Aortic valve repair in adults: long-term clinical outcomes and echocardiographic evolution in different valve repair techniques

Abstract

OBJECTIVES

Aortic valve repair/sparing techniques have been established as effective treatments for aortic regurgitation and/or aortic aneurysms. However, concerns remain regarding long-term durability, reproducibility and patient selection. This study aims to asses long-term clinical and echocardiographic outcomes, with a focus on aortic regurgitation grade and left ventricular ejection fraction evolution, in adults undergoing these procedures.

METHODS

Adult patients in the Heart Valve Society Aortic Valve Database, undergoing any aortic valve repair/sparing technique were included. Time-to-event analyses were used for clinical outcomes and mixed-effects models for left ventricular ejection fraction and aortic regurgitation grade evolution. Techniques: isolated valve repair (group 1), ascending aortic replacement + valve repair (group 2), partial-root replacement ± valve repair (group 3) and valve-sparing root replacement ± valve repair (group 4).

RESULTS

Survival at 10 years was comparable to survival of the matched-general-population in each group. The 10-year cumulative incidence of reintervention was 19.5% [95% confidence interval (CI) 16.70–22.40%] in group 1 [including only double external annuloplasty in group 1; reintervention was not significantly different between techniques (P = 0.112)]; 13.8% (95% CI 10.10–18.10%) in group 2; 12.7% (95% CI 5.50–22.90%) in group 3; and 8.5% (95% CI 7.00–10.10%) in group 4 (P < 0.001). Severe preoperative aortic regurgitation grade [hazard ratio 1.95 (95% CI 1.19–3.21), P < 0.001] and left ventricular end-diastolic diameter [hazard ratio 1.03 (95% CI 1.00–1.05), P < 0.001] were predictive of reintervention in group 4; patch use was a predictor in all groups. The predicted left ventricular ejection fraction (%) initially increased (P < 0.001) and then stabilized.

CONCLUSIONS

This study found that aortic valve repair/sparing techniques provide viable and effective treatment options that should be considered for all eligible patients with aortic regurgitation and/or aortic root/ascending aortic aneurysms, given their potential to restore life expectancy and provide good haemodynamic outcomes with an acceptable hazard of reintervention.

Open in new tabDownload slide

INTRODUCTION

Aortic valve repair/sparing (AVR) techniques have evolved to treat both aortic valve (AV) insufficiency and aortic root aneurysms. Initially used for patients with intact aortic cusps, their promising outcomes have expanded their application to those with cusp abnormalities. Additionally, AVR is now increasingly applied to those with bicuspid aortic valves (BAVs) [1, 2].

In patients with AV and/or aortic root pathology, the type of procedure performed depends on several factors, including pathophysiologic mechanism, AV anatomy, life expectancy, anticoagulation preference and patient and doctor preferences. Although valve replacement is the standard procedure for most patients, selected individuals may benefit from AVR/sparing techniques when performed in experienced centres [3–5]. The complexity of surgical repair varies depending on the type of valvular lesion and the repair technique. Although AVR techniques can preserve the native valve with positive outcomes, inconsistent late results highlight concerns about surgical expertise and patient selection [6, 7]. The goal of the study was to understand the time-related clinical and echocardiographic outcomes of repaired valves after AVR surgery, focusing on the postoperative aortic regurgitation (AR) grade and the evolution of the left ventricular ejection fraction (LVEF).

PATIENTS AND METHODS

Ethical statement

Data access was obtained from the scientific committee of the Heart Valve Society (HVS). Each individual centre had local approval from its medical ethical committee to collect and share the data with the HVS AV database (Erasmus MC: MEC-2020–0890).

Study design

Data were obtained from the international HVS AV database (formerly AVIATOR/LEOPARD), which collects, since January 2013, prospective data on patients undergoing aortic surgery. Data for patients operated on before this date were uploaded retrospectively with prospective follow-up (ambispective design) [8]. Consequently, the recruitment period for this study covers 1995 (the earliest reported operation) to October 2023. Follow-up closed on data of extraction 19 October 2023. Collected outcome and definitions can be found at the HVS website (https://www.hvs.com).

