COMPREHENSIVE GENOMIC PROFILING OF PANCREATIC CANCER TUMOR SPECIMENS: IS MORE BETTER?
Ashley Krepline*1, Mohammed Aldakkak1, Kathleen Christians1, Ben George3, Paul S. Ritch3, William Hall2, Beth A. Erickson2, Douglas B. Evans1, Susan Tsai1
1Surgery, Medical College of Wisconsin, Milwaukee, WI; 2Department of Radiation Oncology, Medical College of Wisconsin, Milwaukee, WI; 3Division of Hematology/Oncology, Medical College of Wisconsin, Milwaukee, WI
Introduction: Comprehensive Genomic Profiling (CGP) is increasingly used to identify somatic alternations that have prognostic and/or predictive relevance. The predictive value of CGP in pancreatic cancer is unknown.
Methods: Surgical specimens from patients treated with neoadjuvant therapy for pancreatic cancer sent to Caris Life Science for CGP were identified from 2016-2017. Genetic variants identified by Next Generation Sequencing (NGS) were classified as benign (B), likely benign (LB), variant of unknown significance (VUS), likely pathogenic (LP), or pathogenic (P). Somatic variants with a potential therapeutic target were identified from the company's molecular profiling report.
Results: The commercial testing consisted of either a 472-gene or 46-gene NGS panel. Of the 472 genes only 9 (1.9%) genes had actionable targeted agents: ATM, BRAF, BRCA1, BRCA2, c-KIT, Her2/Neu, PDGFRA, PIK3CA, and RET. Genomic testing using the 472-gene panel was completed in 49 patients. Of the 472 genes tested, 80 (16.0%) genes had at least one variant identified: 3 (0.6%) B, 20 (4.2%) LB, 65 (13.8%) VUS, 3 (0.6%) LP, 13 (2.8%) P. LP or P variants were identified in 14 (3.0%) genes (Table 1). Available targeted therapy was identified for 4 (0.8%) genes involving the homologous recombination DNA repair pathway (BRCA1, BRCA2, ATM) and 1 (0.2%) gene involving the mTOR (PIK3CA) pathway. Of the patients identified with an actionable variant, no patient had more than one actionable variant. Based on NGS results, the most commonly recommended therapy was a platinum agent (n=12, 24.4%). Off-label treatments recommended by the NGS profiling included everolimus in 1 (2.0%) patient, mitomycin-C in 3 (6.1%), olaparib in 3 (6.1%), and temsirolimus in 1 (2.0%). One patient had both the 472-gene and 46-gene panel was completed from the same surgical specimen. Discordant results were identified in 3 genes: KRAS and TP53 P variants were identified on the 46-gene panel but not on the 472-gene panel and a JAK2 PB variant was identified on the 472-gene panel but not the 46-gene panel.
Conclusion: Using a 472-gene panel, only 3.0% of genes had a LP or P variant in patients with localized pancreatic cancer. Among the 49 patients, actionable variants were identified in 13 (26.5%). Platinum agents were the most common targeted agent identified. Further studies are needed to evaluate the cost-effectiveness of using NGS for pancreatic cancer, as the molecular profiling reports are unlikely to change therapy.
Table 1: Pathogenic and Likely Pathogenic Variants with Available Targeted Therapy.
| Gene | Number of Patients with Variant (%) | Agent Available | |
| Likely Pathogenic | ATM | 1 (2%) | Carboplatin Cisplatin Oxaliplatin |
| FH | 1 (2%) | No Agent Available | |
| TP53 | 6 (12%) | No Agent Available | |
| Pathogenic | ARID1A | 3 (6%) | No Agent Available |
| ATM | 8 (16%) | Carboplatin Cisplatin Oxaliplatin | |
| BRCA1 | 2 (4%) | Carboplatin Cisplatin Oxaliplatin Mitomycin C Olaparib | |
| BRCA2 | 1 (2%) | Carboplatin Cisplatin Oxaliplatin Mitomycin C Olaparib | |
| CDKN2A | 9 (18%) | No Agent Available | |
| GNAS | 1 (2%) | No Agent Available | |
| KRAS | 32 (65%) | No Agent Available | |
| MUTYH | 1 (2%) | No Agent Available | |
| PIK3CA | 1 (2%) | Everolimus Temsirolimus | |
| RNF43 | 2 (4%) | No Agent Available | |
| SMAD4 | 5 (10%) | No Agent Available | |
| SMARCA4 | 1 (2%) | No Agent Available | |
| TP53 | 21 (43%) | No Agent Available |
Back to 2019 Abstracts
