Author + information
- Received September 12, 2016
- Revision received October 26, 2016
- Accepted November 28, 2016
- Published online March 6, 2017.
- John A. Wells IV, BS,
- Jose F. Condado, MD,
- Norihiko Kamioka, MD,
- Andy Dong,
- Andrew Ritter, BS,
- Stamatios Lerakis, MD,
- Stephen Clements, MD,
- James Stewart, MD,
- Bradley Leshnower, MD,
- Robert Guyton, MD,
- Jessica Forcillo, MD,
- Ateet Patel, MD,
- Vinod H. Thourani, MD,
- Peter C. Block, MD and
- Vasilis Babaliaros, MD∗ ()
- Division of Cardiology, Emory University School of Medicine, Atlanta, Georgia; and Division of Cardiothoracic Surgery, Emory University School of Medicine, Atlanta, Georgia
- ↵∗Address for correspondence:
Dr. Vasilis Babaliaros, Emory University Hospital F606, 1364 Clifton Road, Atlanta, Georgia 30322.
Objectives The aim of this study was to compare outcomes of transcatheter intervention (TI) versus surgical intervention (SI) for paravalvular leak (PVL).
Background Data comparing the treatment of PVL with TI and SI are limited.
Methods A retrospective cohort study was conducted comparing baseline characteristics, procedural details, and 1-year survival in consecutive patients who underwent TI or SI for moderate or greater PVL from 2007 to 2016. The primary outcome was a composite of death, reintervention for PVL, or readmission for congestive heart failure–related symptoms at 1 year.
Results Of 114 patients, 56 underwent TI and 58 underwent SI. PVL locations were mitral, aortic, and pulmonary in 69 (60.5%), 39 (34.2%), and 6 (5.3%) patients, respectively. At baseline, TI patients were older (age 71 vs. 62 years; p = 0.010) and had fewer cases of active endocarditis (0.0% vs. 25.9%, p < 0.001) than SI patients. The TI group had a shorter post-operative stay (4 vs. 8 days; p < 0.001), a shorter intensive care unit stay (0 vs. 3 days; p < 0.001), and fewer readmissions at 30 days (8.9% vs. 25.9%; p = 0.017). There were no differences in the primary endpoint (TI 33.9% vs. SI 39.7%; p = 0.526) or 1-year survival (TI 83.9% vs. SI 75.9%; p = 0.283) between groups.
Conclusions In this study, TI for PVL closure had comparable 1-year clinical outcomes with SI, even after adjusting for differences in baseline characteristics, with less in-hospital morbidity and 30-day rehospitalization. Although further study is needed, these findings support the increased implementation of TI for PVL closure at experienced institutions.
Paravalvular leak (PVL) is a frequent and serious complication of both surgical and transcatheter valve replacement, occurring in 5% to 17% of surgical prosthetic valves and at even higher rates following transcatheter aortic valve replacement (TAVR) (1–3). Untreated PVL may lead to congestive heart failure and hemolytic anemia, and recent research has confirmed that PVL is associated with poor clinical outcomes (1,3–5). The PARTNER (Placement of Aortic Transcatheter Valves) II trial of TAVR in intermediate-risk patients found a significant relationship between moderate or greater PVL and late mortality after both surgical valve replacement and TAVR (3).
Until recently, surgical intervention (SI) with valve replacement or repair has been the standard approach to patients with PVL (4,6,7). SI has been successful for leak reduction, but many patients with PVL are high-risk surgical candidates predisposed to perioperative morbidity (4,6–10), and complicated cardiac surgery has been associated with increased costs and high resource use (11,12). Transcatheter intervention (TI) for PVL closure is increasingly performed as an alternative to SI (13,14). Several descriptive series and meta-analyses have shown that TI can reduce the severity of PVL and its symptoms, with low rates of procedural complications and morbidity (10,15–18). Despite its growing application, there are limited data comparing the effectiveness of TI with that of SI (8,10,16).
