| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Cardiothoracic Surgery, St. Mary’s Hospital, Faculty of Medicine, Imperial College, London, England
b Department of Surgical Oncology and Technology, St. Mary’s Hospital, Faculty of Medicine, Imperial College, London, England
c Department of Cardiothoracic Surgery, Royal Brompton Hospital, Faculty of Medicine, Imperial College, London, England
* Address correspondence to Dr Murtuza, Department of Cardiothoracic Surgery, St. Mary’s Hospital, Faculty of Medicine, Imperial College, London, W2 1NY, England (Email: b.murtuza{at}imperial.ac.uk).
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Only four small randomized studies in the literature to date, each with only 20 to 60 patients in each group, compare mAVR with cAVR [3–6]. These studies report disparate findings. The purpose of the present study was to review all randomized and nonrandomized comparative studies published in the literature that describe mAVR versus cAVR and to integrate the data using meta-analysis techniques to draw more useful conclusions about primary outcomes of mortality, cerebrovascular accident (CVA), renal failure, and respiratory failure. In addition, surrogate outcomes of ventilation time, intensive care unit (ICU) stay, and total LOS, as well as a number of other secondary events, were evaluated. We further studied intraoperative variables and examined the heterogeneity between studies.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Data Extraction
All data extraction from the 26 included studies was performed independently by two of us (Bari Murtuza and Thanos Athanasiou). The following study characteristics were extracted: author, study design, geographic location, period of study, year of publication, exclusion and matching criteria, number of patients in each group, mean patient age, and percentage of male patients. Extracted intraoperative variables included the following: type of incision; CCT, CPB, and total operation times; techniques used for cannulation and myocardial protection; venting and strategy for removal of air; minimal access group (MAG) conversions; and type of prosthesis used. Primary outcome data included mortality, CVA, renal failure, and respiratory failure. Secondary outcomes included surrogates of ventilation time, ICU stay, total LOS, and data related to arrthymias, sternal wound complications, minor neurologic complications, pericardial and pleural effusions, pneumothorax, and chest infection.
The study was performed according to guidelines of the Meta-Analysis of Observational Studies in Epidemiology group [29], and the quality of nonrandomized studies was assessed using modified criteria of the Newcastle-Ottawa Scale [30]. High-quality studies were defined as randomized controlled trials (RCTs) and those matched for five comparability criteria or more, including age, sex; preoperative left ventricular ejection fraction (LVEF), native valve diseasse, previous myocardial infarction, chronic obstructive pulmonary disease, neurologic status, peripheral vascular disease, diabetes mellitus, renal impairment, congestive heart failure (CHF), body mass index/body surface area or weight, percentage of nonelective procedures, and Parsonnet score.
Inclusion and Exclusion Criteria for Meta-Analysis
All RCTs and nonrandomized comparative studies were included in the analysis. Studies included at least one of the outcomes of interest and comprised a previously unreported patient group. We included reports with data for both aortic and mitral valve replacement when aortic valve replacement was reported separately and in a comparative manner. All noncomparative studies were excluded, as well as those in which outcomes of interest were not documented. Publications with mixed outcome data combining results for both aortic and mitral valve replacement were excluded. When a value of zero was reported for a particular outcome of interest for both techniques, this outcome measure was excluded from the meta-analysis, as were outcome data when the mean and standard deviation (SD) were not stated or calculable.
Statistical Analysis
Meta-analysis was performed using the odds ratio (OR) or weighted mean difference (WMD) as the summary statistic for binary or continuous variables, respectively. The analysis was performed according to recommendations of the Cochrane Collaboration and the Quality of Reporting of Meta-analyses guidelines [31]. An OR or WMD less than 1 favors mAVR. Random effects models were used because this model assumes variation between studies and is preferred for surgical data because selection criteria and risk profiles for patients differ among centers. P < 0.05 was considered statistically significant.
A sensitivity analysis for quantitative assessment of heterogeneity was performed. For this, two subgroups were considered and the data reanalyzed: studies with sample sizes of at least 70 patients in each group; RCTs, and studies matched for five criteria or more. All data were analyzed using commercially available software (SPSS version 12.0 for Windows; SPSS Inc, Chicago, IL, and Review Manager version 4.2; The Cochrane Collaboration, Oxford, England).
