Mid-term outcomes of aortic valve repair using an anatomically shaped internal annuloplasty ring

Abstract

Open in new tabDownload slide

OBJECTIVES

The Hemispherical Aortic Annuloplasty Reconstructive Technology (HAART) ring is a rigid, internal and geometric device. The objective of this article is to assess the mid-term outcomes of aortic valve repair (AVr) using this prosthesis.

METHODS

A prospectively maintained database was used to obtain outcomes for adult patients undergoing AVr using the HAART ring between September 2017 and June 2023. All aortic patients at our institution undergo life-long surveillance with regular assessment and valve imaging.

RESULTS

Seventy-one patients underwent AVr using the HAART device: 53 had a trileaflet valve and 18 a bicuspid valve. The median age was 54 years, and most were male (79%). Many required concomitant intervention: 46% had a root procedure and 77% an arch repair. There were no in-hospital deaths, and the median postoperative stay was 5 days. At a mean follow-up of 3.9 (±1.1) years, freedom from reoperation was 94%. Late imaging demonstrated: zero trace (25%), 1+ (54%), 2+ (15%) and 4+ (6%) aortic insufficiency (AI). Eleven patients have ≥moderate AI under surveillance, all of whom have a trileaflet valve (21% of trileaflet patients). Four patients required reoperation: 3 for ring dehiscence and 1 for endocarditis.

CONCLUSIONS

Although early results using the HAART device are encouraging, mid-term results raise concern as 21% of trileaflet patients developed recurrent ≥moderate AI by 4 years post-repair. We experienced 3 incidences of ring dehiscence requiring reoperation. Based on this, we recommend caution using the sub-annular approach for stabilization in patients with trileaflet aortic valves. Long-term results are needed to assess outcomes against established techniques.

INTRODUCTION

Aortic valve repair (AVr) has several benefits over aortic valve replacement (AVR) in selected patients with aortic insufficiency (AI). Preservation of native tissue is associated with reduced rates of thromboembolism and endocarditis [1, 2] and superior health-related quality of life [3] and leads to more physiological flow [4]. It is of benefit in younger patients who would otherwise suffer from accelerated structural valve deterioration with a biological prosthesis.

However, AVr techniques and the associated evidence base are not as mature as those underpinning mitral valve repair. Whilst experience and technical improvements are increasing, the Achilles heel of the operation is longevity. This was demonstrated recently in the 2021 European valvular heart disease guidelines, where the committee gave AVr for severe insufficiency only a class IIb (may be considered) recommendation. Further work is needed to establish selection criteria and to assess outcomes of technical improvements as they emerge.

Annular remodelling and stabilization are a critical aspect of AVr, typically achieved through annuloplasty with either an external ring, suture annuloplasty or valve-sparing reimplantation [5]. The Hemispherical Aortic Annuloplasty Reconstructive Technology (HAART) ring (BioStable Science and Engineering, Austin, TX, USA) is a US Food and Drug Administration-approved device first proposed in 2011 [6]. The rings are made of titanium and are covered with a layer of polyester to promote endothelialization. The bicuspid version is circular with 2 sub-commissural posts 180° apart. The trileaflet ring is elliptical, with 3 equally spaced sub-commissural posts. The sub-commissural posts on both rings flare outward by 10°. They are inserted into the sub-annular position with several mattress sutures. The shape of both rings is based on 3D imaging of healthy humans (Fig. 1). Therefore, a theoretical advantage of the HAART ring is remodelling of natural annular shape. As it is placed internally, it avoids the need for external root dissection. The potential disadvantages include lack of long-term data, the risks of ring dehiscence in the outflow tract and the imposition of a single geometry for all patients.

This article adds to the AVr evidence base by presenting the mid-term outcomes of AVr utilizing the HAART ring. This series is the largest single institutional experience internationally (with no commercial affiliation with the manufacturer) and builds on previously published early results [7].

PATIENTS AND METHODS

Ethical statement

This study was ethically approved by the Duke University Institutional Review Board (Pro # 00106459).

Patient population and data collection

The prospectively maintained institutional aortic surgery database at Duke University Medical Centre was used to obtain details for all patients (aged ≥16) undergoing AVr with the HAART ring between September 2017 and June 2023. All patients undergo life-long follow-up at the Duke University Center for Aortic Disease [8], with clinical assessment and valve imaging (either transthoracic echocardiography or magnetic resonance imaging (MRI)) at 6–12 months post-discharge, and then either annually or biennially following this.

