Sjögren’s patient subgroups identified through whole genome DNA methylation profiling

Sjögren’s disease (SjD) is a heterogeneous autoimmune disorder characterized by lymphocytic infiltration of exocrine glands resulting in severe oral and ocular dryness. Previous published work showed DNA methylation (DNAm) can distinguish SjD case subgroups based on clinical features; however, studies used small samples and did not adjust for cellular heterogeneity in labial salivary glands (LSGs). Our objectives were to: (1) identify DNAm clusters from LSGs; (2) investigate cluster clinical characteristics; and (3) identify differential methylation between SjD case subgroups to further understand biological pathways. We identified clinically meaningful subgroups of SjD through hierarchical clustering of DNAm embeddings from a variational autoencoder (VAE) of LSGs, which allows for a low dimensional representation of the high dimensional methylation data. LSGs from 1,059 SjD cases (n = 592) and symptomatic non-cases (n = 467) were profiled using the Illumina HumanMethylationEPIC BeadChip, and cell-type proportions were estimated using a solid-tissue reference. Participants clustered into subgroups with differential SjD features and proportions of predicted cell-types. Comparison of SjD cases within distinct clusters showed evidence for differential methylation in each cell-type. The largest number of differences between subgroups occurred in epithelial and B-cells and were in genes and pathways with relevance to disease pathogenesis. In B-cells, methylation within NR2F2, previously reported to be differentially expressed in lacrimal glands of SjD mouse models, and NDRG2, which increases saliva production in estrogen deficient rats, distinguished subgroups with different clinical manifestations. Additional candidates of interest identified in epithelial and B-cells from SjD cases include genes previously implicated in systemic lupus erythematosus. These findings provide insight into the powerful link between epigenetics and clinical heterogeneity in SjD and contribute to classification of important patient subgroups.