Minimally‐invasive parasternal aortic valve replacement–A slow learning curve towards improved outcomes

We report our starting experience with parasternal minimally‐invasive aortic valve replacement.


| INTRODUCTION
Conventional aortic valve replacement (AVR) has experienced significant changes over the past 10 years. 1 The advent of transcatheter aortic valve implantation (TAVI) has resulted in a decrease in numbers and a growing fraction of patients is operated using minimally-invasive accesses, mainly partial sternotomy. 1 New trials push into lower risk patients and report non-inferior or even superior outcomes to conventional AVR. 2 Although minimallyinvasive approaches were allowed in these trials, the fractions of minimally-invasive procedures were only 24%. 2 Various investigators suggest a benefit of minimally-invasive AVR over sternotomy regarding the speed of recovery 3 or even mortality. 4 However, the vast majority of these investigations compared a partial sternotomy as minimally-invasive approach to sternotomy.
Various reports demonstrate promising results with a parasternal, sternotomy-sparing approach. 5 Miceli et al 6 compared partial sternotomy with a parasternal, sternotomy-sparing approach and report the best outcomes for the parasternal approach. We adopted a parasternal, sternotomy-sparing approach in 2014 and report our total experience of the first 4 years here, focusing on our learning experience.

| Patient selection
We included all patients who underwent minimally-invasive parasternal AVR or repair in our department between 7/14 and 8/18, AVR. Patients were selected based on anatomic feasibility judged by preoperative computed tomography (CT)-scan. Two of six surgeons in the department performed all parasternal procedures. Patients with porcelain aorta, left-sided aortic location (ie, the center of the most right-sided section of the ascending aorta is left of the center of the sternum on CT-scan) or suspicion of abscess formation in cases of endocarditis were not selected for parasternal approach. Since only two of six surgeons performed parasternal AVR, the department's fraction of parasternal approaches during this time was about 40%.
Ethics approval was waived by the committee. Every patient signs an individual consent form for the general use of their personal data for clinical research purposes.

| Operative technique
For parasternal AVR, cardiopulmonary bypass (CPB) was established using percutaneous groin cannulation as described elsewhere. 7 Central cannulation was used for severe peripheral artery disease in three cases. A 5 cm right parasternal incision over the second intercostal space was made and the pleural space was opened. The right internal mammary and vein were ligated and transected and the third rib was detached at the level of the sternal joint. CPB was started and the pericardium was opened. A left atrial vent and a cardioplegia cannula were placed and the aorta was cross-clamed using the Cygnet clamp. Bretschneider solution was used for antegrade cardioplegia as single shot. There were no differences in the way the valve prostheses were sized or selected compared to full sternotomy. We used replica sizing as described before. 8 At the end of the procedure the pericardium was closed, the rib reattached with a 1-0 vicryl suture and the wound closed in layers. CPB was ended and the Proglide system was used to close the arterial puncture site. 7 Figure 1 shows the operative setup, the way a classical stented valve is brought into the situs and the postoperative cosmetic result.

| Statistical analysis
Patients were separated into the first and second group of the entire experience and outcomes were compared using the student t test.
Values are given as means and standard deviation and a P < .05 was considered significant. Table 1 shows the demographic data of the two patient groups. There were 192 patients, which were separated into a first (early) and F I G U R E 1 Intraoperative photographs showing the minimallyinvasive surgical access in relation to a classic sternotomy incision (A, forceps in the thoracotomy and blue line indicating the site where a sternotomy incision would have been performed) and the way a prosthetic valve is brought into the chest through the mini-incision (B) as well as the postoperative results 5 days after surgery in a man (C) and a woman (D, notice the incision under the garment on the right side of the décolleté second (late) group for our total experience until 8/18 (96 patients in each group). There were no statistically significant differences between the two groups, although patients with endocarditis were only present in the first group (n = 4). Patients were on average 67 years old and 60% were male. Body mass index was 29 and mean ejection fraction was 59%. Stenosis was the main indication (85%), followed by regurgitation (15%). The cohort included patients with preoperative stroke (12%), severe chronic obstructive pulmonary disorder (11%), endocarditis (4% in the first group) preoperative reanimation (1.5%, all occurred within 3 weeks before surgery).

