Outcome of open total arch replacement in the modern era Fabrizio Settepani, MD, Antioco Cappai, MD, Alessio Basciu, MD, Alessandro Barbone, MD, and Giuseppe Tarelli, MD, Rozzano, Milan, Italy Objective: To shed light on contemporary results of open total aortic arch surgery, we undertook a systematic review to identify all reports on this procedure published in the last 10 years. Methods: Extensive electronic literature search was undertaken to identify all published articles from 2004 to 2014 that provided results on total aortic arch replacement. According to inclusion and exclusion criteria, 21 relevant studies were selected and meta-analyzed to assess outcomes. Results: The pooled estimate for operative mortality was 5.3%. Permanent and transient neurologic deficit occurred postoperatively at a pooled rate of 3.4% and 5.2%, respectively. Pooled rate of irreversible spinal cord injury was 0.6%, whereas renal failure occurred at a pooled rate of 4.1%. Prolonged intubation occurred at pooled rate of 15.4%. Among elective patients, pooled rate of mortality and permanent neurologic deficit was 2.9% and 2.2%, respectively, with a significant difference compared with urgent/emergency surgery cases. Conclusions: The main findings from this meta-analysis indicate that total aortic arch replacement can be performed with satisfactory mortality and morbidity. The pooled rates of mortality and permanent neurologic deficit among elective cases were surprisingly low, and these data have an even greater prominence when they are compared with outcomes of hybrid arch series. Under urgent/emergency surgery, early mortality and neurologic complications showed an about threefold higher rate. Moderate hypothermic circulatory arrest and early rewarming seem to provide proper renal protection, with an intermediate risk of prolonged intubation. (J Vasc Surg 2016;63:537-45.)
Total aortic arch replacement (TAR) is still to be considered a challenging procedure, although, in recent years, results have substantially improved. The keys to success are the new developments in both surgical and brain protection techniques.1,2 Hybrid arch procedures that combine open brachiocephalic debranching with concomitant antegrade endovascular stent graft placement in the aortic arch in a single-stage procedure have emerged as a treatment option for complicated pathologic processes in high-risk patients.3,4 To shed light on contemporary results of open total aortic arch surgery, we undertook a systematic review to identify all reports on this procedure published in the last 10 years. Eligible studies were combined into an extensive meta-analysis to assess the results. METHODS Search strategy. In December 2014, we queried the PubMed database using the following keywords: total [all fields] AND (“aorta, thoracic” [MeSH Terms] OR (“aorta” [All Fields] AND “thoracic” [All Fields]) OR From the Department of Cardiac Surgery, Humanitas Clinical and Research Center. Author conflict of interest: none. Correspondence: Fabrizio Settepani, MD, Department of Cardiac Surgery, Humanitas Clinical Research Center, Via Manzoni, 56, Rozzano, Milano 20089, Italy (e-mail: [email protected]). The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. 0741-5214 Copyright Ó 2016 by the Society for Vascular Surgery. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jvs.2015.10.061
“thoracic aorta” [All Fields] OR (“aortic” [All Fields] AND “arch” [All Fields]) OR “aortic arch” [All Fields] AND (“reimplantation” [MeSH Terms] OR “reimplantation [All Fields] OR “replacement” [All Fields]). The database was searched from January 2004 to date. Articles were also identified using the function “related articles.” In addition, bibliographies of retrieved papers were searched. Criteria for inclusion and exclusion. Only studies that met the following conditions were considered eligible for this meta-analysis: 1. Reported on a series of total open arch replacement; 2. Reported on a series of at least 25 patients to prevent bias arising from a small sample population; 3. Provided baseline characteristics of the recruited patients; 4. Stated the incidence of at least one of the basic outcome criteria; and 5. Had a publication date later than January 2004. Exclusion criteria included the following: articles in languages other than English, series including partial arch (hemiarch) replacement, series including associated extensive descending thoracic aortic replacement, and series including hybrid procedures. Statistical analysis. Descriptive statistics, reported as weighted means with 95% confidence intervals (CIs), were used to summarize demographic and baseline data of the recruited patients from all eligible studies. Metaanalysis was conducted in accordance with the recommendation of the Meta-analysis of the Observational Studies in Epidemiology (MOOSE) group.5 The primary 537
JOURNAL OF VASCULAR SURGERY February 2016
538 Settepani et al
Table. Descriptive characteristics of eligible studies Author
Study period
No. of patients
Mean age, years
Shrestha et al Ochiai et al Minakawa et al Di Eusanio et al Kondoh et al Spielvogel et al Sasaki et al Schwartz et al Shimizu et al Kulik et al Kaku et al Okita et al Tanaka et al Nishi et al Leshnower et al Choi et al Czerny et al Matsuyama et al Fukui et al Iwasaki et al Nota et al
1982-2012 1988-2003 1989-2006 1995-2002 1998-2011 1999-2005 2000-2005 2001-2004 2001-2011 2002-2010 2002-2011 2002-2012 2003-2011 2004-2011 2004-2012 2004-2012 2005-2012 2006-2008 2006-2010 2008-2009 2008-2013
179 46 122 352 127 150 305 32 203 88 107 423 146 61 145 26 39 119 56 38 116
56.4 61.8 65 64.9 70 63 73 61 67.9 61.5 70 69.2 66.6 69.8 59 54.7 63 68 73.9 71 72.3
Male, No. (%)
Dissection, No. (%)
Urgent/emergent, No. (%)
112 24 86 178 91 91 238 28 159 38 88 313 90 44 93 18 31 94 41 20 89
88 (49) 46 (100) 68 (56) ND 36 (28) 56 (37) 34 (11) ND 69 (34) 38 (43) 12 (11) 153 (36) 69 (47) 19 (31) ND 26 (31) 14 (36) 55 (46) ND 20 (53) 15 (13)
47 (26) 46 (100) 47 (38) 49 (14) ND 0 0 32 (100) 45 (22) ND 14 (13) 135 (32) 61 (42) 14 (23) 23 (16) 26 (100) 0 56 (47) ND 18 (47) 0
(63) (52) (70) (50) (72) (60) (78) (87) (78) (43) (82) (74) (61) (72) (64) (69) (79) (79) (73) (53) (77)
Follow-up, months ND 65 52 ND 46 ND ND ND ND 35 47 29 51 ND 33 54 11 25 ND ND ND
ND, No data.
