Advertisement

Pulmonary function tests do not predict mortality in patients undergoing continuous-flow left ventricular assist device implantation

Open ArchivePublished:April 10, 2017DOI:https://doi.org/10.1016/j.jtcvs.2017.02.069

      Abstract

      Objectives

      To investigate the effect of pulmonary function testing on outcomes after continuous flow left ventricular assist device implantation.

      Methods

      A total of 263 and 239 patients, respectively, had tests of forced expiratory volume in 1 second and diffusing capacity of the lungs for carbon monoxide preoperatively for left ventricular assist device implantations between July 2005 and September 2015. Kaplan-Meier analysis and multivariable Cox regressions were performed to evaluate mortality. Patients were analyzed in a single cohort and across 5 groups. Postoperative intensive care unit and hospital lengths of stay were evaluated with negative binomial regressions.

      Results

      There is no association of forced expiratory volume in 1 second and diffusing capacity of the lungs for carbon monoxide with survival and no difference in mortality at 1 and 3 years between the groups (log rank P = .841 and .713, respectively). Greater values in either parameter were associated with decreased hospital lengths of stay. Only diffusing capacity of the lungs for carbon monoxide was associated with increased intensive care unit length of stay in the group analysis (P = .001). Ventilator times, postoperative pneumonia, reintubation, and tracheostomy rates were similar across the groups.

      Conclusions

      Forced expiratory volume in 1 second and diffusing capacity of the lungs for carbon monoxide are not associated with operative or long-term mortality in patients undergoing continuous flow left ventricular assist device implantation. These findings suggest that these abnormal pulmonary function tests alone should not preclude mechanical circulatory support candidacy.

      Key Words

      Abbreviations and Acronyms:

      CI (confidence interval), DLCO (diffusion capacity of the lungs for carbon monoxide), FEV1 (forced expiratory volume in 1 second), ICU (intensive care unit), INTERMACS (Interagency Registry For Mechanically Assisted Circulatory Support), LOS (length of stay), LVAD (left ventricular assist device), MCS (mechanical circulatory support), PFT (pulmonary function test)
      Figure thumbnail fx1
      Kaplan-Meier survival estimates for patients by FEV1 group.
      Abnormal pulmonary function tests alone should not preclude mechanical circulatory support candidacy.
      Many centers consider abnormal pulmonary function tests a barrier to mechanical circulatory support candidacy, but there is little research on their prognostic implications. In this retrospective series, there is no association of abnormal tests with increased mortality after left ventricular assist device implantation. Pulmonary function tests alone should not preclude mechanical circulatory support.
      See Editorial Commentary page 1971.
      See Editorial page 1957.
      The prevalence of heart failure is increasing steadily in the United States.
      • Go A.S.
      • Mozaffarian D.
      • Roger V.L.
      • Benjamin E.J.
      • Berry J.D.
      • Borden W.B.
      • et al.
      Heart disease and stroke statistics—2013 update: a report from the American Heart Association.
      Mechanical circulatory support (MCS) is a proven therapy in advanced heart failure and is shown to improve survival in selected patients.
      • Yancy C.W.
      • Jessup M.
      • Bozkurt B.
      • Butler J.
      • Casey D.E.
      • Drazner M.H.
      • et al.
      2013 ACCF/AHA Guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association Task Force on practice guidelines.
      The presence of lung disease is associated with increased mortality in the Society of Thoracic Surgeons risk calculator.
      • Shih T.
      • Paone G.
      • Theurer P.F.
      • McDonald D.
      • Shahian D.M.
      • Prager R.L.
      The Society of Thoracic Surgeons adult cardiac surgery database version 2.73: more is better.
      However, in heart failure surgery, studies of pulmonary function tests (PFTs) have found both a negative correlation with mortality and postoperative pulmonary complications as well as an absence of any relationship between the tests and outcomes.
      • Iversen K.K.
      • Kjaergaard J.
      • Akkan D.
      • Kober L.
      • Torp-Pedersen C.
      • Hassager C.
      • et al.
      The prognostic importance of lung function in patients admitted with heart failure.
      • Olson T.
      • Denzer D.
      • Sinnett W.
      • Wilson T.
      • Johnson B.
      Prognostic value of resting pulmonary function in heart failure.
      • de Oliveira E.K.
      • da Silva V.Z.M.
      • Turquetto A.L.R.
      Relationship of postoperative walk test and lung function with the length of heart surgery hospital stay.
      • Henn M.C.
      • Zajarias A.
      • Lindman B.R.
      • Greenberg J.W.
      • Melby S.J.
      • Quader N.
      • et al.
      Preoperative pulmonary function tests predict mortality after surgical or transcatheter aortic valve replacement.
      • Zibrak J.D.
      • O’Donnell C.R.
      • Marton K.
      Indications for pulmonary function testing.
      • Hawkins N.M.
      • Petrie M.C.
      • Jhund P.S.
      • Chalmers G.W.
      • Dunn F.G.
      • McMurray J.J.V.
      Heart failure and chronic obstructive pulmonary disease: diagnostic pitfalls and epidemiology.
      • Ivanov A.
      • Yossef J.
      • Tailon J.
      • Worku B.M.
      • Gulkarov I.
      • Tortolani A.J.
      • et al.
      Do pulmonary function tests improve risk stratification before cardiothoracic surgery?.
      • Cohen A.
      • Katz M.
      • Katz R.
      • Hauptman E.
      • Schachner A.
      Chronic obstructive pulmonary disease in patients undergoing coronary artery bypass grafting.
      • Bevelaqua F.
      • Garritan S.
      • Haas F.
      • Salazar-Schicchi J.
      • Axen K.
      • Reggiani J.-L.
      Complications after cardiac operations in patients with severe pulmonary impairment.
      This may be distinct from noncardiac thoracic surgery, in which preoperative pulmonary function measures are correlated unequivocally with postoperative morbidity and mortality.
      • Berry M.F.
      • Jeffrey Yang C.-F.
      • Hartwig M.G.
      • Tong B.C.
      • Harpole D.H.
      • D’Amico T.A.
      • et al.
      Impact of pulmonary function measurements on long-term survival after lobectomy for stage i non-small cell lung cancer.
      Many heart failure centers evaluate PFTs when considering patients for MCS.
      This study investigates the efficacy of percent predicted forced expiratory volume in 1 second (FEV1) and diffusion capacity of the lungs for carbon monoxide (DLCO) in predicting mortality in patients with heart failure who undergo implantation of a continuous-flow left ventricular assist device (LVAD). Secondary measures such as intensive care unit (ICU) and total hospital lengths of stay (LOS) are also explored.

