Recommended Primary Outcomes for Clinical Trials Evaluating Hemostatic Blood Products and Agents in Patients with Bleeding: Proceedings of a National Heart Lung and Blood Institute and United States Department of Defense Consensus Conference

Introduction

The treatment of conditions associated with bleeding is a common challenge in clinical medicine, and, accordingly, the focus of a substantial volume of clinical research. Interventions may target supportive care to manage the bleeding and prevent complications, or those aimed at controlling the bleeding source.¹ Across multiple clinical settings, there is an increased risk of morbidity and mortality if bleeding occurs at a rapid rate, results in copious volume loss, or occurs in a critical organ system. The importance of timelines is reiterated by data following traumatic injury which indicates the median time to death due to hemorrhage is 2 hr.² Alternatively, in the setting of intracranial hemorrhage, mortality and residual disability are highly correlated with the volume of the lesion.^{3, 4}

High-quality evidence guiding optimal transfusion and other supportive therapies to reduce bleeding are needed to improve outcomes for patients either with severe bleeding or hemostatic disorders that are associated with poor outcomes. Alongside challenges in performing high-quality clinical trials in patient populations that are at risk of bleeding or who are actively bleeding, the interpretation of research evaluating hemostatic agents has been limited by inconsistency in the choice of primary trial outcomes. This lack of standardization of primary endpoints or outcomes decreases the ability of clinicians to assess the validity of endpoints, compare research results across studies, impairs meta-analytic efforts, and ultimately delays the translation of research results into clinical practice.

To address this challenge, an international panel of experts was convened by the National Heart Lung and Blood Institute (NHLBI) and the United States (US) Department of Defense (DoD) on September 23^rd and 24^th 2019 to develop expert opinion, consensus-based recommendations for primary clinical trial outcomes for pivotal trials in pediatric and adult patients with six categories in various clinical settings. This publication documents the conference proceedings from the workshop funded by the National Heart Lung and Blood Institute (NHLBI) and the United States (US) Department of Defense (DoD) that consolidated expert opinion regarding clinically meaningful outcomes across a wide range of disciplines to provide guidance for outcomes of future trials of hemostatic products and agents for patients with active bleeding.

Methods

An international working-group of experts in the clinical treatment of bleeding, hemostasis research, and clinical research methodology was assembled. Members were identified based on current and prior research activities, relevant expertise, and roles within relevant organizations to ensure diversity of perspectives across clinical and scientific domains. The executive committee of the workshop in collaboration with the subgroup leaders finalized the list of members for each subgroup. Members chosen were predominantly from the US with a few exceptions due to the limitation in funding for travel. The members of the working group and their affiliations are listed in the Appendix.

The working group was organized into six subgroups, each addressing the assessment of bleeding in a clinical setting: 1) traumatic injury; 2) intracranial hemorrhage; 3) cardiac surgery (including mechanical circulatory support); 4) gastrointestinal hemorrhage; 5) inherited bleeding disorders; and 6) hypoproliferative thrombocytopenia. Each subgroup was balanced to include a breadth of clinical perspectives, clinical and translational research expertise, clinical trialists, and at least one epidemiologist or statistician.

The recommendations were intended to establish valid, clinically relevant outcomes for late-stage human trials evaluating hemostatic interventions, devices, or agents and to support alignment among investigators, sponsors, and regulators. The scope of the recommendations was limited to primary outcomes in pivotal clinical trials in adults. Some subgroups included pediatric recommendations.

For each bleeding category, working-group members considered prior experience with previously published clinical trial outcomes in their respective fields including strengths and weaknesses of each possible outcome. Particular note was taken of prior work in which the discrepancy between regulatory guidance and human physiology complicated the choice of primary endpoints and contributed to difficulty in clinical interpretation of results.⁵ The selection of primary outcomes was guided by expected or demonstrated sensitivity to beneficial treatment effects, clinical and logistical feasibility, and broad applicability within each bleeding category. Most importantly, the primary outcomes were chosen based on their merits as clinically meaningful endpoints in the opinion of the expert panel. As a result, formal literature searches for each subgroup were not performed since clinical relevance or meaningfulness is qualitative and not quantitative. A literature search of what outcomes have been used before would not have achieved the goal of developing consensus statement for recommendations of outcomes that are of clinical value. Each subgroup did reference the literature to provide biologic rationale for expert opinion recommendations when available.