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
|
With respect to differences in clinical characteristics between included studies, four of five studies reporting operative priority [7, 15, 26, 28] included 78.6% to 94% elective procedures in MAG and 72% to 92% elective procedures in CG. The fifth report, by Sharony and colleagues [16], included a far greater percentage of nonelective procedures: 55.8% in MAG and 57.5% in CG. The percentage of mechanical valves implanted was 23.3% to 94.0% in MAG and 19.4% to 97% in CG; this percentage was not stated in 9 studies [3, 6, 16, 18, 20, 22, 23, 25, 28]. Five studies included a not insignificant proportion of patients with a history of CHF (New York Heart Association functional class III or IV), or low ejection fraction. [15, 16, 25–27].
Meta-Analysis of Primary and Secondary Outcome Events
All meta-analytical data were calculated using random-effects models. Of the major outcome events, we found a small apparent difference favoring MAG for percentage perioperative mortality (Fig 1), although this finding was not robust in subsequent sensitivity analysis taking into account study quality (Fig 2). There were no significant differences between groups for CVA, renal failure, or respiratory failure (Table 2). Twenty studies were included with a total of 4,667 patients for perioperative mortality (OR, 0.72; 95% confidence interval (CI), 0.51-1.00; p = 0.05, and 
2 = 9.29; p = 0.97). The corresponding incidence of mortality across studies was 3.0% and 4.6% for MAG and CG, respectively. Smaller numbers of patients (<2000) were included for ICU stay, total LOS, and ventilation time. Calculated nonweighted means between studies for MAG versus CG were as follows: ICU stay, 1.8 versus 2.4 days (WMD = -0.43; p = 0.02; 95% CI, -0.78 to -0.08; and 2 = 104.02; p < 0.00001); total LOS, 8.8 versus 10.2 days (WMD = -1.23; 95% CI, -1.91 to -0.54; p = 0.0004, and 2 = 89.1; p < 0.00001); ventilation time, 9.4 versus 12.5 hours (WMD = -2.86; 95% CI, -4.04 to -1.68; p < 0.00001, and 2 = 228.18; p < 0.00001).
|
|
|
2 = 380.43; p < 0.00001); CPB, 102.3 versus 90.2 minutes (WMD = 11.22; 95% CI, 6.87-15.57; p < 0.00001, and 2 = 306.11; p < 0.00001); total operating time, 209.4 versus 191.8 minutes (WMD = 15.17; 95% CI, 5.52-24.81; p = 0.002, and 2 = 106.15; p < 0.0001). We found, however, a significantly fewer number of patients requiring transfusion in MAG (46.6% incidence across studies versus 63.5% for CG). This analysis included 1390 patients in 9 studies, with OR = 0.55; 95% CI, 0.42-0.73; p < 0.0001, and 2 = 8.75; p = 0.36. Analysis of all other secondary outcome measures did not reveal any significant differences between groups.
Sensitivity Analysis
When considering high-quality studies defined as RCTs or those matched for five criteria or more, as stated, the data for perioperative mortality was no longer found to be significant. In contrast, data for ICU stay, total LOS, and ventilation times all showed persistent significant differences favoring MAG, with less heterogeneity between studies included in each subgroup analysis, as follows: ICU stay, 1.6 days for MAG versus 2.0 for CG (WMD = -0.39; 95% CI, -0.67 to -0.11; p = 0.007, and 2 = 8.81; p = 0.07); total LOS, 7.9 days for MAG versus 8.6 days for CG (WMD = -0.67; 95% CI, -1.08 to -26; p = 0.001, and 2 = 6.25; p = 0.18); and ventilation time, 10.6 hours for MAG versus 11.3 hours for CG (WMD = -1.02; 95% CI, -1.66 to -0.38; p = 0.002, and 2 = 9.61; p = 0.05). The only other persistent differences were those for CPB and total operating time, which were still less for CG when only higher quality studies were considered: CPB, 104.4 minutes for MAG versus 94.0 minutes for CG (WMD = 9.04; 95% CI, 0.66-17.42; p = 0.03, and 2 = 121.84; p < 0.00001), and total operating time, 211.7 minutes for MAG versus 180.7 minutes for CG (WMD = 29.09; 95% CI, 8.42-49.76; p = 0.006, and 2 = 13.15; p = 0.001).
