Redefining VTE Trial Endpoints: How Precise Definitions Transform Research Outcomes

Can Precise Endpoint Definitions Transform VTE Prevention Trials?

The MARINER trial, which examined rivaroxaban for venous thromboembolism (VTE) prevention in medically ill patients post-hospital discharge, initially failed to meet its primary endpoint. However, a recent independent readjudication study offers compelling insights into how clinical trial endpoint design can significantly impact outcome interpretation. This comprehensive reanalysis demonstrates that more specific endpoint definitions may better capture treatment effects, even with fewer total events – a finding with broad implications for clinical trial methodology across cardiovascular medicine.

The original MARINER trial randomized 12,024 high-risk patients to receive either rivaroxaban (10 mg once daily, or 7.5 mg for those with reduced renal function) or placebo for 45 days following hospital discharge for medical illness. Patients had been hospitalized for heart failure, acute respiratory insufficiency, acute ischemic stroke, or acute infectious/inflammatory disease, with additional VTE risk factors. The primary efficacy outcome – a composite of symptomatic, nonfatal VTE and VTE-related death – showed a non-significant reduction with rivaroxaban (HR: 0.76; 95% CI: 0.52-1.1). Notably, the trial's definition of "VTE-related death" included not only documented fatal VTE events but also deaths where pulmonary embolism (PE) could not be excluded, which comprised approximately half of all death events. This inclusive approach was based on the assumption that VTE would be the predominant cause of mortality in this population – an assumption the researchers sought to examine critically through readjudication.

How Does Detailed Readjudication Redefine Mortality Classification?

The research team, led by investigators from CPC Clinical Research and the University of Colorado Anschutz Medical Campus, reexamined all 241 death events from the MARINER trial using more specific, standardized definitions for VTE-related death. Independent adjudicators blinded to treatment assignment and original adjudication results categorized deaths into seven prespecified causes: confirmed fatal VTE, VTE contributing to death, probable VTE death, possible VTE death, other cardiovascular death, non-cardiovascular death, and death of unknown etiology. This approach allowed for a systematic analysis of how different levels of endpoint specificity affect statistical outcomes. The investigators also assessed the quality and completeness of available information for each death event, categorizing the evidence as source documents, detailed narratives, or non-detailed/no information.

The readjudication results revealed striking insights into both the nature of events in the MARINER population and the methodological implications of endpoint design. Of the 112 deaths originally classified as "cannot rule out PE," only 9 (8%) were readjudicated as confirmed, probable, or possible VTE-related deaths. The majority were redistributed to deaths of unknown etiology (60%), other cardiovascular deaths (21%), or non-cardiovascular deaths (12%). When analyzing trial outcomes using the readjudicated causes of death, the hazard ratio for "confirmed VTE" (including nonfatal VTE, VTE contributing to death, and confirmed fatal VTE) was 0.46 (95% CI: 0.23-0.91), showing a more pronounced treatment effect than the original primary endpoint. This pattern persisted even when expanding the definition to include probable and possible VTE deaths, with HRs of 0.44 (95% CI: 0.23-0.88) and 0.53 (95% CI: 0.29-0.98), respectively. These findings suggest that the more specific endpoint definitions better captured the treatment effect of rivaroxaban on true VTE-related outcomes.

The readjudication effort also highlighted important challenges in determining cause of death in outpatient clinical trials. Deaths of unknown etiology constituted 30% of all deaths after readjudication, second only to non-cardiovascular deaths (32%). Importantly, the level of available evidence strongly influenced adjudication results – while specific death determinations were predominantly supported by source documents, 68.5% of deaths classified as unknown etiology had minimal or no information available. Further analysis revealed that for most of these cases (84%), there was no additional "knowable" information that could have been obtained, as deaths often occurred at home without witnesses or specific symptoms. This finding challenges the notion that high rates of undetermined deaths necessarily reflect poor operational quality in clinical trials, particularly in outpatient settings where circumstances surrounding death may be inherently difficult to document.

The cause of death in the MARINER population, when determinable, generally reflected the reason for the index hospitalization. Deaths among patients admitted for heart failure or stroke were predominantly cardiovascular in nature (53.2% and 40.8%, respectively), while those admitted for respiratory or inflammatory conditions more commonly died from non-cardiovascular causes (47.1% and 49.1%, respectively). This heterogeneity in the study population further underscores the importance of specific endpoint definitions that accurately capture the events most likely to be modified by the intervention under study.

Could Specific Endpoint Design Enhance Detection of Treatment Effects?

The investigators propose that designing clinical trial endpoints with greater specificity to the disease process and intervention mechanism may better preserve the ability to detect treatment effects, despite potentially reducing overall event rates. They suggest that standardized definitions, such as those recently published by the International Society on Thrombosis and Haemostasis for VTE-related death, should be employed across trials to ensure consistency and specificity. For deaths of unknown etiology, which remain an inevitable challenge in outpatient trials, the authors recommend reporting these separately and interpreting them in the context of all-cause mortality, rather than assuming they represent a specific cause of death.

This reanalysis of the MARINER trial has significant implications for clinical trial design and interpretation across cardiovascular medicine. The tension between selecting inclusive endpoints with higher occurrence rates versus more disease-specific endpoints with potentially greater responsiveness to intervention represents a fundamental challenge in trial methodology. As this study demonstrates, the choice of endpoint definition can substantially impact conclusions about treatment efficacy. For clinicians and researchers evaluating trial evidence, understanding these nuances is essential for accurate interpretation and application to patient care. Could more stringent endpoint definitions in other cardiovascular trials reveal treatment effects that were previously obscured by overly inclusive composite endpoints? How might regulatory agencies and trial sponsors balance the need for feasible trial execution with the scientific integrity of endpoint specificity? These questions merit careful consideration as the field continues to refine approaches to generating high-quality evidence for clinical decision-making.

What Limitations Must We Consider in This Reanalysis?

While this analysis provides valuable methodological insights, it is important to note its limitations as an exploratory, retrospective study. The readjudication relied solely on the original source documents obtained during the trial, without the ability to seek additional information from sites. The authors appropriately caution that the hazard ratios and confidence intervals presented are for illustrative purposes only and should not be interpreted as definitive outcomes from the MARINER trial. Nevertheless, the principles demonstrated – particularly regarding the impact of endpoint specificity on effect size and precision – have broad relevance for clinical trial design and interpretation across cardiovascular medicine and beyond.

Summary

The readjudication study of the MARINER trial, which initially showed non-significant results for rivaroxaban in VTE prevention, reveals that more specific endpoint definitions can better capture treatment effects. The reanalysis of 241 death events using standardized definitions showed that only 8% of deaths previously classified as "cannot rule out PE" were actually VTE-related. Using more precise endpoints, the study found a more pronounced treatment effect for rivaroxaban (HR: 0.46; 95% CI: 0.23-0.91). The study also highlighted challenges in determining cause of death in outpatient trials, with 30% of deaths remaining of unknown etiology. These findings suggest that more specific endpoint definitions in clinical trials may better demonstrate treatment efficacy, even with fewer total events.