We hypothesized that outcomes of TI would be similar to those of SI, and thus we compared 1-year outcomes in patients undergoing TI or SI for PVL at a single U.S. academic center.
We reviewed all patients who underwent either TI or SI for clinically significant PVL at our institution between January 2007 and June 2016. Clinically significant PVL was defined as moderate or greater PVL with symptoms of heart failure and/or hemolysis (3,7,18). Standard definitions for grading PVL severity were applied (19–21). Hemolytic anemia was defined by clinical documentation of symptoms in conjunction with laboratory evidence of anemia (hemoglobin <15 mg/dl in men and <13 mg/dl in women) and hemolysis (lactate dehydrogenase >500 U/l or haptoglobin <20 mg/dl) (18). The Society of Thoracic Surgeons (STS) Predicted Risk of Mortality (PROM) score is reported for valve replacement directly from the STS online risk calculator (22). Other patient factors and outcomes were defined according to the STS Adult Cardiac Surgery Database Data Specifications, version 2.81, and the Valve Academic Research Consortium 2 consensus document, as appropriate (23,24).
For patients with multiple PVL interventions, the first intervention within the study period was counted as the index procedure. For all cases, baseline characteristics, procedural details, and procedural outcomes up to 1 year were collected from the electronic medical record. Patients were contacted by telephone follow-up for data not in the electronic medical record.
The study’s primary endpoint was the composite of all-cause mortality, readmission for congestive heart failure, and reintervention for PVL at 1 year. Secondary endpoints included 30-day clinical success, post-operative complications, length of hospital stay, length of intensive care unit stay, and major morbidity. Clinical success was defined as mild or less PVL on echocardiography at the time of 1-month follow-up and 30-day hospital-free survival (18). If 1-month echocardiographic data were unavailable, post-procedural images were used. Major morbidity was defined as a composite of serious in-hospital events, including stroke, renal failure requiring dialysis, seizure requiring antiepileptic treatment, additional operation (e.g., tracheostomy, gastrostomy tube, mediastinal debridement), prolonged ventilation, and cardiac arrest.
Patients were selected for TI on the basis of their surgical risk, leak characteristics, and the availability of TI at our institution. Patients with active endocarditis were not considered for TI. Patients not eligible for SI because of high surgical risk may have been considered for TI. Access route (transfemoral arterial, transapical, transseptal, and transfemoral venous) and device type were determined pre-operatively according to the valve location, arterial anatomy, and leak characteristics. Since the initial TI for PVL closure, our institution has modified the closure strategy. Early in our experience, the PVL space was closed using the body of the closure device in the center of the defect (waist technique) (Figure 1A). In October 2014, we transitioned to closing the PVL space using the disk of the device as a cap to cover the space from the sewing ring to the great vessel or myocardial wall (cap technique) (Figure 1B). An experienced interventional cardiologist (P.C.B., V.B.) performed all procedures.
Surgical technique (replacement vs. repair) was determined according to PVL cause, leak characteristics, and surgeon preference. There was no STS cutoff or risk limitation in the selection of patients, though some patients were excluded from SI by heart team analysis of open surgical risk. The preferred technique at our institution was valve explantation and replacement. Standard surgical techniques were used for repair or replacement of the valve (25). Experienced cardiothoracic surgeons performed all procedures (V.H.T., R.G., B.L.).
Categorical variables are reported as number (percentage) and numeric variables either as median (interquartile range) for highly skewed variables or as mean ± SD for normally distributed ones. Baseline characteristics were compared between TI and SI patients using either chi-square or Fisher exact tests for categorical variables, when applicable, and Student t tests or Mann-Whitney U tests for normally or non-normally distributed continuous variables, respectively. The distribution of continuous variables was determined using the Shapiro-Wilk test. Cumulative incidence curves for the occurrence of the composite outcome at 1 year were created and compared using the log rank test. A Cox proportional hazards model was then created to compare the effects of TI, patient age, active endocarditis, STS PROM score, and moderate or greater PVL at 1 month on the primary outcome. The covariates were evaluated for any significant time interactions. A 2-tailed p value of ≤0.05 was considered to indicate statistical significance in this study. We estimated a sample size of n = 100 and assumed an event rate of 30%, resulting in 80% power to detect a difference of 24.3% in the primary endpoint. Statistical analyses were performed using IBM SPSS version 23 (IBM, Armonk, New York).
