Proceedings of the First Curing Coma Campaign NIH Symposium: Challenging the Future of Research for Coma and Disorders of Consciousness

Current prognostication models for brain injury that focus only on motor response and overt evidence of cognition provide limited ability to predict outcomes, with the best models having only 70–80% discrimination [2]. Treatments for acute brain injury based on phenotype classification are also greatly limited, with harmful, uncertain, or inconsistent outcomes in clinical trials [3‐5]. Furthermore, most clinical trials have not successfully translated preclinical evidence of efficacy into successful therapies for humans. Thus, 60–85% of randomized controlled clinical trials (RCTs) conducted in the setting of acute brain injury have produced inconclusive results [3‐5].

The limitations in accuracy of prognostication and efficacy of therapy point to the larger issue of human biological heterogeneity. Converging research demonstrates the limitations of phenotype-driven detection, diagnosis, classification, treatment selection, and prognosis for patients with severe brain injury and DoC. These approaches may lump biologically heterogeneous patients into a single phenotypic category. Successfully addressing the goal of restoring consciousness and promoting meaningful recovery requires precisely characterizing states of vigilance that distinguish the underlying etiology or pathophysiologic mechanisms affecting consciousness and/or responsiveness. By describing, defining, and classifying DoC in a manner that is more closely aligned with biological mechanisms, we can design more effective RCTs and increase the likelihood of discovering therapies that promote recovery from coma.

Clinical evaluations that focus on motor function and overt cognition fail to capture subtle but highly meaningful differences between patients [6]. One example of this is patients with covert cognition or cognitive motor dissociation (CMD). Covert cognitive states may provide a stronger basis for treatment and prognostication, but they require advanced imaging or neurophysiology to be identified [7‐10].

Patients who share a specific mechanism or cluster of mechanisms that lead to the phenotype of coma are said to belong to an endotype. More generally, an endotype can be defined as a constellation of disease features anchored in a biological mechanism or pathway that is associated with a predictable disease trajectory and treatment response. An endotype is intrinsically more homogeneous than the larger phenotypically defined population. A wide variety of genetic and nongenetic determinants shape biological mechanisms that drive coma. Thus, multiple patients may be in a coma (the phenotype), but each might have a different disease trajectory, recovery potential, and treatment response depending on the underlying endotype. Each endotype will require a nuanced approach to diagnosis and treatment that would not be applicable on the phenotypic scale. The potential for endotypes to transform classification and treatment has been demonstrated in a number of domains, including cancer, asthma, and multiple sclerosis. Endotypes are a fundamental principle in personalized (precision) medicine because they may form the foundation for prognostic enrichment, in which participants are selected on the basis of the likelihood that they will experience a specific outcome, and for predictive enrichment, in which patients are selected on the basis of the likelihood that they will respond to an intervention [11‐20].

Heterogenous pathological states with different underlying mechanisms may be imprecisely distinguished by the clinician. This imprecision leads to errors in diagnosis, prognostication, and treatment, yielding poor patient outcomes.

The overall goal is to increase diagnostic precision, build accurate prognostic models, predict therapeutic responsiveness, and help create proof-of-concept clinical trials by leveraging accurate prognostic and predictive enrichment. This approach will allow smaller mechanistically homogenous groupings of individuals who will benefit substantially more from specific treatments than the larger heterogeneous classification based on clinical phenotypes. The expectation is that this approach would increase the likelihood of recovery when treating patients with coma and DoC.

There is a need to develop novel endotypes based on a mechanistic biological model of DoC that are not considered in current clinical treatment settings. These endotypes will inform therapeutic approaches based on personalized interventions for DoC to increase the rate of positive outcomes (i.e., regained consciousness, functional independence, improved patient-centered outcomes) among patients with DoC.

