Sleep spindle alterations as a novel biomarker for phenotypic stratification in sporadic amyotrophic lateral...

To quantitatively evaluate sleep spindle alterations in sporadic amyotrophic lateral sclerosis (ALS) and explore their potential as biomarkers for diagnosis and phenotypic stratification.

In this cross-sectional study, overnight sleep electroencephalography was recorded in 97 sporadic ALS patients and 73 matched healthy controls. Sleep spindle parameters (amplitude, duration, density, frequency) were automatically analyzed at frontal leads. Multiple comparisons were controlled using the false discovery rate (FDR) approach. We used least absolute shrinkage and selection operator (LASSO) regression for diagnostic modeling and employed K-means clustering to define spindle-based subtypes. Bootstrap internal validation was performed to assess model optimism.

After FDR correction, ALS patients showed significant spindle abnormalities predominantly in the bipolar FP12 derivation, including reduced slow spindle density (p-FDR = 0.007), reduced overall spindle density (p-FDR = 0.007), and shortened slow spindle duration (p-FDR = 0.017). A diagnostic model incorporating Epworth Sleepiness Scale score, wake after sleep onset, sleep efficiency, FP12 slow spindle density, and education years showed promising discriminative ability (apparent AUC = 0.931; optimism-corrected AUC = 0.923). Unsupervised clustering consistently revealed two distinct spindle phenotypes. The "spindle-deficient" phenotype, characterized by poorer spindle integrity, was independently associated with lower ALSFRS-R scores (OR 1.101, 95% CI 1.024-1.202, p = 0.017), lower percentage of predicted forced vital capacity (OR 1.035, 95% CI 1.010-1.065, p = 0.011), and absence of drinking history (OR 3.03, 95% CI 1.02-9.46, p = 0.049).

Sleep spindle alterations may represent a core electrophysiological feature of ALS, potentially reflecting thalamocortical dysfunction. These exploratory findings suggest that spindle parameters could serve as candidate biomarkers for disease stratification, though validation in independent longitudinal cohorts is needed before clinical application.