2017 Volume 81 Issue 3 Pages 330-338
Background: The aim of this study was to compare early and late outcomes of tricuspid valve replacement (TVR) and tricuspid valve repair (TVr) for severe tricuspid regurgitation (TR).
Methods and Results: From 1994 to 2012, 360 patients (mean age, 58±13 years) with severe TR underwent TVR (n=97, 27%) or TVr (n=263, 73%). Among them, 282 patients (78%) had initial rheumatic etiology, and 307 patients (85%) had preoperative atrial fibrillation. The TVR group had higher total bilirubin, higher baseline central venous pressure, and higher incidence of previous cardiac operation. There was no difference in early mortality (TVR:TVr, 3.1%:3.4%, P=0.877). Ten-year overall survival (TVR:TVr, 72%:70%, P=0.532) and 10-year freedom from cardiac death (TVR:TVr, 76%:77%, P=0.715) were not significantly different between groups. After applying stabilized inverse probability of treatment weighting methods, there were still no significant differences in early mortality (P=0.293), overall survival (P=0.649) or freedom from cardiac death (P=0.870). Higher NYHA functional class, total bilirubin (>2 mg/dL), initial central venous pressure, and cardiopulmonary bypass time were independent predictors of early mortality. Older age, LV dysfunction (EF <40%), and hemoglobin <10 g/dL were independent predictors of late cardiac mortality.
Conclusions: Compared with TVr, TVR had acceptable early and late outcomes in patients with severe TR. TVR can be considered as a valid option with acceptable clinical outcomes in patients who are not suitable candidates for TVr.
It is well documented that tricuspid valve replacement (TVR) is associated with worse early and long-term outcomes than tricuspid valve repair (TVr).1–6 Therefore, most efforts are made to salvage the native tricuspid valve during tricuspid valve surgery. Currently, there appears to be several factors to take into consideration with regard to previous studies on tricuspid valve surgery. First, there are commonly great differences in baseline patient characteristics between TVR and TVr.1,3,6 TVR is frequently performed in patients with another simultaneous valve disease or advanced heart failure. Moreover, there is a higher number of cardiac reoperations in TVR recipients.4,6,7 Second, current indications for tricuspid valve surgery are different from those of 2–3 decades ago. With growing consensus that intervention for tricuspid regurgitation (TR) should be performed earlier, patient characteristics are changing and outcomes are improving. In addition, perioperative management for patients in critical cardiac condition has improved. Several recently published articles showed that TVR is not related to extremely higher risk in severe TR patients.4,7,8
The aim of this study was therefore to compare early and long-term outcomes of TVR and TVr performed for severe TR before and after applying stabilized inverse probability treatment weighted methods, and to evaluate the risk factors for early and late mortality in tricuspid valve surgery.
This was a retrospective observational study using data extracted from our prospectively collected database. The institutional review board approved this study, and informed consent was waived based on the retrospective nature of this study (IRB no.: 2014-09-012).
Baseline patient characteristics are listed in Table 1. From August 1994 to December 2012, 372 patients underwent either TVR (first-time cases) or TVr for severe TR (reported as moderate–severe, 29%; or severe, 70.6%, on transthoracic echocardiography). Patients with Ebstein anomaly (n=9) or tetralogy of Fallot (n=3) were excluded. A total of 360 patients were enrolled in this study. The operations were performed by 7 surgeons at the present institution, but mostly by a specialized surgeon in valve surgery (P.W.P.). There were 97 patients (27%) in the TVR group and 263 patients (73%) in the TVr group.
