Long-term age-stratified outcomes after surgical and transcatheter aortic valve replacement: a Dutch cohort study

The NHR is a nationwide quality registry, including data from 15 Dutch heart centres performing TAVI and SAVR, consisting of demographics, clinical characteristics, peri-procedural parameters, complications, mortality status and follow-up data [10]. All patients aged > 64 years receiving isolated SAVR or TAVI in the Netherlands in the period 2013–2021 were eligible for this study. An isolated SAVR or TAVI procedure was defined as a valve replacement that was not combined with any other additional procedure in the same setting, such as coronary artery bypass grafting, left atrial appendage closure or percutaneous coronary intervention in the case of TAVI. All access types for TAVI procedures were included in this analysis. Patients with incomplete follow-up data (e.g. absence of follow-up status or time-to-follow-up) or who had previously undergone SAVR or TAVI were excluded.

The following baseline data were collected: age, sex, body mass index (BMI) and the EuroSCORE II components (New York Heart Association (NYHA) class III/IV, Canadian Cardiovascular Society class, poor mobility, chronic obstructive pulmonary disease (COPD), diabetes mellitus (defined as insulin use), previous cardiac surgery, active endocarditis, critical pre-operative state, unstable angina, recent myocardial infarction (MI), thoracic aortic surgery and urgency of the operation). Additional parameters included history of atrial fibrillation (AF) and previous stroke and left ventricular ejection fraction (LVEF), based on echocardiographic images, either exact or eyeballing, and pulmonary arterial systolic pressure (PASP), invasively measured or estimated by echocardiography.

The primary endpoint was all-cause mortality (maximum 5‑year follow-up). The secondary endpoint was aortic valve re-intervention (maximum 5‑year follow-up).

Categorical variables were compared with the chi-squared or Fisher’s exact test (2-sided) and are reported as n (%). Continuous variables were compared with the t-test or Mann–Whitney U test, and ANOVA or Kruskal–Wallis test for comparison of ≥ 2 groups, and are reported as median (interquartile range; IQR). Because of the known age difference between TAVI and SAVR patients and the different age limits included in the guidelines, the following age groups were defined: 65–75 years, 75–80 years and > 80 years [11].

Due to differences in follow-up time, cumulative crude survival was expressed as n (%). Survival analyses were evaluated by a Kaplan-Meier analysis with log-rank tests and Cox proportional hazard modelling. The assumption of proportional hazards was tested by Schoenfeld residuals and the cox.zph package test in R, and stratification of the non-proportional hazard was performed if needed. To identify an interaction between older age groups and cohorts for both outcomes, an interaction-term was analysed by Cox regression survival analysis.

We included 7879 patients who underwent an isolated SAVR procedure and 14,461 patients who underwent an isolated TAVI procedure in the Netherlands in the period 2013–2021. At baseline, SAVR patients were younger than TAVI patients (median age: 73.0 years; IQR: 69.0–77.0 vs 81.0 years; IQR: 77.0–84.0; p < 0.001), were less frequently female (3447 (43.7%) vs 7232 (50.0%); p < 0.001) and had a higher median BMI (27.3 kg/m²; IQR: 24.7–30.4 vs 26.5 kg/m²; IQR: 23.9–29.8; p < 0.001), a lower NYHA class (class III/IV: 1894 (31.7%) vs 7712 (57.6%); p < 0.001) and lower median EuroSCORE II (1.44; IQR: 1.08–2.13 vs 3.25; IQR: 2.01–5.45; p < 0.001) (Tab. 1). After dividing patients into age groups (65–75, 75–80 and > 80 years old), there was no statistically significant difference in the distribution of sex by age, but TAVI patients consistently had a higher NYHA class, a higher rate of poor mobility and a higher EuroSCORE II compared with SAVR patients (see Table S1 in Electronic Supplementary Material).

TAVI patients showed higher prevalences of COPD, diabetes, AF, dialysis, previous stroke, unstable angina and recent MI than SAVR patients. In addition, they had more frequently undergone previous cardiac surgery, had higher creatinine levels, lower LVEF and higher PASP. After stratifying patients by age, these higher rates in TAVI remained statistically significant, except for diabetes and dialysis in the older age group. SAVR patients, on the other hand, had higher rates of endocarditis, a critical pre-operative condition and higher urgency, which were consistent through all age groups. Tables S2 and S3 in the Electronic Supplementary Material show baseline characteristics of both TAVI and SAVR cohorts in early (2013–2016) and recent (2017–2021) years. Several changes in baseline characteristics were observed over time, which were largely similar between TAVI and SAVR cohorts. Regarding trends in older age groups, the proportions of female sex, age and EuroSCORE II increased in older age groups in both cohorts. The prevalences of COPD, AF, dialysis, recent MI and previous cardiac surgery declined in older age groups in both cohorts and thus were highest in younger age groups (see Table S4 in Electronic Supplementary Material).

After a median follow-up time of 1617 days, 1012/7879 (12.8%) SAVR patients died. In contrast, after 1121 days of follow up, 5140/14,461 (35.5%) TAVI patients died (Tab. 2). Median survival was shorter after TAVI (685 days; IQR: 250–1160 vs 874 days; IQR: 317–1356; p < 0.001 (Fig. 2)). This trend was also seen in both early and recent years (see Tables S2 and S3 in Electronic Supplementary Material) and when including only transfemoral TAVI patients versus SAVR patients (see Table S5 in Electronic Supplementary Material). However, there was no interaction between time, cohort and outcomes (see Table S6 in Electronic Supplementary Material). Aortic valve re-intervention rates were low (135 (1.9%) in SAVR patients vs 114 (0.9%) in TAVI patients; p < 0.001), and the time to aortic valve re-intervention was shorter in TAVI compared with SAVR (52 vs 338 days; p < 0.001).