Patient selection and surgical techniques

All patients who underwent AVR were identified. Exclusion criteria were patients <18 years old, any aortic dissection, active infective endocarditis or unspecified technique. Patients were categorized into 4 groups based on surgical technique and their inherent risk profiles: group 1 (isolated AV repair), group 2 [ascending aorta (AAo) replacement + valve repair], group 3 (partial root replacement ± valve repair) and group 4 [valve-sparing root replacement (VSRR) ± valve repair] as shown in Supplementary Material, Fig. S1. The goal of these techniques is to preserve or restore AV competency and root geometry. Isolated AV is performed for isolated AR due to the abnormality of the cusps. Annulus stabilization can be achieved with single annuloplasty (Cabrol stitches, suture annuloplasty or external/internal rings) or double external annuloplasty [one ring at the subvalvular level and another at the sinotubular juction (STJ)]. It can be combined with AAo replacement (group 2) in the presence of an ascending aortic aneurysm. VSRR includes partial root remodelling, suitable for aortic root aneurysms involving only 1 or 2 sinuses (group 3) and full root procedures (group 4), which include the David technique (full root replacement) and the Yacoub technique (full root remodelling), the latter of which may incorporate a circular subaortic annuloplasty.

Outcomes

The primary outcomes were long-term AR grade, LVEF and the need for reintervention following AVR. Secondary outcomes were early and long-term survival, postoperative bleeding events and the occurrence of endocarditis. Early outcomes were defined as events occurring within 30 days after the operation; major bleeding was defined according to Valve Academic Research Consortium-2 criteria [9].

Statistical analyses

Continuous data are presented as mean (standard deviation) for Gaussian distribution or median (interquartile range) for non-Gaussian distribution. Categorical data are presented as frequencies (percentages). Comparisons among continuous variables were made with the analysis of variance test or the Kruskal–Wallis test, as appropriate. Early survival and categorical variables were assessed using the χ² test or the Fisher exact test, as appropriate. Late survival was calculated and visualized with the Kaplan–Meier method, using the log-rank test to compare survival groups. Median follow-up time was assessed using the inverse of the Kaplan–Meier method. Patient survival was compared to that of the age- and sex-matched general population of the country of origin for every group separately [10]. Cumulative hazards for AV reintervention were calculated using the Aalen–Johansen estimator to account for the competing risks of reintervention and death, with Gray’s test used to compare groups. The multivariable Cox proportional hazards regression model was used to investigate independent predictors for AV reintervention. Included risk factors were chosen based on clinical knowledge, and all were entered in a multivariable model structure. A full model was presented, and no covariate selection was applied, because the goal was to explore potential predictors and not derive a parsimonious prediction model.

Correlation between predictors was assessed by Spearman’s correlation coefficient before multivariable analysis. In case of a strong correlation (e.g. Spearman ≥0.40 or ≤−0.40), the clinically most important variable, determined by the research team, was retained. The proportionality assumption was checked through the Schoenfeld residuals and relaxed using a stepwise time-varying coefficient in case of severe violation upon visual inspection. Included covariates are presented in Supplementary Material, Text S1. A full model was presented, and no covariate selection was applied, because the goal was to explore potential predictors and not derive a parsimonious prediction model. Missing baseline data were imputed using multiple imputation [11]. Details are presented in Supplementary Material, Text S2 and Tables S1 and S2. Estimates were then pooled according to Rubin’s rules [12]. Linear mixed-effect models and continuation ratio mixed models were used to analyse repeated echocardiograms for LVEF and AR [13, 14]. More details are provided in Supplementary Material, Text S3. P-values <0.05 were considered statistically significant. Statistical analyses were performed with R (R-Foundation for Statistical Computing, version 141–4.2.2. Institute for Statistics and Mathematics, Vienna, Austria; https://www.R-project.org/).

RESULTS

This study included 7126 adults (18.9% female) with a median age of 52 years (40–63). Patient characteristics and preoperative echocardiographic features are presented in Table 1.

Intraoperative features

Intraoperative characteristics are summarized in Supplementary Material, Table S3. Isolated AV repair (group 1) was performed in 2167 patients; AAo replacement + valve repair (group 2), in 1080 patients; partial root replacement ± valve repair (group 3) in 212 patient and full VSRR ± valve repair (group 4) in 3667 patients [1190 (33%) David procedures and 2466 (67%) Yacoub procedures]. Overall, annuloplasty was performed in 4258 (60%) patients, with 250 (12%) patients in group 1 undergoing double external annuloplasty. There were no significant differences in the STJ, annulus or tubular aorta diameters among patients who underwent double and single annuloplasties (P = 0.73, P = 0.70, P = 0.82, respectively).

Early outcomes

A total of 69 patients died: 17 in group 1 (0.8%), 17 in group 2 (1.6%), 0 in group 3 and 35 in group 4 (1.0%, P = 0.51). Reintervention was needed in 95 (1.3%) patients: 51 (2.3%) in group 1 [6 (0.2%) for double annuloplasty versus 45 (2%) for others]; 9 (0.7%) in group 2; 3 (1.4%) in group 3; and 32 (0.8%) in group 4 (P ≤ 0.001). The main reason for reintervention was AR in 44 (80%) patients, aortic stenosis in 2 (4%), other procedure-related causes in 7 (12%) and graft-related in 1 (2%). There was 1 case of in-hospital endocarditis that required a reintervention due to a postoperative course complicated by the onset of mediastinitis.