Biomarkers are considered any measurements, including chemical, physical, or biological (cellular, molecular), that relate to a biological system or the interaction between biological systems [35]. Currently available diagnostic biomarkers do not reliably detect preserved brain networks that may support the recovery of consciousness, and early prognostic biomarkers do not reliably predict outcomes [36, 37]. Because of these failings, prognostication is an uncertain art for clinicians treating DoC in the ICU and beyond. Yet many significant treatment decisions are made on the basis of uncertain prognoses. Incorrect prognosis may lead to premature withdrawal of life-sustaining therapy (WLST) or to the maintenance of individuals in a severely disabled state against their wishes. Indeed, many patients experiencing traumatic brain injury (TBI) die of WLST, despite only a portion of these patients demonstrating biomarkers associated with failure to recover consciousness [34, 38]. A quarter of patients with hypoxic-ischemic injury who died of WLST would have survived, and 16% might have made a functional recovery by the time of discharge, on the basis of their prognoses [39]. Early evidence has shown promise for magnetic resonance imaging (MRI) [7‐9] and EEG [10, 40, 41] patterns that may identify individuals with covert consciousness (i.e., purposeful response to stimuli otherwise missed by behavioral assessments) [42, 43]. Biomarkers of covert consciousness may improve prognostic capabilities for patients with coma [10]. Similarly, biomarkers of covert cortical processing (i.e., association cortex responses to language otherwise missed by behavioral assessments) also may predict long-term functional outcomes [44]. A combination of model-based and data-driven analyses may enhance mechanistic understanding of DoC [37].

There is a lack of reliable and reproducible biomarkers for patients with DoC to assist in prognostication and serve as targets for treatments applied in clinical practice, as well as those tested in RCTs, to promote recovery of consciousness.

The overall goal is to develop a mechanistic approach to methodology, phenotyping, outcomes studies, and trial design that is anchored in biomarkers that serve as diagnostic, prognostic, monitoring, or descriptive measures in patients with DoC.

The Curing Coma Campaign Coma Science Working Group has identified five potential therapeutic approaches: pharmacological, electromagnetic, mechanical, sensory, and regenerative. Each provides unique opportunities to identify prognostic or therapeutic biomarkers. Proposed biomarkers can be split into four domains: molecular and cellular, imaging, electrophysiology, and transcranial magnetic stimulation and electroencephalography (TMS-EEG).

There is a need for accurate, reliable, and reproducible biomarkers for patients with DoC that allow for more precise prognostication of recovery and provide end points for RCTs.

Proof-of-concept clinical trials of consciousness-promoting therapies have been performed mostly in the subacute and chronic stages of DoC [72]. However, early interventions may offer opportunities to promote recovery of consciousness during critical and acute care phases of treatment. Development of early interventions for use in the ICU may prevent premature WLST, facilitate self-expression, decrease ICU-related complications, and increase access to rehabilitative care. Conversely, early-stage clinical trials may also disprove promising theoretical interventions that have not yet been fully evaluated in clinical settings. Reasons for the failure of proof-of-concept clinical trials, particularly in acute care settings, include heterogeneity of the population (including the need for better early endotyping), inadequate power, use of nonspecific assessments, timing of the intervention, and failure to account for a variety of individual mitigating factors (e.g., underlying physiology, systemic illness, comorbidities, drug interactions, timing and dosage of therapies) [73]. Additionally, the impact of the context of health care delivery and treatments may present unique challenges for clinical trials regarding psychosocial aspects of care, cultural/religious beliefs, and availability of resources throughout recovery. Over the past 2 decades, coma research has focused largely on diagnosis and prognosis to understand mechanisms of spontaneous recovery of consciousness; yet there has been little attention to mechanisms of induced recovery of consciousness. Throughout the continuum of care, coma researchers must carefully consider these aspects and design more effective clinical trials of therapies at all stages of DoC. A recent search of ClinicalTrials.gov for trials that are registered as being “in progress” used the search terms “consciousness disorder, coma, or disorders of consciousness” and found 73 trials of DoC evaluating treatment, assessment, or prognosis techniques (ClinicalTrials.gov accessed September 6, 2020), with few evaluating DoC throughout the continuum of care.

Patients with DoC or coma in the ICU and subacute to chronic setting are often overlooked in clinical trials, undermining the development of reliable prognostic tools and treatments.

