Real-time intraoperative co-registration of transesophageal echocardiography with fluoroscopy facilitates transcatheter mitral valve-in-valve implantation in cases of invisible degenerated bioprosthetic valves

Abstract

 

Open in new tabDownload slide

OBJECTIVES

Transcatheter mitral valve-in-valve (TMViV) implantation is an alternative treatment to surgery for high-risk patients with degenerated bioprosthetic mitral valves. Some types of bioprostheses are fluoroscopically translucent, resulting in an ‘invisible’ target deployment area. In this study, we describe the feasibility and outcomes of this procedure using intraoperative fusion of transoesophageal echocardiography (TEE) and live fluoroscopy to facilitate valve deployment in cases of invisible bioprosthetic valves.

METHODS

We reviewed all TMViV implantations at our centre from July 2014 to July 2019. Patient, procedure and outcome details were compared between those with a visible bioprosthesis (N = 22) to those with an invisible one (N = 12). Intra-operative TEE and live Fluoroscopy co-registration were used for real-time guidance for all invisible targets.

RESULTS

All valve implantations were completed successfully in both groups without cardiovascular injury, valve migration or left ventricular outflow-tract obstruction. Technical success was 100% in both groups. One-year survival was 83% [95% confidence interval (CI) 70–96] for the entire cohort, with 79% (95% CI 63–100) survival for the visible group and 92% (95% CI 77–100) for the invisible group. Probability of 1-year survival free from mitral valve reintervention, significant valve dysfunction, stroke or myocardial infraction was 78% (95% CI 63–93) for all patients whereby the probability was 72% (95% CI 54–97) in the visible group and 80% (95% CI 59–100) for the invisible group.

CONCLUSIONS

The use of intraoperative TEE and live fluoroscopy image fusion facilitates accurate TMViV among patients with a fluoroscopically invisible target-landing zone.

INTRODUCTION

Structural valve deterioration (SVD) resulting into clinically relevant valve stenosis or regurgitation affects a sizable proportion of bioprosthetic valves in the long term [1]. Recent studies have shown 15-year cumulative reoperation rates of 11.1% and 19.9% at 15 years among patients who received bioprosthetic mitral valve replacement aged of 50–69 and 18–50 years, respectively [2, 3]. SVD is a common indication for reoperation [4, 5].

Transcatheter mitral valve-in-valve (TMViV) implantation for degenerated bioprosthetic valves is a viable alternative therapy in patients with a high risk for conventional surgical reoperation. A report from the transcatheter mitral valve replacement registry showed a 1-year cumulative all-cause mortality of 14.0% among 322 patients with a mean Society of Thoracic Surgeons (STS) score of 9.2% (±7.2%) [6]. A recent meta-analysis that pooled data from 13 studies showed a reduced 30-day mortality following TMVI among this high risk patients as compared to the predicted operative mortality [7].

Both transseptal and transapical approaches have been described [8, 9]. Fluoroscopy and echocardiography are important for precise positioning. Some types of bioprosthetics are fluoroscopically translucent, resulting in an ‘invisible’ target deployment area (Fig. 1).

To facilitate TMViV procedure in such, we have recently employed intraoperative fusion imaging of transoesophageal echocardiography (TEE) and fluoroscopy. This method has been described for the guidance of mitral valve repair using the mitral clip and cardioband devices [10] as well as the closure of paravalvular leaks, atrial septal defects and left atrial appendages [11, 12].

In this report, we describe the feasibility and technical considerations for using this strategy in transapical TMViV among patients with fluoroscopically invisible valves. We report the procedural and clinical outcomes in this patient cohort compared to patients undergoing TMViV for SVD of fluoroscopically visible mitral valve bioprostheses.

METHODS

Ethical statement

The institutional ethics review board approved the retrospective study (Ethics approval number EA1/063/19).

Patient characteristics and outcomes

Records of all patients undergoing TMViV at the German Heart Center Berlin between July 2014 and July 2019 were reviewed. The preoperative characteristics, operative details, procedural and mid-term outcomes were compared among patients with visible (group A) or invisible (group B) degenerated bioprosthetic mitral valves.

