A Novel Model to Predict 1-Year Mortality in Elective Transfemoral Aortic Valve Replacement: The TAVR-Risk Score

Abstract

Objectives. We aimed to develop and validate an effective prediction model for 1-year mortality risk in elective transfemoral transcatheter aortic valve replacement (TAVR), ie, the TAVR-Risk (TARI) model. Background. TAVR is the primary treatment for patients with symptomatic severe aortic valve stenosis; however, risk assessment tools for longer-term outcomes after TAVR remain scarce. Methods. This retrospective cohort study used logistic regression to test univariate and multivariate associations. The German Aortic Valve Registry (GARY) was the derivation (n = 20,704) and the Swedish SWEDEHEART TAVR Registry (SWENTRY) was the validation cohort (n = 3982). The main outcome was the area under the curve (AUC) in the prediction of 1-year mortality. The final model included 12 parameters that were associated with 1-year mortality in a multivariate analysis. Results. The TARI model (AUC, 0.66; 95% confidence interval [CI] 0.65-0.67) performed better as compared with the Society of Thoracic Surgeons (STS) score (AUC, 0.63; 95% CI, 0.62-0.64; P<.001) and logistic EuroSCORE I (AUC, 0.60; 95% CI, 0.59-0.61; P<.001) in the GARY derivation cohort, and discriminated the risk for 1-year mortality better than logistic EuroSCORE I in the SWENTRY validation cohort (AUC, 0.62; 95% CI, 0.60-0.64 vs AUC, 0.59; 95% CI, 0.57-0.61; P=.04). Conclusions. This novel TARI score provides a relatively easy-to-use risk model and offers a superior prediction for 1-year mortality in European TAVR patients.

J INVASIVE CARDIOL 2022;34(11):E776-E783. Epub 2022 October 13.

Key words: TAVR, aortic valve, intervention, long-term, risk assessment

In Western countries, aortic stenosis (AS) is the most common valvular heart disease requiring treatment. Transcatheter aortic valve replacement (TAVR) has emerged as the primary treatment option for patients with symptomatic severe AS who are considered high or intermediate risk for surgical aortic valve replacement (SAVR),^1,2 and, at least, noninferior to surgery in low-risk patients.³

A heart team evaluation assessing operative vs interventional procedural risk, frequently using surgical risk models, is recommended by European and American cardiovascular societies^1,2 to determine the optimal treatment option. The most frequently used tools to assess surgical risk are the logistic EuroSCORE I (LogES I),^4,5 the updated EuroSCORE II (ES II)⁶, and the ­Society of Thoracic Surgeons Predicted Risk of Mortality model (STS-PROM).⁷ All of these surgical scores significantly overestimate 30-day mortality risk in elective transfemoral TAVR.^8-10 TAVR-specific risk models have been developed for the prediction of 30-day mortality (eg, GAVS-II, FRANCE-2, OBSERVANT),^11-13 but have underperformed in external validation.¹⁴ Additionally, attempts to predict 1-year mortality of TAVR have only been performed in small cohorts and lack external validation.^15-17 However, 1-year mortality may be important to evaluate as it reflects survival beyond the procedure, which may have relevant societal and health economic impacts. Thus, there is a need for a validated risk model to predict longer-term risk after TAVR.¹⁸

We aimed to use the German Aortic Valve Registry (GARY)^19,20 to develop a new risk prediction model for 1-year mortality risk in elective transfemoral TAVR. The TAVR Risk (TARI) model was then externally validated using the Swedish SWEDEHEART TAVR Registry (SWENTRY).

Methods

Principles of study conduct. The study was conducted in conformity to the ethical guidelines of the 1975 Declaration of ­Helsinki. The investigators had full access to the data and control of the analysis. Initial approval for GARY was obtained from the Freiburg International Ethical Committee and patients gave written informed consent for inclusion. The regional ethical review board in Stockholm approved the use of SWENTRY registry data for the study.

Data sources. Patients included in GARY were used as the derivation cohort and the validation was performed on patients from SWENTRY. Both are all-comer registries collecting data on all TAVR procedures performed in Germany and Sweden, respectively. GARY is a nationwide registry of patients with aortic valve treatment¹⁹ and its design has been described previously.^3,21,22 Data from 92 hospitals performing aortic valve interventions and/or surgical aortic valve replacements are collected using standardized case report forms to record demographic, clinical, procedural, and follow-up data. The design of SWENTRY has also been described previously.²³ Data from 8 Swedish TAVR centers are collected into an electronic database platform, after patient consent and heart team evaluation. Record linkage through the National Civil Registry ensured full data capture for mortality.

