Abstract 5694: Accurate detection of microsatellite instability (MSI) in matched and unmatched clinical tumor samples using the Pillar oncoReveal MSI panel

Abstract

Abstract Introduction: MSI is caused by mutations in mismatch repair (MMR) genes that result in increased insertions and deletions within simple repeats in the genome. The MSI-High (MSI-H) phenotype is prevalent in Lynch Syndrome, caused by the inheritance of MMR loss-of-function alleles as well as in 15% of all colorectal cancers. Accurate, timely detection of MSI is important for predicting the efficacy of immunotherapies that take advantage of increased expression of neoantigens in MSI-H tumors. The Pillar oncoReveal MSI Panel is a targeted NGS assay comprised of a single-tube multiplex PCR-based chemistry and companion Pillar Variant Analysis Toolkit (PiVAT) software with MSIsensor-based detection of tumor MSI status. We report the accuracy of the PiVAT MSI module in detection of tumor MSI status without the need for a matched normal comparator. Methods: 56 pairs of clinical tumor and matched normal FFPE samples were analyzed using the oncoReveal MSI Panel. MSI status was verified by a standard MSI detection method. Illumina sequencing was performed using a PE2x150 protocol. PiVAT determines MSI status using MSIsensor. Call performance on Burrows-Wheeler Alignment (BWA) BAM files was compared with PiVAT’s paired-end assembled and filtered BAM files (PBAM). In both cases coverage normalization and FDR thresholds were set to 1. In a second experiment, 88 clinical normal tissue samples were combined with the matched tumor/normal samples and analyzed using the PiVAT MSI module’s unmatched calling protocol using a synthetic pooled normal. Results: MSI was first assessed in matched tumor/normal samples using MSIsensor. Preliminary analyses revealed a greater separation of MSI scores between MSI-positive and -negative tumors when MSIsensor was run on PBAM vs BWA-generated BAM files. Due to high individual variability at MS sites, MSI detection is difficult for tumor samples with no available matched normal tissue. Hierarchical clustering demonstrated that a subset of MS sites exhibited low variability between normal samples and showed large differences between MSI-H tumor, and MS stable (MSS) tumor and/or normal samples when compared to a pooled normal baseline. Analyses based on the full target set did not accurately detect differences between MSI-H and MSS tumor or normal samples in the pooled normal calling context. When using a clustered subset of sites, MSI-detection was comparable with the matched tumor/normal context. Conclusions: We developed a robust assay for MSI detection in multiple tumor types that encompasses library generation, data analysis, and diagnosis with 97% sensitivity and 100% specificity. We also demonstrate that accurate MSI-detection is possible in the absence of matched normal tissue, using a pooled normal reference and a curated target set. Citation Format: Jordan Aldersley, Michael Liu, Lixing Qi, Ye Jiao, Akuah Kontor, Yue Ke, Zhaohui Wang. Accurate detection of microsatellite instability (MSI) in matched and unmatched clinical tumor samples using the Pillar oncoReveal MSI panel [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5694.