Preoperative evaluation

All patients underwent a standard institutional diagnostic work-up for a transcatheter heart valve procedure. A TEE was performed to confirm severe SVD, to exclude relevant additional paravalvular regurgitation and prosthetic valve endocarditis and/or thrombosis. Evaluation of the suitability of the device-landing zone (DLZ) for TMViV was assessed using high-resolution ECG gated contrast 4D computed tomography (CT) scans analysed with the 3Mensio structural heart program (version 8.0 and 9.0, Pie Medical Systems, Netherlands). The measured parameters included mitral valve area, mitral valve perimeter, aorto-mitral angle, left ventricular (LV) diameter, neo-left ventricular outflow tract (LVOT) area, feasibility for transseptal or transapical implantation and the optimal fluoroscopic C-arm angulation. A neo-LVOT in late systole of <1.7 cm² was considered an exclusion criterion for TMViV [13]. Based on CT measurements, the type and size of the prosthetic device was chosen using the mitral valve-in -valve app (http://www.ubqo.com/vivmitral). A simulation of the procedure was performed using virtual valve mock-ups.

Standard transapical TMViV procedure

All procedures were performed in a hybrid operating room by a consistent heart-team comprising at least 1 cardiologist, anaesthesiologist and cardiac surgeon. The target intercostal space was chosen based on transthoracic echocardiography and preoperative CT scan analysis. A left-sided mini-thoracotomy was performed and a soft-tissue retractor placed. Under TEE guidance in the long-axis 4-chamber view, the left ventricular apex was identified. The pericardium was incised in a very limited fashion and 2 deep apical U-stitches were placed through the pericardium using pledgeted 3–0 prolene sutures. In patients with a concomitant diagnosis of aortic stenosis the transcatheter aortic valve implantation procedure was conducted first through the same access. After apical puncture, a J-tip Terumo guidewire (Terumo Interventional Systems, NJ, USA) was introduced into the left ventricle, followed by a 6 F multipurpose pigtail catheter. After crossing the mitral valve prosthesis, the guidewire was exchanged for an Innowi pre-shaped stiffer wire (Symedrix GmBH, Germany). The transcatheter valve system was then introduced and positioned across the landing zone. Under a short duration of rapid pacing and with live fluoroscopy guidance, the transcatheter valve was implanted. We aimed to place the stent frame of the transcatheter valve 70–90% in the LV, and 10–30% on the atrial side. Clear fluoroscopic demarcation of the annular ring of the degenerated bioprosthetic valve was therefore critical. The delivery catheter, wire and sheath were then withdrawn and the LV apex closed using the preplaced sutures.

TEE and fluoroscopy real-time fusion

In cases where the degenerated bioprosthesis was fluoroscopically invisible, the EchoNavigator system (Philips Healthcare, MA, USA) was used for guidance. Image fusion not only allows the implanter to concentrate on one screen during implantation but also facilitates precise demarcation of the landing zone. The concept has been described previously [14, 15]. For TMViV guidance, first the TEE echocardiographic view that affords the best perpendicular view of the valve was selected. Typically this is a 2D bicommissural view between 50° and 90°. A simulated preview fusing this TEE view with the pre-calculated optimal fluoroscopic angulation was conducted. The TEE probe location was then registered in the fluoroscopic field using 2 fluoroscopic projections. The pre-calculated fluoroscopic angulation was then implemented and co-registered with the prior optimized TEE field of view to generate a real-time registered image overlay. In the fused image, the implanter had a perpendicular view onto the target-landing zone and the delivery system including the stent (Fig. 2) (Video 1).

Post-procedural anticoagulation therapy

Patients received anticoagulation with vitamin K antagonists with or without Aspirin and Plavix for at least 3 months following TMVI. In the early stage of our experience, we routinely bridged patients taking anticoagulation prior to surgery with heparin. More recently, we refrained from stopping anticoagulation preoperative and immediately restarted it the day after surgery in the absence of bleeding complications.

