Assessment of Circulating Tumor DNA Burden in Patients With Metastatic Gastric Cancer Using Real-World Data

PURPOSE Circulating tumor DNA (ctDNA) has emerged as a promising biomarker with prognostic and potentially predictive value for several tumor types, including gastric cancer (GC). This study uses real-world data (RWD) to investigate the association of pretreatment ctDNA burden with clinical outcomes among patients with first-line (1L)-treated metastatic gastric cancer (mGC) in the United States. METHODS Patients were identified from the GuardantINFORM real-world clinical-genomic database. Adult patients with mGC who underwent testing with the Guardant360 assay from June 2014 to March 2022 and within 60 days before 1L treatment were included. The median of the maximum variant allele fraction (MVAF) was used to classify patients into high or low ctDNA burden groups, with undetectable ctDNA burden included in the low group. Associations with real-world outcome variables derived from claims, including time to treatment discontinuation (rwTTD), time to next treatment (rwTTNT), and overall survival (rwOS), were assessed using log-rank tests and multivariable Cox models. RESULTS A cohort of 824 patients with mGC was identified. Median MVAF was 2.9%, with 91% having detectable ctDNA. Among patients receiving chemo-based treatment (n = 537), rwTTDs were similar in low and high ctDNA burden groups, while those with high ctDNA burden showed significantly shorter rwTTNT and rwOS (median rwTTNT = 4.8 months v 7.4 months, P < .001; median rwOS = 13.2 months v 19.1 months, P < .001). Multivariable Cox analyses showed similar results. ctDNA burden in immunotherapy (n = 100) and trastuzumab-based (n = 99) treatment groups did not have significant associations with outcomes. CONCLUSION We used RWD to demonstrate that high pretreatment ctDNA burden was associated with worse clinical outcomes in a mGC population receiving 1L chemotherapy-based treatments. Our analysis suggests ctDNA burden could be used as a prognostic biomarker for mGC.