Feasibility of a Mitral Annuloplasty With the Capability for Peri- and Postoperative Adjustment

Mitral regurgitation (MR) is a debilitating disease, which affects 2% of the European adult population, with 250,000 new patients diagnosed worldwide per year [1]. MR is the predominant valvular pathology in Western populations, with an incidence which increases rapidly with age [2]. In the UK, 2929 primary surgical mitral valve procedures were performed in 2015 [3], the majority of which were for the treatment of severe MR.

The mitral valve apparatus consists of an annulus from which the two leaflets emanate. Chordae tendineae arising from the free edge of each leaflet are attached to the papillary muscles of the left ventricle. Mitral regurgitation can arise from dysfunction of any of these components. Changes in the mitral valve, such as weakened leaflets and chordae, lead to primary mitral regurgitation and impaired systolic leaflet coaptation. In contrast, in secondary mitral regurgitation, the mitral valve tissue is normal, but the supporting apparatus is altered. Generally, as the left ventricle becomes dilated and dyskinetic from ischemic or nonischemic cardiomyopathies, the mitral valve leaflets fail to appose one another, leading to regurgitation [4]. In primary and secondary MR, systolic backflow of blood into the left atrium alters the ventricular preload and after-load. Initially, these changes may be reversible, but with time, irreversible decompensation occurs resulting in reduced contractility and symptoms of congestive heart failure. In primary MR, the deterioration is gradual: 90% of the patients are alive at 8 years after diagnosis of MR and 50% of these survive without any cardiac related events. In secondary MR, the intrinsic relationship between the mitral valve and left ventricle means that dysfunction of either structures begets further dysfunction such that transition is more rapid. Only 30% of patients survive to 5 years without a cardiac event [5].

Noninotropic and inotropic vasopressors, diuretics, vasodilators, and heart failure drugs have all been used to treat MR, but with no long term or reproducible success. As such, surgical intervention is the mainstay of treatment for MR. Given the complex nature of the mitral valve apparatus, various approaches are available to manage defects. The interaction between annulus, leaflets, chordae, and ventricle has meant that complete valve replacement has been less successful, requires careful patient selection, and is generally avoided if possible. On the other hand, valve repair allows the dysfunctional portion of the valve apparatus to be corrected or compensated for, typically by annuloplasty, leaflet resection, or neo-chordoplasty. The implantation of an annuloplasty ring is considered as the gold standard for valve repair in functional/secondary MR, with a lower mortality rate, improved ventricular function, freedom from reoperation and anticoagulant complications, and superior long term survival than for replacement of the valve [6]. As such, repair is often the preferred intervention where possible.

Carpentier first introduced the mitral annuloplasty ring in 1969 as a means of reshaping the annulus to improve leaflet coaptation [7]. The use of an undersized ring to reduce the annulus in secondary MR was first reported in 1995 [8]. Over this period, over 40 designs of mitral annuloplasty rings have been reported.

Currently, a variety of surgical annuloplasty ring devices are available: they can be classified as rigid or flexible and planar or saddle-shaped. Some rings may be of an “incomplete” C shape, extending around the posterior leaflet, and leaving a space across the base of the anterior leaflet. The physiological annulus is D-shaped when viewed axially and takes a saddle shape in 3D space when loaded during systole. in vivo, in vitro, and in silico studies have suggested that augmenting the saddle form of the mitral valve annulus in a repair promotes optimum leaflet dynamics and reduces annular loading forces, promoting a more favorable outcome. While flexible rings may promote physiological dynamics, their lack of rigidity cannot provide the advantages conferred by a saddle form [9–11].

Although mitral annuloplasty is superior to medical intervention and valve replacement, long term results after annuloplasty are still suboptimal, with recurrence of 2+ MR in 33% of the patients at 1 year and 65% of the patients presenting with recurrent MR at 2 years after surgery [12]. Current surgical mitral valve repair and replacement in severe ischemic MR carry a 20% mortality at 2 years. Among patients alive at 2 years following mitral valve repair, 60% have recurrent moderate or severe MR [13]. Undersized ring annuloplasty may unevenly displace the posterior annulus anteriorly, which leads to increased loading of the posterior leaflet [14]. Essentially, the mitral valve becomes functionally unileaflet. Leaflet tethering decreases leaflet curvature and results in the increased leaflet and chordal stress [15]. Using saddle-shaped rings that are accurately sized to patients' mitral annulus directly affects the durability and effectiveness of mitral valve repairs [16]. Currently, a qualitative approach is taken to annuloplasty shaping and sizing. Advances in image acquisition allow high accuracy reconstructions of the mitral valve apparatus prior to intervention [9], and echocardiography and Doppler allow repairs to be analyzed peri-operatively.

