Circulating tumor DNA predicts clinical benefits of immune checkpoint blockade in HER2-negative patients with advanced gastric cancer.

Background: Immune checkpoint inhibitors (ICIs) are becoming more prominent in the treatment of gastric cancer (GC). However, predictive biomarkers of response to ICIs in HER2-negative patients remain incompletely understood.

Methods: A total of 47 patients diagnosed with HER2-negative advanced GC who underwent ICI regimens were eligible for this study. Plasma samples with paired white blood cells prior to treatments were collected from these 47 patients. Variations of circulating tumor DNA (ctDNA) was evaluated by next-generation sequencing followed by its significance analysis.

Results: A total of 658 somatic mutations involving 203 genes were identified in all ctDNA. Mutations in MEN1, MLH1, CEBPA, ATR, GNAQ, and FOXL2 genes were more frequent in responders (P < 0.05). Compared with wild-type patients, patients with CEBPA or IRS2 mutations had prolonged median progression-free survival (mPFS, P = 0.0056). Patients with co-occurring mutations in IRS2/CEBPA, IRS2/POLD1, TP53/PIK3CA, or POLD1/CEBPA had longer mPFS compared with others (P = 0.003; 0.006; 0.0166; 0.0315; respectively). Both alteration of CDKN2A alone and co-mutations with MSH6 were significantly associated with superior overall survival (OS, P = 0.0289; 0.0355; respectively). In addition, higher on-treatment ctDNA concentration or variant allele frequency (VAF) were associated with poorer response (P < 0.05). Additionally, the increased molecular alterations of POLE, FGFR2 and MDC1 seemed to indicate the acquired resistance to ICIs.

Conclusions: Variation signatures captured by ctDNA as well as the kinetics of ctDNA could predict the clinical benefits of ICB in HER2-negative GC patients, which was worth further validated in large cohort.