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
In evaluating minimal access valve surgery, one may consider the quality of valve surgery, complications of CPB, and inadequate myocardial protection, as well as incision-related morbidity [43]. The quality of mAVR may, in turn, be considered in terms of perioperative mortality, long-term survival, conversion rates, and PVL rates. The best available evidence for mAVR comes from the four published RCTs, which do not show any significant differences in perioperative mortality [3–6]. In the other four high-quality studies we identified as being matched for at least five independent predictors of mortality, findings were similar [7, 15–17]. The largest nonrandomized study, by Mihaljevic and colleagues [25], included more than 500 patients in each group and reached similar conclusions. Most of the other nonrandomized studies, however, suggested lower mortality for MAG. This accounts for an apparent mortality benefit for MAG in our overall meta-analysis, although with an absence of significant effect in the sensitivity analysis ( Fig 2 ).
Six studies reported midterm follow-up data [8, 14–16, 21, 25]. Mihaljevic and colleagues [25] reported an actuarial survival rate of 98%, 94%, and 82% for MAG at 1, 3, and 5 years, respectively, compared with 94%, 90%, and 86% for CG, and these data are comparable with those of Sharony and colleagues, [16] who reported previous myocardial infarction, CHF, and urgent or emergent surgery as risk factors for midterm mortality. Common causes for reoperation during follow-up were endocarditis and constrictive pericarditis [8]. Detter and colleagues [21] reported freedom from reoperation of 100% for CG versus 98.5% for MAG, with freedom from endocarditis of 100% for mAVR versus 96.9% for cAVR. These results seem to suggest that mAVR can be performed with good midterm results comparable to those with cAVR, although the quality of published studies and degree of matching within these studies introduce sources of publication bias that is therefore also inherent in the present type of analysis.
Conversion to full sternotomy during mAVR is a potentially important cause of morbidity and mortality. Fifteen studies reported conversion rates with a low overall incidence [3–9, 15, 17, 18, 20, 21, 23, 26, 27]. While definite conclusions cannot be drawn about the effects of these conversions, a recent report reviewed a large number of cases of mAVR with a particular focus on conversions and found an overall incidence of 2.6% and 4.0% for upper and lower hemisternotomy [41]. The most common reasons for conversion in this series were bleeding and ventricular dysfunction (upper incision) or poor exposure (lower incision).
Although sufficient data were not available for meta-analysis, PVL rates were given in five reports [15, 21, 24, 27, 28]. Of these, two studies reported no early PVL in either group [15, 27] and the other three reported respective early PVL rates of 1.4%, 0%, and 21.4% for MAG and 0%, 2.0%, and 0% for CG [21, 24, 28]. Christiansen and colleagues [15] reported minor PVL at 1-year follow-up in 18.2% of patients in MAG and 13.0% in CG. These data imply that PVL rates may be higher for MAG, although it should be noted that rates of PVL detected depend on the intensity of follow-up, and more rigorous studies will be needed to detect and quantify PVL rates.
Controversy exists in the literature concerning the definition of minimally invasive valve surgery. We consider "minimal access" the preferred term [8, 9, 22] because CPB is still required in all mAVR operations. Some centers use peripheral cannulation techniques in association with described minimally invasive approaches, and these may be considered more invasive because of morbidity related to cannulation through the groin [44]. Further, the effects of femoral versus aortic cannulation on renal, splanchnic, and cerebral perfusion remain unclear. Some have expressed concern that myocardial protection may be compromised as a result of inability for adequate topical cooling of the heart during mAVR, although the three studies reporting data on postoperative LVEF do not suggest this [5, 11, 19].