Each subgroup’s recommendations were developed based on structured discussions through a series of teleconferences over 12 months and augmented with in-person meetings, to create preliminary recommendations for outcomes to consider. The subgroups refined these recommendations based on additional literature review, consideration of clinical relevance, feasibility, and biologic rationale, to create a list of outcomes appropriate for primary and secondary trial outcomes for pivotal clinical trials.

In September 2019, members of the working group met in Bethesda, MD, along with representatives of funding and regulatory agencies and additional meeting registrants, including members of industry, to discuss and finalize recommendations. During the two-day meeting recommendations from all subgroups were discussed and modified. The simultaneous integration of multiple endpoints in a single analysis, using innovative analytic techniques were considered as well (e.g., a gatekeeping approach with multiple endpoints).⁶ Finalization of the recommendations for outcomes occurred in the symposium breakout sessions, followed by a public review by all attendees. All expert opinion recommendations were voted on by subgroup members with at least 85% indicating agreement.

Summary of Recommendations and Rationale

Synopses of the recommendations for primary clinical trial outcomes and the associated rationales are provided below and summarized in Table 1.

Table 1.

Recommendations for primary outcomes for trials examining hemostatic blood products and agents

Clinical Condition	Primary Outcome Recommendations
Traumatic Injury (excluding isolated traumatic brain injury and burn injury)	For patients with traumatic injury, a 3 to 6-hour all-cause mortality is recommended as a primary outcome for hemostatic interventions, with robust evaluation of late safety-related outcomes In children with life-threatening traumatic hemorrhage, a primary outcome of death at either 6 or 24 hours is recommended.
Intracranial Hemorrhage	For patients with spontaneous or traumatic intracranial hemorrhage the recommended optimal primary outcome is a global patient-centered clinical outcome scale measured between 30–180 days after the event. For studies that may lack sufficient power for this, a combined clinical and radiographic endpoint (such as poor outcome associated with ICH expansion) is an acceptable second choice.
Cardiac Surgery/Mechanical Circulatory Support	For patients with bleeding secondary to cardiothoracic surgery, a primary efficacy endpoint of total allogeneic blood products (units vs ml/kg) administered intraoperatively and postoperatively to day 5 or hospital discharge is recommended. For clinical trials of patients on ECMO, the recommended primary outcome is a 5-point ordinal score of both thrombosis and bleeding severity adapted from the consensus definitions in Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. For VAD studies, the recommended primary endpoint is freedom from disabling stroke (CTCAE ≥ Gr3 through 180 days.
Gastrointestinal Hemorrhage	For patients with acute gastrointestinal bleeding, the recommended primary outcome is a composite measure consisting of further bleeding (including persistent bleeding and rebleeding with primary assessment in 7 days),42 and cause-specific or bleed-related mortality at 30–42 days.
Inherited Bleeding Disorders	For the treatment of acute bleeding or for the prevention of surgical bleeding the recommended primary outcome is a composite scoring system incorporating the number of infusions required, rating on a clinical scale, and need for re-intervention For the prevention of bleeding events the recommended primary outcome is an annualized bleeding rate, quantifying the number of bleeding events over a period of time. For the long term and overall impact of bleeding, which is the ultimate goal in patients with inherited bleeding disorders, the recommended primary outcome is a composite of joint function (measured via an objective clinical scale or a patient reported functional score),50 chronic pain (measured via a validated scale), and HRQoL (both generic and specific)46 as proposed in a core outcome set for hemophilia gene therapy trials.
Hypoproliferative Thrombocytopenia	For patients with bleeding secondary to hypoproliferative thrombocytopenia, the recommended primary outcome for trials that examine the efficacy of blood products or hemostatic agents is the WHO Bleeding Assessment Score. Specifically, we recommend a WHO Grade of 2 or greater as a primary outcome in clinical trials of platelet threshold and efficacy.