Patient selection from derivation and validation cohorts. For model development, patients treated with elective transfemoral TAVR procedures from 2011 to 2015 were included from the GARY registry into the derivation cohort with the primary endpoint of 1-year mortality. After 2015, the 1-year mortality data were not available. Patients included in GARY and SWENTRY undergoing elective transfemoral TAVR were selected as shown in Supplemental Figure S1; patients undergoing SAVR or nontransfemoral or nonelective TAVR, as well as those with follow-up <1 year or undergoing concomitant percutaneous coronary intervention (PCI) were excluded. Additional exclusion criteria were patients with double entries or incomplete information regarding demographic data, New York Heart Association (NYHA) class, history of myocardial infarction, lung disease, hypertension, pulmonary hypertension, atrial fibrillation, prior PCI, prior cardiac surgery, diabetes, serum creatinine level, mitral regurgitation (MR), left ventricular ejection fraction (LVEF), prior stroke, or vascular disease, and incomplete information about the 1-year-vital-status. Finally, a total of 20,704 patients were included in the model. Figure 1 shows a summary of patient selection. As outlined in the CONSORT diagram, outcome assessment after 1 year was available for all patients who entered this analysis.

For model validation, all SWENTRY patients undergoing elective transfemoral TAVR for symptomatic severe AS between the January 2008 and April 2019 with available 1-year survival data were included (6022 cases).

Model parameter definitions. All patients underwent baseline transthoracic echocardiography. LVEF was graded as normal (LVEF >50%); mildly or moderately reduced (LVEF 30%-49%) or severely reduced (LVEF <30%). MR severity was graded as none/trivial, mild, moderate, or severe according to the European Society of Cardiology/American College of Cardiology/American Heart Association²⁴ recommendations. Pulmonary artery systolic pressure was calculated by Doppler echocardiography or measured invasively. Baseline electrocardiography (ECG) was acquired and interpreted by the treating physicians. Atrial fibrillation (AF) was coded if documented on ECG or previously included in the patient records (whether paroxysmal, persistent, or permanent). NYHA class was assessed to evaluate the severity of symptoms and was summarized into 2 categories (NYHA I/II or NYHA III/IV). The presence of diabetes mellitus was defined as known or new diabetes mellitus resulting in conservative treatment, medication, or insulin treatment. Chronic obstructive lung disease was defined by any documented use of guideline-recommended medical treatment, such as bronchodilators, inhaled corticosteroids, and others. Occlusive vascular disease was defined as at least 1 of the following entities: peripheral vascular disease with documented stenosis >50% and/or symptoms such as intermittent claudication; central vascular disease with a documented stenosis >50% in the brain supplying arteries; or aortic aneurysm as defined by current guidelines along the ascending or descending aorta. Prior stroke was defined by documented stroke in the past medical history with permanent neurological deficit. Creatinine levels entered the database after quantification in the institutions’ laboratory chemistry according to local standards. Patients on chronic dialysis were coded with a creatinine level of 350 mmol/L to avoid artificially low creatinine levels.

Statistical model development and validation. All statistical computations were performed using either SPSS, version 26 (SPSS, Inc), IBM 2019 (IBM Corp), or Stata statistical software, release 16 (StataCorp, LLC) and MedCalc statistical software, version 19.3.1 (MedCalc Software, Ltd). Categorical variables are presented as frequencies (percentages) and compared using Chi-squared tests. Continuous data are presented as mean ± standard deviation and compared using 2-tailed Student’s t tests or Mann-Whitney U tests, accordingly. Candidate variables were identified based on a review of literature and the clinical judgment of the authors. Non-normally distributed variables were log-transformed to normalize data. Logistic regression was used to test univariate associations of candidate variables with the primary endpoint in the derivation dataset. Variables with a P-value <.10 were used in the final multivariable logistic regression model. Univariate analyses were performed with all available variables, whereas multivariate analyses were performed only with patients who had all variables available. A backward selection was performed using an exclusion criterion of P≥.10. The final model included 12 parameters that were associated with the primary endpoint. A linear calculator was developed to determine the probability of the primary endpoint for each patient. For that, the regression coefficients of the covariables and the intercept of the fitted risk model were summed up. Then, a risk percentage was calculated.²⁵