In considering the surgical incision for mAVR, anticipated benefits of limited access (blood loss, pain, effects on lung function, rehabilitation, and LOS-related costs) must be balanced with the need for adequate exposure, ease of operation, and ease of conversion. It is important to emphasize that a minimal access incision does not equate to minimal invasiveness because some of these surgical approaches result in internal thoracic artery division, costochondral division or resection, opening of pleural cavities, and rib or sternal spreading. We did not find any significant differences between groups in the number of patients who received blood transfusions, the 24-hour blood loss, or the rate of repeat exploratory surgery because of bleeding. Blood loss, however, is multifactorial and related not only to incision but also to CPB and operating time, as well as preoperative and postoperative anticoagulation and antiplatelet drug regimens.
One might expect less postoperative pain with mAVR than with cAVR. Among studies included here, nine commented on postoperative pain scores or analgesia requirements [3–6, 10, 15, 17, 20, 22] and four reported differences favoring MAG [3, 10, 20, 22]. Pain is an important factor influencing postoperative lung function, although CPB, age, history of chronic obstructive pulmonary disease and smoking, ventilation time, and adequacy and duration of mediastinal and pleural drainage are also contributing factors [3, 4, 18, 22]. We did not find any significant difference in incidence of respiratory failure or other respiratory tract complications. Four studies reported small benefits in early postoperative lung function for MAG [3, 4, 18, 22]. By one month after surgery, however, lung function seems to be similar for both MAG and CG [4, 22], suggesting that early effects of less pain and better mobilization become less important later on. It is difficult to interpret the results for postoperative lung function, however, because there were important differences between studies in surgical approach, duration of CPB, pain scores, age, and blood loss.
In terms of patient satisfaction and QOL, only the study by Detter and colleagues [21] formally reported results of QOL as assessed by the 36-item Short-Form Health Survey questionnaire, with no significant difference between groups. Better reporting of pain, QOL, and patient satisfaction using standardized scores such as the Short-Form Health Survey in future randomized studies will help draw more useful conclusions about these data. Overall cost-effectiveness of mAVR versus cAVR was difficult to evaluate because only two studies [17, 27] reported cost data, although neither provided a detailed cost breakdown or analysis. Mihaljevic and colleagues [25] reported a significantly higher percentage of patients discharged to home as opposed to nursing care facilities after mAVR compared with cAVR. Further carefully designed studies that include the rehabilitation period after discharge from the primary hospital will be required to address this.
Considering all of our data, we could not demonstrate any clear advantage or disadvantage for mAVR compared with cAVR in terms of clinical outcomes. The question then arises as to which group of patients should be offered mAVR. Certainly mAVR could be offered on the basis of patient choice. It is important, however, to ask whether mAVR offers benefits in selected groups of patients at high risk such as those with poor left ventricular (LV) function, the elderly, or those with substantially impaired lung function. In particular, any such benefits in patients at high risk could be important for comparison with outcomes using newer percutaneous AVR approaches [45,46]. DeSmet and colleagues [23] examined outcomes in risk-stratified patient groups and found a lower incidence of atrial fibrillation in groups at medium and high risk undergoing mAVR compared with cAVR, with no differences in perioperative mortality, ICU stay, or total LOS. As to poor LV function, we identified five studies that included a substantial percentage of patients with a history of CHF (New York Heart Association class III or IV) or low LVEF, although the definition of low LVEF varied among studies [15, 16, 25–27]. Perioperative mortality was comparable to that in the other studies included. Some centers considered poor LV function a contraindication to mAVR [4, 18, 23]. These differences in exclusion criteria, definitions, and percentages of patients with poor LV consitute an important source of heterogeneity. Tabata and colleagues [40] recently compared mAVR and cAVR in patients with substantial LV dysfunction and found no evidence for a significant difference in operative mortality, CVA, renal failure, ventilation time, or total LOS in these patients.
Although not included in this analysis (a larger, more recent study [16] was included), Sharony and colleagues [47] reported a retrospective comparison of two matched cohorts of elderly patients undergoing mAVR and cAVR. Comparing 189 patients in each group, with a mean age of 75.3 ± 6.4 years for MAG and 75.3 ± 6.7 years for CG, they found no differences in any of the primary outcome events examined in the present work. They did, however, find significant differences in terms of total LOS and percentage of patients discharged to home favoring MAG, and this could have important resource utilization implications if this finding were confirmed in an adequately powered randomized study. No studies have examined the role of mAVR specifically in patients with poor preoperative lung function, and it will be important for future work to evaluate this.
