Personalized differential expression analysis in triple-negative breast cancer

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Accurate identification of individual-level differentially expressed genes (DEGs) is crucial for understanding disease mechanisms and advancing precision medicine.
• Existing algorithms struggle to balance accuracy with sufficient statistical power for individual DEG analysis.
• RankCompV2, initially for population-level DEGs, has been adapted for individual-level analysis.

Purpose of the Study:

• To develop and validate an optimized algorithm, RankCompV2.1, for identifying individual-level DEGs with improved accuracy and statistical power.
• To assess the performance of RankCompV2.1 against existing individualized analysis methods using simulations and real cancer data.
• To apply RankCompV2.1 to triple-negative breast cancer (TNBC) for identifying therapeutic targets and patient stratification.

Main Methods:

• Modification of the RankCompV2 algorithm to RankCompV2.1, incorporating gene rank positions and relative rank differences.
• Comparative analysis of RankCompV2.1 with other individualized DEG algorithms on simulated and real paired cancer-normal datasets from ten cancer types.
• Application of Gene Set Enrichment Analysis (GSEA) and Gene Set Variation Analysis (GSVA) to interpret pathway-level changes.
• Identification of universally deregulated genes in TNBC and analysis of their drug interactions.
• Clustering of TNBC samples based on highly variable deregulated genes and subsequent survival analysis.

Main Results:

• RankCompV2.1 demonstrated superior statistical power and computational efficiency compared to other individualized algorithms, achieving comparable accuracy.
• Analysis revealed specific pathways enriched with up- or down-regulated genes, correlating with higher or lower enrichment scores, respectively.
• Sixteen genes universally deregulated in 966 TNBC samples were identified as potential therapeutic targets, interacting with FDA-approved drugs.
• TNBC samples were stratified into three subgroups with distinct prognoses based on gene deregulation patterns.
• The poorest outcome subgroup exhibited down-regulated immune, signal transduction, and apoptosis pathways, with OAS family genes suggested as potential immunotherapy targets.

Conclusions:

• RankCompV2.1 is a robust tool for identifying individual-level DEGs with high accuracy and statistical power.
• The algorithm facilitates the analysis of carcinogenesis mechanisms and the exploration of novel therapeutic strategies.
• The study identified promising therapeutic targets and patient subgroups within TNBC, paving the way for personalized treatment approaches.