Best oral presentation award: In situ detection of KRAS(G12D) mutations in the spatial context of pancreatic cancer subtypes

Introduction: Pancreatic ductal adenocarcinoma (PDAC) is known for its aggressive biology and lethality. Transcriptomics have identified two major subtypes in human PDAC, termed ‘classical’ and ‘basal-like’, with the latter having the worst prognosis. Expression of GATA6 is elevated in ‘classical’ tumors and considered a valuable marker to distinguish both subtypes. Interestingly, aggressive PDACs have been linked to the highest mutant KRAS levels, activating mutations in KRAS being the drivers of PDAC. Aims & Methods: Here, we aim at implementing a novel assay, BaseScope, that is based on RNAscope™ technology, to stain KRAS(G12D) in paraffin sections from human PDAC and matched organoids. We co-stain GATA6 and other markers, using RNAscope and immunofluorescence, for discriminating PDAC subtypes in their spatial context. BaseScope validation was performed in paraffin sections of pelleted human PDAC cell lines Su86.86 and BxPC3, that are respectively KRAS(G12D) mutant and wild type as previously determined by Sanger sequencing. Clinical samples were obtained from patients undergoing pancreatic surgery at UZ Brussels, Belgium (Ethics approval B.U.N. 143201941917). Next generation sequencing (NGS) for a hot spot panel was used to detect KRAS mutations and the tumors' molecular subtype was determined based on RNA sequencing. Image analysis is performed with HALO (Indica™) software. Results: In the PDAC cell lines, BaseScope for a positive control probe (PPIB, a housekeeping gene) was detected in ≥95% of cells while ≤2% showed signal for the negative control probe (DapB, a gene from soil bacteria). In the homozygous mutant cell line, 77% of cells stained positive for KRAS(G12D) while the KRAS wild-type cell line showed non-specific staining in 5% of the cells. The BaseScope assay was optimized for use on clinical samples resulting in detection of PPIB in 67±7% of cells (n=3) and limiting DapB detection to 5% of the cells (n=1). Using these conditions, KRAS(G12D) was detected in an NGS-confirmed KRAS(G12D) sample while not detecting it in a KRAS(G12R) or a KRAS wild type sample. Heterogenous expression of KRAS(G12D) was noted. A negative correlation was found between the percentage of cells per sample staining positive for GATA6 and positive for KRAS(G12D) (R=-0.5). In addition, there was a negative correlation in the amount of GATA6-positive dots per cell and the amount of KRAS(G12D) dots per cell (R=-0.7). Preliminary observations confirm this also in PDAC derived organoids. Interestingly, both phenotypes (GATA6 ;KRAS(G12D) and GATA6 ;KRAS(G12D) ) could be detected within single patient samples, especially those transcriptomically classified as basal-like, and also within single ducts, demonstrating intra-patient spatial heterogeneity. This observation is in line with recent literature showing coexistence of classical and basal-like regions in a single tumour and is confirmed by co-labelling of GATA6 and basal markers. Conclusion: Classical transcriptomic profiling destroys tissue architecture hampering the ability to study molecular heterogeneity at single cell level. Here, we report the feasibility to successfully stain KRAS(G12D) on clinical PDAC samples and relating it to molecular subtype markers. Spatial subtyping that incorporates in situ detection of mutations will provide critical insights into tumor heterogeneity and plasticity.