Incidence and Outcomes of Infective Endocarditis After Transcatheter or Surgical Aortic Valve Replacement

Abstract

Background

Data comparing the frequency and outcomes of infective endocarditis (IE) after transcatheter (TAVR) to surgical aortic valve replacement (SAVR) are scarce. The objective of this study is to compare the incidence and outcomes of IE after TAVR using a supra‐annular, self‐expanding platform (CoreValve and Evolut) to SAVR.

Methods and Results

Data of 3 randomized clinical trials comparing TAVR to SAVR and a prospective continued TAVR access study were pooled. IE was defined on the basis of the modified Duke criteria. The cumulative incidence of IE was determined by modeling the cause‐specific hazard. Estimates of all‐cause mortality were calculated by means of the Kaplan–Meier method. Outcomes are reported for the valve‐implant cohort. During a mean follow‐up time of 2.17±1.51 years, 12 (0.5%) of 2249 patients undergoing TAVR and 21 (1.1%) of 1828 patients undergoing SAVR developed IE. Patients with IE more frequently had diabetes mellitus than those without (57.6% versus 34.2%; P=0.005). The cumulative incidence of IE was 1.01% (95% CI, 0.47%–1.96%) after TAVR and 1.58% (95% CI, 0.97%–2.46%) after SAVR (P=0.047) at 5 years. Among patients with IE, the rate of all‐cause mortality was 27.3% (95% CI, 1.0%–53.6%) in the TAVR and 51.8% (95% CI, 28.2%–75.3%) in the SAVR group at 1 year (log‐rank P=0.15).

Conclusions

Pooled prospectively collected data comparing TAVR with a supra‐annular, self‐expanding device to SAVR showed a low cumulative risk of IE irrespective of treatment modality, although the risk was lower in the TAVR implant group. Once IE occurred, mortality was high.

Registration

URL: https://www.clinicaltrials.gov; Unique identifiers: NCT01240902, NCT01586910, NCT02701283.

Nonstandard Abbreviations and Acronyms

IE

infective endocarditis

PARTNER

Placement of Aortic Transcatheter Valves

SAVR

surgical aortic valve replacement

SURTAVI

Surgical Replacement and Transcatheter Aortic Valve Implantation

TAVR

transcatheter aortic valve replacement

Clinical Perspective

What Is New?

• This study reports on the incidence and outcomes of infective endocarditis observed in pooled data of 3 randomized controlled trials and a prospective continued access registry comparing transcatheter aortic valve replacement using a self‐expanding platform to surgical aortic valve replacement.

• Although infective endocarditis was infrequent after aortic valve replacement therapies, its cumulative incidence was lower after transcatheter aortic valve replacement with a self‐expanding valve than after surgical aortic valve replacement.

• Annular abscess formation was more frequently encountered in endocarditis after surgical aortic valve replacement than transcatheter aortic valve replacement without evidence of a different spectrum of causative microorganisms in the 2 groups.

What Are the Clinical Implications?

• Among patients undergoing aortic valve replacement, the risk of infective endocarditis is low irrespective of the mode of replacement.

• Future studies should investigate whether the lower rates of endocarditis and abscess formation after transcatheter aortic valve replacement using a self‐expanding valve compared with surgical aortic valve replacement can be corroborated.

Valve replacement is required in patients with symptomatic severe aortic stenosis to relieve symptoms and improve prognosis. ¹ Transcatheter aortic valve replacement (TAVR) and surgical aortic valve replacement (SAVR) are well‐established effective and safe treatment options for patients across the whole spectrum of surgical risk. ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ Infective endocarditis (IE) of the prosthetic valve is a deleterious complication after valve replacement, which can occur early or late and is associated with a high morbidity and mortality according to observational data. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ Previous reports do not suggest an elevated risk of IE after TAVR compared with SAVR ¹⁰ , ¹¹ , ¹³ , ¹⁶ ; however, differences in valve design, such as the structure and composition of the stent frame or the tissue type and processing of the prosthetic leaflets, as well as differences in the preparatory steps, ways of valve delivery and implantation modes may all affect the risk of endocarditis associated with each specific TAVR device. This study aims to compare the frequency, timing, and outcomes of IE after TAVR with devices of a supra‐annular, self‐expanding platform (CoreValve and Evolut) to SAVR based on pooled data of 3 randomized controlled trials and a prospective continued access study in patients with severe symptomatic aortic stenosis covering the whole range of surgical risk.

