was reported by DeBakey et al. in 1957 after previously unsuccessful attempts [1]. Mortality in an early series was 42% [2], however with advancements in technique, especially in the field of cerebral protection, mortality and morbidity rates have decreased [3,4]. Deep hypothermic circulatory arrest (DHCA) was introduced and shown to be effective in proReceived 13 May 2011; accepted 22 July 2011; available online 26 August 2011 ∗
Corresponding author at: C/- Sydney Cardiothoracic Surgeons, 304/100 Carillon Avenue, Newtown, NSW 2042, Australia. Tel.: +61 295501933; fax: +61 295506669. E-mail address: [email protected] (M.P. Vallely).
tecting the brain [5]. However, further research showed the risk of stroke increased significantly after periods of arrest exceeding 40 minutes and mortality increased significantly after 65 minutes [6]. The two main techniques to overcome this have been retrograde cerebral perfusion (RCP) and antegrade cerebral perfusion (ACP) [7]. RCP was introduced in the 1980s with initially satisfactory results, including a mortality rate of 10–19.7% and a 4–11.8% rate of permanent neurological deficit [8,9]. However others have questioned its effectiveness in adequately perfusing the brain [10,11]. A prospective controlled trial has shown significantly higher rates of transient neurological dysfunction with RCP compared with ACP [12]. ACP, which was used in DeBakey’s original operation but subsequently fell from favour, has enjoyed a resurgence, having been shown to produce significant reductions in mortality and morbidity, especially neurological, compared to both RCP and DHCA alone [3]. ACP has increasingly become the primary form of cerebral protection used in complex
© 2011 Australasian Society of Cardiac and Thoracic Surgeons and the Cardiac Society of Australia and New Zealand. Published by Elsevier Inc. All rights reserved.
1443-9506/04/$36.00 doi:10.1016/j.hlc.2011.07.009
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
operations to replace the aortic arch and is seen by many surgeons as ‘best practice’ [3,4,12–16]. Many changes in operative technique have also occurred. The Carrel patch, with retention of the head and neck vessels on an ‘island’ of aorta, which is then anastomosed to the arch graft, was used in earlier series [17]. Criticism of this technique, including that it does not remove all the diseased aorta leading to possible re-dissection or emboli from athrosclerosis, led to the introduction of multibranched grafts [18,19]. Initially three branched, then after 1993 four branched to allow for a perfusion lumen [19]. Others, especially Spielvogel and colleagues, have advocated using trifurcated grafts, believing that it leads to less manipulation of the head and neck vessels, reduced circulatory arrest time to the brain and at least as satisfactory results in regards to morbidity and mortality compared to standard ACP [13,14,20–22]. The ‘Elephant Trunk’ technique, introduced by Borst et al. has been used to allow the distal anastomosis to be completed more proximally where there is greater vision and access as well providing a proximal attachment point for open or, increasingly, endovascular second stage operations for descending thoracic aortic lesions [23]. Improvements both in technique and cerebral protection have gradually lead to improving morbidity and mortality rates for patients undergoing aortic arch operations. Earlier series reported mortality rates of 42% in the 1960s [2] to 23–28% in the 1980s–90s [24,25]. However mortality rates in modern large series have been in the range of 6–16% [3,13] and permanent neurological morbidity has been in the range of 3–9% [4,26]. Our unit’s techniques have evolved over time, utilising most of the techniques mentioned, hence we decided to review our outcomes. Our aim in particular was determine if outcomes have improved with the ongoing modifications to our technique, as well as to determine predictors of mortality.
Materials and Methods Ethical approval was obtained from the Sydney South West Area Health Service Human Research Ethics Committee (Protocol number X10-0164) which included a waiver for consent for the retrospective data presented in this paper. Patients having an aortic arch replacement at our institution between 1989 and 2010 inclusive were retrieved from a departmental database. Patients were included if they had open surgery that involved replacing a circumferential part of their aortic arch and involved the re-implantation of at least one of the three head and neck vessels. These patients’ records were reviewed and data was extracted with regard to their preoperative status, intraoperative variables and their postoperative hospital outcomes. The patients were divided into two groups: Group A (1989–2005) and Group B (2006–2010). This divide represents changes such as the settled use of the current technique as described below – including the routine use of 25 ◦ C hypothermia and the exclusive use of ACP – the use of preformed and clotted grafts, factor VIIa and an increase in the annual number of procedures performed.
705
Surgical Techniques A variety of surgical techniques have been used at our institution during the period included in this study. Early cases frequently used Carrel patches with deep hypothermic circulatory arrest at 13–18 ◦ C. RCP was introduced in the early 1990s, being used in combination with ACP from the mid 1990s. RCP was last used in 2002 with all cases since using a variation of ACP. Occasionally a modification of the trifurcated graft technique has been used, where a Carrel patch has been attached to an intermediate graft then onto the arch graft, rather than a direct anastomosis to the arch graft, frequently using this second graft as a method to supply ACP rather than directly cannulating the head and neck vessels. More recently trifurcated grafts have also been used via a similar method to provide ACP.