Risk model discrimination accuracy was evaluated using receiver operating characteristic (ROC) curve and area under the curve (AUC) calculated for the score derived from the multivariable logistic regression model and established risk scores, both in the derivation and the validation cohort. Risk model calibration accuracy/goodness of fit was evaluated graphically by the stratification of patients into risk quartiles and comparisons of observed vs expected events within risk strata, as well as by using the Kaplan-Meier method and log-rank testing. Also, AUCs were calculated comparing the performance of the novel score in different years of therapy. A directly applicable digital calculator for TARI that easily calculates the individual risk for 1-year mortality was programmed and tested (TARI score calculations shown  in Supplemental Table S1).

Results

Derivation and validation patient cohorts. After selection, a total of 20,704 patients from GARY formed the derivation cohort, while 3982 patients from SWENTRY formed the validation cohort. Complete data were available in 19,642/20,704 patients (94.9%). Patient baseline characteristics of both cohorts are shown in Table 1. Mean age in both cohorts was 81 years, male and female sex frequencies were similar, and typical cardiovascular comorbidities were common. One-year mortality was 15.6% (GARY) and 9.5% (SWENTRY).

Variable identification and TARI modeling. Univariate and multivariate regression analyses were performed in the derivation cohort to identify variables predicting 1-year mortality. Results are presented in Table 2. Age, body mass index (BMI), creatinine level, NYHA class, pulmonary hypertension, atrial fibrillation, diabetes mellitus, vascular disease, chronic obstructive pulmonary disease, prior stroke, LVEF (trichotomized as described above), and MR were identified as significant predictors of 1-year mortality and were incorporated into the TARI model. The univariate odds ratio (OR) represents the OR per unit increase in the logarithm of age, BMI, and creatinine. For LVEF, trichotomized values were applied. Constant and regression coefficients derived from multivariable analysis were combined to form the TARI model.

TARI risk model performance. Mean 1-year mortality risk prediction from the TARI model was 15.1 ± 7.6% and 15.0 ± 7.3% for the derivation and validation cohort, respectively. In the GARY derivation cohort, calibration analysis showed very good agreement of predicted event probability and observed events in risk quartiles. In the SWENTRY validation cohort, TARI overestimated risk only slightly in the lowest-risk quartile, but showed a steady trend to more divergence at higher risk (~1/3 overestimation).

TARI model discrimination performance for 1-year mortality in the GARY derivation cohort (AUC, 0.66; 95% confidence interval [CI], 0.65-0.67) was slightly superior to its results in the SWENTRY validation cohort (AUC, 0.62; 95% CI, 0.60-0.64) (Figure 2A and Figure 2B). The TARI model performed superior to both STS (AUC, 0.63; 95% CI, 0.62-0.64; P<.001 vs TARI) and LogES I (AUC, 0.60; 95% CI, 0.59-0.61; P<.001 vs TARI) in the GARY derivation cohort (Figure 2A) and also discriminated risk for 1-year mortality better than LogES I in the SWENTRY validation cohort (AUC, 0.59; 95% CI, 0.57-0.61; P=.04 vs TARI) (Figure 2B).

Patients were stratified into quartiles according to TARI-­predicted 1-year mortality risk (Q1, 0% to 9%; Q2, >9% to 13%; Q3, >13% to 19%; Q4, >19%). Both in the GARY derivation cohort (log-rank P<.001) (Figure 3) and in the SWENTRY validation cohort (log-rank P<.001) (Figure 4), survival analysis revealed significant increases in mortality risk in higher-risk quartiles over the course of 1-year follow-up.

Despite the fact that TARI was developed for elective cases, the score was also tested on all TAVR cases including nonelective cases from the GARY cohort. TARI performed best (AUC, 0.66; 95% CI, 0.65-0.67; P<.05 against both) compared with STS (AUC, 0.64; 95% CI, 0.63-0.65) and LogES I (AUC, 0.61; 95% CI, 0.60-0.62).

To investigate the performance of TARI during the evolution of TAVR treatment, a comparison was made for different years. Equivalent performance was observed in 2011 (AUC, 0.637; 95% CI, 0.606-0.669), 2012 (AUC, 0.640; 95% CI, 0.613-0.668), 2013 (AUC, 0.666; 95% CI, 0.642-0.691), 2014 (AUC, 0.661; 95% CI, 0.640-0.682), and 2015 (AUC, 0.664; 95% CI, 0.644-0.684).

