Cheng, Davy C. H. MD*; Martin, Janet PharmD, MSc (HTA&M)*†; Lal, Avtar MD, PhD*; Diegeler, Anno MD, PhD‡; Folliguet, Thierry A. MD§; Nifong, L. Wiley MD∥; Perier, Patrick MD‡; Raanani, Ehud MD¶; Smith, J. Michael MD#; Seeburger, Joerg MD**; Falk, Volkmar MD††
Interest in minimally invasive cardiac surgery continues to grow rapidly. Although conventionally mitral valve surgery has been performed via a full incision through the sternum, a variety of technologies and techniques have enabled minimally invasive mitral valve surgery (mini-MVS) to be performed through one or more small incisions in the thorax with assisted vision using cameras. In some cases, robotic surgery may be used for mini-MVS. The goal of mini-MVS is to reduce the surgical trauma to the patient (presumably to reduce pain, scarring, and inflammatory response) while maintaining the proven surgical efficacy of the conventional open approach.
Some, but not all, studies suggest that favorable outcomes are achieved using the minimally invasive approach to mitral valve surgery. A comprehensive meta-analysis of available comparative studies (randomized and nonrandomized) is needed to better assess the risks versus benefits of mini-MVS versus conventional open mitral valve surgery (conv-MVS) through median sternotomy. One previous meta-analysis of mini-MVS by Modi et al1 has been published. However, a number of relevant studies were not identified, and only some outcomes of interest were analyzed in the article. Therefore, we conducted a systematic review with meta-analysis to address the available evidence to date which compares mini-MVS via thoracotomy or parasternal approach with conv-MVS through median sternotomy.
The objective of this systematic review with meta-analysis was to determine whether, in patients undergoing mitral valve surgery, mini-MVS improves postoperative complications and provides comparable or improved long-term clinical outcomes, quality of life, and cost-effectiveness compared with conv-MVS.
To answer this primary objective, the following subquestions were prespecified to guide the systematic review:
1. Does mini-MVS reduce risk of death compared with conv-MVS?
2. Does mini-MVS reduce perioperative complications compared with conv-MVS?
3. Does mini-MVS improve quality of life, functionality, or other patient-reported outcomes?
4. Does mini-MVS reduce total costs, intensive care unit (ICU) and hospital length of stay, need for repeat cardiac surgery, readmissions, and is it cost-effective?
This systematic review with meta-analysis of comparative studies was performed in accordance with state-of-the-art methodological recommendations for randomized and observational studies (ie, as per the “Quality of Reports of Meta-Analyses” and “Meta-Analysis of Observational Studies of Epidemiology,” QUOROM and MOOSE guidelines, respectively)2,3 and according to a protocol that prespecified outcomes, search strategies, inclusion criteria, and statistical analyses.
Clinical endpoints of interest included postoperative all-cause mortality, stroke, myocardial infarction, atrial fibrillation, low cardiac output syndrome, need for antiarrhythmic drugs, need for pacemaker, reexploration for bleeding, major bleeding, transfusions, renal failure, heart failure, New York Heart Association (NYHA) class, reintervention for bleeding, reintervention for valvular repair or replacement, patient satisfaction, functionality, quality of life, cross-clamp time, duration of surgery, ICU length of stay, total hospital length of stay, costs, and cost-effectiveness.
A comprehensive literature search of MEDLINE, EMBASE, Cochrane CENTRAL, CTSnet, and databases of conference abstracts using keywords and variants was performed from the earliest available date to March 2010. Search terms included variants of “surgical procedures,” “minimally invasive,” “mitral valve,” “thoracoscopic,” and “robotic.” No limits were placed on date, study design, or language. This was supplemented with hand search from the selected articles and review articles. Experts were contacted to solicit additional reports of published or unpublished clinical studies of mini-MVS.
To be eligible for inclusion in the systematic review and meta-analysis, studies had to be randomized or nonrandomized comparative studies comparing mini-MVS (performed via thoracotomy through port-access or keyhole, with either direct visualization or with camera or robotic assistance through lateral, parasternal, or xiphoid approaches) versus conv-MVS (performed via median sternotomy or parasternal approach). Published and unpublished studies were eligible in any language.
Two authors extracted and verified by consensus the following data points: baseline demographics, surgical procedure details, and all clinically relevant or resource-related outcomes.
Risk ratios (RRs) and their 95% confidence intervals (CIs) were calculated for discrete data. Weighted mean differences (WMD; 95% CI) were calculated for continuous data. Heterogeneity was explored through the Q-statistic and by calculating the heterogeneity (I2). RR and WMD were calculated using the random effects model to provide a conservative analysis in the case of heterogeneity. When statistical heterogeneity was high (I2 > 75%), potential reasons for heterogeneity were explored. Because data from randomized trials were of greatest interest, subanalysis was planned for randomized controlled trials (RCTs) versus observational studies.
When possible, intention-to-treat was data analyzed. Statistical significance for overall effect was defined as P < 0.05 or a confidence that excluded the value 1.00 for RR and 0 for WMD. Data analysis was performed using Review Manager (REVMAN) 5 COCHRANE Collaboration software.
Of over 1159 citations identified, 208 were identified as potentially relevant and were retrieved for full review. An additional two studies were identified by experts as potentially eligible studies. Of these, 35 studies met the inclusion criteria for the primary analysis of mini-MVS versus conv-MVS.4–38 Two of the included studies were RCTs12,13 and the remainder were nonrandomized studies (Fig. 1). Most of the non-RCTs were retrospective studies. Thirty-four of the studies were published between 1997 and 2010. One37 of the two unpublished studies37,38 suggested by the experts was subsequently published before this manuscript was sent to press. Most of the studies were from the United States (n = 17), followed by Germany (n = 6), France (n = 2), Israel (n = 2), and one each from Austria, Yugoslavia, Slovenia, Japan, China, India, Egypt, and Turkey.
