Rivaroxaban in Peripheral Artery Disease After Revascularization: Worst Events and Net Outcomes in VOYAGER PAD

Abstract

Background

VOYAGER PAD (Efficacy and Safety of Rivaroxaban in Reducing the Risk of Major Thrombotic Vascular Events in Subjects With Symptomatic Peripheral Artery Disease Undergoing Peripheral Revascularization Procedures of the Lower Extremities) demonstrated that antiplatelet therapy with rivaroxaban 2.5 mg twice daily compared with placebo reduced the first event in a composite end point in patients with peripheral artery disease after revascularization. The purpose of this analysis was to (1) evaluate whether rivaroxaban decreased the most severe events and (2) compare the harm and benefit of rivaroxaban on the same scale.

Methods

Two exploratory, hierarchical analyses of the primary efficacy composite outcome and one exploratory analysis of the net clinical benefit composite outcome were prespecified. The global rank method and the win ratio method ranked components of the composite outcome by clinical importance. The net clinical benefit included both efficacy and safety outcomes.

Results

The global rank method demonstrated that participants on rivaroxaban had fewer or later events than their placebo counterparts when events were compared in a hierarchical manner (primary ranking, P=0.0158). The win ratio approach demonstrated fewer or later fatal and nonfatal ischemic events in the rivaroxaban arm (win ratio, 1.16 [95% CI, 1.03–1.30]; P=0.0167). The net clinical benefit analysis revealed an overall beneficial effect of rivaroxaban when different efficacy and safety end points were included (P<0.001).

Conclusions

Evaluation of the VOYAGER PAD primary outcome exploring ranking of components revealed that (1) rivaroxaban decreased the most severe ischemic events and (2) that across different efficacy and safety outcomes, results consistently supported a favorable benefit–risk profile. These results help inform shared decision making on the use of rivaroxaban 2.5 mg twice daily with daily antiplatelet therapy.

Registration

URL: https://www.clinicaltrials.gov; unique identifier: NCT02504216.

Nonstandard Abbreviations and Acronyms

COMPASS

Cardiovascular Outcomes for People Using Anticoagulation Strategies

TIMI

thrombolysis in myocardial infarction

VOYAGER PAD

Efficacy and Safety of Rivaroxaban in Reducing the Risk of Major Thrombotic Vascular Events in Subjects With Symptomatic Peripheral Artery Disease Undergoing Peripheral Revascularization Procedures of the Lower Extremities

Clinical Perspective.

What Is New?

• Further, prespecified, exploratory analyses of VOYAGER PAD (Efficacy and Safety of Rivaroxaban in Reducing the Risk of Major Thrombotic Vascular Events in Subjects With Symptomatic Peripheral Artery Disease Undergoing Peripheral Revascularization Procedures of the Lower Extremities) evaluating low‐dose rivaroxaban versus placebo (both with daily antiplatelet therapy) for patients with symptomatic peripheral artery disease after successful lower extremity revascularization indicate that low‐dose rivaroxaban decreased the most severe events when compared in a hierarchical manner, with the worst event being death.

• Further analysis of net clinical benefit revealed that, when compared on the same scale, the ischemic benefit of low‐dose rivaroxaban was greater than bleeding risk.

What Are the Clinical Implications?

• This information can help guide shared decision making when discussing potential ischemic benefits and ischemic risks in patients who have undergone successful lower‐extremity revascularization for symptomatic peripheral artery disease.

Peripheral artery disease (PAD) affects ≈238 million people worldwide and is associated with functional impairment, morbidity, and death. ¹ VOYAGER PAD (Efficacy and Safety of Rivaroxaban in Reducing the Risk of Major Thrombotic Vascular Events in Subjects With Symptomatic Peripheral Artery Disease Undergoing Peripheral Revascularization Procedures of the Lower Extremities; ClinicalTrials.gov identifier: NCT02504216) was an international, randomized controlled trial that demonstrated that rivaroxaban, on a background of antiplatelet therapy, reduced the composite end point of acute limb ischemia, major amputation of vascular pathogenesis, myocardial infarction, ischemic stroke, or cardiovascular death in patients who underwent successful lower‐extremity revascularization for symptomatic PAD. ² A novel aspect of the study design was the inclusion of major adverse limb events along with traditional major adverse cardiovascular events as part of the composite primary end point. The primary analysis was time to first event, with fatal and nonfatal events weighted equally. ³ Overall, the trial was positive; however, the benefits were numerically driven by major adverse limb events, ultimately leading to approval for those outcomes by the European Medicines Agency ⁴ and the United States Food and Drug Administration ⁵ and endorsement by other professional societies. ⁶

