A74-06 Gene Expression Differences and Derivation of a Peripheral Blood-Based Transcriptomic Risk Score for Extrapulmonary Sarcoidosis

Abstract Rationale Extrapulmonary sarcoidosis can be identified in many tissues but, as a variable manifestation of an already rare disease, it has not been systematically examined. While previous studies have identified demographic and environmental factors that account for some of the variability in sarcoidosis disease course, there have been few studies aimed at identifying biomarkers that are associated with extrapulmonary disease risk and yield insights into underlying pathobiology. We therefore sought to identify gene expression differences in blood and develop a transcriptomic risk score (TRS) to predict extrapulmonary sarcoidosis. Methods We used RNA sequencing data from peripheral blood mononuclear cells from individuals enrolled in the Genomic Research in Alpha-1 Antitrypsin Deficiency and Sarcoidosis (GRADS) study and from whole blood from individuals in an independent cohort from our institution. We performed a differential expression mega-analysis as well as pathway enrichment and cell-type deconvolution, comparing individuals with pulmonary-only sarcoidosis to individuals with extrapulmonary sarcoidosis. In addition, we deployed supervised learning models to develop a TRS to distinguish extrapulmonary sarcoidosis from pulmonary-only disease. Results We identified 594 genes that were differentially expressed at a nominal p 0.05 and |log2FC| 0.2 in pooled cohort mega-analysis. While we did not identify differences in estimated deconvoluted cell-type proportions in the GRADS cohort, we found that individuals in the independent cohort had significantly lower estimated proportions of naïve and memory B cells. After implementing a least absolute shrinkage and selection operator (LASSO) model on the training sample from the GRADS cohort, we identified 15 genes with non-zero weights. Included in the 15 TRS genes were five T cell receptor alpha subunit-related genes and five microRNAs. The TRS had moderate predictive ability for extrapulmonary sarcoidosis in the held-out testing sample from the GRADS cohort (AUC 0.72) but only modest predictive ability (AUC 0.58) in the independent sample (Figure 1). Conclusions Our study demonstrates that a TRS derived from blood-based transcriptomics is able to distinguish between individuals with pulmonary-only and extrapulmonary sarcoidosis. The TRS is also comprised of biologically relevant genes, including a number of T cell receptor alpha subunit genes. Future studies are needed to prospectively validate our findings and potentially expanding their use to identify precision treatments for individuals with extrapulmonary sarcoidosis. This abstract is funded by: CPH is supported by K24HL173667. JLH is supported by R01NS128535 and the CNY Community Foundation. AJG is supported by K08HL168205 and the 2022 ATS/FSR Grant.