      Methods

       Patients and Methods

      Institutional review board approval was obtained to retrospectively review the MCS quality improvement database at a single institution. A total of 310 sequential patients who underwent continuous-flow LVAD implantations for all indications between July 2005 and September 2015 were evaluated. All patients with recorded data for FEV1 and DLCO were included in our primary analysis. All patients with DLCO testing also had FEV1 testing, but the reverse was not true. Patients with no PFT testing were excluded from initial analyses, leaving 263 (85%) and 239 (77%) patients, respectively, for further analysis. Patients were then stratified into groups based on the respective PFTs. A cutoff value of 40% was chosen for the first group, which is a more stringent criterion for pulmonary dysfunction than previously used in the literature.
      • Shih T.
      • Paone G.
      • Theurer P.F.
      • McDonald D.
      • Shahian D.M.
      • Prager R.L.
      The Society of Thoracic Surgeons adult cardiac surgery database version 2.73: more is better.
      • Iversen K.K.
      • Kjaergaard J.
      • Akkan D.
      • Kober L.
      • Torp-Pedersen C.
      • Hassager C.
      • et al.
      The prognostic importance of lung function in patients admitted with heart failure.
      • Ivanov A.
      • Yossef J.
      • Tailon J.
      • Worku B.M.
      • Gulkarov I.
      • Tortolani A.J.
      • et al.
      Do pulmonary function tests improve risk stratification before cardiothoracic surgery?.
      The remaining patients were then stratified accounting for even distribution of patients to allow statistical comparison. For FEV1, the groups were defined as follows: Group 1: FEV1 <40% predicted, Group 2: FEV1 40% to 55% predicted, Group 3: FEV1 56% to 67% predicted, Group 4: FEV1 68% to 78% predicted, and Group 5: FEV1 >78% predicted. DLCO groups were defined as Group 1: DLCO <40% predicted, Group 2: DLCO 40% to 46% predicted, Group 3: DLCO 47% to 53% predicted, Group 4: DLCO 54% to 61% predicted, and Group 5: DLCO >61% predicted.
      Demographic, preoperative, operative, and postoperative variables such as serum creatinine, hemodynamic profile and Interagency Registry For Mechanically Assisted Circulatory Support (INTERMACS) profile, cardiopulmonary bypass time, reoperation status, additional cardiac surgical procedures at the time of implantation, ventilator time, postoperative pneumonia, reintubation status, tracheostomy, postoperative ICU and total hospital LOS, and mortality at 30 days, 1 year, and 3 years were collected.

       Statistical Analysis

      Univariate and multivariable analyses were performed on outcome variables, with adjustment for factors such as age, sex, INTERMACS profile, serum creatinine, reoperation status, ventilator time, postoperative pneumonia, and cardiopulmonary bypass time, which are known risk factors for poor outcomes after cardiac surgery. First, the data were analyzed with FEV1 and DLCO treated as continuous variables. Comparisons were then made across groups using the group with highest observed values as a reference.
      Overall survival analysis was conducted with the Kaplan-Meier method. Multivariable regressions were performed with the Cox proportional hazard model, which was applied to the entire cohort with the categories of FEV1 and DLCO entered into the model. The proportional-hazard assumption was tested, on basis of Schoenfeld residuals from separate Cox proportional hazard models, treating the variables as continuous and then as categorical and was not violated (P = FEV1 and DLCO .413 and .152, and .222 and .090, respectively, for continuous and categorical variables). LOS analysis was performed with negative binomial regressions to accommodate overdispersion of this outcome variable. For time-to-event analyses, patients were followed from implant until death without censoring for LVAD exchange, explant, or heart transplantation. Sensitivity analyses were performed by including patients with no PFT data in the lowest groups. Furthermore, FEV1 and DLCO were included as linear splines to detect possible nonlinear relationship which was absent (Table E1). All statistical analyses were performed with Stata, version 14 (Stata Corporation, College Station, Tex).

      Results

       Baseline Characteristics

      The baseline characteristics are reported in Table 1. The average patient presented for implantation as INTERMACS 3. The average age of patients in the FEV1 cohort was 54 years; 81.7% were male, and 60.5% of these patients were eligible for transplantation, three quarters of whom underwent eventual heart transplantation. The average time from LVAD implantation to heart transplantation was 282 days. There was no difference in rates of heart transplantation across any of the 5 FEV1 or DLCO groups (average 45.6%, P = .56, and 45.2%, P = .99, respectively). FEV1 ranged from 23% to 113% predicted, whereas DLCO ranged from 28% to 101% predicted. Heartmate II (St Jude, St. Paul, Minn) implantations comprised 79% of procedures in the FEV1 group, whereas the remainder underwent HeartWare HVAD (HeartWare, Framingham, Mass) implantation. There was a similar distribution for the DLCO analysis.
      Table 1Demographics and baseline characteristics
      Characteristic12345TotalP value
      FEV1 group
       n1357646762263
       Age, mean (SD)49.4 (14.2)54.9 (13.8)55.2 (12.1)55.8 (13.5)51.9 (14.3)54.2 (13.5).3
       Sex, male10 (77%)46 (81%)57 (89%)55 (82%)47 (76%)215 (81.7%).4
       INTERMACS profile, mean (SD)2.3 (0.5)2.9 (1.2)3.0 (1.1)3.4 (1.3)3.1 (1.0)3.1 (1.2).0124
       Device
      HeartWare3 (23%)10 (18%)14 (22%)12 (18%)16 (26%)55 (20.9%).79
      HeartMate II10 (77%)47 (82%)50 (78%)55 (82%)46 (74%)208 (79.1%)
       Intent of therapy
      Bridge to transplantation1 (8%)21 (33%)32 (50%)33 (53%)32 (53%)119 (45.2%).024
      Bridge to eligibility5 (38%)11 (17%)11 (17%)7 (11%)8 (13%)42 (16.0%)
      Destination therapy7 (54%)32 (50%)21 (33%)22 (35%)20 (33%)102 (38.8%)
      DLCO group
       n4155514646239
       Age, mean (SD)56.6 (14.0)54.7 (13.5)54.8 (11.2)54.6 (13.9)51.3 (14.4)54.4 (13.4).45
       Sex, male38 (93%)46 (84%)41 (80%)37 (80%)34 (74%)196 (82.0%).24
       INTERMACS profile, mean (SD)2.9 (1.3)2.9 (1.1)3.2 (1.1)3.2 (1.3)3.1 (1.0)3.1 (1.2).57
       Device
      Heartware8 (20%)7 (13%)9 (18%)11 (24%)10 (22%)45 (18.8%).66
      Heartmate II33 (80%)48 (87%)42 (82%)35 (76%)36 (78%)194 (81.2%)
       Intent of therapy
      Bridge to transplantation12 (29%)26 (47%)20 (39%)20 (43%)24 (52%)119 (45.2%).38
      Bridge to eligibility6 (15%)6 (11%)11 (22%)9 (20%)6 (13%)42 (16.0%)
      Destination therapy23 (56%)23 (42%)20 (39%)17 (37%)16 (35%)102 (38.8%)
      FEV1, Forced expiratory volume in 1 second; SD, standard deviation; INTERMACS, Interagency Registry For Mechanically Assisted Circulatory Support; DLCO, diffusion capacity of the lungs for carbon monoxide.