Methods

Study Design

Data of 3 multicenter randomized controlled trials comparing TAVR using devices of the self‐expanding CoreValve family to SAVR in patients with symptomatic severe aortic stenosis at high (CoreValve high risk; clinicaltrials.gov, NCT01240902), intermediate (SURTAVI [Surgical Replacement and Transcatheter Aortic Valve Implantation]); clinicaltrials.gov, NCT01586910) and low (Evolut Low Risk; clinicaltrials.gov, NCT02701283) surgical risk, and data of the SURTAVI continued access study (clinicaltrials.gov, NCT01586910) were aggregated. The studies were conducted at tertiary, high‐volume centers across North America, Asia, Europe, and Oceania, and patients were recruited between February 2011 and November 2018. Patients treated with TAVR received a self‐expanding, supra‐annular bioprosthesis of the CoreValve family (CoreValve, Evolut R, or Evolut PRO; Medtronic, Minneapolis, USA). Patients undergoing SAVR were treated with any bioprosthetic surgical valve at the discretion of the operator. All studies were approved by appropriately constituted competent ethics committees, study conduct complied with the Declaration of Helsinki, and all participants provided written informed consent before inclusion. Detailed information on the trials administrative structure and the specific protocols have been previously published. ³ , ⁶ , ⁷ Because of the sensitive nature of the data collected for this study, requests to access the data set from qualified researchers trained in human subject confidentiality protocols may be sent to Medtronic, SH&A Clinical Research & Medical Science (8200 Coral Street. MVS66, Mounds View, MN 55112).

Study Population

The study population comprises patients with symptomatic severe aortic stenosis at high, intermediate, or low surgical risk as assessed by the local heart team and the predicted risk of surgical mortality at 30 days based on the Society of Thoracic Surgeons Predicted Risk of Mortality score. All patients were deemed eligible for both TAVR and SAVR by the heart team, and anatomy had to be suitable for both treatment modalities accordingly. Patients with presence of ongoing sepsis were excluded. Details of the eligibility criteria of each included study have been published. ³ , ⁶ , ⁷

Definitions and Follow‐Up

IE was defined on the basis of the modified Duke criteria: For definite endocarditis, 2 major criteria, or 1 major and 3 minor criteria, or 5 minor criteria were required; for possible endocarditis, 1 major and 1 minor criterion, or 3 minor criteria. ⁹ Endocarditis was classified as early if it occurred within a year of valve replacement, otherwise as late. ⁹ Clinical end points were defined according to the Valve Academic Research Consortium’s (and Valve Academic Research Consortium 2) definitions and adjudicated by an independent clinical events committee. Follow‐up was performed at least 3 times in the first year and yearly thereafter, with a maximum follow‐up time of 5 years.

Statistical Analysis

Analyses were performed in the valve‐implanted population comprising the patients in whom a TAVR valve or a surgical valve was actually implanted; sensitivity analyses were performed in the as‐treated (attempted trial treatment according to allocation) populations as well as in the as‐treated cohort excluding the patients of the SURTAVI continued access study. Patient and procedural characteristics are presented as counts (percentage) for categorical variables and mean (±SD) for continuous ones. P values were derived from Student’s t‐tests for comparisons of continuous data and Fisher’s exact tests when the observed count was <5 and otherwise with the chi‐square test for categorical variables. Cumulative incidence estimates were derived by modeling the cause‐specific hazard, taking into account the competing risk of death, and curves were compared using Gray’s test. ¹⁷ Incidence rates and CIs were obtained using normal approximation to the Poisson distribution. Clinical outcomes after endocarditis were assessed by means of the Kaplan–Meier method stratified by the mode of valve replacement and survival curves were compared using the log‐rank test. All statistical analyses were performed with the use of SAS software, version 9.4 (SAS Institute Inc, Cary, NC).

Results

Baseline and Procedural Characteristics

A total of 4301 patients were randomly assigned to TAVR or SAVR in the trials or enrolled in the continued access study. In 4088 patients the assigned valve replacement procedure was attempted, and in 4077 patients a valve was implanted (Figure 1). In the valve‐implant cohort 33 cases of endocarditis occurred; 12 (0.5%) in 2249 in the TAVR group during a mean follow‐up of 2.15 (±1.49) years and 21 (1.1%) in 1828 patients in the SAVR group during a mean follow‐up of 2.17 (±1.54) years.

