Transcatheter Aortic Valve Replacement in Nonagenarians: A Systematic Review and Meta-Analysis

Background. Nonagenarians represent only a small proportion of patients included in large transcatheter aortic valve replacement (TAVR) trials, but will become a relevant future population in need of treatment due to demographic change. Thus, this study sought to evaluate outcomes of TAVR for the treatment of severe aortic stenosis (AS) in nonagenarian patients. Methods. We screened Medline/Pubmed for studies that stated specific outcomes for nonagenarians undergoing TAVR. A weighted meta-analysis was conducted, calculating pooled estimate rates using a binary random-effects model for dichotomous variables, and comparing non-dichotomous outcomes with a continuous random-effects model. Results. Data from 23 studies including 16,094 nonagenarians were merged; 53.4% were women. Despite reasonable rates of comorbidities, Society of Thoracic Surgeons mortality risk score was 10.2 ± 5.4. Pooled estimate rate of procedural success was 94.1% (95% confidence interval [CI], 91.7-96.6), with major vascular complications occurring in 6.3% (95% CI, 2.7-9.8) and at least moderate postprocedural paravalvular leak in 7.5% (95% CI, 4.4-10.6). The rate of periprocedural stroke or transient ischemic attack was 2.6% (95% CI, 2.0-3.2). At 30 days, the pooled estimate of mortality was 6.1% (95% CI, 4.7-7.4) and a permanent pacemaker was implanted in 12.6% (95% CI, 7.6-17.6). After 1 year, the mortality rate was 20.5% (95% CI, 15.9-25.1). Conclusion. TAVR in nonagenarians is an effective and safe procedure, with encouraging outcomes given the general life expectancy of these patients. Currently, only selected nonagenarians are undergoing TAVR, but their number will grow as life expectancy continues to increase in the developed world. Specific research to identify ideal candidates and techniques in this cohort is needed.

J INVASIVE CARDIOL 2022;34(3):E226-E236.

Key words: nonagenarians, transcatheter aortic valve implantation, transcatheter aortic valve replacement

During the last decade, transcatheter aortic valve replacement (TAVR) has been established as a reliable treatment option, with favorable outcomes and safety for patients with symptomatic aortic stenosis (AS) at all levels of surgical risk.^1-4 While longer-term issues like valve durability and potential valve deterioration are currently of utmost importance in younger patients with greater life expectancy, the focus in older patients (especially those aged ≥90 years) remains on safety and efficacy to determine whether TAVR might be a non-futile treatment in this high surgical risk population with no other treatment options.^5-9 Of note, nonagenarians formed only a minority of the patients enrolled in relevant randomized controlled trials (RCTs) assessing TAVR. However, the number of nonagenarian patients evaluated for TAVR is rapidly growing due to demographic change and the increasing prevalence of severe calcific AS with age.^10-12 Hence, specific data are vital to inform physicians, patients, and their relatives about the efficacy of TAVR in nonagenarians who present with characteristics, comorbidities, and procedural risks substantially different from most TAVR study cohorts. In this systematic literature review and meta-analysis, we evaluated current evidence concerning the safety profile and procedural outcomes of TAVR in nonagenarian patients.

Search strategy. We performed a systematic search of the MEDLINE databases via PubMed from 2008 to November 2019 for all TAVR studies. Our search string included (“TAVR” or “TAVI” or “transcatheter aortic valve replacement” or “transcatheter aortic valve implantation”) AND (“age” or “elderly” or “nonagenarian”). We also hand-searched bibliographies of relevant selected studies, reviews, and meta-analyses to identify further eligible studies. Titles and abstracts were screened for eligibility and full-text articles were accessed accordingly. The search was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement.¹³ Two independent reviewers performed the search and literature screening (OMD and JC), with disputes resolved by consensus following discussion with a third author (AL).

