Multi-modal integration of protein interactomes with genomic and molecular data discovers distinct RA endotypes

Rheumatoid arthritis (RA) is a complex autoimmune disease characterized by clinical and molecular heterogeneity, notably in the presence of anti-cyclic citrullinated peptide antibodies (CCP). CCP positivity in RA patients is associated with more severe disease progression and distinct responses to treatment compared to CCP- patients. Although previous studies have investigated cellular and molecular differences between these RA subtypes, there has been limited exploration of their genetic differences at a systems scale, taking into account underlying molecular networks.

Here, we use a novel multi-scale framework that couples a network-based genome-wide association study (GWAS) to functional genomic data to uncover network modules distinguishing CCP+ and CCP- RA.

We utilized the RACER (Rheumatoid Arthritis Comparative Effectiveness Research) cohort, comprising 555 CCP+/RF+ and 384 CCP-/RF+ RA patients, and uncovered significant differences in heritability between these two disease groups. This was followed by a network-based GWAS which uncovered 14 putative gene modules that explained genetic differences between CCP+/RF+ and CCP-/RF+ RA. Interestingly, these included many genes outside the HLA locus. Further validation through heritability partitioning and multivariate expression analyses underscored the significance of specific modules, highlighting novel genetic loci driving heterogeneity in antibody prevalence in RA. The identified modules were validated in a completely orthogonal cohort from the All of Us program. Functional analysis revealed that these modules captured critical molecular programs that not only relate to serological variation but also underlie broader functional heterogeneity in RA, including differences in synovial cell-type abundance phenotypes and variation in treatment response.

Our findings demonstrate the utility of network-based approaches in elucidating the complex genetic landscape of RA, offering new insights into the differential genetic risk factors underlying CCP+/RF+ and CCP-/RF+ RA, and paving the way for more personalized therapeutic strategies.