Advertisement

Less-invasive ventricular assist device implantation: A multicenter study

Published:December 22, 2020DOI:https://doi.org/10.1016/j.jtcvs.2020.12.043

      Abstract

      Background

      Left ventricular assist device has been shown to be a safe and effective treatment option for patients with end-stage heart failure. However, there is limited evidence showing the effect of the implantation approach on postoperative morbidities and mortality. We aimed to compare left ventricular assist device implantation using conventional sternotomy versus less-invasive surgery including hemi-sternotomy and the minithoracotomy approach.

      Methods

      Between January 2014 and December 2018, 342 consecutive patients underwent left ventricular assist device implantation at 2 high-volume centers. Patient characteristics were prospectively collected. The propensity score method was used to create 2 groups in a 1:1 fashion. A competing risk regression model was used to evaluate time to death adjusting for competing risk of heart transplantation.

      Results

      The unmatched cohort included 241 patients who underwent left ventricular assist device implantation with the conventional sternotomy technique and 101 patients who underwent left ventricular assist device implantation with the less-invasive surgery technique. Propensity matching produced 2 groups each including 73 patients. In the matched groups, reexploration rate for bleeding was necessary in 17.9% (12/67) in the conventional sternotomy group compared with 4.1% (3/73) the less-invasive surgery group (P = .018). Intensive care unit stay for the less-invasive surgery group was significantly lower than for the sternotomy group (10.5 [interquartile range, 2-25.75] days vs 4 [interquartile range, 2-9.25] days, P = .008), as was hospital length of stay (37 [interquartile range, 27-61] days vs 25.5 [interquartile range, 21-42] days, P = .007). Mortality cumulative incidence for conventional surgery was 24% (95% confidence interval, 14.3-34.8) at 1 year and 26% (95% confidence interval, 15.9-37.4) at 2 years for patients without heart transplantation. Mortality cumulative incidence for less-invasive surgery was 22.5% (95% confidence interval, 12.8-33.8) at 1 year and 25.2% (95% confidence interval, 14.5-37.4) at 2 years for patients without heart transplantation. There was no difference in cumulative mortality incidence when adjusting for competing risk of heart transplantation (subdistribution hazard, 0.904, 95% confidence interval, 0.45-1.80, P = .77).

      Conclusions

      The less-invasive surgery approach is a safe technique for left ventricular assist device implantation. Less-invasive surgery was associated with a significant reduction in the postoperative bleeding complications and duration of hospital stay, with no significant difference in mortality incidence.

      Graphical abstract

      Key Words

      Abbreviations and Acronyms:

      CI (confidence interval), CPB (cardiopulmonary bypass), CS (conventional sternotomy), ECMO (extracorporeal membrane oxygenation), INTERMACS (Interagency Registry for Mechanically Assisted Circulatory Support), IQR (interquartile range), LIS (less-invasive surgery), LVAD (left ventricular assist device), RVAD (right ventricular assist device), SMD (standardized mean difference), VAD (ventricular assist device)
      Figure thumbnail fx2
      Cumulative incidence plot stratified by surgical techniques.
      LVAD implantation using the LIS approach is a safe and effective alternative technique.
      Less-invasive methods for LVAD implantation such as hemi-sternotomy and minithoracotomy are becoming more common. This study compares less-invasive methods with conventional sternotomy to evaluate whether there is benefit for mortality, complications, and hospital stay.
      See Commentaries on pages 1919 and 1920.
      Conventional sternotomy (CS) using a cardiopulmonary bypass (CPB) machine is considered a standard approach for left ventricular assist device (LVAD) implantation. Although the number of referrals of ventricular assist device (VAD) implantation is increasing, the complexity of patients and their perioperative risk factors are surgical challenges. Therefore, in an effort to minimize the surgical trauma at the time of VAD implantation, many centers started using other less-invasive and/or off-pump approaches for VAD implantation.
      • Banner N.R.
      • Bonser R.S.
      • Clark A.L.
      • Clark S.
      • Cowburn P.J.
      • Gardner R.S.
      • et al.
      UK guidelines for referral and assessment of adults for heart transplantation.
      • Saeed D.
      • Sixt S.
      • Albert A.
      • Lichtenberg A.
      Minimally invasive off-pump implantation of HeartMate 3 left ventricular assist device.
      • Sileshi B.
      • Haglund N.A.
      • Davis M.E.
      • Tricarico N.M.
      • Stulak J.M.
      • Khalpey Z.
      • et al.
      In hospital outcomes of a minimally invasive off-pump left thoracotomy approach using a centrifugal continuous-flow LVAD.
      Several advantages of less-invasive surgery (LIS) such as shorter hospital stay have been reported in smaller series.
      • Maltais S.
      • Davis M.E.
      • Haglund N.
      Minimally invasive and alternative approaches for long-term LVAD placement: the Vanderbilt strategy.
      However, published studies in the field lack a representative study showing the effect of the implantation approach on postoperative morbidities and mortality. This study aimed to evaluate VAD implantation using CS versus LIS at 2 high-volume cardiac surgery centers.

      Materials and Methods

      Between January 2014 and December 2018, a total of 342 consecutive patients underwent LVAD implantation for end-stage heart failure at 2 high-volume institutions. The patients at both centers were discussed in detail with the heart failure team. The majority of the surgical cases were performed by 2 surgeons. The decision to proceed with sternotomy or LIS approach (hemi-sternotomy or minithoracotomy approach) was based on the operating surgeon's discretion. Both surgical techniques were used for the duration of the study, with the CS technique used more commonly in earlier years. Both centers have a restrictive approach for tricuspid valve repair in VAD candidates. Indications and perioperative management of patients with end-stage heart failure were consistent throughout the recruitment period and between managing centers.
      Research ethics board approval was obtained from both institutions. Retrospective analysis of prospectively collected data was performed. Patient demographics, preoperative and postoperative risk factors, hemodynamic measurements, perioperative risk factors, and clinical outcomes were analyzed. The Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) definitions of right ventricular failure severity were used. Necessity for right ventricular assist device (RVAD) was indicated in acute severe right ventricular failure. The majority of patients had right heart catheterization performed before and after LVAD implantation. Data of the latest right heart catheterization were documented and analyzed. All patients underwent postoperative echocardiography performed by specialized cardiologists. The latest echocardiography performed after LVAD implantation and before discharge was used for analysis. Postoperative right ventricular functional impairment was assessed as mild, moderate, or severe.

