Genome-based classification of pancreatic acinar cell carcinoma reveals similarities to KRAS wild-type PDAC

Pancreatic acinar cell carcinoma (ACC) is a rare primary neoplasm of acinar cell origin. Histologically, it can present as pure ACC or mixed with neuroendocrine or ductal differentiation. We studied whether ACC shared genomic similarities with pancreatic ductal adenocarcinoma (PDAC) or well-differentiated pancreatic neuroendocrine tumors (PNET) using the hidden genome classifier methodology. The hidden genome classifier was trained using next-generation sequencing of KRAS wild-type (KRAS-WT) PDAC (n = 251), KRAS-mutated (KRAS-mut) PDAC (n = 1872) and PNET (n = 127). We applied the classifier to 62 primary and metastatic ACC samples from two institutions to identify the genomic class and its relationship to histologic differentiation and clinical outcomes. According to the classifier, KRAS mutations and tumor mutation burden contributed to the KRAS-mut PDAC class, while ATRX, VHL and SETD2 mutations contributed to the PNET class. Copy number alterations (CNA) in chromosomes 1q and 11q were key factors in the genomic classification of KRAS-WT PDAC. Most ACC (49/62) aligned genomically with KRAS-WT PDAC. Fifteen ACCs had mixed histology, and the majority of those (11/15) also classified as KRAS-WT PDAC. CNA were present in most ACC tumors, even those without mutations in DNA damage repair or chromatin modification pathways. The predicted genomic class was not associated with survival outcomes. The hidden genome classifier found most ACC shares genomic similarities with KRAS-WT PDAC, independent of ductal or neuroendocrine differentiation. This may reflect a common cell of origin in ACC tumorigenesis and explain the improved clinical outcome compared to KRAS-mut PDAC.