Percutaneous Device Closure of Paravalvular LeakClinical Perspectives
Combined Experience From the United Kingdom and Ireland
This article has a correction. Please see:
Background: Paravalvular leak (PVL) occurs in 5% to 17% of patients following surgical valve replacement. Percutaneous device closure represents an alternative to repeat surgery.
Methods: All UK and Ireland centers undertaking percutaneous PVL closure submitted data to the UK PVL Registry. Data were analyzed for association with death and major adverse cardiovascular events (MACE) at follow-up.
Results: Three hundred eight PVL closure procedures were attempted in 259 patients in 20 centers (2004–2015). Patient age was 67±13 years; 28% were female. The main indications for closure were heart failure (80%) and hemolysis (16%). Devices were successfully implanted in 91% of patients, via radial (7%), femoral arterial (52%), femoral venous (33%), and apical (7%) approaches. Nineteen percent of patients required repeat procedures. The target valve was mitral (44%), aortic (48%), both (2%), pulmonic (0.4%), or transcatheter aortic valve replacement (5%). Preprocedural leak was severe (61%), moderate (34%), or mild (5.7%) and was multiple in 37%. PVL improved postprocedure (P<0.001) and was none (33.3%), mild (41.4%), moderate (18.6%), or severe (6.7%) at last follow-up. Mean New York Heart Association class improved from 2.7±0.8 preprocedure to 1.6±0.8 (P<0.001) after a median follow-up of 110 (7–452) days. Hospital mortality was 2.9% (elective), 6.8% (in-hospital urgent), and 50% (emergency) (P<0.001). MACE during follow-up included death (16%), valve surgery (6%), late device embolization (0.4%), and new hemolysis requiring transfusion (1.6%). Mitral PVL was associated with higher MACE (hazard ratio [HR], 1.83; P=0.011). Factors independently associated with death were the degree of persisting leak (HR, 2.87; P=0.037), New York Heart Association class (HR, 2.00; P=0.015) at follow-up and baseline creatinine (HR, 8.19; P=0.001). The only factor independently associated with MACE was the degree of persisting leak at follow-up (HR, 3.01; P=0.002).
Conclusion: Percutaneous closure of PVL is an effective procedure that improves PVL severity and symptoms. Severity of persisting leak at follow-up is independently associated with both MACE and death. Percutaneous closure should be considered as an alternative to repeat surgery.
Paravalvular leak (PVL) is a common and challenging problem that has existed since prosthetic cardiac valves were first implanted.1 PVL occurs in 7% to 17% of mitral valve replacements (MVRs) and 5% to 10% of aortic valve replacements (AVRs)2,3 and can be associated with disabling symptoms related to heart failure or hemolysis.4 Repeat surgery to repair PVL is associated with significant mortality and morbidity.5–8 Tissues around PVL are often calcified and friable, and, therefore, leaks may recur.5,6 Conservative management of mild to moderate leaks is also associated with high adverse event rates because leaks tend to deteriorate over time.9,10
Percutaneous device closure of PVL was first reported in 1992.11 Since then there have been multiple small case series reported describing the various techniques with transvenous, transarterial, and transapical access.12–14 The largest series so far has reported outcomes in 126 patients from a single center of excellence in this procedure.15 However, so far there has been no systematic report of real-world experience from a national registry. We report the combined experience of percutaneous PVL closure from the United Kingdom (UK) and Ireland.
Percutaneous device implantation of PVL has been undertaken in the UK since 2004. Units report numbers of cases to the British Cardiovascular Intervention Society audit office and figures are published annually.16 All centers (n=20) where percutaneous PVL activity has been reported in the UK and Ireland were contacted to participate in this study, and all agreed to participate. Anonymized data were acquired from medical and electronic records regarding patient demographics, clinical features, preprocedure clinical condition, echocardiographic features, procedural characteristics, procedural complications, in-hospital outcomes, adverse events during follow-up, and vital status. The protocol was approved by the Brighton and Sussex University Hospitals Review Board. No patient consent was required because this was a retrospect registry with anonymized data according to local policy.
The majority of procedures were to treat PVL of surgical valves with only a minority to treat PVLs of transcatheter valves (transcatheter aortic valve replacement) or annuloplasty rings. Most procedures were done under general anesthesia with transesophageal echocardiography guidance (TEE). Three-dimensional (3D) TEE contributed to preprocedural and procedural planning (Figures 1 and 2). PVLs were often multiple and had complex morphology.
