Aortic valve performance after remodelling versus reimplantation in a propensity-matched comparison

Abstract

OBJECTIVES

Both aortic root remodelling and aortic valve (AV) reimplantation have been used for valve-sparing root replacement in patients with aortic root aneurysm with or without aortic regurgitation. There is no clear evidence to support one technique over the another. This study aimed to compare remodelling with basal ring annuloplasty versus reimplantation on a multicentre level with the use of propensity-score matching.

METHODS

This was a retrospective international multicentre study of patients undergoing remodelling or reimplantation between 2010 and 2021. Twenty-three preoperative covariates (including root dimensions and valve characteristics) were used for propensity-score matching. Perioperative outcomes were analysed along with longer-term freedom from AV reoperation/reintervention and other major valve-related events.

RESULTS

Throughout the study period, 297 patients underwent remodelling and 281 had reimplantation. Using propensity-score matching, 112 pairs were selected and further compared. We did not find a statistically significant difference in perioperative outcomes between the matched groups. Patients after remodelling had significantly higher reintervention risk than after reimplantation over the median follow-up of 6 years (P = 0.016). The remodelling technique (P = 0.02), need for decalcification (P = 0.03) and degree of immediate postoperative AV regurgitation (P < 0.001) were defined as independent risk factors for later AV reintervention. After exclusion of patients with worse than mild AV regurgitation immediately after repair, both techniques functioned comparably (P = 0.089).

CONCLUSIONS

AV reimplantation was associated with better valve function in longer-term postoperatively than remodelling. If optimal immediate repair outcome was achieved, both techniques provided comparable AV function.

Open in new tabDownload slide

INTRODUCTION

More than 4 decades have passed since Magdi Yacoub first performed valve-sparing aortic root remodelling in patient with annuloaortic ectasia [1]. The possibility to preserve the aortic valve (AV), restore its function and replace the dilated part of ascending aorta has become a game-changing concept in approach to aortic root and/or regurgitant AV. By implementation of standardized cusp repair and aortic annulus stabilization [2, 3], excellent outcomes have been reported regardless of the aortic regurgitation severity [4, 5].

Another valve-sparing aortic root replacement (VSARR) technique was published by Tirone David in 1992 who reimplanted native AV into a tube Dacron graft [6]. The technique has experienced several modifications towards imitating the native aortic root with its sinuses [7, 8], bulged ‘Valsalva’ graft being the most popular [9]. The group of David [10] and others [11, 12] have reported excellent long-term outcomes.

There has been a constant debate over which of the 2 above-mentioned techniques is superior. Advocates of remodelling argue by shorter cross-clamp times [13] and in vitro evidence of better haemodynamic valve characteristics [14]. The supporters of reimplantation believe it might provide better intraoperative haemostasis [15] and superior long-term outcomes in connective tissue disorders (CTD) [16]. None of previous studies has observed a significant difference in long-term AV reintervention rate [12, 17]. Moreover, there is only scarce evidence to compare reimplantation against remodelling with annuloplasty, which are now both considered equal in terms of aortic annulus stabilization [18, 19].

The aim of this study was to compare AV reimplantation (reimplantation; David procedure) and aortic root remodelling including basal ring annuloplasty (remodelling; Yacoub procedure) with regard to the longer-term freedom from AV reoperation/reintervention, recurrence of AV regurgitation, and other major valve-related events.

PATIENTS AND METHODS

Study design and patient selection

This was a retrospective study of all consecutive patients undergoing remodelling at 6 institutions (11 surgeons) and reimplantation at 2 institutions (8 surgeons) between 2010 and 2021 (Supplementary Material, Table S1). Urgent surgery due to aortic dissection or root dimensions was included in the analysis. Root remodelling without basal ring annuloplasty, hemi-Yacoub, Florida sleeve and other alternative techniques were excluded. Postoperatively, all patients were annually followed in the VSARR centres or by their outpatient cardiologists. The data of interest were manually collected into a pre-formatted table sheet at all institutions. The data source were patient in-hospital records and records of outpatient echocardiographic controls. In case of follow-up, missingness, patients, their outpatient cardiologists or general practitioners were contacted by telephone. Patients unreachable by any means for more than 2 years before this study end were considered lost to follow-up. The end of follow-up was set at 31 December 2022.

ETHICS STATEMENT

The study was approved by the institutional review board (Ethics committee at the University Hospital Hradec Kralove, 13 July 2023, ID number 202307 J01). Informed patient consent was waived.