Late outcomes

During the late follow-up period, 438 patients died. Valve-related deaths occurred in 60 (19%) patients; 54 (17%) deaths were due to other cardiac causes and 204 (64%) were due to non-cardiac causes; the causes of the remaining 120 (27%) were unknown. The overall median follow-up time was 4.01 years: 4.47 years for group 1; 3.11 years for group 2; 3.9 years for group 3; and 3.97 years for group 4. The 10-year survival after isolated AV repair, ascending aortic replacement + valve repair, partial root replacement ± valve repair and full VSRR ± valve repair was 85.5% [95% confidence interval (CI) 82.98–88.12%], 86.3% (95% CI 82.36–90.62%), 82.0% (95% CI 71.60–93.75%) and 87.0% (95% CI 84.66–88.80%), respectively (P log-rank = 0.48). The Kaplan–Meier estimates are shown in Supplementary Material, Fig. S2. Overall survival at 10 years after all AVR techniques was 86.9% (95% CI 85.4–88.4%), resulting in a relative survival of 99.4% (97.7–100%) compared with the age-, country of origin- and sex-matched general population (survival in general population 87.4%) as shown in Supplementary Material, Fig. S3. Patient survival curves compared with those of the general population of each group are depicted in Fig. 1.

During follow-up, a total of 486 patients underwent 518 reinterventions: 174 (36%) were due to AR; 51, to (11%) AS; 24 (5%), to endocarditis; 10 (2%), to other procedure-related causes; 11 (2%) were aorta-related and 216 (44%) were of unknown causes. At 10 years, the cumulative incidence of reintervention was 19.5% (95% CI 16.70–22.40%) for isolated AV repair; 13.8% (95% CI 10.10–18.10%) for ascending aortic replacement + valve repair; 12.7% (95% CI 5.50–22.90%) for partial root replacement ± valve repair and 8.5% (95% CI 7.00–10.10%) for full VSRR ± valve repair (P < 0.001), as shown in Fig. 2. When analysing only patients in group 1 who had a double external annuloplasty [with an STJ ring], the cumulative incidence of reintervention was no longer significantly different between groups (P = 0.115), as shown in Supplementary Material, Fig. S4.

The multivariable Cox proportional hazard ratios model revealed that the use of a pericardial patch was associated with greater hazards of reintervention in group 1 [hazard ratio (HR) 2.19 (95% CI 1.53–3.12), P≤0.001], group 2 [HR 2.05 (95% CI 1.29–4.94), P = 0.007] and group 4 [HR 1.92 (95% CI 1.08–3.43), P = 0.026]. Preoperative severe AR grade was associated with greater hazards of reintervention in group 4 [HR 1.95 (95% CI 1.19–3.21), P = 0.009], along with a larger preoperative left ventricular end-diastolic diameter (LVEDD) within 5 years after the operation [HR 1.03 (95% CI 1.00–1.05), P = 0.049]. BAV morphology was not associated with greater hazards of reintervention in any groups. The results of multivariable Cox proportional hazards regression are shown in Table 2. At 10 years, the cumulative incidence of endocarditis was 0.7% (95% CI 0.40–1.20%) for group 1, 0.4% (95% CI 0.10–1.20%) for group 2, 0% group 3 and 0.5% (95% CI 0.20–1.10%) in group 4 (P = 0.33). Similarly, the cumulative incidence of major bleeding at 10 years was 1.2% (95% CI 0.60–2.10%) for group 1, 1.1% (95% CI 0.30–3.00%) for group 2, 3.6% (95% CI 0.70–11.00%) for group 3 and 0.7% (95% CI 0.50–11.00%) in group 4 (P = 0.55).

Bicuspid aortic valve

Survival at 10 years was 95.6% (95% CI 93.20–98.17%) in group 1, 95.3% (95% CI 91.55–99.13%) in group 2, 97.2% (95% CI 93.52–100%) in group 3 and 93.1% (95% CI 90.75–95.54%) in group 4, as shown in Supplementary Material, Fig. S5. The cumulative incidence of late reintervention at 10-years was 23.1% (95% CI 18.20–28.30%) in group 1, 23.7% (95% CI 14.60–34.10%) in group 2, 15.3% (95% CI 5.90–28.70%) in group 3 and 9.9% (95% CI 7.30–12.80%) in group 4 as shown in Supplementary Material, Fig. S6.

