Estrogen-related receptor gene expression associates with sex differences in cortical atrophy in isolated REM sleep behavior disorder

Abstract

Isolated REM sleep behavior disorder, characterized by dream-enacting movements during REM sleep, is a male-predominant parasomnia and the strongest prodromal marker of synucleinopathies. Individuals with this disorder show cortical atrophy whose regional distribution covaries with gene expression patterns measured in the healthy human brain. However, the effect of sex on these brain changes remains unknown. The study objective is to comprehensively assess sex differences in cortical morphology and to characterize the healthy-brain gene expression correlates of brain abnormalities using the largest international multicentric MRI dataset of polysomnography-confirmed patients. Males have significantly more extensive and severe cortical thinning compared to females, despite similar age and clinical features. Imaging transcriptomics analyses indicate that regions affected in female patients map onto areas with higher expression of estrogen-related receptor genes, particularly ESRRG and ESRRA, in the healthy brain. These findings support potential sex-specific neuroprotection in the prodromal stages of synucleinopathies and may inform personalized and targeted therapeutic strategies.

Fig. 1. Participant selection flowchart.

The initial sample included 888 participants. Following exclusions during MRI processing, visual quality control, and age matching, the final sample consisted of 687 participants. P values (two-sided) indicate differences in age distributions between groups before and after matching using independent two-sample t tests. F = female; HC = healthy controls; iRBD = isolated REM sleep behavior disorder; M = male.

Fig. 2. Vertex-wise sex-by-group interaction effect on cortical thickness.

a Clusters showing significant sex-by-group interaction on cortical thickness at a cluster-forming threshold of P < 0.05 and following Monte Carlo permutations. The color bar indicates the statistical significance on a logarithmic scale of P-values (−log10), with positive values (red-yellow scale) indicating the presence of a significantly stronger sex effect in iRBD patients compared to controls. b Average cortical thickness (in mm) across groups in significant clusters, showing significant reduction in cortical thickness in iRBD males (N = 294) compared to iRBD females (N = 49) and controls (131 females, 213 males). The P-values indicate significant differences between groups after one-sided independent two-sample t tests. Box plots display the median (center line), the interquartile range (box; 25th-75th percentiles), and whiskers extending to the most extreme values within 1.5 times the interquartile range from each quartile. Each point represents the mean cortical thickness at cluster peak for each participant. Brain maps were generated using freeview in FreeSurfer. F = females; HC = healthy controls; iRBD = isolated REM sleep behavior disorder; M = males.

Fig. 3. Brain map showing sex-stratified atrophy patterns in iRBD.

a Cortical maps showing the distribution of reduced cortical thickness in females and males with iRBD against age- and sex-adjusted values. The color bar indicates W-scores, representing the severity of regional deviation compared to expected morphology. b Cortical maps showing the significantly atrophied regions in iRBD females and males after conducting one sample t tests for each cortical region. The color bar indicates t-values for regions reaching significance after FDR correction (P_FDR < 0.05). Cortical surface maps were generated with SurfIce. FDR = false discovery rate; iRBD = isolated REM sleep behavior disorder.

Fig. 4. GSEA on sex interaction estimates of cortical thickness in iRBD.

a Violin plots showing the percentage of variance in sex interaction estimates explained by gene expression; the dot represents the empirical variance, and the asterisk indicates the components that were significant against spatial null models. b Scatterplot of the association between the sex interaction estimates and the regional weights of the first and third c latent variables. Pearson’s correlation coefficient (r) and corresponding P-value are reported. The red line shows the linear regression fit, and the shaded error band represents the 95% confidence interval (CI) of the fit. d Brain maps of the sex interaction coefficients and their regional weights for the first and third (f) latent variables. e, g The top ten molecular function terms from the Gene Ontology knowledge base that are enriched in the positively and negatively weighted gene sets associated with sex effect on cortical thickness in iRBD. Terms are ranked based on the normalized enrichment score; darker colored bars present significantly enriched terms after FDR correction. Brain maps were generated with SurfIce. FDR = false discovery rate; GSEA = gene set enrichment analysis; iRBD = isolated REM sleep behavior disorder; LV = latent variable; PLS = partial least square.

Fig. 5. Top 24 tissues expression of ESRRG and ESRRA.

a, b Violin plots of expression values of ESRRG and ESRRA across tissue types, with brain tissues shown in yellow and non-brain tissues shown in lilac. The P-value indicates a significantly higher proportion of brain tissues overexpressing ESRRG compared to ESRRA. Box plots display the median (center line) and the interquartile range (box; 25th-75th percentiles). Points appearing outside of the boxes correspond to outliers above or below 1.5 times the interquartile range. The expression of the top 24 tissue types, out of 54 available, are shown. Gene expression levels for each tissue were obtained from the GTEx Analysis Release V10 (dbGAP Accession phs000424.v10.p2) via the GTEx Portal (accessed 11/22/2024). ESRRA = estrogen-related receptor alpha; ESRRG = estrogen-related receptor gamma; GTEx = genotype-tissue expression; TPM = transcripts per million.