Outcomes and statistics

Outcomes were evaluated according the recommendations of the mitral valve academic research consortium [16]. Continuous variables are summarized as median and interquartile range while categorical variables are characterized as frequencies and percentages. For group comparisons, the Mann–Whitney U test was used for continuous variables and Chi² test or Fisher’s exact test compared categorical data.

RESULTS

Patients

Thirty-four patients received a TMViV via a transapical access. Twenty-two (65%) had a visible degenerated bioprosthesis (group A), while 12 (35%) had an invisible degenerated bioprosthetic valve (group B). Patient characteristics are summarized in Table 1. There were no statistically significant differences in age, sex distribution, body surface area, New York Heart Association or preoperative cardiac rhythm between the 2 groups. The median EuroScore II was 9.0% [6%, 16%] in group A and 9.5 [3.2, 18] in group B. The median STS predicted risk of mortality within 30 days (STS-Predicted Risk of Mortality) was 5.6% [4.1%, 10.6%] and 5% [3.1%, 10.6%] in group B. In group A, the degenerated prosthetic valves included Medtronic Hancock II (N = 12), Carpentier Edwards Perimount (N = 7), Epic Mitral (N = 1), Medtronic Mosaic valve (N = 1) and Sapien XT (N = 1) (Table 2). The degenerated prosthetic valves in group B included Labcor stentless valve (N = 7), Shelhigh bioprosthetic (N = 4) and Biomitral from Biointegral (N = 1). In group A, mitral valve insufficiency was present in 9 (41%) patients, mitral valve stenosis in 5 (22%) patients and mixed stenotic/regurgitant disease in 7 (32%) patients. One (5%) patient with a degenerated surgically placed Sapien XT valve had a trans-stent leak. In group B, 3 (25%) had pure mitral insufficiency, whereas 7 (58%) patients had mitral valve stenosis and 2 (17%) had mixed mitral disease. The median mitral valve area and perimeter were 6.95 [6.1, 7.27] cm² and 95 [89.0, 96.6] mm, respectively, for group A and they were 6.80 [5.35, 9.5] cm² and 101.6 [87.1, 118.1] mm in group B. In 5 (23%) patients in group A and 5 (42%) patients in group B, the TMVIV procedure was combined with a transcatheter aortic valve implantation within the same procedure. Sapien 3 (Edwards Lifesciences) devices were implanted in the majority (94%) of patients. Lotus prostheses (Boston Scientific) were implanted in 2 (6%). All patients with invisible bioprostheses (group B patients) received a Sapien 3 prosthesis.

Early outcomes

Early outcomes up to 30 days are summarized in Table 3. All devices in both groups were successfully implanted in the correct position without any damage to intracardiac structures, valve migration, LVOT obstruction or significant mitral valve insufficiency. The median fluoroscopy time was 7.3 [5.8, 13.5] min in group A and 10.5 [8, 13.6] min in group B. The median dose area product was 3673 [1546, 6607] uGray × m² in group A and 1993 [1350, 4467] uGray × m² in group B. There was no statistically significant difference in fluoroscopic time or dose area product between the 2 groups (Table 2). There were no incidences of stroke, myocardial infarction or acute kidney injury. There was 1 (5%) new pacemaker implantation in group A within 30 days postop and none in group B. New onset of atrial fibrillation occurred in 1 (5%) patient in group A and 2 (17%) patients in group B. There were 4 bleeding related reoperations. Three patients had pericardial bleeding while 1 had inguinal bleeding. One patient in group B received a surgical mitral valve re-replacement on POD 8 due to early valve thrombosis. Thirty-day mortality was 6% (2 patients). One patient in group A died on postoperative day 8 due to left ventricular bleeding and 1 patient in group B died on postop day 22 due to multi-organ failure and sepsis. Technical success was 100% in both groups. Device and procedural success at 30 days were both 91% in group A and 75% in group B respectively (P = 0.255). There was no statistically significant difference in 30-day mortality, technical success, device success or procedural success.