      Operative Techniques

      The LIS approach was sporadically used early in our experience. However, after the implementation of the procedure, the LIS approach is exclusively used unless aortic valve replacement or patent foramen ovale closure was necessary. The patient was positioned with slight elevation of the left chest. The left ventricular apex was first localized using transthoracic echocardiography. A 4- to 6-cm incision and upper L-shaped hemisternotomy (third intercostal space) was performed (Video 1). The majority of patients underwent operation using a CPB machine (MAQUET AG, Rastatt, Germany). Cannulation was performed under transthoracic echocardiography guidance. For arterial cannulation, the ascending aorta was used, and for venous cannulation, the right femoral vein was used. An anterolateral minithoracotomy was performed, the pericardium was opened, pericardial traction sutures were placed, and the sewing ring was placed on the left ventricular apex under transthoracic echocardiography guidance. CPB was initiated. The apex was incised within the sewing ring using an 11 blade. A coring knife was used for appropriate incisions. The device was inserted into the ventricle and fixed. The driveline was tunneled using the C-Technique.
      • Schibilsky D.
      • Benk C.
      • Haller C.
      • Berchtold-Herz M.
      • Siepe M.
      • Beyersdorf F.
      • et al.
      Double tunnel technique for the LVAD driveline: improved management regarding driveline infections.
      The outflow graft was tunneled within the pericardium and anastomosed end-to-side to the ascending aorta after partial clamping of the ascending aorta. De-airing was performed through the outflow graft and ascending aorta. CPB support was reduced, and the pump was turned on. The pump was covered with a polytetrafluoroethylene membrane, and the pericardium was closed or approximated. The sternum and skin were closed in the usual manner. Notably, a few patients underwent off-pump implantation in this series. The decision to proceed with off-pump implantation was made in the operating room on the basis of the patient's hemodynamic stability. The HeartWare (Medtronic, Framingham, Mass) coring knife was used for all off-pump implantations regardless of the implanted pump. Because of the inability to resterilize the HeartWare coring knifes and increasing use of HeartMate III in the recent years, no off-pump implantation was practiced in the later series. Beating heart, rapid pacing, or adenosine was used at the time of coring for off-pump implantations. The same target activated clotting time was used regardless of whether the procedure was performed on or off-pump.
      The postoperative anticoagulation strategy was similar in both centers and included intravenous heparin 24 hours after surgery with target activated partial thromboplastin time of 40-50 seconds. Starting from the third postoperative day, the activated partial thromboplastin time target was 50-60 seconds. Furthermore, every patient received acetylsalicylic acid (100 mg 1-0-0) starting from postoperative day 1. Coumadin was started after removal of the chest tubes and pacing wires.

      Statistical Analysis

      Categorical variables were reported as counts with percentage and analyzed with chi-square or Fisher exact test whenever appropriate. Continuous study variables were assessed for normal distribution using QQ plots and reported as mean with standard deviation if normally distributed or median with interquartile range (IQR) if not normally distributed. Statistical tests were performed according to normality of data with Welch's t test or Wilcoxon signed-rank test whenever appropriate.
      Patients operated using CS approach were compared with patients undergoing LIS approach. Because these 1 groups were unlikely to be similar, a propensity score analysis was computed. Standardized mean differences (SMDs) were calculated for both the unmatched and matched cohorts. For propensity score matching algorithm, we used the nearest neighbor approach using greedy algorithm, caliper size 0.1, without replacement, using a 1:1 ratio, using the R package MatchIt. The following variables were included in the propensity score match based on literature and clinical expertise of the research team: patient age, whether patient was on extracorporeal membrane oxygenation (ECMO) preoperatively, previous cardiac surgery, whether LVAD implantation was performed off-pump, center (Leipzig, Dusseldorf), and model of VAD (HeartWare, HM3). We excluded the HM2 device before performing the matching algorithm because it is not readily used in the thoracotomy approach. Propensity score–matching diagnostics were performed by examining the standardized mean difference (SMD) of covariates in the matched group, comparing the distributional balance of propensity scores between CS and LIS in both unmatched and matched groups using histograms, comparing covariate balance between unmatched and matched groups using a Love plot, and comparing propensity scores of patients using a jitter plot. Diagnostics were performed using the R package cobalt. A sensitivity analysis was performed whereby a multivariable regression model was used to evaluate the effect of the same covariates in the propensity score model including surgical technique using the full unmatched cohort data excluding the HM2 LVAD model.
      In addition, we performed time-to-event analyses to compare survival of patients who underwent CS versus LIS. To adjust for the competing risk of heart transplantation, we performed a competing risk regression. We used the Fine and Grey method
      • Fine J.P.
      • Gray R.J.
      A proportional hazards model for the subdistribution of a competing risk.
      using R package cmprsk. which allows for evaluating the association between surgical technique and the time to death while adjusting for the competing risk of heart transplantation. Model coefficients were exponentiated to obtain the subdistribution hazard rate, which is the hazard of death in the presence of competing events. In addition, we designed a cumulative incidence curve for both CS and LIS groups. Patients with missing data were excluded from statistical analysis. Statistical analysis was performed with R (version 3.6.2).

