- Email: [email protected]
Basic Data From 176 Studies on the Immediate Outcome After Aortic Valve Replacement With or Without Coronary Artery Bypass Surgery
Basic Data From 176 Studies on the Immediate Outcome After Aortic Valve Replacement With or Without Coronary Artery Bypass Surgery Fausto Biancari, MD, PhD,* Marta Martin, MD,* Giulia Bordin, MD,* Elia Vettore, MD,* Giulia Vinco, MD,* Vesa Anttila, MD, PhD,* Juhani Airaksinen, MD, PhD,† and Francesco Vasques, MD* Objective: The aim of this study was to summarize the immediate outcome after aortic valve replacement (AVR) with or without coronary artery bypass grafting (CABG). Design: Systematic review and meta-analysis. Setting: University hospitals. Participants: Participants were 683,286 patients who underwent AVR with or without CABG. Patients undergoing other major cardiac procedures were excluded from this analysis. Interventions: AVR with or without CABG. Measurements and Main Results: Operative mortality after AVR with or without concomitant CABG was 4.3%, stroke 2.1%, pacemaker implantation 5.9%, and dialysis 2.2%. After isolated AVR, operative mortality was 3.3%, stroke 1.7%, pacemaker implantation 3.3%, and dialysis 1.6%. Mortality was increased among very elderly (o60 years: 3.3%, 60-69 years: 2.7%, 70-79 years: 3.8%, Z80 years: 6.1%, p o 0.001). Prevalence of minimally invasive AVR (mini-AVR) was associated with significantly lower operative mortality (p ¼ 0.039, 46 studies). Mini-AVR only
tended toward lower mortality when included in metaregression analysis as a dichotomous variable (mini-AVR 4,367 patients: 2.3%, 95% CI 1.8-2.9% v full sternotomy 11,076 patients: 3.5%, 95% CI 28-4.1%, p ¼ 0.088). Operative mortality after AVR plus CABG was 5.5% (versus isolated AVR: p o 0.001), stroke 3.0%, pacemaker implantation 3.9%, and dialysis 5.6%. Mortality was high in all age strata, particularly among very elderly (mean age o70 years: 4.8%, mean age 70-79 years: 4.7%; mean age Z80 years: 8.4%, p ¼ 0.002). Conclusions: Isolated AVR is associated with low mortality and morbidity. Coronary artery disease requiring concomitant CABG increases the operative mortality. Patients requiring AVR and CABG should be the main target of lessinvasive treatment strategies. & 2013 Elsevier Inc. All rights reserved.
A
surgery were used to broaden the search. The authors applied the guidelines from Stroup et al in Meta-analysis of Observational Studies in Epidemiology (MOOSE).6
ORTIC STENOSIS (AS) is one of the most frequent cardiac diseases among the elderly.1 The expected significant growth of the number of elderly during the next decades may therefore lead to an increase of the number of patients requiring treatment of AS. Operative treatment of AS is a rapidly evolving field as, currently, the exposure to cardiopulmonary bypass and its related adverse events can be avoided or decreased by the use of transcatheter aortic valve replacement (TAVI)2 and sutureless aortic valve prostheses.3 Even if conventional aortic valve replacement (AVR) is considered the gold standard,4 particularly because of its well-proven long-term durability, major concern exists regarding the perceived high risk of immediate postoperative morbidity and mortality. Therefore, less-invasive technology recently has been embraced with enthusiasm, also in low-risk patients,5 even if there is no evidence of its durability. The lack of data effectively summarizing the immediate outcome of patients undergoing conventional AVR may induce such a shift in treatment modalities. This pooled analysis was planned to summarize the results of recent studies on this topic in order to validate the early efficacy of AVR and to identify those patients who may benefit most from less-invasive treatment methods. Furthermore, the aim from less-invasive was to better define the prognostic impact of concomitant CABG and the potential benefits of the minimally invasive technique in patients undergoing AVR. MATERIAL AND METHODS
Study Selection An English language literature review was performed through PubMed up to July 2012 for any study evaluating the outcome after treatment of aortic valve diseases. The words employed in the search were “aortic valve replacement” and “aortic valve and prosthesis.” The reference list of obtained articles and of journals dealing with cardiac
KEY WORDS: cardiac surgery, aortic valve, coronary artery bypass, risk, outcome, AVR + CABF, Mini-AVR, TAVI, aortic stenosis, transcatheter aortic valve implantation
Data Extraction Data were retrieved only from the articles, and no attempt to get missing data from the authors was made. Each study was evaluated independently by 6 coauthors (F.B., F.V., M.M., G.B, E.V., and G.V.) for inclusion or exclusion from this analysis. Then, the following data were extracted: Authors, year of publication, study period and mid-date of study, study design, prevalence of female gender, cerebrovascular disease, pulmonary disease, peripheral vascular disease, renal failure, prior cardiac surgery, urgent/emergency surgery, minimally invasive surgery (either ministernotomy or minithoracotomy) as well as mean values of the patients’s age, left ventricular ejection fraction, logistic EuroSCORE, duration of aortic cross-clamping and cardiopulmonary bypass. The main treatment modality was defined, whenever data were available, according to the intention-to-treat principle.
Inclusion Criteria To enter this analysis, studies had to fulfil the following criteria: (1) prospective or retrospective observational studies including at least 100 patients undergoing AVR with or without CABG, (2) studies published in the English language, (3) studies published as a full article, (4) studies reporting on at least immediate postoperative mortality,
From the *Department of Surgery, Oulu University Hospital, Oulu, Finland; and †Heart Center, Turku University Hospital, Turku, Finland. Address reprint requests to Fausto Biancari, MD, PhD, Oulu University Hospital, Department of Surgery, Kajaanintie 50, PL 21, 90029 Oulu, Finland. E-mail: [email protected] © 2013 Elsevier Inc. All rights reserved. 1053-0770/2601-0001$36.00/0 http://dx.doi.org/10.1053/j.jvca.2013.07.020
Journal of Cardiothoracic and Vascular Anesthesia, Vol ], No ] (Month), 2013: pp ]]]–]]]
1
2
BIANCARI ET AL
Potentially relevant studies identified and screened n = 3909 Non-pertinent studies n = 2232 Studies with no extractable data n = 274 Specific data not available n = 156 Case reports/ technical reports n = 423 Comments, letters and editorials n = 39 Reviews or metaanalyses n = 105 Studies considering fewer 100 patients n = 449 Duplicate studies n = 55
Studies included in the present analysis n = 176
Fig 1.