Discussion

We present the novel TARI risk model for the prediction of 1-year mortality in patients undergoing elective transfemoral TAVR, based on a large national registry on aortic valve treatment in Germany (GARY). After multivariable adjustment, age, BMI, creatinine level, NYHA class, pulmonary hypertension, atrial fibrillation, diabetes mellitus, vascular disease, chronic obstructive pulmonary disease, prior stroke, LVEF, and MR emerged as independent predictors of 1-year mortality and were integrated in the risk model. TARI showed better performance than surgical models in the GARY derivation cohort and was confirmed with external validation in the SWENTRY registry.

During the last decade, transfemoral TAVR has developed from an experimental procedure in inoperable highest-risk patients to an accepted alternative in elderly patients at intermediate²⁶ and low surgical risk.²⁷ Increases in procedure counts and center volumes,²⁸ operator experience,²⁹ and technological advances in devices and delivery systems have all contributed to a continued improvement in outcomes.

One challenge in the TAVR field has been the ability to predict longer-term survival of patients. Dedicated TAVR risk models developed from national registries thus far disappointed in their predictive performance. For the prediction of 30-day mortality, available models such as GAVS-II,¹³ FRANCE-2,¹² or OBSERVANT¹¹ did not perform superior to surgical risk models (LogES I, ES II, STS PROM) and failed in external validation studies.^14,30

Predicting the risk of 1-year mortality may be more relevant to the heart team and the patients than 30-day mortality when deciding the treatment options, but is also considerably more challenging: Available registries with high-quality data on longer-term outcomes are rare. The few TAVR-specific models aimed at the prediction of 1-year mortality have been developed from small patient cohorts (n = 845; n = 3687; n = 511).^15-17 Additionally, they lack external validation, limiting their utility in other patient cohorts. The prediction of 1-year mortality may also assist in guiding decisions of who may not benefit from treatment, such as patients with a predicted 1-year mortality of >50%.

We aimed to improve TAVR 1-year risk prediction by using the large and high-quality GARY cohort¹⁹ for model development. With our selection of predictors of 1-year mortality, the model is fast, simple, and accurate. Patient selection included patients with complete 1-year mortality data who underwent elective transfemoral TAVR. Prespecified external validation in the SWENTRY cohort was used to test model translation to a different European cohort. GARY reflects a typical patient collective and best clinical practice of TAVR in Germany. Both registries feature an “all-comer” design over a period of many years, thus including influence of learning curves and innovations in implantation technology and devices with possible impact on outcomes.

The resulting TARI model discriminates risk for 1-year mortality comparably in derivation and validation cohorts with a significant performance advantage over surgical risk models. Surgical risk models were mainly evaluated to predict 30-day TAVR mortality,¹⁴ which is less challenging than predicting 1-year mortality. However, the AUCs of TARI for 1-year mortality were comparable to the AUCs of 30-day mortality from the surgical scores. Risk quartiles were significantly different in Kaplan-Meier analyses and coherent between both cohorts during follow-up through to 1-year. The model demands further external validation to prove the benefit of 1-year mortality prediction in heart team decision making, where the focus is still primarily on short-term (30-day) risk as compared with longer-term benefits of the TAVR procedure. Changing this may result in more patient-centered decision making, where attention shifts from immediate results of the procedure to the postprocedural improvement in functional status that could be achieved for each individual patient. ­Ultimately, this may lead to identifying patients who may not benefit from treatment due to their high risk of 1-year mortality (>50%).

Study limitations. Patients in the GARY cohort are recruited over a long period of time starting in 2011 with an evolving technology over time. However, over a 5-year period, the analysis provided no evidence of different discriminatory power during this evolving technology, but we cannot safely exclude this in the following years. Early recruited patients might not be treated with state-of-the-art technology in 2021, such as earlier valve generations. Despite the strength of a large cohort, the limitation persists that data after 2015 were not available regarding the main outcome after 1 year. In the SWENTRY cohort, STS score was not available as a comparator to validate superiority, which constitutes another limitation of the study. Another limitation is that there are no sufficient data to include any measurement of frailty in the models. Certainly, frailty can help improve outcome prediction and should be incorporated into the individual assessment of patients beyond scores.³¹ Given these limitations, the most up-to-date available data of 2 large European registries have been used.