Longitudinal echocardiographic analysis

During follow-up, 6936 patients (97%) received an echocardiographic evaluation, with a total of 36 306 echocardiograms [mean (SD) echocardiogram, per patient, 5.13 (2.26)].

Evolution of aortic regurgitation grade

Marginal probabilities of the AR category over time are shown in Fig. 3. Overall, AR severity increased during follow-up, primarily in the first 2 years after surgery, adjusting for patient age, followed by a gradual rise over time. The progression of AR is observed mainly in the early postoperative phase in group 1, whereas in group 2, this progression occurs in the late postoperative phase (Supplementary Material, Fig. S7).

A more detailed view of the progression of AR grades 3 and 4 is provided in Fig. 4, mainly observed in group 1. For patients who underwent isolated AV repair with a double ring annuloplasty, the likelihood of experiencing a higher progression was no longer observed. Results of continuation ratio mixed-effects models for AR grade are presented in Supplementary Material, Table S4.

Evolution of left ventricular ejection fraction

After correcting for age and preoperative LVEF, an increase in the LVEF (%) was observed in the first 2 years after the operation (P < 0.001), with a steady non-significant decline afterwards (P = 0.92) (Fig. 5). Introducing the surgical technique in the model showed a significantly different starting point and a less steep increase in LVEF in groups 2 and 4 compared to group 1. Additionally, groups 2 and 4 maintained a stable LVEF over time during follow-up (Supplementary Material, Fig. S8). Coefficients details are shown in Supplementary Material, Tables S5 and S6. While categorizing patients by preoperative LVEF, we found that those with reduced (<30%) or impaired (30–50%) LVEF showed significant postoperative improvement, though the trend was less pronounced for reduced LVEF and minimal for preserved LVEF (>50%) (Supplementary Material, Fig. S9).

DISCUSSION

The present study evaluated long-term clinical outcomes and echocardiographic evolution following AVR surgery, providing an overall view of these approaches without directly comparing them, because the techniques are inherently different and dictated by the specific lesion present [15, 16].

Ten-year results aligned with those of previous reports [17–20]. Each group demonstrated survival rates comparable to those of their respective matched general populations. However, a closer examination of the curve reveals a slight decline during the earliest period, indicating early deaths. Nevertheless, after this initial dip, the curve aligns with the background deaths, suggesting that if the patient survives the operation, the subsequent deaths are comparable.

A higher likelihood of AR grade progression during the first 2 postoperative years was observed, primarily involving transitions from grade 0 to 1 or 2. Interestingly, the overall prevalence of moderate to severe and severe AR (grades 3 and 4) remained consistently low throughout all the years of follow-up, which aligns with previous findings [16, 21]. Preoperative LVEF is a key predictor for postoperative outcomes. It has been reported to have an association with complications such as persisting impaired LVEF postoperatively, increased operative deaths and diminished postoperative survival rates [22]. Surgical correction induces LV reverse remodelling as a compensatory mechanism of the chamber’s dilatation due to the AR [23]. Our study reported a consistent improvement in LVEF in the first 2 periods, corrected for preoperative LVEF and age, particularly in patients with reduced (<30%) or impaired (30–50%) preoperative LVEF [24].

BAV morphology was not a predictor of reintervention, which was consistent with other findings [16, 25]. However, the use of a pericardial patch emerged as a common risk factor for AV reintervention across all analysed groups. Although the exact mechanism remains unclear, previous studies have suggested that pericardial patches may be susceptible to calcification and shrinkage, which could contribute to structural deterioration over time and increased risk of AV-related reoperations [26–30].

Predictors of failure are therefore essential. When the anticipated durability of valve repair is lower than that of the valve replacement, this consideration should guide decision making, balancing replacement risks and the potential need for reintervention [31]. Valve-related complications such as thromboembolism, bleeding and infective endocarditis were low, which aligns with the results of other studies [16, 25]. We have reported in our study lower 10-year mortality and major adverse event rates compared to biological and mechanical valves in younger patients [32–34]. Comparable outcomes have been observed between VSRR and bio-Bentall, though not consistently in younger patients [35, 36].

