http://www.jhltonline.org
ORIGINAL CLINICAL SCIENCE
Concomitant tricuspid valve surgery during implantation of continuous-flow left ventricular assist devices: A Society of Thoracic Surgeons database analysis Jason O. Robertson, MD, MS,a Maria V. Grau-Sepulveda, MD,b Shoichi Okada, MD,a Sean M. O’Brien, PhD,b J. Matthew Brennan, MD, MPH,b Ashish S. Shah, MD,c Akinobu Itoh, MD,a Ralph J. Damiano, MD,a Sunil Prasad, MD,a and Scott C. Silvestry, MDa From the aDivision of Cardiothoracic Surgery, Washington University, St. Louis, Missouri; bDuke Clinical Research Institute, Duke University, Durham, North Carolina; and the cDivision of Cardiac Surgery, Johns Hopkins University, Baltimore, Maryland.
KEYWORDS: Left ventricular assist device; tricuspid; tricuspid regurgitation
BACKGROUND: Performing concomitant tricuspid valve procedures (TVPs) in left ventricular assist device (LVAD) patients with significant pre-operative tricuspid regurgitation (TR) is controversial, and no studies have been large enough to definitively guide therapy. METHODS: Between January 2006 and September 2012, 2,196 patients with moderate to severe preoperative TR from 115 institutions underwent implantation of a continuous-flow left ventricular assist device (LVAD) as reported by The Society of Thoracic Surgeons National Database. Of these, 588 (27%) underwent a concomitant TVP. Inverse probability weighting based on propensity score was used to adjust for differences between the LVAD alone and LVADþTVP groups, and outcomes were compared. RESULTS: Most patients in the LVADþTVP group underwent an annuloplasty alone (81.1%). Concomitant TVP did not affect risk of post-operative right VAD insertion (risk ratio [RR], 0.81; 95% confidence interval [CI], 0.49–1.36; p ¼ 0.4310) or death (RR, 0.95; 95% CI, 0.68–1.33; p ¼ 0.7658). However, TVP was associated with an increased risk for post-operative renal failure (RR, 1.53; 95% CI, 1.13–2.08; p ¼ 0.0061), dialysis (RR, 1.49; 95% CI, 1.03–2.15; p ¼ 0.0339), reoperation (RR, 1.24; 95% CI, 1.07–1.45; p ¼ 0.0056), greater total transfusion requirement (RR, 1.03; 95% CI, 1.01–1.05; p ¼ 0.0013), and hospital length of stay 4 21 days (RR, 1.29; 95% CI, 1.16–1.43; p o 0.0001). Time on the ventilator and intensive care unit length of stay were also significantly prolonged for the LVADþTVP group. CONCLUSIONS: Performing a concomitant TVP for continuous-flow LVAD patients with moderate to severe TR did not reduce early death or right VAD requirement and was associated with worse early postoperative outcomes. These data caution against routine concomitant TVP based solely on degree of preoperative TR and suggest that additional selection criteria are needed to identify those patients in whom concomitant TVP may prevent post-operative right ventricular failure. J Heart Lung Transplant ]]]];]:]]]–]]] r 2014 International Society for Heart and Lung Transplantation. All rights reserved.
Reprint requests: Scott C. Silvestry, MD, Surgical Director, Heart Transplant and Mechanical Cardiac Support Program, Washington University School of Medicine/Barnes-Jewish Hospital, 660 S Euclid Ave, Box 8234, St. Louis, MO 63110. Telephone: 314-362-6182. Fax: 314747-4216. E-mail address: [email protected]
Patients referred for management with left ventricular assist devices (LVADs) often have some degree of preoperative right-sided heart failure and frequently present with significant tricuspid regurgitation (TR). Some of these patients have right ventricular (RV) failure that persists
1053-2498/$ - see front matter r 2014 International Society for Heart and Lung Transplantation. All rights reserved. http://dx.doi.org/10.1016/j.healun.2014.01.861
2
The Journal of Heart and Lung Transplantation, Vol ], No ], Month ]]]]
post-operatively. Despite improvements with continuousflow LVADs (cfLVADs), post-operative RV dysfunction complicates up to 22% of cases and is associated with significant increases in morbidity and mortality.1 A number of authors have advocated performing TV repair or replacement in the presence of moderate to severe TR as a strategy to reduce post-operative RV failure. The rationale for performing a concomitant tricuspid valve procedure (TVP) during LVAD insertion in patients with significant TR is sound, but the currently available data are limited to reports from relatively small, retrospective, single-institution studies that have been underpowered to address the issue.2–9 Moreover, reports on the prognostic significance of pre-operative TR have produced mixed conclusions, likely due to changes in right heart hemodynamics that occur with insertion of a cfLVAD.10–12 Some have argued that a sub-set of patients with TR will improve from mechanical unloading with an LVAD alone, thereby negating their need for a concomitant TVP.13 As a result, the decision to perform concomitant TVPs remains controversial. This study used The Society for Thoracic Surgeons (STS) Adult Cardiac Surgery Database to address how concomitant TV repair or replacement at the time of LVAD implantation affects short-term outcomes for patients with cfLVADs. This analysis of TVP in a large number of patients with moderate to severe TR undergoing cfLVAD implantation should provide insight about the utility of performing concomitant TV procedures according to the degree of pre-operative TR.
surface area); medical history and risk factors (i.e., diabetes, hypertension, cardiovascular disease, cerebrovascular accident, peripheral vascular disease, dialysis, ejection fraction, last creatinine measurement, immunosuppression, chronic lung disease); cardiac status at the time of operation (i.e., myocardial infarction [MI] o 21 days, unstable angina without MI, cardiogenic shock, atrial fibrillation, congestive heart failure/New York Heart Association class IV); presence of a pre-operative intraaortic balloon pump (IABP) or inotrope infusions; degree of aortic, mitral, and tricuspid valvular insufficiency; pre-operative coronary catheterization data (number of diseased vessels, Z 50% left main disease) and operative variables (i.e., reoperation, other concomitant procedures, categorization of operative urgency, and year of surgery). We also included multiplicative interaction terms between age and reoperation, age and operative status, and sex and body surface area. Missing categoric variables were imputed to the most common category, and missing continuous variables were imputed by
Methods Approval for this study was granted by the STS Database Access and Publications Committee.
Patient population The study population consisted of patients from the STS Adult Cardiac Surgery Database who underwent insertion of a cfLVAD that was indicated for bridge to recovery, bridge to transplant, or destination therapy between January 2006 and September 2012 (Figure 1). Exclusions included patients (1) with a prior VAD, (2) with an isolated right VAD (RVAD) implantation, (3) for whom pre-operative TR data were missing, (4) with no, trivial, or mild TR, (5) who had a concomitant aortic or mitral valve procedure, (6) who received a heart transplant at the time of initial LVAD insertion, or (7) with active endocarditis. This resulted in 2,196 cfLVAD patients with moderate to severe TR, and concomitant TV repair or replacement was performed in 588.
Adjustments for between-group differences Propensity scores to estimate the probability of undergoing a TVP were developed with the use of logistic regression to adjust for between-group differences in baseline characteristics of the patients and hospitals. Variables included in the model were based on previously published valve models and clinical relevance.14 These included annual volume of LVAD implantations; annual volume of TVPs; patient demographics (i.e., age, race, sex, body
Figure 1 Study population. AV, aortic valve; BTR, bridge to recovery; BTT, bridge to transplant; cfLVAD, continuous-flow left ventricular assist device; DT, destination therapy; MV, mitral valve; RVAD, right ventricular assist device; TR, tricuspid regurgitation; TVP, tricuspid valve procedure; VAD, ventricular assist device.
Robertson et al.
Concomitant TVP for TR in cfLVAD Patients
stratifying patients according to related risk factors and then imputing stratum-specific medians. Missing data were rare (o 0.5% for most covariates), but for any variable with 4 5% missing data, we confirmed that the p-value comparing the percentage of missing data between groups was non-significant. Trimming was performed by excluding 35 observations with extreme propensities that did not overlap with the other treatment group. Summaries of the propensity distributions after trimming (minimum, 25th percentile, median, 75th percentile, and maximum) were similar in each group (no TVP: 3.8%, 10.7%, 16.8%, 30.1%, 91.0%; and TVP: 3.8%, 21.0%, 39.4%, 56.4%, 91.1%). Inverse probability weighting (IPW) according to the propensity score was used to adjust for differences between the 2 treatment groups. An advantage of this approach over propensity matching is that all observations are retained in the analysis, and it is, therefore, relatively efficient. In contrast, propensity matching would result in discarding large numbers of observations and would be less efficient. IPW involved weighting each patient who underwent a TVP by the inverse of the probability of being selected for a TVP and weighting each patient who did not undergo a TVP by the inverse of the probability of not being selected for a TVP. The ability of the propensity model to balance the 2 treatment groups was assessed by comparing the distribution of baseline characteristics before and after weighting by the inverse of the propensity score. Because the standardized differences after weighting for all variables were o |10| (or 0.1 standard deviation), we confidently assumed that balance was achieved between groups. The distribution of standardized differences before and after IPW is shown in Figure 2.
Statistics Summary statistics are presented as percentages for categoric variables and and as mean ⫾ standard deviation for continuous variables. Baseline characteristics were compared for patients between groups with the Pearson chi-square test for categoric variables and with the chi-square rank-based group means score statistics, which is equivalent to the Kruskal-Wallis and Wilcoxon tests, for continuous variables. To compare dichotomous outcomes between treatment groups, before and after IPW adjustment, we calculated risk ratios (RR) using generalized linear models with a log-link function.
Figure 2
3 The Huber-White variance estimator was used to compute 95% confidence intervals (CIs) and p-values.15 Continuous outcomes were similarly compared between groups, before and after IPW adjustment, using Wald tests with an empirical (Huber-White) variance estimator. Outcomes were only computed for patients with non-missing values. The primary outcomes considered were operative mortality (death during the same hospitalization as surgery, regardless of timing, or r 30 days of surgery, regardless of venue) and reoperation for RVAD insertion, which serves as proxy evidence of severe post-operative right-sided heart failure. New renal failure/dialysis, hospital length of stay (continuous and dichotomized at 4 21 days), intensive care unit (ICU) length of stay (continuous and dichotomized at the median, 4 172 hours for the overall stay), prolonged ventilation (4 24 hours), transfusion requirements, stroke, other causes for reoperation, and VAD complications (intracranial bleeding, device infection, and device malfunction) were also compared between groups to assess potential differences in operative morbidity and clinical course. Values of p o0.05 were considered significant. We did not adjust for multiple comparisons, and this should be taken into account when interpreting our results, because the probability of finding significant associations by chance increases with the number of outcomes analyzed. For example, if 20 independent hypotheses were to be considered, the probability of observing at least 1 p-value of o 0.05 by chance alone would be 64%. A sensitivity analysis was performed to compare results from the IPW approach with results from traditional multivariable regression analysis for the most relevant outcomes. Specifically, conditional logistic regression was used to compute odds ratios for TVP and each pertinent dichotomous outcome, and the model was stratified by hospital identification to account for potential confounding by center effects.16 Covariates for each model were identical to those used in the propensity model.