| TVR group (n=97) |
TVr group (n=263) |
P-value | Total (n=360) |
|
|---|---|---|---|---|
| Age at operation (years) | 54±13 | 59±12 | 0.001 | 58±13 |
| Female | 74 (76.3) | 170 (64.6) | 0.042 | 244 (67.8) |
| History of AMI (>3 months) | 1 (1.0) | 6 (2.3) | 0.680 | 7 (1.9) |
| Previous cerebrovascular accident | 13 (13.4) | 30 (11.4) | 0.714 | 43 (11.9) |
| Infective endocarditis (active) | 5 (5.2) | 10 (3.8) | 0.560 | 15 (4.2) |
| Extracardiac arteriopathy | 9 (3.4) | 2 (2.1) | 0.734 | 11 (3.1) |
| Preoperative i.v. inotrope use | 19 (19.6) | 42 (16.0) | 0.431 | 61 (16.9) |
| Ventilator, IABP or ECMO dependency before surgery | 1 (1.0) | 7 (2.7) | 0.688 | 8 (2.2) |
| NYHA functional class III or IV | 51 (52.6) | 138 (52.5) | >0.999 | 189 (52.5) |
| Logistic EuroSCORE (%) | 10.8±12.9 | 9.2±11.2 | 0.257 | 9.6±11.7 |
| Atrial fibrillation | 83 (85.6) | 224 (85.2) | 1.000 | 307 (85.3) |
| Estimated RVSP (mmHg)† | 49±16 | 55±19 | 0.015 | 54±19 |
| Initial left-sided valve disease | 0.079 | |||
| Rheumatic | 83 (85.6) | 199 (75.7) | 282 (78.3) | |
| Non-rheumatic | 14 (14.4) | 60 (22.8) | 74 (20.6) | |
| Not applicable | 0 | 4 (1.5) | 4 (1.1) | |
| LV dysfunction (EF <40%) | 7 (7.2) | 24 (9.1) | 0.675 | 31 (8.6) |
| Chronic renal failure (on dialysis) | 1 (1.0) | 7 (2.7) | 0.688 | 8 (2.2) |
| Lab | ||||
| Total bilirubin ≥2 mg/dL | 23 (23.7) | 32 (12.2) | 0.009 | 55 (15.3) |
| eGFR <60 mL/min | 27 (27.8) | 71 (27.0) | 0.894 | 98 (27.2) |
| Hb <10 g/dL | 24 (24.7) | 45 (17.1) | 0.131 | 69 (19.2) |
| No. previous cardiac operations | <0.001 | |||
| 0 | 39 (40.2) | 218 (82.9) | 257 (71.4) | |
| 1 | 44 (45.4) | 40 (15.2) | 83 (23.3) | |
| 2 | 13 (13.4) | 3 (1.1) | 16 (4.4) | |
| 3 | 1 (1.0) | 2 (0.8) | 3 (0.8) | |
| Previous valve implantation | <0.001 | |||
| AVR | 1 (1.0) | 0 | ||
| MVR | 25 (25.8) | 14 (5.3) | ||
| DVR | 27 (27.8) | 14 (5.3) | ||
| Previous TVr | 25 (25.8) | 3 (1.1) | <0.001 | 28 (7.8) |
Data given as mean±SD or n (%). †RVSP was available in 66/97 patients (68%) in the TVR group, and 206/263 patients (78%) in the TVr group. AMI, acute myocardial infarction; AVR, aortic valve replacement; DVR, double valve replacement; ECMO, extracorporeal membrane oxygenator; EF, ejection fraction; eGFR, estimated glomerular filtration rate; Hb, hemoglobin; IABP, intra-aortic balloon pump; LV, left ventricle; MVR, mitral valve replacement; NYHA, New York Heart Association; RVSP, right ventricular systolic pressure; TV, tricuspid valve; TVR, tricuspid valve replacement; TVr, tricuspid valve repair.
Under full median sternotomy, cold antegrade and/or retrograde blood cardioplegia were used. Operations for left-sided valves always preceded tricuspid valve surgery. TVR was chosen when the valve leaflets had a substantial degree of structural deformity, and TVr was chosen when the TR was attributable to annular dilation and/or minimal leaflet deformity. In addition, TVR was considered to be favorable when (1) TVr had already been performed previously; (2) there was right ventricular (RV) dysfunction with severe enlargement of both the right atrium and ventricle; or (3) the patient had already undergone multiple redo operations. In some cases of TVR with hugely dilated RV, a crescent-shaped bovine pericardial patch was sutured at the thinned valve tissue on the conduction area with 6-0 polypropylene, and valve stitches were done on the bovine pericardium (Figure 1). TVr always included ring or suture annuloplasty, and commissuroplasty or coaptation sutures were added as necessary.
Surgical photograph of tricuspid valve replacement using pericardial patch augmentation. (A) When the right ventricle was hugely dilated, a crescent-shaped bovine pericardial patch was sutured at the thinned valve tissue on the conduction area with 6-0 polypropylene. (B,C) Valve stitches on the bovine pericardium. (D) Finished status.
