Short mononucleotide repeat detection of MSI : towards high throughput diagnosis

Abstract

Microsatellites are short repetitive DNA sequences, which are liable to replication errors. Microsatellite instability (MSI) is controlled by the mismatch repair system, and accumulation of microsatellite mutations is used as a diagnostic criterion for tumours where this repair system is compromised, such as those which develop in Lynch Syndrome (HNPCC) patients. Currently, the Amsterdam II screening criteria and revised Bethesda Guidelines are used to identify tumours for MSI testing using both immunohistochemistry and fragment analysis tests. However, because Lynch Syndrome patients are being missed, testing for all colorectal and endometrial cancers is now being recommended. Faster and cheaper MSI testing methods are therefore desirable. Although PCR and sequencing error compromise sequence based typing of the repeats currently used for diagnosis, some short mononucleotide repeats have been identified which show low level instability, suggesting that sequence typing of short repeats may be possible. Here, I investigate the utility of high throughput sequencing (HTS) as the basis for MSI testing. As an initial assessment of the method, I used the MiSeq platform to type 22 previously published short mononucleotide repeats in 4 microsatellite unstable (MSI-H) tumours, and showed that MSI could be detected above background noise in 7-12bp repeats. To identify the most variable short repeat markers for MSI testing, I then analysed MSI in whole genome sequence data from The Cancer Genome Atlas network, and identified a panel of 120 7-12bp informative mononucleotide repeats which were subsequently evaluated on a panel of 5 MSI-H tumours and controls. The most informative 20 markers were further tested on a panel of 58 colorectal tumours to define thresholds for instability calling. Using a panel of eighteen 8-12bp mononucleotides it was possible to distinguish between MSI-H and microsatellite stable (MSS) tumours with a sensitivity and specificity of 100%. Flanking SNPs were also evaluated and identified an excess of allelic bias among MSI-H tumours compared to MSS tumours, a feature that could be integrated into the MSI test. Finally, short mononucleotide repeats with flanking SNPs were assessed for their potential to identify clonal variation in MSI-H tumours. Using a multiple biopsy approach evidence of different sub-clones was found in three MSI-H tumours, suggesting that these markers could be used for analysis of clonal variation and evolution.