Molecular landscape and site of metastasis in PDAC.

person Shafia Rahman The Ohio State University Comprehensive Cancer Center, Columbus, OH info_outline Shafia Rahman, Yasmine Baca, Harshabad Singh, Joanne Xiu, Atrayee B Mallick, Mark P. Rubinstein, Andrew Aguirre, George W. , Michael J. Pishvaian

Authors person Shafia Rahman The Ohio State University Comprehensive Cancer Center, Columbus, OH info_outline Shafia Rahman, Yasmine Baca, Harshabad Singh, Joanne Xiu, Atrayee B Mallick, Mark P. Rubinstein, Andrew Aguirre, George W. , Michael J. Pishvaian Organizations The Ohio State University Comprehensive Cancer Center, Columbus, OH, Caris Life Sciences, Phoenix, AZ, Dana-Farber Cancer Institute, Brookline, MA, Sidney Kimmel Medical College, Thomas Jefferson University, Philadelphia, PA, The Ohio State University, Columbus, OH, Dana-Farber Cancer Institute, Boston, MA, Johns Hopkins University School of Medicine, Washington, DC Abstract Disclosures Research Funding No funding sources reported Background: More than 50% of patients with Pancreatic ductal adenocarcinoma (PDAC) have metastatic disease at the time of diagnosis and liver is the most common site of metastatic spread. Liver metastasis (LM) is associated with poor prognosis. Herein we examine the difference in molecular landscape of PDACs with LM versus other metastatic sites (OM). Methods: Total of 7,979 PDAC tumors underwent next-generation sequencing of DNA (592-gene/whole exome) and RNA (whole transcriptome) at Caris Life Sciences (Phoenix, AZ). Tumors were then evaluated and divided into LM (N=4988) site vs OM (N=3073) sites based on tissue specimen sites. RNA expression data was used to analyze transcriptional signatures as well as tumor immune microenvironment (TME) using Quantiseq. Real-world overall survival (rwOS) information was obtained from insurance claims data and calculated from time of collection or first treatment to last contact. Hazard ratio (HR) was calculated using Cox proportional hazards model, and P values were calculated using log-rank test.Significance for molecular comparisons was calculated using either chi-square, Fisher’s exact, or Mann-Whitney U test, with p-values adjusted for multiple comparisons (q<0.05). Results: Mutations in TP53 (81% vs 67%), KRAS (88% vs 83%), ARID1A (13% vs 11%), KDM6A (4% vs 3%), and BRCA1 (1.3% vs 0.7%) were all significantly higher in LM vs OM (all <0.05). Conversely, TMB-H (3% vs 4%), MSI-H (0.7% vs 1.3%), GNAS (1% vs 3%), STK11 (1% vs 3%), ATM (3% vs 5%), PTPRD (2% vs 9%) and RNF43 (5% vs 6%) mutations were significantly lower in LM vs OM (all q<0.05). In the TME, B cells, Tregs, M1, M2 macrophages, and NK cell infiltration was lower in LM vs OM (FC: 0.62-0.91, q<0.05). However, LM had higher neutrophil infiltration vs OM (FC: 1.1, q<0.05). Both interferon-gamma score (IFG) and T cell inflamed score (TIS) were significantly higher in OM vs LM (q<0.05). OM had better OS than LM (11.4 m vs 6.8 m, HR=0.65 95% CI: 0.62-0.68, p<0.001). This association held for tumors treated with Immune Checkpoint Inhibitors (ICI) (9.6 m vs 4.6 m, HR=0.71 95% CI: 0.52-0.97, p=0.031), Gemcitabine/Nab-paclitaxel treated (15.3 m vs 8.1 m, HR=0.57 95% CI: 0.52-0.61, p<0.001) as well as mFOLFIRINOX treated tumors (20.3 m vs 12.0 m, HR=0.57 95% CI: 0.52-0.63, p<0.001). Conclusions: When comparing pancreatic LM to OM sites, our data reinforces the observation that OS is better in OM vs LM and response to ICI was better in OM vs. LM. Significant differences were observed in molecular landscape, TME and signatures that are predictive of immunotherapy response (TIS and IFG scores). Test Positive N in LM Total N - LM % LM Positive N in OM Total N - OM % OM q-value TP53 mut 3391 4195 80.8% 1631 2430 67.1% 0.0 GNAS mut 42 4214 1.0% 82 2451 3.4% 0.0 KRAS mut 3709 4215 88.0% 2048 2456 83.4% 0.0 STK11 mut 51 4211 1.2% 71 2448 2.9% 0.0 PTPRD mut 3 143 2.1% 8 86 9.3% 0.02 BRCA1 mut 56 4187 1.3% 17 2446 0.7% 0.02 RNF43 mut 194 4217 4.6% 144 2456 5.9% 0.04 ARID1A mut 435 3399 12.8% 205 1917 10.7% 0.04