Figure 1. Flowchart depicting the patient flow from the intention‐to‐treat to the as‐treated and valve‐implanted cohorts in the randomized trials and the SURTAVI continued access study.

FU indicates follow‐up; SAVR, surgical aortic valve replacement; SURTAVI, Surgical Replacement and Transcatheter Aortic Valve Implantation; and TAVR, transcatheter aortic valve replacement.

The mean age of the valve‐implant cohort was 78.3±7.1 years, 42.1% were women (Table 1). Mean Society of Thoracic Surgeons Predicted Risk of Mortality score was 4.1±2.6. Patients with endocarditis experienced on average more heart failure symptoms (New York Heart Association class III/IV: 72.7% versus 51.8%; P=0.022), and were more likely to suffer from diabetes mellitus (57.6% versus 34.2%; P=0.005) at baseline (Table 1). There were no significant differences in the distribution of baseline characteristics comparing TAVR and SAVR patients with IE (Table S1). The mean time in the catheterization laboratory or operating room did not differ between patients developing endocarditis and those who did not (Table S2.) There was a numerical trend toward a higher proportion of concomitant percutaneous revascularization procedures among patients who developed subsequent endocarditis compared with those without endocarditis in the TAVR group (25% versus 8.3%; P=0.07); this was not observed for surgical revascularization in patients treated with SAVR (Table S2).

Table 1.

Baseline Clinical Characteristics Stratified by Outcome

Characteristic	Endocarditis (N=33)	No Endocarditis (N=4044)	P Value
Age, y	77.2±7.3	78.3±7.1	0.37
Female sex	36.4 (12/33)	42.1 (1702/4044)	0.51
Body mass index, kg/m2	30.7±5.0	29.7±6.0	0.35
STS score, %	4.4±2.2	4.1±2.6	0.47
NYHA class			0.024
I	0.0 (0/33)	3.4 (139/4044)
II	27.3 (9/33)	44.7 (180/4044)
III	66.7 (22/33)	45.7 (1847/4044)
IV	6.1 (2/33)	6.2 (249/4044)
Diabetes mellitus	57.6 (19/33)	34.2 (1382/4044)	0.005
Serum creatinine >2 mg/dL	3.0 (1/33)	1.6 (65/4044)	0.42
Chronic lung disease	45.5 (15/33)	30.8 (1226/3982)	0.07
Peripheral vascular disease	48.1 (13/27)	32.7 (864/2643)	0.09
Cerebrovascular disease	18.2 (6/33)	16.4 (662/4035)	0.78
History of hypertension	100.0 (33/33)	89.6 (3622/4042)	0.051
Previous PCI	21.2 (7/33)	21.6 (872/4044)	0.96
Previous CABG	15.2 (5/33)	14.0 (565/4044)	0.85
Previous MI	21.2 (7/33)	13.1 (528/4044)	0.17
Atrial fibrillation/flutter	27.3 (9/33)	25.7 (1036/4038)	0.83
Immunosuppressive therapy	9.1 (3/33)	5.9 (237/4042)	0.44
Preexisting pacemaker or ICD	3.0 (1/33)	9.9 (400/4043)	0.25

Data comprise all patients from the CoreValve Pivotal High Risk, SURTAVI, and Evolut Low Risk trials as well as the SURTAVI continued access registry. Data are presented as % (number/denominator) or as mean±SD. P values are derived from Fisher’s exact tests for categorical variables and Student t‐tests for continuous variables.

CABG indicates coronary artery bypass grafting; CCS, Canadian Cardiovascular Society; ICD, intracardiac defibrillator; MI, myocardial infarction; NYHA, New York Heart Association; PCI, percutaneous coronary intervention; and SURTAVI, Surgical Replacement and Transcatheter Aortic Valve Implantation.