Traumatic Injury

The trauma subgroup focused on hemostatic outcomes of clinical trials for patients with life-threatening bleeding excluding isolated traumatic brain injury (TBI) and burns. Historically, 30-day survival has been the standard primary outcome for all injury trials,^{7, 8} but the biological rationale for this arbitrary endpoint is not consistent with the physiological effects of effective hemostatic interventions.⁹ Assessing mortality at later time points results in more deaths occurring in both the experimental and control groups, for reasons other than lack of hemostasis, more commonly due to TBI and inflammatory complications.^{2, 9} Early, effective hemostatic interventions may result in significant differences within hours of injury, however, the same absolute mortality difference that was statistically significant at 3 to 6 hours, may lose significance at 24 hours or 30 days due to the decrease in statistical power and dilution of the target outcome (hemorrhagic deaths) by other death causes not directly treatable by hemostatic interventions (e.g., TBI, multiple organ failure).^{5, 9, 10}

For patients with traumatic injury, a 3 to 6-hour all-cause mortality is recommended as a primary outcome for hemostatic interventions, with robust evaluation of late safety-related outcomes ¹¹. This recommendation is supported by multiple RCTs and large observational studies that indicate that the vast majority of hemorrhage-related fatalities occur within 3 to 6 hours after injury.^{2, 5 12–20} The recommendation of using all-cause mortality is justified by the known difficulties in objectively ascribing a primary cause of death in patients with combined hemorrhage and TBI. While conceptually appealing, a standard, objective definition of hemorrhagic only death may be impossible given that its major competing risk, TBI, occurs along with hemorrhage up to 38% of the time.^{5, 9} Exclusion of isolated TBI patients based on physical exam alone may be difficult, if not impossible, before randomization, especially in the pre-hospital and very early hospital setting.^{2, 5, 9, 12, 13, 21–23}

In children with life-threatening traumatic hemorrhage, a primary outcome of death at either 6 or 24 hours is recommended ¹¹. In a recent prospective observational study of 208 children younger than 16 years (median 10 years), the median time to death from hemorrhage was 2.9 hours (personal communication, Spinella PC, December 11, 2019). The percent of the deaths that occurred from hemorrhage was 70% by 6 hours and 90% by 24 hours. At both 6 and 24 hours after hospital admission, 60% of the deaths were due to hemorrhage, with the other 40% from CNS injury. Further work is needed to define the primary cause and timing of death based on age groups (18–12, 11–6, and < 6 years).

The FDA has agreed with both above adult and pediatric recommended primary outcomes in recently submitted pre-IND applications for trials comparing whole blood to blood components in both populations (personal communication, Holcomb JB and Spinella PC, March 29, 2021).

Intracranial Hemorrhage

The intracranial hemorrhage subgroup focused on hemostatic outcomes for clinical trials in patients with spontaneous and traumatic intracranial hemorrhage (ICH). ICH is the most devastating form of bleeding that can occur in the human body, with the highest rate of mortality and residual disability.³

Conventional outcomes for hemostatic therapies for intracranial bleeding to date have focused on measurements of hematoma volume – almost universally the change in hemorrhage volume at 24 hours – and global functional outcome at 90 days. Fortunately, intracranial bleeding is easy to quantify with extreme precision using computed tomographic (CT) scanning, allowing for accurate and valid measurement of blood volume to the nearest milliliter.^{24, 25}

For patients with spontaneous or traumatic intracranial hemorrhage the recommended optimal primary outcome is a global patient-centered clinical outcome scale measured between 30–180 days after the event. For studies that may lack sufficient power for this, a combined clinical and radiographic endpoint (such as poor outcome associated with ICH expansion) is an acceptable second choice. Preferred outcome scales include Modified Rankin Scale, Extended Glasgow Outcome Scale, or a similar patient-centered outcome measure that is relevant for the disease entity. The high degree of correlation between final hemorrhage volume, mortality, and functional outcome makes hemorrhage volume or growth a reasonable endpoint for trials of hemostatic therapies when there is insufficient power to use patient-centered outcome scales in the primary outcome.^24–28

Cardiac Surgery/Mechanical Circulatory Support

The cardiac surgery subgroup considered primary outcomes for clinical trials of hemostatic therapies for bleeding associated with cardiac surgery and with the use of mechanical circulatory support (MCS), primarily extracorporeal membrane oxygenation (ECMO) and ventricular assist devices (VADs). For MCS, however, it was noted that both hemorrhage and thrombotic events are clinically important with the risk of thromboembolic events limiting the use of some hemostatic treatments; the MCS subgroup thus also considered thromboembolic endpoints. The lack of consensus definitions for clinically important bleeding and thrombosis makes optimal endpoints for clinical trials of hemostatic strategies in cardiac surgery and MCS challenging.

