Classification of Somatic Variants in Solid Tumors Detected by Next-Generation Sequencing (ngs) and the Need for Clinical Guidelines.

Antonios Papanicolau-Sengos,Edward Hart,Wei Shen,Kenneth F. Grossmann,Ceclly Vaughn,Wade Samowitz,Allle Grossmann
DOI: https://doi.org/10.1200/jco.2014.32.15_suppl.1555
IF: 45.3
2014-01-01
Journal of Clinical Oncology
Abstract:1555 Background: NGS approaches to clinical mutation testing reveal previously uncharacterized, or poorly characterized, somatic variants. A major challenge in reporting is categorization of these variants in the absence of formal guidelines. The process is subjective and the potential for variation between molecular pathologists is significant. Methods: To better characterize variation between molecular pathologists we retrospectively reviewed 95 clinical cases sequenced with the Ion Ampliseq Cancer Hotspot Panel v2 at ARUP Laboratories. We focused on variants that were reclassified upon re-review by three molecular pathologists. Results: Ninety eight mutations and 31 variants of unknown significance (VUS) were reported in 72 total tumors. There were seven reclassifications (5% of total classifications) in seven different tumors (3 urothelial carcinomas, 1 bladder adenocarcinoma, 1 melanoma, 1 lung carcinoma, and 1 unknown). Most of these were point mutations and involved FGFR1 (x1), FGFR3 (x1), CTNNB1 (x1), RB1 (x1) and PIK3CA (x3). Six of seven were reclassified from mutation to VUS based upon insufficient evidence of oncogenicity. Conclusions: Five of the reclassifications involved actionable genes (FGFR1, FGFR3, and PIK3CA). The original classification of these as “mutation” was primarily based on in vitro data. A “mutation” classification implies there is sufficient evidence for an oncogenic change that may be amenable to targeted therapy. In vitro biochemical and cell biologic assays are artificial systems which may not fully recapitulate in vivo tumor behavior. This study emphasizes the subjective nature of variant classification and the need for the development of weighted criteria devised by a multidisciplinary panel of experts. Until then, it is of utmost importance that molecular pathologists develop systematic methods to determine the quality of available data. We propose a simple checklist to optimize the way variants are classified. We suggest that medical oncologists critically scrutinize the classification of variants as they appear in clinical reports.
What problem does this paper attempt to address?