The diagnostic and prognostic utility of electrophysiological responses to speech in post-traumatic coma and prolonged disorders of consciousness

Each day, ten individuals across the United Kingdom enter a coma as a result of a blow to the head (also known as a traumatic brain injury, or TBI). Patients in coma are unresponsive to external stimulation and breathe with support from a mechanical ventilator. While 60% of people survive coma, one third of survivors will develop a prolonged disorder of consciousness (PDOC), such as the vegetative state, in which they appear to be awake but show no signs of being aware of themselves or of their environments. However, if a patient with a PDOC is able to move in response to verbal commands (e.g. "Give me a thumbs-up") they are considered to be at least minimally conscious. Once this ability has been identified, rehabilitation can focus on training the individual to use this action to communicate (e.g. "Thumbs-up for yes").

Current methods for predicting the level of recovery after coma are inadequate. Furthermore, a 40% misdiagnosis rate occurs with current methods for diagnosing the level of awareness of those patients who develop a PDOC. Electroencephalography, or EEG, is a portable form of brain-imaging that records the tiny electrical signals generated by the brain via a series of electrodes placed on the scalp. By measuring the brain's response to external stimuli, it may be possible to improve the accuracy of prognosis in coma (i.e. predicted level of recovery) and diagnosis in PDOC.

To move in response to a verbal command, a series of interactions must occur between the relatively distant brain regions involved in processing the features of speech and those brain regions involved in the control of movement. When you hear speech that describes actions, such as the word 'kick', a similar set of interactions occur. This is known as semantic embodiment, and describes the observation that hearing the word 'kick' causes activity in many of the same brain regions involved in physically kicking your leg. Together, this suggests that the preservation of those brain networks that support the ability to move to verbal commands can be estimated via EEG responses to action speech (e.g. 'kick').

This research has four primary objectives. First, by combining EEG with functional magnetic resonance imaging (fMRI), we will map out the characteristics of the EEG markers of speech processing in healthy individuals. Second, we will develop statistical methods to identify these EEG markers with confidence in individual patients. Third, we will test the predictive power of these EEG markers. Specifically, we will investigate EEG markers of speech processing in a group of 135 patients in coma after a TBI. One year after this assessment, we will determine each patient's level of recovery through interviews with caregivers and thorough behavioural assessments. Finally, we will test whether EEG markers of speech processing can also reduce the high rate of misdiagnosis currently occurring in PDOC.

The more accurate methods of prognosis that come from this research will result in more appropriate provision of limited rehabilitation resources post-coma, thereby reducing healthcare costs. The more accurate methods of PDOC diagnosis that come from this research will ensure that no aware patient is misdiagnosed as unaware, thereby profoundly influencing the lives of patients and their families/caregivers. From a scientific perspective, this research will contribute to our understanding of how the meaning of speech is represented in our brains, and the ways in which the human brain supports consciousness.

The ability to move in response to verbal commands is a diagnostic milestone in recovery from coma that also provides a target for rehabilitation. The relative preservation in coma of the auditory-motor cortical network that supports this ability may predict the potential for that network to support consciousness subsequently. Hearing speech that describes action ('kick') produces rapid and automatic activation in an auditory-motor network, known as semantic embodiment. Due to the overlap between these functional networks, the relative preservation of the network that supports movement to command may be estimated at the bedside via electroencephalographic (EEG) markers of semantic embodiment.

We will test a cohort of post-traumatic coma patients for evidence of semantic embodiment after withdrawal of sedation. We will determine outcome at 1-year through telephone interview (Glasgow Outcome Scale Extended) and multiple assessments with the Coma Recovery Scale-Revised for those progressing to a prolonged disorder of consciousness (PDOC). Multinomial ordinal regression will determine the prognostic value of EEG markers of speech processing, with outcome categorised as good (at least functional communication), command-following (at least MCS plus), or no command-following (at most MCS minus), thus providing greater functional specificity in prognosis that can stratify patients for earlier rehabilitation or inform discussions of best interests and prior wishes.

The diagnostic utility of EEG markers of speech processing will also be tested in PDOC patients at follow-up. To aid in interpretation of each patient's specific neurocognitive preservation, dynamic causal modelling of simultaneous EEG-fMRI data will characterise the network interactions that underlie EEG markers of semantic embodiment in healthy individuals, specifically in relation to the network that supports movement to command.

A range of groups will benefit from the proposed research:

Critical care medicine: To impact upon prognostic practice in critical care, this research must be recognised in the UK guidelines produced by NICE. The proposed research will serve as a proof of principle for subsequent large multi-centre studies necessary to demonstrate clinical utility to NICE. In the short term, NICE citations of outputs from the proposed research alongside recommendations of further investigation will highlight the tools' potential to practitioners nationwide. Prof Belli (collaborator) will promote this research in his advisory role to the trauma board of NICE.

Long-term care: The diagnostic methods that stem from the proposed research may guide rehabilitation resources toward previously misdiagnosed patients. As in critical care, to be considered as a standard diagnostic tool it will be necessary for the outputs of the proposed research to be cited in the Royal College of Physicians' (RCP) diagnostic guidelines. Several members of the RCP Guideline Development Group are collaborators on the proposed research and on related projects, providing a channel through which to advocate for these methods. In the immediate term, I will produce reports outlining research findings for each patient's clinical team. Reports are common practice in this field and will meet the highest ethical standards. The clinical team will carefully discuss the results with the patient and their families, ensuring that they have access to all available information relating to levels of awareness.

Legal system: In England and Wales, court proceedings are required to approve withdrawal of clinically assisted nutrition and hydration (CANH) in PDOC, primarily due to the potential for uncertain diagnoses. The methods that stem from the proposed research may reduce this uncertainty. By engaging with collaborators who serve as experts in these court proceedings, I will promote the first use of non-behavioural diagnostic methods in court, setting a precedent and thereby increasing the effectiveness of UK legal policy.

Healthcare and economy: The cost of care for patients with irreversible PDOCs is estimated at £1.5bn per year in the UK; patients who may live for decades after their injuries, resulting in significant economic impact. More accurate prognoses in critical care may assist families in balancing the expected level of recovery with the patient's best interests and prior wishes, and may allow earlier access to rehabilitation, thereby increasing the effectiveness of public health services and enhancing quality of life of those who progress beyond coma.

Public interest: Disorders of consciousness continue to attract public interest, perhaps due to emotive associations with withdrawal of CANH and misdiagnosis. To ensure continued public engagement, I will give talks and demonstrations at public events (science festivals, open days), promote my research through press releases and interviews, and form a dialogue with families of patients through talks at brain injury support groups.

Capacity building: The PDF, PDRA, and Research Nurse will gain skills in technical research methods and skills that can be applied across a range of sectors, such as managing databases and working with families of patients who lack capacity. The PDF will develop their communication skills through presentations at international conferences and preparing manuscripts. As an early career researcher, the PI will benefit from the infrastructure for patient research that comes from the proposed research, the experience of managing a research team, and an increased standing within the field, thus setting the stage for future collaborations.