Eighteen studies used primarily endoaortic balloon clamping,5,7,12,17–20,22,23,25,27,29,30,32,33,35,36,38 whereas 13 studies used direct/transthoracic aortic clamping,8–11,13–15,21,24,28,31,34 and four studies used both endoaortic and transthoracic clamping.6,16,26,37 Three studies reported that posterior mitral valve repair was performed: 77% versus 73%,16 75% versus 80%,26 and 81% versus 80%19 for mini-MVS versus conv-MVS, respectively. No study provided separate data for anterior mitral valve repair versus posterior mitral valve repair. Three studies included only redo mini-MVS versus conv-MVS,5,6,25 whereas the remainder were of first-time, mixed redo or first-time, or did not report this information.
For studies that reported more than once on an identical or overlapping population, attempts were made to include only the most recent or most complete results. In many cases, it was difficult to determine whether series from the same centers (or involving overlapping authors, but from differing centers) provided distinct data or whether there was overlap between studies. In many nonrandomized studies, patient attrition was inadequately described, and crossover from intended mini-MVS to conv-MVS was rarely described. This does not guarantee that crossovers did not occur.
Baseline characteristics are listed in Table 1. The patients in the mini-MVS group were younger on average by 1.6 (−2.7 to −0.4) years compared with conventional open mitral valve repair/replacement (MVR) (54.0 ± 12.4 years vs 55.8 ± 12.6 years, respectively). The percentage of females was similar for both groups (42%). The mini-MVS group had less baseline renal failure (1% vs 2%), less chronic obstructive pulmonary disease (11% vs 12%), and lower pulmonary artery pressures (42 vs 45 mm Hg) compared with the conv-MVS group. For other baseline characteristics, there were no statistically significant differences detected between mini-MVS and conv-MVS for most reported baseline characteristics; however, some characteristics were reported in only a few studies, and there was significant heterogeneity among the studies for a number of baseline characteristics. Although baseline patient characteristics were similar for randomized and nonrandomized studies, the small number and size of the randomized studies and the inconsistent reporting of baseline characteristics in nonrandomized studies precluded adequate power to confidently rule out whether potential differences exist.
All-cause mortality did not differ between minimally invasive and conventional open MVR (1.2% vs 1.5%; RR 1.03, 95% CI 0.75–1.42; 20 studies).5,6,8,12,13,15–19,21,26,28,31,32,34–38 Two RCTs involving 140 patients reported no mortality in minimally invasive and conventional open MVR.12,13 Subanalysis by endoaortic or transthoracic clamping also showed no significant difference for all-cause mortality between minimally invasive and conventional open MVR. There was no significant heterogeneity among the studies for this outcome (Fig. 2; Table 2).
Neurologic Outcomes Up to 30 Days
Risk of stroke was significantly increased for mini-MVS compared with conv-MVS (2.1% vs 1.2%, RR 1.79, 95% CI 1.35–2.38; 11 studies).6,8,14,16,17,19,21,32,36–38 No RCTs reported stroke; however, subanalysis of two propensity comparison studies also showed significant increase of stroke in minimally invasive compared with conventional open MVR (1.9% vs 0.9%, RR 2.02, 95% CI 1.40–2.94; two studies). The results of these higher quality propensity-adjusted studies were consistent with the remainder of the cohort studies reporting on this outcome (Fig. 3; Table 2).
Subgroup analysis by endoaortic or transthoracic clamp suggested a consistent trend toward increased risk of stroke in the mini-MVS patients of the endoartic clamp subgroup (RR 1.72, 95% CI 0.91–3.23; P = 0.09; five studies)17,19,32,36,38 and significantly increased risk of stroke in the studies reporting mixed use of endoaortic and transthoracic clamp (RR 1.84, 95% CI 1.33–2.55; three studies),6,16,37 whereas no such trend toward increased stroke existed in the studies using only transthoracic clamp (RR 0.80, 95% CI 0.07–8.92; three studies).8,14,21 However, these results are constrained by the usual caveats associated with subanalysis of retrospective data. There was no significant heterogeneity among the studies for the overall as well as all the subgroup analyses (Fig. 4).
Transient neuropathy was not significantly different between mini-MVS and conv-MVS (RR 5.09, 95% CI 0.54–48.11; two studies).8,14 There was no significant heterogeneity among the studies, and no RCTs reported on this outcome.
Permanent Neurologic Deficit
Permanent neurologic deficit was not significantly different for mini- versus conv-MVS; however, only one observational study including 200 patients reported on this outcome and was underpowered to rule out the potential for clinically relevant difference.19
Available outcomes for pain included Visual Analog Scale (VAS) and use of analgesics. Meta-analysis of five observational studies involving 576 patients showed no significant difference in the VAS for pain between mini- and conv-MVS (WMD −0.07 points, 95% CI −0.25 to +0.11 points; five studies).21,23,33,34,38 There was no significant heterogeneity among the studies. Another observational study involving 41 patients reported similar postoperative pain between two groups, except that pain tended to resolve more quickly in minimally invasive than conventional open MVR.18 In addition, minimally invasive patients had less difficulty in coughing and getting out of the bed than patients who underwent conventional open MVR.
