Spatial transcriptomic landscape of invasion patterns in human papillomavirus-associated endocervical adenocarcinoma

Human papillomavirus (HPV)-associated endocervical adenocarcinoma is the second-most common cancer of the uterine cervix. HPV-associated endocervical adenocarcinoma can be classified into histologic Silva patterns of invasion, which are associated with clinical outcome. However, the mechanisms underlying these patterns of invasion are incompletely understood. We used whole transcriptome spatial transcriptomics to examine gene expression differences separately in the tumor epithelium and the surrounding stromal immune microenvironment (SIME). Seven cases were evaluated, focusing on cases with two distinct patterns of invasion within the same tumor, to control for inter-patient heterogeneity. The most strongly upregulated pathways in both higher-risk tumor epithelium and SIME were associated with extracellular matrix (ECM) remodeling. Transcriptomic-based inference of immune cell populations showed an increase in macrophage populations in higher-risk tumor areas, confirmed by immunohistochemistry. Finally, we derived a four-gene signature from genes upregulated in higher-risk tumor epithelium (KRT6A, TNC, LAMC2 and FN1), which was associated with worse clinical outcome in an independent dataset (The Cancer Genome Atlas). Overall, this work demonstrates that ECM remodeling and macrophage presence are important in the progression to high-risk patterns of invasion in HPV-associated endocervical adenocarcinoma. In addition, we established a prognostic four-gene signature that is predictive of poor outcome.