Magnifying Glass on Sinonasal NUT Carcinoma Heterogeneity via Spatial Transcriptomics.

Background: Nuclear protein in testis (NUT) carcinomas (NCs) are rare, clinically aggressive tumors with characteristic translocation involving NUTM1 and BRD4 genes. While NCs are generally characterized by undifferentiated basaloid cells with focal/abrupt squamous differentiation, tumoral heterogeneity remains unexplored. This may have therapeutic implications as NC frequently develops drug resistance and metastasis. Spatial transcriptomics (ST) sequencing technology emerged as an approach to elucidate tumoral heterogeneity by integrating cellular transcriptome and spatial distributions within tissues, providing insights into the interactions among different cell types and overall tissue genomic architecture.

Methods: Three cases of sinonasal NC with formalin-fixed paraffin embedded tissue formed the study material. A representative region of each NC was subjected to 10X Genomics Visium Spatial Gene Expression analysis. Resulting data were analyzed in 10X Genomics Loupe Browser.

Results: All cases demonstrated statistically distinct (graph-neural network based) transcriptomic clusters. These clusters were correlated with histologic characteristics by parsing all spots by region types based on tissue distribution including tumor, limited stroma, and normal epithelium. Cluster constitution and transcriptomic pattern were consistent with the morphologic features of each tissue region. NC#3 had a relatively low number of tumor clusters in comparison to NC#1 and NC#2.

Conclusions: Spatial transcriptomic profiling corresponds to histopathologic features in NC and yet highlights cell type diversity across tumors and in the case of NC#2, within a given tumor. Behavioral correlates are speculative, but NC#3 for which the patient had a rapidly lethal outcome showed a preponderance of neural gene expression which may be of interest in tumor cell progression/evolution.