We propose a mitral annuloplasty repair that is not fixed to a preset geometry, allowing peri-operative adjustment during the implantation procedure, and also postoperative adjustment. Such a device would allow an optimum repair to be achieved and could allow readjustment to ameliorate the effects of ventricular remodeling. To achieve this, the REshapeable Mitral Annuloplasty DevIce (REMADI) consisting of a fully encapsulated low melting temperature alloy is used. The alloy is solid and rigid at body temperature and provides traction to shape the annulus. When heated, the alloy melts and the REMADI becomes malleable.

To enable the REMADI to be implanted percutaneously, attachment to the annulus is suture-less. We designed a set of anchors that automatically deploy when opposed to the tissue. Each anchor consists of two nickel titanium alloy barbs, which become embedded in the annular tissue. Prior to deployment, the barbs are constrained within a stainless-steel sheath. Upon apposition to the tissue surface, the sheath retracts allowing the barbs to return to their original, curved conformations. The barbs become embedded within the mitral valve annulus, thereby securing the annuloplasty ring.

There are several other devices for surgical, percutaneous, or hybrid mitral valve repair, which have been proposed or have reached clinical testing. There is an array of devices of varying geometries, implantation methods, tissue attachment, and some with the facility to modify the geometry peri- or postoperatively; a selection of relevant devices are summarized in Table 1. While several of these devices have reached the first-in-man stage, there have been technical, clinical, or commercial hurdles that have limited adoption. As a consequence, percutaneous repair of mitral regurgitation remains an unsolved problem.

The annuloplasty ring is composed of a low melting temperature alloy core. A bismuth, indium, tin, and zinc alloy (35, 48.6, 16, and 0.4 wt %, respectively) was first proposed for biomedical uses in 2014 as a bone cement. The alloy melts at 57.5 °C [24]. The encapsulating layer is a silicone elastomer MED-4420 (NuSil, Polymer Systems Technology Limited, High Wycombe, UK). Appropriately sized rods were coated while rotating, before being cured in an environmental chamber at 50 °C and 50% relative humidity for 24 h.

Prototype devices used for testing were constructed using nonvulcanized latex rubber (MBFG, Ireland). Appropriately sized mandrels were dip coated in a rubber solution before being rotated around their long axis, for a minimum of 24 h. The rubber tubes were then released from the mandrel.

Attachment of the annuloplasty device to the annulus tissue is achieved using nitinol barbs, shown in Figs. 1(a) and 1(b). The barbs were laser cut from a nickel titanium wire in their straight (predeployed) conformation and shape set to their curved (deployed) form shown in Fig. 2. The shaft of the barbs is secured within the anchor by a welded collar. The anchor is laser cut from 316 L stainless steel and consists of a ring through which the annuloplasty tube passes. The barbs are welded to the outer surface of the ring. Another stainless-steel tube is welded to the inner surface of the ring and is used to engage the annuloplasty device with the delivery tool. Six anchors are evenly spaced along the length of the annuloplasty tube and bonded to the rubber encapsulating layer using cyanoacrylate.

The parts of the assembled device are illustrated in Figs. 1(c) and 1(d). The rubber tube forms the encapsulating layer of the annuloplasty. The tubes were elongated by 200%, filled with molten alloy, and then cooled while in this extended state. The ends of the rubber tubes were sealed using custom stoppers (FormLabs Tough Resin V2).

The prototype delivery tool was designed for in vitro catheter delivery. Six nitinol wires are used to form a basket. Axial compression of the nitinol wire members determines their radial displacement. Each member can be controlled individually, allowing noncircular shapes to be formed. The nitinol wires engage with the anchors of the annuloplasty device, as shown in Figs. 2 and 3. The radial position of each spline can be controlled, and the axial position of each anchor with respect to each spline can be controlled, and as such, any saddle form can be generated for the annuloplasty.

Engagement with the mitral annulus is achieved using the barbs. The force required to deploy and engage the barbs and the force required to dehisce the barbs from the tissue were tested using a position controlled tensile tester (Texture Analyser, Stable Microsystems). Porcine hearts were obtained from a local abattoir and the mitral annulus dissected. The mitral annulus was restrained on the base plate using a Perspex sheet with a 10 mm hole through which the barb passed. Insertion and dehiscence were tested at a minimum of eight points on the atrial side of the mitral annulus. The force required to dehisce the barbs from the tissue was compared with the force required to dehisce a 2-0 monofilament suture from the mitral annulus.