Conventional primary endpoints for cardiac surgery trials include the volume of allogeneic transfusions,²⁹ time to anastomotic hemostasis,³⁰ postoperative drainage tube/chest tube output,³¹ the need for reoperation,³¹ or combinations of these metrics (i.e., bleeding scores).^{32, 33} Limitations of these approaches, with attendant challenges in interpretation, include heterogeneity in bleeding risk among different surgical procedures and variability in surgical technique, thresholds for reoperation, and transfusion practices.

For patients with bleeding secondary to cardiothoracic surgery, a primary efficacy endpoint of total allogeneic blood products (units vs ml/kg) administered intraoperatively and postoperatively to day 5 or hospital discharge is recommended. To control extraneous variation across centers and/or providers, a pre-specified bleeding management protocol/algorithm is critical.^{29, 34–36}

Conventional endpoints for ECMO/VAD trials are derived from INTERMACS,³⁷ ELSO,³⁸ and other registries’ definitions. Previously utilized endpoints are variable and include survival, transfusion requirements, clinically important bleeding, stroke, and thrombotic complications.^{39, 40} Importantly, the optimal hemostatic strategy in the setting of MCS requires a balance between the risks of hemorrhage and thrombosis; any primary outcome that fails to take both into account risks yielding clinically inappropriate results.

For clinical trials of patients on ECMO, the recommended primary outcome is a 5-point ordinal score of both thrombosis and bleeding severity adapted from the consensus definitions in Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. The score should be applied during the initial 10 days of treatment, as patient attrition often occurs over this period; the initial resuscitation phase (24–48 hours) and late maintenance phase (>24–48 hours) should also be evaluated.. Existing ELSO definitions of bleeding or thrombosis should serve as individual safety points.³⁷

For VAD studies, the recommended primary endpoint is freedom from disabling stroke (CTCAE ≥ Gr3) through 180 days. Assessments should be obtained through Day 2, < 30 days and ≥ 30 days. Recommended primary safety endpoints include thrombosis and bleeding complications defined by INTERMACS.³⁷ Thrombotic/bleeding rates may vary over time, so both early (<30 days) and late rates (≥30 days) should be assessed. The 5-point ordinal bleeding/thrombosis scale proposed for ECMO is also feasible to validate in the VAD population.

Gastrointestinal Hemorrhage

Endpoints used in the assessment of treatments for gastrointestinal hemorrhage have included initial or immediate hemostasis based on visual examination. New technologies (e.g., Doppler probe) may be useful in the future to better define when treatments have achieved more durable initial hemostasis at index endoscopy.⁴¹ Definitions for these have varied or are based on technology with limited availability. Furthermore, these immediate endpoints only partially determine subsequent longer-term outcomes, such as patient morbidity and mortality.

The primary outcome we recommended for all types of acute GI bleeding (UGI, small bowel, & colon) was a composite outcome of the following: further bleeding after randomization leading to RBC transfusion, urgent intervention (repeat endoscopy; surgery or interventional radiology) or death. The time period to assess this endpoint was suggested to be 30 days. Persistent bleeding may be defined as the inability to achieve hemostasis after 3 to 5 minutes of observation following what is felt by the operator to be the optimal attempt at endoscopic hemostasis based on visual cues using the studied modality. A proposed consensus definition for rebleeding is to include: 1) the post-intervention development of hematemesis or bloody nasogastric (NG) aspirate > 6 hours after endoscopy, 2) melena after normalization of stool color,3) hematochezia after normalization of stool color or after melena, 4) the development of tachycardia (heart rate ≥ 110 beats per minute) or hypotension (systolic blood pressure ≤ 90 mm Hg) after ≥ 1 hour of hemodynamic stability (i.e. no tachycardia or hypotension) in the absence of an alternative explanation, 5) a hemoglobin drop of ≥ 2 g /dl after two consecutive stable hemoglobin values (< 0.5 g /dl decrease) ≥ 3 hours apart, 6) the presence of tachycardia or hypotension that does not resolve within 8 hours after index endoscopy despite appropriate hemodynamic resuscitation in the absence of an alternative explanation and associated with persistent melena or hematochezia, and 7) a persistently dropping hemoglobin of > 3 g /dl in 24 hours associated with persistent melena or hematochezia.⁴² Mortality is chosen as part of a composite outcome measure owing to its clinical importance and because its direction and incremental decrease evolve with therapy in the same general direction and amplitude (in very general terms) as those of persistent bleeding and rebleeding. The time horizon to be covered by all bleeding-related mortality needs to take into consideration the varying natural history of GI bleeding etiologies, especially non-variceal (30 days) ^{42, 43} versus variceal (usually measured in 42-day cumulative mortality rates).⁴⁴