Results Baseline, clinical, and operative characteristics Baseline patient demographic and clinical characteristics are presented in Table 1. The LVADþTVP group comprised
The distribution of standardized differences before and after inverse probability weighting (IPW).
The Journal of Heart and Lung Transplantation, Vol ], No ], Month ]]]]
4 Table 1
Baseline Patient Demographic and Clinical Characteristics
Variablea Age, years Caucasian Body surface area, m2 Female NYHA class IV Ejection fraction, % Pre-op creatinine 4 2.0 mg/dl Renal failure Dialysis No dialysis Hypertension Dyslipidemia Diabetes mellitus Non–insulin-dependent Insulin-dependent Cerebrovascular accident Number diseased vessels 3 2 1 0 Left main disease Peripheral vascular disease Immunosuppression Atrial fibrillation Myocardial infarction Chronic lung disease (mod/severe) Cardiac reoperation Previous valve surgery Tricuspid insufficiency (severe) Mitral insufficiency (mod/severe) Aortic insufficiency (mod/severe) Cardiogenic shock Resuscitation Pre-op inotropes Pre-op intraaortic balloon pump Cardiac presentation STEMI NSTEMI Unstable angina Stable angina Symptoms unlikely to be ischemia No symptoms or angina Year 2006–2008 2009–2010 2011–2012
LVAD (n ¼ 1,608)
LVADþTVP (n ¼ 588)
55.8 ⫾ 13.5 66.4 (1,067) 2.0 ⫾ 0.2 25.3 (406) 82.3 (1,324) 16.3 ⫾ 6.8 12.3 (198)
55.2 ⫾ 13.7 61.9 (364) 2.0 ⫾ 0.2 27.4 (161) 80.8 (475) 16.2 ⫾ 6.5 16.3 (96)
3.4 (55) 10.7 (172) 60.1 (967) 61.9 (996)
3.4 (20) 14.8 (87) 55.1 (324) 57.7 (339)
17.2 (277) 21.4 (344) 9.5 (153)
18.2 (107) 20.4 (120) 8.2 (48)
18.4 (295) 6.9 (111) 6.3 (102) 61.0 (981) 4.5 (72) 11.1 (178) 4.2 (68) 64.2 (1,032) 36.9 (594) 18.8 (303) 39.3 (632) 7.7 (124) 24.6 (395) 76.6 (1,232) 2.9 (47) 29.9 (480) 1.7 (27) 77.1 (1,239) 30.5 (491)
13.4 (79) 6.5 (38) 6.3 (37) 64.0 (376) 2.2 (13) 7.1 (42) 3.6 (21) 64.1 (377) 31.3 (184) 17.5 (103) 37.9 (223) 9.0 (53) 56.1 (330) 69.4 (408) 1.4 (8) 30.1 (177) 1.9 (11) 76.7 (451) 41.3 (243)
0.8 (12) 1.4 (22) 4.6 (70) 4.6 (70) 50.4 (770) 32.7 (499)
0.4 (2) 0.7 (4) 3.6 (20) 4.3 (24) 43.0 (241) 35.0 (196)
13.9 (223) 42.2 (679) 43.9 (706)
12.9 (76) 36.6 (215) 50.5 (297)
p-value 0.4897 0.4021 0.4316 0.3122 0.4527 0.9206 0.0132 0.0280
0.0261 0.0469 0.8040
0.3283 0.0851
0.0182 0.0060 0.4855 0.9467 0.0136 0.4877 0.5438 0.3230 o0.0001 0.0287 0.0556 0.9497 0.7685 0.9219 o0.0001 0.2245
0.0204
LVAD, left ventricular assist device; mod, moderate; NSTEMI, non-ST elevation myocardial infarction; NYHA, New York Heart Association; STEMI, ST elevation myocardia infarction; TVP, tricuspid valve procedure. a Continuous data are presented as mean ⫾ standard deviation and categoric data as percentage (n).
26.8% of the final study population. Those that received a TVP more commonly had severe TR (56.1% vs 24.6%, p o 0.0001) and had significantly higher incidences of pre-operative renal failure not requiring dialysis (14.8% vs 10.7%, p ¼ 0.0280) and IABP use (41.3% vs 30.5%, p o 0.0001). Conversely, patients receiving a TVP had significantly lower
incidences of hypertension, dyslipidemia, prior MI, coronary artery disease, moderate to severe mitral insufficiency, and peripheral vascular disease. More than 85% of the patients in this study are from 2009 onward, and the number of TVP cases increased in the years 2011 to 2012, with 50.5% of LVADþTVP study patients from that period.
Robertson et al.
Concomitant TVP for TR in cfLVAD Patients
Table 2 compares the operative characteristics between the groups. Most LVADs in both groups were HeartMate II devices (Thoratec, Pleasanton, CA) inserted as a bridge to transplantation. Patients that received a TVP had an 3% higher incidence of full cardiopulmonary bypass (CPB), defined as the need for CPB or coronary perfusion for the entire procedure, and an average increase in perfusion time of 36 minutes (125.0 ⫾ 42.2 vs 88.8 ⫾ 40.2 minutes; p o 0.0001). Most patients in both groups did not require aortic occlusion, but it was more commonly used for those undergoing an LVADþTVP (9.9% vs 6.3%; p ¼ 0.0144), and cross-clamp times were prolonged by an average of 12 minutes. Concomitant surgeries other than mitral, aortic, and TV operations were also more commonly performed in the LVADþTVP group (28.1% vs 19.7%; p o 0.0001). Table 2
Variable
5 These included coronary artery bypass grafting in 0.9% (5 of 588) vs 1.2% (19 of 1,608), pulmonic valve procedures in 0.2% (1 of 588) vs 0.0% (0 of 1,608), LV aneurysm repairs in 0.2% (1 of 588) vs 0.1% (1 of 1,608), ventricular septal defect repairs in 0.0% (0 of 588) vs 0.2% (3 of 1,608), atrial septal defect repairs in 12.2% (72 of 588) vs 3.4% (55 of 1,608), other congenital defect repairs in 1.9% (11 of 588) vs 0.8% (13 of 1,608), ablation of atrial fibrillation in 0.2% (12 of 588) vs 0.01% (12 of 1,608), aortic aneurysm and/or dissection repairs in 0.0% (0 of 588) vs 0.003% (5 of 1,606), other cardiac procedures in 15.7% (92 of 588) vs 12.6% (202 of 1,607), and other non-cardiac procedures in 1.5% (9 of 588) vs 2.0% (32 of 1,608). Among those who received a TVP, annuloplasty was performed 81.1% of the time.
Operative Characteristics a
Status Emergent salvage Emergent Urgent Elective Cardiopulmonary bypassb Full Combination None Perfusion time, min Aortic occlusion Balloon occlusion Aortic cross-clamp None/partial cross-clamp Cross-clamp time, min VAD indication Destination therapy Bridge to recovery Bridge to transplantation LVAD device type HeartMate IIc Thoratec (unspecified)c HeartWare HVADd Heart Assist 5 (Debakey)e Jarvik 2000f DuraHeartg Type of TV procedure Reconstruction Without annuloplasty With annuloplasty Replacement Annuloplasty only Concomitant surgery
LVAD (n ¼ 1,608)
LVADþTVP (n ¼ 588)
0.1 (1) 7.7 (123) 67.7 (1,088) 24.4 (393)
0.3 (2) 9.0 (53) 70.4 (414) 20.2 (119)
95.09 (1,529) 2.2 (36) 2.7 (43) 88.8 ⫾ 40.2
98.0 (576) 1.5 (9) 0.5 (3) 125.0 ⫾ 42.2
0.1 (1) 6.3 (101) 92.6 (1,489) 47.8 ⫾ 40.5 (92)
0.2 (1) 9.9 (58) 89.8 (528) 59.6 ⫾ 40.6 (53)
34.0 (546) 1.7 (28) 64.3 (1,034)
39.8 (234) 2.9 (17) 57.3 (337)
93.8 (1,509) 0.9 (14) 2.9 (46) 0.3 (5) 2.1 (33) 0.1 (1)
95.6 (562) 1.4 (8) 2.7 (16) 0.0 (0) 0.3 (2) 0.0 (0)
NA NA NA NA 19.7 (316)
4.4 (26) 9.2 (54) 5.3 (31) 81.1 (477) 28.1 (165)
p-value 0.0670
0.0040
o0.0001 0.0144
0.0446 0.0059
0.0456
o0.0001
LVAD, left ventricular assist device; NA, not applicable; TV, tricuspid valve; VAD, ventricular assist device. a Continuous data are shown as the mean ⫾ standard deviation and categoric data as percentage (n). b Full cardiopulmonary bypass: cardiopulmonary bypass or coronary perfusion was used for the entire procedure; combination: with or without cardiopulmonary bypass and/or with or without coronary perfusion at any time during the procedure (captures conversions from off-pump to onpump only). c Thoratec, Pleasanton, California. d HeartWare International Inc, Framingham, Massachusetts. e MicroMed Technology, Houston, Texas. f Jarvik Heart Inc, New York, New York. g Terumo Heart Inc, Ann Arbor, Michigan.LVAD, left ventricular assist device; NA, not applicable; TV, tricuspid valve; VAD, ventricular assist device.
The Journal of Heart and Lung Transplantation, Vol ], No ], Month ]]]]
6
Operative outcomes RRs for TVP are presented for dichotomous variables before and after propensity score adjustment using IPW in Table 3. Performing a concomitant TVP did not result in any difference in operative mortality (IPW adjusted RR, 0.95; 95% CI, 0.68–1.33; p ¼ 0.7658) or subsequent RVAD insertion (IPW adjusted RR, 0.81; 95% CI, 0.49–1.36; p ¼ 0.4310). However, after IPW adjustment, concomitant TVP was associated with significantly higher risks for prolonged ICU stay (4 172 hours), hospital length of stay 4 3 weeks, total transfusion requirement, any post-operative complication, particularly prolonged ventilation, new renal failure and new dialysis requirement, and any reoperation. Relative risks for associations between TVP and post-operative transfusion requirement and reoperation for bleeding and/or tamponade were significant before but not after IPW adjustment. Table 4 summarizes treatment group analyses for continuous outcomes before and after adjustment using IPW. Total transfusion requirements were significantly higher for packed red blood cells, fresh frozen plasma, cryoprecipitate, and platelets in the setting of concomitant TVP. No differences in initial or total ICU hours or hospital length of stay were observed when the variables were analyzed as continuous variables rather than dichotomized above and below the median.