       Hemodynamic Profile

      The average patient had a left ventricular ejection fraction of 19.7%, a cardiac index of 2.3 L/min/m2, and a mean pulmonary arterial pressure of 32 mm Hg with a pulmonary capillary wedge pressure of 22 mm Hg. A total of 17.3% were supported with intra-aortic balloon pumps (Table 2). The average serum creatinine level was 1.3 mg/dL. On group level analysis, the only significant difference between groups was found in the INTERMACS profile at presentation (2.3, standard deviation [SD], 0.5, in the lowest FEV1 group vs 3.1 [SD 1.0] in the highest group, P = .01). In the DLCO groups, a greater percentage of patients with lower test values were supported with an intra-aortic balloon pump at implantation (34% in the lowest DLCO group vs 7% in the highest DLCO group, P value .01), and right-sided fill pressures tended to be greater (12.2 mm Hg [SD 5.3] in the lowest FEV1 group vs 7.8 [SD 4.8] mm Hg in the highest FEV1 group, P < .001 for right atrial pressures), where as the trend was towards a greater wedge pressure in groups with lower FEV1 (24.3 mm Hg in the lowest FEV1 group vs 20.3 mm Hg in the highest FEV1 group, P = .057). All other measures were similar across the groups.
      Table 2Hemodynamic profile and laboratory values
      Characteristic12345TotalP value
      FEV1 group
       n1357646762263
       Creatinine, mg/dL1.3 (0.4)1.3 (0.4)1.4 (0.4)1.2 (0.4)1.2 (0.4)1.3 (0.4).14
       Intra-aortic balloon pump, n (%)6 (55)17 (31)8 (14)7 (11)5 (8)43 (17.3)<.001
       Right atrial pressure, mm Hg12.2 (5.3)11.3 (5.3)11.6 (6.7)8.6 (4.2)7.8 (4.8)9.9 (5.5)<.001
       Right ventricular systolic pressure, mm Hg49.5 (9.6)47.9 (12.0)49.7 (13.2)46.9 (14.5)44.1 (15.9)47.3 (13.9).25
       Right ventricular diastolic pressure, mm Hg13.3 (5.2)11.6 (5.6)11.2 (6.3)9.3 (5.6)8.7 (5.4)10.3 (5.8).01
       Mean pulmonary arterial pressure, mm Hg36.4 (6.4)33.1 (7.3)34.0 (8.4)30.8 (9.8)29.9 (10.0)32.1 (9.0).03
       Pulmonary capillary wedge pressure, mm Hg24.3 (5.8)24.1 (7.4)22.7 (7.2)21.7 (8.8)20.0 (7.9)22.2 (7.9).06
       Cardiac output by thermodilution, L/min4.5 (1.0)4.7 (1.3)5.1 (3.1)4.6 (1.4)4.5 (1.2)4.7 (1.9).48
       Cardiac index by thermodilution, L/min/m22.4 (0.4)2.2 (0.6)2.3 (0.7)2.3 (0.7)2.3 (0.6)2.3 (0.6).94
       Left ventricular ejection fraction, %18.9 (9.4)19.0 (7.3)19.9 (7.1)19.5 (6.3)20.6 (6.3)19.7 (6.9).74
      DLCO group
       n4155514646239
       Creatinine, mg/dL1.3 (0.4)1.4 (0.5)1.3 (0.4)1.2 (0.3)1.3 (0.4)1.3 (0.4).12
       Intra-aortic balloon pump, n (%)13 (34)11 (20)8 (17)5 (11)3 (7)40 (17.5).01
       Right atrial pressure, mm Hg9.3 (6.0)10.3 (5.1)10.9 (6.3)9.4 (5.2)10.0 (5.1)10.0 (5.5).67
       Right ventricular systolic pressure, mm Hg46.9 (12.5)47.8 (14.4)50.3 (13.4)46.4 (14.8)46.6 (13.7)47.6 (13.8).65
       Right ventricular diastolic pressure, mm Hg9.9 (6.2)10.5 (5.4)10.3 (5.7)10.5 (5.9)10.8 (6.1)10.4 (5.8).97
       Mean pulmonary arterial pressure, mm Hg31.5 (8.2)32.7 (9.2)33.4 (8.2)31.5 (10.3)32.0 (8.3)32.3 (8.9).83
       Pulmonary capillary wedge pressure, mm Hg21.1 (8.2)22.5 (7.0)22.7 (8.0)22.2 (9.0)23.5 (7.0)22.4 (7.8).71
       Cardiac output by thermodilution, L/min4.8 (1.3)5.0 (3.2)4.7 (1.1)4.5 (1.4)4.8 (1.5)4.8 (1.9).75
       Cardiac index by thermodilution, L/min/m22.4 (0.6)2.3 (0.7)2.3 (0.6)2.2 (0.6)2.3 (0.7)2.3 (0.6).52
       Left ventricular ejection fraction, %21.3 (8.5)19.7 (7.8)19.1 (6.8)19.0 (5.3)20.7 (6.1)19.9 (7.0).48
      FEV1, Forced expiratory volume in 1 second; DLCO, diffusion capacity of the lungs for carbon monoxide.

