Identification of whole-blood gene expression signature in primary Sjögren's syndrome-associated lymphoma

Abstract

Primary Sjögren’s syndrome (pSS) is an autoimmune disease of the exocrine glands. The syndrome is characterized by many systemic features, including a markedly increased risk of lymphoma development. PSS-associated lymphoma was first reported in 1963, however, the mechanisms and the risk factors of lymphoma development in pSS patients remain incompletely understood. The aim of my project is to identify a whole-blood gene expression signature in pSS-associated lymphoma. To achieve this goal, I first evaluated the effect of the depletion of the abundant globin mRNA in whole-blood samples on microarray analyses of pSS. Then I prepared samples (globin mRNA depleted samples) from a “Discovery cohort” which consisted of five subject groups (“pSS (non-lymphoma)”, “pSS-associated lymphoma”, “pSS-paraproteinemia”, “pSS-other cancers” and “healthy controls”) to identify a list of differentially expressed genes (DEGs) between the “pSS (non-lymphoma)” and the “pSS-associated lymphoma” groups. The next step was to confirm the differential expression of these genes using qRT-PCR. This has led to the identification of a potential gene expression signature for pSS-associated lymphoma. To further explore the role of these genes in the pathogenesis of pSS-associated lymphoma, I performed pathway analysis using various algorithms provided by Ingenuity Pathway Analysis (IPA). I also compared the microarray data of different subject groups to investigate whether the potential gene signature was “specific” for pSS-associated lymphoma. I then validated the potential transcriptomic signature “biologically” using an independent cohort (the “Validation cohort”) consisting of two subject groups – “pSS (without lymphoma) and “pSS-associated lymphoma”. Moreover, the potential biosignature was tested in a group of pSS patients with untreated lymphoma. Prediction modelling was used to identify the important genes within the potential biosignature that best predict the development of pSS-associated lymphoma. I showed that globin mRNA depletion of whole-blood samples provided potentially more sensitive microarray data compared with paired non–globin RNA depleted samples. From the microarray analysis of the “Discovery cohort”, 68 DEGs were identified between the lymphoma and non-lymphoma groups (68-DEGs-Mi). qRT-PCR confirmed the differential expression of 26 genes (26-DEGs-MiPCR). Biological validation with an independent xv cohort verified 3 genes (3-gene biosignature), 2 of which were up-regulated (NUDT14, MGST3) and 1 gene was down-regulated (BMS1) in pSS-associated lymphoma. Moreover, 2 genes in addition to NUDT14 (DRAP1, DYNLL1) were also differentially expressed in a cohort of pSS patients with untreated lymphoma. Prediction modelling suggested that NUDT14 was the most important gene in predicting membership in the pSS-associated lymphoma group. Pathway analysis of the differentially expressed genes in pSS-associated lymphoma revealed several canonical pathways such as “Aryl Hydrocarbon Receptor Signalling,” “Histamine Degradation,” “Unfolded protein response,” “Neuregulin Signalling,” and “T Cell Receptor Signalling.” In addition, the Downstream Effects analysis revealed the biological functions in pSS-associated lymphoma and the Upstream Regulators analysis investigates possible gene regulators. Moreover, comparisons of microarray gene-expression data between other pSS subgroups suggest the DEGs were unique to pSS-associated lymphoma. IPA showed that “Interferon Signalling pathway” was the top canonical pathway in all pSS subgroups. Furthermore, similar patterns were seen in the IPA Downstream Effects analyses for the “pSS (non-lymphoma)”, “pSS-paraproteinemia” and “pSS-other cancers” groups, while the “pSS-associated lymphoma” group showed a unique pattern, further indicate that a unique gene expression signature exist in pSS-associated lymphoma.