end points of the meta-analysis were operative mortality,6 permanent neurologic deficit (PND), transient neurologic deficit (TND), spinal cord injuries (SCIs), renal failure requiring dialysis, postoperative bleeding, and prolonged (>48 hours) intubation. To account for undetectable heterogeneity related to the observation design of the studies included, pooled estimates were calculated by using the random-effect model proposed by DerSimonian and Laird.7 Between-study heterogeneity was analyzed by means of I2 index, and value >75% was considered to be indicative of high heterogeneity. To deal with potential selection bias resulting from the retrospective nature of the source studies, we performed separate subanalysis of more homogeneous papers. RESULTS Overview of the studies included. We identified 141 citations meeting our search criteria. After exclusion of 18 duplicate citations, 123 articles were evaluated. Of these, 102 were excluded in a subsequent evaluation on the basis of inclusion and exclusion criteria. Twenty-one articles were deemed eligible for this systematic review,1,8-28 with a total of 2880 patients (Table). All of the included manuscripts were retrospective observational studies without controls (level of evidence 4, according to the Oxford Center for Evidence-Based Medicine). Demographic characteristics. Descriptive characteristics of eligible studies are listed in Table. The entire population underwent TAR. Weighted mean age of the patients was 66.5 6 11.9 years, and the majority of them were male (69.1%). Surgery was performed under emergency/urgent need in 23.4% of the total cases. Aortic disease was dissection in 35.6% of the patients. A trifurcated Dacron graft
was used in the majority of patients (80.4%). TAR was performed under deep, moderate, and mild hypothermic circulatory arrest (HCA) in 15.3%, 80.5%, and 4% of the patients, respectively. All the studies but one23 used antegrade selective cerebral perfusion (ASCP). Weighted averages of cardiopulmonary bypass (CPB) time, myocardial ischemic time, HCA time, and ASCP time were 216.4 6 65.1 minutes (range, 151-331 minutes), 120.8 6 44.8 minutes (range, 85.1-182 minutes), 51.9 6 20.4 minutes (range, 11-118 minutes), and 106.8 6 46.3 minutes (range, 40.4-166 minutes), respectively (Fig 1). Mean rectal temperature was 24.7 C 6 2.7 C (data provided by 10 of 21 studies1,8,9,14,17,21,22,26,27). Among studies reporting on late outcomes (11 of 21),1,8,10,11,17,19,20,24-28 follow-up ranged from 11 to 65 months. Mortality and morbidity analysis. The pooled estimate for overall operative mortality was 5.3% (95% CI, 3.9%-6.6%; Fig 2). A PND was found to occur postoperatively at a pooled rate of 3.4% (95% CI, 2.7%-4.2%; Fig 3). Signs and symptoms referable to TND, data provided by 15 of 21 studies,1,8,11,12,14,16,19-21,23-28 were found at a pooled rate of 5.2% (95% CI, 3.7%-6.6%). Data regarding irreversible SCI were available in 18 studies,1,8-11,13-17,19-21,24-28 with a pooled rate of 0.6% (95% CI, 0.3%-1%). When studies were excluded in which a long (>15 cm) elephant trunk was used,9,25 irreversible SCI was found at a pooled rate of 0.5% (95% CI, 0.2%-0.9%). Seventeen studies provided data regarding postoperative renal failure, defined as temporary or permanent dialysis, with a pooled rate of 4.1% (95% CI, 3%-5.2%).1,8,10,11,13-17,20-23,25-28 Postoperative bleeding requiring re-exploration, data provided by 16 of 21
JOURNAL OF VASCULAR SURGERY Volume 63, Number 2
Settepani et al 539
Fig 1. Operative durations for eligible studies including (A) cardiopulmonary bypass (CPB) time, (B) myocardial ischemic time, (C) hypothermic circulatory arrest (HCA) time, and (D) selective cerebral perfusion (SCP) time.