Flowchart summarizing literature search and selection of studies.
(5) studies published after 1999, and (6) studies including adult patients. The language of the articles was defined as reported in PubMed. The authors did not include in the present analysis unpublished data in the present analysis or reported only in abstract.
Exclusion Criteria The following exclusion criteria were used in this study: (1) studies including only or mostly pediatric patients, (2) studies reporting on any major associated procedure other than CABG and whose specific data could not be retrieved from the article, (3) studies reporting unclear data (for example, data not matching each other between the text and tables), (4) studies without enough information on treatment modalities and immediate postoperative outcome, (5) studies reporting only on aortic
valve endocarditis, and (6) patients who underwent transcatheter aortic valve implantation or any sutureless aortic valve implantation.
Outcome of Interest and Definitions The main outcome measure of this study was immediate postoperative mortality as occurred during the in-hospital stay and/or 30day postoperative period. Secondary outcome endpoints were stroke, de novo dialysis, atrioventricular block requiring pacemaker implantation, and re-exploration for excessive bleeding.
Statistical Analysis Statistical analysis was performed using the freely downloadable software Open Meta-analyst (http://tuftscaes.org/open_meta/help/
3
BASIC DATA ON OUTCOME AFTER AORTIC VALVE REPLACEMENT
Table 1. Pooled Immediate Postoperative Outcome After Aortic Valve Replacement With or Without Coronary Artery Bypass Surgery as Reported in 176 Studies
Outcome Endpoint
Operative mortality Stroke Dialysis Pacemaker implantation Re-exploration for bleeding
Table 3. Pooled Immediate Postoperative Outcome After Isolated Aortic Valve Replacement as Reported in 103 Studies
Outcome Endpoint
Proportion
Number of
Number of
Studies
Patients
I2
(%)
95% CI
176
683,286
97%
4.3
4.0-4.6
84 42 30
234,463 35,605 13,435
91% 80% 98%
2.1 2.2 5.9
1.9-2.4 1.8-2.5 4.2-7.6
65
124,219
79%
4.3
4.0-4.6
Operative mortality Stroke Dialysis Pacemaker inplantation Re-exploration for bleeding
Proportion
Number of
Number of
Studies
Patients
I2
(%)
95% CI
103
158,749
87%
3.3
3.0-3.6
45 27 18
95,337 21,648 7,567
73% 70% 73%
1.7 1.6 3.3
1.5-1.9 1.2-1.9 2.5-4.0
38
61,486
73%
3.9
3.4-4.3
Abbreviation: CI, confidence interval.
Abbreviation: CI, confidence interval.
Outcome in the Overall Series openMA_help.html) and PASW statistical software (version 18, IBM SPSS Inc., Chicago, IL). Results were reported as proportions and 95% confidence intervals (95% CI). This was decided after the authors verified that arcsine of square root proportion provided similar pooled mortality proportions. Heterogeneity across studies was evaluated using the I2 test, assuming that o40% indicates a non-significant heterogeneity. Random-effects method was used in all analyses because this study included prospective and retrospective observational studies with different baseline and operative characteristics, and significant heterogeneity was expected. Sensitivity analysis and univariate metaregression were done to evaluate the impact of baseline covariates and treatment methods on the immediate outcome. Results were confirmed by excluding outliers. When indicated, the authors included no more than 3 covariates for multivariate meta-regression. Any p value less than 0.05 was considered statistically significant.
RESULTS
Outcome after Isolated Aortic Valve Replacement
Literature search provided 176 articles (reference list available from the authors), which where suitable for the present analysis. Literature search and inclusion/exclusion of retrieved studies are summarized in Figure 1.
Table 2. Impact of Baseline and Operative Characterictics on Immediate Postoperative Outcome After Aortic Valve Replacement With or Without Coronary Artery Bypass Surgery (Univariate Metaregression) Variable
Mid-date of study Geographic area Mean age Female gender Cerebrovascular disease Pulmonary disease Peripheral vascular disease Renal failure Prior cardiac surgery Urgent/emergency surgery Mean LVEF Concomitant CABG Mean cross-clamping time Mean CPB time
For this pooled analysis, 683,286 patients were available. Pooled proportion of in-hospital and/or 30-day mortality after AVR with or without concomitant CABG was 4.3% (95% CI 4.0-4.6, I2 97%; arcsine of square root proportion: 4.3%, 95% CI 4.0-4.6, I2 96%). Table 1 summarizes the pooled rates of main postoperative adverse events in these series. The risks of stroke and dialysis after AVR with or without concomitant CABG were exceedingly low (2.1 and 2.2% respectively). Univariate meta-regression showed that most clinically relevant risk factors were associated with increased risk of immediate postoperative mortality (Table 2). Mean patents’ age, prevalence of concomitant CABG, and prevalence of prior cardiac surgery were considered in the meta-regression model; all these risk factors were independent predictors of immediate postoperative mortality (p o 0.001).
One hundred three studies reporting on 158,749 patients were available for analysis of the immediate outcome after isolated AVR. Table 3 summarizes the pooled rates of the main outcome endpoints. Despite the significant heterogeneity of the included studies, pooled mortality after isolated AVR was 3.3%. Interestingly, the rates of stroke and dialysis after surgery Table 4. Impact of Baseline and Operative Characterictics on Immediate Postoperative Outcome After Isolated Aortic Valve Replacement (Univariate Meta-regression)
Number of Patients
p Value
Variable
683,008 661,876 390,023 593,081 328,105 340,271 372,772 376,017 165,026 535,169 55,225 683,286 38,227 39,738
0.223 0.091 o0.001 0.015 0.010 0.013 o0.001 o0.001 o0.001 o0.001 0.016 o0.001 0.054 0.009
Mid-date of study Geographic area Mean age Female gender Cerebrovascular disease Pulmonary disease Peripheral vascular disease Renal failure Prior cardiac surgery Urgent/emergency surgery Mean LVEF Mini-AVR Mean cross-clamping time Mean CPB time
NOTE. Geographic area: Europe (3.9%) v North America (4.6%). Abbreviations: CABG, coronary artery bypass grafting; CPB, cardiopulmonary bypass; LVEF, left ventricular ejection fraction.
