Somatic alterations in circulating cell-free DNA of oesophageal carcinoma patients during primary staging are indicative for post-surgical tumour recurrence

Abstract

Oesophageal cancer (OC) has high mortality. This study aims at determining the feasibility of liquid biopsies for genomic profiling in early stage OC, comparing two different technologies for mutational analysis in circulating cell -free DNA (ccfDNA) and evaluating the clinical impact of these somatic alterations during primary staging. In 25 patients with locally advanced OC, endoscopic tumour biopsies and simultaneous blood samples were taken during primary staging. Genomic DNA from biopsies and ccfDNA were analysed for mutations using a 12 gene panel next-generation sequencing (NGS) assay as well as digital droplet PCR (ddPCR). Genetic data was correlated with patients' outcome. In 21 of the tested biopsies (84%) at least one somatic mutation was detected by NGS. Mutations detected by NGS were detectable by ddPCR with similar allele frequencies. In three out of the 21 patients with proven mutations, the same mutations were also detectable in ccfDNA using NGS (14%). In contrast, ddPCR detected mutations in ccfDNA of five additional patients (8/21, 38%). Post-surgical outcome analysis was performed for those patients who had received complete tumour resection (n = 16). Five of them suffered from an early relapse within the first year after surgery, including four with detectable somatic mutations in ccfDNA during primary staging. Taken together, we showed a higher sensitivity for ddPCR compared to NGS in detecting mutated ccfDNA in OC. Detection of somatically altered ccfDNA during primary staging seems to be indicative for post-surgical tumour recurrence.

Figure 1

Study Workflow showing respective sample numbers for each analysis step. Methodical comparison of next-generation sequencing vs. digital droplet PCR for the detection of tumour DNA in plasma is performed on data of 21 patients with identified mutations. Clinical outcome analysis is performed for 16 patients as five patients had to be excluded due to incomplete surgery.

Figure 2

Comparison of mutation allele frequencies by next-generation sequencing (white bars) and digital droplet PCR (black bars) on genomic DNA. The Pearson approach was used to determine the correlation between the two datasets (R² = 0.91).

Figure 3

Visual representation of a single mutation in gDNA and ccfDNA. The upper panels show results from next-generation sequencing, whereas the lower panels show the same DNA analysed by digital PCR. The left panels show the mutation on genomic DNA, the right panels show the results from corresponding ccfDNA.

Figure 4

Kaplan-Meier plot representing the probability to relapse within 1 year for patients who underwent complete resection and who had mutations detectable in tissue-based analysis (n = 16). Compared are patients with and without detectable levels of circulating tumour DNA (ctDNA) at primary staging.