In the TVR group (n=97), a mechanical prosthesis was used in 69 patients (71%) and a tissue prosthesis in 28 patients (29%; Table 2). Generally, mechanical prosthesis was preferred when the patient had previously implanted mechanical prosthesis in the left heart. Tissue prosthesis was chosen when the life expectancy was <15 years. In the TVr group (n=263), annuloplasty ring was used in 186 patients (71%), while a De Vega annuloplasty was performed in 76 patients (29%) in the earlier period of the study. For modified De Vega procedure, 2-0 monofilament suture with Teflon pledgets at both ends was used from the midportion of the septal leaflet to antero-septal commissure. Two or 3 more pledgets were interposed among the annuloplasty bites to prevent tissue cut-through.
| TVR group (n=97) |
TVr group (n=263) |
P-value | |
|---|---|---|---|
| Initial CVP at operating room (mmHg) | 15±6 | 11±5 | <0.001 |
| CPB time (min) | 186±82 | 157±64 | 0.002 |
| ACC time (min) | 127±63 | 115±49 | 0.094 |
| Total ultrafiltration (L) | 4.1±1.9 | 3.3±1.8 | <0.001 |
| Ultrafiltration/kg (mL/kg) | 95±43 | 77±35 | <0.001 |
| Mechanical valve | 69 (71) | ||
| St. Jude Mechanical | 40 (41) | ||
| On-X | 18 (19) | ||
| ATS | 9 (9) | ||
| Sorin Bicarbon | 2 (2) | ||
| Tissue valve | 28 (29) | ||
| St. Jude Biocor or Epic | 18 (19) | ||
| Carpentier-Edwards Pericardial | 7 (7) | ||
| Medtronic Hancock II | 3 (3) | ||
| Annuloplasty ring used | 186 (71) | ||
| Annuloplasty ring not used | 76 (29) | ||
| Ring usage uncertain | 1 (0.4) | ||
| Concomitant CABG | 2 (2.1) | 13 (4.9) | 0.225 |
| Concomitant maze | 20 (21) | 112 (43) | <0.001 |
| Concomitant valve operations | – | ||
| Tricuspid valve surgery alone | 31 (32.0) | 24 (9.1) | |
| 3rd redo MVR | 3 (3.1) | 0 | |
| 2nd redo AVR | 1 (1.0) | 0 | |
| 2nd redo AVR+redo MVR | 1 (1.0) | 0 | |
| 2nd redo MVR | 0 | 3 (1.1) | |
| 2nd redo MVR+AVR | 2 (2.1) | 0 | |
| 2nd redo MVR+redo AVR | 1 (1.0) | 1 (0.4) | |
| Redo AVR | 10 (10.3) | 3 (1.1) | |
| Redo AVR+MVR | 1 (1.0) | 0 | |
| Redo AVR+Redo MVR | 5 (5.2) | 4 (1.5) | |
| Redo MVR | 7 (7.2) | 9 (3.4) | |
| Redo MVR+AVR | 0 | 2 (0.8) | |
| Redo MVR+AV repair | 0 | 1 (0.4) | |
| Redo PVR | 0 | 1 (0.4) | |
| AVR | 3 (3.1) | 2 (0.8) | |
| AVR+MV repair | 0 | 10 (3.8) | |
| Bentall | 1 (1.0) | 0 | |
| MVR+AVR | 10 (10.3) | 41 (15.2) | |
| MV repair | 5 (5.2) | 60 (22.8) | |
| MV repair+AV repair | 1 (1.0) | 5 (1.9) | |
| MVR | 16 (16.5) | 91 (34.6) | |
| MVR+AV repair | 1 (1.0) | 5 (1.9) | |
| PVR | 0 | 1 (0.4) |
Data given as mean±SD or n (%). ACC, aortic cross-clamp; AV, aortic valve; CABG, coronary artery bypass grafting; CPB, cardiopulmonary bypass; CVP, central venous pressure; MV, mitral valve; PVR, pulmonary valve replacement. Other abbreviations as in Table 1.
Before surgery, chest computed tomography with contrast material or coronary angiography was performed ≥24 h before operation. Prothrombin time (PT) international normalized ratio (INR) was reversed after adding fresh frozen plasma and vitamin K. I.v. heparin was started as required.