       Intraoperative Details

      A total of 19.8% of patients had undergone a previous sternotomy for cardiac surgery before LVAD implantation (Table 3). A total of 31.2% of patients had additional procedures at the time of surgery, including tricuspid valve repair (25.1%) and aortic valve repair. A total of 2.7% of patients required implantation of a right ventricular assist device at the time of surgery. The average cardiopulmonary bypass time was 90.8 minutes.
      Table 3Perioperative details
      Characteristic12345TotalP value
      FEV1 group
       n1357646762263
       Intraoperative
      Reoperation2 (15%)12 (21%)12 (19%)20 (30%)6 (10%)52 (19.8%).074
      Tricuspid valve5 (38%)16 (28%)22 (34%)11 (16%)12 (19%)66 (25.1%).084
      Aortic valve0 (0%)2 (4%)3 (5%)2 (3%)2 (3%)9 (3.4%).94
      Right ventricular assist device1 (8%)0 (0%)1 (2%)3 (5%)2 (3%)7 (2.7%).4
      Cardiopulmonary bypass time, mean (SD)110.5 (42.3)94.3 (34.7)94.3 (41.4)86.6 (52.6)84.5 (44.6)90.8 (44.2).27
       Postoperative
      Ventilator time in hours, median (IQR)24 (17-51)25 (20-45)24 (19-40)24 (22-41)23 (17-38)24 (19-43).47
      Reintubation1 (8%)6 (9%)5 (8%)11 (18%)5 (8%)28 (10.6%).36
      Pneumonia3 (23%)3 (5%)5 (8%)2 (3%)5 (8%)18 (6.8%).12
      Tracheostomy0 (0%)3 (5%)2 (3%)4 (6%)1 (2%)10 (3.8%).61
      DLCO group
       n4155514646239
       Intraoperative
      Reoperation15 (37%)16 (29%)7 (14%)7 (15%)5 (11%)50 (20.9%).009
      Tricuspid valve11 (27%)11 (20%)14 (27%)14 (30%)14 (30%)64 (26.8%).75
      Aortic valve1 (2%)1 (2%)1 (2%)2 (4%)4 (9%)9 (3.8%).36
      Right ventricular assist device1 (2%)1 (2%)1 (2%)0 (0%)4 (9%)7 (3.0%).12
      Cardiopulmonary bypass time, mean (SD)97.2 (43.1)97.0 (44.7)88.1 (37.0)87.8 (35.1)96.4 (60.2)93.3 (44.6).68
       Postoperative
      Ventilator time in hours, median (IQR)24 (20-37)25 (20-49)27 (22-44)24 (17-43)24 (20-32)25 (20-43).5
      Reintubation3 (7%)5 (9%)5 (10%)8 (17%)2 (4%)23 (9.6%).29
      Pneumonia3 (7%)1 (2%)3 (6%)7 (15%)2 (4%)16 (6.7%).094
      Tracheostomy1 (2%)0 (0%)3 (6%)4 (9%)0 (0%)8 (3.3%).07
      FEV1, Forced expiratory volume in 1 second; SD, standard deviation; IQR, interquartile range; DLCO, diffusion capacity of the lungs for carbon monoxide.

       Postoperative Outcomes

      Postoperative outcomes for the patient population are listed in Tables 3 and 4 (additional data are available in Table E1). The average ventilator time was 51.1 hours for the FEV1 cohort with a median of 24 hours (interquartile range, 19-43), reflecting the outliers with very prolonged ventilator times (maximum 1222 hours). Reintubation, postoperative pneumonia, and tracheostomy rates were 10.6%, 6.8%, and 3.8%, respectively. There were no significant differences in these pulmonary function measures across the 5 FEV1 groups (P = .54, .12, .36, and .61 for ventilator time, postoperative pneumonia, reintubation, and tracheostomy rates).
      Table 4Correlation of postoperative outcomes with pulmonary function tests as continuous variables
      OutcomeFEV1DLCO
      RatioP valueRatioP value
      N263239
      Mortality (hazard ratio per 10% change)
      3-year mortality, Cox regression.
       Unadjusted0.99 (0.84-1.18).9250.91 (0.73-1.14).435
       Adjusted1.08 (0.90-1.30).4330.98 (0.77-1.24).838
      Hospital length of stay (ratio of means per 10% change)
      Negative binomial regression.
       Unadjusted0.93 (0.90-0.97).0010.93 (0.88-0.98).008
       Adjusted0.94 (0.91-0.98).0020.93 (0.88-0.97).004
      ICU length of stay (ratio of means per 10% change)
      Negative binomial regression.
       Unadjusted1.00 (0.94-1.06).9360.96 (0.88-1.04).286
       Adjusted1.00 (0.95-1.04).7720.94 (0.88-1.00).055
      Regressions adjusted for age, sex, INTERMACS profile, serum creatinine, reoperation status, ventilator time, postoperative pneumonia, and cardiopulmonary bypass time. FEV1, Forced expiratory volume in 1 second; DLCO, diffusion capacity of the lungs for carbon monoxide; ICU, intensive care unit.
      3-year mortality, Cox regression.
      Negative binomial regression.
      The median ventilator time was 25 hours (interquartile range 20-43) for the DLCO cohort. Reintubation, postoperative pneumonia, and tracheostomy rates were 9.6%, 6.7%, and 3.3%, respectively. There were no significant differences in these pulmonary function measures across the 5 DLCO groups (P = .24, .09, .29, and .07 for ventilator time, postoperative pneumonia, reintubation, and tracheostomy rates, respectively).
      In the linear analysis, there was no association of FEV1 or DLCO with survival (hazard ratio 0.99, confidence interval [CI] 0.84-1.18 and 0.91, CI 0.73-1.14; P = .925 and .435, respectively). This relationship held after adjustments in a multivariable model. However, there was a negative correlation between these PFT parameters and hospital LOS. Each 10%-point increase in FEV1 or DLCO was associated with a 7% decrease in hospital LOS (Table 4). No linear effect was shown for ICU LOS, although analysis of the groups showed a correlation of increased ICU LOS with decreased DLCO for values greater than 46% predicted.
      In the multivariable regression, age >60 years, postoperative pneumonia, and increasing cardiopulmonary bypass and ventilator times were associated with increasing risks of mortality. Female patients with postoperative pneumonia and longer ventilator times experienced longer ICU and hospital LOS.

