Comparison of outcomes between minimally invasive and median sternotomy for double and triple valve surgery: A meta‐analysis

Limited data exists demonstrating the efficacy of minimally invasive surgery (MIS) compared to median sternotomy (MS) for multiple valvular disease (MVD). This systematic review and meta‐analysis aims to compare operative and peri‐operative outcomes of MIS vs MS in MVD.


| INTRODUCTION
Despite the wide use of minimally invasive surgery (MIS) across multiple specialities, registry data highlights the lack of penetration of this approach into most aspects of cardiac surgery. 1 The practical complexity and resulting steep learning curve involved in MIS is the main factor that inhibits its use over median sternotomy (MS). 2 Nevertheless, when compared with MS, numerous studies reviewing MIS techniques have reported benefits including reduced morbidity, shorter hospital-stay, improved cosmesis, and patient satisfaction. 3,4 In the field of MIS for isolated aortic valve (AV) or mitral valve (MV) surgery, there is an accumulating body of evidence (including randomized control trials) demonstrating its efficacy compared to MS. 5 Furthermore, it is not uncommon for patients undergoing valvular heart surgery to have more than one disease-afflicted valve. This could either be as a separate disease processes or a direct consequence (classically mitral incompetence leading to tricuspid valve [TV] incompetence). 6 Multiple valve disease (MVD) is more prevalent in the elderly population. 7 In the 2001 EuroHeart Survey, more than 20% of patients with native valve disease were found to have MVD. 8 More recently, undiagnosed MVD was identified in over 30% of participants aged ≥ 65 years in a largescale echocardiographic screening programme involving 2500 participants. 9 The technical challenges for treating MVD is increased when performing procedures via smaller incisions during MIS.
Reports of acceptable outcomes of MIS for MVD have emerged within the last two decades in smaller groups of patients. However, high quality evidence testing the efficacy for MIS vs MS for MVD is lacking. In this meta-analysis of nonrandomised studies, we aim to evaluate the outcomes of MIS compared to MS for the treatment of MVD.

| Search methodology
An extensive literature search was conducted in accordance with methodological recommendations by PRISMA through PubMed, Ovid, and Embase using a search strategy including the combined terms of: "mini" or "miniature" or "minimally" or "right" or "anterior" AND "sternotomy" or "access" or "incision" or "thoracotomy" AND "aorta" or "aortic" or "double valve" AND "mitral" AND "tricuspid" from their dates of inception to August 2019 as seen in Appendix 1. 10 The studies resulting from this search were distributed among the authors to be screened based on their titles and abstracts for eligibility based on the objectives devised within the predefined inclusion criteria. The reference lists of the pertinent articles were further reviewed to identify any relevant studies that may have not been included in the initial database search.

| Selection criteria
Both retrospective and observational studies involving patients receiving combined MV and AV procedures with or without TV annuloplasty through MIS vs MS were considered. The pathophysiology of valvular disease was not taken into consideration. The exact form of minimal access was not restricted, but was commonly reported as a less than or equal to 6 cm anterolateral thoracotomy (ALT) in the third to fifth intercostal space (ICS) enabling adequate surgical exposure as illustrated in Appendix 2. 11 Studies were excluded if they: 1) Involved patients receiving isolated valve procedures only 2) Did not compare the two treatment modes of interest 3) Were published as case reports, editorials or commentaries Following contemporary practice, articles were restricted to English language and human subjects. Two reviewers (HM and MYS) assessed all titles and abstracts for inclusion. Where a single institution published multiple cohort studies, only the largest and most recent informative studies were included. Any disagreements between reviewers were settled by discussion and where necessary, consensus with the senior author (HV).

| Data extraction and evaluation of quality of evidence
Standardised data extraction forms were devised for the collection of quantitative and qualitative data. The majority of the variables were reported as a mean ± standard deviation (SD) or number (n) and percentage (%). However, where data was presented as medians with interquartile ranges (IQR, 25-75%), we referred to the Cochrane guidelines for expressing medians and IQR into means with SD. 12 In addition, the methodological quality of the pooled nonrandomised studies was assessed using the Newcastle-Ottawa scale as shown in Appendix 3. 13

| Statistical analysis
The odds ratio (OR) or weighted mean difference (WMD) were used as the summary statistic for the incidences of short-term complications. A random effects meta-analysis was used to find an overall OR comparing MIS valve surgery with MS for operative mortality due to the expected heterogeneity between the studies. Heterogeneity was investigated using Cochrane's test and the I 2 statistic, with I 2 values interpreted according to the Cochrane collaboration. 12 Funnel plots were generated to assess for publication bias. Peter's test for small studies was conducted to rule out large effects from potentially nonsignificant studies. Meta-regression analysis was used to investigate the effects of covariates, including patient and operative characteristics. Statistical analysis was conducted using the Stata 13.0 software (Stata Corp, College Station, TX).