Inclusion and exclusion criteria, quality assessment. We considered all TAVR studies. Studies were eligible if they explicitly reported TAVR outcomes for nonagenarians and were written in English. Editorial comments, conference abstracts, meta-analyses, systematic reviews, and animal studies were excluded. The quality of the included studies was assessed using a 20-item checklist based on the STARD statement concerning Standards for Reporting Diagnostic Accuracy studies and the STROBE statement for Strengthening the Reporting of Observational studies in Epidemiology.^14,15

Data extraction and analysis. Two authors (OMD and JC) independently extracted the data from included trials; data were then verified by a third author (MP). We extracted baseline patient characteristics, procedural safety endpoints according to Valve Academic Research Consortium (VARC)-2 criteria, 30-day outcomes regarding mortality, stroke, or transient ischemic attack (TIA), and 1-year outcomes regarding mortality.¹⁶ Baseline characteristics are reported as pooled weighted means and composite standard deviations for continuous variables or cumulative proportions for categorical variables. When data were available only as median and interquartile range, mean and standard deviation were calculated according to Wan et al.¹⁷ A weighted meta-analysis of TAVR outcomes in nonagenarians was performed. According to DerSimonian and Laird, cumulative event rates for dichotomous outcomes were obtained from a pooled analysis among selected studies and pooled estimate rates with 95% confidence intervals (CIs) were calculated using a binary random-effects model.¹⁸ For non-dichotomous outcomes, comparison was performed by means of a continuous random-effects model.¹⁸ Heterogeneity across studies was assessed using Cochrane Q statistics to compute I² (a heterogeneity P-value ≤.10 was considered significant). I² values of <25%, 25%-50%, or >50% indicated low, moderate, or high heterogeneity, respectively. To evaluate the potential influence of growing experience and newer iterations of TAVR devices, subgroup analysis was undertaken in all studies enrolling patients since 2011 (the year of regulatory approval of TAVR as a treatment option for inoperable patients in the United States, resulting in rapid expansion and further development of the procedure). Meta-Analyst Beta 3.13 software (Tufts Evidence-Based Practice Center) was used for statistical computation.

Search results and study quality. The final database search was performed on November 3, 2019. A total of 23 studies were identified for inclusion in the present meta-analysis following literature screening according to the PRISMA statement (Figure 1).^19-41 Characteristics of the 23 studies included are presented in Table 1. Application of the 20-item checklist revealed sufficient allocation of relevant data and information concerning research design for all included studies (Supplemental Table S1). However, we identified occasional instances of potential bias and missing data concerning center details, implanter experience, and clinical management.

Baseline characteristics. Twenty-three studies enrolling 78,858 patients met the inclusion criteria. Of these, a total of 16,094 patients (20.4%) were nonagenarians and formed the study population. Baseline patient characteristics are depicted in Table 2. The mean age of the nonagenarians was 91.2 years and 53.4% were women. The mean Society of Thoracic Surgeons mortality risk score (STS-M) was 10.2 ± 5.4 (Figure 2). The mean left ventricular ejection fraction was 56.6 ± 12.7%, with prior percutaneous coronary intervention in 26.2%, myocardial infarction in 15.4%, coronary artery bypass grafting in 16.4%, and peripheral vascular disease in 25.2%.

Procedural characteristics and safety. Transfemoral access was the most commonly utilized approach (83.0%), followed by transapical (16.7%) and subclavian access (0.02%). Among patients for whom data concerning the type of valve used was available (2071/16,094 patients), the majority (72.7%) received a balloon-expandable device.

Procedural success pooled estimate rates were 94.1% (95% CI, 91.7-96.6; I² = 95.2%; P<.001 for heterogeneity) using weighted meta-analysis. Incidence of major vascular complications was 6.3% (95% CI, 2.7-9.8; I² = 76.6%; P<.001 for heterogeneity), while that of at least moderate postprocedural paravalvular leak (PVL) was 7.5% (95% CI, 4.4-10.6; I² = 60.8%; P<.01 for heterogeneity) (see graphical abstract in Figure 3).

The pooled meta-analysis estimate of periprocedural mortality rate was 5.4% (95% CI, 4.4-6.4; I² = 58.6%; P<.001 for heterogeneity). Further procedural outcomes are presented in Table 3.

