However, some limitations of our study should be mentioned. The major limitation of this study was the small number of expression profiling samples, which potentially weakened the credibility of the data. We admit that biological replicates should be used to eliminate differences among the samples. However, when performing proper nonpooled replicates, the main deterrents, as Tilstone pointed out in an article published in Nature, are the financial constraints of gene expression profiling assays [30]. Therefore, future studies should include larger sample sizes and assess the mechanisms by which metformin regulates certain genes. In addition, we should to measure relevant parameters of fatty liver such as hepatic TG content et al. in future.