| Study selection
Initially, a total of 902 studies were identified through online databases with an additional 95 citations found through other sources.
After screening titles and abstracts, 30 relevant articles remained for full-text review against the inclusion criteria. Of these, 25 articles were excluded on the basis of either being case reports, commentaries, single arm studies, animal studies or including the wrong patient population hence, leaving five studies eligible for review.
Assessment for quality of evidence demonstrated all five studies published between 2010 and 2019 to be of "good" quality. [14][15][16][17][18] Two of the studies 14,18 included propensity-matched cohorts in their analyses, the data of which were included in our analysis in preference over the unmatched data. The total amongst the included studies were 340 patients in the MIS and 414 patients in the MS group. This information is summarized in Figure 1 and Table 1.
The characteristics of the five included studies, three of which were retrospective and the remaining two being observational cohort studies are demonstrated in Table 1. 14-18 Mean follow-up time was recorded for all five studies. A funnel plot analysis revealed little evidence of publication bias as seen in Figure 2.

| Patient characteristics
The overall means of the patient baseline characteristics were similar between both MIS and MS groups (Table 1)

| Interventions
The general indication for operation was reported as combined valvular dysfunction among all studies. The type of surgical procedures varied between studies for the MIS and MS groups as presented in Table 3, although one study failed to clearly report the F I G U R E 1 PRISMA flow chart outlining the search and study selection with reasons for exclusion provided specific valves that were operated on in each cohort. 17 Combined AV and MV surgery was performed in two institutions 14,18 and similarly, combined MV and TV procedures were performed at two other centers. 15,16 Three of these studies also reported operations involving all three valves; aortic, mitral, and tricuspid (AV + MV + TV). 14,15,18 The overall proportion of patients receiving each of these concomitant valvular interventions was higher in the MS group among the four studies that provided the relevant information. [14][15][16]18 A right ALT through the third, fourth or fifth ICS (Appendix 2) was the approach taken in all studies. [14][15][16][17][18] One study also performed a "J" incision. 14 Similar cannulation techniques were reported in all studies involving arterial perfusion achieved via the femoral artery and venous access obtained via the femoral vein. 14-18

| Synthesis of evidence by outcome
The intra-operative and timing outcomes of the included studies comprising cardiopulmonary bypass (CPB) time and aortic crossclamp (AoX) are shown in Table 3 Table 4.

| Early mortality
Early or in-hospital mortality was reported in four studies (Table 4). [14][15][16]18 Postoperative deaths occurred more frequently in the MS group with one study reporting a greater than fivefold increase in deaths of patients receiving MS compared to MIS (2% vs 11%). 15 However, there was moderate heterogeneity between the studies and minimal evidence suggesting any significant difference in postoperative mortality between MIS and MS (OR, 0.72; 95% CI, 0.16-3.30; P = .670; I 2 = 57.5%; P = .070; Figure 4).

| Reoperation for bleeding
There was convincing evidence to suggest the risk of reoperation for bleeding was higher in the MS cohort, however reoperation for bleeding was reported in four studies (Table 4). 14 Figure 4).

| Meta-regression
Due to heterogeneity present among the studies, particularly with the variation in valves undergoing intervention, we performed meta-  29,39,40 This trend is consistent with our review since all included studies apart from one 14 reported reduced hospital stay for MIS cohorts. It has been postulated that preserved sternal stability in MIS may reduce the risk of SSI, improve respiratory function, increase mobility, thus decreasing the predisposition to pneumonia and prolonged hospital stay. 31 We found no significant differences between the two surgical approaches in reducing wound infection. [41][42][43] However, this may be confounded by misdiagnosis accuracy and varying definitions of SSI. Nevertheless, the clinical equivalence offered by MIS is reassuring when compared to MS.

| LIMITATIONS
The evidence included in our meta-analysis is based on singleinstitution, nonrandomised studies used which may be subject to bias. These studies did not separate the outcomes from cohorts of valve repair and valve replacement, follow-up was limited to midterm and none of the studies provided echocardiographic data. Furthermore, none of the studies included assessment of quality of life or patient satisfaction as an important postoperative outcome.

| CONCLUSION
MIS for MVD have similar short-term outcomes compared to MS.
This adds value to the feasibility of minimally invasive methods for multivalvular surgery, despite conferring longer operative times.
However, the paucity in literature and learning curve associated with MIS warrants further evidence, ideally randomized control trials, to support these findings.