      Results

      A total of 342 consecutive patients underwent LVAD implantation (HVAD, HMII, and HMIII) due to end-stage heart failure in both institutions. The LIS approach was performed in 101 patients (29.5%). The LIS approach was feasible in all patients, and none of the patients were converted from LIS to CS. The mean age of patients was 59 ± 11 years. The majority of patients were male (307/342 patients, 90%). Follow-up was complete in 100% of the patients with a total LVAD duration support of 420 patient-years. Four LVADs were explanted during the follow-up after recovery of the left ventricular function.
      A total of 85 patients were classified as INTERMACS 1. Venoarterial ECMO was used in 67 patients (19.6%) before VAD implantation. The ECMO support duration was 9 ± 6 days in the CS group versus 5 ± 3 days in the LIS group (P = .02). A comparison of the hemodynamic parameters between both groups showed a pulmonary artery pulsatility index of 2.0 (IQR, 1.38-3.0) in the CS group versus 2.03 (IQR, 1.39-2.95) in the LIS group and a central venous pressure/pulmonary capillary wedge pressure ratio of 0.47 (IQR, 0.32-0.63) in the CS group versus 0.5 (IQR, 0.34-0.64) in the LIS group.
      Patients in the CS group (n = 241) had an INTERMACS mean of 2.35 ± 1.13 versus 2.83 ± 1.0 in the LIS (n = 101) group (P < .001). A total of 55 patients (23%) in the CS group had a history of cardiac surgeries versus 14 patients (14%) in the LIS group (P = .08). Patient characteristics with corresponding SMDs before propensity score matching are summarized in Table E1.
      Propensity matching identified 73 patients in each treatment group (Table 1). There was no significant difference in the number of LIS versus CS LVAD implantations between 2014 and 2018 (P = .172). The 2 groups were comparable by visual inspection using histograms of propensity scores (Figure E1), Love plot depicting covariate balance (Figure E2), and jitter plot showing distribution of propensity scores (Figure E3). We noted that off-pump implantation SMD of 0.237 was due to only 2 patients receiving off-pump surgery in the CS group compared with zero patients in the LIS group.
      Table 1Preoperative patient characteristics in matched cohort with standardized mean differences comparing groups
      CharacteristicCSLISSMD
      No. of patients7373
      Center0.139
       Duesseldorf28 (38.4%)33 (45.2%)
       Leipzig45 (61.6%)40 (54.8%)
      Age (y)57.47 (12.02)57.38 (11.26)0.007
      Gender0.302
       Female12 (16.4%)5 (6.8%)
       Male61 (83.6)68 (93.2)
      Weight (kg)89.48 (19.76, n = 66)86.78 (18.26, n = 69)0.142
      Pump type0.110
       Heartware35 (47.9%)39 (53.4%)
       HM20 (0%)0 (0%)
       HM338 (52.1)34 (46.6)
      Central venous pressure (mm Hg)13.96 (7.43)14.16 (6.81)0.029
      TAPSE15.34 (4.08, n = 53)15.40 (3.98, n = 42)0.016
      RVEDD (mm)43.73 (9.06, n = 37)43.57 (8.87, n = 44)0.018
      Hemoglobin value (mmol/L)30 (6-51)30 (13-52, n = 69)0.013
      Creatinine (μmol/L)121.29 (42.74, n = 56)130.60 (49.60, n = 60)0.201
      PCWP (mm Hg)25.92 (9.81, n = 38)26.90 (7.38, n = 40)0.113
      Cardiac index (L/min/m2)1.83 (0.51, n = 50)1.83 (0.59, n = 55)0.001
      PAP mean (mm Hg)35.08 (8.80, n = 51)34.90 (9.52, n = 58)0.020
      Mean arterial pressure80.26 (14.37, n = 54)79.04 (12.60, n = 49)0.090
      Off pump VAD implantation2 (2.7%)0 (0%)0.237
      Tricuspid valve insufficiency II-III27 (42.9%, n = 63)26 (37.7%, n = 69)0.106
      Ischemic cardiomyopathy etiology37 (51.4%, n = 72)40 (54.8%, n = 73)0.068
      Preoperative venoarterial ECMO10 (13.7%)10 (13.7%)0.0
      History of prior cardiac surgeries10 (13.7%)9 (12.3%)0.041
      INTERMACS mean (SD)2.73 (1.15)2.68 (0.91)0.040
      SMD is reported as absolute values. Continuous variables reported as mean (standard deviation) or median (quartile 1 - quartile 3). Variables with missing data have indicated number of available data for analysis. CS, Conventional sternotomy; LIS, less-invasive surgery; SMD, standardized mean difference; HM2, HeartMate 2; HM3, HeartMate 3; TAPSE, tricuspid annular plane systolic excursion; RVEDD, right ventricular end-diastolic diameter; PCWP, pulmonary-capillary wedge pressure; PAP, pulmonary arterial pressure; VAD, ventricular assist device; ECMO, extracorporeal membrane oxygenation; INTERMACS, Interagency Registry for Mechanically Assisted Circulatory Support.
      For both surgical treatment groups, these laboratory values were obtained at a median (IQR) of 1 day (0-2) after LVAD implantation. CPB time was available only from 1 center: 76.8 (33.1) minutes in the CS group and 71.4 (24.7) minutes in the LIS group. Preoperative heart rate was 86.8 beats/min in the CS group and 85.3 beats/min in the LIS group (SMD, 0.078). Mean age was 57.5 ± 12.0 years in the CS group and 57.4 ± 11.3 years in the LIS group. There were 61 (83.6%) male patients in the CS and 68 male patients (93.2%) in the LIS group. The type of LVADs in the matched groups were as follows: 35 HeartWare HVAD (Medtronic, Framingham, Mass) and 38 HeartMate 3 (Abbott, Abbott, Abbott Park, Ill) in the CS group; 39 HeartWare HVAD and 34 HeartMate 3 in the LIS group.