Number of Patients
p Value
158,657 149,922 126,882 136,503 65,794 74,043 66,601 72,508 57,690 92,614 42,161 16,393 14,469 15,335
0.809 0.001 o0.001 0.006 0.073 0.008 o0.001 0.001 o0.001 o0.001 0.125 0.039 0.294 0.371
NOTE. Geographic area: Europe (2.8%) v North America (3.9%). Abbreviations: AVR, aortic valve replacement; CPB, cardiopulmonary bypass; LVEF, left ventricular ejection fraction.
4
BIANCARI ET AL
Fig 2. Meta-regression plot showing the impact of mean patients’ age on the immediate postoperative mortality after isolated aortic valve replacement (p o 0.001).
were extremely low (1.7% and 1.6%, respectively). Importantly, isolated AVR was associated with a pooled rate of need of pacemaker implantation of 3.3%. A number of risk factors were associated with increased risk of immediate postoperative mortality (Table 4), most notably mean patients’ age (p o 0.001) (Fig 2), prevalence of prior cardiac surgery (p o 0.001), and urgent/emergency procedure (p o 0.001). Aortic cross-clamping time and cardiopulmonary bypass time were not associated with increased mortality risk. Multilevel analysis showed that mortality was significantly increased only among studies with patients’ mean age Z80 years (pooled rates: mean age o60 years: 3.3%, mean age 60 years: 2.7%; mean age 70 years: 3.8%, mean age Z80 years: 6.1%, p o 0.001). Impact of Minimally Invasive Surgery on Immediate Postoperative Outcome Forty-eight studies reported on details of surgical technique and, therefore, these were stratified according to the prevalence of minimally invasive approach (mini-thoracotomy or ministernotomy). The overall mortality in these series was similar to the overall series (3.2%, 95% CI 2.7-3.7%, I2 79%, 602 out of 16,393 patients). Univariate meta-regression showed that the prevalence of mini-AVR was associated with significantly lower operative mortality (p ¼ 0.039, Fig 3). However, when the technique was included as a dichotomous variable (43 studies), mini-AVR only tended to have a lower operative
mortality (p ¼ 0.088). Pooled mortality after mini-AVR was 2.3% (95% CI 1.8-2.9%, I2 32%, 125 out of 4,367 patients, 16 studies), while it was 3.5% (95% CI 28-4.1%, I2 80%, 402 out of 11,076 patients, 37 studies) after full sternotomy. A significant difference in operative mortality after mini-AVR was observed when adjusted for mean left ventricular ejection fraction (p ¼ 0.010), but not when adjusted for mean patients’ age (p ¼ 0.234) and prior cardiac surgery (p ¼ 0.530). Mini-AVR was associated with a lower rate of postoperative stroke (1.5%, 95% CI 1.0-1.9%, I2 1%, 48/2547 patients, 11 studies) compared with full sternotomy (2.2%, 95% CI 1.5-2.8%, I2 74%, 158/5,839 patients, 20 studies), but such a difference was not statistically significant (p ¼ 0.351). MiniAVR was not associated with a reduced risk of postoperative dialysis (p ¼ 0.921), pacemaker implantation (p ¼ 0.531) or reexploration for bleeding (p ¼ 0.590). Outcome after Aortic Valve Replacement with Concomitant Coronary Artery Bypass Surgery Forty-six studies reporting on 321,073 patients were available for analysis of the immediate outcome after AVR and concomitant CABG. Table 5 summarizes the main outcome endpoints of this study population. Pooled operative mortality was 5.5% (Table 5) and was significantly higher than after isolated AVR (p o 0.001). This finding was confirmed also when adjusted for mean patients’ age (p o 0.001; Fig. 4),
Fig 3. Meta-regression plot showing the impact of prevalence of minimally invasive approach on the immediate postoperative mortality after conventional, isolated aortic valve replacement (p ¼ 0.039).
BASIC DATA ON OUTCOME AFTER AORTIC VALVE REPLACEMENT
5
Table 5. Pooled Immediate Postoperative Outcome After Aortic Valve Replacement With Coronary Artery Bypass Surgery as Reported in 46 Studies
surgery and favour TAVI or the use of sutureless valve prostheses. It is worth noting that shorter cross-clamping time was not associated with better early survival in this pooled analysis. However, the limitations of aggregate data analysis and lack of studies with extremely short cross-clamping time (the one expected with the use of sutureless aortic valve prostheses) may account for such a finding. Subanalysis of patients who underwent minimally invasive isolated AVR, ie, through a mini-thoracotomy or mini-sternotomy approach, versus those operated on through a full sternotomy suggests that mini-AVR may have the potential to improve operative survival, but such evidence is still lacking. A previous doublearm meta-analysis suggested that no difference is expected between these 2 techniques.7 The present meta-regression, starting from a one-arm analysis, indicates that mini-AVR is not associated with lower mortality when adjusted for mean patients’ age or history of prior cardiac surgery. Herein, the authors did not perform a formal 2-arm analysis, because many studies did not include a control group. Despite this, the results of this study, which include a significant number of patients (4,367 patients in the mini-AVR group and 11,076 in the full sternotomy group), indicate that there is still a need for further, well-planned studies to definitely establish the benefits of miniAVR or, even more important, identify the subsets of patients who may benefit most from this technique. The most important finding of the present study was the evident poorer results of patients undergoing AVR with concomitant CABG compared with those undergoing isolated AVR. The poorer outcome of the former patients seems to be independent from mean patients’ age, history of prior cardiac surgery, and aortic cross-clamping time. Importantly, the operative mortality after AVR with concomitant CABG was rather high in all age strata. These and previous findings8 suggest that immediate postoperative mortality may approach 10% among octogenarians. Indeed, the authors were able to identify mean patients’ age as the only determinant of operative mortality. This significantly increased operative risk suggests that patients undergoing AVR with concomitant CABG are the subset that may most benefit from less-invasive procedures. A previous meta-analysis on intermediate survival after TAVI2 showed that centers performing TAVI combined with percutaneous coronary intervention had better survival at 2 years
Outcome Endpoint
Operative mortality Stroke Dialysis Pacemaker implantation Re-exploration for bleeding
Proportion
Number of
Number of
Studies
Patients
I2
(%)
95% CI
46
321,073
77%
5.5
5.2-5.9
18 10 4
116,849 1,935 1,181
70% 83% 0%
3.0 5.6 3.9
2.5-3.4 3.5-7.7 2.8-5.1
13
38,791
79%
5.2
4.2-6.2
Abbreviation: CI, confidence interval.
prevalence of prior cardiac surgery (p o 0.001), and mean aortic cross-clamping time (p o 0.001). A number of risk factors were associated with increased risk of immediate postoperative mortality (Table 6). Multilevel analysis showed that mortality was significantly increased only among studies with a mean patients’ age Z80 years (pooled rates: mean age o70 years: 4.8%, mean age 70 years: 4.7%; mean age Z80 years: 8.4%, p ¼ 0.002).
