By Alexander E. Merkler, MD, MS
Assistant Professor of Neurology and Neuroscience, Weill Cornell Medical College, and Assistant Attending Neurologist, New York-Presbyterian Hospital
In conjunction with serial clinical examinations, electroencephalogram and functional magnetic resonance imaging may be helpful in predicting who will recover consciousness after an acute brain injury. However, in this study, early withdrawal of care leaves much uncertainty regarding the probability of eventual recovery.
Amiri M, Fisher PM, Raimondo F, et al. Multimodal prediction of residual consciousness in the intensive care unit: The CONNECT-ME study. Brain 2023;146:50-64.
Acute brain injury can lead to prolonged disorders of consciousness. A key question in the management and prognosis of patients with acute brain injury is whether and when a patient with a disorder of consciousness will recover. Identifying prognostic factors or novel methods to determine which patients will recover consciousness is vitally important, since decisions regarding ongoing life support vs. withdrawal of life-sustaining care often are based on neurological prognosis and the belief as to whether there will be recovery of consciousness.
Prior studies have identified that advanced neuroimaging and electroencephalogram (EEG) analysis may be helpful in identifying whether consciousness and awareness are present in patients with a clinical state of coma. Prior studies have suggested that advanced imaging/EEG may reveal a cognitive-motor dissociation in which the physical examination may not show any clinical indication of consciousness, but advanced imaging/electrophysiological modalities suggest that consciousness may be present.1,2
In the CONNECT-ME study, the investigators focused on the initial hospitalization for brain injury and evaluated whether functional magnetic resonance imaging (fMRI) and EEG might be helpful in identifying residual consciousness. This is highly relevant, since decisions regarding withdrawal of life support are made during the initial hospitalization for brain injury before life-sustaining measures, such as tracheostomy and percutaneous gastrostomy, are performed. The investigators enrolled patients with acute disorders of consciousness as the result of either traumatic or non-traumatic brain injury. These patients underwent repeated clinical assessments, as well as EEG and fMRI. Machine learning was used to assess whether EEG and fMRI at study enrollment could help distinguish patients in a minimally conscious state (MCS) or better (≥ MCS) from those in coma or unresponsive wakefulness state (UWS or ≤ UWS) both at the time of study enrollment and at intensive care unit (ICU) discharge.
Resting state EEG and EEG response to external stimulations were evaluated, in addition to spectral band analysis and a machine learning classifier (support vector machine). Parameters were manually adjudicated from the EEG and were used in the machine learning algorithm. The predictive performance was assessed with area under the curve (AUC). The authors enrolled 87 patients, of which 59% were ≤ UWS and 41% were ≥ MCS at the time of study enrollment. Thirty-six percent of patients died during their ICU admission, with most deaths caused by the withdrawal of life-sustaining therapy. EEG and fMRI predicted consciousness levels at study enrollment and ICU discharge with maximum AUCs of 0.79 and 0.71, respectively. Models that combined EEG and fMRI predicted consciousness levels at study enrollment and discharge of 0.78 and 0.83, respectively.
Limitations of the study include the fact that a large proportion of patients underwent withdrawal of life-sustaining therapy and, thus, the performance of the multimodal approach of EEG and fMRI may be subject to bias. In addition, EEG characteristics were based on clinical adjudication and also were subject to bias. As the authors acknowledged, fMRI is subject to technical challenges, especially in ICU patients who are critically ill. Finally, it is uncertain whether actual EEG and fMRI were better than standard clinical examination in predicting recovery from coma in the acute critically ill setting, since no direct comparison was made between prognosis based on clinical measures compared to EEG/fMRI findings.
COMMENTARY
CONNECT-ME is an important step in identifying patients who may have covert consciousness and who may recover overt signs of consciousness. This has important implications, since neurological prognosis often is challenging in acute brain-injured patients but is vitally important for decisions regarding ongoing medical care and withdrawal of life-sustaining therapy.
Overall, although some patients may fit the phenotype of cognitive-motor dissociation and, thus, may benefit from a multimodal approach at identifying covert consciousness, the proportion of these patients is uncertain, and future research is necessary to assess the reliability and practicality of these advanced tools.
REFERENCES
- Monti MM, Vanhaudenhuyse A, Coleman MR, et al. Willful modulation of brain activity in disorders of consciousness. N Engl J Med 2010;362:579-589.
- Claassen J, Doyle K, Matory A, et al. Detection of brain activation in unresponsive patients with acute brain injury. N Engl J Med 2019;380:2497-2505.