Time to first occurrence of an event in a composite of events is typically the primary end point in cardiovascular randomized controlled trials ⁷ and is intended to capture the clinically meaningful and aggregate effects of a possible therapy. ⁸ However, there are inherent limitations in using time‐to‐first‐event analyses, including the following: (1) Contributory events are treated with equal importance; and (2) events with the highest incidence will tend to be overrepresented among first events, and consequently, in an event‐driven trial, there is often insufficient power to evaluate treatment effects on individual components. One alternative approach, the global rank method, was first proposed by O'Brien in 1984 ⁹ to deal with multiple outcomes of interest. While not widely adopted, this approach has been proposed to improve the assessment of mechanical circulatory support devices ¹⁰ and the methodology of acute heart failure clinical trials ¹¹ and critical limb ischemia trials. ⁸ A second method to analyze composite end points, which considers seriousness, is the win ratio. This approach ranks the individual events in the composite end point by order of significance and then compares individuals from each treatment arm. ¹² , ¹³ , ¹⁴ , ¹⁵ An advantage of this approach is its ability to incorporate non–time‐to‐event variables in the composite end point. ¹³ Finally, for therapies with safety considerations, analysis of net outcomes may also help clinicians understand overall benefit–risk of an intervention, as efficacy and safety are compared on the same scale. ¹⁶ These 3 alternative analysis methods were prespecified in VOYAGER PAD to better understand the beneficial effect of the primary outcome weighting end point components in different hierarchies, when comparing individual participants from each treatment arm, and in understanding benefit–risk through a composite net outcome.

Methods

The VOYAGER PAD trial was a randomized, double‐blind, placebo‐controlled, event‐driven phase 3 study, of which the primary results have previously been reported. ² A novel feature of the trial was the inclusion of acute limb ischemia and major amputation of vascular pathogenesis in the primary composite end point. ³ Local ethics committees approved the protocol. The protocol was approved by the appropriate ethics committee at each participating site and according to local regulations. All participants provided written informed consent. The authors declare that all supporting data are available within the article and its online supplementary files.

Study Population

Between August 2015 and January 2018, 6564 participants were randomized in a 1:1 manner to rivaroxaban 2.5 mg twice daily or placebo, in addition to background therapy of aspirin 100 mg daily; the use of clopidogrel was at the treating physician's discretion. Median follow‐up was 28 months. Enrolled participants had symptomatic PAD and had undergone a technically successful lower‐extremity revascularization procedure no more than 10 days before randomization. ³

Statistical Analysis

This exploratory analysis data set included all randomized patients in VOYAGER PAD. All 3 analyses were prespecified, ² and results were based on intention to treat.

The first analysis was the global rank method. In this analysis, the components of the primary end point were ranked by order of clinical importance, using prespecified primary and alternative rankings. Each patient was assigned a rank, with worse rank assigned for worse outcome occurring first. For patients with the same outcome, patients with an earlier event were assigned a worse rank. The “winner” of the comparison was assigned a 1 and the “loser” assigned a 0. When neither participant experienced an event, or when the same event severity occurred on the same day, then each participant was assigned a 0.5. The Van Elteren test for differences between groups was applied, stratified by type of procedure and use of clopidogrel (consistent with the primary trial analysis). This test is a nonparametric statistical method to compare the distributions of continuous outcome data between 2 groups while accounting for stratification. It extends the Wilcoxon rank‐sum test by allowing for the analysis of stratified data, thereby controlled for potential confounding variables. This test does not necessitate the normality of the outcome distribution and can also analyze ordered categorical outcomes, making it a suitable alternative when a 2‐way analysis of variance may not be appropriate. ¹⁷