Our sensitivity analysis demonstrated that outcomes were similar when analyzed by traditional multivariate regression compared with IPW using conditional logistic regression (Table 5). Concomitant TVP resulted in significantly increased odds of any reoperation, reoperation for bleeding and/or tamponade, prolonged ventilation, and new renal failure. TVP continued to have no effect on operative mortality or post-operative RVAD requirement.
Discussion The primary finding of this study is that performing a concomitant TVP for LVAD patients with moderate to severe TR fails to reduce early mortality or subsequent RVAD requirement but significantly increases the risk for post-operative renal failure, greater transfusion requirement, reoperation, prolonged ventilation, prolonged ICU stay, and prolonged hospital stay. This is particularly relevant because concomitant TVPs are now used in 25% of patients with moderate to severe TR nationally and in up to 50% of this population in certain centers,2–4,6 even though the benefits of this approach have not been rigorously examined in large, multi-institutional studies. Our data suggest that additional criteria should be identified to select patients who may benefit from concomitant TVP if it is to become an effective strategy to reduce post-operative morbidity and mortality.
Table 3 Dichotomous Variable Risk Ratios for Tricuspid Valve Procedure Before and After Propensity Score Adjustment Using Inverse Probability Weighting Before IPW adjustment Variables Operative mortality ICU hoursa Initial 4 166 Total 4 172 LOS 4 21 days Any blood products Intra-/post-op Post-op Any complicationb Post-op complications Sternal infection Perm/embolic stroke Prolonged ventilation New renal failure New dialysis Reoperation Any Bleeding or tamponade RVAD insertion VAD complications Intracranial bleeding Device infection Device malfunction
After IPW adjustment
No. (events/total)
RR for TVP (95% CI)
221/2,161
1.07 (0.81–1.41)
0.6351
0.95 (0.68–1.33)
0.7658
1,026/2,084 1,045/2,084 1,049/2,152
1.14 (1.04–1.25) 1.16 (1.06–1.27) 1.33 (1.22–1.45)
0.0056 0.0015 o0.0001
1.14 (1.02–1.28) 1.14 (1.02–1.27) 1.29 (1.16–1.43)
0.0165 0.0210 o0.0001
2,018/2,108 1,749/2,110 1,764/2,164
1.03 (1.01–1.05) 1.06 (1.02–1.11) 1.13 (1.09–1.17)
0.0002 0.0027 o0.0001
1.03 (1.01–1.05) 1.04 (0.99–1.09) 1.12 (1.07–1.17)
0.0013 0.1221 o0.0001
17/2,154 67/2,153 1,442/2,154 212/1,822 155/2,088
1.12 0.71 1.13 1.39 1.65
(0.40–3.17) (0.40–1.27) (1.07–1.21) (1.06–1.81) (1.21–2.25)
0.8305 0.2508 o0.0001 0.0154 0.0015
1.31 0.65 1.10 1.53 1.49
699/2,154 345/2,154 98/2,161
1.23 (1.08–1.39) 1.28 (1.04–1.57) 1.02 (0.66–1.57)
23/2,092 29/2,092 21/2,092
0.57 (0.19–1.67) 1.22 (0.56–2.66) 0.84 (0.31–2.29)
p-value
RR for TVP (95% CI)
p-value
(0.43–3.99) (0.32–1.33) (1.02–1.18) (1.13–2.08) (1.03–2.15)
0.6298 0.2412 0.0131 0.0061 0.0339
0.0018 0.0201 0.9296
1.24 (1.07–1.45) 1.16 (0.89–1.51) 0.81 (0.49–1.36)
0.0056 0.2600 0.4310
0.3034 0.6232 0.7405
0.83 (0.24–2.93) 1.21 (0.46–3.16) 0.81 (0.28–2.31)
0.7753 0.6956 0.6933
CI, confidence interval; ICU, intensive care unit; IPW, inverse probability weighting; LOS, length of stay; RR, risk ratio; RVAD, right ventricular assist device; TVP, tricuspid valve procedure; VAD, ventricular assist device. a Median initial ICU stay is 166 hours; median total ICU stay is 172 hours. b Includes post-operative complications, reoperations, and VAD complications.
Robertson et al.
Concomitant TVP for TR in cfLVAD Patients
7
Table 4 Treatment Group Analysis for Continuous Outcome Variables Before and After Propensity Score Adjustment Using Inverse Probability Weighting Before IPW adjustment
Variable ICU hours Initial Total Hospital LOS, days Post-op transfusion Packed RBC, U FFP, U Cryoprecipitate, U Platelets, U Intra-/post-op transfusion Packed RBCs, U FFP, U Cryoprecipitate, U Platelets, U
After IPW adjustment
LVAD (mean)
LVADþTVP (mean)
Difference (mean)
258.76 291.09 25.29
283.00 311.98 28.47
24.25 20.89 3.19
0.1132 0.2128 0.0044
6.89 3.62 0.59 1.71
8.15 4.15 0.83 1.86
1.26 0.53 0.24 0.15
0.0104 0.0973 0.0411 0.4657
9.79 7.44 1.91 3.69
11.74 9.41 2.61 4.45
1.95 1.97 0.70 0.76
0.0008 o0.0001 0.0071 0.0063
p-value
LVAD (mean)
LVADþTVP (mean)
Difference (mean)
p-value
262.63 294.52 25.79
275.81 303.16 29.38
13.18 8.64 3.58
0.4099 0.6251 0.0835
6.921 3.662 0.59 1.71 9.90 7.53 1.95 3.71
7.935 4.119 0.91 1.76 11.45 9.14 2.77 4.49
1.01 0.46 0.31 0.044
0.1025 0.2365 0.0627 0.8266
1.55 1.61 0.82 0.78
0.0222 0.0024 0.0214 0.0156
FFP, fresh frozen plasma; ICU, intensive care unit; IPW, inverse probability weighting; LOS, length of stay; LVAD, left ventricular assist device; RBC, red blood cells; TVP, tricuspid valve procedure.
Prior analyses regarding the prognostic significance of preoperative moderate to severe TR have resulted in varied conclusions,10–12 perhaps partly because of the interplay between native valvular disease and physiologic changes that occur with the initiation of mechanical support. Evidence has shown that implantation of an LVAD by itself can improve TR in some patients by reducing secondarily elevated pulmonary artery pressures.6,10,13,17 Reports suggest that as many as 50% of patients will see improvements in TR post-operatively,17 but in contrast to the performance of a tricuspid ring annuloplasty, LVAD insertion alone does not seem to produce uniform improvement in the degree of TR and, in some cases, it may be a gradual process over a period of months.3,18,19 The TR in another sub-set of patients will stay the same or worsen post-operatively. Acute worsening of TR can
occur from distortion of the tricuspid annulus due to leftward deviation of the intraventricular septum secondary to increased flow after LVAD placement. Significant volume-loading and transfusion requirements that are often necessary in the post-operative period may also exacerbate TR when flows are limited by output across the pulmonary vascular bed.20 Moreover, LVAD patients commonly have implantable cardioverter defibrillator leads traversing the tricuspid valve that may entrap the TV leaflets and lead to an anatomic cause for persistence of TR. Complicating matters, there is evidence that the absence of severe TR pre-operatively does not imply that the tricuspid orifice is free of disease, and the presence of annular dilation could represent a significant substrate for the later development of TR.21 Therefore, because the
Table 5 Sensitivity Analysis Comparing Results from Traditional Multivariate Regression Analysis to Inverse Probability Weighting Using Conditional Logistic Regression Before IPW adjustment
After IPW adjustment
Variables
No. (events/total)
OR for TVP (95% CI)
p-value
OR for TVP (95% CI)
p-value
Operative mortality Reoperation Any Bleeding or tamponade RVAD insertion Prolonged ventilation New renal failure Stroke
224/2,196
1.22 (0.83–1.80)
0.3050
0.95 (0.61–1.47)
0.8177
710/2,189 350/2,189 99/2,196 1,462/2,189 214/1,851 67/2,188
1.39 1.67 0.94 1.48 1.66 1.25
0.0099 0.0012 0.8170 0.0039 0.0121 0.5227
1.46 1.93 0.69 1.40 1.93 1.22
0.0076 0.0002 0.2631 0.0262 0.0002 0.6045
(1.08–1.79) (1.22–2.27) (0.54–1.64) (1.13–1.940) (1.12–2.48) (0.63–2.50)
(1.11–1.93) (1.37–2.72) (0.36–1.32) (1.04–1.89) (1.37–2.72) (0.57–2.62)
CI, confidence interval; IPW, inverse probability weighting; LVAD, left ventricular assist device; OR, odds ratio; RVAD, right ventricular assist device; TVP, tricuspid valve procedure. Variables in adjusted model: atrial fibrillation, age, gender, body surface area, congestive heart failure, New York Heart Association class IV, chronic lung disease, creatinine, cerebrovascular disease, cardiovascular disease, diabetes without insulin, dialysis, hypertension, intraaortic balloon pump, inotropes, immunosuppression, left main disease; NDV, number of diseased vessels o 21 days, unstable angina, peripheral vascular disease; race, reoperation (first, Z 2), cardiogenic shock, status urgent/emergent salvage, year of surgery, mitral valve insufficiency moderate/severe, aortic valve insufficiency moderate/severe tricuspid insufficiency severe, concomitant procedures other than valve. Interaction terms: age reoperation, age status, gender body surface area.