Mitral valve PVLs were generally closed via a transvenous route with transseptal puncture in the anterograde direction. Real-time 3D-TEE images were used where available to guide guidewire and catheter placement across the defect (Figure 1). A steerable transseptal sheath (eg, Agilis, St Jude Medical) was often useful in locating the leak. Through this a smaller catheter (multipurpose or Judkins Right 4 guiding or diagnostic catheter) was passed in a telescoping fashion to provide flexibility and support. Hydrophilic guidewires and delivery sheaths (Flexor Shuttle [Cook Medical] or Destination [Terumo]) were often used. Occasionally, arteriovenous loops were required to support passage of the delivery sheath. Some operators preferred to cross mitral PVLs in a retrograde direction either via transapical access (percutaneous or surgical) or via femoral artery access through the aortic valve, especially for PVLs close to the atrial septum.
Aortic valve PVLs were generally crossed in a retrograde direction with either femoral or radial arterial access. Crossing of the leak was facilitated by fluoroscopy and echocardiography. Devices to close PVLs were selected by physician preference, because, until recently, there were no CE marked devices for this procedure (Figure 3).
Statistical analysis was performed by using SPSS v22.0 (IBM Corp). Where analysis pertained to patient-related outcomes, only final attempts at percutaneous PVL closure were included in the analysis. Normally distributed data are presented as mean±standard deviation and non-Gaussian data are presented as median (first and third quartile [Q1–Q3]). Categorical data are presented as frequency (percentage). Clinical and procedural variables were tested for an association with death and major adverse cardiovascular events (MACE) during follow-up using univariable Cox proportional hazard regression (Figures 4 and 5). For continuous variables, the hazard ratio (HR) represents the increase in hazard per unit rise in that variable. Ordinal variables include baseline leak severity, New York Heart Association (NYHA) class, left ventricular function, and presentation. For the categorical variables, the comparison is between the presence or absence of that factor. Those factors with P<0.1 on univariable analysis were entered into a multivariable model. MACE was defined as a composite of death, stroke, late embolization of device, infective endocarditis, new hemolysis requiring transfusion, new valve leaflet interference, repeat procedure, or valve surgery. Changes in PVL and NYHA class (ordinal variables) were assessed by using Wilcoxon signed rank test.
Three hundred eight PVL closure procedures were attempted in 259 patients in 20 centers (2004–2015). Patients’ clinical variables are detailed in Table 1. Mean age was 67±13 years. The majority of patients were male (72%). The indication for closure was heart failure (80%), hemolysis (16%), or both (2%). The procedure was elective (81%), urgent (17.4%), and emergent (1.6%). The median time from last valve surgery to percutaneous closure attempt was 4.7 (1.4–8.9) years, and 27% of patients had had >1 previous valve surgery.
Procedural variables are shown in Table 2. Devices were successfully implanted in 91% of patients. Device delivery sheath approach was via the transradial (7%), femoral arterial (52%), femoral venous (33%), and transapical (7%) routes. More than 1 closure procedure was required in 19% of patients. The target valve was surgical in 94% of cases (mitral 44%, aortic 48%, both 2%, pulmonic 0.4%). Transcatheter valves (transcatheter aortic valve replacement) accounted for 5% of PVLs treated (Edwards 3.5%, CoreValve 1.5%). One patient (0.4%) had a PVL of a mitral annuloplasty ring treated percutaneously. The majority of PVLs were around mechanical valves (61.5%).
The preprocedural leak was severe (61%), moderate (34%), or mild (5.7%), and was multiple in 37%. PVL leak improved significantly following closure (P<0.001) and was graded as none (33.3%), mild (41.4%) moderate (18.6%), or severe (6.7%) at final follow-up (Figure 6A). Mean NYHA class improved from 2.7±0.8 preprocedure to 1.6±0.8 over a median follow-up of 110 (7–452) days (P<0.001) (Figure 6B). New hemolysis requiring transfusion as a consequence of PVL closure occurred in 1.6% of patients. Hospital mortality among elective patients was 2.9% in comparison with 6.8% for urgent cases and 50% for emergencies (P<0.001, log rank). Overall hospital mortality was 3.9%. During follow-up, 64 patients (24.8%) experienced MACE, which included death (42 patients, 16.2%) and valve surgery (15 patients, 6%). Two patients experienced new hemolysis requiring transfusion (1.6%). Each of the following occurred in 1 patient (0.4%): late embolization of device, infective endocarditis, and new valve leaflet interference. No patients had a stroke.