The second analysis used the unmatched win ratio method, ¹³ which involves 3 steps. Let N _rivaroxaban and N _placebo be the number or participants assigned to rivaroxaban and placebo arms, respectively. The first step is to form N _rivaroxaban  × N _placebo of possible pairs with 1 participant from each arm without matching on patient's risk profiles. Second, a scoring rule is predefined to declare a “winner” or “loser” in each pair according to a comparison in that composite outcome. Given a pair of participants and a composite outcome, the first patient to experience the most important event (in this case, cardiovascular death), “loses” the event. If neither participant in a pair experiences the event within the common follow‐up duration, or both experience the event at the same time, then the next important event is evaluated, and so on. The win ratio is calculated by dividing the number of wins in the rivaroxaban group by the number of losses in the rivaroxaban group. The participant who experienced the most important event first (in this case, cardiovascular death) “lost” the comparison. If neither patient experienced the event, then the next most important event was compared, and so on. The win ratio was calculated by dividing the number of wins by the number of losses for the rivaroxaban group. Finkelstein and Schoenfeld statistics ¹⁸ were used to calculate 1‐sided P values, with CIs provided from bootstrapping. The Finkelstein and Schoenfeld test is a nonparametric statistical method that combines a time‐to‐event and a longitudinal measure so that if there is a substantial difference between treatments in either measure, the test will reject the null hypothesis. It is a modification of the generalized Wilcoxon test. ¹⁸

The third analysis performed was the net clinical benefit, chosen to better understand overall risk versus benefit. This analysis used predetermined efficacy and safety events through 2 days following treatment discontinuation and included all randomized participants who received at least 1 dose of the study treatment. The predetermined efficacy end points included components of the primary composite outcome: cardiovascular death, myocardial infarction, ischemic stroke, acute limb ischemia, and major amputation. The predetermined safety end points included fatal bleeding, intracranial bleed, and thrombolysis in myocardial infarction (TIMI) major bleeding. Four net clinical benefit outcomes were prespecified. The time to first event for the composite outcome was compared between treatment groups by calculating hazard ratios (HRs) with stratified Cox regression models and P values evaluated with stratified log‐rank test, and cumulative incidences estimated by the Kaplan–Meier method. To ascertain whether the different net clinical benefit outcomes differ, pairwise estimates of correlation of the log HR from the primary analysis and the 4 net clinical benefit analyses were obtained using bootstrapping.

Results

Of the 6564 participants randomized, the median age was 67 years, and most were men (74%) and White individuals (82%). Consistent with the randomized design, treatment groups had similar rates of risk factors (including diabetes, current tobacco use, and non–PAD), had similar PAD‐related history, and were on similar medication regimens. Demographic and medical history details of the randomized population have been described previously. ²

The global rank analysis showed consistent benefit when a predefined ranked hierarchy of outcomes was compared, with cardiovascular death as the worst event (Table 1). Using the primary hierarchy result, in which ischemic stroke was assigned the second worst event, followed by acute limb ischemia, major amputation of a vascular pathogenesis and myocardial infarction, rivaroxaban was superior to placebo (P=0.0158) using the Van Elteren test for differences between groups, stratified by procedure type and clopidogrel use. The alternative hierarchy result, in which myocardial infarction was ranked as a worse event than major amputation of a vascular pathogenesis, also demonstrated superiority of rivaroxaban compared with placebo (P=0.0155).

Table 1.

Global Rank Method Exploratory, Prespecified Analysis of the VOYAGER PAD Efficacy Results

	Primary hierarchy	Alternative hierarchy
1	Cardiovascular death	Cardiovascular death
2	Ischemic stroke	Ischemic stroke
3	Acute limb ischemia	Acute limb ischemia
4	Major amputation of a vascular pathogenesis	Myocardial infarction
5	Myocardial infarction	Major amputation of a vascular pathogenesis

Participants outcomes in the rivaroxaban arm were compared with participant outcomes in the placebo arm. Those who experienced a worse event at 3 years according to the above hierarchy were assigned a worse rank. Among participants with the same event type, those having the event earlier were assigned a worse rank. In the primary hierarchy, rivaroxaban was superior to placebo (P=0.0158). In the alternative hierarchy, rivaroxaban was superior to placebo (P=0.0155). VOYAGER PAD indicates Efficacy and Safety of Rivaroxaban in Reducing the Risk of Major Thrombotic Vascular Events in Subjects With Symptomatic Peripheral Artery Disease Undergoing Peripheral Revascularization Procedures of the Lower Extremities.