studies,8-11,13,15-17,19,21-26,28 was found at a pooled rate of 6.1% (95% CI, 3.8%-8.4%). In sixteen studies, prolonged intubation, defined as intubation time longer than 48 hours, was reported, with a pooled rate of 15.4% (95% CI, 11.5%-19.3%).1,8-11,14-17,20-24,26,28 Finally, eight studies reported data on late death, with a pooled rate of 9.4% (95% CI, 5.4%-13.5%).1,13,17,19,20,24-26 Subanalysis of the outcomes. To deal with potential selection bias resulting from the retrospective nature of the source studies, we performed separate subanalysis of more homogeneous papers. However, results did not change significantly when the five studies including fewer than 50 patients were removed11,13,19,20,23 or when the six studies reporting the earliest series from the 2000s were removed.8,11-13,16,21 To evaluate the impact of urgent/emergency surgery on mortality and PND, we performed separate subanalysis of elective procedures (1732 patients) and urgent/emergency procedures (491 patients). Pooled analysis showed significantly different rates of operative mortality: 2.9% (95% CI, 1.8%-4%) for elective cases vs 8.8% (95% CI, 5.9%-11.8%) for urgent/emergency surgery cases; P < .001; Fig 4. Similarly, pooled analysis showed significantly different rates of postoperative PND: 2.2% (95% CI, 1.4%-3.1%) for elective
cases vs 6.5% (95% CI, 3.3%-9.7%) for urgent/emergency surgery cases (P ¼ .03). Being that the Japanese papers were extremely homogeneous in terms of surgical technique and brain protection strategy, a subanalysis of these was performed.1,8-12,14,17,22-25,28 Pooled analysis showed a considerably different rate of operative mortality of Japanese papers compared with the total sample, 3.8% (95% CI, 2.6%-5.1%) vs 5.3% (95% CI, 3.9%-6.6%), whereas the rate of PND did not differ from the overall data: 3.1% vs 3.4%. When the only Japanese paper not using ASCP as a method of brain protection was excluded,23 PND result did not change significantly. When Japanese papers were compared with “nonJapanese” papers, rates of operative mortality were significantly different: 3.8% (95% CI, 2.6%-5.1%) vs 7.8% (95% CI, 5%-10.6%), respectively (P < .001). This trend was confirmed in elective cases with a rate of 2.0% (95% CI, 1.2%-2.9%) vs 6.7% (95% CI, 4.3%-9.2%), respectively (P ¼ .005); whereas for urgent/emergency cases, the rates were similar: 9.4% (95% CI, 5.4%-13.4%) vs 8% (95% CI, 3.1%-12.9%), respectively (P ¼ .21). To evaluate the impact of brain protection strategy on PND, we performed separate subanalysis of ASCP with
540 Settepani et al
JOURNAL OF VASCULAR SURGERY February 2016
Fig 2. Forest plot of in-hospital mortality. The rate from each included study (squares) and from the pooled estimate (diamond) are plotted, each with 95% confidence interval (CI; line length and width of diamond). Heterogeneity: s2 ¼ 0.001; Q value ¼ 55.5, df ¼ 20 (P < 001); I2 ¼ 64%. Ev/Trt, Observed number of events in the treatment group.
triple supra-aortic vessel perfusion (1981 patients) and ASCP with double supra-aortic vessel perfusion (716 patients). Pooled analysis failed to show significantly different rates of PND between the two techniques: 3.2% (95% CI, 2.8%-4.4%) vs 4.8% (95% CI, 2.4%-7.2%), respectively (P ¼ .25). When rates of SCI were investigated, although considerably different results were found, they did not reach statistical significance: 0.7% (95% CI, 0.3%-1.1%) vs 2.9% (95% CI, 1.2%-4.7%), respectively (P ¼ .19). Metaregression analysis of moderators. Intraoperative durations did not influence effect size estimates. Unexpectedly, the rate of operative mortality was significantly higher for younger patients (P < .001), whereas the PND rate was not related to age (P ¼ .15). DISCUSSION The majority of the published series reporting results on arch surgery include variable but frequently consistent numbers of hemiarch replacement,29,30 which, as known, is a less technically demanding and time-consuming operation compared with TAR. This can generate confusion in terms of outcome analysis, given that nowadays, open arch surgery results are often compared with hybrid arch surgery results.31 Therefore, the purpose of this metaanalysis was to shed light on the outcome of open surgical TAR. To make these results comparable with contemporary series of hybrid procedures, we restricted our analysis to studies published in the last 10 years, and to increase
the statistical power of every single study, we excluded series of fewer than 25 patients. This study shows that the results of open TAR are extremely satisfactory in terms of both mortality and morbidity. The overall early mortality pooled rate, settling at around 5% (moderate heterogeneity), suggests that open arch reconstruction can be performed safely. Analyzing the surgical technique used in the 21 eligible studies, it is interesting to note that there are two important points in common: the cerebral protection method, which was ASCP in adjunct to HCA in all the studies but one23; and the technique of supra-aortic vessel reimplantation, which was the branched graft technique (BGT) in most patients (2591 of 2880). The most common surgical technique used for TAR is the one described by Kazui et al2 in 2001. Briefly, after median sternotomy is performed, with an extension of the incision to the left subclavian region, the ascending aorta and the right atrium are cannulated. The CPB is commenced, and after the rectal temperature has decreased to 22 C, under circulatory arrest, the aortic arch aneurysm is opened. Next, ASCP is initiated by placing a 14F or 16F balloon-tipped cannula into the brachiocephalic artery and a 12F cannula in the left common carotid artery. The ASCP flow is maintained at 10 mL/kg/min and the perfusate at a temperature of 22 C. Then, the descending aorta just below the aneurysm is completely transected, and the distal side of a sealed four-branched arch graft is anastomosed to the stump of the descending aorta by a
JOURNAL OF VASCULAR SURGERY Volume 63, Number 2
Settepani et al 541
Fig 3. Forest plot of permanent neurologic deficit (PND). The rate from each included study (squares) and from the pooled estimate (diamond) are plotted, each with 95% confidence interval (CI; line length and width of diamond). Heterogeneity: s2 ¼ 0.00; Q value ¼ 21.8, df ¼ 20 (P ¼ .25); I2 ¼ 8%. Ev/Trt, Observed number of events in the treatment group.
