Lymph Node Transcriptomics in Idiopathic Multicentric Castleman Disease (iMCD) Identifies Pathogenic Mechanisms and Biomarkers Including Increased Clusterin Expression

Idiopathic multicentric Castleman disease (iMCD) is an atypical lymphoproliferative disorder with significant morbidity and mortality and a poorly understood pathophysiology. Defined by a characteristic lymph node (LN) histology, patients can present with a range of symptoms from thrombocytopenia, anasarca, fever, renal dysfunction, and organomegaly (iMCD-TAFRO) to thrombocytosis, hypergammaglobulinemia, and plasmacytosis (iMCD-IPL); and patients not falling into either group are considered iMCD-NOS. The molecular mechansims of iMCD, particularly within these clinical subtypes and in comparison to other inflammatory disorders, has not been elucidated. Additionally, there have been no distinguishing biomarkers identified and transcriptional programs governing pathology are largely unknown, especially in LN tissue, the site of defining histopathologic changes in iMCD.

We implemented a two-tiered analytical approach using bulk RNA-sequencing followed by targeted gene expression quantificiation to characterize the LN transcriptome(s) of iMCD clinical subtypes (iMCD-TAFRO, n=12, iMCD-NOS, n=4, iMCD-IPL, n=3), clinico-pathologically overlapping disorders (SLE, n=4), neoplastic conditions (DLBCL, n=5), and healthy controls (n=8). Using bulk RNA-sequencing, we found 425 genes were up-regulated, and 108 genes were down-regulated (log₂FC>±1, adj. P <0.05) in iMCD compared to healthy controls. Gene set enrichment analysis (GSEA) identified 18 enriched gene sets (normalized enrichment score, NES>1.5, and FDR q-val<0.1) including angiogenesis, mTORC1, IL-6-JAK-STAT3, and TNF signaling. The targeted gene expression data revealed shared DEGs across clinical subtypes of iMCD including SPP1 (a regulator of chemotaxis and cytokine production), XBP1 (a transcription factor driving plasma cell differentiation), and CLU (Clusterin, a follicular dendritic cell identifying marker). Importantly, CLU was uniformly upregulated in all iMCD clinical subtypes, but not upregulated in other inflammatory, neoplastic, or healthy controls. We validated Clusterin as a potential distinguishing biomarker by comparing Clusterin levels in iMCD lymph nodes (n=10) vs healthy controls (n=9) via immunohistochemistry (IHC). There was a 1.6X significant increase in cytoplasmic Clusterin in iMCD lymph node tissue (P=0.03). To determine potential pathogenic cell types within the LN tissue, we predicted cell type proportions in the LN using a computational pipeline that leverages single-cell transcript levels in different cell types from publicly available single-cell RNA sequenced LN samples (n=12) to predict the cell type proportions in bulk RNA sequencing data. This analysis predicted an increased proportion of plasma cells (P=0.007) and monocytes (P=0.05) in iMCD-TAFRO LN tissue. We then identified potential therapeutic targets that reverse the gene signatures in iMCD LN tissue using the LINCS1000 database, a database of gene expression profiles after pharmaceutical intervention at different time points and doses. A VEGF inhibitor, Tivozanib, and an NF-kB inhibitor, Parthenolide, were two of the top therapies predicted to reverse the iMCD-TAFRO gene signature, neither of which have been reported in the literature for use in iMCD-TAFRO.

Taken together, we have identified biomarkers of iMCD-TAFRO LNs such as Clusterin, gene pathways that may be driving iMCD pathology, and predicted treatments based on these analyses, such as Tivozanib. We also validated that a biomarker, Clusterin, is indeed upregulated in iMCD LNs compared to healthy controls. Clusterin may differentiate iMCD from other inflammatory and neoplastic conditions, but further work needs to be done to confirm this finding. We also used in-silico methods to identify increased numbers of plasma cells and monocytes in iMCD LN tissue, and identified VEGF and TNF inhibition as potential therapeutic invterventions in iMCD.