Isolated aortic valve repair

Both early and late aortic reintervention rates were higher in these patients. However, in a subgroup undergoing isolated AV repair with double external ring annuloplasty, rates were similar to those of other groups. The choice of double annuloplasty was independent of annulus and STJ size, suggesting potential superiority over single annuloplasty, warranting further research. Progression to grade 3 and 4 AR paralleled the reintervention curve, with this group showing a higher rate of grade 3 progression, potentially explaining the increased need for intervention. This situation was not observed in the double annuloplasty group (aligning with the curve of reintervention). Additionally, group 1 had the highest rate of patch reconstruction, identified as a predictor of reintervention. In the third period, a decline in LVEF was noted in group 1, whereas it remained stable in the other groups. This trend may be due to initially larger LVEDD, limiting LV remodelling. These multiple conditions could contribute to the higher rate of reintervention in this group. Despite the small sample size, double ring annuloplasty seems to perform better within the isolated AV repair group, aligning with other studies [37].

Ascending aorta replacement + valve repair

Patients in group 2 were the oldest, with the highest rates of arterial hypertension, chronic obstructive pulmonary disease and poor mobility. However, despite the older age, no survival differences were observed, possibly due to the relatively short follow-up time. Patch use was also identified as a predictor of reintervention in this group, showing similar AR grade 3 progression to group 1. This technique is often applied to fragile patients or to those with non-excessive annular dilatation.

Partial and full root replacement ± valve repair

Both partial and full root replacement provided satisfactory outcomes in survival, AR stability and freedom from reintervention, including for patients with BAV. Partial root replacement was less preferred for those with connective tissue disease, despite similar referral reasons to full root replacement, where it was mainly used. Patients in group 4 mostly had severe preoperative AR, a known risk factor for reintervention, as noted by other authors [25]. However, attributing this risk specifically to the David or Yacoub procedure is challenging. Additionally, the use of a patch was a common predictor of a higher reintervention rate in this group.

Strengths and limitations

This retrospective analysis has limitations due to the differences in characteristics and anatomical lesions among the groups. Additionally, using the surgical technique as a starting point is challenging, because it is typically determined by pathology technique preferences. The observed outcomes may reflect the expertise of high-volume centres, which may not be easily replicable in lower-volume settings or by less experienced surgeons. Adequate case volume is crucial to overcome the learning curve, and performing a sufficient number of AVR/sparing procedures annually is recommended to achieve good outcomes [38]. Nevertheless, despite these limitations, these results provide valuable insights into best practices and outcomes achievable under optimal conditions. Reintervention should be viewed as a shared decision rather than an event, because not all patients follow the same decision-making process. Additionally, multiple testing may have resulted in significant findings purely by chance, which we acknowledge as a limitation. However, it is one of the largest cohorts and with mid- to long-term follow-up providing overall insights into these techniques.

CONCLUSION

This study provides a contemporary overview of outcomes following AVR techniques, highlighting satisfactory long-term results in AR grade stability and LVEF preservation. However, the use of pericardial patches consistently correlates with higher reoperation rates. Each technique demonstrated a survival rate comparable to that of the age- and sex-matched general population, suggesting that AVR may restore life expectancy, if patients survive the early postoperative phase. Given its potential to achieve this goal, along with good haemodynamic outcomes and an acceptable reintervention risk, this operation should be considered for all eligible patients with AR and/or aortic root or ascending aortic aneurysms.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

FUNDING

This research project has been made possible thanks to the HVS AV Database data team, funded by Edwards Lifesciences.

Conflict of interest: none declared.

DATA AVAILABILITY

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

Author contributions

Kevin Veen: Conceptualization; Methodology; Supervision; Validation; Visualization; Writing—review & editing. Giovanni Melina: Conceptualization; Supervision; Validation; Writing—review & editing. Emmanuel Lansac: Validation; Writing—review & editing. Hans-Joachim Schaefers: Validation; Writing—review & editing. Laurent de Kerchove: Validation; Writing—review & editing. Johanna Takkenberg: Conceptualization; Methodology; Supervision; Validation; Visualization; Writing—review & editing. Jolanda Kluin: Conceptualization; Methodology; Supervision; Validation; Visualization; Writing—review & editing. Mostafa Mokhles: Conceptualization; Methodology; Supervision; Validation; Visualization; Writing—review & editing

Reviewer information

The European Journal of Cardio-Thoracic Surgery thanks Luca Di Marco, Leila Louise Benhassen, Mateo Marin-Cuartas and the other anonymous reviewers for their contributions to the peer review process of this article.

REFERENCES

ABBREVIATIONS

ABBREVIATIONS
 
• AAo

  Ascending aorta

 
• AR

  Aortic regurgitation

 
• AV

  Aortic valve

 
• AVR

  Aortic valve repair/sparing

 
• BAV

  Bicuspid aortic valves

 
• CI

  Confidence interval

 
• HR

  Hazard ratio

 
• HVS

  Heart Valve Society

 
• LVEF

  Left ventricular ejection fraction

 
• STJ

  Sinotubular junction

 
• VSRR

  Valve-sparing root replacement