In the operating room, central venous pressure (CVP) was checked after insertion of a central venous catheter. When CVP was >10–12 mmHg, continuous injection of dobutamine and nitroglycerin was started, especially in redo operations. During operation, the choice of TVR or TVr was made as quickly as possible. Repeated repair attempts were avoided, especially in high-risk patients. Conventional and modified ultrafiltration was applied during and after cardiopulmonary bypass (CPB).9 Target hematocrit after modified ultrafiltration was >33–35%. CVP was kept below 10–12 mmHg. Since 2000, oral sildenafil (10–20 mg 3–4 times per day) has been added as soon as possible after the patient is transferred to the intensive care unit (ICU).
The target INR for TVR with mechanical valves was the same regardless of valve product type. We preferred a target PT-INR of 2.5–3.5 during the first year after surgery, and it was then maintained at 2.5–3.0.
Follow-upAmong the 360 patients enrolled in this study, 271 patients (75%) were alive at closure of follow-up; 215 patients (80%) were still being followed at the present institution, while 28 (10%) were followed by other hospitals. Survival data for the remaining 27 patients (10%) was collected from Korea National Medical Insurance information and Statistics Korea. Survival data covered 359/360 cases (99.7%) except for 1 foreigner. Follow-up echocardiography was available in 90.1% at 1 year, 72.5% at 3 years, 68.5% at 5 years, and 67.3% at 7 years. Mean follow-up period was 82±55 months (range, 0.1–228 months). Recurrence of TR was defined as moderate or greater degree of TR on echocardiography during follow-up. Other definitions of morbidity and mortality followed the guidelines for reporting mortality and morbidity after cardiac valve interventions.10
Statistical AnalysisPASW 18 (IBM SPSS, Armonk, NY, USA) and R (version 3.1.1, R Foundation for Statistical Computing) with the survival and survey packages were used. Student’s t-test and chi-squared test were used for comparison of continuous and categorical variables, respectively. A logistic regression model was constructed to calculate the propensity score for each case, for disposition in either of the 2 groups (TVR and TVr). Variables in this propensity score model included sex, age, body mass index, diabetes, hypertension, history of cerebrovascular accident, chronic obstructive pulmonary disease, New York Heart Association (NYHA) functional class, cardiac reoperation, left ventricular (LV) dysfunction (ejection fraction [EF] <40%), anemia (hemoglobin <10 g/dL), hyperbilirubinemia (total bilirubin >2 mg/dL), and estimated glomerular filtration rate <60 mL/h. Using the propensity scores derived from this model, stabilized inverse probability treatment weighting (sIPTW) was applied. After applying sIPTW, key baseline characteristics were well balanced, with non-significant P-values and a standardized mean difference <10% (Table S1). In order to compare survival data after the application of sIPTW, weighted Kaplan-Meier and weighted Cox regression analyses were performed.
Another logistic regression model was utilized to identify risk factors of operative mortality. Variables with P<0.1 on univariate analysis were included in the multivariable analysis, and the variables were filtered again with backward conditional selection. The variables that passed these steps were then entered into the final model along with the most important variable of TVR vs. TVr. Cox proportional hazards ratio (Cox regression) models were used to identify risk factors of long-term mortality, cardiac death, and major adverse cardiovascular events (MACE). First, a univariate analysis was performed for each variable. Variables with P<0.1 on univariate analyses were chosen for the next step. After another variable selection process using the backward conditional method, the last set of variables were entered into the final model. As in the logistic regression analysis, the variable of interest (TVR vs. TVr) was included in the multivariable analysis for analytical purposes.
Mean patient age was 58±13 years and 244 patients (68%) were female. Before intergroup balancing with sIPTW, patients in the TVr group were older, and more likely to be male (Table 1). The number of patients with hyperbilirubinemia (total bilirubin ≥2 mg/dL) was higher in the TVR group. The number of patients with previous open heart surgery was significantly different between groups: 45/263 patients (17%) in the TVr group had at least 1 previous open heart surgery compared with 58/97 (60%) in the TVR group (P<0.001). Among 103 patients with a history of previous cardiac surgery, 85 patients (83%) had previous operation in another hospital. In the TVR group, 25 patients (26%) had undergone previous TVr, compared with only 3 patients (1%) in the TVr group (P<0.001). The proportion of patients with severe TR was significantly higher in the TVR group (TVR:TVr, 88.9%:63.0%, P<0.001).