      The reexploration rate for bleeding was necessary in 17.9% (12/67) of the CS group compared with 4.1% (3/73) of the LIS group (P = .018; Table 2). Intensive care unit stay for the LIS group was significantly lower than in the CS group (10.5 [IQR, 2-25.75] days vs 4 [IQR, 2-9.25] days, P = .008). There was also a significant reduction in hospital length of stay (37 [IQR, 27-61] days vs 25.5 [IQR, 21-42] days, P = .007). Postoperative capillary wedge pressure was lower in the LIS group (14.3 [6.1]) than in the CS group (18.2 [5.95], P = .008). However, no significant difference in postoperative severe acute right ventricle dysfunction requiring RVAD was observed (CS 19.4% vs LIS 9.6%, P = .147). Postoperative incidence of pump thrombosis, driveline infection, and gastrointestinal bleeding at the time of index hospitalization were not significantly different between groups. Mortality cumulative incidence for CS was 24% (95% confidence interval [CI], 14.3-34.8) at 1 year and 26% (95% CI, 15.9-37.4) at 2 years for patients without heart transplantation. For patients who underwent heart transplantation, mortality cumulative incidence for CS was 6.3% (95% CI, 2.0-14.1) at 1 year and 14% (95% CI, 5.8-25.8) at 2 years. Mortality cumulative incidence for LIS was 22.5% (95% CI, 12.8-33.8) at 1 year and 25.2% (95% CI, 14.5-37.4) at 2 years for patients without heart transplantation. For patients who underwent heart transplantation, mortality cumulative incidence for LIS was 7.9% (95% CI, 2.8-16.5) at 1 year and 13.6% (95% CI, 5.6-25.3) at 2 years. There was no significant difference in survival between the groups when adjusting for the competing risk of heart transplantation (subdistribution hazard: 0.904, 95% CI, 0.45-1.80, P = .77). Figure 1 depicts cumulative incidence curves for the outcome mortality in patients without the competing risk heart transplantation. The sensitivity analysis using multivariable regression on the unmatched cohort (excluding HM2 device) adjusting for the same covariates as in propensity match is reported in Table E2.
      Table 2Postoperative outcomes in conventional surgery and less-invasive surgery groups in matched cohort
      OutcomeCSLISP value
      No. of patients7373
      Postoperative BUN (mmol/L)53.62 (25.88, n = 62)56.07 (32.98, n = 69).635
      Postoperative eGFR62.29 (22.94, n = 56)64.00 (30.44, n = 66).724
      Postoperative PCWP (mm Hg)18.23 (5.95, n = 31)14.30 (6.11, n = 40).008
      Postoperative SvO2 (%)66.61 (10.90, n = 54)68.54 (9.52, n = 57).325
      Postoperative cardiac index (L/min/m2)2.98 (0.64, n = 32)3.07 (0.84, n = 44).577
      Reexploration for bleeding12/67 (17.9%)3/73 (4.1%).018
      Postoperative stroke2/69 (2.9%)9/73 (12.3%).074
      Driveline infection7/66 (10.6%)2/73 (2.7%).085
      GI bleeding1/68 (1.5%)6/73 (8.2%).117
      Pump thrombosis2/65 (3.1%)1/73 (1.4%).601
      Any form of postoperative RV failure26 (35.6%)17 (23.3%).102
      Postoperative RV failure severity.140
       047 (64.4%)56 (76.7%)
       12 (2.7%)2 (2.7%)
       21 (1.4%)4 (5.5%)
       38 (11.0%)4 (5.5%)
       415 (20.5%)7 (9.6%)
      RVAD implantation15/73 (20.5%)7/73 (9.6%).147
      TAPSE (mm)12.51 (3.76, n = 37)12.29 (3.42, n = 34).798
      RVEDD (mm)42.58 (9.15, n = 50)41.71 (8.52, n = 55).616
      Right ventricle dysfunction (echocardiography).055
       Mild18/53 (34.0%)15/62 (24.2%)
       Moderate19/53 (35.8%)36/62 (58.1%)
       Severe16/53 (30.2%)11/62 (17.7%)
      ICU length of stay (d)10.5 (2-25.75, n = 70)4 (2-9.25, n = 72).008
      Hospital length of stay (d)37 (27-61)25.5 (21-42, n = 70).007
      Heart transplantation during follow-up9 (12.3%)9 (12.3%)1.0
      30-d mortality5 (6.8%)6 (8.2%)1.0
      1-y mortality.77
      Cumulative incidence regression analysis adjusting for competing risk of heart transplantation.
       without heart transplantation24% (95% CI, 14.3-34.8)22.5% (95% CI, 12.8-33.8)
       with heart transplantation6.3% (95% CI, 2.0-14.1)7.9% (95% CI, 2.8-16.5)
      2-y mortality.77
      Cumulative incidence regression analysis adjusting for competing risk of heart transplantation.
       without heart transplantation26% (95% CI, 15.9-37.4)25.2% (95% CI, 14.5-37.4)
       with heart transplantation14% (95% CI, 5.8-25.8)13.6% (95% CI, 5.6-25.3)
      Variables with missing data have indicated number of available data for analysis, otherwise no missing data. CS, Conventional sternotomy; LIS, less-invasive surgery; BUN, blood urea nitrogen; eGFR, estimated glomerular filtration rate; PCWP, pulmonary capillary wedge pressure; SvO2, mixed venous oxygen saturation; GI, gastrointestinal; RV, right ventricle; RVAD, right ventricular assist device; TAPSE, tricuspid annular plane systolic excursion; RVEDD, right ventricular end-diastolic diameter; ICU, intensive care unit.
      Cumulative incidence regression analysis adjusting for competing risk of heart transplantation.
      Figure thumbnail gr1
      Figure 1Cumulative incidence plot with corresponding 95% CI bands, stratified by surgical techniques (LIS; CS) for outcome of mortality in patients without competing risk of heart transplantation. Below the cumulative incidence plot, a table demonstrates the number of patients at risk at each follow-up time point. When adjusting for the competing risk of heart transplantation, the regression model demonstrates no significant difference in time to mortality between LIS and CS (subdistribution hazard LIS: 0.904, 95% CI, 0.453-1.8, P = .77).