Comparing TAVI and SAVR patients who died during follow-up, showed that TAVI patients were older and more frequently female and had lower BMI, more often NYHA class III/IV at the time of the procedure, poor mobility and higher EuroSCORE II (see Table S7 in Electronic Supplementary Material). TAVI patients who had died had a higher prevalence of chronic lung disease, AF, stroke and previous cardiac surgery. LVEF was lower in TAVI patients, whereas PASP was higher. SAVR patients, in contrast, more frequently were in critical pre-operative condition and had previous thoracic aortic surgery and higher urgency.

Mortality rates increased with increasing age for both procedures, but this was more pronounced after SAVR than TAVI (9.7% for 65–75 years, 15.7% for 75–80 years and 24.5% for > 80 years in SAVR; p < 0.001, compared with 32.8%, 32.1% and 38.3%, respectively, in TAVI; p < 0.001) (Tab. 3, and see Tables S8 and S9 in Electronic Supplementary Material). Interaction analysis between cohort and age group revealed that the hazard associated with TAVI decreased relative to SAVR in higher age groups (see Table S10 in Electronic Supplementary Material). Aortic valve re-intervention rates were low and declined with increasing age (2.4%, 1.4% and 0.5% in SAVR; p < 0.001, compared with 1.5%, 0.9% and 0.7% in TAVI; p < 0.001).

The differences in age, sex and prevalence of comorbidities, such as COPD, diabetes, AF and previous stroke, we observed are in line with results from other nationwide registries in the literature [7‐9]. The significantly higher age of TAVI patients was expected and can be explained by the use of local, national and international guidelines. In the Netherlands, TAVI procedures are only reimbursed for patients > 80 years or those at high surgical risk [5]. Our study showed that several characteristics of both patients undergoing SAVR or TAVI changed over time. In addition, there was no change in baseline characteristics before and after the Dutch indication document for reimbursement was effectuated in 2021, as has been shown in a recent publication based on the same NHR registries [12]. The same study indicated that age > 80 years was the most common reason for choosing TAVI in the Netherlands. However, even within the stratified age groups, there were still several remarkable differences between TAVI and SAVR patients. For example, the prevalences of all comorbidities within each age category were greater in the TAVI group. This difference between TAVI and SAVR was most prominent in the younger age group, indicating that younger patients with more comorbidities or a high-risk profile underwent TAVI. This is perhaps not surprising given that patients with more comorbidities tend to be declined for surgical treatment by the Heart Team. Currently, there is an emerging trend towards treating younger patients with fewer comorbidities with TAVI in many European countries and in the USA [13]. Several low-risk trials involving younger and relatively healthier patients are currently in the follow-up phase and should give more insight into the outcomes in this patient group in the coming years [1, 3, 4, 14‐16].

In patients undergoing SAVR, long-term mortality rates clearly increased with age, which is in line with previously published data [17]. This trend was less obvious in patients undergoing TAVI. This may be explained again by the common practice that patients with a high rate of comorbidities are more likely to undergo TAVI and these patients have lower life expectancy, even when the procedure is performed at younger age. The cause of death was unknown in the current registry. However, it has been shown that long-term mortality after both procedures is generally mostly non-valve-related [18‐20].

The aortic valve re-intervention rate was low across all age groups and slightly lower compared with other (randomised) cohorts [1, 21]. The aortic valve re-intervention rate was lowest in older patients, possibly due to competing risks; either older patients die before the valve is able to degenerate or they are too frail to undergo a second AVR procedure. Another possibility is the earlier and rapid degeneration of bioprosthetic valves in younger patients, although the evidence is conflicting [22, 23]. In this registry, we did not have information on the reasons for withholding aortic valve re-intervention, and other cohorts of patients undergoing TAVI or SAVR unfortunately also did not capture this information [24, 25]. Information on valvular degeneration and symptomatology is essential in the follow-up of the predominantly older AVR population, as symptomatology may frequently be determined by comorbidities. When post-AVR patients become symptomatic, the decision for aortic valve re-intervention requires a balanced consideration of expected decrease of symptoms and expected risk of procedural complications.

This study has some limitations. First, there was a high risk of residual confounding due to the selective collection of variables. Second, due to the eligibility criteria for TAVI in the Netherlands, there was a selection bias, as younger patients without risk factors are declined for TAVI. The aforementioned low-risk trials should provide new information regarding the outcomes of younger patients. Lastly, by design, we excluded patients with additional procedures and SAVR patients with concomitant coronary artery disease, inducing a potential additional bias.

There are substantial differences in baseline characteristics between patients undergoing SAVR and TAVI. Based on our study and other previously reported data [7, 25], younger patients with fewer comorbidities seem to be underrepresented in Dutch and other TAVI cohorts. It could be questioned whether comparison of outcomes after both procedures is even possible. Real-world and randomised trials of patients aged < 75 years and with fewer comorbidities are needed and should include long-term follow-up. In addition, mortality may not be an appropriate long-term outcome comparator since this endpoint can be driven by non-valvular-related factors and other outcomes, such as decrease of symptoms, hospitalisation for heart failure, valve degeneration and re-intervention, may be more meaningful.