Background:

Idiopathic multicentric Castleman Disease (iMCD) is a rare polyclonal lymphoproliferative disorder of unknown etiology with an annual incidence of approximately 1500 individuals in the United States and a 25-35% 5 year mortality rate. iMCD is characterized by a wide range of clinical and pathological features including multifocal lymphadenopathy with unique histopathologic features and cytokine-driven inflammation. Within iMCD, there exists multiple clincal phenotypes. Patients presenting with thrombocytopenia, anasarca, fever/elevated C-reactive protein (CRP), reticulin myelofibrosis, renal dysfunction, and organomegaly have been defined as  iMCD-TAFRO. iMCD with idiopathic plasmacytic lymphadenopathy (iMCD-IPL) is defined by thrombocytosis, hypergammaglobulinemia, and a milder clinical course. Patients not meeting these clinical criteria are classified as iMCD–not otherwise specified (iMCD-NOS), which is most often characterized by milder clincal features. Patients can also be classified by histopathological subtype, including hyaline vascular/hypervasular, plasmacytic, and a mixed variant that includes features of both, although the clinical implcations of histopathological subtypes remains unclear.

The specific etiology of iMCD is not well understood and the heterogeneous presentation of iMCD may argue for multiple etiologies being able to contributing to the phenotype. Given that the systemic inflammation and multiorgan involvement in iMCD is quite non-specific, many features of iMCD are observed in other inflammatory disorders as well as lymphomas, although gene expression differences in lymph node tissue between these clinico-pathologically related disorders have not been thoroughly explored. In a subset of patients, the systemic inflammatory process is driven by interleukin-6 (IL-6), a pro-inflammatory cytokine that leads to activation of signaling pathways associated with cell survival and proliferation. The discovery of IL-6 as a disease driver in a subset of iMCD patients led to the development and use of siltuximab, a monoclonal antibody directed against IL-6, which remains the only US Food and Drug Administration (FDA) approved therapy. However, 66% of patients with iMCD do not respond to anti-IL6 therapy suggesting potential overlapping and undiscovered disease mediators and treatment targets in iMCD9,10.

Most discovery based investigations of gene expression networks or proteomic signatures involved in iMCD have been done in circulating peripheral blood mononuclear cells (PBMCs) or in plasma/serum. In a large serum proteomics based study involving 88 iMCD patients and a number of clinico-pathologically related disorders, PI3K-AKT-mTOR, angiogenesis, IL-6-JAK-STAT3, and CXCL13 were some of the most enriched pathways and up-regulated cytokines in iMCD serum. In PBMCs,  interferon stimulated genes (ISGs) were found to be upregulated and strongly correlated with mTOR signaling in a number of immune cell types in iMCD patients. More recently, Wing et. al and Horna, et. al explored gene expression changes in MCD lymph node tissue compared to healthy controls. These studies identified plasma cell differentiation (XBP1), follicular dendritic cell markers (CXCL13, clusterin), angiogenesis (VEGF, VEGFR), mTORC1 pathway genes, and complement cascade signatures that were significantly increased in MCD lymph node tissues. However,  sample size was limited in these studies, HHV-8 status was not defined in Wong et al, and subsetting based on clinical manifestations of MCD was not performed in either study.

Here, we report the investigation of the lymph node transcriptome in multiple iMCD subtypes (iMCD-TAFRO, iMCD-IPL, iMCD-NOS) and compare these profiles to both healthy controls and clinico-pathologically overlapping autoimmune (systemic lupus erythematosus, SLE) and neoplastic (diffuse large B cell lymphoma, DLBCL) conditions. Characterizing the lymph node transcriptome of iMCD subtypes, related disorders, and healthy controls.

Research Methods:

Patient samples and histologic review

We identified samples from 19 patients enrolled in the ACCELERATE natural history registry (NCT02817997) that met clinical and laboratory criteria for iMCD based on current international diagnostic criteria. Based on clinical and laboratory data collected at time of diagnosis and based on current guidelines, patients were categorized into either iMCD-TAFRO (n=12), iMCD-IPL (n=3), or iMCD-NOS (n=4) subtypes. Additionally, 8 sentinel lymph node samples with histomorphological features demonstrating unremarkable lymph node tissue were used as controls.  We obtained lymph node biopsy samples from 9 patients with comparator diseases, including diffuse large B cell lymphoma (DLBCL, n=5) and  systemic lupus erythematosus (SLE, n=4). These samples were obtained from a University of Pennsylvania pathology core service.

RIN (RNA integrity) values  were used to select a targeted cohort of high quality RNA samples that were used for bulk RNA sequencing as described below. For the remainder of the FFPE samples comprising the iMCD subtypes and related inflammatory disorder samples, a targeted gene expression quantification approach, Nanostring nCounter, was used.