Study end-points

Primary end-point was defined as longer-term postoperative AV performance expressed by freedom of AV or root reoperation/reintervention, freedom from recurrence of at least moderate AV regurgitation and both of them together as a composite end-point. Secondary end-points included perioperative outcomes, longer-term survival, freedom from endocarditis and from major bleeding/thromboembolism. These end-points were defined in accordance with guidelines reported by Akins et al. [20].

Surgical techniques

Both remodelling and reimplantation were described in detail elsewhere. Briefly, in remodelling, aortic wall was resected down to the level of AV and replaced by tongues of Dacron Valsalva graft (Terumo Aortic, Vascutek Ltd, Renfrewshire, Scotland, UK). Aortic annuloplasty was performed in all cases either by an expandable annuloplasty ring (Coroneo, Extra-Aortic, Montreal, QC, Canada) [4], a Dacron ring [18] or a polytetrafluorethylene suture (Gore-Tex CV-0, WL Gore and Associates, Munich, Germany) based on surgeon preference [13].

In reimplantation, AV with its commissures was reimplanted into a Valsalva Dacron graft (Terumo Aortic, Vascutek Ltd, Renfrewshire, Scotland, UK or Cardioroot Woven Aortic, Getinge AB, Göteborg, Sweden). Several graft-sizing strategies were described over the years [9, 21].

In both techniques, AV cusps were addressed whenever necessary based on standardized intraoperative assessment of geometric and effective height in all patients [2]. Cusp prolapse was corrected by means of central plication. Triangular resection, shaving or decalcification were performed rarely (Table 2). Coronary arteries were reimplanted, and distal aortic anastomosis was performed. A wide range of concomitant procedures were performed as indicated (see Table 1, Supplementary Material, Tables S2 and S4).

Statistical analysis

Continuous variables were expressed as median and interquartile range, categorical variables as absolute and relative frequencies. All comparisons in brackets were ordered as remodelling versus reimplantation.

The nonparametric Wilcoxon test (unpaired, two-sided alternative) with Bonferroni correction was used to assess differences between groups for all numeric variables and Fisher’s exact test for categorical variables. Missing variables were limited in number and considered random—Mice package for multivariate imputation by chained equations was used for prediction. Propensity-score matching (PSM) was performed using the R package MatchIt (using 23 preoperative variables, see Supplementary Material, Figs S1 and S2). The method of ‘matching with caliper’ was selected to minimize propensity scores. Propensity scores were assessed by general linear model and k-nearest neighbours’ method was then applied. Maximum difference in matched cohorts was set at 0.15. Standardized mean differences were reported for each variable.

Time-to-event analyses were assessed using the Kaplan–Meier method with R packages survminer and survival. P-value was determined using the log-rank test. Hazard ratios (HRs) for AV reinterventions were estimated using the Cox proportional hazards regression model, including also the need for central plication, decalcification and degree of immediate postoperative AV regurgitation. Secondarily, a sub-analysis of the patients leaving the operating room with optimal repair outcome only [i.e. excluded patients with worse than mild AV regurgitation on transoesophageal echocardiography (TEE) in the operating room after the procedure] was performed with the use of Kaplan–Meier method. Another sub-analysis of reinterventions excluding the reoperations for endocarditis was also performed. Fine Gray analysis of competitive risk with age, sex, mortality and body mass index was performed using the R package riskRegression version 2023.12.21.

A P-value under 0.05 was considered significant. All statistical analyses were performed in R (The R Foundation for Statistical Computing, Vienna, Austria, version 4.0.3) using RStudio (RStudio, Inc., Version 1.2.5042) and the packages tidyverse, stats, ggthemes, cowplot, MatchIt, mice, FSA, dlookr and ggprism.

RESULTS

Throughout the study period, 297 patients underwent remodelling and 281 reimplantation (Fig. 1). Before PSM, there were several substantial differences in cohort characteristics between the groups (Supplementary Material, Table S2). Patients in remodelling group had more pronounced annular and root dilation, while in reimplantation, a dilation of sinotubular junction and supracoronary ascending aorta were more prominent. Remodelling group had a higher prevalence of bicuspid AV and a higher AV regurgitation degree. Reimplantation was used more frequently for acute aortic dissection, and patients had a higher operative risk in terms of EuroSCORE II. The prevalence of CTD was comparable (Marfan, Ehlers-Danlos, Loeys-Dietz and Turner syndrome, only genetically proven syndromes included).