Incidence of IE

The incidence rate of IE amounted to 3.74 (95% CI, 2.46–5.01) per 1000 patient‐years in the total cohort, 2.47 (95% CI, 1.07–3.87) per 1000 patient‐years in the TAVR, and 5.28 (95% CI, 3.02–7.54) per 1000 patient‐years in the SAVR group. The overall estimated cumulative incidence of endocarditis at 5 years amounted to 1.28% (95% CI, 0.83%–1.88%), taking into account the competing risk of death. In the TAVR group, the cumulative incidence of endocarditis at 5 years was 1.01% (95% CI, 0.47%–1.96%) and 1.58% (95% CI, 0.97%–2.46%) in the SAVR group (P= 0.047) (Figure 2). Sensitivity analyses showed that in the as‐treated cohort cumulative incidence was 1.05% (95% CI, 0.50%–1.99%) in the TAVR and 1.59% (95% CI, 0.98%–2.46%%) in the SAVR group at 5 years (P=0.07) (Figure S1); cumulative incidence estimates were 1.00% in the TAVR and 1.59% in the SAVR group at 5 years if patients of the SURTAVI continued access study were excluded (P=0.049) (Figure S2). Cumulative incidence rates at 2 years stratified by surgical risk categories are reported in Table S3. There was no indication of a difference in the incidence of IE according to bioprosthetic leaflet tissue type (Figure S3).

Figure 2. Cumulative incidence of endocarditis taking into account the competing risk of death in the SAVR group amounted to 0.66% (95% CI, 0.35%–1.15%) at 1 year, and 1.58% (95% CI, 0.97%–2.46%) at 5 years, in the TAVR group to 0.23% (95% CI, 0.12%–0.61%) at 1 year and 1.01% (95% CI, 0.47%–1.96%) at 5 years.

SAVR indicates surgical aortic valve replacement; and TAVR, transcatheter aortic valve replacement.

Characteristics of IE

A total of 27 (81.8%) of 33 patients fulfilled the modified Duke criteria for definite IE, the remaining the criteria for possible endocarditis (Table 2). Half of the IE cases occurred early (≤365 days), the other half late (>365 days) after valve replacement. Among patients with endocarditis, abscess formation was observed more frequently after SAVR than after TAVR (47.6% versus 8.3%; P=0.027) (Table 2). The most frequent causative microorganisms were Streptococcus (33.3%) and Enterococcus species (30.3%), followed by coagulase‐negative staphylococci (18.2%), and Staphylococcus aureus (15.2%) (Table 2). No notable differences were observed between the TAVR and SAVR groups with regard to the detected microorganisms. In both groups, roughly two‐thirds of the patients with IE were treated conservatively with antibiotic treatment only, whereas one‐third underwent surgical intervention in addition (Table 2).

Table 2.

Characteristics of Endocarditis Stratified by Mode of Valve Replacement

Characteristic	TAVR (N=12), n (%)	SAVR (N=21), n (%)	P Value
Early*	6 (50)	11 (52.4)	>0.99
Late*	6 (50)	10 (47.6)	>0.99
Definite †	10 (83.3)	17 (81)	>0.99
Possible †	2 (16.7)	4 (19)	>0.99
Echocardiographic findings
Vegetation	10 (83.3)	11 (52.4)	0.13
Abscess	1 (8.3)	10 (47.6)	0.027
Moderate or more valve regurgitation	2 (16.7)	5 (23.8)	0.99
Microorganism ‡
Gram‐positive bacilli	12 (92.3)	20 § (100)	0.39
Staphylococcus aureus	2 (15.4)	3 (15)	>0.99
Coagulase‐negative staphylococci	2 (15.4)	4 (20)	>0.99
Streptococcus species	5 (38.5)	5 § (25)	0.46
Viridans group streptococci	5 (38.5)	2 (10)	0.08
Non‐viridans group streptococci	0 (0)	2 (10)	0.51
Enterococcus species	3 (23.1)	7 (35)	0.70
Gram‐negative bacilli	1 (7.7)	0 (0)	0.39
Polymicrobial (≥2 microorganisms)	1 (7.7)	3 (15)	>0.99
Not documented	0 (0)	4 (20)	0.14
Treatment
Antibiotic only	8 (66.7)	13 (61.9)	>0.99
Valve surgery	4 (33.3)	8 (38.1)	>0.99

^*Early ≤365 days, late >365 days after the index intervention.

^† According to modified Duke criteria.

^‡ In this section, percentages refer to total number of identified microorganisms and not patients.

^§ One microorganism not further specified.