Current Technique The operation is usually conducted via a median sternotomy, with a clam-shell incision or left thoracotomy used on occasion. Intraoperative transoesophageal echocardiogram is performed routinely as is intrathoracic use of carbon dioxide. The placement of the arterial cannula varies at the discretion of the surgeon, usually utilising central, femoral or axillary sites. Axillary or femoral cannulation is preferred for use in reoperative procedures. Antegrade and retrograde cold blood cardioplegia solution is given every 20 minutes throughout the procedure. Moderate hypothermia of 25 ◦ C is used. During the period of cooling preparations are made for an aortic valve or aortic root replacement as required. When a nasopharyngeal temperature of 25 ◦ C is reached the systemic circulation is arrested and the aorta opened. ACP is commenced either via cannulation of the innominate and left common carotid arteries or via clamping of the innominate artery with axilliary perfusion and cannulation of the left common carotid artery. When using direct cannulation the vessel is slung with a vessel loop or nylon tape to prevent displacement. Cerebral arrest is generally limited to less than 5 minutes. If femoral cannulation is being used, the aorta is cross clamped distal to the arch and lower body perfusion is continued, otherwise lower body perfusion is arrested until the completion of the distal anastomosis. A 4 branched gelatine impregnated arch graft (Gelwave; Vascutek; Inchinnan, Scotland) is used. The distal anastomosis of the arch graft to the proximal descending aorta is performed first, with Teflon reinforcement and pledgets as necessary. The graft is then cross clamped proximally and circulation is then restored to the body via the perfusion branch of the graft. The arch vessels are then anastomosed to the branches of the graft in sequence, first the left subclavian artery, then left common carotid artery and then the innominate artery. Whilst each of the anastomoses is being performed, perfusion is maintained to the other vessels via the arch graft or ACP, as appropriate. After each of the head and neck vessels has been anastomosed any other concomitant procedures – such as aortic valve replacement, aortic root replacement or coronary artery bypass grafting – are completed whilst the patient is being rewarmed. Finally the proximal anastomosis of the arch
ORIGINAL ARTICLE
Heart, Lung and Circulation 2011;20:704–711
ORIGINAL ARTICLE
706
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
graft is performed to the distal ascending aorta or the distal end of the aortic root graft as applicable.
Analysis of Results Statistical analysis was performed using SPSS v19.0 (IBM; Chicago, Illinois, USA). All categorical variables were expressed as number of patients and the percentage of valid cases for that variable (excluding patients with missing data). Continuous variables were expressed as means and standard deviations or medians and interquartile range where appropriate. The Kolmogorov–Smirnov Z test was used to test normality. Data from the earlier half of the series (Group A: 1989–2005) were compared with the latter half (Group B: 2006–2010) using an independent sample t test or Mann–Whitney-U test for continuous variables and using a Pearson’s Chi Squared test for categorical variables, with p < 0.05 being deemed significant in all cases. All preoperative and perioperative variables were included in a univariate analysis looking for predictors of mortality. This was performed using the Chi squared statistic for categorical variables and an independent sample t test or Mann–Whitney-U test for continuous variables. All variables with a p < 0.2 were considered eligible for inclusion into a multivariate analysis to form a model to look for independent predictors of mortality. The multivariate model was performed using Binominal Logistic Regression. If more univariate predictors were found than could be included in the multivariate analysis due to a lack of power, then potential predictors were selected for inclusion on a combination of their univariate significance and their clinical significance, selecting approximately one variable for every ten patients. Cross clamp time was defined from the time the cross clamp was placed on the aorta until it was released, or when body circulation ceased and/or restarted if there was circulatory arrest and the cross clamps were not applied. Circulatory arrest time was defined as when the brain was not being perfused from any source. ACP and RCP time were defined from when separate perfusion of the brain from the rest of the body was commenced until it was ceased. Permanent neurological dysfunction (PND) was defined as central neurological injury with residual symptoms at discharge, or central neurological injury found using clinical, radiological or post mortem pathological examination where the patient did not survive, whilst transient neurological dysfunction (TND) was defined as temporary and resolving central neurological symptoms, including confusion, delirium or agitation. Blood products reported include all blood products used from entry to theatres for the aortic arch operation until discharge from hospital.