Outcomes at 30 days and 1 year. The pooled estimate of 30-day mortality rate was 6.1% (95% CI, 4.7-7.4; I² = 71.4%; P<.001 for heterogeneity) (Figure 4) and incidence of stroke or transient ischemic attack was 2.8% (95% CI, 1.9-3.6; I² = 58.7%; P<.001 for heterogeneity). The estimate of permanent pacemaker (PPM) implantation rate at 30 days was 12.6% (95% CI, 7.6-17.6; I² = 97.8%; P<.001 for heterogeneity).

After 1 year, the pooled estimate of mortality rate was 20.5% (95% CI, 15.9-25.1; I² = 93.2%; P<.001 for heterogeneity) (Figure 5). Table 4 summarizes the outcomes at different follow-up points.

Subgroup analysis of studies enrolling patients since 2011. Six studies that enrolled patients from 2011 onward, including a total of 13,297 nonagenarians, were identified.^{24,28,31,33,37,38} The procedural success pooled estimate rate was 93.5% (95% CI, 91.4-95.6; I² = 82.3%; P<.001 for heterogeneity). A PPM was required in 13.6% (95% CI, 1.4-25.7; I² = 99.5%; P<.001 for heterogeneity). The mortality rate in these patients was 5.5% at 30 days (95% CI, 3.4-7.6; I² = 84.7%; P<.001 for heterogeneity) and 16.1% at 1 year (95% CI, 8.6-23.6; I² = 97.7%; P<.001 for heterogeneity).

In this study, we have shown the following: (1) TAVR in nonagenarians is effective and safe, with high procedural success rates and favorable outcomes up to 1 year; and (2) nonagenarians undergoing TAVR have a surgical risk profile comparable with patients enrolled in the inoperable and high-risk TAVR trials. However, since patients in these trials were noticeably younger, nonagenarians seem to be a selected cohort whose operative risk is driven by reasonable comorbidities and age.

Patient characteristics of nonagenarians undergoing TAVR. The overall pooled baseline characteristics of patients in our study were comparable to those enrolled in the high-risk TAVR trials. However, the prevalence of previous percutaneous coronary intervention, myocardial infarction, coronary artery bypass grafting, or peripheral vascular disease was in keeping with intermediate-risk TAVR study patients.^2,42-44 Conversely, their surgical risk profile was substantially elevated and comparable to inoperable-risk TAVR trial cohorts (which enrolled patients almost a decade younger, suggesting that age is the main factor driving operative risk in nonagenarians) (Figure 2).^45,46 We also observed a trend toward more female nonagenarians undergoing TAVR when compared with high-risk TAVR cohorts (53.4% in nonagenarians vs 42.2% in the PARTNER high-risk trial vs 46.4% in the CoreValve high-risk trial), likely reflecting longer life expectancy in women.^42,43,47 The baseline risk-profile data of women undergoing TAVR are heterogeneous. Previous studies have demonstrated that while comorbidities such as coronary artery disease or diabetes mellitus are less prevalent, overall surgical risk profile is significantly higher in females.^48-50 However, it is important to note that this higher surgical risk did not translate into worse survival in women in all studies assessing these baseline differences. In contrast to findings for surgical aortic valve replacement, women demonstrated significantly lower mortality following TAVR than men at 1-year follow-up.^48-50 Consistent with the so-called “obesity paradox” (a predictor of improved outcomes following TAVR in overweight or even obese patients), the body mass index (BMI) of the pooled nonagenarian cohort was rather low (24.8 ± 4.1 kg/m², corresponding to normal weight).^51-53 BMI declines naturally with age (especially above 80 years and beyond) and may dilute the established effect of BMI on outcomes in elderly patients, necessitating a broader approach including assessment of physical capability, mobility, and frailty.^54,55 In addition to these findings, it is important to consider potential survival (healthy patients, still alive at ≥90 years) and selection biases (healthy patients, scheduled for the procedure) when considering the risk profile of nonagenarians currently undergoing TAVR.