Factors associated with death and MACE are shown in Figures 4 and 5, respectively. Age per decade (HR, 1.28; P=0.02), presentation (HR, 1.78; P=0.01), creatinine (HR, 3.75; P=0.001), number of leaks (HR, 0.57; P=0.053), NYHA class at follow-up (HR, 1.57; P=0.008), and leak severity at follow-up (HR, 1.78; P=0.035) were entered into the multivariable model for association with death. The only factors independently associated with death were the degree of leak at follow-up (HR, 2.87 [1.07–7.73]; P=0.037), NYHA class at follow-up (HR, 2.00 [1.14–3.50]; P=0.015), and baseline creatinine (HR, 8.19 [2.39–28.08]; P=0.001).
Age per decade (HR, 1.17; P=0.092), presentation (HR, 2.05; P=0.001), hemolysis (HR, 1.93; P=0.006), creatinine (HR, 2.41; P=0.017), target valve MVR (HR, 1.78; P=0.012), procedural success (HR, 0.43; P=0.005), NYHA class at follow-up (HR, 1.88; P<0.001), and leak severity at follow-up (HR, 2.73; P<0.001) were entered into the multivariable model for association with MACE. The only factor independently associated with MACE was the degree of leak at follow-up (HR, 3.01 [1.48–6.12]; P=0.002). Kaplan-Meier plots for survival and survival free from MACE according to residual PVL leak are shown in Figures 7 and 8, respectively.
For those patients who had a PVL leak of a single surgical valve, PVL leak of the mitral valve was associated with increased MACE (HR, 1.83 [1.15–2.91]; P=0.011) and trend toward death (HR, 1.63 [0.99–2.69]; P=0.055). Mechanical valves were not associated with MACE (HR, 1.17 [0.68–2.00]; P=0.57) or death (HR, 0.65 [0.39–1.09]; P=0.10).
The types of devices and the frequency with which they were used are shown in Table 2 and Figure 3. The shape of the PVL, although complex and varied, is often crescentic (Figure 2A). Therefore, oblong devices (Amplatzer Vascular Plug 3, St Jude Medical, and Paravalvular Leak Device, Occlutech) carry a theoretical advantage for some PVLs over other devices with a circular cross section because they may conform better to the anatomy of the PVL. The first Amplatzer Vascular Plug 3 device was used in the UK on July 4, 2008. Before that time, only 26 other devices (mostly Amplatzer Duct Occluder and muscular ventricular septal defect occluders) had been implanted. From that date, the Amplatzer Vascular Plug 3 accounted for 68% of devices implanted. The Occlutech Paravalvular Leak Device was first used in the UK on June 19, 2013 and represents 3.9% of devices implanted. Use of oblong devices was associated with a nonsignificant trend for less leak at follow-up (none or mild) (HR, 1.37 [0.96–1.96]; P=0.079) and also NYHA I to II symptoms at follow-up (HR, 1.30 [0.93–1.83]; P=0.13). Both oblong devices are currently CE marked in Europe, but have not achieved Food and Drug Administration approval.
This series of unselected cases undergoing percutaneous PVL closure represents the combined UK and Ireland experience of this procedure. As in the previous largest report, the residual leak postclosure is important in determining MACE.15 However, residual leak in our series is also shown to be independently associated with death. The present analysis also differs in that it reports results of PVL closure with oblong devices that accounted for approximately two-thirds of devices in our series. As in prior percutaneous and surgical PVL closures series, this is a high-risk cohort with high attrition rates at follow-up despite technically and clinically successful procedures.6,7,15
Paravalvular Leak: Etiology
PVL is a frequent occurrence after valve surgery.2,3 However, the occurrence of PVL can be difficult to predict. One series suggests that postoperative PVL is present in 33% of MVRs immediately postoperatively.3 At 1.8-year follow-up, this had dropped to 15%. Similar figures for PVLs of AVRs were 6% postoperatively and 10% at 1.8-year follow-up. Therefore, some PVLs may close with time, whereas new ones will form. The same series suggested that PVL occurrence was not associated with the extent or severity of annular calcification (seen at surgery), use of pledgeted sutures, or identity of the operator. The only factors associated with PVL were MVR position, supra-annular AVR, and continuous sutures in the mitral position. Continuous suturing of the MVR was frequently associated with PVL (41%–67% of MVRs).3,17,18
The cause of PVL is uncertain and likely to be multifactorial. The regression of some PVLs in the early postoperative period, especially with MVRs, may relate to endothelialization of small PVLs or indeed suture tracts.3 In our series, the median time from valve surgery to PVL closure was 4.7 years. Such a long interval raised the possibility that there is some breakdown of cardiac tissue or sutures that progresses with time.