DISCUSSION
The present analysis indicated that isolated AVR can be performed with rather low rates of operative mortality and major morbidity. This strengthens the concept that AVR remains the gold standard in patients without coronary artery disease.4 Meta-regression confirmed to a larger extent the prognostic importance of the patient’s age, female gender, pulmonary disease, peripheral vascular disease, renal failure, prior cardiac surgery, and urgent/emergency surgery (Table 4). This means that a judicious scrutiny is needed before deciding alternative methods to AVR, which are more expensive and with unproven durability. Figure 2 suggests that advanced age may pose an indication for less-invasive procedures, particularly in those patients aged 480 years. Still, the operative mortality after isolated AVR in the very elderly can be less than 10%; therefore, other concomitant morbidities along with advanced age may formally contraindicate conventional
Fig 4. Meta-regression plot showing the impact of mean patients’ age on the immediate postoperative mortality after aortic valve replacement and coronary artery bypass grafting (p o 0.001).
6
BIANCARI ET AL
Table 6. Impact of Baseline and Operative Characterictics on Immediate Postoperative Outcome After Aortic Valve Replacement With Coronary Artery Bypass Surgery (Univariate Meta-regression) Variable
Mid-date of study Geographic area Mean age Female gender Cerebrovascular disease Pulmonary disease Peripheral vascular disease Renal failure Prior cardiac surgery Urgent/emergency surgery Mean LVEF Mean cross clamping time Mean CPB time
Number of Patients
p Value
148,072 145,178 200,792 233,408 210,459 182,024 213,666 212,204 36,679 212,526 2,314 4,576 5,100
0.781 0.013 0.005 0.096 0.873 0.625 0.038 0.270 0.790 0.068 0.874 0.344 0.283
NOTE. Geographic area: Europe (5.2%) v North America (6.2%). Abbreviations: CPB, cardiopulmonary bypass; LVEF, left ventricular ejection fraction.
(74%) compared with centers performing only isolated TAVI. This also was consistent with survival rates at 30 days and 1 year. A recent study by Pasic et al9 showed that, in a series of 46 patients with a mean age of 81 years, mean logistic EuroSCORE of 40% and STS score of 23%, TAVI combined with percutaneous coronary intervention for only the most relevant coronary lesions resulted in survival rates at 30 days, 1 year, and 2 years of 96%, 87%, and 70%, respectively. Taken together, the results of this meta-analysis and those of small series reporting on TAVI and concomitant percutaneous coronary intervention suggest that subsets of patients requiring combined treatment of aortic valve stenosis and coronary artery disease may be better treated with less-invasive methods. In this setting, very elderly patients (Fig 4) may benefit from TAVI and percutaneous coronary intervention, particularly in the early and intermediate postoperative period. The authors
may speculate that in these patients, the use of sutureless aortic valve prostheses may open further possibilities of significantly shorter periods of cardiopulmonary bypass when combined with off-pump coronary artery revascularization. Furthermore, a number of non-surgical interventions during the perioperative period recently have been identified to improve the outcome of surgical procedures and are expected to have a significant impact on the results of aortic valve surgery in high-risk patients.10 The authors acknowledge that a number of limitations may affect the validity of the present results. An effort has been made to exclude any duplicate data, but they cannot be sure that a number of patients have been included in different series. However, the authors verified the validity of their results by excluding outliers and by leave-one-out analysis. The results of meta-regression can be affected also by intrinsic limitation of evaluating aggregate data, but still the results of this regression analysis were clinically sound and confirmed the results of previous studies. The authors did not compare either the results of institutional versus multicenter studies or those of prospective versus retrospective studies since they expected significant heterogeneities in all these series and in pooled analyses. Importantly, studies reported variably on 30-day or inhospital mortality or combined 30-days/in-hospital mortality. The authors verified that in most studies reporting on both 30day and in-hospital mortalities, the differences were not significant, and, therefore, they considered any 30-day or inhospital mortality as operative mortality. In conclusion, isolated AVR currently is associated with low mortality and morbidity rates. In this setting, the benefit of minimally invasive technique remains unclear in this analysis. Coronary artery disease requiring concomitant CABG may significantly increase the operative mortality of AVR, which may approach 10% in the very elderly. These observations indicate that patients requiring AVR and CABG should be the target of future treatment strategies such as endovascular and/or hybrid procedures.