SAVR indicates surgical aortic valve replacement; TAVR, transcatheter aortic valve replacement.

Outcomes of IE

In the overall cohort, all‐cause mortality after the occurrence of IE was 42.3% (95% CI, 24.5%–60.1%) at 1 year. In the TAVR cohort, 1‐year all‐cause mortality amounted to 27.3% (95% CI, 1.0%–53.6%), in the SAVR group to 51.8% (95% CI, 28.2%–75.3%) (P=0.15) (Figure 3). The composite of all‐cause mortality and stroke occurred in 55.0% (95% CI, 24.5%–85.5%) in the TAVR and 64.6% (95% CI, 32.1%–97.1%) in the SAVR group at 2 years (P=0.71) (Figure S4). Mortality in the cohort including only patients of the randomized trials is shown in Figure S5. One‐year mortality did not differ between patients with endocarditis with and without abscess formation (45.5% versus 40.7%; P=0.50) (Figure S6). One‐year mortality rates stratified by surgical risk category are reported in Table S4.

Figure 3. Kaplan–Meier curves depicting all‐cause mortality after endocarditis stratified by mode of valve replacement.

In the TAVR cohort, 1‐year all‐cause mortality was 27.3% (95% CI, 1.0%–53.6%) and in the SAVR group 51.8% (95% CI, 28.2%–75.3%). SAVR indicates surgical aortic valve replacement; and TAVR, transcatheter aortic valve replacement.

Discussion

This analysis of pooled data of 3 large randomized clinical trials and a prospective continued access study showed a higher prevalence of diabetes mellitus and symptoms of heart failure in patients developing IE after aortic valve replacement at baseline than in those who do not. Incidence rates of IE were 2.47 per 1000 person‐years with TAVR and 5.28 per 1000 person‐years with SAVR. In the valve‐implanted cohort, the cumulative incidence at 5 years was lower in those who underwent TAVR (1.01%) than those who received SAVR (1.58%). Half of the endocarditis cases occurred within a year of valve replacement. Abscess formation was more frequently reported in patients with endocarditis after SAVR than after TAVR (47.6% versus 8.3%). Streptococcus and Enterococcus species were the most frequent causative microorganisms followed by Staphylococcus aureus. About a third of the endocarditis patients underwent surgical intervention. All‐cause mortality after endocarditis was 42.3% at 1 year, with a numerically higher rate in the SAVR than the TAVR group (51.8% versus 27.3%).

The reported incidence rates of IE after TAVR and SAVR range between 1% and 2% per year in the vast majority of observational studies. ¹⁰ , ¹² , ¹⁵ , ¹⁶ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ An analysis of pooled data encompassing 8530 patients included in the PARTNER (Placement of Aortic Transcatheter Valves) I and II trial series and registries reported a lower overall incidence rate of IE with 0.5% per year. ¹¹ The incidence rates observed in our analysis are in line with these latter findings (0.4% per year). Whether the rates observed in randomized controlled trials are lower because of a reduction in misclassification as a consequence of the independent adjudication of events or rather because of an underreporting of events in these trials as endocarditis was merely a secondary outcome remains unknown.

Several studies have compared the incidence of IE after TAVR with SAVR. Neither crude incidence rates nor studies that performed adjustment for potential confounders by means of regression analysis or propensity‐score matching suggest significantly different rates of IE after surgical or transcatheter valve replacement. ¹⁰ , ¹¹ , ¹⁶ , ¹⁸ , ²² In this context, the lower cumulative incidence of endocarditis observed in the TAVR group in this study has to be interpreted with caution.

The potential predictors of IE after aortic valve replacement reported in the literature vary considerably ¹¹ , ¹² , ¹³ , ¹⁵ , ¹⁶ , ²¹ , ²² ; on the one hand, this heterogeneity can be explained by the fact that endocarditis is a rare event, and on the other hand by a lack of granularity of data with respect to patient‐ and procedure‐related factors.

