← Back to List

Program Number: AHNS16
Session Name: Scientific Session 4 - Cancer Biology
Session Date: Thursday, May 15, 2025
Session Time: 9:00 AM - 9:45 AM

scRNA-seq and genomics analyses reveal key mechanisms of inverted papilloma-associated sinonasal squamous cell carcinoma malignant transformation

Tolani Olonisakin, MD, PhD¹; Kenny Vu, MS²; Xinping Yang²; Aaron Udager, MD³; Clint Allen, MD²; Nyall London, MD, PhD²; ¹Johns Hopkins Medicine; ²National Cancer Institute/National Institutes of Health; ³University of Michigan

Background: Sinonasal inverted papilloma (IP) is a benign sinonasal tumor with malignant transformation to IP-associated sinonasal squamous cell carcinoma (IP-SNSCC) occurring in up to 10% of patients. The mechanisms underlying malignant transformation from IP to IP-SNSCC and the interplay between tumor cells and tumor immune microenvironment (TIME) in this transformation remain incompletely understood. We hypothesized that conducting the first single cell RNA sequencing (scRNA-seq) analysis of IP and IP-SNSCC would identify insights into these molecular mechanisms.

Methods: Patients with IP (n=12) and IP-SNSCC (n=4) were enrolled and tumor tissue and matched peripheral blood mononuclear cells (PBMC) were obtained. Fresh tumor tissue underwent tissue dissociation followed by single-cell capture (5’ GEM, 10X Genomics) and subsequent single-cell transcriptomic analysis. Publicly available normal sinonasal mucosa scRNA-seq datasets were integrated into the bioinformatics analysis to ascertain tumor cell of origin and evaluate mechanisms of transformation from normal mucosa, to IP, to IP-SNSCC. Clustering and differential expression analysis were performed using Seurat. Snap frozen tumor tissue and matched PBMCs underwent whole-exome sequencing (WES). Somatic mutations and copy number variations were detected from WES. Low-risk human papillomavirus in situ hybridization was performed on tumor tissue.

Results: As previously reported, WES identified frequent EGFR exon 20 mutations in IP tissue. Stop-gain mutation or copy number variant loss in CDKN2A were defining genomic features of IP-SNSCC transformation. Using scRNA-seq analysis with clustering (Figure 1) and differential gene expression (Figure 2), IP-SNSCC were found to be highly enriched with a variety of highly activated immune cells, in particular myeloid cells and T cells. Of the highly upregulated tumor-derived genes, CXCL14 emerged as the only inflammatory cytokine upregulated in IP-SNSCC tumor cells compared to IP tumor cells (p = 1.90E-118) (Figure 3), suggesting that CXCL14 may be a key driver of IP to IP-SNSCC transformation and immune cell activation. Lastly, integration of normal sinonasal mucosa datasets supported basal cells of the respiratory epithelium as the cell of origin for IP development.

Conclusions: The highly conserved chemokine CXCL14 was found to be the sole significantly differentially upregulated inflammatory cytokine or chemokine in IP-SNSCC tumor cells when compared with IP cells. As a secreted protein, CXCL14 is a highly druggable target with potential for development of neutralizing antibodies, small molecule inhibitors or exogenous mimetic. Further evaluation is necessary to ascertain the direct mechanistic impact of CXCL14 on the TIME during malignant transformation.

 

Figure 1: Cell populations in IP-SNSCC tumor microenvironment. Uniform manifold approximation and projection (UMAP) plot showing cell clusters-malignant epithelial “pap_tumor” cells, endothelial cells, fibroblasts, pericyte, T cells, B cells, mast cells, myeloid cells, and macrophages-based on gene expression in IP-SNSCC relative to IP. Notation “Myeloid.0, Myeloid.1,…” indicates subpopulations within each cluster.

 

 

Figure 2: Differential gene expression by cell cluster. Heat map depicting the top 10 differentially expressed genes for each cell cluster.

 

 

Figure 3: Volcano plot of tumor cell gene expression in tumors obtained from IP (n = 12) and IP-SNSCC (n = 4).

 

← Back to List