       FEV1 as Predictor of Outcomes

      Patients with FEV1 less than 40% predicted (n = 13) had no operative deaths in this study. Mortality and 95% CIs at 1 and 3 years were 0.09 (0.01-0.60) and 0.11 (0.04-0.34) deaths per person year, respectively. This did not differ significantly from patients with normal FEV1 values. In line with this, Kaplan-Meier analysis showed no difference in survival across groups (Figure 1, log rank P = .841). Hospital LOS was significantly greater for patients with lower FEV1. Patients with FEV1 less than 40% predicted stayed 10 days longer in the hospital on average than patients with FEV1 greater than 78% predicted (ratio 1.43, 95% CI 1.03-2.0, P = .04). Patients with FEV1 68% to 78% predicted showed no significantly greater LOS than the normal group (Supplementary material). ICU LOS varied from 7.2 to 8.7 days in the different FEV1 groups, but no significant statistical difference existed between groups.
      Figure thumbnail gr1
      Figure 1Kaplan-Meier survival estimates for patients by FEV1 groups (Group 1: <40%, Group 2: 40%-55%, Group 3: 56%-67%, Group 4: 68%-78%, Group 5: >78%). FEV1, Forced expiratory volume in 1 second.

       DLCO as Predictor of Outcomes

      In the group of patients with DLCO less than 40% predicted (n = 41), operative and medium term mortality rates and 95% CIs at 1 and 3 years were 0, 0.14 (0.06-0.33), and 0.11 (0.06-0.22) deaths per person-year. This was statistically similar to observed rates in the other groups and is reflected in the Kaplan-Meier analysis (Figure 2, log rank test P value = .713). Hospital LOS was significantly longer in the groups with lower DLCO (Figure 3). Patients with DLCO less than 40% predicted stayed 7 days longer on average than those with DLCO >61% predicted (ratio 1.44, 95% CI 1.16-1.78, P < .001). However, patients with DLCO between 40% and 46% predicted (n = 55) had no different LOS than patients with DLCO greater than 61% predicted (n = 40).
      Figure thumbnail gr2
      Figure 2Kaplan-Meier survival estimates for patients by DLCO groups (Group 1: <40%, Group 2: 40%-46%, Group 3: 47%-53%, Group 4: 54%-61%, Group 5: >61). DLCO, Diffusion capacity of the lungs for carbon monoxide.
      Figure thumbnail gr3
      Figure 3Lengths of stay of different patient groups. FEV1 Groups (Group 1: <40%, Group 2: 40%-55%, Group 3: 56%-67%, Group 4: 68%-78%, Group 5: >78%). DLCO Groups (Group 1: <40%, Group 2: 40%-46%, Group 3: 47%-53%, Group 4: 54%-61%, Group 5: >61). FEV1, Forced expiratory volume in 1 second; DLCO, diffusion capacity of the lungs for carbon monoxide.
      ICU LOS was significantly greater in Group 3, representing DLCO of 47% to 53% predicted compared with the group with DLCO >61% predicted (ratio 1.78, 95% CI 1.3-2.4, P < .001). However, patients with the lowest DLCO values in this study did not show increased ICU LOS compared with the reference group (Table E1).

       Sensitivity Analysis Results

      A total of 47 patients had no testing for FEV1, whereas 71 had none for DLCO. These patients were on average 4 years younger than patients who underwent testing (P = .042), more likely to have worsening cardiac and renal function on inotropes (INTERMACS profile 2), and more likely to be supported with an intra-aortic balloon pump at implantation (P value < .001). In a Kaplan-Meier survival analysis, patients with no FEV1 testing had worse survival than the cohort of patients who were tested before implantation (log rank test P = .046). When the former were included entirely in the lowest FEV1 group (FEV1 <40% predicted), there were no observed changes in the relationships between groups. There was no observed difference in survival (ratio 1.68, 95% CI 0.80-3.53, log rank test P = .462) between the group inclusive of patients with no testing and the reference group (FEV1> 78% predicted). The ratio of ICU LOS of patients with FEV1 less than 40% predicted to the reference increased from 1.08 (.7-1.7) to 1.53 (1.11-2.11) with inclusion of patients with no FEV1 testing, but this remained statistically insignificant (P = .062). Hospital LOS difference remained statistically significant (ratio 1.53, 95% CI 1.22-1.92 from 1.43 (1.0-2.0), P = .03 and P < .001). Patients without DLCO testing had similar survival to the cohort of patients who were tested (log rank test, P value .706).