Results On review of patient databases and records 75 patients were found to be eligible for the study and their files were analysed. There were 40 in Group A and 35 in Group B. The overall preoperative characteristics of the patients and
Heart, Lung and Circulation 2011;20:704–711
a comparison between Groups A and B (early and late series respectively) is presented in Table 1. Of note there were significantly more emergency cases (66.7% vs 34.3%), acute dissections (42.5% vs 14.3%) and history of myocardial infarctions (13.3% vs 0%) and stroke (18.2% vs 0%) in Group A, representing an overall higher risk cohort. Overall 23 (32.9%) patients had had previous cardiothoracic surgery. There was no significant difference between the groups, with Group A having 11 (31.4%) patients and Group B 12 (34.3%) patients. The overall intraoperative statistics and a comparison between Groups A and B are presented in Table 2. There were significant changes in the way the operation was performed earlier in comparison to later patients. Variations of techniques utilising Carrel patches were much more common in Group A (A: 66.6% vs B: 20.0%), whilst branched grafts were much more common in Group B (A: 15.4% vs B: 57.1%). Also methods of cerebral protection varied between Groups A and B. DHCA in isolation was used exclusively in the earliest of Group A patients, and techniques involving RCP were also only used in Group A patients, whilst forms of ACP, including axillary perfusion were much more common in Group B patients. Lowest nasopharyngeal temperatures were significantly lower in Group A patients (16.65 ◦ C vs 21.88 ◦ C) and all the intraoperative deaths were in Group A patients as well. The overall postoperative outcomes and a comparison between Groups A and B is presented in Table 3. The main outcome measure was 30 day mortality. There were 23 (30.7%) deaths in that period. There were six deaths out of 34 elective cases (17.6%). The causes of death for these patients were: eight (36.4%) haemorrhage, five (22.7%) cerebrovascular accident (CVA), five (22.7%) multiorgan failure (MOF), two (8.7%) were unable to be weaned from bypass, one (4.3%) pulmonary embolism (PE) and one (4.3%) from a bradycardic arrhythmia leading to cardiac arrest. There was a significantly greater mortality amongst patients in Group A (20–50%) than Group B (3–8.6%). The causes of death in Group A patients were: eight (42.2%) haemorrhage, four (21.1%) CVA, three (15.8%) MOF, two (10.0%) were unable to be weaned from bypass, one (5.0%) PE and one (5.0%) from a bradycardic arrhythmia leading to cardiac arrest. There were five deaths amongst 12 elective cases in Group A (41.6%). The causes of death in the Group B patients were: two (66.7%) MOF and one (33.3%) CVA. Of the three Group B mortalities, only one was in an elective case, giving a late series elective mortality of 1/22 (4.5%). Neurological outcomes were the secondary outcome measure. There were 14 (23.7%) patients who sustained a PND. Of these 14 patients seven (50%) died within 30 days. There were significantly more PND in Group A (10–40%) than Group B (4–11.8%). In Group A five (50%) died within 30 days whilst in Group B two (50%) died within 30 days. There were 17 (28.8%) patients who suffered from TND whose symptoms resolved by discharge. There were relatively similar numbers between Group A (7–28%) and Group B (10–29.4%) patients. There were 10 (16.4%) patients who sustained recurrent laryngeal nerve
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
707
Table 1. Preoperative Characteristics. Variables Agea Sex (Males) BMIa <20 >30 Elective Operation Aetiologyc Aneurysm Acute Dissection Chronic Dissection Previous CTx Surgery Pre-existing Arrhythmia Hypertension Cardiac Failure Ejection Fractiona Diabetes Mellitus (Total) Type 1 Type 2 Smoker (Ever) Current Former Hypercholesterolaemia Previous MI Pre-Operative Creatinineb >200 Family History Aortic/CT Disease Ischaemic Heart Disease Previous Carotid/CVA CVD PVD COPD Marfan’s Disease
Overall % (n) Mean/Median (SD/IQR)
Group A (1989–2005) % (n) Mean/Median (SD/IQR)
57.52 (±14.52) 74.7% (56) 27.74 (±4.58) 4.5% (3) 25.4% (17) 49.3% (35)
59.75 (±13.92) 70.0% (28) 26.46 (±3.87) 9.4% (3) 15.6% (5) 33.3% (12)
Group B (2005–2010) % (n) Mean/Median (SD/IQR)
p
54.97 (±14.97) 80.0% (28) 28.91 (±4.92) 0% (0) 34.3% (12) 65.7% (23)
n/s n/s 0.03 n/s n/s 0.01
72.0% (54) 29.3% (22) 17.3% (13) 32.9% (23) 11.8% (8) 76.7% (56) 4.5% (3) 57.91 (±11.08) 10.6% (7) 1.5% (1) 9.1% (6) 66.2% (43) 16.9% (11) 49.2% (32) 50% (30) 6.2% (4) 91.00 (82.00–113.00) 6.2% (4)
67.5% (27) 42.5% (17) 17.5% (7) 31.4% (11) 12.1% (4) 81.6% (31) 9.4% (3) 61.93 (±18.24) 12.9% (4) 0% (0) 12.9% (4) 76.7% (23) 26.7% (8) 50.0% (15) 65.4% (17) 13.3% (4) 96.00 (81.75–135.50) 10% (3)
77.1% (27) 14.3% (5) 17.1% (6) 34.3% (12) 11.4% (4) 71.4% (25) 0% (0) 56.21 (±5.59) 8.6% (3) 2.9% (1) 5.7% (2) 57.1% (20) 8.6% (3) 48.6% (17) 37.1% (13) 0% (0) 90.00 (87.0–100.0) 2.9% (1)
n/s 0.01 n/s n/s n/s n/s n/s n/s n/s n/s n/s n/s n/s n/s
11.7% (7) 8.3% (5) 8.8% (6) 24.6% (17) 17.2% (11) 7.6% (5) 4.5% (3)