In the win ratio analysis (Table 2), rivaroxaban “won” the pairing (ie, the patient taking placebo experienced the event first) 14.8% of the time (4.4% of cardiovascular death events and 10.4% of nonfatal events), and placebo “won” the pairing 12.8% of the time (4.9% of cardiovascular death events and 7.9% of nonfatal events); a tie occurred 72.4% of the time (Figure 1). This pairwise comparison, with cardiovascular death as the most severe event followed by noncardiovascular ischemic events, showed that rivaroxaban was superior to placebo, with a win ratio of 1.16 (95% CI, 1.03–1.30; P=0.0167). Alternatively, the loss ratio (or reciprocal of the win ratio) was 0.86 (95% CI, 0.77–0.97).

Table 2.

Method to Calculate Win Ratio of VOYAGER PAD Efficacy Outcomes

Outcomes	Outcome category
Patient in rivaroxaban arm had cardiovascular death first	A Placebo win
Patient in placebo arm had cardiovascular death first	B Rivaroxaban win
Patient in rivaroxaban arm had nonfatal	C
primary efficacy event first	Placebo win
Patient in placebo arm had nonfatal primary	D
efficacy event first	Rivaroxaban win
No cardiovascular death or nonfatal primary efficacy events	None of the 4 categories
Win ratio for rivaroxaban: [(B+D)/(A+C)]

VOYAGER PAD indicates Efficacy and Safety of Rivaroxaban in Reducing the Risk of Major Thrombotic Vascular Events in Subjects With Symptomatic Peripheral Artery Disease Undergoing Peripheral Revascularization Procedures of the Lower Extremities.

Figure 1. Win ratio exploratory, prespecified analysis of the VOYAGER PAD efficacy outcomes.

Participants from the rivaroxaban arm were compared with participants in the placebo group at 3 years. The participant who experienced death first lost the event for his/her treatment arm. When death did not occur, the participant who experienced the first nonfatal end point lost the event. VOYAGER PAD indicates Efficacy and Safety of Rivaroxaban in Reducing the Risk of Major Thrombotic Vascular Events in Subjects With Symptomatic Peripheral Artery Disease Undergoing Peripheral Revascularization Procedures of the Lower Extremities.

Regarding the overall net clinical benefit, rivaroxaban was superior to placebo at 3 years when efficacy end points included either cardiovascular or all‐cause death, and safety end points included either intracranial/fatal bleeding or TIMI major bleeding (Figure 2). Similar rates were observed regardless of the type of death (cardiovascular or all‐cause) or type of bleed (intracranial/fatal or TIMI major). The first prespecified net clinical outcome used the primary efficacy outcome components (including cardiovascular death) and the safety end points of intracranial or fatal bleeding. At 3 years, participants on rivaroxaban experienced a cardiovascular death, ischemic stroke, myocardial infarction, acute limb ischemia, major amputation, or intracranial/fatal bleeding 14.1% of the time, versus 17.9% (cumulative incidences at 3 years), respectively (HR, 0.76 [95% CI, 0.66–0.87]; P<0.001; number needed to treat, 26). A second definition including a broader safety outcome—all TIMI major bleeding—yielded a similar result: 15.4% of participants assigned rivaroxaban and 18.5% participants assigned placebo experienced an end point event (HR, 0.80 [95% CI, 0.70–0.92, P=0.0012; number needed to treat, 32). In the third net clinical outcome, using the primary efficacy end points and all‐cause death (instead of cardiovascular death), and the safety outcome of intracranial/fatal bleeding, cumulative incidences at 3 years were 15.0% for rivaroxaban and 19.2% for placebo (HR, 0.75 [95% CI, 0.66–0.86]; P<0.001; number needed to treat, 24). In the fourth net clinical outcome, the primary efficacy outcome included all‐cause death, and the safety outcome was TIMI major bleeding, and the cumulative incidences at 3 years were 16.3% and 19.7% for rivaroxaban and placebo, respectively (HR, 0.79 [95% CI, 0.70–0.90]; P<0.001; number needed to treat, 29). Despite different components for each composite end point, the correlations of log HR among the primary efficacy end point and the 4 net clinical benefit outcomes are highly correlated (correlation coefficients are >0.9 with 95% CIs excluding 0; Table S1).