Patients undergoing concomitant proximal aortic procedures (including ascending aorta, selective sinus [9], remodelling valve-sparing root and hemi/total arch replacement) were included. Patient demographics, operative details and postoperative outcomes were collected, using Society of Thoracic Surgeons definitions [10].

Indications for surgery were in accordance with internationally published guidelines [11], including severe AI with symptoms or LV dilatation, moderate AI in patients undergoing a proximal aortic repair as the primary indication or annular stabilization in patients with lesser degrees of AI undergoing remodelling valve-sparing root replacement.

Operative techniques

The specific operative techniques used have been previously described in detail [7]. Since September 2019, Schäfers callipers (Genesee BioMedical Inc., Denver, CO) have been used to assess effective/geometric leaflet heights. Contra-indications to repair included the following known risk factors for early repair failure: geometric height of <20 mm for trileaflet valves and <24 mm for bicuspid valves, type C bicuspid commissural orientation (highly asymmetric), cusp calcification and multiple/large fenestrations [12]. Cusp prolapse was treated with plication using techniques described by Schäfers and El Khoury [5].

Video 1 demonstrates a summary of the key operative steps in a patient with a type 1 (R–L fusion) bicuspid valve. The aortic valve (AV) is inspected for any contra-indications to repair, and height is assessed. Annulus is measured with a Hegar dilator, and the free margin of the leaflets is assessed using the HAART ring sizers. The sub-commissural stitches are placed and the prosthesis is lowered into place. The specific methods for this and the looping body sutures are elaborated on in the discussion section. After leaflet prolapse is addressed, the result is assessed, and the proximal graft size is chosen based on a combination of annular and sinotubular junction diameters.

Outcomes and statistics

The primary outcome was the degree of AI (graded in accordance with American Society of Echocardiography guidelines) [13] and transvalvular gradient on late follow-up imaging, and a composite outcome of freedom from reoperation or ≥2+ AI. Secondary outcomes included 30-day/in-hospital complications, need for reoperation for recurrent AI and New York Heart Association (NYHA) functional class. Categorical variables are presented as frequencies. Continuous variables are presented as median with interquartile range. The Kaplan–Meier method was used to understand the survival distribution for the composite outcome (freedom from reoperation or ≥2+ AI). A log-rank test was used to compare differences between those with tri- or bileaflet valves. All analyses were performed using SPSS 27 (IBM, Armonk, NY).

RESULTS

Patient demographics

A total of 71 patients underwent AVr using the HAART device, of whom 53 had a trileaflet valve and 18 had a congenital bicuspid valve. Baseline demographics and procedural characteristics are listed in Table 1. The median age of the cohort was 54 years, and most patients were male (79%). Predictably, the bicuspid group had a lower median age of 44 compared to the trileaflet group (60). Nine of the patients had a confirmed connective tissue disorder on genetic testing (3 with Marfan syndrome, 3 with Loeyz–Dietz syndrome and 3 undefined). At baseline, most of the patients had NYHA I (62%) or NYHA II (32%) class symptoms. The median preoperative annular diameter was 27 mm in both bi- and tricuspid patients. The median maximal aortic diameter of patients undergoing proximal aortic repair was 5.3 cm. Of the 18 bicuspid patients, 1 had a type 0, 16 had a type 1 (R/L fusion) and 1 had a type 1 (R/N fusion) valve.

Procedural characteristics, mortality and morbidity

Most patients required some form of concomitant aortic procedure, with only 2 undergoing isolated AVr: 49% required supracoronary ascending aorta replacement, 46% had an aortic root procedure and 77% required a concomitant arch repair. When compared to the preoperative annular median diameter of 27 mm, the median ring size of 21 mm in both bi- and tricuspid patients demonstrates the degree of annular dilatation in the cohort (the HAART ring sizers give the predicted correct anatomical size of the annulus based on leaflet free margin measurements). On completion of surgery, transoesophageal echocardiography (TEE) demonstrated that most patients had zero to trace AI (87%), with just 13% being graded as 1+. The median gradient on leaving the operating room was 11 mmHg. There were no in-hospital deaths and the median postoperative stay was 5 days. Four patients from the cohort suffered from minor stroke, with no major long-term impairment. Table 2 summarizes the procedural characteristics and early postoperative outcomes.

