Bioinformatic detection of microsatellite instability using whole transcriptome analysis of colorectal cancer samples

BACKGROUND: Tumor microsatellite instability/microsatellite stability (MSI/MSS) status is a crucial parameter determining both disease prognosis and potential response to immunotherapy. Patients are referred for MSI typing based on clinical indications; however, only a small proportion of patients meet these criteria. Consequently, most sequencing datasets available in repositories and collected locally do not inherently contain MSI information in their metadata. Project funding limitations and insufficient biomaterial quantities often preclude DNA isolation and subsequent MSI typing using polymerase chain reaction methods. RNA sequencing results from tumor tissue may also be used to determine MSI status, but this requires incorporating a bioinformatics tool for MSI assessment into the tumor whole transcriptome analysis protocol. AIM: To evaluate the availability and applicability of bioinformatics tools for determining MSI status in locally derived datasets of whole transcriptomes of colorectal cancer samples. METHODS: Publicly available bioinformatics tools designed for MSI status assessment using RNA sequencing data were selected for analysis. These tools were tested on a locally derived dataset of whole tumor transcriptomes from 13 patients following primary colorectal tumor resection. The number of somatic mutations was assessed as a surrogate marker of MSI/MSS status, along with MSI/MSS status based on the results of bioinformatics tool testing, and their correlation. RESULTS: We tested two bioinformatics tools, PreMSIm and MIRACLE, designed to determine MSI status using transcriptomic data. When using MIRACLE, microsatellite instability was detected in 3 out of 13 samples. The MSI status determined by MIRACLE correlated with tumor mutational burden (TMB) (mean, 2163 mutations in MSI samples vs. 122.9 in MSS samples) and reliably identified unstable samples. PreMSIm also detected MSI in 3 samples, but its results showed limited concordance with TMB. For two of the three samples with high TMB, we identified known pathogenic and likely pathogenic variants in MSH2 and MLH1 genes associated with Lynch syndrome, confirming MSI status. For one sample, we proposed sporadic MSI etiology due to MLH1 gene hypermethylation. CONCLUSION: PreMSIm and MIRACLE demonstrate different sensitivity and specificity profiles for MSI status determination when using TMB as a surrogate MSI/MSS marker in colorectal adenocarcinoma. The MIRACLE tool can be easily integrated into whole-transcriptome tumor analysis protocols and provides biologically plausible MSI/MSS assessments that correlate with transcriptome-derived TMB.