REFERENCES 1. Lindroos M, Kupari M, Heikkilä J, et al: Prevalence of aortic valve abnormalities in the elderly: An echocardiographic study of a random population sample. J Am Coll Cardiol 21:1220-1225, 1993 2. Messori A, Trippoli S, Biancari F: Early and intermediate survival after transcatheter aortic valve implantation: Systematic review and meta-analysis of 14 studies. BMJ Open 3:e001770, 2013. doi:10.1136/ bmjopen-2012-001770. 3. Santarpino G, Pfeiffer S, Schmidt J, et al: Sutureless aortic valve replacement: First-year single-center experience. Ann Thorac Surg 94: 504-508, 2012 4. Walther T, Blumenstein J, van Linden A, et al: Contemporary management of aortic stenosis: Surgical aortic valve replacement remains the gold standard. Heart 98Suppl 4:iv23-iv29, 2012 5. Thyregod HG, Søndergaard L, Ihlemann N, et al: The nordic aortic valve intervention (NOTION) trial comparing transcatheter versus surgical valve implantation: Study protocol for a randomised controlled trial. Trials 14:11, 2013
6. Stroup DF, Berlin JA, Morton SC, et al: Meta-analysis of observational studies in epidemiology: A proposal for reporting. JAMA 283:2008-2012, 2000 7. Brown ML, McKellar SH, Sundt TM, et al: Ministernotomy versus conventional sternotomy for aortic valve replacement: A systematic review and meta-analysis. J Thorac Cardiovasc Surg 137:670-679, 2009 8. Vasques F, Lucenteforte E, Paone R, et al: Outcome of patients aged Z80 years undergoing combined aortic valve replacement and coronary artery bypass grafting: A systematic review and meta-analysis of 40 studies. Am Heart J 164:410-418, 2012 9. Pasic M, Dreysse S, Unbehaun A, et al: Combined elective percutaneous coronary intervention and transapical transcatheter aortic valve implantation. Interact Cardiovasc Thorac Surg 14:463-468, 2012 10. Landoni G, Rodseth RN, Santini F, et al: Randomized evidence for reduction of perioperative mortality. J Cardiothorac Vasc Anesth 26:764-772, 2012
tended toward lower mortality when included in metaregression analysis as a dichotomous variable (mini-AVR 4,367 patients: 2.3%, 95% CI 1.8-2.9% v full sternotomy 11,076 patients: 3.5%, 95% CI 28-4.1%, p ¼ 0.088). Operative mortality after AVR plus CABG was 5.5% (versus isolated AVR: p o 0.001), stroke 3.0%, pacemaker implantation 3.9%, and dialysis 5.6%. Mortality was high in all age strata, particularly among very elderly (mean age o70 years: 4.8%, mean age 70-79 years: 4.7%; mean age Z80 years: 8.4%, p ¼ 0.002). Conclusions: Isolated AVR is associated with low mortality and morbidity. Coronary artery disease requiring concomitant CABG increases the operative mortality. Patients requiring AVR and CABG should be the main target of lessinvasive treatment strategies. & 2013 Elsevier Inc. All rights reserved.
A
surgery were used to broaden the search. The authors applied the guidelines from Stroup et al in Meta-analysis of Observational Studies in Epidemiology (MOOSE).6
ORTIC STENOSIS (AS) is one of the most frequent cardiac diseases among the elderly.1 The expected significant growth of the number of elderly during the next decades may therefore lead to an increase of the number of patients requiring treatment of AS. Operative treatment of AS is a rapidly evolving field as, currently, the exposure to cardiopulmonary bypass and its related adverse events can be avoided or decreased by the use of transcatheter aortic valve replacement (TAVI)2 and sutureless aortic valve prostheses.3 Even if conventional aortic valve replacement (AVR) is considered the gold standard,4 particularly because of its well-proven long-term durability, major concern exists regarding the perceived high risk of immediate postoperative morbidity and mortality. Therefore, less-invasive technology recently has been embraced with enthusiasm, also in low-risk patients,5 even if there is no evidence of its durability. The lack of data effectively summarizing the immediate outcome of patients undergoing conventional AVR may induce such a shift in treatment modalities. This pooled analysis was planned to summarize the results of recent studies on this topic in order to validate the early efficacy of AVR and to identify those patients who may benefit most from less-invasive treatment methods. Furthermore, the aim from less-invasive was to better define the prognostic impact of concomitant CABG and the potential benefits of the minimally invasive technique in patients undergoing AVR. MATERIAL AND METHODS
Study Selection An English language literature review was performed through PubMed up to July 2012 for any study evaluating the outcome after treatment of aortic valve diseases. The words employed in the search were “aortic valve replacement” and “aortic valve and prosthesis.” The reference list of obtained articles and of journals dealing with cardiac
KEY WORDS: cardiac surgery, aortic valve, coronary artery bypass, risk, outcome, AVR + CABF, Mini-AVR, TAVI, aortic stenosis, transcatheter aortic valve implantation
Data Extraction Data were retrieved only from the articles, and no attempt to get missing data from the authors was made. Each study was evaluated independently by 6 coauthors (F.B., F.V., M.M., G.B, E.V., and G.V.) for inclusion or exclusion from this analysis. Then, the following data were extracted: Authors, year of publication, study period and mid-date of study, study design, prevalence of female gender, cerebrovascular disease, pulmonary disease, peripheral vascular disease, renal failure, prior cardiac surgery, urgent/emergency surgery, minimally invasive surgery (either ministernotomy or minithoracotomy) as well as mean values of the patients’s age, left ventricular ejection fraction, logistic EuroSCORE, duration of aortic cross-clamping and cardiopulmonary bypass. The main treatment modality was defined, whenever data were available, according to the intention-to-treat principle.
Inclusion Criteria To enter this analysis, studies had to fulfil the following criteria: (1) prospective or retrospective observational studies including at least 100 patients undergoing AVR with or without CABG, (2) studies published in the English language, (3) studies published as a full article, (4) studies reporting on at least immediate postoperative mortality,
From the *Department of Surgery, Oulu University Hospital, Oulu, Finland; and †Heart Center, Turku University Hospital, Turku, Finland. Address reprint requests to Fausto Biancari, MD, PhD, Oulu University Hospital, Department of Surgery, Kajaanintie 50, PL 21, 90029 Oulu, Finland. E-mail: [email protected] © 2013 Elsevier Inc. All rights reserved. 1053-0770/2601-0001$36.00/0 http://dx.doi.org/10.1053/j.jvca.2013.07.020
Journal of Cardiothoracic and Vascular Anesthesia, Vol ], No ] (Month), 2013: pp ]]]–]]]
1
2
BIANCARI ET AL
Potentially relevant studies identified and screened n = 3909 Non-pertinent studies n = 2232 Studies with no extractable data n = 274 Specific data not available n = 156 Case reports/ technical reports n = 423 Comments, letters and editorials n = 39 Reviews or metaanalyses n = 105 Studies considering fewer 100 patients n = 449 Duplicate studies n = 55
Studies included in the present analysis n = 176
Fig 1.
Flowchart summarizing literature search and selection of studies.
(5) studies published after 1999, and (6) studies including adult patients. The language of the articles was defined as reported in PubMed. The authors did not include in the present analysis unpublished data in the present analysis or reported only in abstract.
Exclusion Criteria The following exclusion criteria were used in this study: (1) studies including only or mostly pediatric patients, (2) studies reporting on any major associated procedure other than CABG and whose specific data could not be retrieved from the article, (3) studies reporting unclear data (for example, data not matching each other between the text and tables), (4) studies without enough information on treatment modalities and immediate postoperative outcome, (5) studies reporting only on aortic
valve endocarditis, and (6) patients who underwent transcatheter aortic valve implantation or any sutureless aortic valve implantation.