Gene Expression Quantification

Bulk RNA sequencing

Libraries for whole transcriptome RNA sequencing were prepared using the Stranded Total RNAseq with Ribo-zero Plus kit (Illumina, San Diego,CA) as per manufacturer’s instructions starting with an input of 100ng of RNA and 15 cycles of final PCR amplification. Library size was assessed using the 2100 Bioanalzyer and the High-Sensitivity DNA assay (Agilent, Santa Clara, CA). Concentration was determined using the Qubit Fluorometer 2.0. Next Generation Sequencing with a paired-end 2x100bp run length was done on the NextSeq2000 platform (Illumina, San Diego, CA). A minimum of 30M reads per sample was acquired for each sample.

Nanostring nCounter

RNA was isolated from Formalin Fixed Paraffin Embedded (FFPE) tissue sections fixed on slides using the RNeasy DSP FFPE kit (Qiagen, Hilden, Germany). Briefly, sections 10 microns thick, were scraped from 5 slides of each sample and submerged in Deparaffinization Solution (DSP) as per protocol. During this process tissue digestion using Proteinase K was also done. DNAseI treated RNA was subsequently isolated using Qiagen’s proprietary spin column method as per protocol. RNA integrity was assessed using the TapeStation RNA ScreenTape (Agilent, Santa Clara, CA) and concentration was determined used the Qubit 2.0 Fluorometer (ThermoFisher, Waltham, MA). Since FFPE RNA is expected to be highly degraded, a DV200 value, which is the percentage of RNA fragments above 200bp, was determined for each sample.

100ng of RNA fragments 200bp and over were used for the Nanostring nCounter assay (Nanostring, Seattle, WA).  Briefly, RNA was hybridized at 65C for 19 hours with proprietary fluorescently labeled reporter and capture probes to bind target genes. The hybridized mix of probes and target genes were washed, enriched, and isolated using the Nanostring Prep station and scanned for florescent intensity using the Nanostring Digital Analyzer. The nCounter Immunology panel (v2) was used to quantify 594 genes that are known to be involved in autoimmune/immune response and also including 15 internal reference control genes to aid in normalization.

Informatic analysis and differential gene expression and gene set enrichment analysis

Bulk RNAseq

RNA-seq reads were demultiplexed using bcl2fastq (Illumina, San Diego). Demultiplexed FASTQ files were aligned to Human reference GRCh38 using the STAR aligner16 (v.2.6.1) using default settings. Generated BAM files were read in to the R statistical computing environment, and gene counts were quantified using the Genomic Alignments package17. Preliminary QC and differential expression analysis was performed using the R/Bioconductor package DESeq218. Read counts were filted to include only genes with a read count >=20 in greater than 33% of samples. Patients were grouped by phenotype (iMCD-TAFRO, iMCD-NOS, SLE, DLBCL, and Sentinel) and differentially expressed genes were considered for further investigation using a log2FC > -/+1 and a Benjamini Hochberg corrected P-value <0.05.

Nanostring RNA counting assay

Reporter code counts (RCC) were read into the R computing environment using NanostringQCPro19. Quality control and normalization were performed using RUVSeq20,21. After normalization, DESeq2 was used to perform differential expression analysis and gene level exploration. Significantly differentially expressed genes were selected based on -/+1 log2FC and  unadjusted P-value <0.05 thresholds.

Gene Set Enrichment Analysis (GSEA)

Gene Set Enrichment analysis22 (GSEA, v.4.3.2) was performed for the bulkRNAseq iMCD-TAFRO (n=7) vs. Sentinel lymph nodes (n=7) using phenotype permutation settings. Gene sets were considered significant if FDR q-val <0.25. Several gene sets databases were used from the Molecular Signatures Database (MSigDB) gene set collection23 including gene ontology-molecular function (GO-MF), gene ontology biological process (GO-BP), Biocarta, Hallmark gene sets, Kyoto Encyclopedia of Genes and Genomes (KEGG)24, Reactome pathways25, and Wikipathways26.

Estimation and comparison of cell type composition in bulk RNAseq data

Single-cell RNA sequencing data from lymph node tissue of 12 deceased organ donors was downloaded from the CellTypist immune atlas (Celltypist.org, accessed June 30, 2023)27. iMCD bulk RNA sequencing samples were deconvoluted into cell types using Bisque to estimate cell type composition using a non-negative least-squares regression model28. Cell type composition comparisons between iMCD and Sentinel samples were performed using a T-test to determine statistical significance.

Querying LINCS1000 database to predict drug perturbation efficacy

Immunohistochemistry

Formalin fixed/paraffin-embedded tissue sections were immunostained on a Benchmark XT autostainer (Ventana Medical Systems, Inc, Tucson, AZ), using a primary anti-C4d rabbit polyclonal antibody (American Research Products, Belmont, Massachusetts, catalog #: 12-5000) or a primary anti-CXCL13 mouse monoclonal antibody (R&D systems, Minneapolis, MN, clone 53610), as previously described.