      Discussion

      In this multicenter, propensity-adjusted study comparing VAD implantation with the CS compared with the LIS approach, we illustrated that the LIS approach was associated with a lower reexploration rate for bleeding and shorter duration of hospital stay. In addition, there was no significant difference in cumulative incidence of mortality at final follow-up between these surgical techniques (Figure 2).
      Figure thumbnail gr2
      Figure 2Retrospective analysis of patients with terminal left heart failure who underwent LVAD implantation through LIS or CS. The LIS approach showed shorter hospital length of stay and lower reexploration for bleeding. There was no difference in mortality between LIS and CS.
      Mechanical circulatory support systems are considered an important therapy option for patients with end-stage heart failure. As with many other cardiac surgery procedures, the gold standard approach for LVAD implantation is the CS approach. However, with the downsizing of devices and consistent improvement of minimally invasive surgery outcomes, surgeons have been implementing LVAD implantation via LIS approach with the notion that clinical outcomes are similar or superior to the sternotomy approach.
      • Saeed D.
      • Sixt S.
      • Albert A.
      • Lichtenberg A.
      Minimally invasive off-pump implantation of HeartMate 3 left ventricular assist device.
      ,
      • Gregoric I.D.
      • La Francesca S.
      • Myers T.
      • Cohn W.
      • Loyalka P.
      • Kar B.
      • et al.
      A less invasive approach to axial flow pump insertion.
      • Anyanwu A.C.
      Technique for less invasive implantation of Heartmate II left ventricular assist device without median sternotomy.
      • Schechter M.A.
      • Patel C.B.
      • Blue L.J.
      • Welsby I.
      • Rogers J.G.
      • Schroder J.N.
      • et al.
      Improved early survival with a nonsternotomy approach for continuous-flow left ventricular assist device replacement.
      However, since that time, there have been no randomized trials comparing the full sternotomy approach with the LIS approach. Previous studies comparing LIS with CS were limited because of small sample size, lack of adjustment for key patient characteristics,
      • Kazui T.
      • Lick S.D.
      • Smith R.G.
      • Avery R.J.
      • Suryanarayana P.G.
      • Juneman E.B.
      • et al.
      Minimally invasive off-pump HVAD vs full sternotomy on-pump LVAD placement: a better option?.
      or no comparison of LIS versus CS implantation technique.
      • Sileshi B.
      • Haglund N.A.
      • Davis M.E.
      • Tricarico N.M.
      • Stulak J.M.
      • Khalpey Z.
      • et al.
      In-hospital outcomes of a minimally invasive off-pump left thoracotomy approach using a centrifugal continuous-flow left ventricular assist device.
      Accordingly, this study brings several strengths to the current literature. First, it is to our knowledge the largest study examining the influence of surgical approach on postoperative morbidity and mortality. Second, it includes data from 2 centers with procedures performed by experienced surgeons. Given the learning curve associated with surgical interventions, it is critical for each intervention to be performed by an expert to avoid spurious results secondary to surgeon skill rather than the true effect of the intervention.
      • Devereaux P.J.
      • Bhandari M.
      • Clarke M.
      • Montori V.M.
      • Cook D.J.
      • Yusuf S.
      • et al.
      Need for expertise based randomised controlled trials.
      Furthermore, the patients were matched according to several covariates selected on the basis of previous literature and expert input from the surgical team.
      Given the association with shorter hospital length of stay, LVAD implantation with LIS may be suggested to be cost-effective when compared with implantation with full sternotomy. This is supported by our results favoring LIS for hospital length of stay and reexploration for bleeding, both of which are costly. However, this study was not intended to compare the costs between the groups. Further studies are necessary to evaluate the potential differences of costs. These findings of our study mirror the findings of the LATERAL trial, which showed a significant reduction in length of stay and hospital costs in centrifugal-flow LVAD group implanted using minimally invasive surgery.
      • Mcgee E.
      • Danter M.
      • Strueber M.
      • Mahr C.
      • Mokadam N.A.
      • Wieselthaler G.
      • et al.
      Evaluation of a lateral thoracotomy implant approach for a centrifugal-flow left ventricular assist device: the LATERAL clinical trial.
      We speculate that LVAD implantation using the LIS approach leads to minimal heart displacement and maintenance of pericardial integrity. It is important to note that all patients in this study underwent only limited pericardial opening with the pericardium adapted or closed after VAD implantation. In cases where pericardial closure was challenging because of limited space, the pump was covered and the pericardium was adapted. Complete closure of pericardium over the aorta was feasible in all cases.
      Our results are similar to the results of Gosev and colleagues.
      • Gosev I.
      • Wood K.
      • Ayers B.
      • Barrus B.
      • Knight P.
      • Alexis J.D.
      • et al.
      Implantation of a fully magnetically levitated left ventricular assist device using a sternal-sparing surgical technique.
      They conducted an unmatched single-center study of 105 total patients comparing HeartMate III implantation using LIS versus CS. Their results showed that the LIS approach is associated with a lower incidence of severe right ventricle failure, decreased blood product transfusions, and a shorter hospital stay,
      • Gosev I.
      • Wood K.
      • Ayers B.
      • Barrus B.
      • Knight P.
      • Alexis J.D.
      • et al.
      Implantation of a fully magnetically levitated left ventricular assist device using a sternal-sparing surgical technique.
      which agrees with our multicenter propensity-matched cohort study of 146 patients.
      Similar results were observed in the Lateral trial by Mcgee and colleagues.
      • Mcgee E.
      • Danter M.
      • Strueber M.
      • Mahr C.
      • Mokadam N.A.
      • Wieselthaler G.
      • et al.
      Evaluation of a lateral thoracotomy implant approach for a centrifugal-flow left ventricular assist device: the LATERAL clinical trial.
      This multicenter, prospective, nonrandomized trial evaluated the lateral thoracotomy implantation of HVAD and compared these results with predefined goal from historical sternotomy data. Mean length of initial hospital stay was 18 days and was significantly shorter than the predefined performance goal of 26.1 days. In addition, the overall patient survival was good, and bleeding requiring reoperation was significantly less frequent than that observed in previous studies using the sternotomy approach.
      • Mcgee E.
      • Danter M.
      • Strueber M.
      • Mahr C.
      • Mokadam N.A.
      • Wieselthaler G.
      • et al.
      Evaluation of a lateral thoracotomy implant approach for a centrifugal-flow left ventricular assist device: the LATERAL clinical trial.
      Another important observation in our study was the fact that even after matching the groups, the LIS group was associated with lower reexploration rates for bleeding (4% vs 18%, P = .001). This difference may be explained by limited surgical trauma for patients with LIS. It has been well documented that postoperative bleeding is a significant risk factor for morbidity and mortality after LVAD implantation.
      • Slaughter M.S.
      • Rogers J.G.
      • Milano C.A.
      • Russell S.D.
      • Conte J.V.
      • Feldman D.
      • et al.
      HeartMate II Investigators
      Advanced heart failure treated with continuous-flow left ventricular assist device.
      The difference in favor for LIS also may be due to the LIS approach reducing bleeding events. This is of particular importance for patients with a history of cardiac surgeries. Patients undergoing LVAD implantation using the LIS approach experience the advantages of avoiding right ventricle adhesion preparations. Therefore, we speculate that patients with a history of cardiac surgery obtain a greater advantage when undergoing LVAD implantation using LIS compared with the CS approach.