Surgical data are summarized in Table 2. Initial CVP measured in the operating room was significantly higher in the TVR group (P<0.001). CPB time was longer in the TVR group, but aortic cross-clamp time was not significantly different between groups (Table 2). The total amount of ultrafiltration was greater in the TVR group (P<0.001). There were a variety of combinations of concomitant surgery, except 31 (32%) and 24 (9%) lone tricuspid valve operations in the TVR and TVr group, respectively. Mitral valve replacement (n=16, 17%) was the most frequently performed concomitant operation in the TVR group, while mitral valve repair (n=60, 23%) was the most frequently performed in the TVr group (Table 2).
Early OutcomeEarly and late outcomes are summarized in Table 3. Early mortality (≤30 days) was not significantly different between groups, with 3 (3.1%) and 9 (3.4%) in the TVR and TVr groups, respectively (P=0.877). Among the 3 cases of early mortality in the TVR group, 2 involved emergency cardiac reoperations due to prosthetic valve endocarditis (Table S2). There were 3 (0.8%) early reoperations requiring CPB: 1 (1.0%) in the TVR group and 2 (0.8%) in the TVr group.
| TVR group (n=97) |
TVr group (n=263) |
P-value | |
|---|---|---|---|
| Postoperative ventilator dependency (h) | 49 (5–1,296) | 31 (4–720) | 0.270 |
| ICU stay (days) | 5 (1–54) | 4 (1–65) | 0.179 |
| Early mortality (≤30 days) | 3 (3.1) | 9 (3.4) | 0.877 |
| Early morbidities | |||
| Reoperation | 1 (1.0) | 2 (0.8) | >0.999 |
| Paravalvular leak of any valve | 3 (3.1) | 1 (0.4) | 0.061 |
| Tracheostomy | 2 (2.1) | 4 (1.5) | 0.722 |
| Bleeding (re-exploration) | 4 (4.1) | 8 (3.0) | 0.612 |
| New-onset heart block | 2 (2.1) | 2 (0.8) | 0.296 |
| Cerebral infarction | 1 (1.0) | 6 (2.3) | 0.446 |
| Cerebral hemorrhage | 7 (7.2) | 8 (3.0) | 0.079 |
| AKI requiring dialysis | 2 (2.1) | 9 (3.4) | 0.506 |
| IABP or ECMO | 4 (4.1) | 11 (4.2) | 0.980 |
| Delayed sternal closure | 1 (1.0) | 3 (1.1) | 0.930 |
| Mediastinitis | 1 (1.0) | 0 | 0.269 |
| Late morbidities | |||
| Cardiac reoperation | 3 (3.1) | 14 (5.3) | 0.576 |
| Infective endocarditis during follow-up | 3 (3.1) | 7 (2.7) | 0.733 |
| Anticoagulation-related hemorrhage | 12 (12) | 22 (8.4) | 0.309 |
| Paravalvular leak of any valves | 4 (4.1) | 7 (2.7) | 0.496 |
| Thrombosis of any prosthetic valve | 3 (3.1) | 1 (0.4) | 0.061 |
Data given as median (range) or n (%). AKI, acute kidney injury; ICU, intensive care unit. Other abbreviations as in Table 1.
Early mortality, duration of postoperative ventilator use, and ICU stay were not significantly different between groups after balancing with sIPTW (Table S1). Independent predictors of early mortality were higher NYHA functional class, elevated CVP, longer CPB time, and preoperative total bilirubin >2 mg/dL (Table 4). Longer aortic cross-clamp time, however, was not a risk factor of early mortality.