16% (4) 20% (5) 18.2% (6) 32.4% (11) 24.1% (7) 6.5% (2) 3.1% (1)
8.6% (3) 0% (0) 0% (0) 17.1% (6) 11.4% (4) 8.6% (3) 5.7% (2)
n/s 0.01 0.01 n/s n/s n/s n/s
0.03 n/s n/s
Continuous Variable – expressed as mean ± Standard Deviation (SD). Skewed Continuous variable expressed as median and Interquartile Range (IQR). All other data is categorical – expressed as number (n) and percentage of valid cases (%). n/s = difference between early and late series not statistically significant (p > 0.05). c Patients with both an aneurysm and dissection will be represented in both categories as appropriate. BMI = Body Mass Index; CTx = Cardiothoracic; MI = Myocardial Infarction; CT Disease = Connective Tissue Disease; CVA = Cerebrovascular Accident; CVD = Cardiovascular Disease; PVD = Peripheral Vascular Disease; COPD = Chronic Obstructive Pulmonary Disease. a
b
injury intraoperatively. There was a larger number of patients in Group B (8–22.9%) who suffered this injury than in Group A (2–7.7%) however the difference was not significant. In respect to other outcomes, length of stay and ICU time were similar between groups. Inotropes were frequently used in both groups post operatively and there was a significantly longer ventilator time (38.0 hours vs 18.9 hours) and a greater chest drainage blood loss (2880.0 ml vs 680.0 ml) in Group A as opposed to Group B. Post operative atrial arrhythmias, especially atrial fibrillation, were very common in both Group A and B patients. Both groups had similar rates of respiratory complications and red blood cell and plasma transfusions, however significantly more platelets were transfused in Group A in comparison to Group B. The preoperative and perioperative variables were entered into a univariate analysis as described earlier. Statistically significant univariate predictors of mortality are presented in Table 4. Amongst these predictors increasing age, increasing circulatory arrest time and decreasing
lowest core body temperature were all found to be significant predictors of mortality. Aneurysm (as opposed to dissection) as an indication and ACP only (as opposed to other methods of cerebral protection including combination with RCP) were found to have significant protective effects. Due to the large number of significant predictors (16), not all could be included in the multivariate analysis. Therefore seven predictors were chosen based on their univariate statistical significance and clinical importance (Table 5). Due to circulatory arrest time and lowest nasopharyngeal temperature being very highly correlated (Pearson Correlation = −0.506; p < 0.001), it was decided to only include circulatory arrest time as it is more indicative of the clinical difficulty of the case and hence more clinically relevant. The multivariate model (Table 5) found that a history of cardiovascular disease and an increasing circulatory arrest time were both independent predictors of mortality for patients undergoing aortic arch replacement in our series.
ORIGINAL ARTICLE
Heart, Lung and Circulation 2011;20:704–711
ORIGINAL ARTICLE
708
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
Heart, Lung and Circulation 2011;20:704–711
Table 2. Perioperative Data. Variables Vessel Attachment Method Carrel Patch Trifurcated Graft Branched Graft Carrel Patch to Graft to Arch Other CPB Time (min)a Cross Clamp Time (min)a Cerebral Protection Method DHCA Only RCP RCP and ACP Combined Axillary ACP Direct ACP Circ. Arrest Time (min)b ACP Time (min)a RCP Time (min)a Lowest NP Temp. (◦ C)a Concomitant Procedures AVR Root Replacement Ascending Replacement CABG Intraoperative Mortality
Overall % (n) Mean/Median (SD/IQR)
Group A (1989–2005) % (n) Mean/Median (SD/IQR)
37.8% (28) 5.4% (4) 35.1% (26) 6.8% (5) 14.9% (11) 276.15 (±72.01) 199.44 (±74.59)
53.8% (21) 2.6% (1) 15.4% (6) 12.8% (5) 15.4% (6) 274.33 (±88.91) 183.35 (±83.17)
8.0% (6) 13.3% (10) 8.0% (6) 24.0% (18) 46.7% (35) 9.50 (2.75–38.00) 92.04 (±45.31) 41.30 (±23.63) 19.16 (±3.77) 42.7% (32) 46.7% (35) 64.0% (48) 12.0% (9) 13.3% (10)
15.0% (6) 25.0% (10) 15.0% (6) 12.5% (5) 32.5% (13) 36.00 (4.00–51.00) 88.77 (±54.67) 41.30 (±23.63) 16.65 (±2.83) 35.0% (14) 40.0% (16) 62.5% (25) 17.5% (7) 25% (10)
Group B (2005–2010) % (n) Mean/Median (SD/IQR)
p
20.0% (7) 8.6% (3) 57.1% (20) 0% (0) 14.3% (5) 227.71 (±54.89) 175.97 (±67.14)
<0.01 n/s <0.01 0.03 n/s n/s n/s
0% (0) 0% (0) 0% (0) 37.1% (13) 62.9% (22) 6.00 (1.00–13.00) 94.21 (±38.61) n/a 21.88 (±2.59)
0.02 <0.01 0.02 0.01 0.01 <0.01 n/s n/a <0.01
51.4% (18) 54.3% (19) 65.7% (23) 5.7% (2) 0% (0)
n/s n/s n/s n/s <0.01
Continuous Variable – expressed as mean ± Standard Deviation (SD). Skewed Continuous variable expressed as median and Interquartile Range (IQR). All other data is categorical – expressed as number (n) and percentage of valid cases (%). n/s = difference between early and late series not statistically significant (p > 0.05). CPB = Cardiopulmonary Bypass; DHCA = Deep Hypothermic Circulatory Arrest; RCP = Retrograde Cerebral Perfusion; ACP = Antegrade Cerebral Perfusion; NP = Nasopharyngeal; AVR = Aortic Valve Replacement; CABG = Coronary Artery Bypass Grafts.
