Site-Specific Outcomes and Genomics in Metastatic Pancreatic Ductal Adenocarcinoma

ABSTRACT Background Clinical and molecular features of pancreatic adenocarcinoma (PDAC) according to involved metastatic sites have not been well characterized. Materials and Methods We used the Foundry platform to query information for patients with metastatic PDAC at MD Anderson Cancer Center. Involved metastatic sites were extracted using natural language processing from imaging reports and verified by chart reviews. Overall survival (OS) and next generation sequencing (NGS) data were analyzed. Results In a cohort of 1,095 patients (median overall survival [OS]: 22.8 months), lung-only metastasis was associated with longer median OS (57.6 months, p<0.0001). Patients with metastases not involving the liver had OS of 41.3 months, while liver-only (19.5 months) and multiple sites including liver (19.3 months) had the poorest survival. TP53 mutation rates were higher in liver-involved groups (84–85%) versus lung-only (73%) and non-liver metastases (67%, p<0.0001). Lower rates of GNAS mutations were detected in liver-only (5%) and liver-involved (2%) cases, higher in lung-only (8%) and non-liver (10%, p<0.01). The frequency of KRAS mutation and mutant allele distribution were not significantly different in tissue based NGS. However, in liquid biopsy analysis (n = 240), lung-only metastasis showed the lowest mutation detection (50%), TP53 (23%) and KRAS (10%), compared to liver-only detection rates of 65%, 46%, and 52% respectively (p values: 0.009, 0.02, <0.001). Conclusions Patients with PDAC without liver metastasis have significantly improved OS and lower rates of TP53 mutation. Patients with lung-only metastasis had longer OS, a lower liquid-biopsy positivity rate, and a lower rate of KRAS detection by liquid biopsy; because liquid biopsy may under-detect KRAS in lung-only disease, tumor-tissue sequencing should be prioritized in these patients as KRAS-directed therapies emerge.