The PVL jet often follows a complex path and may be directed perpendicular to the usual direction of flow of blood, across the face of the surgical valve. Therefore, quantification of the leak severity by conventional echocardiographic criteria is difficult and severity is often underestimated, although 3D-TEE helps overcome some of these issues. The size of the arc of valvular dehiscence is another useful metric in determining suitability for percutaneous closure and reflects both valvular stability and PVL size. It is best estimated on 3D-TEE.
PVL of MVR versus AVR
PVLs are not homogeneous and seem to occur in a higher proportion of MVRs than AVRs.3 The reasons are unclear, although it may simply relate to the longer circumference of the sewing ring. Our data suggest that PVLs of the MVR are associated with increased MACE. This, however, may relate to confounding factors such as comorbidities, because valve type was not associated with MACE after multivariable adjustment. Data from surgical treatment of PVLs show that surgery for mitral PVLs is twice as frequent as for aortic PVLs and is associated with a nonsignificant trend for increased 30-day mortality (MVR 10.7% versus AVR 5%, P=0.1).6,8 Freedom from cardiac mortality at 12 years was less for surgical treatment of PVLs of MVRs than for surgical treatment of AVRs (56% versus 78%, P=0.01).6
Importance of the Residual Leak
Although baseline creatinine and NYHA functional class at follow-up were both independently associated with death at follow-up, the only factor independently associated with both MACE and death was the degree of leak at follow-up. As such, it seems that achieving a PVL no greater than mild in severity at follow-up is key to achieving a good outcome. This occurred in 74.7% of our patients.
Outcomes of PVLs
Both surgical and percutaneous series suggest that patients in whom PVLs occur are a high-risk cohort.6–8,13,15 The current gold standard of surgical reoperation is associated with 30-day/in-hospital mortality of 8.8% to 11.5%.6,7 One series reported perioperative stroke in 5.1%.8 After repeat surgery, PVLs recurred in 16% to 37% of patients.7,8 Long-term survival from surgical correction of PVLs remains poor at 30% to 57.8% and may reflect comorbidity rather than the disease process itself, although clear data on this are lacking. One surgical series reports 12-year actuarial survival at 39% but freedom from cardiac death at 12 years of 54%.6
Data on conservative management is sparse. One series suggests that it is associated with high event rates and generally fails with intervention being required.9 Another nonrandomized series suggests that, even in patients without severe symptoms related to the PVLs, surgical correction was still associated with improved survival in comparison with conservative treatment (log rank P=0.03).10 In this series, patients waited a median 4.6 years from diagnosis of the PVL before undergoing corrective surgery.
It is therefore clear that both patients and surgery are high risk, PVLs may recur despite surgical correction, and long-term outcomes remain poor. It is on this background that the result of our series should be interpreted. An overall hospital mortality of 3.9% compares favorably with the surgical outcomes, especially given that one-fifth of patients had urgent or emergent clinical presentations. In the present analysis, the percutaneous approach results in good improvements in PVL severity and NYHA classification and therefore is efficacious. Although the follow-up is not long-term, it is clear that there is a significant attrition rate despite successful procedures, and this is consistent with surgical outcomes as discussed above and likely reflects comorbidities.
Despite the varied morphologies of PVLs, it is clear from this series that oblong devices are preferentially selected by implanting cardiologists for closing PVLs in the UK and Ireland. Once they were available in the UK and Ireland, they accounted for 72% of devices used. Patients who had oblong devices had a nonsignificant trend to less leak (HR, 1.37; P=0.079) and better NYHA class (HR, 1.30; P=0.13) at follow-up. Although the Amplatzer Vascular Plug 3 and the Occlutech Paravalvular Leak the Device have a European CE mark for vascular occlusion and paravalvular leaks closure, respectively, neither device has been approved by the US Food and Drug Administration.19
Hemolysis at baseline was the principle indication for PVL closure in 16% of patients. It was associated with MACE on univariable analysis but not after multivariable adjustment. New hemolysis requiring transfusion as a consequence of PVL closure only occurred in 1.6% of patients, and, therefore, it seems that this feared complication of PVL closure is of less clinical importance.
Percutaneous PVL Closure: Technical Aspects
Percutaneous closure of PVL can be a technically demanding procedure and the techniques have been discussed in detail above. It is clear that PVLs of the MVR are more challenging to treat, and transseptal or transapical access is generally required. This procedure should only be undertaken by a team of structural and imaging cardiologists with a specialized interest in the technique and experience in advanced structural interventions. Careful preprocedural planning and imaging is mandatory as is periprocedural imaging.
Existing Literature on PVL Device Closure
The previous largest report of percutaneous PVL closure described PVL in 126 patients from a single center of excellence of this technique.15 This series similarly highlighted the importance of the final leak severity in determining MACE but did not link it with mortality. Because this series was from the United States, it did not include oblong devices. Other authors have presented smaller series including the transapical approach (percutaneous13 and surgical20), also with good outcomes.