The overall goal is to offer accurate prognostication and treatment strategies that promote recovery of consciousness throughout the continuum of care to as many patients as possible.

Barriers for the development of strong proof-of-concept trials for curing coma science include lack of proper endotyping, lack of quality measurement tools for therapeutic response, need to account for spontaneous consciousness recovery, need for control groups, and lack of accounting for socioeconomic factors, local environment and resources, and cultural variation.

A key deliverable is to develop trials that define biomarkers, end points, and feasible treatment approaches that build on clearly defined patient-based phenotypes and endotypes. There is a need for the design of adaptive, exploratory interventional trials that inform development of larger pragmatic trials to identify interventions that support recovery of consciousness throughout the continuum of care and improve patient-centered outcomes. Simple master protocols could facilitate development and use of these deliverables. Once effective interventions are identified, further work by using an implementation science approach will be needed to integrate findings into routine clinical care.

Prognostication herewith refers to the ability of clinicians to provide surrogate decision-makers with an outlook regarding a patient’s future: the potential for recovery, the likelihood for disability, and the prospects for awakening from coma. Prognostication requires a synthesis of both data and clinical experience to inform critical decisions by surrogate decision-makers [101]. To this end, improving the accuracy of prognostication is an important aim in research focused on descriptions of clinical predictors, characterization of imaging findings, and the development of blood-based biomarkers [102].

Prognostication seeks to accurately predict whether a patient will awaken and whether they will achieve a satisfactory QOL consistent with their values and preferences. To achieve this, investigators must speak the same language as defined by CDEs. For example, “awakening” or “following commands” may be defined in different ways, as emphasized by recent work highlighting the CMD detected by EEG [10], fMRI [9], or multidimensional clinical assessments [9, 103]. Efforts to define CDEs specific to DoC are underway [1].

Yet improvements in prognostic precision are often hindered by medical decisions regarding patient triage, care limitations, or WLST. Decisions for WLST limit the unbiased study of the natural trajectories of recovery and the ability to compare these trajectories to those of patients who ultimately recover from coma [38, 39, 104]. WLST decisions after cardiac arrest are made within 72 h in as many as 63% of patients, despite findings that nearly one third of patients experience delayed awakening at a median of 93 h following arrest [39, 105]. The implications for this are grave: after cardiac arrest, early WLST results in an estimated excess of 2300 deaths in the United States each year; nearly two thirds of these patients might have had functional recovery if allowed to survive [39]. Yet significant variability exists in the rates of WLST worldwide [100]. After severe TBI, the rates of WLST vary broadly across centers [34], and extraneous factors unrelated to the medical conditions of patients may even play a role in these decisions, including geographic region, race, payment status, and other factors [106]. Ultimately, WLST decisions may result in a self-fulfilling prophecy in which false certainty about an outcome leads independently to an increase in the probability of that outcome [107].

Typically, prognostic considerations center on mortality and functional disability. However, satisfactory outcome after awakening from coma should ideally hinge on QOL, which is based on the individual patient’s values and preferences [108]. Existing research uses common functional outcome scales, such as the modified Rankin Scale or the Glasgow Outcome Scale, which focus on a limited set of disabilities, such as difficulties in walking or performing activities of daily living. These, by their nature, do not capture QOL, nor do they account for the preferences or values of patients or their families.

Recent work has highlighted cognitive or mental health disabilities that develop after critical illness, termed the “post-intensive-care syndrome” [109, 110], which may impact QOL. However, the majority of studies focused on cognitive or mental health outcome end points after critical illness have excluded patients with brain injuries [111]. Up to 20% of patients requiring care in a neuro-ICU subsequently exhibit cognitive impairment, and more than one third have at least moderate problems with anxiety or depression [112]. Those with significant cognitive or physical disability cannot be evaluated by using traditional neuropsychological testing. In a study of patient-reported outcomes after stroke, only 11.5% of those able to participate had a modified Rankin Scale score of 3 or greater [113]. The combined impacts of brain injury and critical illness on multidimensional outcomes, including physical functioning, cognition, and mental health, after recovery from coma have not been adequately studied.