Survival, freedom from aortic insufficiency and reoperation

After hospital discharge, 4 patients died late of unrelated causes, all of whom had uncomplicated postoperative courses. The first of these was a 65-year-old male who died of an intracranial haemorrhage 8 weeks after surgery. He had a history of hypertension and morbid obesity. He was not on any anticoagulation. The second of these was a 65-year-old male who died of sepsis and myocardial infarction 4 years postoperatively. His death was not thought to be valve related. The third was a 48-year-old male who died of unknown causes 4 and a half years postoperatively. Lastly, a 39-year-old male died in a car crash 5 years postoperatively.

Post-discharge imaging in the form of a transthoracic echocardiogram or MRI was available for 96% of the patients, with 3 of the patients not attending their first follow-up appointment. At the study completion date of 1 June 2023, the mean follow-up duration was 3.9 (±1.1) years, at which point completeness of follow-up and imaging was available for 94% of the patients. At this point, most of the patients were NYHA class I (85%) or II (14%) symptomatically. Imaging demonstrated zero trace (25%), 1+ (54%), 2+ (15%) and 4+ (6%) aortic insufficiency. The median AV gradient (mean) was 10 mmHg. A total of 4 patients (6%) required reoperation on the AV and are described below. Eleven patients (15%) had ≥2+ AI at follow-up and are under surveillance. Of these 11 patients, all had a trileaflet valve such that 21% of the trileaflet cohort have developed recurrent ≥2+ AI. Freedom from the composite outcome of reoperation or ≥2+ AI was 80% at a mean follow-up of 3.9 (±1.1) years (Fig. 2a). Table 3 summarizes the mid-term outcomes. A log-rank test was run to determine if there were differences in the distribution for tri- and bicuspid valves for the composite outcome, which demonstrates a trend towards significance, χ²(2) = 2.730, P = 0.098 (Fig. 2b).

Detailed description of the 4 patients requiring reoperation

Patient 1

49-year-old male with Marfan syndrome, a trileaflet AV with moderate AI, a 31-mm annulus and a 5.3-cm aortic root who underwent valve-sparing root (Yacoub) and hemi-arch replacement. A 23-mm HAART 300 ring was implanted along with plication of the non-coronary cusp for cusp prolapse. Postoperative TEE demonstrated trace AI with minimal transvalvular gradient. At routine 29-month surveillance follow-up, postoperative imaging demonstrated severe AI. At reoperation, he was found to have dehiscence of the HAART ring beneath the right-non commissure due to breakage of the adjacent commissural plication suture, with a hole at the base of the non-coronary cusp (Fig. 3). The patient underwent redo sternotomy for ring explant and AVR with a 23-mm mechanical valve (On-X, Cryolife, Kennesaw, GA, USA) and is NYHA class I now 30 months postoperatively.

Patient 2

35-year-old male with a bicuspid AV, moderate AI, a 25-mm annulus and a 5-cm ascending aorta who underwent AVr, ascending and hemi-arch replacement and repair of partial anomalous pulmonary venous return. A 21-mm HAART 200 ring was implanted with plication of both the non-coronary and fused right-left cusps. Completion TEE demonstrated no AI with a 16-mmHg mean gradient. At routine 9-month surveillance follow-up, he developed NYHA II symptoms and postoperative imaging demonstrated severe AI, a new peri-membranous ventricular septal defect (VSD) and moderate mitral regurgitation (MR) originating from the base of the anterior mitral valve leaflet (Video 2). At reoperation, it was found that he had dehiscence of the HAART ring along the non-coronary cusp due to breakage of the looping suture nearest the right non-commissure. Similar to the prior case of ring dehiscence, there was a large perforation at the base of the non-coronary cusp (presumably secondary to repeated impact with the dehisced ring). There was also a perforation of the anterior leaflet of the mitral valve and a peri-membranous VSD, both presumably also due to local trauma from the dehisced ring. The patient underwent redo sternotomy for AVR with a 21-mm On-X mechanical valve, mitral valve repair (primary repair of defect) and primary repair of the VSD and is NYHA class I now 39 months postoperatively.