Study Limitations
In considering ICU stay, total LOS, ventilation time, CCT, CPB, and total operating times, we found significant heterogeneity between studies in the overall meta-analysis (Table 2). This was markedly less when considering only RCTs and high-quality reports (Table 3). The small number of published RCTs included, each with relatively few patients, also limits somewhat the robustness of the overall findings of our study. Discrepancies in the clinical characteristics of patients among studies and selection bias owing to differences in operative strategy, number of centers and surgeons, and learning curve effects mean that we cannot completely eliminate the effects of confounding factors because the MAGs and CGs were not comparable for all of the factors that could potentially alter the outcomes of interest. Further, most studies included in the meta-analysis did not have random allocation of patients to the two groups. Publication bias is inherent in the present type of study because data showing positive findings tend to favor publication. In considering LOS, there may also be a type of bias introduced as a result of perhaps more aggressive management of patients with the goal of early discharge in those in MAGs. Other sources of heterogeneity and limitations of this analysis include differences in operative strategies, in particular for myocardial protection, and percentages of patients with poor LV function.
|
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Glauber, A. Miceli, S. Bevilacqua, and P. A. Farneti Minimally invasive aortic valve replacement via right anterior minithoracotomy: Early outcomes and midterm follow-up J. Thorac. Cardiovasc. Surg., December 1, 2011; 142(6): 1577 - 1579. [Full Text] [PDF]
|
|
|
|
 |
|
 E. Khoshbin, S. Prayaga, J. Kinsella, and F. W. H. Sutherland Mini-sternotomy for aortic valve replacement reduces the length of stay in the cardiac intensive care unit: meta-analysis of randomised controlled trials BMJ Open, November 24, 2011; 1(2): e000266 - e000266. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. M. Pineda, O. Santana, C. Zamora, A. M. Benjo, G. A. Lamas, and J. Lamelas Outcomes of a Minimally Invasive Approach Compared With Median Sternotomy for the Excision of Benign Cardiac Masses Ann. Thorac. Surg., May 1, 2011; 91(5): 1440 - 1444. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
O. Santana, J. Reyna, R. Grana, M. Buendia, G. A. Lamas, and J. Lamelas Outcomes of Minimally Invasive Valve Surgery Versus Standard Sternotomy in Obese Patients Undergoing Isolated Valve Surgery Ann. Thorac. Surg., February 1, 2011; 91(2): 406 - 410. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Lamelas, A. Sarria, O. Santana, A. M. Pineda, and G. A. Lamas Outcomes of Minimally Invasive Valve Surgery Versus Median Sternotomy in Patients Age 75 Years or Greater Ann. Thorac. Surg., January 1, 2011; 91(1): 79 - 84. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Vaughan, N. Fenwick, and P. Kumar Assisted venous drainage on cardiopulmonary bypass for minimally invasive aortic valve replacement: is it necessary, useful or desirable? Interact CardioVasc Thorac Surg, June 1, 2010; 10(6): 868 - 871. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Scarci, C. Young, and H. Fallouh Is ministernotomy superior to conventional approach for aortic valve replacement? Interact CardioVasc Thorac Surg, August 1, 2009; 9(2): 314 - 317. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Modi, A. Hassan, and W. R. Chitwood Jr. Minimally invasive mitral valve surgery: a systematic review and meta-analysis Eur J Cardiothorac Surg, November 1, 2008; 34(5): 943 - 952. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
0.05) if it does not touch the central vertical line (ie, outside the 95% CI). Squares for each included study show point estimates of treatment effect (OR), with the size of the square representing the weight attributed to each study; horizontal bars show 95% CIs for these studies. These results suggest that minimal access aortic valve replacement (mAVR) may have small beneficial effects in terms of mortality. (cAVR = conventional aortic valve replacement; CG = conventional group; df = degrees of freedom; MAG = minimal access group.)