Microsatellite instability detection using a large next-generation sequencing cancer panel across diverse tumour types

Abstract

AimMicrosatellite instability (MSI), a hallmark of DNA mismatch repair deficiency, is a key molecular biomarker with multiple clinical implications including the selection of patients for immunotherapy, identifying patients who may have Lynch syndrome and predicting prognosis in patients with colorectal tumours. Next-generation sequencing (NGS) provides the opportunity to interrogate large numbers of microsatellite loci concurrently with genomic variants. We sought to develop a method to detect MSI that would not require paired normal tissue and would leverage the sequence data obtained from a broad range of tumours tested using our 467-gene NGS Columbia Combined Cancer Panel (CCCP).MethodsAltered mononucleotide and dinucleotide microsatellite loci across the CCCP region of interest were evaluated in clinical samples encompassing a diverse range of tumour types. The number of altered loci was used to develop a decision tree classifier model trained on the retrospectively collected cohort of 107 clinical cases sequenced by the CCCP assay.ResultsThe classifier was able to correctly classify all cases and was then used to analyse a test set of clinical cases (n=112) and was able to correctly predict their MSI status with 100% sensitivity and specificity. Analysis of recurrently altered loci identified alterations in genes involved in DNA repair, signalling and transcriptional regulation pathways, many of which have been implicated in MSI tumours.ConclusionThis study highlights the utility of this approach, which should be applicable to laboratories performing similar testing.