      Study Limitations

      Despite the advantages demonstrated in LIS, it has some limitations. First, LIS is technically more challenging. Specifically, the measurement of the length of the outflow graft may be more difficult. In addition, concomitant procedures, such as valve repair or replacement, may be challenging in the LIS approach. Accordingly, CS remains the gold standard for LVAD implantation in cases where concomitant procedures are required. Notably, considering the possible need of RVAD implantation in patients with LIS, RVAD implantation can be performed safely using the LIS approach.
      • Maxhera B.
      • Albert A.
      • Westenfeld R.
      • Hoffmann T.
      • Lichtenberg A.
      • Saeed D.
      • et al.
      Minimally invasive right ventricular assist device implantation in a patient with Heartware left ventricular assist device.
      A ProtekDuo cannula (CardiacAssist, Pittsburgh, Pa) can be alternatively used. Some centers tend to use venoarterial ECMO for right ventricle failure in these patients.
      Our study has some limitations. First, the analysis of data was conducted retrospectively. However, the majority of data were collected prospectively, which was inputted into a database with standardized variables to be collected. Additionally, this was not a randomized study and the postoperative morbidities were examined using only index hospitalization instead of long-term follow-up. Third, despite matching our surgery groups using several covariates including INTERMACS, there still may be confounding by indication. For instance, the preoperative covariates center, gender, and creatinine had SMDs exceeding 0.1, which suggests imbalance and may influence outcomes. There is also the potential in for unknown covariates to be unbalanced, as is the case in nonrandomized studies. However, we do note that other markers of cardiac function such as cardiac, moderate to severe tricuspid valve insufficiency, central venous pressure, and INTERMACS profile were distribute similarly between groups. We assume that the operative approach and not the differences in the preoperative condition explains the differences in the outcome. The bias confounding by indication was minimized as both centers were subject to strict and similar indications and quality control. Accordingly, the center-level effects on outcomes are likely less influential than the effect of the implantation techniques themselves. Our match algorithm was also not meant to parisominous; we matched for all variables that our research team agreed to be important for patient prognosis after LVAD implantation. Nevertheless, we caution readers to evaluate the covariate balance between our groups when evaluating our outcomes and conclusions. Fourth, in our cohort, data on concomitant procedures were not collected, and the groups were not adjusted according to the concomitant procedures. Therefore, we cannot comment on the potentially effect of concomitant procedures on the need for blood product transfusion and type of blood product transfused during surgery. As previously mentioned, concomitant procedures included in this study period were limited to aortic valve replacement and patent foramen ovale closure. Notably, previous studies have demonstrated that concomitant procedures may increase morality compared with LVAD implantation only.
      • Veenis J.F.
      • Yalcin Y.C.
      • Brugts J.J.
      • Constantinescu A.A.
      • Manintveld O.C.
      • Bekkers J.A.
      • et al.
      Survival following concomitant aortic valve procedure during left ventricular assist device surgery: an ISHLT Mechanically Assisted Circulatory Support (IMACS) registry analysis.
      ,
      • Robertson J.O.
      • Naftel D.C.
      • Myers S.L.
      • Prasad S.
      • Mertz G.D.
      • Itoh A.
      • et al.
      Concomitant aortic valve procedures in patients undergoing implantation of continuous-flow left ventricular assist devices: an INTERMACS database analysis.
      Fifth, data were missing in some covariates for the matched and unmatched groups, which suggests the possibility that the similarity between matched groups may vary if patients with missing data were different between surgical technique groups. There was also some missing data in outcomes, resulting in denominators less than the total number of patients in each matched group. For the purpose of transparency and comprehensiveness, we reported these outcomes; however, we add caution when interpreting these outcomes. In addition, we note that intensive care unit length of stay, hospital length of stay, and reexploration for bleeding are at risk of bias due to medical decision. The clinical decisions that resulted in a patient experiencing these outcomes may have been influenced by the LVAD implantation technique. Finally, the cohort of patients are highly selected for these centers, which may limit generalizability. Our centers perform a high volume of surgeries for heart failure. Therefore, the generalizability of these results to other surgical centers may be limited to high-volume centers.

      Conclusions

      LIS is a safe alternative technique for LVAD implantation. It was associated with a lower reexploration rate for bleeding, shorter intensive care unit length of stay, and shorter length of hospital stay. There was no significant difference in incidence of mortality between these techniques. Randomized trials are necessary to confirm our findings.

      Conflict of Interest Statement

      The authors reported no conflicts of interest.
      The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest.