      Discussion

      Patient selection for advanced heart failure therapy, and more specifically, MCS, has been a focus of clinicians as devices have improved in reliability and durability. As a community, we generally have been successful in identifying patients who were ill enough to benefit from therapy. This has led to improved outcomes with 2-year survival in excess of 70%.
      • Kirklin J.K.
      • Naftel D.C.
      • Pagani F.D.
      • Kormos R.L.
      • Stevenson L.W.
      • Blume E.D.
      • et al.
      Sixth INTERMACS annual report: a 10,000-patient database.
      • Lietz K.
      • Miller L.W.
      Improved survival of patients with end-stage heart failure listed for heart transplantation.
      However, our judgment has been challenged repeatedly in the assessment of those too ill to benefit from this therapy. This is because one of our major risk indicators, noncardiac frailty, cannot be expected to improve with improved hemodynamics, and the greater the noncardiac frailty, the less the potential benefit.
      • Flint K.M.
      • Matlock D.D.
      • Lindenfeld J.
      • Allen L.A.
      Frailty and the selection of patients for destination therapy left ventricular assist device.
      Noncardiac frailty can take many forms, including physical debilitation, cachexia, and end-organ dysfunction. Although there are well-studied physical debilitation measures such as handgrip strength, gait speed, and unintentional weight loss, the prognostic value of end organ damage is still under investigation. There have been attempts at assessment of liver disease, which have suggested acceptable outcomes with liver dysfunction.
      • Demirozu Z.T.
      • Hernandez R.
      • Mallidi H.R.
      • Singh S.K.
      • Radovancevic R.
      • Segura A.M.
      • et al.
      HeartMate II left ventricular assist device implantation in patients with advanced hepatic dysfunction.
      • Sargent J.E.
      • Dardas T.F.
      • Smith J.W.
      • Pal J.D.
      • Cheng R.K.
      • Masri S.C.
      • et al.
      Periportal fibrosis without cirrhosis does not affect outcomes after continuous flow ventricular assist device implantation.
      Although continuous-flow LVAD implantation results in improved postoperative renal function,
      • Sandner S.E.
      • Zimpfer D.
      • Zrunek P.
      • Rajek A.
      • Schima H.
      • Dunkler D.
      • et al.
      Renal function and outcome after continuous flow left ventricular assist device implantation.
      patients with preoperative renal dysfunction have been shown to have worse survival with MCS.
      • Coffin S.T.
      • Waguespack D.R.
      • Haglund N.A.
      • Maltais S.
      • Dwyer J.P.
      • Keebler M.E.
      Kidney dysfunction and left ventricular assist device support: a comprehensive perioperative review.
      • Yoshioka D.
      • Sakaguchi T.
      • Saito S.
      • Miyagawa S.
      • Nishi H.
      • Yoshikawa Y.
      • et al.
      Predictor of Early mortality for severe heart failure patients with left ventricular assist device implantation.
      However, lung disease has not been well investigated in the MCS population and to date, there has been no concerted effort to evaluate the efficacy of PFTs as predictors of mortality in this patient population. Ghotra and colleagues
      • Ghotra A.S.
      • Hussain Z.
      • Bhatia N.
      • Perez R.
      • Cheng A.
      • Slaughter M.S.
      • et al.
      Impact of pulmonary function tests on outcomes in patients after left ventricular assist device implantation.
      presented preliminary data suggesting a significant association between DLCO less than 50% predicted and increased mortality in their patients undergoing MCS at a single center.
      In this study, patients with very low PFTs were implanted with ventricular assist devices. A significant percentage of patients underwent redo sternotomy, and a third had additional procedures during device implantation. However, we found no association between either DLCO or FEV1 with mortality. These findings may be anticipated; by modifying patients' heart failure with treatment, the natural history of their PFTs could be altered. For example, it is known that DLCO and FEV1 worsen with heart failure as the heart dilates and left ventricular end-diastolic pressure increases, leading to pulmonary interstitial edema and hydrostatic alveolar injury,
      • Apostolo A.
      • Giusti G.
      • Gargiulo P.
      • Bussotti M.
      • Agostoni P.
      Lungs in heart failure.
      but data exist on the improvement of these tests with treatment of heart failure and particularly after heart transplantation.
      • Mettauer B.
      • Lampert E.
      • Charloux A.
      • Zhao Q.M.
      • Epailly E.
      • Oswald M.
      • et al.
      Lung membrane diffusing capacity, heart failure, and heart transplantation.
      Therefore, we would expect that with MCS and particularly subsequent transplant (for those transplant-eligible), the pulmonary morbidity would be improved. This would lead to better outcomes in these patients than in patients who undergo nonheart failure surgery where the expected postoperative improvement in pulmonary function may be more limited.
      Review of the patients with the lowest FEV1 values shows that pulmonary optimization such as drainage of pleural effusions and oxygen wean contributed at least partially to increased LOS in one half of the patients with prolonged hospital stays. However, other significant factors like severe right ventricular failure, renal dysfunction, and exercise deconditioning were other etiologies that contributed. Again, there was no observed difference in postoperative pulmonary morbidity as characterized by reintubation rates, ventilator time, tracheostomy rates, and pneumonia. This finding further supports the principle that measured preoperative PFTs may be indicators of cardiac, more so, than pulmonary disease in this patient population.
      Figure thumbnail fx2
      Video 1Minimally invasive, off-pump implantation of a HeartWare HVAD (HeartWare, Inc., Framingham, Mass). The patient is a 31-year-old man with nonischemic cardiomyopathy and a dilated left ventricle. Right ventricular function was reasonable for isolated LVAD implantation. The ventricular apex was marked with a transthoracic echocardiogram. An 8-cm incision was made in the left sixth interspace, the seventh rib was divided, a soft-tissue retractor was placed, and a pericardial well created. Simultaneously, a right third interspace incision was made and the rib was divided. The mammary vessels were ligated. A soft-tissue retractor was placed and a pericardial well was created. An umbilical tape was passed between the incisions for later use as a guide for the outflow graft. The apex was elevated into the field by placing a sponge behind the heart. Radial 2-0 braided sutures with pledgets were placed through the sewing cuff and then tied down (note: the anchoring screw should be loosened but not removed at this point). A pacing wire was placed for rapid ventricular pacing (optional). The pump was primed and prepared on the back table. Sizing of the incision was confirmed with the pump, and the patient was heparinized with a goal activated clotting time of 300 seconds. A cruciate incision was made in the apex under rapid ventricular pacing. The coring tool was inserted atraumatically (note: once the coring tool is flush with the sewing cuff, it is generally hemostatic). Coring was then completed and the LVAD inserted with the outflow graft unclamped for deairing (note: during this process, approximately 100 mL of blood can be lost). The umbilical tape was tied to the outflow graft and tunneled to the right thoracotomy incision. The driveline was tunneled out the abdominal wall leaving the velour buried, looped in the pericardium, and then attached to the controller. A side-biting clamp was placed on the aorta, and the outflow graft anastomosis was performed with a 5-0 monofilament suture. A de-airing needle was placed in the outflow graft. The LVAD was then started and slowly increased to nominal speed of 2400 rpm. The apex was covered with a Gore-Tex (Gore, Newark, De) pericardial membrane and the right thoracotomy pericardial incision was closed primarily. A paracostal suture was placed over the apex and the wound closed in layers. The right third rib was reapproximated and the wound closed in layers. Transesophageal echocardiography confirmed excellent inflow orientation; a topogram demonstrates the overall lay of the pump. The patient made an excellent recovery and awaits heart transplant. LVAD, Left ventricular assist device. Video available at: http://www.jtcvsonline.org/article/S0022-5223(17)30675-X/fulltext.
      These findings may help explain the observations in our group analysis, where we found that while generally showing increased ICU LOS with worsening DLCO, patients in the lowest DLCO group showed no significantly longer ICU LOS than those in the normal group, although they did not have increased early mortality. The major contributor to these patients' preoperative performance may be their severe cardiac dysfunction that is then rapidly improved postoperatively, allowing more rapid transit through the ICU.
      The economic impact of longer LOS with poor PFTs cannot be ignored. LVAD surgery is expensive, as demonstrated in a Columbia University study, which reported an initial hospitalization cost of $169,535 for a 37-day stay.
      • Miller L.W.
      • Guglin M.
      • Rogers J.
      Cost of ventricular assist devices: can we afford the progress?.
      Thus, although the overall survival was the same in our analysis, the increase of 10 days in overall hospital LOS for patients with the worst FEV1 values can have a meaningful impact on hospital and personnel resource use. Perhaps future studies can clarify what interventions, if any, when applied to patients with the worst PFT values can decrease the LOS differences observed. Since this study, our practice has evolved in the use of PFTs as part of preoperative evaluation of patients undergoing LVAD implantation. When patients have concerning PFT values, we work collaboratively with our pulmonary medicine colleagues, obtain cross-sectional imaging for evaluation of pulmonary parenchymal disease, determine obvious clinical morbidity such as oxygen dependence, and perform functional tests such as the 6-minute walk test and cardiopulmonary exercise testing as tolerated. The integration of this assessment ultimately guides patient selection with respect to pulmonary comorbidity.