4-0 monofilament running suture with Teflon felt reinforcement. The arch graft proximal to the fourth limb is cross-clamped, and a systemic perfusion to the lower half of the body is started from the fourth limb of the graft. The third limb is then anastomosed to the left subclavian artery by a 5-0 monofilament running suture. Then, the arch graft proximal to the third limb is cross-clamped, and rewarming by means of CBP is started. The proximal side of the arch graft is sutured to the stump of the ascending aorta by a 4-0 monofilament running suture with Teflon felt reinforcement, and the aortic arch is declamped. The first limb of the arch graft is anastomosed to the brachiocephalic artery with a 5-0 monofilament running suture, and the second limb is anastomosed to the left common carotid artery in the same fashion. After the CPB is terminated, the fourth limb of the arch graft, which is used for antegrade systemic perfusion, is resected. More recently, the cannulation of the right axillary artery as opposed to the ascending aorta and ASCP of all three supra-aortic vessels has become the preferred technique of many aortic surgeons.1,12 The advantages of this method are becoming widely recognized. Antegrade cerebral perfusion, whose optimum flow was established by Tanaka et al32 in 1995, is now considered to be the most reliable brain protection method during circulatory arrest.33,34 Although monohemispheric perfusion can be considered acceptable for short circulatory arrest time, bihemispheric
perfusion allows extensive and meticulous reconstruction of the aortic arch.2,35 Because of the frequent incompleteness of the circle of Willis that has been reported to be as high as 58%,36 the trend among many aortic surgeons is to perfuse not only the brachiocephalic and left common carotid artery but also the left subclavian artery. Furthermore, the left subclavian artery is often a supplier of the collateral vessel to the spinal cord.37 ASCP was achieved by the perfusion of all three supra-aortic vessels in 13 studies,1,8-12,14,16,17,22,24,25,28 whereas in five, the perfusion was limited to the brachiocephalic artery and left common carotid artery.13,15,19-21 In one study, monohemispheric perfusion was performed.27 Finally, in only one study, cerebral protection was achieved exclusively by deep HCA.23 BGT, as a method to reimplant the supraaortic vessels, presents several advantages. Replacing the proximal portion of the arch vessels where clots, atheroma, and calcification are often located can reduce the cerebral embolic risk. As the proximal anastomosis of the ascending aorta can precede the arch anastomosis, CPB time and myocardial ischemic time are shorter compared with the en bloc technique,21 and prolonged pump time is a known risk factor for hospital mortality and morbidity. Finally, by dividing the arch cuff into three buttons, more liberal exposure of the aortic arch can be obtained. PND and TND occurred at a pooled rate of 3.4% (low heterogeneity) and 5.2% (moderate heterogeneity),
542 Settepani et al
JOURNAL OF VASCULAR SURGERY February 2016
Fig 4. Forest plot of in-hospital mortality of elective patients vs urgent/emergency surgery patients. The rate from each included study (squares) and from the pooled estimate (diamond) are plotted, each with 95% confidence interval (CI; line length and width of diamond). Heterogeneity: Q value ¼ 55.3, df ¼ 22 (P ¼ .001); I2 ¼ 60%. Ev/Trt, Observed number of events in the treatment group.