Figure 2. Net clinical benefit of the VOYAGER PAD efficacy and safety outcomes.

Outcomes using various combined efficacy and safety end points at 3 years, comparing rivaroxaban to placebo. All analyses included myocardial infarction, ischemic stroke, acute limb ischemia, and major amputation. A, Included cardiovascular death and intracranial or fatal bleeding. B, Included cardiovascular death and TIMI major bleeding. C, Included all‐cause death and intracranial or fatal bleeding. D, Included all‐cause death and TIMI major bleeding. ASA indicates acetylsalicylic acid (aspirin); TIMI indicates thrombolysis in myocardial infarction; and VOYAGER PAD, Efficacy and Safety of Rivaroxaban in Reducing the Risk of Major Thrombotic Vascular Events in Subjects With Symptomatic Peripheral Artery Disease Undergoing Peripheral Revascularization Procedures of the Lower Extremities.

Discussion

VOYAGER PAD demonstrated that vascular‐dose rivaroxaban significantly reduced the primary composite end point consisting of acute limb ischemia, major amputation of a vascular pathogenesis, myocardial infarction, ischemic stroke, and cardiovascular death in patients with symptomatic PAD who have undergone successful lower‐extremity revascularization. ² Results of the primary time‐to‐first‐event analysis were driven by acute limb ischemia, which was the most frequent component of the composite and the only one that was significantly reduced as an individual event. As such, a logical question is whether rivaroxaban reduced the other ischemic end points. Because the study was not powered for the individual components of the composite end point, prespecified exploratory analyses were performed to further ascertain the relative impact of fatal and nonfatal events on the composite outcome using different hierarchies, as well as to better understand the overall benefit–risk profile. The results show consistent superiority and net clinical benefit of rivaroxaban in this population when evaluated as (1) a ranked hierarchy of outcomes using the global rank method, with cardiovascular death as the worst event; (2) a win ratio approach ranking cardiovascular death as the worst event; and (3) net outcomes including death, ischemic events, and major bleeding.

The use of a primary composite outcome, as was done in VOYAGER PAD, is common in cardiovascular clinical trials, with most trials including 3 to 5 distinct components that may vary in clinical significance. ⁷ One of the main reasons for this approach is pragmatic: An end point with >1 component allows for higher event rates, thereby requiring smaller sample sizes or shorter follow‐up. ¹⁹ To work successfully, an essential consideration in using a composite outcome is that the components must be “associated with the primary objective.” ²⁰ Even when this occurs, however, an important criticism is the potential for a composite outcome to convey an exaggerated view of how well an intervention actually works. ²¹ VOYAGER PAD clearly illustrates this issue: The overall results were driven by the most frequent event (acute limb ischemia), with limited power to assess any individual event, particularly the least frequent event. Furthermore, there is potential for total event rates to be skewed by death, which is a definitive terminal event. Analytic approaches including ranking, weighting, and net clinical outcomes can help clinicians to better understand the strength of results, the relative impact of fatal and nonfatal events, and overall risk–benefit for patients.

The global rank and win ratio methods belong to a larger group of tests called generalized pairwise comparisons, first introduced in 1999 ¹⁸ and meant to help elucidate the results of multiple trial end points. ²² The global rank method uses a predetermined hierarchy of end points based on clinical importance. ⁸ An advantage of this method is that it more adeptly deals with component end points that vary in directionality (ie, decreased risk of 1 component but increased risk of another), since all end points are included in the final statistic. An example in which the global rank method can optimize clinical trials is pediatric patients with congenital heart disease, given the challenges of patient enrollment²²¹ Additionally, more important (predetermined) clinical events have greater influence on the final statistic. One limitation of this approach is that a true survival curve is not generated for the composite, although survival estimates can be shown for each component of the end point. ⁸ In this analysis of the VOYAGER outcomes, participants on low‐dose rivaroxaban with daily antiplatelet therapy had fewer or later events compared to their placebo counterparts when events were ranked (a priori) in order of severity.