8
The Journal of Heart and Lung Transplantation, Vol ], No ], Month ]]]]
degree of pre-operative TR often changes after implantation of a cfLVAD, intervening based on this criterion alone will result in many unnecessarily performed concomitant TVPs, or, in some patients, failure to operate when it may have been beneficial. None of the currently available studies have reported a significant early or late survival benefit to concomitant TVP.2–7,9 The current report supports these data with respect to early mortality. This could possibly reflect the effectiveness of medical treatment of RV failure with pulmonary vasodilators and inotropic agents. Another explanation, however, is that, for the reasons discussed, patient selection for concomitant TVP failed to identify patients who would have had significant post-operative TR and exacerbations of RV failure. With respect to improvements in right heart function after a LVADþTVP in patients with significant TR, Piacentino et al6,7 showed reductions in RV failure, defined as postoperative inotrope duration 4 14 days or RVAD insertion. One of their 2 studies demonstrated a significant decrease in the incidence of post-operative RVAD requirement with concomitant TVP of 21% (6 of 28) vs 3% (1 of 33; p o 0.05), but other groups have failed to replicate these findings.2–4 In particular, the 21% incidence of RVAD insertion seems high compared with other reports and could have been artificially inflated by small sample sizes4,12,22 and might also reflect different strategies for pre-operative optimization of patients with significant TR and/or preoperative RV failure. Our study found the incidence of reoperation for RVAD implantation reported nationally for patients with moderate to severe TR was only 4.5% (98 of 2,161) and was not affected by performing a concomitant TVP. Unfortunately, we cannot comment on the duration of post-operative inotrope used in the present study. Moreover, concern exists that prolonged back pressure from non-regression or progression of TR could lead to renal or hepatic congestion with resultant coagulopathy. Of multiple studies addressing these issues,2,4,7,9 1 study demonstrated a lower incidence of post-operative renal dysfunction (39% vs 21%, p ¼ 0.05).7 Surprisingly, our data show that performance of a concomitant TVP is significantly associated with an increased risk for new postoperative renal failure and dialysis. Transfusion requirement in our study is also significantly increased for the concomitant TVP group. A study by Saeed et al9 that compared LVAD (n ¼ 42) with LVADþTVP (n ¼ 8) for patients with moderate to severe TR is consistent with this latter finding, demonstrating a higher median number of packed red blood cells transfused for the group that received a concomitant TVP (4 vs 13 units; p ¼ 0.03). Although these differences could possibly reflect the effects of the right heart on end-organ function, it is more likely that the higher incidence of renal failure and the greater transfusion requirement are consequences of prolonged time on CPB and longer cross-clamp times. The necessary addition of an atriotomy, additional dissection, and the need for bicaval cannulation in the LVADþTVP group may also have affected bleeding risk.
The present study has further shown that patients who undergo a LVADþTVP are at increased risk for significantly prolonged ventilation, longer ICU stays, longer hospital stays, and reoperation. This is in contrast to a single-institution study that demonstrated shortened hospital length of stay with concomitant TVP (26 vs 19 days; p ¼ 0.02). No other study has detected a difference with respect to hospital stay, ICU stay, or ventilator duration.2–4,9 Given the wide variation in these parameters between patients, previous studies were underpowered to adequately detect differences. The finding of increased risk of reoperation, particularly for bleeding and tamponade, is novel. The reasons for these data are uncertain, but the role of prolonged time on CPB and the addition of an atriotomy are unlikely to account for all of the observed differences in the treatment groups. The relative effects of these factors vs the potential benefit of an immediate reduction in central venous pressure are not known. Although the results do not support performing concomitant TVPs in cfLVAD patients with moderate to severe TR, several important questions remain. First, although this STS database data is able to address peri-operative outcomes, we are unable to assess long-term differences in mortality or morbidity between groups. Analyses using the Interagency Registry for Mechanically Assisted Circulatory Support database are ongoing and will help elucidate longterm outcomes. Still, for this important clinical question, it remains necessary to validate results with different approaches in multiple large registries to best address the absence of randomized data. Second, some authors have suggested that parameters such as annular dilation, annular motion, and geometry of the RV are predictors of post-operative RV failure. Whether selection of candidates for concomitant TVP by one or a combination of these parameters would improve outcomes over selection according to the degree of pre-operative TR alone remains unclear. Third, we do not know whether a concomitant TVP in patients requiring temporary RVAD support would facilitate weaning in patients with moderate to severe TR and/or other echocardiographic parameters of annular disease. This study also has other limitations. First, we do not have specific data on the temporal relationship between surgery and the pre-operative determination of TR severity and do not know the loading conditions or clinical status of the patients at the time it was measured. Second, preoperative and post-operative hemodynamic data in the groups are insufficient to eliminate potential unmatched treatment bias. We are unable to measure the effect of preoperative intrinsic RV function or its interaction with TR and any subsequent effect on post-operative RV failure. Further, we are unable to assess inotrope usage or hemodynamic indices of RV failure post-operatively. However, in the absence of negative effects on other measured variables in the LVAD-alone group, the duration of inotrope usage has diminishing significance unless it is prolonged by concomitant TVP. Third, serial echocardiographic studies are not available, so we are unable to comment on how the TR changes over time in these patients, with or
Robertson et al.
Concomitant TVP for TR in cfLVAD Patients
without TVP. Available evidence suggests that even with a concomitant TVP, as a many as 10% of patients may continue to have significant TR after insertion of a cfLVAD,6,23 and, as discussed, insertion of an LVAD alone can lead to significant post-operative TR in certain patients with insignificant preoperative TR. This suggests that some fraction of patients in both groups likely had significant post-operative TR, but how this affects outcomes is uncertain. Lastly, there are insufficient data to examine the potential effect of aortic cross-clamp on post-operative morbidity and mortality. However, this study is substantially larger than any previous report on this question and is a multi-institutional study using a prospectively collected database. As such, this report eliminates single-center bias and has greater power to detect significant differences between groups independent of center bias and surgeon preference. In conclusion, this study examined a real-world, multiinstitutional cohort of cfLVAD patients who had moderate to severe pre-operative TR and compared patients who underwent implantation of a LVAD alone with those who had a LVADþTVP. After matching patient cohorts by IPW, concomitant TVP did not improve the risk of short-term mortality or RVAD insertion and was associated with increased risk for prolonged ventilation, prolonged ICU and hospital stays, post-operative renal failure/dialysis, transfusion, and reoperation. As such, these data suggest that the practice of performing a concomitant TVP based on the degree of pre-operative TR alone should be reexamined. Although concomitant TVP might benefit some patients, the data suggest that routine TV intervention in cfLVAD patients may impart additional short-term morbidity for the population as a whole. The medium-term and longer-term benefits remain to be defined. The strategy of performing a concomitant TVP may still be beneficial if a more appropriate method of patient selection is identified. Owing to the demonstrated short-term morbidity of concomitant TVP in this study, further data are required to determine the magnitude and timing of any potential benefit as well as criteria for selection of candidates for TVP.
Disclosure statement Funding for statistical support was provided by the STS Database after review and acceptance of our research proposal. Dr. Silvestry is a consultant for Thoratec and HeartWare. None of the other authors has a financial relationship with a commercial entity that has an interest in the subject of the presented manuscript or other conflicts of interest to disclose.
References 1. Kormos RL, Teuteberg JJ, Pagani FD, et al. Right ventricular failure in patients with the HeartMate II continuous-flow left ventricular assist device: incidence, risk factors, and effect on outcomes. J Thorac Cardiovasc Surg 2010;139:1316-24. 2. Koeckert M, Comas G, Yang J, et al. Tricuspid valve repair during left ventricular assist device insertion. J Heart Lung Transplant 2011;30:S43-4.
9 3. Krishan K, Nair A, Pinney S, Adams DH, Anyanwu AC. Liberal use of tricuspid-valve annuloplasty during left-ventricular assist device implantation. Eur J Cardiothorac Surg 2012;41:213-7. 4. Maltais S, Topilsky Y, Tchantchaleishvili V, et al. Surgical treatment of tricuspid valve insufficiency promotes early reverse remodeling in patients with axial-flow left ventricular assist devices. J Thorac Cardiovasc Surg 2012;143:1370-6. 5. Pal JD, Klodell CT, John R, et al. Low operative mortality with implantation of a continuous-flow left ventricular assist device and impact of concurrent cardiac procedures. Circulation 2009;120: S215-9. 6. Piacentino V 3rd, Ganapathi AM, Stafford-Smith M, et al. Utility of concomitant tricuspid valve procedures for patients undergoing implantation of a continuous-flow left ventricular device. J Thorac Cardiovasc Surg 2012;144:1217-21. 7. Piacentino V 3rd, Troupes CD, Ganapathi AM, et al. Clinical impact of concomitant tricuspid valve procedures during left ventricular assist device implantation. Ann Thorac Surg 2011;92:1414-8; discussion: 1418–9. 8. Potapov EV, Schweiger M, Stepanenko A, et al. Tricuspid valve repair in patients supported with left ventricular assist devices. ASAIO J 2011;57:363-7. 9. Saeed D, Kidambi T, Shalli S, et al. Tricuspid valve repair with left ventricular assist device implantation: is it warranted? J Heart Lung Transplant 2011;30:530-5. 10. Piacentino V 3rd, Williams ML, Depp T, et al. Impact of tricuspid valve regurgitation in patients treated with implantable left ventricular assist devices. Ann Thorac Surg 2011;91:1342-6; discussion: 1346–7. 11. Potapov EV, Stepanenko A, Dandel M, et al. Tricuspid incompetence and geometry of the right ventricle as predictors of right ventricular function after implantation of a left ventricular assist device. J Heart Lung Transplant 2008;27:1275-81. 12. Puwanant S, Hamilton KK, Klodell CT, et al. Tricuspid annular motion as a predictor of severe right ventricular failure after left ventricular assist device implantation. J Heart Lung Transplant 2008;27:1102-7. 13. Lee S, Kamdar F, Madlon-Kay R, et al. Effects of the HeartMate II continuous-flow left ventricular assist device on right ventricular function. J Heart Lung Transplant 2010;29:209-15. 14. O’Brien SM, Shahian DM, Filardo G, et al. The Society of Thoracic Surgeons 2008 cardiac surgery risk models: part 2—isolated valve surgery. Ann Thorac Surg 2009;88:S23–42. 15. White H. Maximum likelihood estimation of misspecified models. Econometrica 1982;50:1-25. 16. Localio AR, Berlin JA, Ten Have TR, Kimmel SE. Adjustments for center in multicenter studies: an overview. Ann Intern Med 2001;135:112-23. 17. Morgan JA, Paone G, Nemeh HW, et al. Impact of continuous-flow left ventricular assist device support on right ventricular function. J Heart Lung Transplant 2013;32:398-403. 18. McCarthy PM, Bhudia SK, Rajeswaran J, et al. Tricuspid valve repair: durability and risk factors for failure. J Thorac Cardiovasc Surg 2004;127:674-85. 19. Rao V, Slater JP, Edwards NM, Naka Y, Oz MC. Surgical management of valvular disease in patients requiring left ventricular assist device support. Ann Thorac Surg 2001;71:1448-53. 20. Chumnanvej S, Wood MJ, MacGillivray TE, Melo MF. Perioperative echocardiographic examination for ventricular assist device implantation. Anesth Analg 2007;105:583-601. 21. Dreyfus GD, Corbi PJ, Chan KM, Bahrami T. Secondary tricuspid regurgitation or dilatation: which should be the criteria for surgical repair? Ann Thorac Surg 2005;79:127-32. 22. Kirklin JK, Naftel DC, Pagani FD, et al. Long-term mechanical circulatory support (destination therapy): on track to compete with heart transplantation? J Thorac Cardiovasc Surg 2012;144:584-603; discussion: 597–8. 23. Deo SV, Hasin T, Altarabsheh SE, et al. Concomitant tricuspid valve repair or replacement during left ventricular assist device implant demonstrates comparable outcomes in the long term. J Card Surg 2012;27:760-6.