Noncontact heating of the mitral annuloplasty was performed using an inductor (FlexHeat 2.0, RDO Induction, Washington, NJ). In this feasibility study, induction heating was performed with an external coil around the left atrium and pulmonary vein, as illustrated in Fig. 4. The induction heater is driven by a 2 kW power supply. The copper coil was internally cooled at a minimum flowrate of 2 L min⁻¹. The heater module is tuned to the resonant frequency of the coil and load. The alternating current through the coil induces an alternating magnetic field, which is strongest within the coil, where the annuloplasty device is located. The alternating magnetic field penetrates the conductive alloy core of the annuloplasty, inducing eddy currents which heat the metal core of the annuloplasty through joule heating. This is used peri-operatively to soften the device and alter it from a straight configuration for delivery, to a C-shape for implantation, and then for any fine adjustment of the saddle form. This configuration of induction heating coil is incompatible with in vivo testing, but allows a minimum induction current to be used for feasibility testing. The final device would utilize an internal coil mounted on the delivery catheter.

Heat from the metal core of the annuloplasty ring will be dissipated through the encapsulating layer and into the environment.

To test delivery and feasibility of the REMADI, a porcine passive heart model was used [25] (Cardiac Biosimulator, LifeTec Group, Eindhoven, The Netherlands). The passive heart model uses porcine hearts from a local abattoir. The hearts are frozen to reproduce their physiological mechanical properties. The hearts are mounted to the servocontrolled pump, in line with compliance and resistance modules to simulate pulsatile flow in the absence of active cardiac contraction. A port is placed at the apex, which provides the necessary pressure gradients to simulate diastolic and systolic flows through the heart. Without the requirement to pump blood, transparent fluids can be used, allowing direct endoscopic vision to be used. Absolute and transvalvular differential pressure measurements were taken during testing.

The nitinol anchoring devices underwent in vitro testing to assess implantation and pullout force using a 2-0 polypropylene sutures as a control. A position-controlled tester implanted the anchors into a fresh porcine mitral annulus. Sample data from this experiment are shown in Fig. 5. A single prototype anchor provided 2 ± 0.5 N. The prototype device would be secured by a minimum of 6 anchors. When combined, the anchors would be sufficient to withstand the typical forces (1.2 ± 0.9 N cyclic force, with a peak at 2.4 ± 4.5 N) exerted upon a ring anchored in the mitral annulus under a systolic pressure of 150 mmHg [26]. The 2-0 polypropylene suture provided at least 10 N.

A proof-of-concept REMADI was assembled to demonstrate the feasibility of the REMADI. Complete reversibility of the soft-rigid state allowed the REMADI to be shaped to fit into a catheter and shaped to the annulus once in the left atrium. For the proof-of-concept, a 32Fr catheter was used.

Four hearts were characterized on the passive heart platform. Gross and endoscopic images are shown in Figs. 6 and 7. Four implantations were performed. Three annuloplasties were implanted off-pump under direct vision with a transatrial approach. One annuloplasty was implanted under simulated flow conditions using the catheter delivery device through the apex of the left ventricle. There was an issue passing the chordae tendineae during the apical implantation, causing some damage. Figure 8 shows the ring being unsheathed within the left atrium. Once unsheathed, the induction heating system was used to soften the alloy core of the annuloplasty. The softened annuloplasty was then expanded using the basket, as shown in Fig. 9. Retracting the annuloplasty device toward the apex brought the device in contact with the annulus. As the undeployed barbs make contact with the annulus, the restraining sheaths slide back into the device, allowing the barbs to become buried in the tissue of the annulus, as shown by the dissection image in Fig. 10. With the REMADI attached to the annulus, the diameter of the device was reduced and a saddle form produced.

The left atrial pressure, left ventricular pressure, and transvalvular pressure gradients are shown in Fig. 11. The minor difference between the traces may reflect a change in valve dynamics induced by the ring, but given that the baseline was a healthy, not a lesion model, no change in hydrodynamics was sought. The regurgitant fraction could not be accurately measured on this setup as it was not possible to perform echocardiography, nor mount a high frequency ultrasonic flow probe on the pulmonary vein. Temperature probes placed in the annular tissue and the left ventricle did not detect any change in temperature during heating.

The REMADI device tested in this study was designed to represent the key components required for an assessment of feasibility. In this regard, could the device be manufactured, fit into a delivery catheter, be transitioned to a soft state while in a physiological flow (before and after engagement with the annulus), be shaped to a saddle form during delivery, engage with the mitral annulus, and provide a rigid deformation of the annulus.

The barb devices described and tested in this paper are a novel means of implanting an annuloplasty device. The absolute pull-out strength has been characterized and is of an appropriate order of magnitude; however, the effect of cyclic stress upon the engagement between the anchor and tissue has not been tested. Subsequent iterations of the REMADI would use a Dacron coating to promote full tissue integration into the mitral annulus. The design of the barbs of the anchors was optimized using linear estimates of the strain and stiffness. Finite element analysis may be used to optimize the dimensions of the barbs, whose dimensions are a critical factor for the overall profile of the device. A significant limitation of the barb method is the lack of control when deploying and difficulty in retracting the barbs once deployed. The deployed barbs may be retracted; however, this functionality is not built into the delivery device. Compared to the helical anchor method used by the Cardioband device [27], the REMADI anchors can be deployed more quickly.

