MSIMEP: Predicting microsatellite instability from microarray DNA methylation tumor profiles.

Area of Science:

• Oncology
• Computational Biology
• Genomics

Background:

• Deficiency in DNA mismatch repair (MMR) activity leads to microsatellite instability (MSI), a hypermutator phenotype in tumors.
• MSI is crucial for Lynch syndrome screening and as a predictive biomarker for anti-PD-1 therapies across diverse cancer types.
• Existing computational methods for MSI inference use DNA or RNA, but MSI-high tumors often show hypermethylation.

Purpose of the Study:

• To develop and validate MSIMEP, a computational tool for predicting MSI status from microarray DNA methylation profiles.
• To assess MSIMEP's performance in colorectal cancer (CRC) and its consistency in other MSI-prevalent cancers.
• To compare MSIMEP's efficacy against MLH1 promoter methylation-based prediction in CRC.

Main Methods:

• Development of MSIMEP using DNA methylation profiles from colorectal cancer samples.
• Validation of MSIMEP's optimized and reduced models across different CRC cohorts.
• Testing MSIMEP's performance in gastric and endometrial cancer datasets.

Main Results:

• MSIMEP demonstrated high performance in predicting MSI status in colorectal cancer cohorts.
• The tool showed consistency in predicting MSI in gastric and endometrial cancers.
• MSIMEP models outperformed MLH1 promoter methylation-based prediction in colorectal cancer.

Conclusions:

• MSIMEP is a high-performing computational tool for predicting MSI status from DNA methylation data.
• The tool is effective across multiple cancer types, including colorectal, gastric, and endometrial cancers.
• MSIMEP offers a promising alternative to methylation-based biomarkers for MSI prediction in cancer.