The Emory University Institutional Review Board approved this study.
We reviewed 133 consecutive patients who underwent either TI (n = 68) or SI (n = 65) for PVL during the study period. TI for PVL became available at our institution in May 2008. Eleven SI cases occurred prior to this date.
Twelve patients in the TI group were excluded. Two were excluded because the procedure was aborted without PVL closure, 9 were treated with valve-in-valve after TAVR, and 1 was treated only with post-dilation after TAVR. Of the 56 patients included in the TI cohort, 54 had undergone prior prosthetic valve procedures and subsequently developed PVL, and 2 developed PVL during TAVR and underwent concurrent TI. The baseline valve types included surgical mechanical valves (n = 24), surgical bioprosthetic valves (n = 26), and TAVR (n = 6).
Seven patients in the surgical group were excluded because of intraoperative findings of central regurgitation without PVL. All 58 patients in the SI cohort had undergone valve replacement during previous hospitalizations: surgical mechanical valves (n = 21), surgical bioprosthetic valves (n = 36), or TAVR (n = 1).
Baseline characteristics are presented in Table 1. Compared with SI patients, TI patients were older, with fewer cases of prior as well as active endocarditis. There were no differences between groups in the frequency of hemolytic anemia, the distribution of New York Heart Association classes, the number of emergent procedures, or median STS PROM score. There were trends toward more prior TAVR and mechanical valves in the TI group. Baseline echocardiographic parameters were similar between groups. A majority of all patients (71.9%) had severe PVL on baseline echocardiography, and the remainder (28.1%) had moderate PVL, with no differences between TI and SI (p = 0.473).
The procedural details are presented in Table 2. The majority of TI cases were performed percutaneously from the femoral artery or vein. Almost one-half of these patients were treated using the cap technique. The preferred surgical technique was replacement over repair. Room time was longer with SI patients, with less blood transfusion and a trend toward less intraoperative hemodynamic support in the TI cohort.
Hospital, 30-day, and 1-year outcomes are presented in Table 3. Among hospital outcomes, there was less major morbidity among patients with TI, as well as shorter post-operative hospital stay and intensive care unit stay. TI patients also required fewer transfusions in the post-operative period.
At 30 days, clinical success had been achieved in a majority of patients, with no difference between TI and SI cohorts. There were no differences in PVL reduction, while patients undergoing TI had significantly fewer readmissions. Thirty-day mortality was equivalent between groups, as were rates of stroke, wound infection, renal failure, pneumonia, and new pacemaker placement.
At 1 year, there were no differences between groups in the primary endpoint. One-year mortality was similar between groups. A Kaplan-Meier survival function based on the primary endpoint at 1-year was equivalent (Figure 2A) and remained equivalent after excluding patients with active endocarditis (Figure 2B). Multivariate analysis showed that high (≥8%) STS PROM score was predictive of the primary endpoint (hazard ratio: 4.941; 95% confidence interval: 2.272 to 10.746; p < 0.001). In the same model, moderate or greater PVL at 30 days did not reach significance (hazard ratio: 2.625; 95% confidence interval: 0.858 to 8.037; p = 0.091). Intervention (TI vs. SI), age, and active endocarditis (hazard ratio: 0.96; 95% confidence interval: 0.33 to 2.77; p = 0.998) had no predictive power. Exclusion of patients with active endocarditis did not affect the primary endpoint (p = 0.921).