There are several shortcomings to the delivery device used for implantation of the REMADI. The form of the basket delivery tool impinges upon the mitral valve and prevents the leaflets from closing during systole. As such, the open basket impeded closure of the mitral valve. Optimization of the basket wire shapes can reduce this limitation in future designs. The transapical route of delivery permits direct access to the mitral valve, which is necessary for this delivery device. It also affords larger bore access and fewer closure related complications and avoids atherosclerotic peripheral vessels; however, it is perceived as being more invasive than transfemoral delivery [28]. Table 1 shows several direct annuloplasty devices that are delivered via a transseptal route with transfemoral access, which has become the first choice of implantation method for valves [29]. The option for transfemoral implantation would be preferred and may ease adoption of the device.

In testing of the REMADI prototypes, we encountered problems with the formation of cracks in the metal core upon refreezing. To mitigate this issue, the elastic encapsulating layer is overfilled, such that there is a positive pressure on the metal core. This is achieved by filling the encapsulating tube in an elongated state. Furthermore, the starting length of the encapsulating tube is less than the minimum desirable circumference of the mitral annulus; the device is stretched in order to engage with the mitral annulus before the remodeling step reduces both the device length and the annulus diameter.

where δ is the penetration depth (m), f is the frequency (Hz), μ is the magnetic permeability of the material (H/m), and σ is the electrical conductivity of the material (S/m). When the penetration depth exceeds the thickness of the material, eddy currents are not generated and there is minimal energy transfer. As such, smaller metal components, such as the barbs, are not heated.

Although the core of the REMADI does not make contact with any biological elements, nonetheless the material has been tested for its hemo- and biocompatibility, which is sufficient for long term implantation in accordance with ISO 10993 [24]. The encapsulating layer is an elastomeric silicone material with ISO 10993 compliance for long term implantation in blood contacting devices.

In this paper, we demonstrated that the shape of the REMADI can be altered after implantation. As such, the REMADI could potentially accept a transcatheter mitral valve implantation. Transitioning the prosthesis to a soft state allows the REMADI to conform to the perimeter of the valve implant, minimizing paravalvular leak and uneven transcatheter mitral valve implantation expansion which may occur with rigid annuloplasty devices. The valve-in-ring concept emulates the accepted clinical pathway of the transcatheter aortic valve implant-in-valve [30]. The second benefit of postoperative adjustment of a mitral annuloplasty ring is to reshape the annulus in a controlled fashion to eliminate any subsequent recurrent mitral regurgitation, this has not been demonstrated in this paper and requires the catheter to reengage with the device.

Peri-operative adjustment of the mitral annuloplasty ring allows for patient specific optimization of the repair. The exact MR phenotype takes a different form in every patient [9], as such the ability to perform a patient specific annuloplasty (with peri-operative feedback) may be of significant benefit. Carpentier described an optimal repair as (1) preserving or restoring full leaflet motion, (2) creating a large surface of coaptation, and (3) remodeling and stabilizing the entire annulus [31]. Points (1) and (2) may be altered by manipulation of the annuloplasty ring after implantation, which can be evaluated using peri-operative echocardiography. As such, the REMADI may confer a benefit over the other devices proposed elsewhere, as summarized in Table 1, which are not capable of long term adjustment and may only be adjusted to a predetermined size.

To complete testing of the REMADI and demonstrate its effectiveness, further testing is warranted in an MR lesion model, both in vitro and in vivo. Finally, assessment of the durability of the REMADI and its components requires further evaluation.

The REMADI is a novel device for the treatment of mitral regurgitation with the capability to be reshaped peri-operatively to achieve an optimal saddle shaped conformation and postoperatively to accept a transcatheter mitral valve. It may also be capable of postoperative adjustment to correct any future MR recurrence as a result of ventricular remodeling. The solid–liquid phase transition of a low melting temperature alloy is used to achieve a soft and rigid state allowing the device to fit within a catheter, as well as adopt any required mitral annulus form. The nitinol anchor system has demonstrated feasibility to engage with mitral annular tissue and secure a device. The prototype REMADI was constructed and then tested in a beating heart model where the annulus could be reshaped by the device.

• British Heart Foundation (Translational Award TG/15/4/31891; Funder ID: 10.13039/501100000274).

• MRC (Confidence in Concept Award, Round 6, 2017; Funder ID: 10.13039/501100000265).

• University of Cambridge (Armstrong Trust Ph.D. Studentship; Funder ID: 10.13039/501100000735).