Abstract 730: Validation of a NGS-based MSI testing workflow in liquid biopsy

Abstract

Abstract Background Microsatellite instability (MSI) is a molecular phenotype characterized by the lengthening or shortening of tandem DNA repeats within the human genome. MSI phenomenon has been reported in various solid tumors as the consequence of DNA mismatch repair deficiency. It is generally associated with good prognosis and effective immunotherapy treatment. The current MSI tests are based on various methodologies, including immunohistochemistry, polymerase chain reactions, or next-generation sequencing (NGS). Each testing methodology has its own technical advantages, but they either fail at liquid biopsy specimen or implement complicated and expensive experimental interventions. Methods A designated analysis pipeline was developed individually for tissue biopsy and liquid biopsy to evaluate the stability of MSI sites within a NGS panel testing targeting over 400 genes at a common depth of coverage (<1000X for tissue biopsy and <5000X for liquid biopsy). The liquid biopsy algorithm implements a signal enhancing module based on a pool of normal samples to boost assay performance. We validated the tissue biopsy testing with an exploratory cohort of 90 gastrointestinal tract cancer patients. Retrospective review of over 10,000 pan-cancer patient-matched tissue-cfDNA sample pairs permitted a comprehensive evaluation of the liquid biopsy MSI testing performance. Results For tissue biopsy specimen validation, the assay system displayed 0.97 area-under-the-curve (AUC) in receiver operative characteristics (ROC) curve analysis against the comparator assay, out-performing publicly available MSI analysis software. Unfortunately, the tissue MSI analysis pipeline and publicly available software fails with less than 10% sensitivity and 0.55 AUC when applied to liquid biopsy specimen with the patient-matched tissue MSI status used as the gold standard. In contrast, the dedicated cfDNA analysis pipeline significantly boosted the assay performance to 0.73 without additional experimental interventions, such as wild-type allele depletion or molecular barcode. Conclusions MSI testing in liquid biopsy specimen remains challenging due to the low fraction of tumor originated cfDNA and consequently weak signal intensity. The implementation of the signal enhancing MSI analysis algorithm demonstrated high proficiency at improving the signal-to-noise ratio. This bioinformatics approach offered a cost-effective and non-invasive alternative for MSI testing without significantly modifying the NGS library preparation protocols. Citation Format: Xiangyuan Ma, Hua Bao, Xiaoling Tong, Zhili Chang, Xunbiao Liu, Xiuyan Zhang, Ryan S. Robetyore, Xue Wu, Yang W. Shao. Validation of a NGS-based MSI testing workflow in liquid biopsy [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 730.