Mid-term follow-up

The median follow-up was 14.73 (1–51) months (Fig. 3). One-year survival was 83% [95% confidence interval (CI) 70—96] and survival free from mitral valve dysfunction, significant valve dysfunction, stroke or myocardial infarction was 78% (95% CI 63—92) at 1 year. For group A, 1-year survival was 79% (95% CI 63–100) and it was 92% (95% CI 77–100) for group B. Probability of 1-year survival free from mitral valve reintervention, significant valve dysfunction, stroke or myocardial infraction was 72% (95% CI 54–96) and 80% (95% CI 59–100). There were 4 episodes of valve thrombosis on late follow-up, N = 2 in group A and N = 2 in group B. Three patients without long-term oral anticoagulation developed valve thrombosis and were successfully medically treated while 1 underwent a redo TMIViV at 14.2 months.

DISCUSSION AND LIMITATIONS

TMViV replacement for degenerated mitral valve prosthesis in patients who are at high risk for surgical reoperation results in acceptable outcomes [6]. A precise visualization of the DLZ during implantation is mandatory to achieve device success and to safely deploy the valve at the optimal location.

There are bioprosthetic valves, such as the Labcor stentless valve, the Shelhigh bioprosthetic and the Biomitral that are radiographically invisible. In our cohort, these comprised more than one-third of TMViV cases performed via transapical access during this period. Radiographically translucent bioprosthetic valves lack a metallic frame. In some clinical settings, for example in cases of endocarditis or cases of heavy annular calcification, these tissue-only bioprostheses may be preferable to frame mounted devices. As such, they will likely continue to comprise a sizable proportion of surgically implanted mitral bioprosthetic valves.

The challenge of correct positioning of a transcatheter valve within the target-landing zone of a fluroscopically invisible bioprosthesis may put off some heart teams from attempting TMViV in these patients.

We here demonstrate feasibility of intraoperative real-time fusion of 2D TEE and fluoroscopic images for procedural guidance and accurate valve implantation in such cases. We also demonstrate comparable short and mid-term outcomes in this group of patients compared to those with visible bioprostheses where such an intraoperative image fusion is not required. The 1-year survival in our series is comparable to the results of a larger cohort of patients reported in the TMViV registry [6].

While fluoroscopically invisible valves are perfectly visible in echocardiography, it is difficult to precisely visualize the delivery and stent system in TEE as it will produce echogenic artefacts. It may therefore be difficult to judge the correct implant depth. The implantation balloon can be filled with echogenic contrast to allow for better visualization. This may allow TMViV based on TEE alone. There are only anecdotal reports with this method [17] and we have also performed only 1 case prior to the introduction of image fusion technology.

Alternatively, 1 may identify the position of the mitral valve by performing levocardiography in the implanter’s fluoroscopic angulation and then use this image as a ‘roadmap’ during implantation. This method has the disadvantage that due to the lack of motion compensation the exact location of the valve is not known and implantation is therefore less precise.

This series of 12 patients is the first report to specifically examine the feasibility and mid-term outcomes in patients undergoing TMVIV with fluoroscopically invisible valves. Mitral valve academic research consortium device and procedural success at 30 days and at mid-term were comparable. The most common complications in our series were not related to prosthesis positioning, demonstrating that image fusion facilitates good DLZ visualization. All other complications and risks for transcatheter interventions continue to apply.

For invisible DLZs, we have up to now only used the transapical route because the short distance to the DLZ helps to simplify image guidance. To further reduce invasiveness and surgical trauma in critical ill patients, we and others are increasingly performing TMViV in visible bioprostheses and rings using the transfemoral transseptal route [8]. With increasing experience using image fusion tools, the intention is to extend the less invasive transseptal approach also to treat invisible mitral valve bioprostheses.

Of the available possible devices for off-label use, we have typically preferred the Sapien 3 device due to its easier handling. In a few cases where the existing degenerated bioprostheses has an exceedingly small area we have also implanted the lotus valves. The benefit of the Lotus device is its re-sheathing and retrieval ability. The Lotus system can currently only be implanted via the transapical access route in the mitral valve position.