Conclusion

The novel TARI score is the first dedicated risk model to provide externally validated prediction for 1-year mortality in large cohorts of European TAVR patients while providing better predictive performance compared with classical surgical scores. Compared with commonly used scores, the calculation needs considerably fewer variables, allowing a significantly faster calculation. This tool may help to support clinical risk stratification and decision making in heart teams.

Impact on daily practice. Competency in patient care and ­procedural skills: transcatheter aortic valve replacement (TAVR) is the primary treatment for patients with symptomatic severe aortic valve stenosis determined to have high or intermediate perioperative risk. The novel TARI score is the first dedicated and easy-to-use risk model to provide externally validated prediction for 1-year mortality. Translational outlook: the score should be incorporated into heart team protocols to support clinical risk stratification and decision making in heart teams.

Availability of data. The anonymized data are available from the authors upon reasonable request and after approval from the GARY and SWENTRY executive boards.

Affiliations and Disclosures

*Joint first authors.

From ¹Heinrich-Heine-University Duesseldorf, Department of Cardiology, Pulmonology and Vascular Medicine, Düsseldorf, Germany; ²Department of Cardiac Thoracic Vascular Surgery, University of Lübeck, Lübeck, Germany; ³German Center for Cardiovascular Research, DZHK, Partner Site Lübeck, Lübeck, Germany; ⁴Department of Medicine, Unit of Cardiology, Karolinska Institutet, and Heart and Vascular Theme, Department of Cardiology, Karolinska University Hospital, Stockholm, Sweden; ⁵Department of Department of Anaesthesiology, Perioperative Medicine and Intensive Care Medicine, Paracelsus Medical University, Salzburg, Austria; ⁶Heinrich-Heine-University Duesseldorf, Department of Cardiovascular Surgery, Düsseldorf, Germany; ⁷Department of Cardiovascular Surgery, University Heart Center Freiburg–Bad Krozingen and Albert Ludwigs University, Freiburg, Germany; ⁸Department of Cardiology, Sana Klinikum Offenbach, Offenbach, Germany; ⁹Department of Cardiology, Robert-Bosch-Krankenhaus, Stuttgart, Germany; ¹⁰Clinic for Thoracic and Cardiovascular Surgery, Heart and Diabetes Centre North Rhine Westphalia, Bad Oeynhausen, Germany; ¹¹German Society of Thoracic, Cardiac and Vascular Surgery (Deutsche Gesellschaft für Thorax, Herz- und Gefäßchirurgie, DGTHG), Berlin, Germany; ¹²Department of Cardiology, University of Lübeck, Lübeck, Germany; ¹³Department of Cardiology, St-Johannes-Hospital Dortmund, Dortmund, Germany; ¹⁴Department of Cardiothoracic Surgery, University Hospital Frankfurt, Frankfurt, Germany; ¹⁵Department of Cardiology, Goethe University Hospital Frankfurt, Frankfurt am Main, Germany; ¹⁶Department of Cardiology, University of Giessen, Giessen, Germany; ¹⁷Division of Cardiology, Piedmont Hospital, Atlanta, Georgia; and ¹⁸Cardiovascular Research Institute Düsseldorf, Heinrich Heine University Düsseldorf, Düsseldorf, Germany.

Funding. The authors acknowledge the support of the Transregio “TRR 259 - Aortic Disease” by the German Research Council. This work was supported by the Forschungskommission of the Medical Faculty of the Heinrich-Heine-University Düsseldorf, No. 2018-32 to GW and No. 2020-21 to RRB for a Clinician Scientist Track.

Disclosure: The authors have completed and returned the ICMJE Form for Disclosure of Potential Conflicts of Interest. The authors report no conflicts of interest regarding the content herein.

Manuscript accepted May 9, 2022.

Address for correspondence: Christian Jung, MD, PhD, Division of Cardiology, Pulmonology, and Vascular Medicine, University Duesseldorf, Moorenstraße 5, 40225 Duesseldorf, Germany. Email: christian.jung@med.uni-duesseldorf.de. Twitter: @cjungMD

References

1. Baumgartner H, Falk V, Bax JJ, et al. 2017 ESC/EACTS Guidelines for the management of valvular heart disease. Eur Heart J. 2017;38(36):2739-2791. doi:10.1093/eurheartj/ehx391

2. Nishimura RA, Otto CM, Bonow RO, et al. 2017 AHA/ACC Focused update of the 2014 AHA/ACC guideline for the management of patients with valvular heart disease: a report of the American College of Cardiology/American Heart Association task force on clinical practice guidelines. J Am Coll Cardiol. 2017;135(25):e1159-e1195. doi:10.1161/CIR.0000000000000503