The patients after remodelling had longer aortic cross-clamp time, higher degree of AV regurgitation on the TEE immediately post-procedure and higher blood loss (Supplementary Material, Table S3). Significantly more reoperations/reinterventions were observed after remodelling than after reimplantation (32 vs 5 events) over median follow-up of 6.1 vs 5.5 years (follow-up loss of 5.1% vs 3.9%). Estimated 10-year freedom from event was 78.4% vs 93.3% in favour of reimplantation (P < 0.0001; Supplementary Material, Fig. S3). Similarly, there were more recurrences of at least moderate aortic regurgitation (P = 0.0002; Supplementary Material, Fig. S4) and composite end-point (P < 0.0001; Supplementary Material, Fig. S5) after remodelling. Longer-term survival (P = 0.64), freedom from endocarditis (P = 0.15) and freedom from bleeding/thromboembolism (P = 0.43) were comparable (Supplementary Material, Figs S6–S8).

Using the PSM, 112 pairs were formed. The matching successfully eliminated all preoperative differences between the groups (Table 1). The remodelling group exhibited longer aortic cross-clamp times (147 vs 131 min; P < 0.001) and patients had a marginally higher risk of re-exploration for bleeding (19% vs 5%; P = 0.06). We did not find a statistically significant difference in perioperative mortality, degree of AV regurgitation postoperatively, or perioperative complications between the groups (Table 2).

Over the median follow-up of 6.5 vs 5.1 years (follow-up loss 6.2% vs 2.7%), remodelling was associated with significantly lower freedom from reintervention than reimplantation (P = 0.016; Fig. 2). Estimated freedom from reoperation/reintervention was 89.9% vs 97.7% at 5 years and 84.1% vs 95.9% at 10 years, respectively. We observed 13 reinterventions after remodelling post-matching: 5 cases of endocarditis requiring root replacement with homograft or Bentall-De Bono procedure, 3 times tearing of central plication, 2 cases of cusp restriction, 1 new prolapse (all treated by valve replacement), 1 prolapse and root dilation in patient with Marfan syndrome treated with mechanical Bentall-De Bono procedure and 1 valve degeneration with stenosis treated with transcatheter valve implantation. Median time to reoperation/reintervention was 2.4 years. In matched reimplantation group, there were 3 reinterventions caused by tearing of central plication, all treated by surgical valve replacement. Median time to reintervention was 5.1 years.

Cox proportional hazard analysis was applied as an alternative adjustment method to evaluate the longer-term reoperation/reintervention risk (Fig. 3). Three significant risk factors were identified by this method, and thus the remodelling technique (HR 3.27; P = 0.02), the need for decalcification (HR 4.53; P = 0.03) and the degree of immediate postoperative AV regurgitation (HR 2.03; P < 0.001).

With regard to this finding, Kaplan–Meier analysis of longer-term AV reintervention risk was repeated, while patients with suboptimal repair outcome were excluded from the comparison (pre- and post-matching). Although the difference remained in favour of reimplantation pre-matching (P = 0.001; Supplementary Material, Fig. S9), we did not find a significant difference between the surgical techniques post-matching (92.4% vs 97.7% at 5 years; 88.6% vs 95.9% at 10 years; P = 0.089; Fig. 4). According to Fine Gray competitive risk analysis, age was shown to be a statistically significant competitive risk factor to reintervention in the longer-term postoperatively (see Supplementary Material, Table S5).

Recurrence of at least moderate aortic regurgitation in the follow-up and composite primary end-point also exhibited a similar trend in favour of reimplantation, although not statistically significant (P = 0.17 and P = 0.06; Fig. 5A and B). We did not find a statistically significant difference in longer-term survival: 96.9% vs 97.4% at 5 years and 92.4% vs 80.4% at 10 years (P = 0.59; Fig. 5C). Freedom from endocarditis and major bleeding/thromboembolism were also comparable (Supplementary Material, Figs S10 and S11). An analysis of AV reoperations without procedures for endocarditis remained in favour of reimplantation pre-matching (P < 0.001), but not post-matching (P = 0.11; Supplementary Material, Fig. S12 and S13).

DISCUSSION

VSARR is an attractive alternative to Bentall-De Bono procedure providing better long-term freedom from valve-related complications and eventually survival [19, 22]. Both AV reimplantation and aortic root remodelling, particularly with annuloplasty, are nowadays considered comparable in terms of long-term outcomes and choice of technique is based on surgeon/centre preference. However, there is still a lack of reliable data on this topic and the debates go on.