In an intention-to-treat analysis, there were no differences in 30-day clinical success or the primary endpoint between the TI and SI groups (Table 4). In this analysis, SI patients had better PVL reduction but higher rates of readmission at 30 days. TI patients had higher rates of reintervention at 1 year.
In a separate analysis of the TI cohort divided according to waist versus cap treatment (Table 5), there were no significant differences between groups. However, there were trends toward greater PVL reduction and reduced rates of reintervention in the cap cohort.
We found that TI for PVL has equivalent 1-year clinical outcomes to SI, with significantly less resource use, bleeding, and perioperative morbidity. We also report our preliminary experience with new techniques of TI, with encouraging results. These findings add to existing evidence regarding the effectiveness of TI (7,8,17,18,26) and provide quality data supporting TI as a first treatment strategy for PVL.
Previous reports of transcatheter PVL closure have used varying definitions of procedural success, making comparisons across studies difficult. A 2015 meta-analysis of 12 nonrandomized studies of transcatheter PVL closure defined procedural success as immediate PVL reduction by ≥1 degree of severity, regardless of 30-day clinical outcomes (15). In that study, the global rate of procedural success was 76.5%, and the global mortality rate was 22.7%, with mean follow-up ranging from 1 to 41 months (15). We used a stringent definition of clinical success, which incorporated both procedural success (PVL reduction to mild or less severity) and outcomes at 30 days. In the largest U.S. series to date, Sorajja et al. (18) defined success as reduction in PVL to mild or less severity in the absence of major procedural complications and reported a success rate of 77%. These results were based on an intention-to-treat design. In our intention-to-treat analysis, we found a procedural success rate of 77% and a clinical success rate of 72% based on 30-day outcomes, which is a comparable result.
Results of SI for PVL closure are not well reported (4,6,7). The most recent study, from Taramasso et al. (7), excluded patients with active endocarditis and reported acute procedural success and mortality rates of 98% and 10.7%, respectively. Other surgical series have reported similar results, with short-term mortality of <10% and high rates of leak resolution (4,6). The SI cohort in our study had comparable outcomes with these series, with 30-day rates for procedural success and mortality of 95% and 7%, respectively, despite the inclusion of patients with active endocarditis. Thirty-five percent of patients in the Taramasso et al. (7) study were treated with valve replacement rather than repair, slightly lower than other series we reviewed, which reported rates of valve replacement ranging from 40% to 52% (4,6,7). Our rate of valve replacement was 90%.
There have been 3 studies comparing TI and SI directly (8,10,16). Our study is the first to report a comparison of TI and SI at an academic, U.S. institution and the first to report a comparison of 2 cohorts with more than 50 patients. In a recent study published by Angulo-Llanos et al. (8) in 2016, patients undergoing TI had better long-term clinical improvement than SI patients, with an odds ratio of clinical improvement of 9.1:1. However, the very poor outcomes in the SI cohort limit the utility of their comparative analysis. In-hospital mortality was >30% among SI patients, despite a mean STS risk of only 4.6%, and 67.6% of surgical patients experienced the composite endpoint at 1 year (8). Our results improve upon the limitations of that study, as we found distinct advantages to TI in early clinical outcomes (i.e., resource use, blood loss, morbidity, and readmission) and no differences in long-term clinical outcomes compared with the SI cohort. Strengthening our comparison, the SI cohort in our study had low rates of 30-day mortality and composite endpoint occurrence compared with historical controls, despite the high ratio of valve replacement to repair.
Just as we report improvements in surgical strategies for PVL, there have been improvements in transcatheter techniques. Two cases from early in our experience were aborted (interference with mechanical valve [n = 1] and inability to cross defect [n = 1]) that most likely would have been successfully completed with current experience. Additionally, placement of the Amplatzer Vascular Plug II (St. Jude Medical, St. Paul, Minnesota) has changed in our practice so that the disk of tightly woven nitinol is used to cover the defect (cap technique) rather than using the body of the device to fill the defect (waist technique). Our experience is that the cap technique has improved PVL reduction and reintervention rate, but the present study is underpowered to show statistical differences. The Amplatzer Vascular Plug III may further improve the ease and success of TI for PVL (27), strengthening the argument for TI as the initial approach to PVL.