An observational study involving 128 patients of transthoracic clamping reported the use of analgesics and found that the total intake of acetaminophen/oxycodone (in milligrams) for the first 3 postoperative days was slightly less in the mini-MVS group compared with conv-MVS (WMD −3.35 mg, 95% CI −5.74 to −0.96 mg).4 However, there was no difference in the use of morphine in the first 3 postoperative days between mini- and conv-MVS in this study.4
Transfusions Up to 30 Days
RBC Units Transfused
Units of red blood cells (RBCs) transfused in the mini-MVS groups were significantly less than in the conv-MVS group (1.5 ± 1.8 U vs 3.5 ± 2.9 U, WMD −1.85 U, 95% CI −2.48 to −1.22 U; 10 studies).5,6,8,12,14,17,18,22,34,35 However, there was significant heterogeneity among the studies. The one RCT that reported this outcome found no significant difference in the amount of RBCs transfused between groups (WMD −0.55 U, 95% CI −1.21 to +0.11 U) (Table 3).12
Patients Transfused RBCs
The percentage of patients receiving blood was similar for mini-MVS and conv-MVS groups (39% vs 42%, RR 0.97, 95% CI 0.81–1.17; seven studies).8,12,14,15,32,37,38 There was no significant heterogeneity among the studies (Table 3).
Units of Fresh Frozen Plasma Transfused
Significantly fewer units of fresh frozen plasma were transfused for mini-MVS compared with conv-MVS (0.3 ± 1.4 U vs 0.7 ± 1.5 U, WMD −0.44 U, 95% CI −0.81 to −0.08 U; two studies).12,14 There was no significant heterogeneity among the studies. One RCT reported the percentage of patient receiving fresh frozen plasma and found no significant difference between mini-MVS versus conv-MVS (20% vs 25%, RR 0.80, 95% CI 0.25–2.55; one RCT) (Table 3).12
Platelet Transfusion and Cryoprecipitate Transfusion
A single study reported platelet units transfused and showed a tendency toward lesser units of platelets transfused for mini-MVS versus conv-MVS (WMD 1.10 U, 95% CI −2.28 to 0.08 U; one study).14 This study also reported a tendency toward fewer units of cryoprecipitate transfused (WMD −1.10 U, 95% CI −0.21 to +0.01 U; one study) (Table 3).
The percentage of patients receiving platelets transfusion was significantly less with mini-MVS compared with conv-MVS (16% vs 19%, RR 0.80, 95% CI 0.73–0.88; two studies).37,38 There was no significant heterogeneity between the studies. The percentage of patients receiving cryoprecipitate did not differ between groups (RR 3.0, 95% CI 0.12–72.8; one study) (Table 3).38
Hemorrhagic or Chest Tube Drainage
Chest tube drainage was significantly decreased with mini-MVS versus conv-MVS (mean 578 ± 406 mL vs 871 ± 711 mL, WMD −266 mL, 95% CI −437 to −95 mL; 14 studies).5,6,8,11–15,17,18,21,24,31,34 The studies generally did not report whether chest tube drainage was serous or sanguineous. The two RCTs reporting this outcome together suggested no significant reduction in total drainage.12,13 There was significant heterogeneity among the studies for the overall as well for the RCTs and non-RCT subgroups considered alone (Table 3).
Reoperation for Bleeding
Reoperation for bleeding was not significantly different between mini-MVS and conv-MVS groups (3.5% vs 2.9%, RR 0.91 95% CI 0.59–1.41; 14 studies).6,8,12,14,15,21,26,28,32,34–38 There was no significant heterogeneity among the studies (Fig. 5; Table 3).
Atrial fibrillation was significantly reduced with mini-MVS versus conv-MVS (18% vs 22%; RR 0.87, 95% CI 0.77–0.99; eight studies).7,8,14,18,32,36–38 There was no significant heterogeneity among the studies for this outcome. No randomized studies reported this outcome. Need for pacemaker implantation was not significantly different between mini-MVS and conv-MVS (RR 0.47, 95% CI 0.05–4.09; two studies).8,12 Only one study, a RCT involving 40 patients, reported on ventricular fibrillation and found no statistically significant difference between mini- and conv-MVS (RR 0.33, 95% CI 0.01–7.72; Table 4).12
Aortic Dissection and Injury
Risk of aortic dissection was significantly increased for mini-MVS versus conv-MVS (0.2% vs 0%, RR 6.04, 95% CI 1.06–34.47; seven studies).5,19,20,26,32,36,37 There was no significant heterogeneity among the studies for this outcome. No RCTs reported this outcome. When aortic dissection and aortic injury (iatrogenic injury) were considered together, there was a significant risk of either dissection or injury for mini-MVS versus conv-MVS (RR 5.68, 95% CI 1.23–26.17; nine studies; Fig. 6; Table 4).5,8,19,20,26,28,32,36,37
Cardiac Events in 30 Days
Perioperative Myocardial Infarction
Perioperative myocardial infarction was not significantly different between groups (0.5% vs 2.0%, RR 0.39, 95% CI 0.06–2.46; three studies).12,14,26 The one RCT that reported this outcome did not find a difference for myocardial infarction between groups.12 There was no significant heterogeneity among the studies (Table 4).
The proportion of patients with mitral insufficiency postoperatively was not significantly different between mini-MVS and conv-MVS groups (2.4% vs 2%, RR 1.16, 95% CI 0.38–3.57; five studies).12,15,21,28,38 One of the studies was an RCT, and it showed no significant difference between groups for mitral regurgitation.12 There was no significant heterogeneity among the studies (Table 4).
Thromboembolic Events and Vascular Complications
There was no significant difference in the percentage of patients experiencing thromboembolic events or vascular complications between mini-MVS versus conv-MVS (0.3% vs 0.2%, RR 1.78, 95% CI 0.82–3.87; seven studies), and there was no significant heterogeneity among the studies.8,14,15,18,20,36,37 No RCT reported this outcome (Table 4).
Other Cardiovascular Outcomes
There was no difference in other cardiac outcomes including low cardiac output syndrome, cardiac tamponade, pericardial effusion, ventricular rupture, pericardiotomy syndrome, need for inotropes, and need for intra-aortic balloon pump implantation between groups; however, in most cases, only one or two studies reported these outcomes (Table 4).
