Clinical, genomic, and spatial transcriptomic features of pancreatic adenosquamous carcinoma support subtype-specific...

Pancreatic adenosquamous carcinoma (ASQ) is a rare variant with poorly defined clinical and molecular features compared to conventional pancreatic ductal adenocarcinoma (PDAC). This study aimed to characterize the clinical outcomes, genomic landscape, and spatial transcriptomic architecture of ASQ.

We retrospectively evaluated clinicopathologic features, treatment patterns, and survival of ASQ patients treated at Johns Hopkins from 2013-2023 (n=178). Genomic landscapes of ASQ (n=244) versus PDAC (n=29,021) were compared using the FoundationCore database. Surgically resected ASQ tumors (n=11 from 8 patients) were analyzed using Visium spatial transcriptomics with a comparator PDAC dataset (n=4).

ASQ patients had poor outcomes in both resected and advanced disease. FoundationCore analysis revealed similar KRAS mutation frequency, higher MTAP loss, and enrichment of low-frequency immune biomarkers (MSI-high, PD-L1+, TMB ³10 muts/Mb) in ASQ compared with PDAC. Spatial transcriptomics revealed broadly similar immune cell proportions across subtypes, with higher regulatory T-cell abundance observed in ASQ in the context of sparse detection in PDAC. Differences in spatial co-localization patterns of immune populations were observed between ASQ and PDAC, while distance analyses within ASQ showed immune infiltrates closest to squamous regions and farthest from glandular regions. Furthermore, ASQ showed increased squamous/basal lineage expression compared with PDAC, with intratumoral squamous regions enriched for epithelial-mesenchymal transition, apical junction, inflammatory, and stress response pathways relative to the glandular niche.

Despite shared genomic drivers with PDAC, ASQ exhibits transcriptionally distinct features with spatially organized tumor and immune heterogeneity, supporting the need for subtype-specific therapeutic strategies.