Outcome of Interest and Definitions The main outcome measure of this study was immediate postoperative mortality as occurred during the in-hospital stay and/or 30day postoperative period. Secondary outcome endpoints were stroke, de novo dialysis, atrioventricular block requiring pacemaker implantation, and re-exploration for excessive bleeding.
Statistical Analysis Statistical analysis was performed using the freely downloadable software Open Meta-analyst (http://tuftscaes.org/open_meta/help/
3
BASIC DATA ON OUTCOME AFTER AORTIC VALVE REPLACEMENT
Table 1. Pooled Immediate Postoperative Outcome After Aortic Valve Replacement With or Without Coronary Artery Bypass Surgery as Reported in 176 Studies
Outcome Endpoint
Operative mortality Stroke Dialysis Pacemaker implantation Re-exploration for bleeding
Table 3. Pooled Immediate Postoperative Outcome After Isolated Aortic Valve Replacement as Reported in 103 Studies
Outcome Endpoint
Proportion
Number of
Number of
Studies
Patients
I2
(%)
95% CI
176
683,286
97%
4.3
4.0-4.6
84 42 30
234,463 35,605 13,435
91% 80% 98%
2.1 2.2 5.9
1.9-2.4 1.8-2.5 4.2-7.6
65
124,219
79%
4.3
4.0-4.6
Operative mortality Stroke Dialysis Pacemaker inplantation Re-exploration for bleeding
Proportion
Number of
Number of
Studies
Patients
I2
(%)
95% CI
103
158,749
87%
3.3
3.0-3.6
45 27 18
95,337 21,648 7,567
73% 70% 73%
1.7 1.6 3.3
1.5-1.9 1.2-1.9 2.5-4.0
38
61,486
73%
3.9
3.4-4.3
Abbreviation: CI, confidence interval.
Abbreviation: CI, confidence interval.
Outcome in the Overall Series openMA_help.html) and PASW statistical software (version 18, IBM SPSS Inc., Chicago, IL). Results were reported as proportions and 95% confidence intervals (95% CI). This was decided after the authors verified that arcsine of square root proportion provided similar pooled mortality proportions. Heterogeneity across studies was evaluated using the I2 test, assuming that o40% indicates a non-significant heterogeneity. Random-effects method was used in all analyses because this study included prospective and retrospective observational studies with different baseline and operative characteristics, and significant heterogeneity was expected. Sensitivity analysis and univariate metaregression were done to evaluate the impact of baseline covariates and treatment methods on the immediate outcome. Results were confirmed by excluding outliers. When indicated, the authors included no more than 3 covariates for multivariate meta-regression. Any p value less than 0.05 was considered statistically significant.
RESULTS
Outcome after Isolated Aortic Valve Replacement
Literature search provided 176 articles (reference list available from the authors), which where suitable for the present analysis. Literature search and inclusion/exclusion of retrieved studies are summarized in Figure 1.
Table 2. Impact of Baseline and Operative Characterictics on Immediate Postoperative Outcome After Aortic Valve Replacement With or Without Coronary Artery Bypass Surgery (Univariate Metaregression) Variable
Mid-date of study Geographic area Mean age Female gender Cerebrovascular disease Pulmonary disease Peripheral vascular disease Renal failure Prior cardiac surgery Urgent/emergency surgery Mean LVEF Concomitant CABG Mean cross-clamping time Mean CPB time
For this pooled analysis, 683,286 patients were available. Pooled proportion of in-hospital and/or 30-day mortality after AVR with or without concomitant CABG was 4.3% (95% CI 4.0-4.6, I2 97%; arcsine of square root proportion: 4.3%, 95% CI 4.0-4.6, I2 96%). Table 1 summarizes the pooled rates of main postoperative adverse events in these series. The risks of stroke and dialysis after AVR with or without concomitant CABG were exceedingly low (2.1 and 2.2% respectively). Univariate meta-regression showed that most clinically relevant risk factors were associated with increased risk of immediate postoperative mortality (Table 2). Mean patents’ age, prevalence of concomitant CABG, and prevalence of prior cardiac surgery were considered in the meta-regression model; all these risk factors were independent predictors of immediate postoperative mortality (p o 0.001).
One hundred three studies reporting on 158,749 patients were available for analysis of the immediate outcome after isolated AVR. Table 3 summarizes the pooled rates of the main outcome endpoints. Despite the significant heterogeneity of the included studies, pooled mortality after isolated AVR was 3.3%. Interestingly, the rates of stroke and dialysis after surgery Table 4. Impact of Baseline and Operative Characterictics on Immediate Postoperative Outcome After Isolated Aortic Valve Replacement (Univariate Meta-regression)
Number of Patients
p Value
Variable
683,008 661,876 390,023 593,081 328,105 340,271 372,772 376,017 165,026 535,169 55,225 683,286 38,227 39,738
0.223 0.091 o0.001 0.015 0.010 0.013 o0.001 o0.001 o0.001 o0.001 0.016 o0.001 0.054 0.009
Mid-date of study Geographic area Mean age Female gender Cerebrovascular disease Pulmonary disease Peripheral vascular disease Renal failure Prior cardiac surgery Urgent/emergency surgery Mean LVEF Mini-AVR Mean cross-clamping time Mean CPB time
NOTE. Geographic area: Europe (3.9%) v North America (4.6%). Abbreviations: CABG, coronary artery bypass grafting; CPB, cardiopulmonary bypass; LVEF, left ventricular ejection fraction.
Number of Patients
p Value
158,657 149,922 126,882 136,503 65,794 74,043 66,601 72,508 57,690 92,614 42,161 16,393 14,469 15,335
0.809 0.001 o0.001 0.006 0.073 0.008 o0.001 0.001 o0.001 o0.001 0.125 0.039 0.294 0.371
NOTE. Geographic area: Europe (2.8%) v North America (3.9%). Abbreviations: AVR, aortic valve replacement; CPB, cardiopulmonary bypass; LVEF, left ventricular ejection fraction.