respectively, and considering that 23% of the patients had an urgent/emergency operation, this is to be considered an extremely satisfactory outcome. Subgroup analysis showed a significantly higher pooled rate of mortality and PND among urgent/emergency surgery cases compared with elective. Although this finding, to a certain extent, was expected, it is interesting to note that elective cases showed a surprisingly low pooled rate of mortality and PND (2.9% and 2.2%, respectively), suggesting that in this setting, the procedure can be performed safely. A subanalysis of the 13 Japanese papers was performed, as they were extremely homogeneous in terms of brain protection method and supra-aortic vessel reimplantation technique. In fact, all the studies but one23 used triple supra-aortic vessel ASCP, and the technique of supra-aortic vessel reimplantation was invariably BGT. The pooled rate of overall operative mortality among Japanese studies was as low as 3.8% (and significantly lower that among non-Japanese studies). These data are consistent with a recently published experience based on the Japan Adult Cardiovascular Surgery Database, reporting outcomes of TAR,38 and is possibly explained by the large experience of the Japanese surgeons in this procedure. However, for emergency cases, Japanese studies showed a mortality rate of 9.4%
(nonsignificantly different compared with non-Japanese studies), indicating that even for experienced centers, the gap between elective and emergency mortality is still relatively large. The inverse correlation between operative mortality and age that was found by metaregression is otherwise difficult to explain, and it could be the result of statistical randomness. If we exclude this possibility, we can presume that the younger patients were more likely to involve acute dissection as opposed to aneurysmal disease, which, as known, is affected by a higher operative mortality. Unfortunately, the majority of the studies did not provide the mean age of the patients who underwent urgent procedures, so this hypothesis cannot be validated. The pooled rate of paraplegia,
Total aortic arch replacement (TAR) is still to be considered a challenging procedure, although, in recent years, results have substantially improved. The keys to success are the new developments in both surgical and brain protection techniques.1,2 Hybrid arch procedures that combine open brachiocephalic debranching with concomitant antegrade endovascular stent graft placement in the aortic arch in a single-stage procedure have emerged as a treatment option for complicated pathologic processes in high-risk patients.3,4 To shed light on contemporary results of open total aortic arch surgery, we undertook a systematic review to identify all reports on this procedure published in the last 10 years. Eligible studies were combined into an extensive meta-analysis to assess the results. METHODS Search strategy. In December 2014, we queried the PubMed database using the following keywords: total [all fields] AND (“aorta, thoracic” [MeSH Terms] OR (“aorta” [All Fields] AND “thoracic” [All Fields]) OR From the Department of Cardiac Surgery, Humanitas Clinical and Research Center. Author conflict of interest: none. Correspondence: Fabrizio Settepani, MD, Department of Cardiac Surgery, Humanitas Clinical Research Center, Via Manzoni, 56, Rozzano, Milano 20089, Italy (e-mail: [email protected]). The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. 0741-5214 Copyright Ó 2016 by the Society for Vascular Surgery. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jvs.2015.10.061
“thoracic aorta” [All Fields] OR (“aortic” [All Fields] AND “arch” [All Fields]) OR “aortic arch” [All Fields] AND (“reimplantation” [MeSH Terms] OR “reimplantation [All Fields] OR “replacement” [All Fields]). The database was searched from January 2004 to date. Articles were also identified using the function “related articles.” In addition, bibliographies of retrieved papers were searched. Criteria for inclusion and exclusion. Only studies that met the following conditions were considered eligible for this meta-analysis: 1. Reported on a series of total open arch replacement; 2. Reported on a series of at least 25 patients to prevent bias arising from a small sample population; 3. Provided baseline characteristics of the recruited patients; 4. Stated the incidence of at least one of the basic outcome criteria; and 5. Had a publication date later than January 2004. Exclusion criteria included the following: articles in languages other than English, series including partial arch (hemiarch) replacement, series including associated extensive descending thoracic aortic replacement, and series including hybrid procedures. Statistical analysis. Descriptive statistics, reported as weighted means with 95% confidence intervals (CIs), were used to summarize demographic and baseline data of the recruited patients from all eligible studies. Metaanalysis was conducted in accordance with the recommendation of the Meta-analysis of the Observational Studies in Epidemiology (MOOSE) group.5 The primary 537
JOURNAL OF VASCULAR SURGERY February 2016
538 Settepani et al
Table. Descriptive characteristics of eligible studies Author
Study period
No. of patients
Mean age, years
Shrestha et al Ochiai et al Minakawa et al Di Eusanio et al Kondoh et al Spielvogel et al Sasaki et al Schwartz et al Shimizu et al Kulik et al Kaku et al Okita et al Tanaka et al Nishi et al Leshnower et al Choi et al Czerny et al Matsuyama et al Fukui et al Iwasaki et al Nota et al
1982-2012 1988-2003 1989-2006 1995-2002 1998-2011 1999-2005 2000-2005 2001-2004 2001-2011 2002-2010 2002-2011 2002-2012 2003-2011 2004-2011 2004-2012 2004-2012 2005-2012 2006-2008 2006-2010 2008-2009 2008-2013
179 46 122 352 127 150 305 32 203 88 107 423 146 61 145 26 39 119 56 38 116
56.4 61.8 65 64.9 70 63 73 61 67.9 61.5 70 69.2 66.6 69.8 59 54.7 63 68 73.9 71 72.3
Male, No. (%)
Dissection, No. (%)
Urgent/emergent, No. (%)
112 24 86 178 91 91 238 28 159 38 88 313 90 44 93 18 31 94 41 20 89
88 (49) 46 (100) 68 (56) ND 36 (28) 56 (37) 34 (11) ND 69 (34) 38 (43) 12 (11) 153 (36) 69 (47) 19 (31) ND 26 (31) 14 (36) 55 (46) ND 20 (53) 15 (13)
47 (26) 46 (100) 47 (38) 49 (14) ND 0 0 32 (100) 45 (22) ND 14 (13) 135 (32) 61 (42) 14 (23) 23 (16) 26 (100) 0 56 (47) ND 18 (47) 0
(63) (52) (70) (50) (72) (60) (78) (87) (78) (43) (82) (74) (61) (72) (64) (69) (79) (79) (73) (53) (77)
Follow-up, months ND 65 52 ND 46 ND ND ND ND 35 47 29 51 ND 33 54 11 25 ND ND ND
ND, No data.