This was a retrospective study with inherent limitations. Although the SI cohort included patients with active endocarditis, this variable was not predictive of outcomes in the multivariate model, and the exclusion of these patients did not affect the Kaplan-Meier analysis or the primary endpoint. We performed an as-treated analysis because at the start of the study period, SI was a mature technique and TI was not. As part of this analysis, we excluded 2 TI patients whose procedures were aborted, and 4 cases crossed treatment arms after a failed first attempt (3 in the TI group, 1 in the SI group). In a separate intention-to-treat analysis (Table 4), we found no differences in the primary endpoint or 30-day clinical success. The study was not powered to detect small differences in the primary outcome. Finally, our analysis was limited to 1 year of follow-up, and the impact of residual PVL may be underestimated.
In the largest study of its kind to date, we found that TI and SI for PVL are similar with respect to 1-year mortality, reintervention rate, and readmission for congestive heart failure. With improving techniques and experience, TI may be a better initial treatment strategy, as resource use and perioperative morbidity are significantly improved. Further study is needed on this topic.
WHAT IS KNOWN? It is known that moderate or greater PVL is associated with poor clinical outcomes, but few studies have compared transcatheter techniques of repairing PVL with traditional surgery.
WHAT IS NEW? Our study provides new evidence regarding outcomes of TI versus SI for PVL at a U.S. institution. In a small sample size, we found similar 1-year outcomes between TI and SI, with significantly less morbidity and resource use.
WHAT IS NEXT? Larger, multicenter studies are needed to detect differences between TI and SI and to better define the clinical indications for each.
Dr. Block is a stockholder in Direct Flow Medical and a consultant for Medtronic and St. Jude Medical. Dr. Thourani is a consultant for Edwards Lifesciences, Maquet, St. Jude Medical, Sorin, and Direct Flow Medical; and is also a cofounder and stockholder of Apica. Dr. Babaliaros is a consultant for Direct Flow Medical and St. Jude Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- Predicted Risk of Mortality
- paravalvular leak
- surgical intervention
- Society of Thoracic Surgeons
- transcatheter aortic valve replacement
- transcatheter intervention
- Received September 12, 2016.
- Revision received October 26, 2016.
- Accepted November 28, 2016.
- 2017 American College of Cardiology Foundation
- Genoni M.,
- Franzen D.,
- Vogt P.,
- et al.
- Angulo-Llanos R.,
- Sarnago-Cebada F.,
- Rivera A.R.,
- et al.
- Said S.M.
- Taramasso M.,
- Maisano F.,
- Latib A.,
- et al.
- Hourihan M.,
- Perry S.B.,
- Mandell V.S.,
- et al.
- Reed G.W.,
- Tuzcu E.M.,
- Kapadia S.R.,
- Krishnaswamy A.
- Ruiz C.E.,
- Jelnin V.,
- Kronzon I.,
- et al.
- Sorajja P.,
- Cabalka A.K.,
- Hagler D.J.,
- Rihal C.S.
- Lancellotti P.,
- Tribouilloy C.,
- Hagendorff A.,
- et al.
- Zamorano J.,
- Goncalves A.,
- Lancellotti P.,
- et al.
- Zoghbi W.A.,
- Chambers J.B.,
- Dumesnil J.G.,
- et al.
- Shahian D.M.,
- Jacobs J.P.,
- Edwards F.H.,
- et al.
- Kappetein A.P.,
- Head S.J.,
- Genereux P.,
- et al.
- Cohn L.H.
- Saia F.,
- Martinez C.,
- Gafoor S.,
- et al.