No significant difference was found between mini-MVS versus conv-MVS for patients experiencing postoperative renal failure or insufficiency (RR 1.08, 95% CI 0.81–1.43; five studies), and there was no significant heterogeneity among the studies.8,14,22,26,37
There was no significant difference between groups for pneumonia, pleural effusion, pneumothorax, pneumonitis, or overall pulmonary complications, and these outcomes were generally reported by only one or two nonrandomized studies (Table 5).
Phrenic Nerve Palsy
Although the proportion of patients experiencing phrenic nerve palsy was significantly increased for mini-MVS compared with conv-MVS (3% vs 0%, RR 7.61, 95% CI 1.30–44.70; three studies, no heterogeneity among the studies), this result was limited by few studies reporting on the outcome, and presumably number of these studies had zero events.8,14,26 In some cases, there was prolonged impact on ventilation time and ICU stay, whereas in other cases the clinical impact was limited to phrenic nerve elevation or was not described (Fig. 7; Table 5).
There was no significant difference in gastrointestinal events between mini-MVS versus conv-MVS (1.3% vs 3%, RR 0.54, 95% CI 0.12–2.32; two studies).8,14 There was no statistically significant heterogeneity among the studies.
Infections and Mediastinitis
As expected, significantly fewer patients experienced sternal infections with mini-MVS compared with conv-MVS (0.04% vs 0.27%, RR 0.34, 95% CI 0.12–0.95; seven studies), and there was no heterogeneity across studies. Of the seven studies reporting this outcome,12–14,31,34,35,37 two were RCTs and together they suggested a trend toward reduction (Table 6).12,13 Although it seems counterintuitive that sternal wound infections could occur in mini-MVS, a total of two cases of sternal wound infections were reported in two studies involving 4377 patients.14,37 In both of these studies, robotic assistance was used in at least a proportion of the patients in the mini-MVS group.
There was no significant difference in the risk of mediastinitis between mini-MVS and conv-MVS (2% vs 1%, RR 1.85, 95% CI 0.15–23.07; two studies). There was no statistically significant heterogeneity among the studies (Table 6).6,26
Groin Infections, Hematoma, or Hydrocele
Groin infections, hematoma, and hydrocele were confined to mini-MVS, and none of these occurred in conv-MVS (2% vs 0%, RR 5.62, 95% CI 1.26–25.13; five studies).12,14,15,26,36 There was no statistically significant heterogeneity among the studies (Table 6).
Deep Infection and Wound Dehiscence
There was no significant difference between mini-MVS and conv-MVS for deep infection and wound dehiscence (0.7% vs 1.3%, RR= 1.09, 95% CI 0.12–10.20; three studies). There was statistically significant heterogeneity among the studies for the overall comparison (Table 6).16,31,32
Superficial Infection and Keloid Formation
One observational study reported no significant difference in superficial infections (0% vs 5.8%, RR 0.14, 95% CI 0.01–2.70) and keloid formation (0% vs 3.8%, RR 0.20, 95% CI 0.01–4.07) between mini-MVS and conv-MVS (Table 6).31
Incision Length and Scar Dissatisfaction
Mean length of incision in the mini-MVS group was 6.3 ± 1.5 cm (reported in nine studies).5,6,8,13,18,23,25,34,36 However, one of these studies reported a disproportionately longer length of incision in mini-MVS (11.5 ± 0.7 cm), which does not reflect usual practice for mini-MVS.13 Only two studies reported the length of incision in both groups, and together they suggested that the mean incision length was significantly reduced for mini-MVS versus conv-MVS by >16 cm (8.4 vs 24.8 cm for mini- vs conv-MVS; WMD −16.35 cm, 95% CI −19.48 to −13.21 cm).18,34 There was a significant heterogeneity among the studies.
Patient Dissatisfaction With Scar
Significantly fewer patients were dissatisfied with their scar with mini-MVS versus conv-MVS (0% vs 19.2%, RR 0.05, 95% CI 0–0.79; one study) in the one observation study that reported this outcome (Table 6).31
There was no significant difference between mini-MVS and conv-MVS for the percentage of patients who experienced no hospital complications (77% vs 80%, RR 0.92, 95% CI 0.65–1.28; two studies) (Table 6). There was significant heterogeneity among the studies, and no RCTs reported this outcome.19,26 When overall complications were defined as STS complications, there was no significant difference between mini-MVS versus conv-MVS for the percentage of patients with Society of Thoracic Surgeons (STS) defined complications (8.5% vs 9.4%, RR 0.90, 95% CI 0.38, 2.13; one study).38
Quality of Life
One observational study reported that quality of life tended to improve similarly after mini-MVS and conv-MVS, with no significant difference between groups.33
There was no significant difference in the percentage of patients having repeat procedure between mini-MVS and conv-MVS (RR 0.61, 95% CI 0.12–3.25; one study) (Table 6).36
Readmission Within 30 Days
There was no significant difference in the percentage of patients readmitted within 30 days between mini-MVS and conv-MVS (4.7% vs 9.0%, RR 0.53, 95% CI 0.24–1.15; one study) (Table 6).36
Conversion to Sternotomy
Ten studies involving 866 patients undergoing minimally invasive MVR reported that 3.7% of the patients undergoing minimally invasive patients were converted to open sternotomy.5,8,12,14,15,18,19,21,36,38 In one RCT reporting this outcome, there was no incidence of conversion from planned mini-MVS to conv-MVS.12
Cross-clamp time was significantly increased with mini-MVS versus conv-MVS (95 ± 39 minutes vs 74 ± 36 minutes, WMD 21.41 minutes, 95% CI 10.14 to 32.69 minutes; 24 studies).4,7,8,11–15,17,18,21,22,25–29,31,32,36–38 Another study30 reported that in the beginning of the learning curve, cross-clamp time was 25 minutes longer in the mini-MVS group compared with conv-MVS. However, with experience cross-clamp time improved in their center but still remained 15% longer in the minimally invasive group. However, subanalysis of RCTs showed shorter cross-clamp time with mini-MVS versus conv-MVS (56 ± 16 minutes vs 60 ± 19 minutes, WMD −4.0 minutes, 95% CI −7.2 to −0.7 minutes; two RCTs), which is opposite to what the nonrandomized studies reported (Fig. 8; Table 7).12,13
Cardiopulmonary Bypass Time
Cardiopulmonary bypass (CPB) time was significantly increased with mini-MVS versus conv-MVS (144 vs 111 minutes, WMD 33.0 minutes, 95% CI 18.9 to 47.1 minutes; 27 studies).4–8,11–15,17–19,21,22,25–29,31,32,34–38 There was significant heterogeneity across studies for this outcome. When subanalyzed, the two RCTs that reported this outcome also showed significantly increased CPB time with mini-MVS versus conv-MVS (Table 7).12,13
There was a significant increase in procedure time for mini-MVS compared with conv-MVS (4.5 vs 3.7 hours, WMD 0.79 hours, 95% CI 0.41 to 1.16 hours; 13 studies).5,6,12,14,15,18,21,25–27,34,37 The single RCT that reported this outcome showed no significant difference in procedure time between groups (WMD 0.24 hours, 95% CI −0.31 to +0.79 hours; one study).12 There was significant heterogeneity among the studies for this outcome (Fig. 9; Table 7).