end points of the meta-analysis were operative mortality,6 permanent neurologic deficit (PND), transient neurologic deficit (TND), spinal cord injuries (SCIs), renal failure requiring dialysis, postoperative bleeding, and prolonged (>48 hours) intubation. To account for undetectable heterogeneity related to the observation design of the studies included, pooled estimates were calculated by using the random-effect model proposed by DerSimonian and Laird.7 Between-study heterogeneity was analyzed by means of I2 index, and value >75% was considered to be indicative of high heterogeneity. To deal with potential selection bias resulting from the retrospective nature of the source studies, we performed separate subanalysis of more homogeneous papers. RESULTS Overview of the studies included. We identified 141 citations meeting our search criteria. After exclusion of 18 duplicate citations, 123 articles were evaluated. Of these, 102 were excluded in a subsequent evaluation on the basis of inclusion and exclusion criteria. Twenty-one articles were deemed eligible for this systematic review,1,8-28 with a total of 2880 patients (Table). All of the included manuscripts were retrospective observational studies without controls (level of evidence 4, according to the Oxford Center for Evidence-Based Medicine). Demographic characteristics. Descriptive characteristics of eligible studies are listed in Table. The entire population underwent TAR. Weighted mean age of the patients was 66.5 6 11.9 years, and the majority of them were male (69.1%). Surgery was performed under emergency/urgent need in 23.4% of the total cases. Aortic disease was dissection in 35.6% of the patients. A trifurcated Dacron graft
was used in the majority of patients (80.4%). TAR was performed under deep, moderate, and mild hypothermic circulatory arrest (HCA) in 15.3%, 80.5%, and 4% of the patients, respectively. All the studies but one23 used antegrade selective cerebral perfusion (ASCP). Weighted averages of cardiopulmonary bypass (CPB) time, myocardial ischemic time, HCA time, and ASCP time were 216.4 6 65.1 minutes (range, 151-331 minutes), 120.8 6 44.8 minutes (range, 85.1-182 minutes), 51.9 6 20.4 minutes (range, 11-118 minutes), and 106.8 6 46.3 minutes (range, 40.4-166 minutes), respectively (Fig 1). Mean rectal temperature was 24.7 C 6 2.7 C (data provided by 10 of 21 studies1,8,9,14,17,21,22,26,27). Among studies reporting on late outcomes (11 of 21),1,8,10,11,17,19,20,24-28 follow-up ranged from 11 to 65 months. Mortality and morbidity analysis. The pooled estimate for overall operative mortality was 5.3% (95% CI, 3.9%-6.6%; Fig 2). A PND was found to occur postoperatively at a pooled rate of 3.4% (95% CI, 2.7%-4.2%; Fig 3). Signs and symptoms referable to TND, data provided by 15 of 21 studies,1,8,11,12,14,16,19-21,23-28 were found at a pooled rate of 5.2% (95% CI, 3.7%-6.6%). Data regarding irreversible SCI were available in 18 studies,1,8-11,13-17,19-21,24-28 with a pooled rate of 0.6% (95% CI, 0.3%-1%). When studies were excluded in which a long (>15 cm) elephant trunk was used,9,25 irreversible SCI was found at a pooled rate of 0.5% (95% CI, 0.2%-0.9%). Seventeen studies provided data regarding postoperative renal failure, defined as temporary or permanent dialysis, with a pooled rate of 4.1% (95% CI, 3%-5.2%).1,8,10,11,13-17,20-23,25-28 Postoperative bleeding requiring re-exploration, data provided by 16 of 21
JOURNAL OF VASCULAR SURGERY Volume 63, Number 2
Settepani et al 539
Fig 1. Operative durations for eligible studies including (A) cardiopulmonary bypass (CPB) time, (B) myocardial ischemic time, (C) hypothermic circulatory arrest (HCA) time, and (D) selective cerebral perfusion (SCP) time.