Procedural performance in nonagenarians. Despite the high surgical risk, with an expected mortality of over 10% according to the STS-M score, the observed perioperative mortality of 6.1% fell noticeably below this prediction. The modest difference between periprocedural (5.4%) and 30-day mortality (6.1%) infers that 30-day mortality is principally driven by periprocedural events, thereby emphasizing the importance of careful patient selection in nonagenarians and the potential scope for technical improvement with greater operator experience. It is important to note that only about a quarter of the patients were older than 75 years in the original cohort used to develop the STS-M scoring algorithm, inferring that the proportion of nonagenarians was even smaller.⁵⁶ Hence, the efficacy (and even validity) of these risk predictors in nonagenarians undergoing TAVR remains unclear. According to recent Social Security life tables, the probability of dying within 1 year at the age of 90 or 93 years (the standard deviation range for pooled age in our study cohort) is 16.4% or 22.0% for men and 13.0% or 17.8% for women, respectively.⁴⁷ Thus, with a pooled estimate of 20.5% 1-year mortality, life expectancy in nonagenarians (albeit selected) with severe AS undergoing TAVR is almost comparable to the general population. Such outcomes are encouraging given the high mortality rate  (approximately 40% within 1 year) if severe AS is left untreated.⁵⁷ The observed rate of major vascular complications (6.3%) is promising, as elevated risks of this adverse outcome have been previously observed in elderly patients.⁵⁸ Furthermore, women, who were well represented in our study and who tend to have more vascular complications, still demonstrated significantly better survival 1 year after TAVR.^48-50

The low rate of periprocedural stroke or transient ischemic attack (2.6%) is encouraging, especially since an increase in aortic calcification (more common in older patients) is associated with higher risk of embolism during TAVR.^59,60 Functional outcomes are of utmost interest in nonagenarians undergoing TAVR, which is why low stroke rates (which may potentially be reduced using cerebral protection devices) are extremely valuable.⁶¹ However, low stroke rates may also be the result of selection bias—for example, operators may have selected patients with favorable anatomy and less calcification from within this high-risk population. Importantly, only very few studies have separately reported rates of major stroke, hindering further elucidation of different strokes and their respective impact.

The rate of PPM implantation at 30 days (12.6%) was rather high in the pooled nonagenarian cohort. As the deleterious effect of PPM implantation is noticeable already at 1 year (when almost 80% of our cohort were still alive), the negative impact on ventricular synchronization becomes potentially relevant in a substantial number of patients.^62,63 While extreme- and high-risk trials evaluating self-expanding devices found substantial new PPM rates at 30 days (21.6% and 19.8%, respectively), rates in the corresponding trials assessing balloon-expandable devices were markedly lower (3.4% in inoperable patients and 3.8% in high-risk patients).^42,43,45,46 Unfortunately, information concerning the implanted TAVR prosthesis was only available in a small number of patients, hindering further assessment of the PPM findings. However, the PPM implantation rate in nonagenarians does not seem to markedly exceed the risk found in other contemporary (mostly younger) cohorts.

At least moderate postprocedural PVL is a second important factor impacting the long-term prognosis after TAVR and was present in 7.5% of the pooled nonagenarian cohort.⁶⁴ These data are similar to the early experience in extreme- and high-risk patients (30-day mortality rates, 9.7%/9.1% in the CoreValve extreme-risk/high-risk trials; 11.8%/12.2% in the PARTNER inoperable/high-risk trials). However, it is important to note that successive iterations of TAVR devices have enabled a notable reduction of postprocedural PVL.^42,43,45,46 Since we could not extract detailed data concerning procedural factors that influence the trade-off between PPM implantation and postprocedural PVL (valve-sizing, implantation depth, or postdilation), we could not further elucidate whether implanters might adapt their procedural strategies to influence these outcomes in nonagenarians.

Procedural complications and mortality have fallen dramatically over the period of TAVR evolution, and we postulate that the use of newer device iterations and improved implantation techniques may also have improved outcomes in nonagenarians.⁶⁵ However, in our subgroup analysis of studies enrolling patients after 2011, we were unable to identify a uniform trend regarding procedural performance endpoints. While survival rates at 30 days and 1 year improved in more recent cohorts, rates of procedural success and PPM implantation were slightly worse. Of note, the more recent subgroup (13,297 patients) formed a major proportion of our total pooled cohort, complicating comparison with older studies. On the other hand, the fact that a large proportion of patients were treated using contemporary techniques and devices potentially explains the favorable overall results in our analysis. Further subgroup analysis was unfortunately limited by lack of patient-level data, particularly concerning procedural factors such as implantation techniques and implanted devices.

