How to establish an accurate neurological prognosis in a patient suffering from a severe disorder of consciousness, coma, vegetative state, or minimal consciousness? This is one of the most difficult challenges in intensive care, both medically, scientifically and ethically.
“These patients in intensive care, with organ replacement machines (artificial ventilation type), can take a long time to regain a state of consciousness. We can then wonder if they will really recover, and if so, with what after-effects”comments to the Daily Dr Benjamin Rohaut, neurologist-resuscitator (Sorbonne University/AP-HP). Obtaining the most precise answers possible about the patient’s cognitive recovery abilities is a real challenge. “We are trying to predict the future, for a damaged brain, while the cognitive dimensions are complex and very imperfectly known,” summarizes the specialist.
Currently, the neurological prognosis is established based on several indicators, including examination of the anatomy of the brain (CT scan and MRI) and its functioning (electroencephalogram [EEG]). “Even with this information, there often remains some uncertainty in the prognosis, which can have consequences on medical decision-making. However, these patients are often very fragile and exposed to numerous complications, which each time raises the question of the intensity of care. explains Dr. Rohaut.
When uncertainty is great, European and American recommendations argue in favor of a multimodal assessment, combining clinical, electrophysiology and neuroimaging. But this has never yet been validated in practice. A group of clinical researchers from the Brain Institute and the Pitié Salpêtrière Hospital (AP-HP), led by Professor Lionel Naccache (Sorbonne University/AP-HP), Drs Jacobo Sitt (Inserm) and Benjamin Rohaut therefore evaluated the performance of such an approach, and showed how it allows better prognoses to be made. Results published in Nature Medicineon May 30.
Association with functional results at one year
Researchers evaluated 349 patients referred to the neurologically oriented intensive care and resuscitation department at Pitié Salpêtrière between 2009 and 2021. More than 63% were men, with a median age of 53.2 years, with a medical history in more than 70% of cases. The disorder of consciousness was caused in 36% of cases of anoxia, in 19% of traumatic brain injury and in 14% of cerebrovascular accident (CVA). “These patients were referred to us because the first-line assessment did not allow us to confidently make a prognosis”specifies Dr. Benjamin Rohaut.
Made up of neurologists, neurophysiologists, neuroradiologists, neuroscientists, the “DOC” team (for Disorders of Consciousness) has, over the years, enriched its neuroprognostic assessment with new markers, increasing their number from four to 12*. “For disorders of consciousness, no single marker is sufficiently effective to establish a reliable prognosis”recalls Dr Rohaut.
Researchers have therefore combined clinical assessments, multivariate EEG analysis algorithms, quantitative MRI known as DTI, cognitive evoked potentials (electrical responses to sensory stimulation), etc. At the end of each evaluation, the researchers formulated a prognostic opinion: “good” in 22% of cases, “uncertain” in 45.5% of cases and “unfavorable” for the remaining 32.5%.
The more we increase the number of tools for evaluating a disorder of consciousness, the more we are able to make a precise prognosis.
Dr Benjamin Rohaut, neurologist-resuscitator (Sorbonne University/AP-HP)
First result, the prognosis thus achieved turns out to be significantly associated with functional results at one year. Patients with a “good prognosis” had a 33% chance of seeing their cognitive abilities improve favorably (defined by a score between 4 and 8 on the Glasgow GOS-E scale ranging from 0 to 8); the others with a prognosis considered “uncertain” had only a 20% chance, and none of the patients assessed as “unfavorable” had regained consciousness after one year. Note that in the group with a good prognosis there were fewer vegetative states (1.8%) than in those with an uncertain (5.6%) or unfavorable prognosis (13.5%), but a greater proportion of patients with severe disabilities making them totally dependent.
Second observation: this prognostic performance was correlated with the number of modalities used: the more indicators used, the greater the precision of the prognosis. The proportion of uncertain diagnoses drops from 57.5% to 32% between an evaluation based on fewer than six modalities, and another based on more than six. The accuracy of predictions increases from 66% to 84%.
“We could have feared that increasing the number of examinations, by increasing the probability of having discordant results, would increase the uncertainty and complexity of a choice. But no! The more we increase the number of tools for evaluating a disorder of consciousness, the more we are able to make a precise prognosis. insists Dr Rohaut. “The care plan for the patient is all the more supported, even if the neuroprognosis is far from being the only determining factor: we also take into account the patient’s wishes when we have access to them, discussions with the family, “institutional and material environment…”, he adds. And to emphasize the importance of weekly meetings where the situations of each patient are discussed, with the multidisciplinary team trying to make sense of all the results of the examinations. “In the future, we could imagine an AI that takes all this data into account and delivers an overall prognostic score, but critical human analysis will have to have the last word”considers Dr. Benjamin Rohaut.
Standardization of the delicate multimodal approach
“This long-term study shows for the first time the benefit of the multimodal approach, which constitutes essential information for intensive care units around the world. It also makes it possible to empirically validate recent recommendations from academies. European And American of neurology”, details Dr. Jacobo Sitt in a press release.
But the generalization of the multimodal approach remains a challenge and the study refrains from recommending a list of examinations to be performed. “Every team depends on the tools they have access to. For the moment the evaluations are specific to each institution », recognizes Dr. Benjamin Rohaut. If the majority of centers have clinical expertise for first-line assessments, EEG, MRI and scanner, the Pitié Salpêtrière hospital stands out for its experience in cognitive evoked potentials (which makes it possible to classify brain function in four levels according to the response, observed via EEG, of the patient to sounds) and a multidimensional analysis of the EEG (complexity of the signal, power in different spectral limits, or markers of functional connectivity between different brain regions). The team also uses a tool for quantifying the fraction of anisotropy of the white matter (which provides information on its alteration) and carries out research into cognitive-motor dissociation by analyzing the brain activity (by EEG) of a patient who is asked to move their hand.
“We propose to build a network of collaborations at national and European level. Thanks to the use of telemedicine tools and automated analysis of EEG or brain imaging, all intensive care units could have a first level of access to multimodal assessment. If it proves insufficient, recourse to an expert regional center would provide a more in-depth assessment. Finally, in the most complex situations, it would be possible to call on all available experts, wherever they are,” concludes Professor Lionel Naccache in the press release.
*The first four markers are: CRS-r (Coma Recovery Scale revised), EEG (Electroencephalography) SSEP (Somatosensory Evoked Potential), and ERP (Event related potential). Then in 2011, Four (Full Outline of Unresponsiveness score) was integrated, in 2013, RS-fMRI: Resting State-functional IRM, in 2015, FA (Fractional Anisotropy), 2016, Pet-index, 2020, Doc-Feeling, and hASR (habituation of Auditory Startle Reflex), and 2021, motor task and Cognitive Motor Dissociation.
Rohaut B. et al., Nature Medicine, 2024. DOI:10.1038/s41591-024-03019-1