studies,8-11,13,15-17,19,21-26,28 was found at a pooled rate of 6.1% (95% CI, 3.8%-8.4%). In sixteen studies, prolonged intubation, defined as intubation time longer than 48 hours, was reported, with a pooled rate of 15.4% (95% CI, 11.5%-19.3%).1,8-11,14-17,20-24,26,28 Finally, eight studies reported data on late death, with a pooled rate of 9.4% (95% CI, 5.4%-13.5%).1,13,17,19,20,24-26 Subanalysis of the outcomes. To deal with potential selection bias resulting from the retrospective nature of the source studies, we performed separate subanalysis of more homogeneous papers. However, results did not change significantly when the five studies including fewer than 50 patients were removed11,13,19,20,23 or when the six studies reporting the earliest series from the 2000s were removed.8,11-13,16,21 To evaluate the impact of urgent/emergency surgery on mortality and PND, we performed separate subanalysis of elective procedures (1732 patients) and urgent/emergency procedures (491 patients). Pooled analysis showed significantly different rates of operative mortality: 2.9% (95% CI, 1.8%-4%) for elective cases vs 8.8% (95% CI, 5.9%-11.8%) for urgent/emergency surgery cases; P < .001; Fig 4. Similarly, pooled analysis showed significantly different rates of postoperative PND: 2.2% (95% CI, 1.4%-3.1%) for elective
cases vs 6.5% (95% CI, 3.3%-9.7%) for urgent/emergency surgery cases (P ¼ .03). Being that the Japanese papers were extremely homogeneous in terms of surgical technique and brain protection strategy, a subanalysis of these was performed.1,8-12,14,17,22-25,28 Pooled analysis showed a considerably different rate of operative mortality of Japanese papers compared with the total sample, 3.8% (95% CI, 2.6%-5.1%) vs 5.3% (95% CI, 3.9%-6.6%), whereas the rate of PND did not differ from the overall data: 3.1% vs 3.4%. When the only Japanese paper not using ASCP as a method of brain protection was excluded,23 PND result did not change significantly. When Japanese papers were compared with “nonJapanese” papers, rates of operative mortality were significantly different: 3.8% (95% CI, 2.6%-5.1%) vs 7.8% (95% CI, 5%-10.6%), respectively (P < .001). This trend was confirmed in elective cases with a rate of 2.0% (95% CI, 1.2%-2.9%) vs 6.7% (95% CI, 4.3%-9.2%), respectively (P ¼ .005); whereas for urgent/emergency cases, the rates were similar: 9.4% (95% CI, 5.4%-13.4%) vs 8% (95% CI, 3.1%-12.9%), respectively (P ¼ .21). To evaluate the impact of brain protection strategy on PND, we performed separate subanalysis of ASCP with
540 Settepani et al
JOURNAL OF VASCULAR SURGERY February 2016
Fig 2. Forest plot of in-hospital mortality. The rate from each included study (squares) and from the pooled estimate (diamond) are plotted, each with 95% confidence interval (CI; line length and width of diamond). Heterogeneity: s2 ¼ 0.001; Q value ¼ 55.5, df ¼ 20 (P < 001); I2 ¼ 64%. Ev/Trt, Observed number of events in the treatment group.
triple supra-aortic vessel perfusion (1981 patients) and ASCP with double supra-aortic vessel perfusion (716 patients). Pooled analysis failed to show significantly different rates of PND between the two techniques: 3.2% (95% CI, 2.8%-4.4%) vs 4.8% (95% CI, 2.4%-7.2%), respectively (P ¼ .25). When rates of SCI were investigated, although considerably different results were found, they did not reach statistical significance: 0.7% (95% CI, 0.3%-1.1%) vs 2.9% (95% CI, 1.2%-4.7%), respectively (P ¼ .19). Metaregression analysis of moderators. Intraoperative durations did not influence effect size estimates. Unexpectedly, the rate of operative mortality was significantly higher for younger patients (P < .001), whereas the PND rate was not related to age (P ¼ .15). DISCUSSION The majority of the published series reporting results on arch surgery include variable but frequently consistent numbers of hemiarch replacement,29,30 which, as known, is a less technically demanding and time-consuming operation compared with TAR. This can generate confusion in terms of outcome analysis, given that nowadays, open arch surgery results are often compared with hybrid arch surgery results.31 Therefore, the purpose of this metaanalysis was to shed light on the outcome of open surgical TAR. To make these results comparable with contemporary series of hybrid procedures, we restricted our analysis to studies published in the last 10 years, and to increase
the statistical power of every single study, we excluded series of fewer than 25 patients. This study shows that the results of open TAR are extremely satisfactory in terms of both mortality and morbidity. The overall early mortality pooled rate, settling at around 5% (moderate heterogeneity), suggests that open arch reconstruction can be performed safely. Analyzing the surgical technique used in the 21 eligible studies, it is interesting to note that there are two important points in common: the cerebral protection method, which was ASCP in adjunct to HCA in all the studies but one23; and the technique of supra-aortic vessel reimplantation, which was the branched graft technique (BGT) in most patients (2591 of 2880). The most common surgical technique used for TAR is the one described by Kazui et al2 in 2001. Briefly, after median sternotomy is performed, with an extension of the incision to the left subclavian region, the ascending aorta and the right atrium are cannulated. The CPB is commenced, and after the rectal temperature has decreased to 22 C, under circulatory arrest, the aortic arch aneurysm is opened. Next, ASCP is initiated by placing a 14F or 16F balloon-tipped cannula into the brachiocephalic artery and a 12F cannula in the left common carotid artery. The ASCP flow is maintained at 10 mL/kg/min and the perfusate at a temperature of 22 C. Then, the descending aorta just below the aneurysm is completely transected, and the distal side of a sealed four-branched arch graft is anastomosed to the stump of the descending aorta by a
JOURNAL OF VASCULAR SURGERY Volume 63, Number 2
Settepani et al 541
Fig 3. Forest plot of permanent neurologic deficit (PND). The rate from each included study (squares) and from the pooled estimate (diamond) are plotted, each with 95% confidence interval (CI; line length and width of diamond). Heterogeneity: s2 ¼ 0.00; Q value ¼ 21.8, df ¼ 20 (P ¼ .25); I2 ¼ 8%. Ev/Trt, Observed number of events in the treatment group.