Patient 3

66-year-old male with a trileaflet AV, severe AI, a 27-mm annulus and a 5.8-cm ascending aorta who underwent valve-sparing Yacoub selective sinus replacement (right and non-coronary sinuses replaced). A 23-mm HAART 300 ring was implanted along with plication of the right coronary cusp for right cusp prolapse. At his 1-year follow-up visit, transthoracic echocardiography demonstrated zero AI. At 22 months postoperatively, he developed enterococcal endocarditis possibly related to orthopaedic interventions on his hip and knee, which was initially managed medically with clearance of his infection. He subsequently developed severe AI with NYHA class III symptoms, and in view of his co-morbidities including stage III chronic kidney disease (CKD), underwent successful valve-in-ring transcatheter aortic valve replacement (TAVR) [14, 15] with a 23-mm Edwards Sapien 3 Ultra valve (Edwards Lifesciences, Irvine, CA, USA). Video 3 demonstrates the flexibility of the HAART ring (which allows it to conform for valve-in-ring TAVR) and the patient’s subsequent successful TAVR. The patient is NYHA class I now 16 months postoperatively.

Patient 4

49-year-old male with a trileaflet AV, severe AI, moderate LV impairment, a 29-mm annulus and a dilated ascending aorta who underwent urgent AVr, ascending and hemi-arch replacement. A 23-mm HAART 300 ring was implanted with no concomitant leaflet plication. Postoperative TEE demonstrated zero AI and a 10-mmHg mean gradient. His postoperative stay was complicated by acute kidney injury in the setting of pre-existing stage III CKD. At his 2-year follow-up, transthoracic echocardiography demonstrated severe AI. As he was asymptomatic, this was initially managed conservatively. At 48 months postoperatively, in view of progressive symptoms, he underwent valve-in-ring transcatheter AVR with a 26-mm Edwards Sapien 3 Ultra Resilia valve (Edwards Lifesciences); open operative intervention was not considered in view of morbid obesity (body mass index 48 kg/m²), severe lung disease and stage IV CKD. After deployment, TEE demonstrated paravalvular AI despite a second balloon dilatation. The residual AI was in the region of the left coronary cusp secondary to ring dehiscence, which had not been fully appreciated on his preoperative imaging. In view of this, as a separate procedure a few days later, 2 Amplatzer vascular plugs (AVP-2 10 mm × 7 mm and AVP-2 8 mm × 7 mm; Boston Scientific Corporation, USA) were successfully deployed in the region of ring dehiscence, with the distal discs sitting below the HAART ring. TEE subsequently demonstrated mild residual paravalvular AI. Video 3 demonstrates the procedural and imaging aspects of the above procedures. The patient is NYHA class II now 1 month postoperatively.

Potential impact of learning curve

Of the 71 HAART procedures performed, ring dehiscence occurred in cases 15, 42 and 62. To assess the impact of learning curve, the cohort was split into tertiles: the composite end point of freedom from reoperation or ≥2+ AI occurred a total of 4, 5 and 5 times in the consecutive tertiles, suggesting that learning curve did not have an impact at the mid-term.

DISCUSSION

AVr offers potential benefits over replacement including a reduction in thromboembolism and endocarditis risk [16], superior health-related quality of life [3] and a reduction in the need for anticoagulation. There is often a superior haemodynamic profile compared to the use of a stented prosthesis. The last 20 years have seen progress in the conceptual basis behind AVr. The key developments include: development of the valve-sparing root procedure [17, 18], development of cusp prolapse repair techniques [19, 20], an improved understanding of the predictors of repair failure [21], an improved understanding of left ventricular outflow tract anatomy [22] and the development of specialized instruments to measure effective and geometric heights [5]. Despite these developments, the procedure is done in a minority of eligible patients. Further, the most recent European guidelines give isolated AVr for severe insufficiency only a class IIb (may be considered) recommendation, demonstrating the need for further work in this field.

Similar to the development of mitral valve repair in the 1980s, the importance of annular stabilization for AVr has become clear. Three observations have contributed towards this: slightly higher failure rates of Yacoub valve-sparing root replacement in some patients secondary to late annular dilatation [23], studies demonstrating larger annular diameter as a risk factor for failure after repair [21] and studies demonstrating the improvement of surgeons’ results following the introduction of annular stabilization [24]. The introduction of annular stabilization during AVr, in addition to other techniques, has led to outstanding results, with some authors reporting >95% freedom from reintervention at 8 years [2].

