Abstract
Carroll EE, Shen Q, Kansara V, Casson N, Michalak A, Niesvizky-Kogan I, Lim J, Postelnik A, Viereck MJ, Egawa S, Kahan J, Carmona JC, Kruger L, Song YL, Velazquez A, Schevon CA, Connolly ES, Ghoshal S, Agarwal S, Roh D, Park S, Kent P, Claassen J. Nat Med. 2025 May;31(5):1578–1585. doi: 10.1038/s41591-025-03578-x. Epub 2025 Mar 3. PMID: 40033114. Cognitive motor dissociation (CMD) can improve the accuracy to predict recovery of behaviorally unresponsive patients with acute brain injury, but acquisition and analysis of task-based electroencephalography (EEG) are technically challenging. N2 sleep patterns, such as sleep spindles on EEG, have been associated with good outcomes, rely on similar thalamocortical networks as consciousness and could provide less technically challenging complementary outcome predictors. In this prospective observational cohort study of 226 acutely brain injured patients, well-formed sleep spindles (WFSS) were more likely present in those with CMD than in those without CMD, often preceding the detection of CMD. WFSS were associated with a shorter time to recovery of consciousness, and both CMD and WFSS independently predicted recovery of independence, controlling for age, admission neurological status and injury type. WFSS are seen in approximately every third behaviorally unresponsive patient after acute brain injury, frequently precede detection of CMD and are a promising complementary predictor for recovery of consciousness and functional independence.
Commentary
Managing patients with severe acute brain injury (ABI) with prolonged impairment of consciousness is often challenging. It often leads to a pessimistic outlook. Historically, prognostication research after severe ABI has focused on identifying predictors of poor outcomes. This is especially true after conditions such as anoxic brain injury. This focus aimed to support decisions to limit or withdraw aggressive treatment and avoid prolonging futile care. Yet, the onward and upward march of modern medicine is, at its core, a journey to improve patient outcomes. This journey can be constrained by how we frame problems and where we look for solutions.
In recent years, the field has witnessed a welcome broadening of this lens. Increasing attention is now focused on identifying positive prognostic markers, features that signal the potential for a meaningful recovery. The level of consciousness impairment is traditionally determined by clinical examination. More recently, advanced neurophysiological testing has identified patients who are unresponsive on behavioral evaluation but can willfully modulate their brain activity to command. This cognitive motor dissociation (CMD) can be present in up to a quarter of ABI patients lacking an observable response to commands. 1
Electroencephalography (EEG) has emerged as a more helpful tool for assessing CMD. Its high temporal resolution and wide availability at the bedside allow repeat assessments. 2 A landmark study found that 15% of unresponsive patients showed signs of brain activation on EEG, as analyzed using a machine learning algorithm, in response to commands to move their hands. Remarkably, patients with CMD showed more than 4 times increased odds of living partially independently at 12 months. 3 Despite its promise, the time-locked EEG data acquisition and complex analysis are not widely feasible. To help translate this paradigm shift into routine clinical practice, the current study by Carroll et al 4 focused on raw EEG features, specifically sleep spindles.