4-0 monofilament running suture with Teflon felt reinforcement. The arch graft proximal to the fourth limb is cross-clamped, and a systemic perfusion to the lower half of the body is started from the fourth limb of the graft. The third limb is then anastomosed to the left subclavian artery by a 5-0 monofilament running suture. Then, the arch graft proximal to the third limb is cross-clamped, and rewarming by means of CBP is started. The proximal side of the arch graft is sutured to the stump of the ascending aorta by a 4-0 monofilament running suture with Teflon felt reinforcement, and the aortic arch is declamped. The first limb of the arch graft is anastomosed to the brachiocephalic artery with a 5-0 monofilament running suture, and the second limb is anastomosed to the left common carotid artery in the same fashion. After the CPB is terminated, the fourth limb of the arch graft, which is used for antegrade systemic perfusion, is resected. More recently, the cannulation of the right axillary artery as opposed to the ascending aorta and ASCP of all three supra-aortic vessels has become the preferred technique of many aortic surgeons.1,12 The advantages of this method are becoming widely recognized. Antegrade cerebral perfusion, whose optimum flow was established by Tanaka et al32 in 1995, is now considered to be the most reliable brain protection method during circulatory arrest.33,34 Although monohemispheric perfusion can be considered acceptable for short circulatory arrest time, bihemispheric
perfusion allows extensive and meticulous reconstruction of the aortic arch.2,35 Because of the frequent incompleteness of the circle of Willis that has been reported to be as high as 58%,36 the trend among many aortic surgeons is to perfuse not only the brachiocephalic and left common carotid artery but also the left subclavian artery. Furthermore, the left subclavian artery is often a supplier of the collateral vessel to the spinal cord.37 ASCP was achieved by the perfusion of all three supra-aortic vessels in 13 studies,1,8-12,14,16,17,22,24,25,28 whereas in five, the perfusion was limited to the brachiocephalic artery and left common carotid artery.13,15,19-21 In one study, monohemispheric perfusion was performed.27 Finally, in only one study, cerebral protection was achieved exclusively by deep HCA.23 BGT, as a method to reimplant the supraaortic vessels, presents several advantages. Replacing the proximal portion of the arch vessels where clots, atheroma, and calcification are often located can reduce the cerebral embolic risk. As the proximal anastomosis of the ascending aorta can precede the arch anastomosis, CPB time and myocardial ischemic time are shorter compared with the en bloc technique,21 and prolonged pump time is a known risk factor for hospital mortality and morbidity. Finally, by dividing the arch cuff into three buttons, more liberal exposure of the aortic arch can be obtained. PND and TND occurred at a pooled rate of 3.4% (low heterogeneity) and 5.2% (moderate heterogeneity),
542 Settepani et al
JOURNAL OF VASCULAR SURGERY February 2016
Fig 4. Forest plot of in-hospital mortality of elective patients vs urgent/emergency surgery patients. The rate from each included study (squares) and from the pooled estimate (diamond) are plotted, each with 95% confidence interval (CI; line length and width of diamond). Heterogeneity: Q value ¼ 55.3, df ¼ 22 (P ¼ .001); I2 ¼ 60%. Ev/Trt, Observed number of events in the treatment group.
respectively, and considering that 23% of the patients had an urgent/emergency operation, this is to be considered an extremely satisfactory outcome. Subgroup analysis showed a significantly higher pooled rate of mortality and PND among urgent/emergency surgery cases compared with elective. Although this finding, to a certain extent, was expected, it is interesting to note that elective cases showed a surprisingly low pooled rate of mortality and PND (2.9% and 2.2%, respectively), suggesting that in this setting, the procedure can be performed safely. A subanalysis of the 13 Japanese papers was performed, as they were extremely homogeneous in terms of brain protection method and supra-aortic vessel reimplantation technique. In fact, all the studies but one23 used triple supra-aortic vessel ASCP, and the technique of supra-aortic vessel reimplantation was invariably BGT. The pooled rate of overall operative mortality among Japanese studies was as low as 3.8% (and significantly lower that among non-Japanese studies). These data are consistent with a recently published experience based on the Japan Adult Cardiovascular Surgery Database, reporting outcomes of TAR,38 and is possibly explained by the large experience of the Japanese surgeons in this procedure. However, for emergency cases, Japanese studies showed a mortality rate of 9.4%
(nonsignificantly different compared with non-Japanese studies), indicating that even for experienced centers, the gap between elective and emergency mortality is still relatively large. The inverse correlation between operative mortality and age that was found by metaregression is otherwise difficult to explain, and it could be the result of statistical randomness. If we exclude this possibility, we can presume that the younger patients were more likely to involve acute dissection as opposed to aneurysmal disease, which, as known, is affected by a higher operative mortality. Unfortunately, the majority of the studies did not provide the mean age of the patients who underwent urgent procedures, so this hypothesis cannot be validated. The pooled rate of paraplegia,
