Electroencephalogram prediction of propofol effects on neuromodulation in disorders of consciousness

Objective This study aimed to characterize electroencephalogram (EEG) responses to low-dose propofol anesthesia in patients with disorders of consciousness (DoC) of distinct etiologies—traumatic brain injury (TBI), anoxic ischemic encephalopathy (AIE), and cerebrovascular accident (CVA)—and explore their prognostic relevance for recovery after spinal cord stimulation (SCS). Methods A retrospective cohort of 40 DoC patients (TBI: 15, CVA: 14, AIE: 11) undergoing SCS under propofol anesthesia was analyzed. Pre- and post-anesthesia 19-lead EEG recordings were evaluated for power spectral density (PSD) in δ (0.5–4 Hz), θ (4–8 Hz), α (8–13 Hz), β (13–30 Hz), and γ (30–45 Hz) bands, alongside permutation entropy (PE). Consciousness levels were quantified using the Coma Recovery Scale-Revised (CRS-R) preoperatively and 3 months post-SCS. Etiology-stratified analyses compared neurophysiological and clinical outcomes. Results Propofol universally suppressed β- ( p < 0.001–0.05) and γ-band ( p < 0.001–0.05) power across all groups. Etiology-specific EEG patterns emerged: AIE patients displayed reduced frontal α-power (Δ = −0.23, p = 0.03), while TBI/CVA patients showed prefrontal-parietal β/γ suppression (Δβ = −0.41, Δγ = −0.38; p < 0.001). Significant PE reduction (ΔPE = −0.21, p < 0.001) correlated with CRS-R improvement ( r = −0.67, p = 0.003) in TBI/CVA subgroups but not in AIE (ΔPE = −0.05, p = 0.12). Three-month outcomes varied by etiology: 20% of TBI patients achieved a minimally conscious state (CRS-R ≥ 10) with enhanced motor (Δ = +0.25, p < 0.01) and visual function (Δ = +0.19, p = 0.03). CVA patients exhibited partial motor (Δ = +0.20, p = 0.007) and arousal gains (Δ = +0.17, p = 0.01), whereas AIE patients showed negligible improvement (mean ΔCRS-R = 0.4 ± 0.3). Conclusion Propofol-induced EEG modulation reflects etiology-dependent neural network vulnerabilities in DoC. TBI/CVA patients demonstrated entropy reduction linked to clinical recovery, suggesting transient network stabilization that may enhance SCS efficacy. In contrast, AIE-associated static dynamics imply irreversible structural damage. Integrated PSD/PE analysis holds prognostic potential for predicting SCS responsiveness, particularly in TBI/CVA cohorts. These findings advocate etiology-tailored neuromodulation strategies, though multicenter validation is imperative for clinical translation.