a
b
Discussion This study has shown much improved outcomes from aortic arch replacement at our institution over time. Mortality has substantially improved from 50% in the earlier patients to 8.6% in all cases and 4.5% in elective cases in recent years. These recent results are in keeping with much of the recent international literature of around 6–16% [3,4,13,26]. Rates of PND have also significantly improved from 40% in the Group A patients to 11.8% in the Group B patients, which is comparable to results in the recent literature [3,4]. There have been many changes in the technical aspects of the procedure and patient selection which are believed to have contributed to this improvement in outcomes. In the very earliest patients only DHCA was used for cerebral protection, then RCP was introduced, sometimes combined with ACP and finally since 2003 ACP, which has become the ‘gold standard’, has been used exclusively. This improvement in cerebral protection, with concomitant decreases in circulatory arrest time, has allowed for less hypothermia. Hypothermia and circulatory arrest have been associated with alterations in coagulation leading to diffuse intravascular coagulation (DIC), consumption and dysfunction of platelets and coagulation factors, end organ damage and increased haemorrhage [27]. The significantly shorter periods of circulatory arrest in the later series of patients may well be contributing to the improved morbidity and mortality rates, as suggested by the univariate analysis, due to the known association
between prolonged circulatory arrest and poorer outcomes [6]. In the Group A patients the predominant method used was a Carrel patch, whilst in the Group B patients the predominant method used was branched grafts with a combination of other techniques, such as trifurcated grafts, used in both periods as well. The benefits of branched grafts include the complete excision of the diseased aorta as well as short circulatory arrest times. In this current study, neither Carrel patches nor branched grafts were significant in the univariate analysis of mortality. Except for one non-randomised study indicating that trifurcated grafts may be better than a modified Carrel patch technique, there is little research comparing various techniques of aortic arch replacement – such as branched grafts, Carrel patches or trifurcated grafts – to determine their relative benefits [14]. The current grafts are supplied preconstructed and are made of preclotted and more recently gelatin impregnated Dacron. These advances reduce the potential for graft leaking and haemorrhage, leading to reduced blood loss, as well as reduced operation time due to the prefabricated grafts not having to be constructed. In more recent years, in the Group B patients, there has been a dramatic increase in the rate of which aortic arch replacements have been performed at our institution. In fact almost as many procedures have been performed in the five years of Group B (35) as in the 17 years of Group A (40). This has given an average of seven operations per year for the last five years; compared to the average of just
709
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
Table 3. Postoperative Outcomes. Variables 30 Day Mortality Length of Stay (days)b Return to Theatre Haemorrhage Other Arch Related Other Non Cardiac ICU Time (hr)b Ventilator Time (hr)b Inotropes >4 h Post Op Drain Losses (ml)b Post Op Creatinineb Required Dialysis PND Post Operatively TND Post Operatively RLN Injury Post Op Arrhythmia Atrial Ventricular Resp. Complications LRTI/Pneumonia Pleural Effusion APO Pneumothorax Other Blood Products PRBC (units)b FFP (units)a Platelets (units)b
Overall % (n) Mean/Median (SD/IQR)
Group A (1989–2005) % (n) Mean/Median (SD/IQR)
30.7% (23) 11.00 (7.0–19.0)
50.0% (20) 9.00 (2.25–21.75)