Length of Ventilation
Ventilation time was significantly reduced for mini-MVS compared with conv-MVS (12.6 vs 19.9 hours, WMD −2.07 hours, 95% CI −3.39 to −0.75 hours; 18 studies).5,7,8,11,12,14,15,17,18,27–29,31,32,34–37 There was significant heterogeneity across studies for this outcome. The one RCT that reported this outcome showed no difference in the length of ventilation between mini- and conv-MVS (WMD −3.0 hours, 95% CI −7.4 to +1.4 hours) (Table 7).12
Length of Stay in ICU
Length of stay in ICU was significantly reduced for mini-MVS versus conv-MVS (1.6 vs 2.4 days, WMD −0.50 days, 95% CI −0.68 to −0.32 days; 18 studies).5,7,8,11,12,14,15,17,21,24,27–29,31,33,35–37 There was significant heterogeneity for this outcome. The one RCT that reported this outcome showed no significant difference in the length of stay in ICU between mini- and conv-MVS (WMD −0.30 hours, 95% CI −92 to +0.32 hours) (Table 7).12
Length of Stay in Hospital
Length of hospital stay was significantly reduced with mini-MVS versus conv-MVS (mean 6.9 vs 8.9 days, WMD −1.60 days, 95% CI −2.09 to −1.11 days; 26 studies).4,5,7,8,10–15,17,18,21,24,26–29,31–38 Subanalysis of the two RCTs reporting this outcome showed no difference in the length of stay in hospital between groups (WMD −0.04 days, 95% CI −0.8 to +0.7 days; two studies).4,13 There was significant heterogeneity among the studies for this outcome (Fig. 10; Table 7).
A study from the United States reported that compared with patients undergoing conv-MVS, patients undergoing mini-MVS had lower mean hospital charges (WMD −$11,430, 95% CI −$12,420 to −$10,440; one study), and lower mean hospital costs (WMD −$9,160, 95% CI −9,870 to −8,460; one study).8 Operating room charges in this study were procedure based, and this may not have accounted for the time-dependent costs. This is an important consideration, because operating room time for mini-MVS was approximately 1.6 hours longer than conv-MVS in this study. This could have masked the time-dependent cost variances. Another study from Slovenia reported that patients undergoing mini-MVS had lower mean costs than patients undergoing conv-MVS (WMD −664 Slovenia tolars, 95% CI −664 to −663 Slovenia tolars) (Table 8).17 None of the studies provided a comprehensive cost-effectiveness analysis of the incremental costs and benefits of mini-MVS versus conv-MVS.
Longer term risk of all-cause mortality was reported in few trials, and it did not show a significant difference between mini-MVS and conv-MVS at 1 year (0.9% vs 1.3%, RR 1.07, 95% CI 0.19 to 6.05; two studies),8,34 at 3 years (0.5% vs 0.5%, RR 1.06, 95% CI 0.07 to 16.79; one study)34 and up to 9 years (0% vs 3.7%, RR 0.19, 95% CI 0.01 to 3.64; one study).26 There was no significant heterogeneity among the studies, and no RCTs reported long-term mortality (Table 9).
Mitral Valve Leak or Insufficiency
No significant difference was found for the risk of long-term mitral valve leak between mini-MVS and conv-MVS (RR 9.5, 95% CI 0.4 to 226.7).8,21,34 There was significant heterogeneity among the studies, and no RCT reported long-term mitral valve outcomes. In addition, there was no significant difference in the grade of mitral insufficiency between mini-MVS and open-MVS (WMD 0.04, 95% CI −0.15 to +0.2).19,26,35 There was no significant heterogeneity among all these studies, and no RCT reported this outcome (Table 9).
There was no significant difference in the percentage of patients developing infective endocarditis between mini-MVS and open-MVS (0.5% vs 0.7%, RR 0.94, 95% CI 0.31–2.86; four studies).8,16,19,26 There was no significant heterogeneity among the studies, and no RCT reported this outcome (Table 9).
There was a small but significant decrease in NYHA class after 1 year for mini-MVS compared with conv- MVS (mean class 1.32 vs 1.52, WMD −0.26, 95% CI −0.27 to −0.25).19,26 There was no significant heterogeneity among the studies, and no RCTs reported this outcome (Table 9).