       Limitations

      This study was retrospective and nonrandomized and thus limited by the usual confinements of such studies. The data obtained were abstracted from a quality improvement database, which had limited fields of data. Importantly, it is likely that some patients with very low PFTs and other comorbidities were evaluated but not offered MCS during the period under review and thus not analyzed. The small number of patients in the lowest FEV1 group may not be sufficient to demonstrate small survival differences. Patients did not routinely undergo postimplant PFTs, and, therefore, we are unable to evaluate changes in these measures over time. Furthermore, not all patients underwent pulmonary function testing before LVAD implantation.

      Conclusions

      This study finds no association between abnormal PFTs and increased mortality after continuous flow LVAD implantation. Our program performs a multidisciplinary evaluation that takes patient comorbidities into consideration. Our experience suggests that carefully selected patients, including those with PFT abnormalities, can undergo MCS with excellent survival. Therefore, PFTs alone should not preclude advanced heart failure therapies to such patients.

       Conflict of Interest Statement

      Dr Mahr is a consultant to St Jude, HeartWare, and Abiomed. Dr Mokadam is a consultant to St Jude, HeartWare, and Syncardia Systems Inc. Dr Smith reports a relationship with HeartWare and Transmedics. Drs Pal and Levy report research grants from HeartWare. Dr Dardas reports research support from International Society of Heart Lung Transplant/HeartWare. All other authors have nothing to disclose with regard to commercial support.
      We thank Orvalho Augusto for assistance with statistical analysis.

      Appendix

      Table E1Correlation of postoperative outcomes with pulmonary function test categories
      Characteristic12345P value
      FEV1 group
       n1364646260
       Mortality
      Unadjusted, HR (95% CI)1.11 (0.3–4.0)1.03 (0.5–2.4)0.95 (0.4–2.2)1.05 (0.5–2.4)1.999
      Adjusted, HR (95% CI)0.83 (0.2–3.2)0.75 (0.3–1.7)0.71 (0.3–1.7)0.90 (0.4–2.1)1.934
      Spline analysis, HR (95% CI)1.02 (0.8–1.3)0.96 (0.9–1.0)1.05 (1.0–1.1)1.01 (0.96–1.05)1.713
       Hospital length of stay, d, mean (SD)28.2 (17.4)23.7 (15.2)22.6 (12.6)21.2 (14.3)18.0 (11.8)
      Unadjusted ratio (95% CI)1.55 (1.1–2.1)1.32 (1.1–1.6)1.25 (1.0–1.5)1.10 (1.0–1.5)1.012
      Adjusted ratio (95% CI)1.43 (1.0–2.0)1.27 (1.1–1.5)1.22 (1.0–1.5)1.07 (0.9–1.3)1.04
       ICU length of stay, mean (SD)8.5 (4.1)7.5 (6.7)8.7 (9.8)8.7 (11.4)7.1 (8.1)
      Unadjusted ratio (95% CI)1.18 (0.7–1.9)1.09 (0.8–1.4)1.20 (0.9–1.6)1.15 (0.9–1.5)1.719
      Adjusted ratio (95% CI)1.08 (0.7–1.7)1.05 (0.8–1.4)1.13 (0.9–1.5)1.03 (0.8–1.4)1.915
      DLCO group
       n4155554840
       Mortality
      Unadjusted, HR (95% CI)1.20 (0.5–3.1)1.00 (0.4–2.6)1.55 (0.6–3.7)1.69 (0.6–4.4)1.837
      Adjusted, HR (95% CI)0.95 (0.4–2.6)0.83 (0.3–2.2)1.46 (0.6 –3.6)1.13 (0.4–3.0)1.754
      Spline analysis, HR (95% CI)0.90 (0.8–1.0)1.2 (1.0–1.4)0.89 (0.7–1.1)1.0 (0.9–1.1)1.395
       Hospital length of stay, d, mean (SD)23.9 (13.1)19.4 (10.0)24.4 (16.9)21.6 (13.3)17.0 (11.0)
      Unadjusted ratio (95% CI)1.41 (1.1–1.8)1.14 (0.9–1.4)1.44 (1.2–1.8)1.27 (1.0–1.6)1.005
      Adjusted ratio (95% CI)1.44 (1.2–1.8)1.16 (0.9–1.4)1.46 (1.2–1.8)1.33 (1.1–1.6)1<.001
       ICU length of stay, mean (SD)6.8 (3.5)6.8 (4.2)11.1 (15.7)8.5 (8.3)5.8 (4.8)
      Unadjusted ratio (95% CI)1.16 (0.8–1.6)1.17 (0.9–1.6)1.89 (1.4–2.5)1.46 (1.1–2.0)1<.001
      Adjusted ratio (95% CI)1.19 (0.9–1.6)1.16 (0.9–1.5)1.78 (1.3–2.4)1.46 (1.1–2.0)1<.001
      FEV1, Forced expiratory volume in 1 second; HR, hazard ratio; CI, confidence interval; SD, standard deviation; ICU, intensive care unit; DLCO, diffusion capacity of the lungs for carbon monoxide.

