Theta oscillation attenuation as an electroencephalogram biomarker for intracranial infections in hematologic malignancies

Abstract Background: Patients with hematologic malignancies frequently experience immune suppression that predisposes them to intracranial infections, which carry high morbidity and mortality. Early, noninvasive monitoring biomarkers are urgently needed. This study aims to analyze resting-state neural oscillation changes to identify potential biomarkers for intracranial infections in critically ill hematological patients. Methods: From May to December 2024, 46 critically ill hematologic patients at the Institute of Hematology, Chinese Academy of Medical Sciences, were enrolled: 23 with confirmed intracranial infection (Inf group) and 23 without (Con group). We collected demographics, clinical scores (Glasgow Coma Score GCS,and Sequential Organ Failure Assessment SOFA), laboratory tests, neuroimaging (MRI/CT), and resting 19-channel EEG. Power spectral density (PSD) and spatial distribution of neural oscillations were computed via Fourier transform. Between-group comparisons focused on whole-brain and regional PSD in theta (4–8 Hz) and alpha (8–13 Hz) bands. Results: (1) The most common type of hematological disease in the Inf group was leukemia (56.52%). The Inf group had a significantly lower median Glasgow Coma Scale score compared with controls [9 (IQR 6-12) vs. 14 (IQR 12-15), P < 0.001], and a significantly higher median SOFA score [10 (IQR 7-15) vs. 6 (IQR 4-9), P < 0.001]. The mortality was higher in the Inf group than in the Con group (26.09% vs. 4.35%, P=0.040). (2) In both groups, theta-band PSD was the dominant frequency band. The Inf group exhibited significantly lower mean whole-brain PSD in the theta band (P = 0.030) and in the alpha band (P = 0.004) compared with the Con group.(3) In the Con group, frontal-region theta PSD was significantly higher than in other regions (ANOVA, F = 5.373, P = 0.0213), making it the dominant brain region. Among these, Fz channel showed the highest theta-band PSD, making it the characteristic channel for the theta band. Frontal theta-band PSD was significantly reduced compared with controls (P = 0.0039), and theta-band PSD at the Fz channel of the Inf group(1.493 ± 0.294) was significantly lower than that of the Con group (2.026 ± 0.250) (P < 0.05).Conclusion: Critically ill patients with hematologic malignancies complicated by intracranial infection demonstrate marked attenuation of theta-band oscillations in the medial prefrontal cortex (Fz). This reduction in Fz theta-band PSD may serve as a sensitive, noninvasive biomarker for early detection of intracranial infection in this high-risk population.