Fries et al. examined the hydrodynamics of remodelling versus reimplantation on porcine roots in vitro [14]. They described a significantly smoother valve movement in remodelling configuration, which could eventually lead to slower valve degeneration over time. In contrast, Di Leonardo et al. recently performed a similar comparison including remodelling, reimplantation and reimplantation into Valsalva graft [23]. The last-mentioned option provided the most similar characteristics to native root with regard to energy loss and valve opening. Nevertheless, clinical implications of these findings remain questionable.

Most clinical studies elaborating on this topic are retrospective, single-centre and with limited statistical adjustment [11, 18, 24, 25]. A large meta-analysis by Arabkhani et al. [12] and best evidence topic by Maskell et al. [17] both concluded that remodelling and reimplantation are comparable in terms of long-term survival and risk of AV reintervention. Despite the fact that both studies were robust, they did not consider eventual slight differences in cohort characteristics and patient selection. The inequalities in root and ventricle dimensions, degree of regurgitation, valve configuration, presence of congenital tissue disorders [16] and acute aortic dissection [26] play substantial roles in outcome and must be considered for proper comparison of surgical strategies. Moreover, only the work of Lenoir et al. has analysed reimplantation against remodelling with annuloplasty [18].

To the best of our knowledge, this study is the first to provide multicentre detailed comparison of both surgical strategies in their current technical form, and to include extensive echocardiographic and demographic patient characteristics in statistical adjustment.

We observed several substantial differences in baseline cohort characteristics, particularly in terms of root dimensions, prevalence of bicuspid AV and urgent settings. Reimplantation was more liberally used in acute aortic dissection and despite this, the postoperative blood loss was significantly lower than after remodelling. Moreover, there was a clinically important difference in the need of re-exploration for bleeding between the groups (although the statistical difference was merely marginal after Bonferroni correction). This outcome may support the hypothesis that reimplantation is more haemostatic than remodelling [15].

An important observation was the difference in degree of immediate postoperative AV regurgitation on TEE in favour of reimplantation. Whether this was a consequence of limited versatility of remodelling technique, more advanced AV/root disease, proceeding with repair in dubious valve morphology or a surgical flaw will remain a matter of debate. Nevertheless, this outcome was identified as a significant risk factor for later AV reoperation/reintervention (Fig. 3). It thus leaves a strong exclamation point that the patient must not leave the operating room with greater AV regurgitation than mild. The valve must be immediately re-repaired or replaced in order to provide durable patient benefit (Fig. 4).

The long-term risk of reoperation/reintervention is the most debated issue when it comes to VSARR. This analysis demonstrates a significant difference in favour of reimplantation, which is in contrast with previous studies [12, 17, 18, 25]. A similar finding was reported by David et al. [27], although not statistically significant (HR 3.37; P = 0.07). Klotz et al. [24] also observed a non-significant difference in favour of reimplantation (11.7 vs 5.8% at 10 years postoperatively; P = 0.65). Very recently, a large meta-analysis was published by Sá et al. and favoured reimplantation in terms of long-term survival (HR = 2.15; P < 0.001) and the risk of AV reintervention (HR = 1.49; P = 0.019) [28]. In our cohort, the most frequent reason for reoperation was infectious endocarditis, and its association with remodelling is questionable. All but 1 of those patients left the operating room with optimal repair outcome. Only 1 endocarditis patient had required early re-exploration for bleeding, and endocarditis incidence was distributed among several centres. A sub-analysis not considering the endocarditis reoperations preserved the same trend as whole cohort, particularly pre-matching. The incidence of tearing of central plication was equal in both remodelling and reimplantation group, and again, was fairly distributed among different centres (1 tear per centre in remodelling; 2 vs 1 in reimplantation). However, several cases of new cusp retraction, prolapse or degeneration were observed purely after remodelling, but not reimplantation. Annular dilation was not an issue after remodelling as annuloplasty was employed. Need for central plication and decalcification was comparable in both groups. Importantly, if the patient left the operating room with an optimal repair, we did not observe a significant difference between the techniques anymore (shown in Fig. 4). When it comes to the preference of straight tube versus Valsalva graft for reimplantation, conflicting outcomes have been reported [9, 27] and it remains a matter of debate [29].

When evaluating the AV regurgitation recurrence, this was in favour of reimplantation before PSM (76.7% vs 93.8% at 10 years; P < 0.001). The same trend was preserved post-matching, although not statistically significant (79.3% vs 89.2% at 10 years; P = 0.17).

