Circulating tumor DNA (ctDNA) analysis by low-coverage whole genome sequencing (lcWGS) of resectable esophageal adenocarcinoma (rEAC) patients.

4033 Background: ctDNA is becoming an established marker to assess tumor burden, relapse after surgery, and to identify responders in immunotherapy studies. In the phase II PERFECT trial rEAC patients were treated with neoadjuvant chemoradiotherapy (nCRT) and a PD-L1 inhibitor (van den Ende et al. CCR. 2021). Here we evaluated the potential of cell-free DNA (cfDNA) to predict pathological complete response (pCR) and recurrence. Methods: The cohort consisted of 40 patients and 145 plasma samples. EDTA blood samples were drawn at baseline (B, N = 40), in week 5 of nCRT (W5, N = 40), before surgery (OR, N = 33) and 3 months after surgery (FU, N = 32). cfDNA was isolated by affinity columns (CNAkit, QIAgen) quantified by spectrofluorometer (BioAnalyzer, Agilent), sequencing libraries were prepared for lcWGS ( < 5-fold coverage, Tag-seq, Takara) and sequenced on a NovaSeq (S4, PE150). Sequencing data were processed with an in-house pipeline. Copy number aberrations (CNA) and the tumor fraction were estimated using the ichorCNA tool. Insert sizes were recovered and we determined a Tumor Enriched Fragment Fraction (TEFF), calculated by doing the ratio of fragments between 90-150 bp and 250-320 bp (enriched in tumor signal) and fragments between 150-250 bp and 320-360 bp (poor in tumor signal). ichorCNA and TEFF were used to quantify the ctDNA fraction in plasma samples. pCR was defined as ypT0N0. Residual tumor, progression or death before surgery were considered non-pCR. Relapse-free survival (RFS) was defined as the time after surgery until recurrence. Results: The pCR rate was 25% (10/40). The median fold change TEFF between B and W5 was -0.15 (range -0.67 to 0.44) in the pCR group and 0.16 (range -1.40 to 0.76) in the non-pCR group (Mann–Whitney U; p = 0.047). Of the 17 patients in whom ctDNA was detected (TEFF≥0.3 and/or ichorCNA≥0.03) in the FU sample, 13 (76%) showed a recurrence. Of the 15 patients with no ctDNA detected 5 (33%) showed a recurrence. Patients with ctDNA detected at FU had worse RFS, HR = 2.72 (95%CI 0.96-7.71; p = 0.050). Recurrences were detected earlier by FU ctDNA than by imaging due to physical complaints with a median of 312 days (163-798 days). Conclusions: lcWGS appears to be a useful tool to predict pCR and recurrence in resectable esophageal cancer. These lcWGS results will be further combined with fragmentomics analysis and targeted mutational data (Ion Torrent next-generation sequencing) in order to assess response to immunotherapy. Clinical trial information: NCT03087864.