ORIGINAL CLINICAL SCIENCE
Concomitant tricuspid valve surgery during implantation of continuous-flow left ventricular assist devices: A Society of Thoracic Surgeons database analysis Jason O. Robertson, MD, MS,a Maria V. Grau-Sepulveda, MD,b Shoichi Okada, MD,a Sean M. O’Brien, PhD,b J. Matthew Brennan, MD, MPH,b Ashish S. Shah, MD,c Akinobu Itoh, MD,a Ralph J. Damiano, MD,a Sunil Prasad, MD,a and Scott C. Silvestry, MDa From the aDivision of Cardiothoracic Surgery, Washington University, St. Louis, Missouri; bDuke Clinical Research Institute, Duke University, Durham, North Carolina; and the cDivision of Cardiac Surgery, Johns Hopkins University, Baltimore, Maryland.
KEYWORDS: Left ventricular assist device; tricuspid; tricuspid regurgitation
BACKGROUND: Performing concomitant tricuspid valve procedures (TVPs) in left ventricular assist device (LVAD) patients with significant pre-operative tricuspid regurgitation (TR) is controversial, and no studies have been large enough to definitively guide therapy. METHODS: Between January 2006 and September 2012, 2,196 patients with moderate to severe preoperative TR from 115 institutions underwent implantation of a continuous-flow left ventricular assist device (LVAD) as reported by The Society of Thoracic Surgeons National Database. Of these, 588 (27%) underwent a concomitant TVP. Inverse probability weighting based on propensity score was used to adjust for differences between the LVAD alone and LVADþTVP groups, and outcomes were compared. RESULTS: Most patients in the LVADþTVP group underwent an annuloplasty alone (81.1%). Concomitant TVP did not affect risk of post-operative right VAD insertion (risk ratio [RR], 0.81; 95% confidence interval [CI], 0.49–1.36; p ¼ 0.4310) or death (RR, 0.95; 95% CI, 0.68–1.33; p ¼ 0.7658). However, TVP was associated with an increased risk for post-operative renal failure (RR, 1.53; 95% CI, 1.13–2.08; p ¼ 0.0061), dialysis (RR, 1.49; 95% CI, 1.03–2.15; p ¼ 0.0339), reoperation (RR, 1.24; 95% CI, 1.07–1.45; p ¼ 0.0056), greater total transfusion requirement (RR, 1.03; 95% CI, 1.01–1.05; p ¼ 0.0013), and hospital length of stay 4 21 days (RR, 1.29; 95% CI, 1.16–1.43; p o 0.0001). Time on the ventilator and intensive care unit length of stay were also significantly prolonged for the LVADþTVP group. CONCLUSIONS: Performing a concomitant TVP for continuous-flow LVAD patients with moderate to severe TR did not reduce early death or right VAD requirement and was associated with worse early postoperative outcomes. These data caution against routine concomitant TVP based solely on degree of preoperative TR and suggest that additional selection criteria are needed to identify those patients in whom concomitant TVP may prevent post-operative right ventricular failure. J Heart Lung Transplant ]]]];]:]]]–]]] r 2014 International Society for Heart and Lung Transplantation. All rights reserved.
Reprint requests: Scott C. Silvestry, MD, Surgical Director, Heart Transplant and Mechanical Cardiac Support Program, Washington University School of Medicine/Barnes-Jewish Hospital, 660 S Euclid Ave, Box 8234, St. Louis, MO 63110. Telephone: 314-362-6182. Fax: 314747-4216. E-mail address: [email protected]
Patients referred for management with left ventricular assist devices (LVADs) often have some degree of preoperative right-sided heart failure and frequently present with significant tricuspid regurgitation (TR). Some of these patients have right ventricular (RV) failure that persists
1053-2498/$ - see front matter r 2014 International Society for Heart and Lung Transplantation. All rights reserved. http://dx.doi.org/10.1016/j.healun.2014.01.861
2
The Journal of Heart and Lung Transplantation, Vol ], No ], Month ]]]]
post-operatively. Despite improvements with continuousflow LVADs (cfLVADs), post-operative RV dysfunction complicates up to 22% of cases and is associated with significant increases in morbidity and mortality.1 A number of authors have advocated performing TV repair or replacement in the presence of moderate to severe TR as a strategy to reduce post-operative RV failure. The rationale for performing a concomitant tricuspid valve procedure (TVP) during LVAD insertion in patients with significant TR is sound, but the currently available data are limited to reports from relatively small, retrospective, single-institution studies that have been underpowered to address the issue.2–9 Moreover, reports on the prognostic significance of pre-operative TR have produced mixed conclusions, likely due to changes in right heart hemodynamics that occur with insertion of a cfLVAD.10–12 Some have argued that a sub-set of patients with TR will improve from mechanical unloading with an LVAD alone, thereby negating their need for a concomitant TVP.13 As a result, the decision to perform concomitant TVPs remains controversial. This study used The Society for Thoracic Surgeons (STS) Adult Cardiac Surgery Database to address how concomitant TV repair or replacement at the time of LVAD implantation affects short-term outcomes for patients with cfLVADs. This analysis of TVP in a large number of patients with moderate to severe TR undergoing cfLVAD implantation should provide insight about the utility of performing concomitant TV procedures according to the degree of pre-operative TR.
surface area); medical history and risk factors (i.e., diabetes, hypertension, cardiovascular disease, cerebrovascular accident, peripheral vascular disease, dialysis, ejection fraction, last creatinine measurement, immunosuppression, chronic lung disease); cardiac status at the time of operation (i.e., myocardial infarction [MI] o 21 days, unstable angina without MI, cardiogenic shock, atrial fibrillation, congestive heart failure/New York Heart Association class IV); presence of a pre-operative intraaortic balloon pump (IABP) or inotrope infusions; degree of aortic, mitral, and tricuspid valvular insufficiency; pre-operative coronary catheterization data (number of diseased vessels, Z 50% left main disease) and operative variables (i.e., reoperation, other concomitant procedures, categorization of operative urgency, and year of surgery). We also included multiplicative interaction terms between age and reoperation, age and operative status, and sex and body surface area. Missing categoric variables were imputed to the most common category, and missing continuous variables were imputed by
Methods Approval for this study was granted by the STS Database Access and Publications Committee.
Patient population The study population consisted of patients from the STS Adult Cardiac Surgery Database who underwent insertion of a cfLVAD that was indicated for bridge to recovery, bridge to transplant, or destination therapy between January 2006 and September 2012 (Figure 1). Exclusions included patients (1) with a prior VAD, (2) with an isolated right VAD (RVAD) implantation, (3) for whom pre-operative TR data were missing, (4) with no, trivial, or mild TR, (5) who had a concomitant aortic or mitral valve procedure, (6) who received a heart transplant at the time of initial LVAD insertion, or (7) with active endocarditis. This resulted in 2,196 cfLVAD patients with moderate to severe TR, and concomitant TV repair or replacement was performed in 588.
Adjustments for between-group differences Propensity scores to estimate the probability of undergoing a TVP were developed with the use of logistic regression to adjust for between-group differences in baseline characteristics of the patients and hospitals. Variables included in the model were based on previously published valve models and clinical relevance.14 These included annual volume of LVAD implantations; annual volume of TVPs; patient demographics (i.e., age, race, sex, body
Figure 1 Study population. AV, aortic valve; BTR, bridge to recovery; BTT, bridge to transplant; cfLVAD, continuous-flow left ventricular assist device; DT, destination therapy; MV, mitral valve; RVAD, right ventricular assist device; TR, tricuspid regurgitation; TVP, tricuspid valve procedure; VAD, ventricular assist device.
Robertson et al.
Concomitant TVP for TR in cfLVAD Patients
stratifying patients according to related risk factors and then imputing stratum-specific medians. Missing data were rare (o 0.5% for most covariates), but for any variable with 4 5% missing data, we confirmed that the p-value comparing the percentage of missing data between groups was non-significant. Trimming was performed by excluding 35 observations with extreme propensities that did not overlap with the other treatment group. Summaries of the propensity distributions after trimming (minimum, 25th percentile, median, 75th percentile, and maximum) were similar in each group (no TVP: 3.8%, 10.7%, 16.8%, 30.1%, 91.0%; and TVP: 3.8%, 21.0%, 39.4%, 56.4%, 91.1%). Inverse probability weighting (IPW) according to the propensity score was used to adjust for differences between the 2 treatment groups. An advantage of this approach over propensity matching is that all observations are retained in the analysis, and it is, therefore, relatively efficient. In contrast, propensity matching would result in discarding large numbers of observations and would be less efficient. IPW involved weighting each patient who underwent a TVP by the inverse of the probability of being selected for a TVP and weighting each patient who did not undergo a TVP by the inverse of the probability of not being selected for a TVP. The ability of the propensity model to balance the 2 treatment groups was assessed by comparing the distribution of baseline characteristics before and after weighting by the inverse of the propensity score. Because the standardized differences after weighting for all variables were o |10| (or 0.1 standard deviation), we confidently assumed that balance was achieved between groups. The distribution of standardized differences before and after IPW is shown in Figure 2.
Statistics Summary statistics are presented as percentages for categoric variables and and as mean ⫾ standard deviation for continuous variables. Baseline characteristics were compared for patients between groups with the Pearson chi-square test for categoric variables and with the chi-square rank-based group means score statistics, which is equivalent to the Kruskal-Wallis and Wilcoxon tests, for continuous variables. To compare dichotomous outcomes between treatment groups, before and after IPW adjustment, we calculated risk ratios (RR) using generalized linear models with a log-link function.
Figure 2
3 The Huber-White variance estimator was used to compute 95% confidence intervals (CIs) and p-values.15 Continuous outcomes were similarly compared between groups, before and after IPW adjustment, using Wald tests with an empirical (Huber-White) variance estimator. Outcomes were only computed for patients with non-missing values. The primary outcomes considered were operative mortality (death during the same hospitalization as surgery, regardless of timing, or r 30 days of surgery, regardless of venue) and reoperation for RVAD insertion, which serves as proxy evidence of severe post-operative right-sided heart failure. New renal failure/dialysis, hospital length of stay (continuous and dichotomized at 4 21 days), intensive care unit (ICU) length of stay (continuous and dichotomized at the median, 4 172 hours for the overall stay), prolonged ventilation (4 24 hours), transfusion requirements, stroke, other causes for reoperation, and VAD complications (intracranial bleeding, device infection, and device malfunction) were also compared between groups to assess potential differences in operative morbidity and clinical course. Values of p o0.05 were considered significant. We did not adjust for multiple comparisons, and this should be taken into account when interpreting our results, because the probability of finding significant associations by chance increases with the number of outcomes analyzed. For example, if 20 independent hypotheses were to be considered, the probability of observing at least 1 p-value of o 0.05 by chance alone would be 64%. A sensitivity analysis was performed to compare results from the IPW approach with results from traditional multivariable regression analysis for the most relevant outcomes. Specifically, conditional logistic regression was used to compute odds ratios for TVP and each pertinent dichotomous outcome, and the model was stratified by hospital identification to account for potential confounding by center effects.16 Covariates for each model were identical to those used in the propensity model.