This is a retrospective registry rather than a randomized trial, and this limits the conclusions that can be drawn.
Percutaneous device closure of PVL represents an effective approach to tackling a difficult problem in a high-risk cohort and compares favorably with results from surgical series. Those with a reduction in PVL leak to none or mild (74.7%) had less MACE and death. Use of oblong devices were key in the evolution of this procedure in the UK and Ireland. Despite significant improvements in symptoms and leak severity, these patients have a poor long-term prognosis that probably reflects their comorbidities. Percutaneous device closure of PVL should be considered in patients with suitable anatomy and in the absence of infective endocarditis.
Sources of Funding
Dr Calvert has received funding from the Academy of Medical Sciences. Dr Daniels is supported by the Wellcome Trust (WT-098519MA).
Drs Calvert, Hildick-Smith, de Giovanni, Mullen, Turner, Ormerod, and Shapiro proctor for St Jude Medical. St Jude Medical has covered travel, registration, and accommodation expenses for Dr Calvert to attend international cardiology congresses.
Sources of Funding, see page 942
Circulation is available at http://circ.ahajournals.org.
- Received March 24, 2015.
- Accepted August 5, 2016.
- © 2016 American Heart Association, Inc.
- Hammermeister K,
- Sethi GK,
- Henderson WG,
- Grover FL,
- Oprian C,
- Rahimtoola SH
- Ionescu A,
- Fraser AG,
- Butchart EG
- Rihal CS,
- Sorajja P,
- Booker JD,
- Hagler DJ,
- Cabalka AK
- Taramasso M,
- Maisano F,
- Denti P,
- Guidotti A,
- Sticchi A,
- Pozzoli A,
- Buzzatti N,
- De Bonis M,
- La Canna G,
- Alfieri O
- Genoni M,
- Franzen D,
- Vogt P,
- Seifert B,
- Jenni R,
- Künzli A,
- Niederhäuser U,
- Turina M
- Hourihan M,
- Perry SB,
- Mandell VS,
- Keane JF,
- Rome JJ,
- Bittl JA,
- Lock JE
- Sorajja P,
- Cabalka AK,
- Hagler DJ,
- Rihal CS
- Ruiz CE,
- Jelnin V,
- Kronzon I,
- Dudiy Y,
- Del Valle-Fernandez R,
- Einhorn BN,
- Chiam PT,
- Martinez C,
- Eiros R,
- Roubin G,
- Cohen HA
- Okuyama K,
- Jilaihawi H,
- Kashif M,
- Soni V,
- Matsumoto T,
- Yeow WL,
- Nakamura M,
- Cheng W,
- Kar S,
- Makkar RR
- Sorajja P,
- Cabalka AK,
- Hagler DJ,
- Rihal CS
- 16.↵BCIS British Cardiovascular Intervention Society. http://www.bcis.org.uk
- Skudicky D,
- Skoularigis J,
- Essop MR,
- Rothlisberger C,
- Sareli P
- McElhinney DB
- Taramasso M,
- Maisano F,
- Latib A,
- Denti P,
- Guidotti A,
- Sticchi A,
- Panoulas V,
- Giustino G,
- Pozzoli A,
- Buzzatti N,
- Cota L,
- De Bonis M,
- Montorfano M,
- Castiglioni A,
- Blasio A,
- La Canna G,
- Colombo A,
- Alfieri O
What Is New?
This is the first multinational series of percutaneous paravalvular leak (PVL) closure.
This is the first large series that also includes results of procedures undertaken with oblong devices that may be better shaped to deal with the crescentic shape of most PVLs.
Patients in whom a persisting PVL of mild or less was achieved at follow-up had less death and less major adverse cardiovascular events.
For patients with PVL of a single surgical valve, mitral valves were associated with increased major adverse cardiovascular events on univariable analysis, but not after multivariable adjustment.
What Are the Clinical Implications?
PVL is a common and difficult problem after valve surgery and patients are high risk.
Percutaneous PVL closure is feasible and compares favorably with results of repeat operation.
This is a challenging procedure that should be undertaken by specialized structural interventional teams with advanced imaging support.
Operators should endeavor to achieve PVL severity of mild or less at follow-up (associated with less death and less major adverse cardiovascular events).
Once available, oblong devices accounted for 72% of devices implanted and resulted in a nonsignificant trend for less leak at follow-up (hazard ratio, 1.37 [0.96–1.96]; P=0.079). The oblong devices used in this series are not currently approved by the Food and Drug Administration.