Inherited Bleeding Disorders

The subgroup considered outcomes for patients with hemophilia as a model since they had been comprehensively studied. They are usually grouped into 6 domains: 1) bleeding and hemostasis; 2) joint health; 3) co-morbidities and mortality; 4) overall physical functioning and participation; 5) health-related quality of life (HRQoL); and 6) cost and resource use.^{45, 46}

For the treatment of acute bleeding or for the prevention of surgical bleeding the recommended primary outcome is a composite scoring system^{45, 47} incorporating the number of infusions required, rating on a clinical scale, and need for re-intervention ⁴⁸. These outcomes have been successfully used to evaluate replacement therapy over several decades and recommended by the International Society of Thrombosis and Hemostasis Scientific Subcommittee on Factor VIII and IX.⁴⁷

For the prevention of bleeding events the recommended primary outcome is an annualized bleeding rate, quantifying the number of bleeding events over a period of time.⁴⁵ There is a body of evidence linking the bleeding rate to the progression of joint disease as measured by joint scores.^{48, 49}

For the long term and overall impact of bleeding, which is the ultimate goal in patients with inherited bleeding disorders, the recommended primary outcome is a composite of joint function (measured via an objective clinical scale or a patient reported functional score),⁵⁰ chronic pain (measured via a validated scale),⁵¹ and HRQoL (both generic and specific)⁴⁶ as proposed in a core outcome set for hemophilia gene therapy trials.^{48, 52}

Hypoproliferative Thrombocytopenia

Bleeding is common in patients with bone marrow failure, and an important risk factor is severe thrombocytopenia; the strength of this relationship might vary depending on the clinical situation.^{53, 54} Conventional endpoints for hemostatic therapies for hypoproliferative thrombocytopenia have included platelet count increments as a surrogate marker of bleeding risk and scoring systems for clinical bleeding. For patients with bleeding secondary to hypoproliferative thrombocytopenia, the recommended primary outcome for trials that examine the efficacy of blood products or hemostatic agents is the WHO Bleeding Assessment Score. Specifically, we recommend a WHO Grade of 2 or greater as a primary outcome in clinical trials of platelet threshold and efficacy.^55–57

Limitations to the use of the WHO Bleeding Assessment Score for data collection and analysis of bleeding have been recognized.^{58, 59} WHO Grade 2 bleeding, the most common grade, includes bleeding into the skin which is difficult to quantify and daily changes may be subjective, whereas Grade 2 mucosal bleeding may be more readily evaluated. Widespread endothelial and mucosal damage occur in patients with hematologic malignancy as a result of disease, infection, and therapy. Whether visible mucosal and cutaneous bleeding represent “window” into vascular and hemostatic health or merely the manifestation of mild and clinically insignificant blood loss is not known. The extent or severity of microvascular bleeding actually may not be fully captured by examination of the skin or mouth. As a composite outcome, WHO Grade 2 or greater bleeding includes Grades 3 (gross blood loss) and Grade 4 (debilitating blood loss), but the incidences of these clinically more meaningful events are low. There is uncertainty whether Grade 2 bleeding is predictive of more severe forms of Grades 3 and 4 bleeding, and whether isolated Grade 3 and 4 bleeding can occur without notable Grade 2 bleeding.^{60, 61}

Conclusions

Advancements in treatment for hemorrhage to improve patient outcomes, require high quality trials rigorously evaluating effects on valid, clinically relevant endpoints. Expert opinion from an international group of clinical experts and trialists has been leveraged to provide recommendations for primary outcomes for pivotal trials of hemostatic therapies. These recommendations attempt to harmonize the objectives of investigators, funding agencies, and regulatory agencies to facilitate the development and performance of high quality trials examining hemostatic blood products and agents in bleeding patients.