Impact of a surgical approach for implantation of durable left ventricular assist devices in patients on extracorporeal life support

The aim of this study was to evaluate the impact of the surgical approach on the postoperative outcome in patients who underwent left ventricular assist device (LVAD) implantation after having received veno‐arterial extracorporeal life support (va‐ECLS) using data from a European registry (ECLS‐VAD). Five hundred and thirty‐one patients were included.


| INTRODUCTION
Continuous-flow left ventricular assist devices (LVADs) are a standard treatment of patients with severe end-stage heart failure, 1 even in extremis for patients on temporary circulatory support. Implantation of durable LVADs in patients on veno-arterial extracorporeal life support (va-ECLS) is a challenging scenario associated with a high mortality rate, [2][3][4] and all efforts should be put in place to decrease perioperative morbidity and mortality. For several years, the routine approach for LVAD implantation has involved a full sternotomy on cardiopulmonary bypass. However, less invasive surgical techniques, including a sternumsparing approach 5 or partial sternotomy, employing ECLS for circulatory support during implantation or even off-pump or a combination thereof, 5,6 are becoming increasingly popular 1 as they appear to be associated with a reduction in surgical traumas, blood loss, and hospital stays. 5,7 Additionally, preserving the retrosternal part of the pericardium around the right ventricle (RV) during less invasive surgery (LIS; e.g., left anterior thoracotomy combined with right anterior thoracotomy or partial upper sternotomy) may help prevent postoperative right ventricular failure (RVF). Thus, avoiding a full sternotomy in critically ill patients would be beneficial. However, the impact of a full sternotomy versus LIS with a sternotomy-sparing or partial sternotomy approach remains unclear.
Understanding the challenges and the impact on outcomes in a large "real-world" cohort of patients is paramount. To analyze the impact of a surgical approach (full sternotomy vs. LIS) on the outcome in critically ill patients who underwent LVAD implantation on va-ECLS based on data from the largest European registry, we created a model where all patients in the full sternotomy group would be potential candidates for LIS, and assumed that the decision was based solely on the surgeon's preference.

| Patient population
The ECLS-VAD registry is a multicentre retrospective study that gathered data on consecutive patients who underwent implantation of durable mechanical circulatory support (MCS) devices after va-ECLS between January 2010 and August 2018 in eleven high-volume European centres. 8 Patients who underwent durable MCS implantation after va-ECLS support were eligible to participate. At all institutions the main goal after va-ECLS implantation was to wean the patient off mechanical support. Patients who did not meet weaning criteria were considered for durable MCS after an adequate neurological evaluation.
All va-ECLS implantations were performed as emergency procedures; patients with postcardiotomy heart failure were also included in the register. The data collection and retrospective analysis were performed after obtaining approval from the institutional review board of each participating center. Internally validated preoperative, intraoperative, and postoperative data were collected from hospital charts from a total of 531 patients. The Interagency Registry for Mechanically Assisted POTAPOV ET AL. For the presented study patients who received 1. a total artificial heart (n = 19), 2. pulsatile devices (implantable, n = 2 and paracorporeal, n = 5), 3. concomitant intracardiac procedures (valve surgery, closure of septal defects or patent foramen ovale, thrombectomy, aneurysmectomy, and similar, n = 116), and 4. ECLS for postcardiotomy heart failure or patients with central cannulation (n = 69) were excluded from the analysis.
The remaining 363 patients were divided into an FS group (full median sternotomy, n = 324) and a LIS group (patients in whom the apex was approached through left thoracotomy and the ascending aorta through right thoracotomy or through partial sternotomy, n = 39); all of them had been on peripheral va-ECLS and received a durable continues-flow LVAD.
The main endpoint was survival; postoperative bleeding, the need for blood products and surgical re-exploration as well as complications and end organ dysfunction/failure including RVF, stroke, and infections were also analyzed.
Patient follow-up was completed as of June 1st, 2020.

| SURGICAL PROCEDURES
The LVAD implantation, postoperative blood product, and factor administration were performed according to institutional protocols which vary between the participating institutions. LIS was performed using left anterior sternotomy to access the apex combined with upper partial sternotomy or right anterior thoracotomy to access the ascending aorta. 20.1% of the patients had a history of cardiac surgery (years or months before index hospitalization). These patients were included in the analysis. In case of severe RVF during the LVAD implantation, temporary RV support was established using a directly cannulated right ventricular assist device (RVAD) in patients with FS approach, with anastomosis of the 10 mm graft to the pulmonary artery and tunnel to the outside allowing for venous drainage from the femoral vein, and a peripheral va-ECLS in patients with LIS approach. In case of postoperative RVF, a percutaneous RVAD was employed, with one cannula inserted into the pulmonary artery and venous drainage from the right atrium through the femoral vein. According to institutional protocols, anticoagulation was commenced in the ICU as soon as bleeding had subsided (chest tube output and need for blood products). Competing risk analyses were used to evaluate the incidence of first stroke, infection, bleeding, and pump thrombosis with death, heart transplantation, and LVAD weaning as competing outcomes. For these outcomes, subdistribution hazard ratios (SHRs) were calculated using the Fine-Gray model with the propensity score as a covariate. 9 The influence of LIS on survival was estimated in a Cox regression adjusting for the propensity score. E values for the point estimator and the confidence limit nearest to nil were calculated to assess the impact of unmeasured confounding on a risk ratio scale, with high E values indicating a robust treatment-outcome association. 10 We assumed a p value of <0.05 as the threshold for statistical significance. The analysis was explorative in nature. R software, version 3.5.2 was used for statistical analyses.

| Patient characteristics
LIS was performed in 6 of 11 study centers, with the proportion varying between 3% and 46% within the center.
The preoperative baseline data of both groups are presented in

| Postoperative outcomes
Postoperative outcomes are shown in Table 2. The chest tube output during the first 24 h was similar in both groups. The incidence of T A B L E 1 Preoperative characteristics in non-adjusted cohort  Figure 1). Causes of death in both groups are presented in Table 3.  We believe that a combination of more re-exploration due to bleeding, possible excessive procoagulation and a greater use of

| LIMITATIONS
This was a retrospective and non-randomized study. Due to the multicentre nature of the study, surgical techniques and perioperative decisions may differ between centres and may to some degree influence the study results. However, the propensity score adjustment eliminates the majority of such effects. Nevertheless, a coincidental relationship between stroke and mortality in our study cannot be excluded.
It should also be noted that our study cohort might lack the power to show a significant difference in outcome due to the small population in the LIS group. Future randomized, controlled studies are warranted.

| CONCLUSION
The intermediate-term survival is comparable between both groups. A less invasive surgical approach for LVAD implantation directly from va-ECLS is safe and is associated with a lower incidence of re-exploration due to bleeding, fewer blood products administered immediately after surgery and absence of perioperative stroke compared to FS. Future randomized, controlled studies are warranted.