12.5% (9) 8.3% (6) 19.4% (14) 88.5 (47.0–150.5) 27.5 (15.0–111.3) 69.5% (41) 1215.0 (492.5–2900.0) 106.0 (89.5–165.0) 11.1% (7) 23.7% (14) 28.8% (17) 16.4% (10)
10.8% (4) 10.8% (4) 16.2% (6) 96.00 (62.5–172.0) 38.00 (20.0–145.0) 62.5% (15) 2880.0 (1400.0–5635.0) 109.0 (95.25–221.00) 10.7% (3) 40.0% (10) 28.0% (7) 7.7% (2)
Group B (2005–2010) % (n) Mean/Median (SD/IQR) 8.6% (3) 12.00 (8.0–18.0) 14.3% (5) 5.7% (2) 22.9% (8) 81.00 (37.0–155.0) 18.00 (12.0–81.0) 22.9% (8) 680.0 (310.0–1400.0) 104.0 (87.0–143.0) 11.4% (4) 11.8% (4) 29.4% (10) 22.9% (8)
p <0.01 n/s n/s n/s n/s n/s 0.02 n/s <0.01 n/s n/s 0.01 n/s n/s
55.9% (33) 15.4% (10)
54.2% (13) 20% (6)
57.1% (20) 11.4% (4)
n/s n/s
19.1% (13) 16.2% (11) 5.9% (4) 4.4% (3) 8.8% (6)
12.1% (4) 21.2% (7) 0% (0) 3.0% (1) 15.2% (5)
25.7% (9) 11.4% (4) 11.4% (4) 5.7% (2) 2.9% (1)
n/s n/s 0.04 n/s n/s
5.0 (4.0–11.0) 8.75 (±8.42) 2.0 (1.0–4.0)
5.00 (4.00–14.00) 9.66 (±10.49) 3.00 (0.00–10.50)
5.00 (2.75–10.25) 7.97 (±6.12) 2.00 (1.00–3.25)
n/s n/s 0.02
Continuous Variable – expressed as mean ± Standard Deviation (SD). Skewed Continuous variable expressed as median and Interquartile Range (IQR). All other data is categorical – expressed as number (n) and percentage of valid cases (%). n/s = difference between early and late series not statistically significant (p > 0.05). ICU = Intensive Care Unit; PND = Permanent Neurological Dysfunction; TND = Transient Neurological Dysfunction; RLN = Recurrent Laryngeal Nerve; PRBC = Packed Red Blood Cells; FFP = Fresh Frozen Plasma; LRTI = Lower Respiratory Tract Infection; APO = Acute Pulmonary Oedema.
a
b
over two per year in the earlier patients of this series and the 10 [3] to 25 [22] per year of other contemporary series. This increase in volume has led to an increase in practical experience, allowing greater fine tuning and consistency in the procedures for looking after these highly complex
patients, both in the operating theatre and post operatively. The fact that some of the best reported outcomes come from studies involving large numbers of patients in relatively short periods of time [15] indicates that high volumes are associated with good results. This reinforces
Table 4. Positive Univariate Predictors. Variable
Odds Ratio
Agea Emergency Aneurysm Dissection CCF Hypertension Myocardial Infarct Family History of IHD Pre Op Creatinine >200 Previous Carotid/CVA CVD PVD Circ Arrest Timea ACP and RCP Combined ACP Only Lowest NP Temp. (◦ C)a
1.056 3.867 0.349 5.352 n/c n/c 10.071 6.750 n/c 6.267 6.000 8.556 1.041 14.167 0.293 0.754
a
95% CI 1.010–1.104 1.290–11.589 0.121–1.006 1.796–15.951 n/c n/c 0.970–104.609 0.994–45.854 n/c 1.042–37.689 1.831–19.660 2.063–35.481 1.014–1.069 1.549–129.560 0.102–0.840 0.637–0.894
p 0.003 0.013 0.047 0.002 0.005 0.002 0.022 0.03 0.001 0.027 0.002 0.001 0.004 0.004 0.019 <0.001
Continuous Variable, odds ratio expressed as per unit increase in variable. n/c = Odds Ratio incalculable due to zero patients in one or more categories. CCF = Congestive Cardiac Failure; IHD = Ischaemic Heart Disease; CVA = Cerebrovascular Accident; CVD = Cardiovascular Disease; PVD = Peripheral Vascular Disease; NP = Nasophayngeal; ACP = Antrograde Cerebral Perfusion; RCP = Retrograde Cerebral Perfusion.
ORIGINAL ARTICLE
Heart, Lung and Circulation 2011;20:704–711
ORIGINAL ARTICLE
710
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
Heart, Lung and Circulation 2011;20:704–711
Table 5. Multivariate Predictors. Variable Significant Predictors CVD Circ Arrest Time (min)a Other Predictors in Final Equation Hypertension Pre Op creatinine >200 Other Predictors Entered in Model Agea Emergency Dissection
Odds Ratio
95% CI
11.106 1.066
p
1.437–85.816 1.015–1.120
0.021 0.011
– –
– –
0.998 0.999
– – –
– – –
– – –
a
Continuous Variable, odds ratio expressed as per unit increase in variable. CVD = Cardiovascular Disease.