4
BIANCARI ET AL
Fig 2. Meta-regression plot showing the impact of mean patients’ age on the immediate postoperative mortality after isolated aortic valve replacement (p o 0.001).
were extremely low (1.7% and 1.6%, respectively). Importantly, isolated AVR was associated with a pooled rate of need of pacemaker implantation of 3.3%. A number of risk factors were associated with increased risk of immediate postoperative mortality (Table 4), most notably mean patients’ age (p o 0.001) (Fig 2), prevalence of prior cardiac surgery (p o 0.001), and urgent/emergency procedure (p o 0.001). Aortic cross-clamping time and cardiopulmonary bypass time were not associated with increased mortality risk. Multilevel analysis showed that mortality was significantly increased only among studies with patients’ mean age Z80 years (pooled rates: mean age o60 years: 3.3%, mean age 60 years: 2.7%; mean age 70 years: 3.8%, mean age Z80 years: 6.1%, p o 0.001). Impact of Minimally Invasive Surgery on Immediate Postoperative Outcome Forty-eight studies reported on details of surgical technique and, therefore, these were stratified according to the prevalence of minimally invasive approach (mini-thoracotomy or ministernotomy). The overall mortality in these series was similar to the overall series (3.2%, 95% CI 2.7-3.7%, I2 79%, 602 out of 16,393 patients). Univariate meta-regression showed that the prevalence of mini-AVR was associated with significantly lower operative mortality (p ¼ 0.039, Fig 3). However, when the technique was included as a dichotomous variable (43 studies), mini-AVR only tended to have a lower operative
mortality (p ¼ 0.088). Pooled mortality after mini-AVR was 2.3% (95% CI 1.8-2.9%, I2 32%, 125 out of 4,367 patients, 16 studies), while it was 3.5% (95% CI 28-4.1%, I2 80%, 402 out of 11,076 patients, 37 studies) after full sternotomy. A significant difference in operative mortality after mini-AVR was observed when adjusted for mean left ventricular ejection fraction (p ¼ 0.010), but not when adjusted for mean patients’ age (p ¼ 0.234) and prior cardiac surgery (p ¼ 0.530). Mini-AVR was associated with a lower rate of postoperative stroke (1.5%, 95% CI 1.0-1.9%, I2 1%, 48/2547 patients, 11 studies) compared with full sternotomy (2.2%, 95% CI 1.5-2.8%, I2 74%, 158/5,839 patients, 20 studies), but such a difference was not statistically significant (p ¼ 0.351). MiniAVR was not associated with a reduced risk of postoperative dialysis (p ¼ 0.921), pacemaker implantation (p ¼ 0.531) or reexploration for bleeding (p ¼ 0.590). Outcome after Aortic Valve Replacement with Concomitant Coronary Artery Bypass Surgery Forty-six studies reporting on 321,073 patients were available for analysis of the immediate outcome after AVR and concomitant CABG. Table 5 summarizes the main outcome endpoints of this study population. Pooled operative mortality was 5.5% (Table 5) and was significantly higher than after isolated AVR (p o 0.001). This finding was confirmed also when adjusted for mean patients’ age (p o 0.001; Fig. 4),
Fig 3. Meta-regression plot showing the impact of prevalence of minimally invasive approach on the immediate postoperative mortality after conventional, isolated aortic valve replacement (p ¼ 0.039).
BASIC DATA ON OUTCOME AFTER AORTIC VALVE REPLACEMENT
5
Table 5. Pooled Immediate Postoperative Outcome After Aortic Valve Replacement With Coronary Artery Bypass Surgery as Reported in 46 Studies
surgery and favour TAVI or the use of sutureless valve prostheses. It is worth noting that shorter cross-clamping time was not associated with better early survival in this pooled analysis. However, the limitations of aggregate data analysis and lack of studies with extremely short cross-clamping time (the one expected with the use of sutureless aortic valve prostheses) may account for such a finding. Subanalysis of patients who underwent minimally invasive isolated AVR, ie, through a mini-thoracotomy or mini-sternotomy approach, versus those operated on through a full sternotomy suggests that mini-AVR may have the potential to improve operative survival, but such evidence is still lacking. A previous doublearm meta-analysis suggested that no difference is expected between these 2 techniques.7 The present meta-regression, starting from a one-arm analysis, indicates that mini-AVR is not associated with lower mortality when adjusted for mean patients’ age or history of prior cardiac surgery. Herein, the authors did not perform a formal 2-arm analysis, because many studies did not include a control group. Despite this, the results of this study, which include a significant number of patients (4,367 patients in the mini-AVR group and 11,076 in the full sternotomy group), indicate that there is still a need for further, well-planned studies to definitely establish the benefits of miniAVR or, even more important, identify the subsets of patients who may benefit most from this technique. The most important finding of the present study was the evident poorer results of patients undergoing AVR with concomitant CABG compared with those undergoing isolated AVR. The poorer outcome of the former patients seems to be independent from mean patients’ age, history of prior cardiac surgery, and aortic cross-clamping time. Importantly, the operative mortality after AVR with concomitant CABG was rather high in all age strata. These and previous findings8 suggest that immediate postoperative mortality may approach 10% among octogenarians. Indeed, the authors were able to identify mean patients’ age as the only determinant of operative mortality. This significantly increased operative risk suggests that patients undergoing AVR with concomitant CABG are the subset that may most benefit from less-invasive procedures. A previous meta-analysis on intermediate survival after TAVI2 showed that centers performing TAVI combined with percutaneous coronary intervention had better survival at 2 years
Outcome Endpoint
Operative mortality Stroke Dialysis Pacemaker implantation Re-exploration for bleeding
Proportion
Number of
Number of
Studies
Patients
I2
(%)
95% CI
46
321,073
77%
5.5
5.2-5.9
18 10 4
116,849 1,935 1,181
70% 83% 0%
3.0 5.6 3.9
2.5-3.4 3.5-7.7 2.8-5.1
13
38,791
79%
5.2
4.2-6.2
Abbreviation: CI, confidence interval.
prevalence of prior cardiac surgery (p o 0.001), and mean aortic cross-clamping time (p o 0.001). A number of risk factors were associated with increased risk of immediate postoperative mortality (Table 6). Multilevel analysis showed that mortality was significantly increased only among studies with a mean patients’ age Z80 years (pooled rates: mean age o70 years: 4.8%, mean age 70 years: 4.7%; mean age Z80 years: 8.4%, p ¼ 0.002).