      Supplementary Data

      Appendix E1

      Figure thumbnail fx4
      Figure E1Histograms demonstrating distribution of propensity scores between LIS and CS groups. Visual inspection of histograms shows improved balance in matched group. CS, Conventional sternotomy; LIS, less-invasive surgery.
      Figure thumbnail fx5
      Figure E2Love plot demonstrating covariate level of balance in matched and unmatched groups. Matched group shows good covariate balance. ECMO, Extracorporeal membrane oxygenation; INTERMACS, Interagency Registry for Mechanically Assisted Circulatory Support; VAD, ventricular assist device; HM3, HeartMate 3.
      Figure thumbnail fx6
      Figure E3Jitter plot showing distribution of propensity scores between unmatched treatment patients, matched treatment patients, matched control patients, and unmatched control patients. Unmatched cohort demonstrates noticeable imbalance in propensity score, and matched cohort demonstrates good balance in propensity score.
      Table E1Preoperative patient characteristics in unmatched cohort
      CharacteristicsCSLISSMDP value
      No. of patients241101
      Center0.432<.001
       Dusseldorf7452
       Leipzig16749
      Age (y)58.5 (10.9, n = 240)58.8 (11.0)0.028.816
      Gender0.0161.0
       Female2510
       Male21691
      Weight (kg)87.7 (18.6, n = 223)85.7 (17.9, n = 97)0.106.380
      Pump type0.403.009
       HeartWare9654
       HM2191
       HM312646
      Central venous pressure (mm Hg)13.2 (6.5, n = 178)13.6 (6.9, n = 80)0.047.747
      TAPSE14.8 (4.2, n = 167)15.4 (3.9, n = 54)0.143.354
      RVEDD (mm)44.9 (9.6, n = 115)42.5 (8.5, n = 59)0.258.102
      Hemoglobin value (mmol/L)29 (3-58, n = 237)31 (18-51, n = 97)0.05.298
      Creatinine (μmol/L)124.3 (49.3, n = 186)130.1 (53.8, n = 87)0.113.393
      PCWP (mm Hg)26.5 (8.8, n = 120)25.9 (8.1, n = 57)0.067.674
      Cardiac index (L/min/m2)1.89 (0.55, n = 151)1.82 (0.59, n = 82)0.132.344
      PAP mean (mm Hg)35.3 (10.1, n = 153)34.9 (9.5, n = 85)0.033.808
      Mean arterial pressure79.7 (13.7, n = 170)78.9 (12.3, n = 68)0.059.676
      Off-pump VAD implantation2140.516<.001
      Tricuspid valve insufficiency II-III78 (35.8%, n = 218)39 (41.1%, n = 95)0.109.448
      Ischemic cardiomyopathy etiology108 (45.2%, n = 239)540.166.201
      Preoperative venoarterial ECMO57100.374.006
      History of cardiac surgeries55140.233.08
      INTERMACS mean (SD)2.3 (1.1, n = 237)2.8 (1.0)0.475<.0001
      SMD is reported as absolute values. Continuous variables reported as mean (standard deviation) or median (quartile 1 - quartile 3). Variables with missing data have indicated number of available data for analysis. CS, Conventional sternotomy; LIS, less-invasive surgery; SMD, standardized mean difference; HM2, HeartMate 2; HM3, HeartMate 3; TAPSE, tricuspid annular plane systolic excursion; RVEDD, right ventricular end-diastolic diameter; PCWP, pulmonary-capillary wedge pressure; PAP, pulmonary arterial pressure; VAD, ventricular assist device; ECMO, extracorporeal membrane oxygenation; INTERMACS, Interagency Registry for Mechanically Assisted Circulatory Support.
      Table E2Sensitivity analysis using multivariable regression on unmatched cohort (excluding HM2 device) adjusting for the same covariates as in propensity match: Patient age, whether patient was on ECMO preoperatively, previous cardiac surgery, whether LVAD implantation was performed off-pump, center (Leipzig, Dusseldorf), and model of VAD (HeartWare, HM3)
      OutcomeCSLISCoefficient estimate if linear regressionOR if logistic regression for LIS vs CS (reference)P value
      No. of patients222100
      Postoperative BUN (mmol/L)60.11 (28.15)54.91 (30.90)−1.6685.68
      Postoperative eGFR61.16 (28.82)61.21 (28.64)2.2826.56
      Postoperative PCWP (mm Hg)16.41 (6.01)14.98 (6.32)−3.54792.0006
      Postoperative SvO2 (%)66.87 (10.55)69.42 (9.24)0.51932.73
      Postoperative cardiac index (L/min/m2)2.92 (0.72)3.03 (0.84)0.030477.83
      Reexploration for bleeding4940.11.0003
      Postoperative stroke10124.81.003
      Driveline infection1770.40.13
      GI bleeding1171.49.53
      Pump thrombosis41Model did not converge because of low event rate
      Any form of postoperative RV failure105240.44.005
      RVAD implantation55100.30.004
      TAPSE (mm)11.76 (3.41)12.75 (3.68)0.39.53
      RVEDD (mm)0.89.45
      Any right ventricle dysfunction (echo)126580.76.43
      ICU length of stay (d)10 (3.5-31)5 (3-9)−13.45<.0001
      Hospital length of stay (d)40 (24.5-65.5)27 (22-42)−14.49.006
      Heart transplantation during follow-up31150.63.26
      30-d mortality2070.86.76
      1-y mortality57160.49.06
      2-y mortality62190.47.09
      CS, Conventional sternotomy; LIS, less-invasive surgery; OR, odds ratio; BUN, blood urea nitrogen; eGFR, estimated glomerular filtration rate; PCWP, pulmonary-capillary wedge pressure; SvO2, mixed venous oxygen saturation; GI, gastrointestinal; RV, right ventricle; RVAD, right ventricular assist device; TAPSE, tricuspid annular plane systolic excursion; RVEDD, right ventricular end-diastolic diameter; ICU, intensive care unit.

