Back

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

Olonisakin, T. F.; Zamuner, F. T.; Vu, K.; Gunti, S.; Ding, Y.; Hou, Y.; Yang, X.; Viswanathan, R.; Huynh, A.; Comoglio, F.; Notaro, M.; Cherry, C.; Patatanian, M.; Wong, N.; DeKlotz, T. R.; Kim, J.; Koch, W.; Lane, A. P.; Ramanathan, M.; Rowan, N. R.; Lechner, M.; Udager, A.; Izumchenko, E.; Allen, C. T.; London, N. R.

2026-06-04 genomics
10.64898/2026.06.01.729345 bioRxiv
Show abstract

Sinonasal inverted papilloma (IP) carries a 10% risk of malignant transformation to IP-associated sinonasal squamous cell carcinoma (IP-SNSCC), yet the molecular and immune drivers of this progression remain poorly defined. This study integrates single-cell RNA sequencing, multiplexed spatial proteomics, whole-exome sequencing, and functional assays across IP and IP-SNSCC cohorts to define mechanisms of malignant transformation. Respiratory epithelial basal cells are identified as the putative cell of origin, with progression marked by recurrent CDKN2A loss and TP53 mutations. Spatial profiling reveals immune reorganization at the lesional interface in IP-SNSCC, characterized by enrichment of alternatively activated M2 macrophages. CXCL14 is shown to directly induce an immunosuppressive myeloid phenotype that suppresses T-cell IFN{gamma} production through an IDO-pathway-dependent mechanism. Integration of these multimodal datasets defines a previously unrecognized CXCL14-IDO mechanism that constrains anti-tumor immunity at the tumor-stroma interface. These findings establish IDO-targeted immunomodulation as a rational adjuvant strategy and provide a comprehensive molecular framework for understanding IP-SNSCC pathogenesis.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.