Prognostic and predictive impact of circulating tumor DNA in advanced gastric cancer treated with immune checkpoint blockade.

Abstract

e16019 Background: The utility of circulating tumor DNA (ctDNA) as a biomarker in patients with advanced gastric cancer (GC) receiving immune checkpoint blockade (ICB) is uncertain. Methods: Blood samples from advanced GC patients who received ICB were collected. The next-generation sequencing(NGS)–based targeted 733-gene and 61-gene panels (3D Medicines Inc.) were used to characterize somatic genomic alterations for baseline and on-treatment blood samples (4-12 weeks) respectively. For each sample, somatic single-nucleotide variants (SNV) and insertion/deletion mutations (indels) were used to calculate the mean variant allele frequency (VAF) and total mutation count. A VAF of 0.1% was the limit of detection, with the exception of variants detected in both pretreatment and on-treatment samples such that variants with VAF < 0.1% in one sample were retained if the exact variant was also detected in the paired sample with VAF ≥0.1%. The variant genes at baseline were filtered against 61-genes panel list for calculating paired-baseline-VAF. The delta-VAF was calculated per patient as the sum of the on-treatment VAF minus the paired-baseline-VAF for each detected SNV or indels divided by the number of detected SNVs or indels. The percentage of molecular response was calculated per patient as delta-VAF divided by paired-baseline-VAF. Results: A total of 16 patients were was enrolled. The detection rate was 93.8% (15/16). The median total mutation count and baseline-VAF were 8 (range 0-44) and 3.13% (range 0-40.4%) respectively. The patients with top 50% baseline-VAF had shorter median PFS, compared to that with bottom 50% baseline-VAF (2.55 vs 4.85 months, p = 0.022, HR 2.98, 95%CI: 0.96-9.26). For baseline blood samples (N = 16), the median total mutation count and paired-baseline-VAF was 4 and 3.89% (range 0-40.4%) respectively. For on-treatment blood samples (N = 10), the median total mutation count and VAF were 3 (range 1-12) and 5.94% (0.30-37.12%) respectively. All the patients with progressive disease (PD, N = 6) have increased delta-VAF and all patients with stable disease (SD, N = 3) or partial response (PR, N = 1) have decreased delta-VAF. The median delta-VAF for PR, SD and PD were was -6.73%, -2.13% and 3.95% respectively. Compared with paired-baseline and on-treatment-VAF, the median percentage of molecular response for PR, SD and PD were -95.53%, -47.66% (range 7.08%-59.07%) and 138.73% (range 66.40%-347.18%). Compared with patients with increased VAF during treatment, the patients with decreased VAF have longer PFS (median 3.9 vs 1.75 months, p 0.004, HR, 0.24, 95%CI 0.06 to 0.97). Conclusions: This preliminary analysis contributes to understanding the role of ctDNA as a prognostic and predictive biomarker and its potential to complement radiologic endpoints. Further studies are warranted in larger cohort to confirm such observations for better management of GC.