      References

        • Banner N.R.
        • Bonser R.S.
        • Clark A.L.
        • Clark S.
        • Cowburn P.J.
        • Gardner R.S.
        • et al.
        UK guidelines for referral and assessment of adults for heart transplantation.
        Heart. 2011; 97: 1520-1527
        • Saeed D.
        • Sixt S.
        • Albert A.
        • Lichtenberg A.
        Minimally invasive off-pump implantation of HeartMate 3 left ventricular assist device.
        J Thorac Cardiovasc Surg. 2016; 152: 1446-1447
        • Sileshi B.
        • Haglund N.A.
        • Davis M.E.
        • Tricarico N.M.
        • Stulak J.M.
        • Khalpey Z.
        • et al.
        In hospital outcomes of a minimally invasive off-pump left thoracotomy approach using a centrifugal continuous-flow LVAD.
        J Heart Lung Transplant. 2015; 34: 107-112
        • Maltais S.
        • Davis M.E.
        • Haglund N.
        Minimally invasive and alternative approaches for long-term LVAD placement: the Vanderbilt strategy.
        Ann Cardiothorac Surg. 2014; 3: 563-569
        • Schibilsky D.
        • Benk C.
        • Haller C.
        • Berchtold-Herz M.
        • Siepe M.
        • Beyersdorf F.
        • et al.
        Double tunnel technique for the LVAD driveline: improved management regarding driveline infections.
        J Artif Organs. 2012; 15: 44-48
        • Fine J.P.
        • Gray R.J.
        A proportional hazards model for the subdistribution of a competing risk.
        J Am Stat Assoc. 1999; 94: 496-509
        • Gregoric I.D.
        • La Francesca S.
        • Myers T.
        • Cohn W.
        • Loyalka P.
        • Kar B.
        • et al.
        A less invasive approach to axial flow pump insertion.
        J Heart Lung Transplant. 2008; 27: 423-426
        • Anyanwu A.C.
        Technique for less invasive implantation of Heartmate II left ventricular assist device without median sternotomy.
        Semin Thorac Cardiovasc Surg. 2011; 23: 241-244
        • Schechter M.A.
        • Patel C.B.
        • Blue L.J.
        • Welsby I.
        • Rogers J.G.
        • Schroder J.N.
        • et al.
        Improved early survival with a nonsternotomy approach for continuous-flow left ventricular assist device replacement.
        Ann Thorac Surg. 2015; 99: 561-566
        • Kazui T.
        • Lick S.D.
        • Smith R.G.
        • Avery R.J.
        • Suryanarayana P.G.
        • Juneman E.B.
        • et al.
        Minimally invasive off-pump HVAD vs full sternotomy on-pump LVAD placement: a better option?.
        J Heart Lung Transplant. 2017; 36: S442
        • Sileshi B.
        • Haglund N.A.
        • Davis M.E.
        • Tricarico N.M.
        • Stulak J.M.
        • Khalpey Z.
        • et al.
        In-hospital outcomes of a minimally invasive off-pump left thoracotomy approach using a centrifugal continuous-flow left ventricular assist device.
        J Heart Lung Transplant. 2015; 34: 107-112
        • Devereaux P.J.
        • Bhandari M.
        • Clarke M.
        • Montori V.M.
        • Cook D.J.
        • Yusuf S.
        • et al.
        Need for expertise based randomised controlled trials.
        BMJ. 2005; 330: 88
        • Mcgee E.
        • Danter M.
        • Strueber M.
        • Mahr C.
        • Mokadam N.A.
        • Wieselthaler G.
        • et al.
        Evaluation of a lateral thoracotomy implant approach for a centrifugal-flow left ventricular assist device: the LATERAL clinical trial.
        J Heart Lung Transplant. 2019; 38: 344-351
        • Gosev I.
        • Wood K.
        • Ayers B.
        • Barrus B.
        • Knight P.
        • Alexis J.D.
        • et al.
        Implantation of a fully magnetically levitated left ventricular assist device using a sternal-sparing surgical technique.
        J Heart Lung Transplant. 2020; 39: 37-44
        • Slaughter M.S.
        • Rogers J.G.
        • Milano C.A.
        • Russell S.D.
        • Conte J.V.
        • Feldman D.
        • et al.
        • HeartMate II Investigators
        Advanced heart failure treated with continuous-flow left ventricular assist device.
        N Engl J Med. 2009; 361: 2241-2251
        • Maxhera B.
        • Albert A.
        • Westenfeld R.
        • Hoffmann T.
        • Lichtenberg A.
        • Saeed D.
        • et al.
        Minimally invasive right ventricular assist device implantation in a patient with Heartware left ventricular assist device.
        ASAIO J. 2015; 61: e42-e43
        • Veenis J.F.
        • Yalcin Y.C.
        • Brugts J.J.
        • Constantinescu A.A.
        • Manintveld O.C.
        • Bekkers J.A.
        • et al.
        Survival following concomitant aortic valve procedure during left ventricular assist device surgery: an ISHLT Mechanically Assisted Circulatory Support (IMACS) registry analysis.
        Eur J Heart Fail. 2020; 22: 1878-1887
        • Robertson J.O.
        • Naftel D.C.
        • Myers S.L.
        • Prasad S.
        • Mertz G.D.
        • Itoh A.
        • et al.
        Concomitant aortic valve procedures in patients undergoing implantation of continuous-flow left ventricular assist devices: an INTERMACS database analysis.
        J Heart Lung Transplant. 2015; 34: 797-805

      Linked Article

      • Commentary: Less invasive, more incisions, more better?
        The Journal of Thoracic and Cardiovascular SurgeryVol. 164Issue 6
        • Preview
          Left ventricular assists devices (LVADs) offer a life-saving therapy for patients with both acute and chronic heart failure. Although the implantation technique is not complex, the management of these patients' condition remains challenging. Ultimately, the goals are improving quality and quantity of life. Appropriately, the field is focused on improving outcomes and this centers on improving device durability, developing algorithms to optimize device function, and eliminating the noxious problem of right ventricular (RV) failure.
        • Full-Text
        • PDF
      • Commentary: Sometimes less is more: The minimally invasive argument for ventricular assist devices
        The Journal of Thoracic and Cardiovascular SurgeryVol. 164Issue 6
        • Preview
          Minimally invasive approaches to implanting left ventricular assist devices (LVADs) may potentially improve outcomes in patients requiring mechanical circulatory support for advanced heart failure, particularly for patients with extensive comorbidities.1 Although conventional sternotomy continues to remain the gold standard, LVAD implantation via combined hemi-sternotomy and mini-thoracotomy has been shown to be safe and feasible, enabled by miniaturized devices such as the HeartMate 3 (Abbott, Abbott Park, Ill) and the HeartWare HVAD (Medtronic, Framingham, Mass).
        • Full-Text
        • PDF