Coma prognostication struggles to define numerous important parameters. Major gaps include defining the end points, ascertaining and accounting for WLST in coma research, defining meaningful outcomes for patients and clinicians, and establishing the validity and timing for implementation of prognostic tools. In addition, current prediction models contain high levels of uncertainty and imprecision due to a lack of complexity and multiple sources of bias. How to properly and effectively formulate and communicate prognostic judgments to families is also unknown. Physicians view prognostication as one of the most difficult parts of their profession [114, 115]; however, they receive little to no training in it, and no guidelines for how to communicate prognosis have been developed.

The overall goal is to develop tools that allow for accurate prognostication of well-defined, relevant end points for patients with DoC and to better understand the optimal methods for communicating prognosis to surrogate decision-makers.

There is a need to develop (1) accurate, reliable, and reproducible prognostic tools for patients with DoC and (2) empirically grounded interventions for high-quality prognostic communication that promotes shared decision-making between clinicians and decision-makers.

Survival of patients with coma who previously would have died is now common because of improvements in early resuscitation, interventions, and ICU management. TBI outcomes have improved through advances in prehospital triage, rehabilitation care, and compliance with published clinical treatment guidelines [144, 145]. In addition, survivors of cardiac arrest with hypoxic-ischemic brain injury are a new population for neurorehabilitation specialists to treat because of increased survival secondary to mainstream use of therapeutic hypothermia and through public access defibrillation efforts and compression-only resuscitation education [146, 147]. Evidence cited in the 2018 American Academy of Neurology, American Congress of Rehabilitation Medicine, and National Institute on Disability, Independent Living, and Rehabilitation Research practice guideline update on DoC makes clear that recovery from coma continues longer than previously believed, leading to meaningful functional improvement in a substantial minority of those affected [148]. The potential for good recovery in those with DoC supports the overarching goal of prospectively studying long-term recovery trajectories for this population.

Acute treatment of coma and other DoC appropriately focuses on short-term survival and recovery of gross neurological function but often fails to consider changes that occur during the post acute course, ultimately influencing long-term outcome. This can result in underestimation of prognosis and inappropriate treatment decisions. In the 1994 New England Journal of Medicine report on the persistent vegetative state, outcome only extended to 1 year after treatment [149]. A few facilities have developed specialized programs for patients with DoC, but admissions are constrained by fiscally driven gatekeeping policies, providing limited opportunity for systematic outcomes research on this population. Patients without detectable signs of recovery at the time of acute care discharge may be thought to be destined for unfavorable recovery and, thus, may not be referred to or accepted into rehabilitation centers. Many are routed to nursing facilities and home care settings that are ill-equipped to manage the complex medical and neurobehavioral consequences of the injury.

There is a need for evidence-based post acute care that spans a variety of settings. A study of early functional outcomes of 396 patients with traumatic DoC and without the ability to follow commands who were admitted to inpatient rehabilitation showed that 68% of patients were able to follow commands, 23% of patients had recovered from posttraumatic amnesia, and 7–14% of patients were independent on a range of self-care and mobility tasks (dependent on skill area assessed) prior to rehabilitation discharge. There is also an imbalance in funded research for patients with traumatic versus nontraumatic injuries, leading to a dearth of data regarding recovery in patients with nontraumatic DoC.

Active medical management by clinicians with expertise in brain injury reduces the rate of new complications in patients with DoC [150]. Acute care physiatry input can help initiate early DoC prognostication efforts, prevent complications, support early efforts to improve level of consciousness, and promote safe transitions of care [151].

Research conducted on long-term functional outcome after TBI through the TBI Model System Program found that by 5 years post injury, 74% of patients who were admitted to inpatient rehabilitation and unable to follow commands had regained this ability, approximately 20% were able to live without in-home supervision, and 19% were rated as capable of competitive employment [152]. Measurable gains in independence continued out to 10 years in a portion of the same cohort [153], particularly those who recovered command-following during their rehabilitation stay. The recovery trajectory for those with nontraumatic DoC differs but is less well understood. Pooled analysis of patients with prolonged nontraumatic vegetative state/unresponsive wakefulness syndrome suggests that 17% will recover consciousness at 6 months and an additional 7.5% beyond 6 months [154].