| P-value | OR or HR (95% CI) | |
|---|---|---|
| Early mortality (all patients) | ||
| TVR | 0.065 | 0.200 (0.036–1.108) |
| NYHA functional class | 0.007 | 3.555 (1.413–8.944) |
| Initial CVP at operating room (mmHg) | 0.029 | 1.135 (1.013–1.272) |
| CPB time (min) | 0.001 | 1.013 (1.006–1.020) |
| Preop. total bilirubin >2 mg/dL | 0.046 | 3.958 (1.026–15.269) |
| Overall mortality (all patients) | ||
| TVR | 0.736 | 1.093 (0.652–1.831) |
| Age | <0.001 | 1.063 (1.040–1.087) |
| Preop. inotrope use | 0.040 | 1.652 (1.023–2.668) |
| Preop. LVEF <40% | 0.001 | 2.671 (1.514–4.712) |
| Preop. Hb <10 g/dL | 0.004 | 2.002 (1.256–3.190) |
| Cardiac death (all patients) | ||
| TVR | 0.687 | 0.877 (0.462–1.664) |
| Age | <0.001 | 1.056 (1.031–1.082) |
| Preop. LVEF <40% | <0.001 | 3.018 (1.627–5.600) |
| Preop. Hb <10 g/dL | 0.068 | 1.676 (0.963–2.916) |
| MACE (all patients) | ||
| TVR | 0.474 | 0.841 (0.523–1.351) |
| Hypertension | 0.008 | 2.033 (1.206–3.426) |
| Finally replaced left side valves | ||
| No left-sided valve replacement | Ref. | |
| MVR status | 0.218 | 1.409 (0.816–2.432) |
| AVR status | 0.034 | 2.899 (1.086–7.737) |
| DVR status | 0.017 | 1.997 (1.131–3.528) |
| TR recurrence† (TVr group only) | ||
| Ring annuloplasty | 0.001 | 0.245 (0.109–0.550) |
| Finally replaced left side valves | ||
| No left-sided valve replacement | Ref. | |
| MVR status | 0.205 | 0.549 (0.217–1.389) |
| AVR status | 0.007 | 6.518 (1.661–25.573) |
| DVR status | 0.286 | 1.679 (0.648–4.350) |
| TR grade at discharge | <0.001 | 1.593 (1.278–1.984) |
†Defined as moderate or greater degree of TR on echocardiography during follow-up. EF, ejection fraction; HR, hazards ratio; LV, left ventricular; MACE, major adverse cardiovascular events; TR, tricuspid regurgitation. Other abbreviations as in Tables 1,2.
There was no significant difference between groups in the incidence of other major late complications (Table 3), including reoperation, and anticoagulation-related complication. There were 3 prosthetic tricuspid valve thromboses in the TVR group during follow-up, and all of them were successfully treated with medical thrombolysis.
Kaplan-Meier analysis was performed to compare overall mortality, cardiac death, MACE, and TR recurrence (Figure 2A,C,E). None of those comparisons were significantly different between the TVR and TVr groups: 10-year overall survival: TVR, 72±6% and TVr, 70±4% (P=0.532); 10-year freedom from cardiac mortality: TVR, 76±6% and TVr, 77±3% (P=0.715); 10-year freedom from MACE: TVR, 64±7% and TVr, 66±4% (P=0.593); and 10-year freedom from TR recurrence (≥moderate): TVR, 90±4% and TVr, 87±3% (P=0.127).
(A,C,E) Kaplan-Meier and (B,D,F) weighted Kaplan-Meier analysis (i.e., stabilized inverse probability treatment weighting) for (A,B) overall survival, (C,D) freedom from cardiac death, and (E,F) freedom from major adverse cardiovascular events (MACE) for tricuspid valve replacement (TVR) vs. tricuspid valve repair (TVr).
The 15-year freedom from TR recurrence was 90±4% (TVR) and 75±7% (TVr). Although this difference did not reach statistical significance, the steep divergence of the 2 curves after 10 years was obvious (Figure 3A). TR recurrence in the TVR group was mainly attributable to tissue valve use (10-year freedom from TR recurrence [≥moderate]: mechanical TVR, 96±3% and tissue TVR, 59±25%; P=0.007).
Freedom from tricuspid regurgitation (TR) recurrence (Kaplan-Meier analysis) for (A) tricuspid valve replacement (TVR) vs. tricuspid valve repair (TVr); (B) annuloplasty with vs. without a ring in the TVr group; and (C) TVR vs. TVr (ring annuloplasty only). TR recurrence was defined as detection of TR of moderate or greater degree during follow-up.
Overall mortality, cardiac death, MACE, and TR recurrence were compared after balancing groups using sIPTW (Figure 2B,D,F). On weighted Cox regression analysis there was no significant difference between groups for the 4 endpoints overall mortality (hazard ratio [HR], 1.11; 95% CI: 0.71–1.74, P=0.649), cardiac death (HR, 1.04; 95% CI: 0.62–1.76, P=0.870), MACE (HR, 0.94; 95% CI: 0.61–1.46, P=0.786), and TR recurrence (HR, 0.91; 95% CI: 0.46–1.79, P=0.788).
