Sensitive Detection of Cell-Free Tumour DNA Using Optimised Targeted Sequencing Can Predict Prognosis in Gastro-Oesophageal Cancer

SIMPLE SUMMARY: Cancer in the stomach and oesophagus is deadly when discovered at a late stage. There are no good biomarkers for its detection or for making a prognostic prediction. In this study, we evaluate the analysis of cell-free DNA as a prognostic cancer biomarker. Cell-free DNA is DNA released from any tissue to a body fluid. When there is a tumour in the body, some of the cell-free DNA will come from that tumour, and it can be detected in a blood sample. We show that the detection of cell-free DNA from the cancer correlates to a worse prognosis than when no tumour DNA is detected. We also show that the method of analysis is important. Either a tissue biopsy must be included as a validation of the genetic variants detected or analysis of the blood cells or another blood sample after tumour resection needs to be analysed to improve detection. ABSTRACT: In this longitudinal study, cell-free tumour DNA (a liquid biopsy) from plasma was explored as a prognostic biomarker for gastro-oesophageal cancer. Both tumour-informed and tumour-agnostic approaches for plasma variant filtering were evaluated in 47 participants. This was possible through sequencing of DNA from tissue biopsies from all participants and cell-free DNA from plasma sampled before and after surgery (n = 42), as well as DNA from white blood cells (n = 21) using a custom gene panel with and without unique molecular identifiers (UMIs). A subset of the plasma samples (n = 12) was also assayed with targeted droplet digital PCR (ddPCR). In 17/31 (55%) diagnostic plasma samples, tissue-verified cancer-associated variants could be detected by the gene panel. In the tumour-agnostic approach, 26 participants (59%) had cancer-associated variants, and UMIs were necessary to filter the true variants from the technical artefacts. Additionally, clonal haematopoietic variants could be excluded using the matched white blood cells or follow-up plasma samples. ddPCR detected its targets in 10/12 (83%) and provided an ultra-sensitive method for follow-up. Detectable cancer-associated variants in plasma correlated to a shorter overall survival and shorter time to progression, with a significant correlation for the tumour-informed approaches. In summary, liquid biopsy gene panel sequencing using a tumour-agnostic approach can be applied to all patients regardless of the presence of a tissue biopsy, although this requires UMIs and the exclusion of clonal haematopoietic variants. However, if sequencing data from tumour biopsies are available, a tumour-informed approach improves the value of cell-free tumour DNA as a negative prognostic biomarker in gastro-oesophageal cancer patients.