Identification of Crohn’s disease subtypes in single cell RNA sequencing signatures of treatment naïve samples across the gastrointestinal tract

Crohn’s disease (CD) is characterized by chronic inflammation throughout the intestines. Previous single cell transcriptomic studies investigated the effect of CD post therapeutic intervention, however, the extent to which therapy modifies the intestinal inflammatory environment remains to be elucidated. To this end, a treatment naïve pediatric CD cohort was recruited to investigate initial disease onset across the intestines, highlighting shared and tissue-specific consequences of disease. Ileum, colon, and rectum biopsies were obtained, and single cell RNA-sequencing was performed. A clustering stability assessment workflow was developed to ensure clustering and downstream results were robust. Tensor decomposition was utilized to identify clinically meaningful groups within the intestines. Other bioinformatic tools were utilized to investigate differences across group status within each tissue. Expected cell types were identified after the clustering stability assessment was performed. Inflammation did not strongly influence cellular proportion profiles due to heterogeneity across donor and tissue. Tensor decomposition revealed distinct mesenchymal and cytotoxic T cell-mediated sources of disease pathology, corresponding to previously identified fibrotic and pro-inflammatory disease progression. Integrating transcriptomics and genome wide association summary statistics for CD suggested myeloid cells and T cells drive disease, highlighting potential cellular therapeutic targets. Tensor decomposition stratified donors into clinically meaningful groups based on their transcriptomic profile, suggesting these signatures can be utilized for personalized medicine.