Freedom From Valve-Related Reoperation
There was no significant difference in freedom from reoperation between groups at 1 year (RR 1.03, 95% CI 0.97 to 1.10; one study).19 However, at 8 years, there was significant improvement in freedom from reoperation for mini-MVS compared with conv-MVS (RR 1.04, 95% CI 1.01 to 1.08; one study).16 Freedom from valve-related reoperation was significantly improved when all studies evaluating this outcome at 1 to 8 years follow-up were combined (RR 1.04, 95% CI 1.01 to 1.06; three studies); however, differences in follow-up between groups challenge the validity of this finding.16,19,26 There was no significant heterogeneity among the studies, and no RCT reported this outcome (Table 9).
Return to Normal Activity
Return to normal activity was significantly faster for mini-MVS versus conv-MVS (6.3 vs 12.3 weeks, WMD −4.96 weeks, 95% CI −6.39 to −3.52 weeks; three studies).4,18,34 There was significant heterogeneity across the studies for this outcome (Table 9).
Quality of Life
Quality of life was reported in only one observational study, and no significant difference was found in the change in quality of life between mini- and conv-MVS (Table 9).33
No study provided the cost or cost-effectiveness of mini-MVS versus conv-MVS over the long-term period.
This systematic review with meta-analysis suggests that mini-MVS is associated with decreased hemorrhage, blood product transfusion, atrial fibrillation, chest wound infection, ventilation time, ICU stay, hospital length of stay, and reduced time to return to normal activity, without detected adverse impact on long-term need for valvular reintervention and survival (though this meta-analysis was underpowered for long-term outcomes). However, an increased risk of stroke, aortic dissection or aortic injury, phrenic nerve palsy, groin infections, and increased CPB, cross-clamp, and procedure time was found.
No significant differences in mortality were found in this meta-analysis. This result is similar to another systematic review,1 which also found no difference in the overall mortality between mini- MVS and conv- MVS (odds ratio 0.46, 95% CI 0.15–1.42). Of note, an analysis of a large propensity-matched comparison study involving 8644 patients reported significant increase in the composite of major hospital morbidity or mortality with mini-MVS compared with conv-MVS (RR 1.17, 95% CI 1.05–1.32),37 raising the potential for concern about the safety of the minimally invasive approach. Unfortunately, other studies did not report death and major complications as a composite outcome, and we were unable to provide a synthesis for this outcome across all studies. Given that the above-mentioned study is large and propensity matched (in comparison to many of the studies in this analysis being unmatched retrospective comparisons), it is important to keep this potential for increased overall adverse events in mind. On the other hand, some centers contributing data to this propensity-matched comparison study were low-volume centers. Further contemporary comparative studies should be conducted to measure the overall risk of death and complications.
Our meta-analysis showed increase in stroke in mini-MVS compared with conv-MVS (RR 1.79, 95% CI 1.35 to 2.38). In contrast, Modi et al1 reported no difference in stroke between mini-MVS and conv-MVS (odds ratio 0.66, 95% CI 0.23–1.93). However, Modi included fewer studies in his meta-analysis and also included mini-sternotomy studies within the definition of mini-MVS. The cause of increased stroke with mini-MVS is not known and is beyond the scope of this meta-analysis. Potential hypotheses may be related to the restricted assess and technically challenging de-airing and/or the increase in procedure time and CPB time. In our subgroup analysis of aortic clamping approach, we found that the majority of increased strokes occurred in the studies using endoaortic cross-clamping. In the studies using primarily transthoracic clamping, there was no excess of strokes. This raises the hypothesis that mini-MVS using the endoaortic clamp may pose greater risk for patients compared with the transthoracic approach. However, it must be stressed that there may be other unexplored factors that could contribute to the increased risk of stroke, and this hypothesis needs further testing in well-conducted, adequately powered comparative trials before the true association can be concluded.
Phrenic Nerve Palsy
Our meta-analysis showed significant increase in the risk of phrenic nerve palsy/diaphragm elevation for mini-MVS versus conv-MVS (3% vs 0%). This may be a potentially serious adverse effect; however, the details of the clinical significance and duration of the palsy were incomplete.8,14 In addition, it is important to note that a number of the studies failing to report specifically on this outcome may have had zero events, although we could not assume this for the meta-analysis. Therefore, the interpretation of this result is limited by the small number of studies reporting on this outcome.