Risk factor analyses are summarized in Table 4. Older age, preoperative i.v. inotrope use, preoperative LV dysfunction (EF <40%), and preoperative hemoglobin <10 g/dL were significant risk factors for overall mortality. Older age, preoperative LV dysfunction, and preoperative hemoglobin <10 g/dL were independent predictors of cardiac death. TVR was not a significant risk factor for any of the 3 long-term endpoints.
Subgroup AnalysisIn the TVR group, overall survival, freedom from cardiac death, and freedom from MACE were not significantly different between mechanical and tissue valve groups (log-rank, P=0.052, 0.103, and 0.909, respectively). On multivariable analysis, the use of a mechanical valve was not an independent predictor of overall survival, freedom from cardiac death, or MACE in the TVR group.
As mentioned earlier, the difference in TR recurrence between the TVR and TVr groups did not reach statistical significance. In the TVr group, however, TVr with ring annuloplasty had better freedom from TR recurrence than suture annuloplasty (P=0.004, Figure 3B). Furthermore, on multivariable analysis for TR recurrence in the TVr group, suture annuloplasty, AVR status (compared with tricuspid valve surgery only), and higher postoperative TR grade at discharge were risk factors for TR recurrence (Table 4).
Given that there was a trend for suture annuloplasty to have the higher incidence of TR recurrence, we additionally compared TR recurrence between the TVR and TVr groups excluding suture annuloplasty (Figure 3C). On analysis, TVr with ring annuloplasty had equivalent TR recurrence to TVR (P=0.935). Also, overall mortality (P=0.179), freedom from cardiac death (P=0.301), and freedom from MACE (P=0.567) were not significantly different between groups (TVR vs. TVr excluding suture annuloplasty).
There are 3 main findings of the present study. First, TVR was not related to an increased risk of early mortality and morbidity. Second, there was no difference in long-term outcomes such as overall survival, cardiac death, and MACE between TVR and TVr groups. Third, TVr with suture annuloplasty was related to an increased risk of TR recurrence.
For the last few decades, surgeons and physicians have tried to avoid TVR in patients with severe TR given the significantly worse clinical outcomes of TVR.1–5,11 Due, however, to the difficulty in compiling a sufficient number of cases from which to draw a reliable conclusion, there are still only a limited number of papers comparing TVR with TVr,1,4,6 and those studies dealt with tricuspid valve surgery performed in different patient populations with varying severity of TR.
The greatest limitation in the comparison of clinical outcomes between TVR and TVr is that a randomized controlled trial is fundamentally impossible. In studies that compared TVR with TVr using adjustment or matching, the results showed an early mortality that was too high in the TVR group, or the number of patients was too small.1,4,12 Surgeons always do their best to perform TVr instead of TVR. Therefore, TVR is frequently the result of failed repair attempts or extensive valve deformity that contraindicates repair. In these circumstances, the comparison between TVR and TVr has been accompanied by a certain degree of skewing. Even worse, reports on the inferiority of TVR have made surgeons more reluctant to do TVR. Hence, TVR has become an operation that is performed only in the worst situations. Kim et al recently reported that there was no meaningful difference between TVR and TVr as a treatment for isolated severe TR.13 This suggests that the clinical outcomes of tricuspid valve surgery are mainly related to preoperative characteristics, not to the type of surgery itself. To date, some studies have already drawn the conclusion that there appears to be no significant difference in clinical outcomes between the 2 approaches,4,7,8 but the study period ranged from the 1970 s, or the operative mortality was high from the viewpoint of current standard practice. In order to compare the outcomes of TVR and TVr, we performed balancing using sIPTW. Major early and long-term outcomes were not significantly different between groups in both the real-world and sIPTW comparisons.
According to Singh et al, the discrepancy in overall survival between TVR and TVr is mainly attributable to a difference in early mortality.1 Therefore, we have focused on reducing early mortality to achieve better outcomes in the TVR group. In order to do that, deliberate perioperative management was of the greatest importance.
Although RV function can be reflected in a number of ways clinically (pleural effusion, ascites, peripheral edema etc.), CVP is one of the objective and measurable parameters that reflects RV preload and function.14 In the present study, higher CVP was related to an increased risk of early mortality, which is meaningful because CVP was an independent predictor even after adjusting for the effect of increased total bilirubin. Both increased CVP and higher total bilirubin (>2 mg/dL) were independent risk factors of early mortality, respectively.