Both 2D and 3D TEE can be co-registered with fluoroscopy in real time. For the live implantation we find the 2D view preferable to provide adequate visualization of the DLZ while maintaining the familiar ‘implanter’s’ view. The echocardiographer needs to determine a 2D imaging plane that produces the least angle dependent offset between itself and the selected fluoroscopic angulation for implantation, leading to a fusion image that is free of deformation and foreshortening. The mitral valve can be readily visualized in many different angles from a midesophageal position, maintaining an almost perpendicular beam angle to the valvular plane. This renders the fusion technology exceptionally useful for the mitral valve in particular. Another clinical scenario where real-time intraoperative image fusion may also be of benefit is transcatheter mitral valve in replacement in patients with severe mitral annular calcification. It is conceivable that at least the complications related to challenges in valve positioning among this group of patients could be reduced using this technology.

In our experience, having a consistent dedicated team of multidisciplinary specialists and maintaining good team dynamics, are important for effective utilization of this new imaging modality. In our team, not only the interventional cardiologists and surgeons are involved in case selection and planning, but also the interventional echocardiographers. This ensures that all aspects of both imaging modalities are thoroughly discussed early in the decision-making process, with inputs from experts from each field.

Future improvements to the sharpness of the visualization masks, and the availability real-time image manipulation controls for the fused images in the sterile workspace, would further help embed this technology in the clinical workflow.

CONCLUSION

It can be expected that a sizable proportion of implanted bioprosthetic mitral valves are fluoroscopically invisible. In case of degeneration of these valves, TMViV implantation may be indicated in patients who are at high risk for surgery. Live intraoperative fusion of TEE and fluoroscopy allows safe and accurate valve positioning, resulting in comparable outcomes to those with visible target-landing zones. Long-term outcomes especially concerning the risks for in-valve thrombosis or degeneration remain to be seen.

ACKNOWLEDGEMENTS

We thank Mrs. Svetlanda Sonnabend and Ms. Bernd Enksaihkhan for database and patient follow-up assistance. We also thank Mr. Helge Haselbach for graphics.

Conflict of interest: Prof. Jörg Kempfert and Dr. Axel Unbehaun are proctors for Abbott GmbH, Boston Scientific Inc., Edwards Lifesciences and Medtronic GmbH. Prof. Volkmar Falk reports a financial relationship with Abbott GmbH, Medtronic GmbH, Boston Scientific Inc., Novartis Pharma, Berlin Heart GmbH, Biotronik Se & Co., Edwards Lifesciences, JOTEC GmbH, Zurich Heart and ETH Zurich in relation to research and study funds and fees for lectures and speeches.

Author contributions

Isaac Wamala: Conceptualization; Data curation; Formal analysis; Methodology; Project administration; Writing—original draft. Axel Unbehaun: Conceptualization; Data curation; Formal analysis; Methodology; Supervision; Writing—original draft. Christoph Klein: Investigation; Methodology; Writing—review & editing. Marian Kukucka: Investigation; Methodology; Writing—review & editing. Dirk Eggert-Doktor: Investigation; Writing—review & editing. Semih Buz: Investigation; Methodology; Writing—review & editing. Julia Stein: Formal analysis; Methodology; Writing—review & editing. Simon Sündermann: Conceptualization; Investigation; Methodology; Writing—review & editing. Volkmar Falk: Conceptualization; Methodology; Supervision; Writing—review & editing. Jörg Kempfert: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Supervision; Writing—original draft.

Reviewer information

Interactive CardioVascular and Thoracic Surgery thanks Clarence Pienteu Pingpoh, Alberto Guido Pozzoli and David Schibilsky for their contribution to the peer review process of this article.

REFERENCES

ABBREVIATIONS

 
• CI

  Confidence interval

 
• CT

  Computed tomography

 
• DLZ

  Device-landing zone

 
• LV

  Left ventricular

 
• LVOT

  Left ventricular outflow tract

 
• STS

  Society of Thoracic Surgeons

 
• SVD

  Structural valve deterioration

 
• TEE

  Transoesophageal echocardiography

 
• TMViV

  Transcatheter mitral valve-in-valve