Abstract / Summary
Circulating tumor DNA (ctDNA), obtained serially during cancer treatment, can capture tumor burden trends to help interpret response and detect resistance to systemic therapy for patients with advanced disease. Traditionally, tumor-informed assays track somatic variants identified from an initial tumor biopsy and have demonstrated prognostic utility for early-stage disease. However, tumor-informed ctDNA assays are potentially limited by logistical and methodologic considerations in the setting of advanced disease. In contrast, methylation-based tumor-agnostic assays offer an alternative approach, interrogating thousands of differentially methylated regions that are broadly shared across cancer cells, without requiring a bespoke panel. A comparison of the two approaches for monitoring ctDNA in advanced cancer is lacking. In this study, a head-to-head comparison was performed between commercially available tumor-agnostic and tumor-informed ctDNA assays. The quantitative correlation of ctDNA between the two assays was determined to be high ( R 2 = 0.79 ) across 99 samples from 33 patients, and controlling for patient-specific platform difference, resulted in a mixed-effect analysis conditional R 2 = 0.97 . Interval ctDNA dynamics were categorized as increase, decrease or no-change/negative, with a concordance of 67/71 (94%) between the paired assays. Discordances between tumor-informed and tumor-agnostic ctDNA results were investigated further by clinical correlation. In all discordant cases, the clinical correlation supported tumor-agnostic ctDNA results, indicating that tumor-informed ctDNA findings were either false-positives or false-negatives. Overall, tissue-informed and tissue-free ctDNA assays were highly consistent in measuring both quantitative ctDNA level and qualitative directional changes. The superiority of the tumor-agnostic ctDNA approach was demonstrated in several discordant cases where its utility in detecting metastatic tumor evolution provided a practical advantage.