Genetic and immune divergence in pancreatic cancer lesions at the time of metastatic relapse compared to primary tumor.

Abstract

4140 Background: Relapse of pancreatic ductal adenocarcinoma (PDA) is common even after complete resection and adjuvant therapy. Compared to the resected tumor, the biological characteristics of metastatic tumors at the time of first relapse are poorly understood. Methods: Whole-exome sequencing (WES) (250x) and bulk RNA sequencing were performed on samples from 30 patients with PDA. Paired primary tumor samples were obtained after R0 or R1 resection, and metastatic tumor samples were obtained by biopsy at the time of first relapse. 74.1% of patients had received adjuvant chemotherapy and radiation therapy, 7.4% had received adjuvant chemotherapy only, and 3.7% had received adjuvant radiation therapy only. Most common metastatic sites were liver and lung. The cohort was 60% male with a median age at diagnosis of 64 years. The vast majority of patients had stage IIA or IIB disease at diagnosis. Median disease-free survival was 481 days. Analysis used Freebayes (somatic variant calling), Kallisto (transcript quantification), Danaher et al. (cell type deconvolution), and antigen.garnish (neoantigen prediction). Results: High-quality WES and/or RNA sequencing were available for 27/30 patients. Among these were 16 pairs of primary and metastatic samples for WES and 15 paired samples for RNA sequencing. Median tumor purity was 32% (primary) and 42% (metastatic). KRAS mutations were present in 43/48 evaluable samples, with conserved KRAS mutations in 14/16 primary-metastatic pairs. Tumors were otherwise highly variable, with 13/16 patients developing oncogenic mutations in metastatic tumors that were undetected in primary tumors (BRCA1 [3/16], AKT3 [3/16], TP53 [2/16], ROS1 [2/16]). Overall, primary and metastatic tumors had similar tumor mutation burden and neoantigen production rate. However, neoantigens were highly variable at the peptide and gene level, with conservation rates of 2.73% and 11.57%, respectively, across primary-metastatic pairs. PDA transcriptomic subtype also differed across primary-metastatic pairs in all cases. Furthermore, metastatic tumors contained lower immune suppressive signal by transcripts and deconvolution (CTLA4: p = 0.0012, FOXP3: p = 0.0026, PDCD11: p = 0.012, regulatory T cells: p = 0.012), while myeloid cells were higher (CD33: p = 0.0067). Conclusions: With the exception of KRAS, metastatic PDA tumors at relapse contain new oncogenes, distinct neoantigens, and lower immune-suppressive signal compared to primary PDA tumors. These data suggest a potential clinical utility for tumor biopsies at the time of first metastatic relapse and caution against clinical decisions for relapsed, metastatic patients based solely on sequencing of the originally resected tumor.