Future of TAVR in nonagenarians. As life expectancy increases (probably exceeding 90 years in women by the end of this decade), health care solutions for future elderly populations need to be established.⁶⁶ The prevalence of AS is directly linked to age, suggesting significant growth in the number of nonagenarian patients referred for TAVR evaluation.^12,67 Thus, the identification of ideal nonagenarian TAVR candidates and optimal implantation techniques will be necessary. The findings of our study show that TAVR seems to be undertaken in relatively healthy nonagenarians who develop severe AS threatening their further life expectancy. When screening nonagenarians with AS, it is important to bear in mind that elevated surgical risk scores might not necessarily be associated with increased mortality following TAVR, as indicated by the low perioperative mortality compared with the predicted rate in our study. More general aspects, such as frailty, quality of life, technical feasibility, and anatomical feasibility may well be of greater importance in guiding the choice of treatment.

Study limitations. As a study-level meta-analysis, our research was limited by the potentially pronounced impact of single large studies. Furthermore, valid assessment was complicated by the fact that some variables were available in only a subset of the included studies. The merger of some variables was also hindered by the heterogeneous terminology used by different authors (eg, major and/or minor bleeding/vascular complications). Importantly, selection and survival biases may be present in current nonagenarian cohorts undergoing TAVR and it remains unclear whether the favorable outcomes found in our study are generally applicable to all nonagenarian AS patients. However, inclusion of a large proportion of patients from real-world registries in this meta-analysis reflects everyday clinical practice and outcomes.  Besides the endpoints assessed in our study, quality of life is an important outcome measure in elderly patients that we were unable to evaluate. Finally, the overlapping enrollment periods of the included studies spanned a substantial period of TAVR evolution; therefore, proving temporal trends in TAVR performance was not possible.

TAVR in nonagenarians is a viable and safe treatment option. The mortality rate at 1-year suggests that TAVR eliminates the negative impact of severe AS and can restore life expectancy to that of the age-matched population. While only selected nonagenarians currently undergo TAVR, it is important to further identify ideal candidates and treatment strategies given that the number of nonagenarian patients with AS will significantly increase during the next decade.

*Joint first authors.

From the ¹Department of Cardiology, St Thomas’ Hospital, London, United Kingdom; ²Charité University Medical Care, Department of Cardiology, Campus Benjamin Franklin, Berlin, Germany; ³Cardiac Intensive Care Unit, IRCCS San Raffaele Scientific Institute, Milan, Italy; ⁴Interventional Cardiology Unit, New Tokyo Hospital, Chiba, Japan; ⁵Invasive Cardiology Unit, Humanitas Clinical and Research Center IRCCS, Rozzano, Milan, Italy; ⁶Department of Cardiac Surgery, Montefiore Medical Center, New York, New York; ⁷Department of Cardiology, Montefiore Medical Center, New York, New York; and the ⁸Division of Cardiology, Department of Medicine, University of Cape Town, Cape Town, South Africa.

Disclosure: The authors have completed and returned the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Latib has served on the advisory boards or as a consultant for Medtronic, Edwards Lifesciences, and Abbott Vascular. Dr Prendergast has received speaker fees from Edwards Lifesciences and Abbott; consultant for Anteris. The remaining authors report no conflicts of interest regarding the content herein.

Manuscript accepted May 20, 2021.

Address for correspondence: Dr Ozan M. Demir, Department of Cardiology, St Thomas’ Hospital, London SE1 7EH, United Kingdom. Email: ozan.demir@kcl.ac.uk

• Alternative Access Versus Transfemoral Transcatheter Aortic Valve Replacement in Nonagenarians 
• Quality of Life Outcomes After Transcatheter Aortic Valve Replacement in Nonagenarians
• Clinically Significant Incidental Findings on CT Imaging During TAVI Work-up: A Systematic Review and Meta-Analysis