Background: Osteoarthritis is a chronic musculoskeletal disease characterized by age-related gradual thinning and a high risk in females. Recent studies have shown that DNA methylation plays important roles in osteoarthritis. However, the genome-wide pattern of methylation in enhancers in osteoarthritis remains unclear. Methods: To explore the function of enhancers in osteoarthritis, we quantified CpG methylation in human enhancers based on a public dataset that included methylation profiles of 470,870 CpG probes in 108 samples from patients with hip and knee osteoarthritis and hip tissues from healthy individuals. Combining various bioinformatics analysis tools, we systematically analyzed aberrant DNA methylation of the enhancers throughout the genome in knee osteoarthritis and hip osteoarthritis. Results: We identified 16,816 differentially methylated CpGs, and nearly half (8111) of them were from enhancers, suggesting major DNA methylation changes in both types of osteoarthritis in the enhancer regions. A detailed analysis of hip osteoarthritis identified 2426 differentially methylated CpGs in enhancers between male and female patients, and 84.5% of them were hypomethylated in female patients and enriched in phenotypes related to hip osteoarthritis in females. Next, we explored the enhancer methylation dynamics among patients with knee osteoarthritis and identified 280 differentially methylated enhancer CpGs that were enriched in the human phenotypes and disease ontologies related to osteoarthritis. Finally, a comparison of enhancer methylation between knee osteoarthritis and hip osteoarthritis revealed organ source-dependent differences in enhancer methylation. Conclusion: Our findings indicate that aberrant methylation of enhancers is related to osteoarthritis phenotypes, and a comprehensive atlas of enhancer methylation is useful for further analysis of the epigenetic regulation of osteoarthritis and the development of clinical drugs for treatment of osteoarthritis.

¹⁸ F-MPG PET/CT enables noninvasive identification of NSCLC patients who may benefit from EGFR-TKI therapy and can be used to monitor treatment outcome.

Chemoresistance has been a major challenge in the treatment of patients with breast cancer. The diverse omics platforms and small sample sizes reported in the current studies of chemoresistance in breast cancer limit the consensus regarding the underlying molecular mechanisms of chemoresistance and the applicability of these study findings. Therefore, we built two transcriptome datasets for patients with chemotherapy-resistant breast cancers—one comprising paired transcriptome samples from 40 patients before and after chemotherapy and the second including unpaired samples from 690 patients before and 45 patients after chemotherapy. Subsequent conventional pathway analysis and new subpathway analysis using these cohorts uncovered 56 overlapping upregulated genes (false discovery rate [FDR], 0.018) and 36 downregulated genes (FDR, 0.016). Pathway analysis revealed the activation of several pathways in the chemotherapy-resistant tumors, including those of drug metabolism, MAPK, ErbB, calcium, cGMP-PKG, sphingolipid, and PI3K-Akt, as well as those activated by Cushing’s syndrome, human papillomavirus (HPV) infection, and proteoglycans in cancers, and subpathway analysis identified the activation of several more, including fluid shear stress, Wnt, FoxO, ECM-receptor interaction, RAS signaling, Rap1, mTOR focal adhesion, and cellular senescence (FDR < 0.20). Among these pathways, those associated with Cushing’s syndrome, HPV infection, proteoglycans in cancer, fluid shear stress, and focal adhesion have not yet been reported in breast cancer chemoresistance. Pathway and subpathway analysis of a subset of triple-negative breast cancers from the two cohorts revealed activation of the identical chemoresistance pathways.

Hepatocellular carcinoma (HCC) is the most prevalent primary cancer and a highly aggressive liver malignancy. Liver cancer cells reprogram their metabolism to meet their needs for rapid proliferation and tumor growth. In the present study, we investigated the alterations in the expression of the genes involved in glucose metabolic pathways as well as their association with the clinical stage and survival of HCC patients. We found that the expressions of around 30% of genes involved in the glucose metabolic pathway are consistently dysregulated with a predominant down-regulation in HCC tumors. Moreover, the differentially expressed genes are associated with an advanced clinical stage and a poor prognosis. More importantly, unsupervised clustering analysis with the differentially expressed genes that were also associated with overall survival (OS) revealed a subgroup of patients with a worse prognosis including reduced OS, disease specific survival, and recurrence-free survival. This aggressive subtype had significantly increased expression of stemness-related genes and down-regulated metabolic genes, as well as increased immune infiltrates that contribute to a poor prognosis. Collectively, this integrative study indicates that expressions of the glucose metabolic genes could be used as potential prognostic markers and/or therapeutic targets, which might be helpful in developing precise treatment for patients with HCC.

Alternatively-activated pathways have been observed in biological experiments in cancer studies, but the concept had not been fully explored in computational cancer system biology. Therefore, an alternatively-activated pathway identification method was proposed and applied to primary breast cancer and breast cancer liver metastasis research using microarray data. Interestingly, the results show that cytokine-cytokine receptor interaction and calcium signaling were significantly enriched under both conditions. TGF beta signaling was found to be the hub in network topology analysis. In total, three types of alternatively-activated pathways were recognized. In the cytokine-cytokine receptor interaction pathway, four active alteration patterns in gene pairs were noticed. Thirteen cytokine-cytokine receptor pairs with inverse activity changes of both genes were verified by the literature. The second type was that some sub-pathways were active under only one condition. For the third type, nodes were significantly active in both conditions, but with different active genes. In the calcium signaling and TGF beta signaling pathways, node E2F5 and E2F4 were significantly active in primary breast cancer and metastasis, respectively. Overall, our study demonstrated the first time using microarray data to identify alternatively-activated pathways in breast cancer liver metastasis. The results showed that the proposed method was valid and effective, which could be helpful for future research for understanding the mechanism of breast cancer metastasis.

In humans, despite the rapid increase in disease‐associated gene discovery, a large proportion of disease‐associated genes are still unknown. Many network‐based approaches have been used to prioritize disease genes. Many networks, such as the protein–protein interaction (PPI), KEGG, and gene co‐expression networks, have been used. Expression quantitative trait loci (eQTLs) have been successfully applied for the determination of genes associated with several diseases. In this study, we constructed an eQTL‐based gene–gene co‐regulation network (GGCRN) and used it to mine for disease genes. We adopted the random walk with restart (RWR) algorithm to mine for genes associated with Alzheimer disease. Compared to the Human Protein Reference Database (HPRD) PPI network alone, the integrated HPRD PPI and GGCRN networks provided faster convergence and revealed new disease‐related genes. Therefore, using the RWR algorithm for integrated PPI and GGCRN is an effective method for disease‐associated gene mining.