3. Bekeredjian R, Szabo G, Balaban U, et al. Patients at low surgical risk as defined by the Society of Thoracic Surgeons score undergoing isolated interventional or surgical aortic valve implantation: in-hospital data and 1-year results from the German aortic valve registry (GARY). Eur Heart J. 2019;40(17):1323-1330. doi:10.1093/eurheartj/ehy699

4. Roques F, Michel P, Goldstone AR, Nashef SA. The logistic EuroSCORE. Eur Heart J. 2003;24(9):881-882. doi:10.1016/s0195-668x(02)00799-6

5. Nashef SA, Roques F, Michel P, Gauducheau E, Lemeshow S, Salamon R. European system for cardiac operative risk evaluation (EuroSCORE). Eur J Cardiothorac Surg. 1999;16(1):9-13. doi:10.1016/s1010-7940(99)00134-7

6. Nashef SA, Roques F, Sharples LD, et al. EuroSCORE II. Eur J Cardiothorac Surg. 2012;41(4):734-44; discussion 744-745. doi:10.1093/ejcts/ezs043

7. O’Brien SM, Shahian DM, Filardo G, et al. The Society of Thoracic Surgeons 2008 cardiac surgery risk models: part 2—isolated valve surgery. Ann Thorac Surg. 2009;88(1 Suppl):S23-S42. doi:10.1016/j.athoracsur.2009.05.056

8. Piazza N, Wenaweser P, van Gameren M, et al. Relationship between the logistic EuroSCORE and the Society of Thoracic Surgeons predicted risk of mortality score in patients implanted with the CoreValve ReValving system—a Bern-Rotterdam study. Am Heart J. 2010;159(2):323-329. doi:10.1016/j.ahj.2009.11.026

9. Martin GP, Sperrin M, Ludman PF, et al. Inadequacy of existing clinical prediction models for predicting mortality after transcatheter aortic valve implantation. Am Heart J. 2017;184:97-105. doi:10.1016/j.ahj.2016.10.020

10. Wang TKM, Wang MTM, Gamble GD, Webster M, Ruygrok PN. Performance of contemporary surgical risk scores for transcatheter aortic valve implantation: a meta-analysis. Int J Cardiol. 2017;236:350-355. doi:10.1016/j.ijcard.2016.12.188

11. Capodanno D, Barbanti M, Tamburino C, et al. A simple risk tool (the OBSERVANT score) for prediction of 30-day mortality after transcatheter aortic valve replacement. Am J Cardiol. 2014;113(11):1851-1858. doi:10.1016/j.amjcard.2014.03.014

12. Iung B, Laouenan C, Himbert D, et al. Predictive factors of early mortality after transcatheter aortic valve implantation: individual risk assessment using a simple score. Heart. 2014;100(13):1016-1023. doi:10.1136/heartjnl-2013-305314

13. Schiller W, Barnewold L, Kazmaier T, et al. The German aortic valve score II. Eur J Cardiothorac Surg. 2017;52(5):881-887. doi:10.1093/ejcts/ezx282

14. Wolff G, Shamekhi J, Al-Kassou B, et al. Risk modeling in transcatheter aortic valve replacement remains unsolved: an external validation study in 2946 German patients. Clin Res Cardiol. 2021;110(3):368-376. doi:10.1007/s00392-020-01731-9

15. Seiffert M, Sinning JM, Meyer A, et al. Development of a risk score for outcome after transcatheter aortic valve implantation. Clin Res Cardiol. 2014;103(8):631-640. doi:10.1007/s00392-014-0692-4

16. Hermiller JB, Yakubov SJ, Reardon MJ, et al. Predicting early and late mortality after transcatheter aortic valve replacement. J Am Coll Cardiol. 2016;68(4):343-352. doi:10.1016/j.jacc.2016.04.057

17. Debonnaire P, Fusini L, Wolterbeek R, et al. Value of the “TAVI2-SCORe” versus surgical risk scores for prediction of one year mortality in 511 patients who underwent transcatheter aortic valve implantation. Am J Cardiol. 2015;115(2):234-242. Epub 2014 Oct 29. doi:10.1016/j.amjcard.2014.10.029

18. Martin GP, Sperrin M, Mamas MA. Pre-procedural risk models for patients undergoing transcatheter aortic valve implantation. J Thorac Dis. 2018;10(Suppl 30):S3560-S3567. doi:10.21037/jtd.2018.05.67