the idea that arch replacement should be conducted in specialised centres performing a relatively high volume of cases. Patients in Group A were also of much higher risk than the patients in Group B. There were significantly more elective operations and significantly less acute dissections in the later patients. The later patients also happened to show a trend towards lower prevelance for most of the comorbidities. The fact that there are generally fewer cases of the type typically regarded as especially high risk would also help explain the improved results, as well as the poorer results seen in earlier patients. The increase in numbers, especially in the number of elective cases, supports the idea that patients who were not previously offered aortic arch replacements are now being offered surgery. Contributing factors for our unit’s more aggressive approach to surgery include the improving outcomes of surgery, since operations are only offered if the risk of the procedure is outweighed by the risks of the underlying disease. Hence with reduced risks of an operation, more patients become eligible for arch replacement, giving rise to greater number of elective, as opposed to the more emergent and therefore higher risk cases than in previous eras. There are limitations to this current study which impact on how strongly conclusions can be drawn. It is a retrospective study based partially on prospectively collected data therefore it shares the limitations of all similar studies. Part of this includes missing data that could lead to potential biases in comparing earlier and later patients. Our study also includes the experience of multiple surgeons, using many changing techniques over a relatively long period of time. This results in great heterogeneity between the included cases. Whilst this is an advantage in allowing the overall effect of the improved techniques over time to be shown in a ‘real world’ manner, it makes it difficult to ascertain the exact causes of changing results. To overcome these limitations a prospective randomised trial between specific techniques is required, although this would be very difficult to practically accomplish at the rates, and in the circumstances, that aortic arch replacements are typically done. Areas of especial interest for further research include the relative merits of branched versus trifurcated grafts, the benefits of axillary perfusion vs other ACP strategies and determining the optimal level of hypothermia for use with modern techniques.
Another limitation to the study is the relatively small numbers of patients. This reduced the statistical power of the study, especially in regards to constructing the multivariate model. This meant that not all the univariate predictors could be included in this model, thereby reducing its validity in producing truly independent predictors of mortality. Instead the seven most clinically and statistically significant univariate predictors were chosen for inclusion. Since circulatory arrest time and lowest nasopharyngeal temperature were highly correlated, as mentioned in the results it was decided to only include circulatory arrest time in the model. Therefore the model is unable to determine what relative effect each of those two variables has in predicting mortality. However it is consistent with the great improvement in results over time with changed techniques, which have allowed operations which minimise both the period of circulatory arrest and the degree of cooling required. The circulatory arrest time is also biased due to the more complex nature of, and pathology in, some of the procedures with long circulatory arrest times. These procedures would be independently associated with a more adverse outcome due to their nature as well as the clinical effects of a long circulatory arrest. Due to the limitations in its construction, which are intrinsic to this study, the model should be regarded as an attempt try to explain some of the predictors of mortality whilst accounting for some, but not all, of the confounding factors.
Conclusion Our experience has shown that the improvements in technique over the last two decades, as well as increased experience, have led to greatly improved results from aortic arch surgery. Key amongst these improvements have been antegrade cerebral perfusion, higher core body temperatures, use of factor VIIa and prefabricated aortic arch grafts, especially multibranched grafts. In combination these techniques have allowed complete replacement of the diseased aortic arch whilst minimising the time the brain is without perfusion. These improvements have also led to aortic arch replacement being a viable option in the more aggressive management of elective surgical patients with aortic aneurysms affecting the aortic arch. We believe that future research is vital, especially in determining the relative benefits of the different forms of antegrade
Davies et al. Evolution in the Techniques and Outcomes of Aortic Arch Surgery: A 22 Year Single Centre Experience
cerebral perfusion and the different graft designs, as well as determining the outcomes of patients both in the long term and in different at risk groups.
Acknowledgements The authors would like to thank Dr. John Wynter and Prof. Richard Walsh for their assistance in collating perfusion chart data and Lisa Turner for her assistance in management of the departmental database and ethics application. There is no external financial support or conflicts of interest for any of the authors.
References [1] De Bakey M, Crawford E, Cooley D, Morris Jr G. Successful resection of fusiform aneurysm of aortic arch with replacement by homograft. Surg Gynecol Obstet 1957;105(6): 657. [2] De Bakey ME, Henly WS, Cooley DA, Crawford ES, Morris Jr GC, Beall Jr AC. Aneurysms of the aortic arch: factors influencing operative risk. Surg Clin North Am 1962;42:1543–54. [3] Sundt 3rd TM, Orszulak TA, Cook DJ, Schaff HV. Improving results of open arch replacement. Ann Thorac Surg 2008;86(3):787–96. [4] Kazui T, Yamashita K, Washiyama N, Terada H, Bashar AHM, Suzuki K, et al. Aortic arch replacement using selective cerebral perfusion. Ann Thorac Surg 2007;83(2):S796–8. [5] Griepp R, Stinson E, Hollingsworth J, Buehler D. Prosthetic replacement of the aortic arch. J Thorac Cardiovasc Surg 1975;70(6):1051. [6] Svensson L, Crawford E, Hess K, Coselli J, Raskin S, Shenaq S, et al. Deep hypothermia with circulatory arrest. Determinants of stroke and early mortality in 656 patients. J Thorac Cardiovasc Surg 1993;106(1):19–28. [7] Milewski RK, Pacini D, Moser GW, Moeller P, Cowie D, Szeto WY, et al. Retrograde and antegrade cerebral perfusion: results in short elective arch reconstructive times. Ann Thorac Surg 2010;89(5):1448–57. [8] Ueda Y, Okita Y, Aomi S, Koyanagi H, Takamoto S. Retrograde cerebral perfusion for aortic arch surgery: analysis of risk factors. Ann Thorac Surg 1999;67(6):1879–82, discussion 91–4. [9] Usui A, Abe T, Murase M. Early clinical results of retrograde cerebral perfusion for aortic arch operations in Japan. Ann Thorac Surg 1996;62(1):94–103. [10] Ehrlich MP, Hagl C, McCullough JN, Zhang N, Shiang H, Bodian C, et al. Retrograde cerebral perfusion provides negligible flow through brain capillaries in the pig. J Thorac Cardiovasc Surg 2001;122(2):331–8. [11] Moon MR, Sundt 3rd TM. Influence of retrograde cerebral perfusion during aortic arch procedures. Ann Thorac Surg 2002;74(2):426–31. [12] Okita Y, Minatoya K, Tagusari O, Ando M, Nagatsuka K, Kitamura S. Prospective comparative study of brain protection in total aortic arch replacement: deep hypothermic circulatory arrest with retrograde cerebral perfusion or selective antegrade cerebral perfusion. Ann Thorac Surg 2001;72(1):72–9.