Longer CPB time was another risk factor for early mortality in the present study. Based on our experience, we did not spend too much time on TVr attempts. Instead, we aggressively corrected the left-side valve problem (e.g., increased transvalvular pressure gradient in aortic prosthesis), because suboptimal left-side valve hemodynamics exacerbate late progression of TR in patients who undergo TVr.15 Also, TR grade at late follow-up after mechanical double valve replacement correlates with transaortic mean pressure gradient.16
Additionally, at the present hospital, full use of conventional and modified ultrafiltration during and after CPB has always been made. This strategy was based on reports showing that body fluid removal and hemoconcentration improve ventricular function and reduce pulmonary vascular resistance in patients with congestive heart failure.17 We believe that immediate reduction of RV afterload at the time of CPB weaning is the most crucial point in the perioperative management of TV surgery. This management protocol may prevent renal dysfunction and excessive use of inotropes.
Every effort must be made to prevent postoperative bleeding, a common and serious problem in multiple reoperation patients with hepatic dysfunction. These efforts include preoperative complete recovery of normal PT-INR, aggressive ultrafiltration, adequate transfusion of blood products, meticulous bleeding control and early detection of postoperative abnormal bleeding tendency. We frequently observed elevated PT-INR in the immediate postoperative period even after preoperative complete recovery. Optimal body weight should be maintained postoperatively. Some patients with severe right-sided heart failure had >5 kg excessive body fluid due to pleural effusion, ascites and tissue edema.
Although the majority of the present patients had concomitant left-side valve surgery, the present approach led to early mortality as low as 3–4% in both the TVR and TVr groups, which is lower compared with the 10–25% in previous reports.11,18–20 The present 360 patients, however, include all patients treated since the establishment of the present institution to the end of 2012, with >15 years of follow-up. Some of our unique results in TVR were published earlier.15,21,22 Possible reasons for the low mortality in the current study are relatively younger age, early referral of TV surgery, lower threshold for choosing TVR rather than TVr, and improvement of perioperative management including aggressive ultrafiltration.
We also examined 2 sub-issues in TVR and TVr: comparison between mechanical and tissue valves in TVR, and comparison between ring and suture annuloplasty in TVr. Neither comparison identified any difference in major endpoints. Considering that the 10-year survival in the TVR group was approximately 70%, it is still questionable whether TVR with mechanical valve is reasonable or not. Patients who were implanted with tissue valve in the tricuspid position might have an extremely high operative risk in the future. In addition, TVr using ring annuloplasty was related to a lower risk of TR recurrence. These findings are consistent with previous reports.19,23–25
In addition to the limited number of patients, the present study had a number of other limitations. First, baseline patient characteristics were different between the TVR and TVr groups. As noted here, the choice between TVR and TVr could not be randomized due to different indications and varying clinical scenarios. Second, we did not include preoperative parameters that represent RV function due to incomplete data (e.g., tricuspid annulus size, RV function). Although we began to use magnetic resonance imaging (MRI) for evaluation of RV dysfunction preoperatively, the number of patients in this study was too small to evaluate. Therefore, we used initial CVP data after anesthetic induction, and noted a significant relationship between CVP and early mortality. Third, we performed suture annuloplasty in approximately 30% of the TVr patients, and this was done mainly in the early period of the study. This could influence the results of TR recurrence in the TVr group. Fourth, the majority of the present patients had multiple valve disease and initial rheumatic etiology, which is rare in Western countries. Further studies to compare late outcomes of TVR and TVr in patients with degenerative disease are required.
Compared with TVr, TVR was not a risk factor for early or long-term outcomes in patients with severe TR. Even with several disadvantageous preoperative characteristics such as longer CPB time, more complex concomitant procedures, and higher proportion of previous cardiac surgeries, TVR was equivalent to TVr in outcomes. In order to prevent TR recurrence, ring annuloplasty is strongly recommended in TVr procedures. TVR is still a valid option when the tricuspid valve has hostile pathology or the patient is in a critical condition.
Joo Min Hwang (RN) helped in the collection and management of the data. Biomedical statistician Joong Hyun Ahn advised on statistical analysis.
None.
Supplementary File 1
Table S1. sIPTW-adjusted data
Table S2. Case summary of early mortality (≤30 days)
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-16-0961