There is some evidence that reimplantation is superior to remodelling in patients with CTD. Elbatarny et al. reported a significantly higher risk of reintervention after remodelling than reimplantation (18% vs 0% after 20 years; P = 0.018). The dominant reason for reintervention was valve failure due to aortic annulus dilation (not stabilized in original remodelling technique). In our whole cohort, the prevalence of CTD was comparable in both groups. We observed no reintervention after reimplantation. In remodelling group, there were 5 reoperations (pre-matching): 2 for endocarditis, 1 for acute aortic dissection, 1 for aortic cusp restriction and annular dilation and 1 complex finding (aortic root dilation, dislocation of annuloplasty ring and a new cusp prolapse).

Regarding the topic of VSARR in conditions of acute aortic dissection, the groups of Sievers et al. [26] and Subramanian et al. [30] did not observe any significant difference between techniques. Both showed an acceptable postoperative survival and valve performance. Our study did not provide enough substrate to better analyse this topic. In all our patients, the indication for VSARR was extension of dissection into the aortic root. None of 38 dissection patients in reimplantation group required reintervention, 1 had a recurrence of aortic regurgitation 1 year postoperatively. Of 7 patients undergoing remodelling for dissection, 2 developed a new aortic regurgitation (5 and 7 years postoperatively), one of them requiring reoperation.

Aside from AV reinterventions, the longer-term postoperative survival remained very acceptable in both groups. The risk of major bleeding or thromboembolism was very low, thus confirming that both valve-sparing alternatives provide effective reduction of adverse valve-related events in comparison to Bentall-de Bono procedure [19, 22] and should be preferred strategy at dedicated centres.

Study limitations

The most important limitation of our study is its retrospective character. Despite the meticulous effort towards balancing input variables, many potential confounders remain: degree of commissural orientation in bicuspid valves, quality of cusp tissue and an individual surgeon threshold of valve repairability. There was no echocardiographic core-lab to re-evaluate the pre-, peri-operative and follow-up exams. Postoperative echocardiographic parameters of AV repair (root dimensions, effective and geometric height, coaptation height, etc.) were not generally measured, particularly in the earlier stages of the study period. We compared data of 2 high volume reimplantation centres with 6 centres performing remodelling with various level of experience, indirectly demonstrated by longer operation times and higher tendency towards bleeding in this group. However, there was no difference in longer-term outcomes amongst the remodelling centres, the freedom from AV failure and reoperation were proportionally equal amongst them. The volume–outcome association remains a matter of debate, nevertheless.

CONCLUSION

In the international multicentre propensity score-matched comparison, AV reimplantation was associated with better freedom from AV reoperation/reintervention than aortic root remodelling in the longer-term postoperatively. If the patient left the operating room with an optimal AV repair (no greater than mild AV regurgitation on TEE), we did not observe a significant difference between the techniques anymore. Survival and freedom from major adverse events were comparable among the groups.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

ACKNOWLEDGEMENTS

The authors thank Dr Aneta Mazouchova for performing the statistical analysis in this manuscript.

FUNDING

The work was supported by an institutional program Cooperatio, area Cardiovascular science.

Conflict of interest: Tomas Holubec discloses receiving consultancy fees and honoraria from Getinge AB, Göteborg, Sweden; however, not specifically for this work. The other authors have nothing to declare.

DATA AVAILABILITY

Dataset underlying this manuscript will be shared on a reasonable request to the corresponding author.

Author contributions

Jan Gofus: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Writing—original draft. Jan Vojacek: Funding acquisition; Supervision; Validation; Writing—review & editing. Mikita Karalko: Investigation. Pavel Zacek: Supervision; Writing—review & editing. Adrian Kolesar: Investigation; Supervision; Validation. Tomas Toporcer: Investigation. Martin Urban: Investigation; Supervision. Filip Glac: Investigation. Stepan Cerny: Investigation; Supervision; Validation. Pavel Homola: Investigation; Validation; Writing—review & editing. Jaroslav Hlubocky: Investigation; Supervision; Validation. Andrey Slautin: Investigation. Petr Fila: Investigation; Supervision; Validation. Daniela Zakova: Investigation. Jan Sterba: Investigation. Hiwad Rashid: Investigation. Arnaud Van Linden: Investigation. Tomas Holubec: Conceptualization; Investigation; Methodology; Supervision; Writing—original draft.

Reviewer information

European Journal of Cardio-Thoracic Surgery thanks the anonymous reviewer(s) for their contribution to the peer review process of this article.

REFERENCES

ABBREVIATIONS

ABBREVIATIONS
 
• AV

  Aortic valve

 
• CTD

  Connective tissue disorders

 
• PSM

  Propensity-score matching

 
• TEE

  Transoesophageal echocardiography

 
• VSARR

  Valve-sparing aortic root replacement