Single-cell transcriptomic architecture for deciphering the complexity of tumor microenvironment in ampulla of Vater carcinoma.

e16474 Background: Ampulla of Vater carcinoma (AVC) is a rare gastrointestinal tumor that is associated with a high mortality rate and it’s often diagnosed at later stages due to lack of clinical symptoms. Elucidating complex ecosystems and molecular features of AVC is pivotal to proactive cancer prevention and optimal therapeutic intervention. Methods: We performed single-cell RNA sequencing (scRNA-seq) of treatment-naïve AVC biopsies (N = 8), and bulk RNA sequencing of surgically resected tissues (N = 62) to investigate the transcriptomic signature in tumor microenvironment (TME) of AVC. Results: We analyzed the single-cell transcriptome of 34,672 cells and epithelial cells were classified into 4 subtypes by consensus non-negative matrix factorization (cNMF) based on their gene expression profiles. Among them, the KRAS high subtype showed more increased copy number variation than other subtypes and was less differentiated with a high stemness score. Moreover, the KRAS high subtype was reversely related to overall survival (OS) and enriched with granzyme K + CD8 T cells. Deconvolution of bulk RNA-seq data validated the poor OS and progression-free survival (PFS) in the AVC patients with the KRAS high subtype. Also, it was revealed that this subtype was associated with early tumor recurrence in AVC. Conclusions: We understood the heterogenous TME of AVC and found the prognostic biomarker to predict postoperative recurrence and survival in AVC.