Genetic correlates of sleep spindle abnormalities in schizophrenia

Previous studies have confirmed a robust and multifactorial genetic basis for the normal human electroencephalogram (EEG). Schizophrenia (SCZ), a common mental disorder with an estimated heritability of approximately 80%, exhibits varying levels of abnormalities in both wake and sleep EEG activities. Among these EEG metrics, sleep spindles have been reported to contribute more to disease classification than others, but it remains unknown whether they have a stronger genetic correlation with SCZ susceptibility. Notably, two subtypes of sleep spindles, namely slow spindles (SS) and fast spindles (FS), exhibit distinct and independent scalp distributions, clinical relevance, and transferability across populations. Among them, the FS density exhibited the most stable between-group differences and effect size. Differences in characteristics between SS and FS can be attributed to distinct genetic underpinnings. Pulling together existing evidence, we hypothesize that: 1) Sleep spindles have a significant and higher degree of correlation with polygenic risk scores (PRS) of SCZ compared to other EEG indicators; 2) Within sleep spindles, FS has a closer correlation with SCZ PRS than SS, with FS density displaying the strongest correlation among all indicators. A total of 72 Chinese SCZ patients and 58 healthy controls (HC) underwent a whole-night high-density EEG recording, as well as several well established event-related potentials (ERP) paradigms during wakefulness, including P50 sensory gating, P3 novelty, auditory steady-state response, and mismatch negativity. Sleep spindle features were extracted using Luna software (v0.25.5), while other EEG metrics were analyzed using Brainvision Analyzer 2.2. Group differences of all EEG metrics were estimated using the logistic regression model. Blood samples have been recently sequenced via a combined genome and exome product named Blended Genome-Exome, which enables imputation and variant calling from a single sample. SCZ PRS will be computed based on GWAS summary statistics from East Asian populations, using the approaches we recently reported. Besides, polygenic scores (PGS) for height will be calculated in a similar manner as a negative control measure. Correlations with both PRSs will be analyzed and compared across EEG variables and sleep spindles subtypes. Additionally, linear regression analysis will be adopted to determine the independent contribution of each EEG variable to the SCZ PRS by controlling for potential confounders such as sex, age, and body mass index. Compared to HC, individuals with SCZ showed significant deficits in the majority of wake and sleep EEG metrics, with effect sizes ranging from 0.59 to 1.27 standard deviations (SD). Sleep EEG indicators generally exhibited higher effect sizes compared to those observed during wakefulness, and notably, FS density displayed the broadest decrease in SCZ. The calculation of PRSs and further PRS-based analyses, which are valuable for understanding the genetic correlations between different EEG variables and SCZ, are currently underway. This study validated the abnormalities of sleep spindles and several other EEG metrics in SCZ. By comparing their correlations with and independent contributions to SCZ PRS, we can gain insights into the potential distinct genetic underpinnings of sleep spindle subtypes, thereby providing valuable implications for further genetic research on sleep spindles and enhancing our understanding of their role in SCZ.