Two‐thirds of the endocarditis cases observed in our cohort were caused by typical microorganisms as defined by the modified Duke criteria. ⁹ The high proportion of enterococci species observed as causative microorganisms is consistent with previous reports; the proportion of Staphylococcus aureus appears lower, but inferences are precluded by the low number of overall cases. ¹¹ , ¹² , ¹⁴ , ¹⁵ , ¹⁶ , ²⁰ , ²³ , ²⁴

Data comparing the incidence of periannular abscess formation observed in endocarditis cases after SAVR to TAVR are scarce. An observational study that investigated endocarditis cases after aortic valve replacement in Finland also reported higher rates of abscesses detected by echocardiography in the SAVR group (0% versus 32.1% [P=0.011]). ²² Rates of periannular aortic abscesses detected in patients undergoing TAVR with diagnosed endocarditis range between 3.6% to 19.1% in the literature, ¹² , ¹⁶ , ²⁰ , ²⁵ whereas reported rates in patients undergoing SAVR vary from 30% to 55%. ²⁶ , ²⁷ , ²⁸ Whether the higher proportion of periannular abscesses found in the SAVR group is related to procedural differences such as the resection of the native aortic valve and deeper wound trauma incurred during SAVR or whether this finding is by chance or caused by detection bias remains unknown and warrants further investigation in future studies.

In contrast to the discrepancies observed between randomized and observational studies regarding the incidence rates of endocarditis, the 1‐year all‐cause mortality rate of 42.3% in our cohort is in accordance with the rates observed in observational studies. ¹² , ¹⁴ , ¹⁸ , ²¹ , ²² Studies consistently report a rapid increase in mortality during the first months after the occurrence of endocarditis, and a mortality of 30% to 50% of the affected population at 1 year and 50% to 70% at 2 years. ¹² , ¹⁴ , ¹⁸ , ²¹ , ²² The numerical difference in 1‐year mortality after TAVR and SAVR observed in our study did not reach statistical significance and was not attributable to the higher prevalence of abscess formation observed in patients with endocarditis after SAVR.

Although this analysis was based on data obtained from rigorously conducted prospective randomized trials and studies, there are certain limitations. Notwithstanding the independent adjudication of all events by an independent clinical event committee, diagnosis of endocarditis is complex and misclassification is possible as the diagnostic value of the modified Duke criteria is limited, and multimodality imaging, which could enhance diagnostic sensitivity, is not performed frequently enough. ²⁹ , ³⁰ Treatment crossovers may distort results of the comparison between TAVR and SAVR; however, robustness of findings in the valve‐implant cohort was assessed by performing sensitivity analyses in the as‐treated study population and by excluding the patients of the nonrandomized continued access study. A further limitation of the presented analysis is the lack of information on antimicrobial prophylaxis. The low number of cases precludes the inference of predictors, a more detailed analysis of causative microorganisms in relation to the timing of endocarditis as well as assessment of differences in IE rates between surgical risk categories.

In conclusion, this analysis of pooled prospective data comparing TAVR with a supra‐annular, self‐expanding device to SAVR showed a low cumulative risk of IE in both groups, although it was lower in the TAVR implant group. If endocarditis occurred, mortality rates were high irrespective of the mode of valve replacement.

Sources of Funding

Medtronic (Minneapolis, MN, USA) funded the CoreValve Pivotal High‐Risk, SURTAVI, and Evolut Low Risk trials and the SURTAVI continued access study.

Disclosures

Dr Reardon reports fees for educational services from Medtronic. Dr Pilgrim reports research grants to the institution from Edwards Lifesciences, Boston Scientific, and Biotronik; personal fees from Biotronik and Boston Scientific; and other from HighLife SAS. Dr Stortecky has received research grants to the institution from Edwards Lifesciences, Medtronic, Abbott Vascular, and Boston Scientific; serves as consultant for BTG and Teleflex; and has received speaker fees from BTG and Boston Scientific. Dr Chetcuti reports grant support, fees for proctoring, and consulting fees from Medtronic and JenaValve Technology Inc. Dr Yakubov reports research grant support and fees for proctoring services from Medtronic. Dr Gleason reports grant support paid to his institution from Medtronic and Boston Scientific and serves as an advisor to Abbott. Dr Huang is an employee of Medtronic. Dr Windecker reports research and educational grants to the institution from Abbott, Amgen, BMS, Bayer, Boston Scientific, Biotronik, Cardinal Health, CSL Behring, Daiichi Sankyo, Edwards Lifesciences, Johnson & Johnson, Medtronic, Querbet, Polares, Sanofi, Terumo, and Sinomed. The remaining authors have no disclosures to report.

Supporting information

Tables S1–S4

Figures S1–S6