Abstract 98P
Background
DNA mismatch repair deficiency usually causes microsatellite loci length alterations during DNA replication, which is known as microsatellite instability (MSI). Recently, Next-Generation Sequencing (NGS) based MSI detection methods are very popular. However, they often show poor consistency across sequencing batches, platforms and cancer types. There is great need to develop a unified method to detect MSI status with high producibility in different NGS labs.
Methods
We construct a mixed reference dataset by mixing multiple population of MSI-negative samples from different sequencing batches, platforms and cancer types. We define the characteristic value for each microsatellite markers within each sample by the number of repeats of different lengths. Therefore, it is possible to match and acquire an optimal subset of reference for each sample from the constructed mixed reference datasets. We calculated the baseline reference value based on the mean and standard deviation value of the characteristic value at each microsatellite loci across the reference subset. Then we compared characteristic value against baseline reference value at each locus to assess the instability of microsatellite loci. Finally, the fraction of unstable loci out of the total number of loci was calculated to assess the instability of each sample.
Results
We evaluated 337 MSI-positive samples and 905 MSI-negative samples from 3 cancer types (colorectal cancer, gastric cancer and endometrial cancer), 6 sequencing chips (MiSeq, MiniSeq-High Throughput, NextSeq-Mid Throughput, NextSeq-High Throughput, MGI200 and MGI2000) from 2 sequencing platforms (Illumina and MGISEQ) and 17 sequencing batches in total. Compared to the well-established gold standard PCR plus capillary electrophoresis methodology, our approach shows high sensitivity (99.41%) and specificity (100%).
Conclusions
Our NGS-based MSI detection method provides an effective, robust and accurate performance to determine MSI status compared to traditional method, which can apply to samples from different cancer types with different sequencing platforms and chips, without the need to creat specific reference set for each NGS lab.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
AmoyDx.
Funding
AmoyDx.
Disclosure
C. Lin, S. Chen, B. Jin, A. Zhang, L. Ruan, H. Dong, X. Li: Financial Interests, Institutional, Full or part-time Employment: AmoyDx.