19. Hamm CW, Beyersdorf F. GARY—the largest registry of aortic stenosis treatment worldwide: the German aortic valve registry (GARY) established in 2010 has been accumulating data for a decade now. Eur Heart J. 2020;41(6):733-735. doi:10.1093/eurheartj/ehaa048

20. Lange R, Beckmann A, Neumann T, et al. Quality of life after transcatheter aortic valve replacement: prospective data from GARY (German aortic valve registry). JACC Cardiovasc Interv. 2016;9(24):2541-2554. doi:10.1016/j.jcin.2016.09.050

21. Abdel-Wahab M, Fujita B, Frerker C, et al. Transcatheter versus rapid-deployment aortic valve replacement: a propensity-matched analysis from the German aortic valve registry. JACC Cardiovasc Interv. 2020;13(22):2642-2654. doi:10.1016/j.jcin.2020.09.018

22. Fujita B, Schmidt T, Bleiziffer S, et al. Impact of new pacemaker implantation following surgical and transcatheter aortic valve replacement on 1-year outcome. Eur J Cardiothorac Surg. 2020;57(1):151-159. doi:10.1093/ejcts/ezz168

23. Feldt K, De Palma R, Bjursten H, et al. Change in mitral regurgitation severity impacts survival after transcatheter aortic valve replacement. Int J Cardiol. 2019;294:32-36. doi:10.1016/j.ijcard.2019.07.075

24. Lancellotti P, Moura L, Pierard LA, et al. European Association of Echocardiography recommendations for the assessment of valvular regurgitation. Part 2: mitral and tricuspid regurgitation (native valve disease). Eur J Echocardiogr. 2010;11(3):223-244. doi:10.1093/ejechocard/jeq031

25. Pavlou M, Ambler G, Seaman SR, et al. How to develop a more accurate risk prediction model when there are few events. BMJ. 2015;351:h3868. doi:10.1136/bmj.h3868

26. Reardon MJ, Van Mieghem NM, Popma JJ, et al. Surgical or transcatheter aortic-valve replacement in intermediate-risk patients. N Engl J Med. 2016;374(17):1609-1620.  doi:10.1056/NEJMoa1700456

27. Mack MJ, Leon MB, Thourani VH, et al. Transcatheter aortic-valve replacement with a balloon-expandable valve in low-risk patients. N Engl J Med. 2019;380(18):1695-1705. Epub 2019 Mar 16. doi:10.1056/NEJMoa1814052

28. Wassef AWA, Rodes-Cabau J, Liu Y, et al. The learning curve and annual procedure volume standards for optimum outcomes of transcatheter aortic valve replacement: findings from an international registry. JACC Cardiovasc Interv. 2018;11(17):1669-1679. doi:10.1016/j.jcin.2018.06.044

29. Salemi A, Sedrakyan A, Mao J, et al. Individual operator experience and outcomes in transcatheter aortic valve replacement. JACC Cardiovasc Interv. 2019; 12(1):90-97. Epub 2018 Dec 12. doi:10.1016/j.jcin.2018.10.030

30. Pilgrim T, Franzone A, Stortecky S, et al. Predicting mortality after transcatheter aortic valve replacement: external validation of the transcatheter valve therapy registry model. Circ Cardiovasc Interv. 2017;10(11):e005481. doi:10.1161/CIRCINTERVENTIONS.117.005481

31. Bruno RR, Wolff G, Wernly B, Kelm M, Jung J. Frailty assessment in patients undergoing aortic valve replacement: be quick and be sure. JACC Cardiovasc Interv. 2020;13(16):1965-1967. doi:10.1016/j.jcin.2020.06.006

Related Reading

• Outcomes of Radial Versus Femoral Access in Patients With Severe Aortic Stenosis Undergoing Percutaneous Coronary Intervention Prior to Transcatheter Aortic Valve Replacement
• Transcatheter Aortic Valve Replacement for Mixed Aortic Valve Disease: A Propensity Score-Adjusted Analysis From the RISPEVA Registry
• Generational Differences in Outcomes of Self-Expanding Valves for Transcatheter Aortic Valve Replacement
• Unplanned Hospital Readmissions After Transcatheter Aortic Valve Replacement in the Era of New-Generation Devices
• Reframing Optimal Implantation of the Sapien 3 Transcatheter Heart Valve