The HAART ring is a US Food and Drug Administration-approved circumferential annuloplasty device with 2 novel features [6]. First, it not only down-sizes the annulus to its predicted correct size, but it also remodels it to its natural shape. This is not currently possible with external solutions. Second, it is inserted into the sub-annular position, meaning that external root dissection is not required [24]. The ease with which sizing is performed also allows a degree of standardization in the repair process. As with any new technology, a thorough assessment of both early and late outcomes is required. This study is the largest single institutional experience presenting the mid-term results of this device. Previous similar-sized series have presented results pooled from multiple institutions [25]. Further, those prior studies were funded by the device manufacturer, who oversaw the data analysis, thereby introducing the potential for conscious or unconscious bias [26]. Finally, our dedicated aortic clinic allows reliable follow-up and imaging surveillance [8]. Overall, 4 specific themes have emerged from our experience.

First, the safety and short-term results of the HAART device have been excellent. There were no in-hospital deaths, and the median postoperative stay was 5 days. No patients left the operating room with more than trace or mild AI. The safety profile of the device is encouraging in the context of the complex cohort: well over half required a concomitant arch procedure and around half required a leaflet prolapse repair, reflective of our referral base as a tertiary aortic centre. In our previously published results, at a median follow-up of a year, there were no valve-related reoperations, and only 1 patient was under surveillance for moderate AI [7]. Although these early results match the highest quality valve repair outcomes in the literature in general, a caveat to this is that the average mean gradient on completion of the repairs was 11 mmHg, higher than that seen in some of the large published series utilizing other more traditional techniques [27].

Second, the mid-term results are not as encouraging. At a mean follow-up of 3.9 (±1.1) years, 11 of the patients (15%) have ≥moderate aortic insufficiency at the most recent follow-up. They have no clinically significant symptoms (i.e. >NYHA II) and are under close surveillance. However, it will be critical to follow the trajectory of these patients as they are clearly at risk of needing reintervention in the future. As all of these patients have trileaflet valves (21% of trileaflet patients), it suggests that this may be a risk factor for early repair failure when using the HAART ring. It is unclear why this is, but we hypothesize that it may have a biomechanical basis. Since bicuspid patients undergoing repair are likely to have almost symmetric commissures, the anatomical fit of a HAART ring (and the resulting biomechanics) is likely to be optimal. In tricuspid patients, the imposition of a fixed geometry (when commissural positions may not be perfectly equally spaced) could increase the chances of biomechanical failure. A detailed review of the postoperative imaging of these 11 patients suggests that the AI is predominantly central (i.e. secondary to an annular issue), supporting the above hypothesis.

Thirdly, 3 of the patients required reoperation due to ring dehiscence. This complication is particularly problematic with the HAART prosthesis, as understandably nearly all patients with ring dehiscence are likely to require open reoperation rather than valve-in-ring TAVR, although the complication was able to be managed satisfactorily in 1 high-risk patient with solely transcatheter techniques. Further, as demonstrated in our experience described above, dehiscence in the sub-annular position can lead to additional damage of the ventricular septum and both the aortic and mitral valve cusps, which must be addressed at the time of open reoperation. This additional damage of surrounding structures in the setting of ring dehiscence may be made more likely by the 10° outward flaring of the commissural posts of the rings, which will repeatedly contact the surrounding structures over multiple cardiac cycles. In addition to the above 3 patients, we are closely following an 18-year-old with Marfan syndrome (who underwent AVr with a 21-mm HAART 200 ring and Yacoub valve-sparing root) who has postoperative moderate MR at his 2-year follow-up imaging. MRI suggests that this is secondary to a small perforation in the aorto-mitral continuity at the base of the A2 segment (Video 4). Other authors have had similar experience with dehiscence [25]. It will be important for all implanters of the HAART prosthesis to report incidences of dehiscence over the coming years. Further experiences may suggest that the manufacturer considers examining the sizing algorithm (i.e. making the degree of downsizing more conservative or putting a limit on the degree of preoperative annular dilatation for which the device can be used), or the methods they recommend to use fixing the ring in place (i.e. the number and strength of sutures used).

