Meta-Analysis of Microarray Expression Studies on Metformin in Cancer Cell Lines

Several studies have demonstrated that metformin (MTF) acts with variable efficiency as an anticancer agent. The pleiotropic anticancer effects of MTF on cancer cells have not been fully explored yet. By interrogating the Gene Expression Omnibus (GEO) for microarray expression data, we identified eight eligible submissions, representing five different studies, that employed various conditions including different cell lines, MTF concentrations, treatment durations, and cellular components. A compilation of the data sets of 13 different conditions contained 443 repeatedly up- and 387 repeatedly down-regulated genes; the majority of these 830 differentially expressed genes (DEGs) were associated with higher MTF concentrations and longer MTF treatment. The most frequently upregulated genes include DNA damage inducible transcript 4 ( DDIT4), chromodomainhelicase DNA binding protein 2 ( CHD2), endoplasmic reticulum to nucleus signaling 1 ( ERN1), and growth differentiation factor15 ( GDF15). The most commonly downregulated genes include arrestindomain containing 4 (ARRDC4), and thioredoxin interacting protein( TXNIP). The most significantly (p-value < 0.05, Fisher’s exact test) overrepresented protein class was entitled, nucleic acid binding. Cholesterol biosynthesis and other metabolic pathways were specifically affected by downregulated pathway molecules. In addition, cell cycle pathways were significantly related to the data set. Generated networks were significantly related to, e.g., carbohydrate and lipid metabolism, cancer, cell cycle, and DNA replication, recombination, and repair. A second compilation comprised genes that were at least under one condition up- and in at least another condition down-regulated. Herein, the most frequently deregulated genes include nuclear paraspeckleassembly transcript 1 (NEAT1) and insulin induced gene 1 (INSIG1). The most significantly overrepresented protein classes in this compilation were entitled, nucleic acid binding, ubiquitin-protein ligase, and mRNA processing factor. In conclusion, this study provides a comprehensive list of deregulated genes and biofunctions related to in vitro MTF application and individual responses to different conditions. Biofunctions affected by MTF include, e.g., cholesterol synthesis and other metabolic pathways, cell cycle, and DNA replication, recombination, and repair. These findings can assist in defining the conditions in which MTF exerts additive or synergistic effects in cancer treatment.