Pain and Functionality
It was anticipated that mini-MVS would be shown to induce less postoperative pain compared with conv-MVS. However, our meta-analysis found no significant difference in the VAS for postoperative pain between groups. Although the trend was toward reduction, the difference was very small with less than 1 point reduction on the VAS (WMD −0.07, −0.25 to 0.11 points on a 10-point VAS).21,23,33,34 Similarly, although intake of acetaminophen/oxycodone (in milligrams) was significantly less for mini-MVS compared with conv-MVS, the overall mean difference of 3 mg over the study period was very small and of questionable clinical relevance (WMD −3.35 mg, 95% CI-5.74 to −0.96 mg). There was no significant reduction in the use of morphine in the postoperative period.4 Therefore, although reduction in pain remains one of the key rationales for mini-MVS, it remains unproven by the current evidence base. Of all the studies published on mini-MVS versus conv-MVS, very few have reported on pain-related outcomes, and this should be an important part of any future research agenda. Meta-analysis of three studies showed faster return to normal activity of ∼5 to 6 weeks (6.3 ± 4.8 weeks vs 12.3 ± 3.8 weeks) for mini-MVS versus conv-MVS.4,18,34
Bleeding and Reoperation Outcomes
It is anticipated that the smaller incision of mini-MVS would lead to less hemorrhage; although there is concern that this could be counteracted by the longer procedure time required. Our meta-analysis showed a decrease in overall hemorrhage and chest tube drainage of ∼300 mL for mini-MVS compared with conv-MVS. This translated into significantly less patients receiving RBC and platelet transfusions. However, there was no statistically significant difference in the percentage of patients undergoing reoperation for bleeding between minimally invasive and conventional open mitral valve surgery.6,8,12,14,15,21,26,28,32,34–38 In contrast, reoperation for any cause (bleeding or any other reason) was significantly increased by ∼1% for mini-MVS versus conv-MVS (6.6% vs 5.5%, RR 1.20, 95% CI 1.01–1.41), although this was reported in just one study (propensity-matched comparison study involving 8644 patients).37
Our meta-analysis found a significant reduction in atrial fibrillation with mini-MVS compared with conv-MVS (absolute risk reduction 4%). This reduced risk of arrhythmia may be related to the less traumatic surgical approach used during mini-MVS resulting in lesser inducement of inflammatory mediators. Although atrial fibrillation was significantly reduced in our meta-analysis, there was no significant reduction in pacemaker implantation; however, this outcome was less commonly reported in the studies. Other cardiac outcomes of interest in our meta-analysis did not reach significant differences between groups, including myocardial infarction, left heart decompensation, hypotension, need for inotropes, intraaortic balloon pump implantation, ventricular rupture, cardiac tamponade, pericardial effusion, venous thromboembolism, arterial embolism, and in-hospital mitral insufficiency.
Our meta-analyses showed a significant increased risk of aortic dissection or iatrogenic aortic injury; however, the absolute risk increase of 0.2% was small (0.2% vs 0% for mini- vs conv-MVS, RR 6.04, 95% CI 1.06 to 34.47) The reason for the increased risk of aortic dissection and aortic injury remains unclear (ie, differences in clamping or added difficulties in manipulating the aorta during minimally invasive surgery).
Other Clinical Complications
Our meta-analysis showed no significant difference for postoperative renal failure or renal dysfunction between groups. Because few studies measured the incidence of renal failure, the current evidence remains underpowered to safely rule out whether important differences may exist between mini-MVS and conv-MVS for this clinically relevant outcome.
Pulmonary events were not significantly different between groups, including pneumonia, pneumothorax, pneumonitis, mediastinitis, and pleural effusion. Similarly, gastrointestinal complications did not differ between groups.
Infections and Scar Length
As expected, our meta-analysis found a significant decrease in sternal infections and dehiscence, but an increase in groin infections, hematoma, or hydrocele. Incision length was significantly reduced by ∼16 cm between groups, although these data were rarely provided for both groups in the studies and may not be reflective of the average practice difference between mini-MVS and conv-MVS. Only one study examined patient satisfaction with the scar and reported an improvement in satisfaction in patients undergoing mini-MVS compared with conv-MVS.
Overall Complications and Readmissions
Our meta-analyses showed no significance in the freedom from all hospital morbidity, quality of life, requirement of repeat procedure, and readmission within 30 days of postoperative period.
Conversion to open sternotomy was reported in 3.7% of patients; however, relatively few studies reported this outcome (12 studies, including 866 patients). The absence of reporting on this outcome does not guarantee that no conversions occurred, and the concern remains that retrospective studies are less likely to ascertain this information because it would be difficult to determine retrospectively from patient charts whether patients who received open sternotomy were originally intended to receive mini-MVS. The inadvertent inclusion of converted patients in the conventional open MVS arm may unfairly bias the results of this meta-analysis toward favoring mini-MVS because converted patients tend to have worse outcomes than patients undergoing their originally intended surgery.
Our meta-analyses found significantly increased cross-clamp time of ∼20 minutes, (95 ± 39 minutes vs 74 ± 36 minutes), increased CPB time of ∼33 minutes (144 ± 52 minutes vs 111 ± 52 minutes), and increased procedure time of close to 1 hour (4.5 ± 1.8 hours vs 3.7 ± 1.7 hours) for mini-MVS versus conv-MVS. Ventilation time was reduced by 2 hours (12.6 ± 17.7 hours vs 19.0 ± 36.3 hours). Length of stay in ICU was reduced by 0.5 days (1.6 ± 1.7 days vs 2.4 ± 2.4 days), and hospital stay was significantly reduced by ∼2 days (6.9 ± 4.2 days vs 8.9 ± 5.1 days) for mini-MVS versus conv-MVS. These results were of similar direction as in the systematic review by Modi et al,1 although the analysis by Modi et al had insufficient power to detect the difference for some of these outcomes. There was significant heterogeneity for each of these outcomes, which is likely due to the different learning curves across the different centers.
It is notable that in the randomized trials, the differences between groups for procedure-related outcomes and length of stay were less stellar than in the observational studies. For example, the randomized trials suggested that cross-clamp time was slightly decreased by 4 minutes (rather than increased by 23 minutes as in the observational studies), procedure time did not differ significantly, and length of stay in ICU and hospital were not reduced significantly for mini-MVS versus conv-MVS. The differences between randomized trials (with comparable groups at baseline) and nonrandomized trials (with potentially important differences between groups at baseline) may be related to the differences in surgeon experience with mini-MVS, different settings, and patient selection criteria (which may unfairly bias the results of observational studies if different criteria are used for mini- vs conv-MVS). The lack of differences in the randomized trials may also be related to their small sample sizes (n = 140 in total).
