Comprehensive Genomic Profiling in Patients with Advanced Cancer in a Large US Healthcare System

Document Type


Publication Date


Publication Title

The Journal of Molecular Diagnostics; Association for Molecular Pathology 2020 Annual Meeting Abstracts




Introduction: The rapid evolution of precision medicine treatment options in cancer care has warranted broad molecular characterization of solid tumors. We are assessing the impact of implementing comprehensive genomic profiling (CGP) on a broad scale in a large community-based healthcare system on the management and outcomes of oncology patients. The present study describes genomic testing results and actionability in a large real-world cohort.

Methods: Patients diagnosed with advanced cancer were tested for genomic aberrations using small panel tests (2015 to 2017) or CGP tests (2018 to 2020) across multiple sites in the Providence St. Joseph Health system. Mutation data as well as deidentified electronic medical record data were abstracted for use in this study. Biomarker actionability was assessed based on current FDA and NCCN guidelines, OncoKB, and natural language processing (NLP)-based large-scale curation of investigational targets from the literature.

Results: This study included 4,244 genomic tests on 3,821 advanced cancer patients tested with CGP (n = 2,724) or small panel tests (n = 1,520). The median age of patients was 66 years; more than half of the patients were female (53%) and white (73%). The most frequently tested tumor type was lung (32%), followed by colorectal (13%), and melanoma (6%). More than half of the tumor specimens (51%) tested with CGP and 27% of specimens tested with small panels harbored a biomarker associated with an FDA-approved or standard-of-care therapy (NCCN-recommended, OncoKB levels 1 or 2, AMP/CAP/ASCO levels A or B). In addition, 72% of CGP-tested and 30% of small-panel-tested tumors had 1 or more biomarkers associated with clinical trial eligibility. NLP-based literature curation also revealed 64% of tumors tested by CGP and 52% by small panels had a biomarker associated with an investigational therapy. The number of variants of unknown significance (VUS) was 5 per report for CGP and 1 per report via small panel testing. CGP also assessed “genomic signatures” associated with tumor typeagnostic therapies: 2.5% of patients had high microsatellite instability and 10.4% exhibited a high tumor mutational burden (≥10 mut/Mb).

Conclusions: The present study demonstrates that CGP testing increases the proportion of patients with potentially actionable biomarkers across tumor types. In ongoing work, we are investigating the impact of CGP on biomarker-informed treatment decisions and patient outcomes.


Earle A. Chiles Research Institute