| RESULTS
There was no emergency operation. The Euroscore II tended to increase from 3.4% in the first to 3.7% in the second group of patients, but the difference was not significant. Table 2 shows the key operative data. The main implanted prosthesis type was biological (90%). Aortic annulus size (mean 24 mm) and implanted prosthesis size (mean 25 mm) was not different between groups. In both groups, most patients received a prosthesis that was at least one size larger than suggested based on intra-annular sizing. Operating times in the first group were just under 3 hours, with bypass times of 129 ± 32 minutes and clamp times of 77 ± 20 minutes. In the second group of the experience, these times changed only minimally, with stable clamp times and only 10 minutes reduction in bypass times. Table 3 shows the operative outcomes. There were three conversions to sternotomy in the first, but only one in the second group. Similarly, three patients died in the first, but only one in the second group. One patient died due to myocardial infarction

| DISCUSSION
We demonstrate in this study that parasternal minimally-invasive AVR is a feasible technique associated with a slow learning curve but the potential to achieve improved outcomes. Considering the transcatheter alternatives, the relative risk reduction may be worth investigating in future comparative trials.
The findings of reduced O/E ratios in a similar patient population in the second group compared to the first are promising. Although the number of patients with just under 100 in each group is not suitable to draw any solid conclusions, the improvement in outcomes is remarkable and the reductions in complications can be attributed to a learning curve with relative certainty.
For instance, the revisions for bleeding dropped significantly from the first to the second group. So did the conversion rate, which was in two cases due to uncontrollable bleeding form the aortic suture line in the first group. We had initially closed the pericardium and reattached the rib before coming off bypass, but after having experienced these complications, we switched to coming off bypass first and then close the pericardium and the chest. That explains the decrease in CPB times without changes in operating times. It also explains the lower rates of revisions for bleeding and conversions.
The anticipated problem with the need to interrupt ventilation after CPB to close the chest did not turn out to be an obstacle.
Although these changes resulted in better outcomes, they did not cause significant changes in the operating times. We would have expected a much faster change in our operating room times as we have noticed with other operative procedures (unpublished observations). It is our impression, that this parasternal approach is an elegant way leading to a high grade of patient satisfaction with outstanding cosmetic results ( Figure 1). However, it is also our impression that the learning curve is rather long and the technical challenge is much greater compared to a standard sternotomy or even partial sternotomy. This argument is supported by the rather long clamp times, which did not really decrease in our first 200 cases.
The argument is also supported by other surgeons who have reported learning experiences of this technique with limited decrease in operating times. 9 Glauber et al 10 reports his large 10 year experience with clamp times similar to ours in the first group. It may therefore also be argued that from the perspective of operating times, there is practically no learning curve because we were already within "normal range" in our initial half of the experience. Nevertheless, we believe that the learning curve is there and strive to shorten operating times in the future. This conclusion is further supported by our most recent experience with all cases that we performed after we closed the database for this analysis. In 50 additional cases we experienced no additional mortality and were able to reduce operating times by about 10 minutes.
Clamp times have always been considered a risk factor for mortality in cardiac surgery. 11 Thus, one may even suspect worse outcomes with this procedure. Although we did not perform a comparison to sternotomy patients in this learning curve analysis (where clamp times of the same surgeons would be substantially shorter), the fact that we achieved an O/E ratio of under 0.3 would not support this concern. In addition, all larger analyses addressing outcomes of minimally-invasive to sternotomy AVR did also not find a negative impact of the mini-approaches. 6,12,13 However, a specific analysis of this relationship in minimally-invasive cases is required.
Other concerns that may arise with any minimally-invasive approach are related to limited exposure and maneuverability within the field leading to potentially suboptimal quality of surgery at site. In other words, the limited access may lead to for instance the implantation of smaller aortic prostheses with poorer hemodynamic performance post op. We record our sizing strategy and the implanted sizes for all our patients. We demonstrated before that using replica sizing may allow implanting prostheses larger than suggested by the intra-annular sizing approach in the majority of cases. 8 We found the same fraction of "upsizing" in this minimallyinvasive patient population. From a technical standpoint it has to be mentioned that the valves have to be routinely taken off from their holders and need to be "shoehorned" into the root. This shoehorning is sometimes also necessary with the open technique. However, the small incision limits the maneuverability of the valve on its holder and the chance to deliver the valve directly is therefore smaller.
The majority of prostheses implanted were classic conventional types ( Table 2). We also used a small series of the Enable sutureless valve, until it was taken off the market. We have not used Perceval or Intuity with this approach since the delivery systems are so large that we are afraid not to be able to place the valve correctly through the limited incision and the at times distorted exposure in some cases (again limited maneuverability due to small incision). However, others have made remarkable experiences with these valves using the same approach. 14 These authors were also able to reduce clamp times using these new prostheses. Since concern has been raised recently with these new valves with respect to pacemaker rate, paravalvular leaks and pressure gradients, 15 problem. In addition, patient and possibly surgeon selection may have influenced the outcomes. For instance, our patient population had a high incidence of preoperative strokes (which were often old and no new neurological deficits were noted postoperatively). From a technical standpoint, we are in the process of conducting a randomized single center trial aimed at comparing our parasternal approach to sternotomy with the same surgeons performing the procedure.

ACKNOWLEDGMENTS
We would like to thank Christina Schenkl and Benjamin May for expert technical assistance in formatting and submission of the manuscript.