These findings indicate that there is substantial recovery over a long period of time for a sizable minority of individuals with prolonged DoC. Although surviving patients often have a high comorbidity burden, effectively managing comorbidity appears to improve long-term outcome [150, 151]. Emerging treatments in the field may enhance these positive long-term outcomes further [72]. This potential for long-term recovery needs to be considered when delivering acute care early after the injury as well as when triaging patients into the ideal care setting to support long-term recovery.

Although understanding of the long-term recovery trajectory of patients with DoC has expanded considerably over the last decade, predicting the recovery trajectory and functional outcome remains imprecise in individual patients, particularly in the very early period when many urgent treatment decisions are being made.

Long-term, researchers need to characterize the clinical trajectory of a broad range of DoC patients over a long time frame, taking patient heterogeneity and contextual factors into account. As a short-term goal, studying those with good and poor recovery after DoC may be a critical first step to help demonstrate and define the biomarkers, early treatments, and personal or environmental factors that represent, enhance, and discern potential for favorable outcome.

To be successful, longitudinal outcome research involving patients with DoC must address the following considerations:

Research involving patients with DoC is hampered by multiple factors, including the following:

Deliverables include (1) determining the natural history of recovery from DoC across the lifetime, (2) identifying accurate biomarkers and clinical predictors of favorable and unfavorable outcome, and (3) understanding the effectiveness of specific interventions on DoC recovery trajectory.

Large data sets, or databases, can help address multifaceted challenges of DoC research, which includes medical, scientific, technical, ethical, and social dimensions. Large data sets will help coma researchers by (1) fostering international consensus on issues such as definitions, causes, and confounders of DoC through implementation of CDEs [156]; (2) facilitating the design of operative diagnostic tools and meaningful outcome time points [157]; (3) enabling better research, including epidemiological studies that take advantage of between-center differences that explore processes of care and structural/organizational factors that may impact outcome and treatment heterogeneity effects [158‐160]; (4) organizing and federating an international network of coma scientists who share a common language and connection [100]; (5) improving patient care by standardizing results and improving expertise [161]; and (6) helping family members and caregivers of patients with coma provide better decision-making tools based on improved knowledge of acute- and post-acute-care trajectories [162]. A survey of the Scientific Advisory Council of the Curing Coma Campaign and the broader NCS found that clinicians and investigators’ interests focus on research involving patients affected by coma and acute DoC. These individuals generally have access to clinical data but lack access to other data types (e.g., anatomical data, clinical outcomes, safety).

The panelists surveyed current data sets to determine whether existing resources could be expanded or absorbed into a large international data set (see Table 1). Currently available resources are almost exclusively limited to patients with TBI or cardiac arrest. Available limited international data are disease specific (e.g., TBI, seizures), with relatively small numbers of patients with coma; moreover, these data pay little attention to the origins of DoC (e.g., toxic metabolic).

Limited large data sets of patients with DoC exist at this time. Gaps in available data sets include limited imaging and electrophysiologic data, lack of CDEs tailored to DoC, focus on primary disease states (i.e., TBI, intracerebral hemorrhage, etc.) rather than coma, lack of applicable biomarkers for coma and DoC, fragmented acute and chronic care data, and regional data restricted primarily to the United States and Europe. The key challenges to creating a large coma science data set are lack of uniform global data standards and collection across both substantive measures (i.e., what is collected) and data handling and storage mode.

The overall goal is to gain insights into mechanisms, predictors, and trajectory of recovery and lay the foundation for interventional trials by creating a large international data set of patients with DoC.

There are several major challenges for creating an international data set:

Deliverables include (1) creation of data dictionaries based on CDEs, common language and terminology, and data structure framework and (2) a large simple international database on patients with DoC that is built on CDEs.