Results Baseline, clinical, and operative characteristics Baseline patient demographic and clinical characteristics are presented in Table 1. The LVADþTVP group comprised
The distribution of standardized differences before and after inverse probability weighting (IPW).
The Journal of Heart and Lung Transplantation, Vol ], No ], Month ]]]]
4 Table 1
Baseline Patient Demographic and Clinical Characteristics
Variablea Age, years Caucasian Body surface area, m2 Female NYHA class IV Ejection fraction, % Pre-op creatinine 4 2.0 mg/dl Renal failure Dialysis No dialysis Hypertension Dyslipidemia Diabetes mellitus Non–insulin-dependent Insulin-dependent Cerebrovascular accident Number diseased vessels 3 2 1 0 Left main disease Peripheral vascular disease Immunosuppression Atrial fibrillation Myocardial infarction Chronic lung disease (mod/severe) Cardiac reoperation Previous valve surgery Tricuspid insufficiency (severe) Mitral insufficiency (mod/severe) Aortic insufficiency (mod/severe) Cardiogenic shock Resuscitation Pre-op inotropes Pre-op intraaortic balloon pump Cardiac presentation STEMI NSTEMI Unstable angina Stable angina Symptoms unlikely to be ischemia No symptoms or angina Year 2006–2008 2009–2010 2011–2012
LVAD (n ¼ 1,608)
LVADþTVP (n ¼ 588)
55.8 ⫾ 13.5 66.4 (1,067) 2.0 ⫾ 0.2 25.3 (406) 82.3 (1,324) 16.3 ⫾ 6.8 12.3 (198)
55.2 ⫾ 13.7 61.9 (364) 2.0 ⫾ 0.2 27.4 (161) 80.8 (475) 16.2 ⫾ 6.5 16.3 (96)
3.4 (55) 10.7 (172) 60.1 (967) 61.9 (996)
3.4 (20) 14.8 (87) 55.1 (324) 57.7 (339)
17.2 (277) 21.4 (344) 9.5 (153)
18.2 (107) 20.4 (120) 8.2 (48)
18.4 (295) 6.9 (111) 6.3 (102) 61.0 (981) 4.5 (72) 11.1 (178) 4.2 (68) 64.2 (1,032) 36.9 (594) 18.8 (303) 39.3 (632) 7.7 (124) 24.6 (395) 76.6 (1,232) 2.9 (47) 29.9 (480) 1.7 (27) 77.1 (1,239) 30.5 (491)
13.4 (79) 6.5 (38) 6.3 (37) 64.0 (376) 2.2 (13) 7.1 (42) 3.6 (21) 64.1 (377) 31.3 (184) 17.5 (103) 37.9 (223) 9.0 (53) 56.1 (330) 69.4 (408) 1.4 (8) 30.1 (177) 1.9 (11) 76.7 (451) 41.3 (243)
0.8 (12) 1.4 (22) 4.6 (70) 4.6 (70) 50.4 (770) 32.7 (499)
0.4 (2) 0.7 (4) 3.6 (20) 4.3 (24) 43.0 (241) 35.0 (196)
13.9 (223) 42.2 (679) 43.9 (706)
12.9 (76) 36.6 (215) 50.5 (297)
p-value 0.4897 0.4021 0.4316 0.3122 0.4527 0.9206 0.0132 0.0280
0.0261 0.0469 0.8040
0.3283 0.0851
0.0182 0.0060 0.4855 0.9467 0.0136 0.4877 0.5438 0.3230 o0.0001 0.0287 0.0556 0.9497 0.7685 0.9219 o0.0001 0.2245
0.0204
LVAD, left ventricular assist device; mod, moderate; NSTEMI, non-ST elevation myocardial infarction; NYHA, New York Heart Association; STEMI, ST elevation myocardia infarction; TVP, tricuspid valve procedure. a Continuous data are presented as mean ⫾ standard deviation and categoric data as percentage (n).
26.8% of the final study population. Those that received a TVP more commonly had severe TR (56.1% vs 24.6%, p o 0.0001) and had significantly higher incidences of pre-operative renal failure not requiring dialysis (14.8% vs 10.7%, p ¼ 0.0280) and IABP use (41.3% vs 30.5%, p o 0.0001). Conversely, patients receiving a TVP had significantly lower
incidences of hypertension, dyslipidemia, prior MI, coronary artery disease, moderate to severe mitral insufficiency, and peripheral vascular disease. More than 85% of the patients in this study are from 2009 onward, and the number of TVP cases increased in the years 2011 to 2012, with 50.5% of LVADþTVP study patients from that period.
Robertson et al.
Concomitant TVP for TR in cfLVAD Patients
Table 2 compares the operative characteristics between the groups. Most LVADs in both groups were HeartMate II devices (Thoratec, Pleasanton, CA) inserted as a bridge to transplantation. Patients that received a TVP had an 3% higher incidence of full cardiopulmonary bypass (CPB), defined as the need for CPB or coronary perfusion for the entire procedure, and an average increase in perfusion time of 36 minutes (125.0 ⫾ 42.2 vs 88.8 ⫾ 40.2 minutes; p o 0.0001). Most patients in both groups did not require aortic occlusion, but it was more commonly used for those undergoing an LVADþTVP (9.9% vs 6.3%; p ¼ 0.0144), and cross-clamp times were prolonged by an average of 12 minutes. Concomitant surgeries other than mitral, aortic, and TV operations were also more commonly performed in the LVADþTVP group (28.1% vs 19.7%; p o 0.0001). Table 2
Variable
5 These included coronary artery bypass grafting in 0.9% (5 of 588) vs 1.2% (19 of 1,608), pulmonic valve procedures in 0.2% (1 of 588) vs 0.0% (0 of 1,608), LV aneurysm repairs in 0.2% (1 of 588) vs 0.1% (1 of 1,608), ventricular septal defect repairs in 0.0% (0 of 588) vs 0.2% (3 of 1,608), atrial septal defect repairs in 12.2% (72 of 588) vs 3.4% (55 of 1,608), other congenital defect repairs in 1.9% (11 of 588) vs 0.8% (13 of 1,608), ablation of atrial fibrillation in 0.2% (12 of 588) vs 0.01% (12 of 1,608), aortic aneurysm and/or dissection repairs in 0.0% (0 of 588) vs 0.003% (5 of 1,606), other cardiac procedures in 15.7% (92 of 588) vs 12.6% (202 of 1,607), and other non-cardiac procedures in 1.5% (9 of 588) vs 2.0% (32 of 1,608). Among those who received a TVP, annuloplasty was performed 81.1% of the time.
Operative Characteristics a
Status Emergent salvage Emergent Urgent Elective Cardiopulmonary bypassb Full Combination None Perfusion time, min Aortic occlusion Balloon occlusion Aortic cross-clamp None/partial cross-clamp Cross-clamp time, min VAD indication Destination therapy Bridge to recovery Bridge to transplantation LVAD device type HeartMate IIc Thoratec (unspecified)c HeartWare HVADd Heart Assist 5 (Debakey)e Jarvik 2000f DuraHeartg Type of TV procedure Reconstruction Without annuloplasty With annuloplasty Replacement Annuloplasty only Concomitant surgery
LVAD (n ¼ 1,608)
LVADþTVP (n ¼ 588)
0.1 (1) 7.7 (123) 67.7 (1,088) 24.4 (393)
0.3 (2) 9.0 (53) 70.4 (414) 20.2 (119)
95.09 (1,529) 2.2 (36) 2.7 (43) 88.8 ⫾ 40.2
98.0 (576) 1.5 (9) 0.5 (3) 125.0 ⫾ 42.2
0.1 (1) 6.3 (101) 92.6 (1,489) 47.8 ⫾ 40.5 (92)
0.2 (1) 9.9 (58) 89.8 (528) 59.6 ⫾ 40.6 (53)
34.0 (546) 1.7 (28) 64.3 (1,034)
39.8 (234) 2.9 (17) 57.3 (337)
93.8 (1,509) 0.9 (14) 2.9 (46) 0.3 (5) 2.1 (33) 0.1 (1)
95.6 (562) 1.4 (8) 2.7 (16) 0.0 (0) 0.3 (2) 0.0 (0)
NA NA NA NA 19.7 (316)
4.4 (26) 9.2 (54) 5.3 (31) 81.1 (477) 28.1 (165)
p-value 0.0670
0.0040
o0.0001 0.0144
0.0446 0.0059
0.0456
o0.0001
LVAD, left ventricular assist device; NA, not applicable; TV, tricuspid valve; VAD, ventricular assist device. a Continuous data are shown as the mean ⫾ standard deviation and categoric data as percentage (n). b Full cardiopulmonary bypass: cardiopulmonary bypass or coronary perfusion was used for the entire procedure; combination: with or without cardiopulmonary bypass and/or with or without coronary perfusion at any time during the procedure (captures conversions from off-pump to onpump only). c Thoratec, Pleasanton, California. d HeartWare International Inc, Framingham, Massachusetts. e MicroMed Technology, Houston, Texas. f Jarvik Heart Inc, New York, New York. g Terumo Heart Inc, Ann Arbor, Michigan.LVAD, left ventricular assist device; NA, not applicable; TV, tricuspid valve; VAD, ventricular assist device.
The Journal of Heart and Lung Transplantation, Vol ], No ], Month ]]]]
6
Operative outcomes RRs for TVP are presented for dichotomous variables before and after propensity score adjustment using IPW in Table 3. Performing a concomitant TVP did not result in any difference in operative mortality (IPW adjusted RR, 0.95; 95% CI, 0.68–1.33; p ¼ 0.7658) or subsequent RVAD insertion (IPW adjusted RR, 0.81; 95% CI, 0.49–1.36; p ¼ 0.4310). However, after IPW adjustment, concomitant TVP was associated with significantly higher risks for prolonged ICU stay (4 172 hours), hospital length of stay 4 3 weeks, total transfusion requirement, any post-operative complication, particularly prolonged ventilation, new renal failure and new dialysis requirement, and any reoperation. Relative risks for associations between TVP and post-operative transfusion requirement and reoperation for bleeding and/or tamponade were significant before but not after IPW adjustment. Table 4 summarizes treatment group analyses for continuous outcomes before and after adjustment using IPW. Total transfusion requirements were significantly higher for packed red blood cells, fresh frozen plasma, cryoprecipitate, and platelets in the setting of concomitant TVP. No differences in initial or total ICU hours or hospital length of stay were observed when the variables were analyzed as continuous variables rather than dichotomized above and below the median.