711
[13] Bischoff MS, Brenner RM, Scheumann J, Bodian CA, Griepp RB, Lansman SL, et al. Long-term outcome after aortic arch replacement with a trifurcated graft. J Thorac Cardiovasc Surg 2010;140(6 Suppl.):S71–6. [14] Strauch JT, Spielvogel D, Lauten A, Galla JD, Lansman SL, McMurtry K, et al. Technical advances in total aortic arch replacement. Ann Thorac Surg 2004;77(2):581–9. [15] Minatoya K, Ogino H, Matsuda H, Sasaki H, Tanaka H, Kobayashi J, et al. Evolving selective cerebral perfusion for aortic arch replacement: high flow rate with moderate hypothermic circulatory arrest. Ann Thorac Surg 2008;86(6):1827–31. [16] Touati GD, Marticho P, Farag M, Carmi D, Szymanski C, Barry M, et al. Totally normothermic aortic arch replacement without circulatory arrest. Eur J Cardiothorac Surg 2007;32(2):263–8. [17] Crawford ES, Saleh SA. Transverse aortic arch aneurysm: improved results of treatment employing new modifications of aortic reconstruction and hypothermic cerebral circulatory arrest. Ann Surg 1981;194(2):180–8. [18] Sakamoto S, Matsubara J, Nagayoshi Y, Nishizawa H, Takeuchi K, Nonaka T, et al. Clinical results of aortic arch replacement using a four branched prosthetic graft. J Cardiovasc Surg 2003;44(6):751–5. [19] Kazui T, Washiyama N, Muhammad B, Terada H, Yamashita K, Takinami M, et al. Total arch replacement using aortic arch branched grafts with the aid of antegrade selective cerebral perfusion. Ann Thorac Surg 2000;70(1):3. [20] Spielvogel D, Strauch JT, Minanov OP, Lansman SL, Griepp RB. Aortic arch replacement using a trifurcated graft and selective cerebral antegrade perfusion. Ann Thorac Surg 2002;74(5):S1810–4. [21] Spielvogel D, Halstead JC, Meier M, Kadir I, Lansman SL, Shahani R, et al. Aortic arch replacement using a trifurcated graft: simple, versatile, and safe. Ann Thorac Surg 2005;80(1):90–5. [22] Spielvogel D, Etz CD, Silovitz D, Lansman SL, Griepp RB. Aortic arch replacement with a trifurcated graft. Ann Thorac Surg 2007;83(2):S791–5. [23] Borst H, Frank G, Schaps D. Treatment of extensive aortic aneurysms by a new multiple-stage approach. J Thorac Cardiovasc Surg 1988;95(1):11. [24] Ergin MA, O’Connor J, Guinto R, Griepp RB. Experience with profound hypothermia and circulatory arrest in the treatment of aneurysms of the aortic arch. Aortic arch replacement for acute arch dissections. J Thorac Cardiovasc Surg 1982;84(5):649–55. [25] Ehrlich M, Grabenwoger M, Luckner D, Simon P, Laufer G, Wolner E, et al. Operative management of aortic arch aneurysm using profound hypothermia and circulatory arrest. J Cardiovasc Surg 1996;37(6 Suppl. 1):63–4. [26] Schwartz JP, Bakhos M, Patel A, Botkin S, Neragi-Miandoab S, Schwartz JP, et al. Repair of aortic arch and the impact of cross-clamping time, New York Heart Association stage, circulatory arrest time, and age on operative outcome. Interact Cardiovasc Thorac Surg 2008;7(3):425–9. [27] Wilde JT. Hematological consequences of profound hypothermic circulatory arrest and aortic dissection. J Card Surg 1997;12(2 Suppl.):201–6.
ORIGINAL ARTICLE
Heart, Lung and Circulation 2011;20:704–711