      Supplementary Data

      References

        • Go A.S.
        • Mozaffarian D.
        • Roger V.L.
        • Benjamin E.J.
        • Berry J.D.
        • Borden W.B.
        • et al.
        Heart disease and stroke statistics—2013 update: a report from the American Heart Association.
        Circulation. 2013; 127: e6-e245
        • Yancy C.W.
        • Jessup M.
        • Bozkurt B.
        • Butler J.
        • Casey D.E.
        • Drazner M.H.
        • et al.
        2013 ACCF/AHA Guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association Task Force on practice guidelines.
        Circulation. 2013; 128: e240-e327
        • Shih T.
        • Paone G.
        • Theurer P.F.
        • McDonald D.
        • Shahian D.M.
        • Prager R.L.
        The Society of Thoracic Surgeons adult cardiac surgery database version 2.73: more is better.
        Ann Thorac Surg. 2015; 100: 516-521
        • Iversen K.K.
        • Kjaergaard J.
        • Akkan D.
        • Kober L.
        • Torp-Pedersen C.
        • Hassager C.
        • et al.
        The prognostic importance of lung function in patients admitted with heart failure.
        Eur J Heart Fail. 2010; 12: 685-691
        • Olson T.
        • Denzer D.
        • Sinnett W.
        • Wilson T.
        • Johnson B.
        Prognostic value of resting pulmonary function in heart failure.
        Clin Med Insights Circ Respir Pulm Med. 2013; 7: 35-43
        • de Oliveira E.K.
        • da Silva V.Z.M.
        • Turquetto A.L.R.
        Relationship of postoperative walk test and lung function with the length of heart surgery hospital stay.
        Rev Bras Cir Cardiovasc. 2009; 24: 478-484
        • Henn M.C.
        • Zajarias A.
        • Lindman B.R.
        • Greenberg J.W.
        • Melby S.J.
        • Quader N.
        • et al.
        Preoperative pulmonary function tests predict mortality after surgical or transcatheter aortic valve replacement.
        J Thorac Cardiovasc Surg. 2016; 151: 578-586.e2
        • Zibrak J.D.
        • O’Donnell C.R.
        • Marton K.
        Indications for pulmonary function testing.
        Ann Intern Med. 1990; 112: 763-771
        • Hawkins N.M.
        • Petrie M.C.
        • Jhund P.S.
        • Chalmers G.W.
        • Dunn F.G.
        • McMurray J.J.V.
        Heart failure and chronic obstructive pulmonary disease: diagnostic pitfalls and epidemiology.
        Eur J Heart Fail. 2009; 11: 130-139
        • Ivanov A.
        • Yossef J.
        • Tailon J.
        • Worku B.M.
        • Gulkarov I.
        • Tortolani A.J.
        • et al.
        Do pulmonary function tests improve risk stratification before cardiothoracic surgery?.
        J Thorac Cardiovasc Surg. 2016; 151: 1183-1189.e3
        • Cohen A.
        • Katz M.
        • Katz R.
        • Hauptman E.
        • Schachner A.
        Chronic obstructive pulmonary disease in patients undergoing coronary artery bypass grafting.
        J Thorac Cardiovasc Surg. 1995; 109: 574-581
        • Bevelaqua F.
        • Garritan S.
        • Haas F.
        • Salazar-Schicchi J.
        • Axen K.
        • Reggiani J.-L.
        Complications after cardiac operations in patients with severe pulmonary impairment.
        Ann Thorac Surg. 1990; 50: 602-606
        • Berry M.F.
        • Jeffrey Yang C.-F.
        • Hartwig M.G.
        • Tong B.C.
        • Harpole D.H.
        • D’Amico T.A.
        • et al.
        Impact of pulmonary function measurements on long-term survival after lobectomy for stage i non-small cell lung cancer.
        Ann Thorac Surg. 2015; 100: 271-276
        • Kirklin J.K.
        • Naftel D.C.
        • Pagani F.D.
        • Kormos R.L.
        • Stevenson L.W.
        • Blume E.D.
        • et al.
        Sixth INTERMACS annual report: a 10,000-patient database.
        J Heart Lung Transplant. 2014; 33: 555-564
        • Lietz K.
        • Miller L.W.
        Improved survival of patients with end-stage heart failure listed for heart transplantation.
        J Am Coll Cardiol. 2007; 50: 1282-1290
        • Flint K.M.
        • Matlock D.D.
        • Lindenfeld J.
        • Allen L.A.
        Frailty and the selection of patients for destination therapy left ventricular assist device.
        Circ Heart Fail. 2012; 5: 286-293
        • Demirozu Z.T.
        • Hernandez R.
        • Mallidi H.R.
        • Singh S.K.
        • Radovancevic R.
        • Segura A.M.
        • et al.
        HeartMate II left ventricular assist device implantation in patients with advanced hepatic dysfunction.
        J Card Surg. 2014; 29: 419-423
        • Sargent J.E.
        • Dardas T.F.
        • Smith J.W.
        • Pal J.D.
        • Cheng R.K.
        • Masri S.C.
        • et al.
        Periportal fibrosis without cirrhosis does not affect outcomes after continuous flow ventricular assist device implantation.
        J Thorac Cardiovasc Surg. 2016; 151: 230-235
        • Sandner S.E.
        • Zimpfer D.
        • Zrunek P.
        • Rajek A.
        • Schima H.
        • Dunkler D.
        • et al.
        Renal function and outcome after continuous flow left ventricular assist device implantation.
        Ann Thorac Surg. 2009; 87: 1072-1078
        • Coffin S.T.
        • Waguespack D.R.
        • Haglund N.A.
        • Maltais S.
        • Dwyer J.P.
        • Keebler M.E.
        Kidney dysfunction and left ventricular assist device support: a comprehensive perioperative review.
        Cardiorenal Med. 2015; 5: 48-60
        • Yoshioka D.
        • Sakaguchi T.
        • Saito S.
        • Miyagawa S.
        • Nishi H.
        • Yoshikawa Y.
        • et al.
        Predictor of Early mortality for severe heart failure patients with left ventricular assist device implantation.
        Circ J. 2012; 76: 1631-1638
        • Ghotra A.S.
        • Hussain Z.
        • Bhatia N.
        • Perez R.
        • Cheng A.
        • Slaughter M.S.
        • et al.
        Impact of pulmonary function tests on outcomes in patients after left ventricular assist device implantation.
        J Card Fail. 2015; 21: S93
        • Apostolo A.
        • Giusti G.
        • Gargiulo P.
        • Bussotti M.
        • Agostoni P.
        Lungs in heart failure.
        Pulm Med. 2012; 2012: 1-9
        • Mettauer B.
        • Lampert E.
        • Charloux A.
        • Zhao Q.M.
        • Epailly E.
        • Oswald M.
        • et al.
        Lung membrane diffusing capacity, heart failure, and heart transplantation.
        Am J Cardiol. 1999; 83: 62-67
        • Miller L.W.
        • Guglin M.
        • Rogers J.
        Cost of ventricular assist devices: can we afford the progress?.
        Circulation. 2013; 127: 743-748