Background: Actionable genomic alterations can be identified through either a biopsy of solid tissue or the detection of circulating tumor DNA from plasma. Little is known about the concordance rates of pathogenic variants between cell free DNA (cfDNA) and solid biopsies, including how concordance varies over time, by cancer type, or by treatment. Here we examine the concordance of pathogenic variants identified in solid tissue biopsies to patient-matched cfDNA biopsies in one of the largest pan-cancer datasets.
Methods: De-identified records of cfDNA (xF) and solid (xT) biopsies were analyzed in 2418 stage 4 patients across 5 cancer types: breast (N=459), colon (N= 564), non-small cell lung cancer (NSCLC) (N=750), pancreatic (N=353) and prostate (N=292). Patients were required to have ≥1 pathogenic variant detected in the solid biopsy and a cfDNA biopsy occurring on the same day or after the solid biopsy. Pathogenic SNVs and indels within overlapping probe regions of xF and xT meeting assay limits of detection were included for analysis. Only one cfDNA and solid biopsy were analyzed per patient.
Results: Overall, the total number of pathogenic variants identified in solid tissue and cfDNA were highly similar within each cancer type (breast 881 vs 831; colon 1983 vs 1673; NSCLC 1672 vs 1470; pancreas 873 vs 703; prostate 480 vs 384; solid biopsy vs. cfDNA, respectively). Subsetting to patients with ≥1 pathogenic variant in both assays (77%), >40% of patients had additional pathogenic variants identified in their cfDNA which were not found in their solid tissue profiling. Of these subsetted patients which also had cfDNA collected >1 year after their solid biopsy, >25% of NSCLC and >20% of breast cancer patients had mutually exclusive variants (all solid tissue variants were undetectable while new pathogenic variants were found in cfDNA). When comparing samples taken between one week and over one year from each other, the percentage of solid tissue variants identified in cfDNA decreased in a time dependent manner (breast 73% vs. 48%; colon 76% vs. 55%; NSCLC 75% vs. 43%; pancreas 76% vs. 36%; prostate 72% vs. 42%; ≤1 week vs. >1 year, respectively).
Conclusions: To our knowledge, this is the largest dataset of matched solid and cfDNA biopsies. >40% of patients with ≥1 pathogenic variants in both assays had additional pathogenic variants found only in their cfDNA. Further, cfDNA was capable of identifying the majority (>70%) of solid tissue variants when samples were taken within one week of each other, though this concordance decreased with time. This data strongly supports the use of matched solid and cfDNA testing, and further exploration of the clinical utility of cfDNA to identify pathogenic variants over the course of a patient’s disease.