Sleep spindles, first described in 1935 as rhythmic bursts of 12–16 Hz activity, are a hallmark of nonrapid eye movement sleep. 5 Steriade et al 6 elucidated their neurophysiological basis and origin in thalamocortical circuits, specifically through rhythmic interactions between thalamic reticular nucleus neurons and thalamocortical relay neurons. The advent of continuous EEG (cEEG) over the past two decades has enabled neurophysiologists and critical care neurologists to observe electrographic changes after ABI and correlate these with disorders of consciousness. A critical advance came when specific EEG spectral patterns were shown to correlate with clinical outcomes after anoxic brain injury, with these patterns interpreted as indirect markers of corticothalamic network integrity. 7 Studies have suggested that EEG spindles, noted in one-third of intensive care unit patients, are associated with favorable outcomes across a range of etiologies. 8
The study by Carroll et al 4 takes this work a critical step further by rigorously evaluating well-formed sleep spindles (WFSS) as scalable, bedside EEG markers of recovery potential in patients with severe ABI. They included 226 ABI patients, median age 63 years, with prolonged impaired consciousness and a Glasgow Coma Scale (GCS) score of 6. Almost 40% had intracranial hemorrhage, and around 15% each had suffered cardiac arrest, subarachnoid hemorrhage, and traumatic brain injury with hemorrhage. The EEG recording was performed a median of 4 days after ABI. CMD was detected in 15% of patients using their previously validated EEG-based motor command-following paradigm, applying machine learning algorithms. 3 Demographics, injury type, and standard EEG features did not significantly differentiate patients with and without CMD. However, patients with CMD had 2.7 times higher odds of showing WFSS on EEG (52% vs 28%). Overall, 31% of the study cohort had WFSS and 26% had rudimentary sleep spindles. Notably, WFSS frequently preceded CMD detection, suggesting that their presence on routine EEG review may serve as an early, accessible marker of CMD. The multivariable analysis, adjusting for GCS, etiology, and age, found that only WFSS, and not CMD, was significantly associated with time to recovery of consciousness by hospital discharge (HR 1.1). The postdischarge functional outcome was assessed using the Glasgow Outcome Scale—Extended (GOS-E), with a score cut-off of ≥4, which translates to living 8 h without assistance. After accounting for death as a competing risk, censoring patients who underwent withdrawal of life-saving treatment, and adjusting for GCS, etiology, and age, the presence of WFSS and CMD was independently associated with 1.2 and 2.6 times faster recovery to a GOS-E of ≥4, respectively, at 12 months after injury.
The value of the Carroll et al 4 study lies not in reaffirming the association between CMD and favorable outcomes, which is already convincingly confirmed.1,3,9 Instead, it shows that the detection of sleep spindles, even rudimentary ones, provides complementary prognostic information across severe ABIs. Adding WFSS to models that already include CMD improves the discrimination of long-term functional outcomes by 10 percentage points. Unlike CMD, WFSS is associated with both short-term and long-term outcomes. Notably, WFSS was most informative in patients without CMD, helping to identify covert recovery potential that was missed by standard testing. In contrast, WFSS did not distinguish further outcomes in CMD patients, likely because the latter is already a strong prognostic marker. However, even in CMD-positive patients, the presence of rudimentary spindles was associated with better outcomes, suggesting that partially preserved sleep architecture still adds prognostic value. These findings support the idea that sleep spindles, reflecting thalamocortical integrity, can not only predict CMD status but also independently forecast recovery, positioning them as a promising addition to multimodal prognostication frameworks.
Not overlooking the role of IV anesthetics, Carroll et al 4 found that the use of minimal, low, or moderate levels of sedative medications was associated with a reduced likelihood of detecting WFSS, but sedation did not influence CMD detection. Notably, after excluding all assessments conducted under moderate sedation, WFSS remained significantly more common in CMD-positive patients and continued to predict outcomes independently. However, before applying the findings from this study in routine clinical practice, it is essential to understand the clinical scenarios or aspects where their conclusions may not be applicable. Patients with uncontrolled seizures or receiving barbiturate coma were excluded. It is unclear whether findings apply equally across etiologies. Almost 14% of the 139 patients without WFSS and CMD also recovered consciousness. This suggests that these two cannot be considered the sole predictors of prognosis. A multimodal approach remains the cornerstone of managing patients with prolonged disorders of consciousness. 2
An unsung hero of this study is EEG, especially cEEG, and its expanding role in dynamic brain monitoring and prognostication. Initially deployed to detect electrographic seizures and status epilepticus, cEEG has evolved, as demonstrated by Carroll et al, 4 into a powerful tool for assessing thalamocortical network integrity and tracking recovery in patients with severe ABI. Remarkably, 90 years after their first description, sleep spindles are now emerging as accessible, bedside prognostic biomarkers in this population. EEG continues to reveal knowledge that has been hidden in plain sight, nearly a century after its first use. As our understanding of sleep architecture in critical illness deepens, incorporating spindle assessment into routine cEEG interpretations could enhance prognostic precision in disorders of consciousness.