DISCUSSION
The present analysis indicated that isolated AVR can be performed with rather low rates of operative mortality and major morbidity. This strengthens the concept that AVR remains the gold standard in patients without coronary artery disease.4 Meta-regression confirmed to a larger extent the prognostic importance of the patient’s age, female gender, pulmonary disease, peripheral vascular disease, renal failure, prior cardiac surgery, and urgent/emergency surgery (Table 4). This means that a judicious scrutiny is needed before deciding alternative methods to AVR, which are more expensive and with unproven durability. Figure 2 suggests that advanced age may pose an indication for less-invasive procedures, particularly in those patients aged 480 years. Still, the operative mortality after isolated AVR in the very elderly can be less than 10%; therefore, other concomitant morbidities along with advanced age may formally contraindicate conventional
Fig 4. Meta-regression plot showing the impact of mean patients’ age on the immediate postoperative mortality after aortic valve replacement and coronary artery bypass grafting (p o 0.001).
6
BIANCARI ET AL
Table 6. Impact of Baseline and Operative Characterictics on Immediate Postoperative Outcome After Aortic Valve Replacement With Coronary Artery Bypass Surgery (Univariate Meta-regression) Variable
Mid-date of study Geographic area Mean age Female gender Cerebrovascular disease Pulmonary disease Peripheral vascular disease Renal failure Prior cardiac surgery Urgent/emergency surgery Mean LVEF Mean cross clamping time Mean CPB time
Number of Patients
p Value
148,072 145,178 200,792 233,408 210,459 182,024 213,666 212,204 36,679 212,526 2,314 4,576 5,100
0.781 0.013 0.005 0.096 0.873 0.625 0.038 0.270 0.790 0.068 0.874 0.344 0.283
NOTE. Geographic area: Europe (5.2%) v North America (6.2%). Abbreviations: CPB, cardiopulmonary bypass; LVEF, left ventricular ejection fraction.
(74%) compared with centers performing only isolated TAVI. This also was consistent with survival rates at 30 days and 1 year. A recent study by Pasic et al9 showed that, in a series of 46 patients with a mean age of 81 years, mean logistic EuroSCORE of 40% and STS score of 23%, TAVI combined with percutaneous coronary intervention for only the most relevant coronary lesions resulted in survival rates at 30 days, 1 year, and 2 years of 96%, 87%, and 70%, respectively. Taken together, the results of this meta-analysis and those of small series reporting on TAVI and concomitant percutaneous coronary intervention suggest that subsets of patients requiring combined treatment of aortic valve stenosis and coronary artery disease may be better treated with less-invasive methods. In this setting, very elderly patients (Fig 4) may benefit from TAVI and percutaneous coronary intervention, particularly in the early and intermediate postoperative period. The authors
may speculate that in these patients, the use of sutureless aortic valve prostheses may open further possibilities of significantly shorter periods of cardiopulmonary bypass when combined with off-pump coronary artery revascularization. Furthermore, a number of non-surgical interventions during the perioperative period recently have been identified to improve the outcome of surgical procedures and are expected to have a significant impact on the results of aortic valve surgery in high-risk patients.10 The authors acknowledge that a number of limitations may affect the validity of the present results. An effort has been made to exclude any duplicate data, but they cannot be sure that a number of patients have been included in different series. However, the authors verified the validity of their results by excluding outliers and by leave-one-out analysis. The results of meta-regression can be affected also by intrinsic limitation of evaluating aggregate data, but still the results of this regression analysis were clinically sound and confirmed the results of previous studies. The authors did not compare either the results of institutional versus multicenter studies or those of prospective versus retrospective studies since they expected significant heterogeneities in all these series and in pooled analyses. Importantly, studies reported variably on 30-day or inhospital mortality or combined 30-days/in-hospital mortality. The authors verified that in most studies reporting on both 30day and in-hospital mortalities, the differences were not significant, and, therefore, they considered any 30-day or inhospital mortality as operative mortality. In conclusion, isolated AVR currently is associated with low mortality and morbidity rates. In this setting, the benefit of minimally invasive technique remains unclear in this analysis. Coronary artery disease requiring concomitant CABG may significantly increase the operative mortality of AVR, which may approach 10% in the very elderly. These observations indicate that patients requiring AVR and CABG should be the target of future treatment strategies such as endovascular and/or hybrid procedures.
REFERENCES 1. Lindroos M, Kupari M, Heikkilä J, et al: Prevalence of aortic valve abnormalities in the elderly: An echocardiographic study of a random population sample. J Am Coll Cardiol 21:1220-1225, 1993 2. Messori A, Trippoli S, Biancari F: Early and intermediate survival after transcatheter aortic valve implantation: Systematic review and meta-analysis of 14 studies. BMJ Open 3:e001770, 2013. doi:10.1136/ bmjopen-2012-001770. 3. Santarpino G, Pfeiffer S, Schmidt J, et al: Sutureless aortic valve replacement: First-year single-center experience. Ann Thorac Surg 94: 504-508, 2012 4. Walther T, Blumenstein J, van Linden A, et al: Contemporary management of aortic stenosis: Surgical aortic valve replacement remains the gold standard. Heart 98Suppl 4:iv23-iv29, 2012 5. Thyregod HG, Søndergaard L, Ihlemann N, et al: The nordic aortic valve intervention (NOTION) trial comparing transcatheter versus surgical valve implantation: Study protocol for a randomised controlled trial. Trials 14:11, 2013
6. Stroup DF, Berlin JA, Morton SC, et al: Meta-analysis of observational studies in epidemiology: A proposal for reporting. JAMA 283:2008-2012, 2000 7. Brown ML, McKellar SH, Sundt TM, et al: Ministernotomy versus conventional sternotomy for aortic valve replacement: A systematic review and meta-analysis. J Thorac Cardiovasc Surg 137:670-679, 2009 8. Vasques F, Lucenteforte E, Paone R, et al: Outcome of patients aged Z80 years undergoing combined aortic valve replacement and coronary artery bypass grafting: A systematic review and meta-analysis of 40 studies. Am Heart J 164:410-418, 2012 9. Pasic M, Dreysse S, Unbehaun A, et al: Combined elective percutaneous coronary intervention and transapical transcatheter aortic valve implantation. Interact Cardiovasc Thorac Surg 14:463-468, 2012 10. Landoni G, Rodseth RN, Santini F, et al: Randomized evidence for reduction of perioperative mortality. J Cardiothorac Vasc Anesth 26:764-772, 2012