Few studies reported on the relative cost of mini-MVS versus conv-MVS. One study each from the United States and Slovenia suggested that mini-MVS was associated with significantly lesser costs than conv-MVS. However, the relevance of these findings to other settings in other countries remains uncertain. In addition, these studies were cost analyses only and did not attempt to calculate the incremental cost-effectiveness of mini-MVS versus conv-MVS over the time horizon of a patient lifetime, or even over the midterm, by including follow-up costs during the midterm.8,17
Longer Term Outcomes
Few studies provided long-term outcomes, and conclusions remain uncertain based on the results provided due to the paucity of data provided on completeness of follow-up in these studies. Of the studies that reported longer term outcomes (1 to 9 years of follow-up), there was no significant difference between groups for long-term mortality, quality of life, presence of mitral valve leak, grade of mitral insufficiency, risk of infective endocarditis, and need for reoperation. There was a small but significant decrease in NYHA class (WMD −0.26 class, 95% CI −0.27 to −0.25 class) and decreased need for reoperation for mini- versus conv-MVS; however, the clinical relevance and generalizability are uncertain given that the reductions were small and were based on a single study.
Strengths and Limitations
This meta-analysis is based on a comprehensive search of several databases to identify all relevant comparative data, and it complies with the latest methodologic recommendations for comprehensive systematic reviews of observational and randomized studies. In addition, this meta-analysis reports on all available clinically relevant and resource-related outcomes, rather than selectively reporting only a few outcomes. However, this meta-analysis is limited by the type of studies and the quality of the data provided in those studies. Most studies in this meta-analysis were retrospective series that compared minimally invasive experience at one center with their own (or others') experience with conv-MVS. In some studies, the series for mini- and conv-MVS were noncontemporaneous series, and the more historic nature of the conventional group may bias the results toward favoring mini-MVS if more contemporary settings provide better outcomes in general (regardless of which procedure would have been provided). In addition, in many series the surgeons who performed mini-MVS were different than the surgeons who provided conv-MVS. Differences in skill sets and differences in progression through the learning curve could bias the results for or against mini-mitral. Improved characterization of the learning curve is also needed. Documentation of the improved procedure-related outcomes over time due to the learning curve effect has been reported in at least one study.30 High-volume centers should perform comparative trials to determine whether better outcomes can be attained with experience.
Several outcomes were reported inconsistently across studies, including some of the most basic outcomes which would provide the rationale for mini-MVS including differences in pain, functionality, patient satisfaction, and quality of life. Also of key concern, most studies failed to provide data on outcomes beyond the patient's discharge from hospital.
Another significant limitation was the potential for overlap among the studies, which was difficult to ascertain in a number of cases due to insufficient reporting of dates of patient recruitment and repeated publications from the same authors or institutions. This issue requires further exploration to elucidate the independence among these patient sets. In addition, the largest data set comes from the STS database analysis by Gammie et al, and it may overlap significantly with some of the most recent studies from U.S. centers. Removing the smaller U.S. center studies from the graphs in the cases where both the study by Gammie et al and the potentially overlapping studies have reported on the same outcome did not have any material impact on the conclusions (because the weight of the study by Gammie et al was large, and the weight of the other singular studies was small). There are a number of U.S. studies that clearly did not overlap the study by Gammie et al because their date of recruitment was earlier than that covered by Gammie et al.
Implication for Practice and Further Research Required
Given the results of this imperfect evidence base, definitive conclusions regarding the overall benefit versus risk of mini-MVS versus conv-MVS are not possible. It seems that mini-MVS improves some important outcomes, but may increase the risk of other serious outcomes. Future studies should assess the volume-outcome relationship for mini-MVS versus conv-MVS. Further research should be encouraged. To ensure that any true differences between mini-MVS and conv-MVS can be detected, and that the magnitude of benefit versus risk can be measured objectively, adequately powered randomized trials should be undertaken to measure stroke, pain, patient satisfaction, overall major complications, need for reoperation, return to normal activity, and quality of life while in hospital and over the longer term. In addition, any future nonrandomized studies should increase validity through prospective and long-term follow-up for clinically important outcomes.
The authors are aware that recruitment for such a randomized trial may be perceived to be difficult in institutions where mini-MVS have been adopted before comparative evidence being available; however, this was equally true for other precedents in surgery, including the eventual conduct of randomized trials to test the widely accepted extracranial to intracranial (EC/IC) bypass to prevent ischemic stroke, even after it had been chosen as standard of care by many centres. Importantly, the eventual fair testing through randomized studies showed that EC/IC bypass surgery actually increased the risk of stroke compared with medical management.40 Although some centers report excellent short- and long-term results, these reports are usually noncomparative case series with uncertainty regarding consecutive recruitment. On the other hand, the studies in this meta-analysis include a mixture of new and older studies, and the early parts of the learning curve may have contributed to the results. Future study should evaluate the volume-outcome relationship for mini-MVS.39
Current evidence suggests that mini-MVS may be associated with decreased bleeding, blood product transfusion, atrial fibrillation, sternal wound infection, scar dissatisfaction, ventilation time, ICU stay, hospital length of stay, and reduced time to return to normal activity, without detected adverse impact on long-term need for valvular reintervention and survival beyond 1 year (although few studies reported long-term outcomes). However, these potential benefits for mini-MVS may come with an increased risk of stroke, aortic dissection or aortic injury, phrenic nerve palsy, groin infections/complications, and increased cross-clamp, CPB, and procedure time. Available evidence is largely limited to retrospective small cohort comparisons of mini-MVS versus conv-MVS that provide only short-term outcomes. Given these limitations, randomized controlled trials with adequate power and duration of follow-up to measure clinically relevant outcomes are recommended to determine the balance of benefits and risks.
The authors thank Ms. Aurelie Alger and Ms. Elizabeth Chouinard for their assistance in organizing the consensus conference. They thank Jennifer Podeszwa DeOliveira and Erin Boyce for facilitating literature searches and retrieval. They also thank Brieanne McConnell for her assistance with citation management.
© 2011 Lippincott Williams & Wilkins, Inc.