Our sensitivity analysis demonstrated that outcomes were similar when analyzed by traditional multivariate regression compared with IPW using conditional logistic regression (Table 5). Concomitant TVP resulted in significantly increased odds of any reoperation, reoperation for bleeding and/or tamponade, prolonged ventilation, and new renal failure. TVP continued to have no effect on operative mortality or post-operative RVAD requirement.
Discussion The primary finding of this study is that performing a concomitant TVP for LVAD patients with moderate to severe TR fails to reduce early mortality or subsequent RVAD requirement but significantly increases the risk for post-operative renal failure, greater transfusion requirement, reoperation, prolonged ventilation, prolonged ICU stay, and prolonged hospital stay. This is particularly relevant because concomitant TVPs are now used in 25% of patients with moderate to severe TR nationally and in up to 50% of this population in certain centers,2–4,6 even though the benefits of this approach have not been rigorously examined in large, multi-institutional studies. Our data suggest that additional criteria should be identified to select patients who may benefit from concomitant TVP if it is to become an effective strategy to reduce post-operative morbidity and mortality.
Table 3 Dichotomous Variable Risk Ratios for Tricuspid Valve Procedure Before and After Propensity Score Adjustment Using Inverse Probability Weighting Before IPW adjustment Variables Operative mortality ICU hoursa Initial 4 166 Total 4 172 LOS 4 21 days Any blood products Intra-/post-op Post-op Any complicationb Post-op complications Sternal infection Perm/embolic stroke Prolonged ventilation New renal failure New dialysis Reoperation Any Bleeding or tamponade RVAD insertion VAD complications Intracranial bleeding Device infection Device malfunction
After IPW adjustment
No. (events/total)
RR for TVP (95% CI)
221/2,161
1.07 (0.81–1.41)
0.6351
0.95 (0.68–1.33)
0.7658
1,026/2,084 1,045/2,084 1,049/2,152
1.14 (1.04–1.25) 1.16 (1.06–1.27) 1.33 (1.22–1.45)
0.0056 0.0015 o0.0001
1.14 (1.02–1.28) 1.14 (1.02–1.27) 1.29 (1.16–1.43)
0.0165 0.0210 o0.0001
2,018/2,108 1,749/2,110 1,764/2,164
1.03 (1.01–1.05) 1.06 (1.02–1.11) 1.13 (1.09–1.17)
0.0002 0.0027 o0.0001
1.03 (1.01–1.05) 1.04 (0.99–1.09) 1.12 (1.07–1.17)
0.0013 0.1221 o0.0001
17/2,154 67/2,153 1,442/2,154 212/1,822 155/2,088
1.12 0.71 1.13 1.39 1.65
(0.40–3.17) (0.40–1.27) (1.07–1.21) (1.06–1.81) (1.21–2.25)
0.8305 0.2508 o0.0001 0.0154 0.0015
1.31 0.65 1.10 1.53 1.49
699/2,154 345/2,154 98/2,161
1.23 (1.08–1.39) 1.28 (1.04–1.57) 1.02 (0.66–1.57)
23/2,092 29/2,092 21/2,092
0.57 (0.19–1.67) 1.22 (0.56–2.66) 0.84 (0.31–2.29)
p-value
RR for TVP (95% CI)
p-value
(0.43–3.99) (0.32–1.33) (1.02–1.18) (1.13–2.08) (1.03–2.15)
0.6298 0.2412 0.0131 0.0061 0.0339
0.0018 0.0201 0.9296
1.24 (1.07–1.45) 1.16 (0.89–1.51) 0.81 (0.49–1.36)
0.0056 0.2600 0.4310
0.3034 0.6232 0.7405
0.83 (0.24–2.93) 1.21 (0.46–3.16) 0.81 (0.28–2.31)
0.7753 0.6956 0.6933
CI, confidence interval; ICU, intensive care unit; IPW, inverse probability weighting; LOS, length of stay; RR, risk ratio; RVAD, right ventricular assist device; TVP, tricuspid valve procedure; VAD, ventricular assist device. a Median initial ICU stay is 166 hours; median total ICU stay is 172 hours. b Includes post-operative complications, reoperations, and VAD complications.
Robertson et al.
Concomitant TVP for TR in cfLVAD Patients
7
Table 4 Treatment Group Analysis for Continuous Outcome Variables Before and After Propensity Score Adjustment Using Inverse Probability Weighting Before IPW adjustment
Variable ICU hours Initial Total Hospital LOS, days Post-op transfusion Packed RBC, U FFP, U Cryoprecipitate, U Platelets, U Intra-/post-op transfusion Packed RBCs, U FFP, U Cryoprecipitate, U Platelets, U
After IPW adjustment
LVAD (mean)
LVADþTVP (mean)
Difference (mean)
258.76 291.09 25.29
283.00 311.98 28.47
24.25 20.89 3.19
0.1132 0.2128 0.0044
6.89 3.62 0.59 1.71
8.15 4.15 0.83 1.86
1.26 0.53 0.24 0.15
0.0104 0.0973 0.0411 0.4657
9.79 7.44 1.91 3.69
11.74 9.41 2.61 4.45
1.95 1.97 0.70 0.76
0.0008 o0.0001 0.0071 0.0063
p-value
LVAD (mean)
LVADþTVP (mean)
Difference (mean)
p-value
262.63 294.52 25.79
275.81 303.16 29.38
13.18 8.64 3.58
0.4099 0.6251 0.0835
6.921 3.662 0.59 1.71 9.90 7.53 1.95 3.71
7.935 4.119 0.91 1.76 11.45 9.14 2.77 4.49
1.01 0.46 0.31 0.044
0.1025 0.2365 0.0627 0.8266
1.55 1.61 0.82 0.78
0.0222 0.0024 0.0214 0.0156
FFP, fresh frozen plasma; ICU, intensive care unit; IPW, inverse probability weighting; LOS, length of stay; LVAD, left ventricular assist device; RBC, red blood cells; TVP, tricuspid valve procedure.
Prior analyses regarding the prognostic significance of preoperative moderate to severe TR have resulted in varied conclusions,10–12 perhaps partly because of the interplay between native valvular disease and physiologic changes that occur with the initiation of mechanical support. Evidence has shown that implantation of an LVAD by itself can improve TR in some patients by reducing secondarily elevated pulmonary artery pressures.6,10,13,17 Reports suggest that as many as 50% of patients will see improvements in TR post-operatively,17 but in contrast to the performance of a tricuspid ring annuloplasty, LVAD insertion alone does not seem to produce uniform improvement in the degree of TR and, in some cases, it may be a gradual process over a period of months.3,18,19 The TR in another sub-set of patients will stay the same or worsen post-operatively. Acute worsening of TR can
occur from distortion of the tricuspid annulus due to leftward deviation of the intraventricular septum secondary to increased flow after LVAD placement. Significant volume-loading and transfusion requirements that are often necessary in the post-operative period may also exacerbate TR when flows are limited by output across the pulmonary vascular bed.20 Moreover, LVAD patients commonly have implantable cardioverter defibrillator leads traversing the tricuspid valve that may entrap the TV leaflets and lead to an anatomic cause for persistence of TR. Complicating matters, there is evidence that the absence of severe TR pre-operatively does not imply that the tricuspid orifice is free of disease, and the presence of annular dilation could represent a significant substrate for the later development of TR.21 Therefore, because the
Table 5 Sensitivity Analysis Comparing Results from Traditional Multivariate Regression Analysis to Inverse Probability Weighting Using Conditional Logistic Regression Before IPW adjustment
After IPW adjustment
Variables
No. (events/total)
OR for TVP (95% CI)
p-value
OR for TVP (95% CI)
p-value
Operative mortality Reoperation Any Bleeding or tamponade RVAD insertion Prolonged ventilation New renal failure Stroke
224/2,196
1.22 (0.83–1.80)
0.3050
0.95 (0.61–1.47)
0.8177
710/2,189 350/2,189 99/2,196 1,462/2,189 214/1,851 67/2,188
1.39 1.67 0.94 1.48 1.66 1.25
0.0099 0.0012 0.8170 0.0039 0.0121 0.5227
1.46 1.93 0.69 1.40 1.93 1.22
0.0076 0.0002 0.2631 0.0262 0.0002 0.6045
(1.08–1.79) (1.22–2.27) (0.54–1.64) (1.13–1.940) (1.12–2.48) (0.63–2.50)
(1.11–1.93) (1.37–2.72) (0.36–1.32) (1.04–1.89) (1.37–2.72) (0.57–2.62)
CI, confidence interval; IPW, inverse probability weighting; LVAD, left ventricular assist device; OR, odds ratio; RVAD, right ventricular assist device; TVP, tricuspid valve procedure. Variables in adjusted model: atrial fibrillation, age, gender, body surface area, congestive heart failure, New York Heart Association class IV, chronic lung disease, creatinine, cerebrovascular disease, cardiovascular disease, diabetes without insulin, dialysis, hypertension, intraaortic balloon pump, inotropes, immunosuppression, left main disease; NDV, number of diseased vessels o 21 days, unstable angina, peripheral vascular disease; race, reoperation (first, Z 2), cardiogenic shock, status urgent/emergent salvage, year of surgery, mitral valve insufficiency moderate/severe, aortic valve insufficiency moderate/severe tricuspid insufficiency severe, concomitant procedures other than valve. Interaction terms: age reoperation, age status, gender body surface area.