With regard to the aggressiveness of the sizing algorithm, it is notable that no patients in the entire series sized for a 25-mm ring (Table 2) when following the manufacturer’s sizing algorithm, despite some patients having an annulus of >30 mm in diameter. As per the ring sutures, we have switched from using the manufacturer-recommended 4–0 polypropylene sutures to anchor the ring [6] to using a 3–0 polypropylene suture for the post sutures and 2–0 braided nylon sutures (identical to those used in surgical AVR) for the looping sutures. The further advantage of the use of these braided nylon looping sutures (Video 1) is that they avoid the need for the use of the manufacturer-recommended ‘lateral suture fixation’ technique [6], since there is no risk of the softer nylon sutures perforating the AV cusps should contact occur, which shortens the aortic cross-clamp time.

For completeness, a systematic root cause analysis was performed to identify potential causative factors for ring dehiscence using the framework outlined by Charles et al. [28]. This identified the following potential causative factors: patient related (severely dilated aortic annulus), prosthesis related (material properties of the ring, manufacturer’s sizing algorithm, appropriateness of placing the ring in the sub-annular position and the appropriateness of the associated pre-approval testing process) and implant related (type of suture material used and appropriateness of pre-op suture strength testing by manufacturer) (see Supplementary Material, Addendum S1 for further details). Post-approval failure of cardiovascular medical devices is not uncommon and historically requires numerous voluntary reports to regulators as well as published evidence before further investigation is undertaken [29, 30].

Lastly, 2 patients requiring reoperation underwent a successful valve-in-ring TAVR with an Edwards Sapien 3 Ultra. Other groups have also had similar successes [14, 15]. Based on the material properties of the HAART ring, we believe that the Sapien (i.e. balloon expandable) may be the valve of choice in this scenario as the ring will conform to the shape of the valve (Video 3).

Limitations

This study has several limitations. Despite being the largest single institutional experience, the overall numbers are modest (particularly in the bicuspid cohort) and further follow-up is needed, especially given the 21% incidence of recurrent ≥2+ AI (under surveillance) noted in the trileaflet cohort at only 3.9 years postoperatively. As a tertiary referral centre for a large population catchment area, our practice is not necessarily reflective of real-world practice, as demonstrated in the complexity of the procedures required (only 2 patients from the entire cohort required isolated AVr). Longer-term results are required before coming to firm conclusions on the safety and effectiveness of annular stabilization using the HAART device.

CONCLUSION

Although the early results of AVr using the HAART device were encouraging, there is reason for concern at the mid-term. We experienced 3 incidences of ring dehiscence which required reoperation. In addition, almost 15% of the cohort has developed moderate AI requiring surveillance. Based on this, we suggest caution in using the sub-annular approach for stabilization, particularly in trileaflet-valved patients.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

Conflict of interest: none declared.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Author contributions

Omar A. Jarral: Data curation; Formal analysis; Project administration; Writing—original draft; Writing—review & editing. Christopher W. Jensen: Data curation; Formal analysis. Julie W Doberne: Data curation; Formal analysis; Writing—original draft. Peter S. Downey: Data curation; Formal analysis. J.D. Serfas: Formal analysis; Validation; Writing—review & editing. Andrew M. Vekstein: Formal analysis; Writing—original draft. G. Chad Hughes: Conceptualization; Formal analysis; Investigation; Methodology; Project administration; Resources; Supervision; Validation; Writing—original draft; Writing—review & editing.

Reviewer information

European Journal of Cardio-Thoracic Surgery thanks Ruggero De Paulis, Takashi Kunihara and the other, anonymous reviewers for their contribution to the peer review process of this article.

REFERENCES

ABBREVIATIONS

ABBREVIATIONS
 
• AI

  Aortic insufficiency

 
• AV

  Aortic valve

 
• AVr

  Aortic valve repair

 
• AVR

  Aortic valve replacement

 
• CKD

  Chronic kidney disease

 
• HAART

  Hemispherical Aortic Annuloplasty Reconstructive Technology

 
• NYHA

  New York Heart Association

 
• TEE

  Transoesophageal echocardiography

 
• VSD

  Ventricular septal defect