Like the global rank method, the win ratio method prioritizes each end point. Outcomes from individuals based on treatment arm are compared. Once all comparisons have been made, a win ratio can be calculated: the ratio of wins to losses for the treatment arm. ¹⁸ Alternatively, a loss ratio (the reciprocal of the win ratio) may be calculated and can be compared with the hazard ratio from the Cox proportional hazards model. ²³ In this analysis, the loss ratio was nearly identical to the HR (0.85 [95% CI, 0.76–0.96]) in the primary study. ² Factors that may influence the win ratio include the duration of follow‐up, the censoring distribution, and competing risks. ²⁴ Since its publication in 2012, ¹³ many cardiovascular clinical trials have used the win ratio as an analysis strategy (a priori and exploratory) to better understand how deaths, clinical events, and patient‐reported outcomes contribute to composite trial end points. ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ The win ratio analysis has also been proposed for critical limb‐threatening ischemia trials given its ability to incorporate non–time‐to‐event variables in the composite end point, such as wound healing and pain resolution. A main caveat, though, is the importance of establishing a clinically meaningful treatment effect up front, as a pathophysiologic measure with a small treatment but insignificant clinical effect can dramatically increase statistical power and thus sway the overall study results. ³⁴ The win ratio analysis for VOYAGER PAD revealed that participants on rivaroxaban had fewer or later cardiovascular deaths and nonfatal ischemic events than their placebo counterparts.

Since the global rank and win ratio methods only consider a treatment's effectiveness, a net clinical benefit analysis was used to better understand the trade‐off between treatment risk and efficacy. This analysis uses an “exchange rate,” or relative value of benefit and harm, to place benefits and harms on the same scale. ¹⁶ Net clinical benefit analyses have previously been used, not surprisingly, in examining anticoagulant ³⁶ , ³⁷ , ³⁸ and antiplatelet therapies. ³⁹ In fact, it has been shown that net clinical benefit and win ratio results can complement each other to better demonstrate treatment effect. ¹² A prespecified net clinical benefit analysis was performed in the COMPASS (Cardiovascular Outcomes for People Using Anticoagulation Strategies) trial, which examined rivaroxaban 2.5 mg twice daily plus low‐dose aspirin versus aspirin in a population with known, stable cardiovascular disease. ⁴⁰ The analysis revealed that participants in the combined rivaroxaban–aspirin arm experienced a better net clinical benefit, which was driven by decreased stroke and cardiovascular death due to the low frequency of major bleeding events. The net clinical benefit analysis performed here for VOYAGER PAD supports similar findings: that participants on rivaroxaban with daily antiplatelet therapy had overall better net outcomes due to more frequent ischemic events compared with bleeding events, but in a higher‐risk population than in the COMPASS trial.

Limitations

Although prespecified, this was an exploratory analysis, so results are hypothesis generating. Additionally, for the ranked analyses, cardiovascular death (as a terminal event) was assigned the most serious event, but some patients may consider shorter survival with fewer nonfatal events (ie, preferring better quality of life over longer survival) the priority. ²² Similarly, there may be disagreement regarding the chosen hierarchy of nondeath events (eg, some may consider a myocardial infarction a worse event than stroke). An additional limitation is that results reflect the weighting as determined prior to database lock and unblinding and informed by scientific opinion as well as event‐related disability and death. Other approaches to ranking and weighting could have provided variable results. In the win ratio analysis, a majority of comparisons resulted in a tie, meaning a potential overestimation of treatment effect. ²² In the future, a win odds method (which takes into account the proportion of ties) may be used. ²⁴

Conclusions

This exploratory analysis of VOYAGER PAD revealed consistent superiority and net clinical benefit of rivaroxaban in a symptomatic PAD population after successful lower‐extremity revascularization when considered as a global ranked hierarchy of outcomes and a win ratio approach, assigning cardiovascular death as the worst event, as well as net clinical outcome incorporating death, ischemic events, and bleeding events. These data further support the use of vascular‐dose rivaroxaban in addition to aspirin to decrease cardiovascular death and ischemic event rates in this patient population.