8
The Journal of Heart and Lung Transplantation, Vol ], No ], Month ]]]]
degree of pre-operative TR often changes after implantation of a cfLVAD, intervening based on this criterion alone will result in many unnecessarily performed concomitant TVPs, or, in some patients, failure to operate when it may have been beneficial. None of the currently available studies have reported a significant early or late survival benefit to concomitant TVP.2–7,9 The current report supports these data with respect to early mortality. This could possibly reflect the effectiveness of medical treatment of RV failure with pulmonary vasodilators and inotropic agents. Another explanation, however, is that, for the reasons discussed, patient selection for concomitant TVP failed to identify patients who would have had significant post-operative TR and exacerbations of RV failure. With respect to improvements in right heart function after a LVADþTVP in patients with significant TR, Piacentino et al6,7 showed reductions in RV failure, defined as postoperative inotrope duration 4 14 days or RVAD insertion. One of their 2 studies demonstrated a significant decrease in the incidence of post-operative RVAD requirement with concomitant TVP of 21% (6 of 28) vs 3% (1 of 33; p o 0.05), but other groups have failed to replicate these findings.2–4 In particular, the 21% incidence of RVAD insertion seems high compared with other reports and could have been artificially inflated by small sample sizes4,12,22 and might also reflect different strategies for pre-operative optimization of patients with significant TR and/or preoperative RV failure. Our study found the incidence of reoperation for RVAD implantation reported nationally for patients with moderate to severe TR was only 4.5% (98 of 2,161) and was not affected by performing a concomitant TVP. Unfortunately, we cannot comment on the duration of post-operative inotrope used in the present study. Moreover, concern exists that prolonged back pressure from non-regression or progression of TR could lead to renal or hepatic congestion with resultant coagulopathy. Of multiple studies addressing these issues,2,4,7,9 1 study demonstrated a lower incidence of post-operative renal dysfunction (39% vs 21%, p ¼ 0.05).7 Surprisingly, our data show that performance of a concomitant TVP is significantly associated with an increased risk for new postoperative renal failure and dialysis. Transfusion requirement in our study is also significantly increased for the concomitant TVP group. A study by Saeed et al9 that compared LVAD (n ¼ 42) with LVADþTVP (n ¼ 8) for patients with moderate to severe TR is consistent with this latter finding, demonstrating a higher median number of packed red blood cells transfused for the group that received a concomitant TVP (4 vs 13 units; p ¼ 0.03). Although these differences could possibly reflect the effects of the right heart on end-organ function, it is more likely that the higher incidence of renal failure and the greater transfusion requirement are consequences of prolonged time on CPB and longer cross-clamp times. The necessary addition of an atriotomy, additional dissection, and the need for bicaval cannulation in the LVADþTVP group may also have affected bleeding risk.
The present study has further shown that patients who undergo a LVADþTVP are at increased risk for significantly prolonged ventilation, longer ICU stays, longer hospital stays, and reoperation. This is in contrast to a single-institution study that demonstrated shortened hospital length of stay with concomitant TVP (26 vs 19 days; p ¼ 0.02). No other study has detected a difference with respect to hospital stay, ICU stay, or ventilator duration.2–4,9 Given the wide variation in these parameters between patients, previous studies were underpowered to adequately detect differences. The finding of increased risk of reoperation, particularly for bleeding and tamponade, is novel. The reasons for these data are uncertain, but the role of prolonged time on CPB and the addition of an atriotomy are unlikely to account for all of the observed differences in the treatment groups. The relative effects of these factors vs the potential benefit of an immediate reduction in central venous pressure are not known. Although the results do not support performing concomitant TVPs in cfLVAD patients with moderate to severe TR, several important questions remain. First, although this STS database data is able to address peri-operative outcomes, we are unable to assess long-term differences in mortality or morbidity between groups. Analyses using the Interagency Registry for Mechanically Assisted Circulatory Support database are ongoing and will help elucidate longterm outcomes. Still, for this important clinical question, it remains necessary to validate results with different approaches in multiple large registries to best address the absence of randomized data. Second, some authors have suggested that parameters such as annular dilation, annular motion, and geometry of the RV are predictors of post-operative RV failure. Whether selection of candidates for concomitant TVP by one or a combination of these parameters would improve outcomes over selection according to the degree of pre-operative TR alone remains unclear. Third, we do not know whether a concomitant TVP in patients requiring temporary RVAD support would facilitate weaning in patients with moderate to severe TR and/or other echocardiographic parameters of annular disease. This study also has other limitations. First, we do not have specific data on the temporal relationship between surgery and the pre-operative determination of TR severity and do not know the loading conditions or clinical status of the patients at the time it was measured. Second, preoperative and post-operative hemodynamic data in the groups are insufficient to eliminate potential unmatched treatment bias. We are unable to measure the effect of preoperative intrinsic RV function or its interaction with TR and any subsequent effect on post-operative RV failure. Further, we are unable to assess inotrope usage or hemodynamic indices of RV failure post-operatively. However, in the absence of negative effects on other measured variables in the LVAD-alone group, the duration of inotrope usage has diminishing significance unless it is prolonged by concomitant TVP. Third, serial echocardiographic studies are not available, so we are unable to comment on how the TR changes over time in these patients, with or
Robertson et al.
Concomitant TVP for TR in cfLVAD Patients
without TVP. Available evidence suggests that even with a concomitant TVP, as a many as 10% of patients may continue to have significant TR after insertion of a cfLVAD,6,23 and, as discussed, insertion of an LVAD alone can lead to significant post-operative TR in certain patients with insignificant preoperative TR. This suggests that some fraction of patients in both groups likely had significant post-operative TR, but how this affects outcomes is uncertain. Lastly, there are insufficient data to examine the potential effect of aortic cross-clamp on post-operative morbidity and mortality. However, this study is substantially larger than any previous report on this question and is a multi-institutional study using a prospectively collected database. As such, this report eliminates single-center bias and has greater power to detect significant differences between groups independent of center bias and surgeon preference. In conclusion, this study examined a real-world, multiinstitutional cohort of cfLVAD patients who had moderate to severe pre-operative TR and compared patients who underwent implantation of a LVAD alone with those who had a LVADþTVP. After matching patient cohorts by IPW, concomitant TVP did not improve the risk of short-term mortality or RVAD insertion and was associated with increased risk for prolonged ventilation, prolonged ICU and hospital stays, post-operative renal failure/dialysis, transfusion, and reoperation. As such, these data suggest that the practice of performing a concomitant TVP based on the degree of pre-operative TR alone should be reexamined. Although concomitant TVP might benefit some patients, the data suggest that routine TV intervention in cfLVAD patients may impart additional short-term morbidity for the population as a whole. The medium-term and longer-term benefits remain to be defined. The strategy of performing a concomitant TVP may still be beneficial if a more appropriate method of patient selection is identified. Owing to the demonstrated short-term morbidity of concomitant TVP in this study, further data are required to determine the magnitude and timing of any potential benefit as well as criteria for selection of candidates for TVP.
Disclosure statement Funding for statistical support was provided by the STS Database after review and acceptance of our research proposal. Dr. Silvestry is a consultant for Thoratec and HeartWare. None of the other authors has a financial relationship with a commercial entity that has an interest in the subject of the presented manuscript or other conflicts of interest to disclose.
References 1. Kormos RL, Teuteberg JJ, Pagani FD, et al. Right ventricular failure in patients with the HeartMate II continuous-flow left ventricular assist device: incidence, risk factors, and effect on outcomes. J Thorac Cardiovasc Surg 2010;139:1316-24. 2. Koeckert M, Comas G, Yang J, et al. Tricuspid valve repair during left ventricular assist device insertion. J Heart Lung Transplant 2011;30:S43-4.
9 3. Krishan K, Nair A, Pinney S, Adams DH, Anyanwu AC. Liberal use of tricuspid-valve annuloplasty during left-ventricular assist device implantation. Eur J Cardiothorac Surg 2012;41:213-7. 4. Maltais S, Topilsky Y, Tchantchaleishvili V, et al. Surgical treatment of tricuspid valve insufficiency promotes early reverse remodeling in patients with axial-flow left ventricular assist devices. J Thorac Cardiovasc Surg 2012;143:1370-6. 5. Pal JD, Klodell CT, John R, et al. Low operative mortality with implantation of a continuous-flow left ventricular assist device and impact of concurrent cardiac procedures. Circulation 2009;120: S215-9. 6. Piacentino V 3rd, Ganapathi AM, Stafford-Smith M, et al. Utility of concomitant tricuspid valve procedures for patients undergoing implantation of a continuous-flow left ventricular device. J Thorac Cardiovasc Surg 2012;144:1217-21. 7. Piacentino V 3rd, Troupes CD, Ganapathi AM, et al. Clinical impact of concomitant tricuspid valve procedures during left ventricular assist device implantation. Ann Thorac Surg 2011;92:1414-8; discussion: 1418–9. 8. Potapov EV, Schweiger M, Stepanenko A, et al. Tricuspid valve repair in patients supported with left ventricular assist devices. ASAIO J 2011;57:363-7. 9. Saeed D, Kidambi T, Shalli S, et al. Tricuspid valve repair with left ventricular assist device implantation: is it warranted? J Heart Lung Transplant 2011;30:530-5. 10. Piacentino V 3rd, Williams ML, Depp T, et al. Impact of tricuspid valve regurgitation in patients treated with implantable left ventricular assist devices. Ann Thorac Surg 2011;91:1342-6; discussion: 1346–7. 11. Potapov EV, Stepanenko A, Dandel M, et al. Tricuspid incompetence and geometry of the right ventricle as predictors of right ventricular function after implantation of a left ventricular assist device. J Heart Lung Transplant 2008;27:1275-81. 12. Puwanant S, Hamilton KK, Klodell CT, et al. Tricuspid annular motion as a predictor of severe right ventricular failure after left ventricular assist device implantation. J Heart Lung Transplant 2008;27:1102-7. 13. Lee S, Kamdar F, Madlon-Kay R, et al. Effects of the HeartMate II continuous-flow left ventricular assist device on right ventricular function. J Heart Lung Transplant 2010;29:209-15. 14. O’Brien SM, Shahian DM, Filardo G, et al. The Society of Thoracic Surgeons 2008 cardiac surgery risk models: part 2—isolated valve surgery. Ann Thorac Surg 2009;88:S23–42. 15. White H. Maximum likelihood estimation of misspecified models. Econometrica 1982;50:1-25. 16. Localio AR, Berlin JA, Ten Have TR, Kimmel SE. Adjustments for center in multicenter studies: an overview. Ann Intern Med 2001;135:112-23. 17. Morgan JA, Paone G, Nemeh HW, et al. Impact of continuous-flow left ventricular assist device support on right ventricular function. J Heart Lung Transplant 2013;32:398-403. 18. McCarthy PM, Bhudia SK, Rajeswaran J, et al. Tricuspid valve repair: durability and risk factors for failure. J Thorac Cardiovasc Surg 2004;127:674-85. 19. Rao V, Slater JP, Edwards NM, Naka Y, Oz MC. Surgical management of valvular disease in patients requiring left ventricular assist device support. Ann Thorac Surg 2001;71:1448-53. 20. Chumnanvej S, Wood MJ, MacGillivray TE, Melo MF. Perioperative echocardiographic examination for ventricular assist device implantation. Anesth Analg 2007;105:583-601. 21. Dreyfus GD, Corbi PJ, Chan KM, Bahrami T. Secondary tricuspid regurgitation or dilatation: which should be the criteria for surgical repair? Ann Thorac Surg 2005;79:127-32. 22. Kirklin JK, Naftel DC, Pagani FD, et al. Long-term mechanical circulatory support (destination therapy): on track to compete with heart transplantation? J Thorac Cardiovasc Surg 2012;144:584-603; discussion: 597–8. 23. Deo SV, Hasin T, Altarabsheh SE, et al. Concomitant tricuspid valve repair or replacement during left ventricular assist device implant demonstrates comparable outcomes in the long term. J Card Surg 2012;27:760-6.