Sources of Funding

VOYAGER PAD was supported by a grant from Bayer AG and Janssen Research & Development, LLC.

Disclosures

Drs Hogan, Nehler, Berkowitz, Bonaca, and Capell all receive salary support from CPC, a nonprofit academic research organization affiliated with the University of Colorado, that receives or has received research grant/consulting funding between July 2021 and July 2023 from the following organizations: Abbot Laboratories, Agios Pharmaceuticals, Inc., Alexion Pharma Godo Kaisha, Amgen Inc., Anthos Therapeutics, Inc., ARCA Biopharma, Inc., AstraZeneca Pharma India, AstraZeneca Pharmaceuticals LP, AstraZeneca UK Ltd, AstraZeneca, Produtos Farmaceuticos, Lda, Atentiv, LLC, Bayer, Bayer (Proprietary) Limited, Bayer Aktiengesellschaft, Bayer Pharma AG, Beth Israel Deaconess Medical Center, Better Therapeutics, Boston Clinical Research Institute, LLC, Bristol‐Myers Squibb, CellResearch Corporation Pte Ltd, Cleerly, Inc., Colorado Dept of Public Health and Environment, Cook Regentec LLC, CSL Behring LLC, Eidos Therapeutics, Inc., EPG Communication Holdings Ltd., Esperion Therapeutics, Inc, Faraday Pharmaceuticals, Inc., HeartFlow Inc, Insmed, Ionis Pharmaceuticals, IQVIA Inc., Janssen Pharmaceuticals, Inc, Janssen Research & Development, LLC, Janssen Scientific Affairs LLC, Lexicon Pharmaceuticals, Inc., LSG Corporation, MedImmune Limited, Medpace, Inc., Medscape, Merck Sharp & Dohme Corp., Nectero Medical Inc., Novartis Pharmaceuticals Corporation, Novo Nordisk, Osiris Therapeutics, Inc., Pfizer, PPD Development, L.P., Prothena Biosciences Limited, Regeneron, Regents of the University of Colorado (aka UCD), Sanifit Therapeutics S.A., Sanofi, Silence Therapeutics PLC, Stanford University, Stealth BioTherapeutics Inc., The Brigham and Women's Hospital, Inc., Thrombosis Research Institute, University of Colorado Denver, University of Pittsburgh, VarmX, WraSer, LLC. Dr Debus reports grants and personal fees from Bayer AG, institutional grants from Cook LTD, and grants from Terumo Aortic during the conduct of the study. Dr Nehler reports grants from Bayer and Janssen during the conduct of the study. Dr Anand reports personal fees from Bayer AG and personal fees from Janssen during the conduct of the study, and personal fees from Bayer AG and personal fees from Janssen outside the submitted work. Dr Anand holds an endowed chair in population health from the Heart and Stroke Foundation of Canada, and a Canada research chair in ethnicity and cardiovascular disease. Dr Patel receives grant support, advisory board fees, and consulting fees from Bayer. Dr Haskell reports funding support from Janssen Pharmaceuticals LLC during the conduct of the study, and funding support from Johnson & Johnson outside the submitted work. Dr Muehlhofer is employed by Bayer AG. Dr Bauersachs receives consulting fees and lecture fees from Bayer. Dr Berkowitz was employed as a clinical research physician by Bayer at the time VOYAGER PAD was performed. Dr Bauersachs reports personal fees from Bayer during the conduct of the study, and personal fees from Bristol Myers Squibb, Daiichi‐Sankyo, and Pfizer outside the submitted work. Dr Bonaca receives support from the AHA SFRN under award numbers 18SFRN3390085 (BWH‐DH SFRN Center) and 18SFRN33960262 (BWH‐DH Clinical Project). Dr Bonaca also reports stock in Medtronic and Pfizer.

Supporting information

Table S1

References [41, 42, 43]