Epigenetics is a heritable phenotypic change that occurs without changes in the genomic DNA sequence. In normal biological processes, epigenetics alters the expression of genes, allowing cells to adapt to the surrounding environment and exert their unique functions. Epigenetic dysregulation is also involved in the processes of tumorigenesis, proliferation and metastasis, epigenetic changes in tumor cells. Abnormal gene expression and biological processes are conducive to tumor survival and facilitate characteristic alterations to tumors. There is great potential for the application of epigenetics in early tumor screening, treatment guidance and prognosis prediction. The main epigenetic modifications include DNA methylation, histone modification and genomic imprinting. As the most common epigenetic mechanism, DNA methylation is commonly altered in tumors. Hypomethylation of DNA leads to gene activation. Many CpG islands are usually methylated in somatic tissue. These methylated islands may become undermethylated in cancer, which can activate adjacent genes. Examples of genes affected by hypomethylation include oncogenes, such as RAS [19] and Phospholipase C epsilon 1 (PLCE1)[20], whose hypomethylation indicates a promising therapeutic target. However, some tumor suppressor genes show hypermethylation in tumors, resulting in decreased gene expression and induction of tumorigenesis and development; these hypermethylated tumor suppressors include Ras association domain family member 1 (RASSF1) methylation in bladder cancer, O-6-methylguanine-DNA methyltransferase (MGMT) methylation in non-small-cell lung cancer and retinoic acid receptor beta (RARB) methylation in prostate cancer [21]. RASSF1 has been characterized as a tumor suppressor gene because it has the potential to inhibit cell proliferation, control the cell cycle and promote apoptosis. The MGMT protein has shown DNA repair ability because it prevents the formation of DNA adducts of carcinogens. RARB has been shown to function as a tumor suppressor gene, and inactivation of its promoter via DNA methylation is associated with lung, breast and prostate cancer. In addition to the hypermethylation of the RARB gene, it was recently reported that Kruppel-like factor 4 (KLF4) is also involved in the biological process of prostate cancer. As a new tumor suppressor gene, KLF4 has a hypermethylated promoter sequence, and its gene expression is downregulated in prostate cancer, which promotes tumor proliferation and drug resistance. Downregulation of KLF4 or hypermethylation of KLF4 gene promoters can be considered markers for predicting the early recurrence of cancer[22, 23]. GSTP1 is functionally inactivated by promoter methylation, which increases the sensitivity to oxidative stress and increases the risk of prostate cancer progression [24]. Studies have shown that GSTP1 can be used as a biomarker of fluid biopsy in prostate cancer, but the results show that it has high specificity and low sensitivity [25]. A post hoc analysis of data from a Phase 3 trial showed that the detection of serum free methylated GSTP1 (mGSTP1) DNA can predict the outcome of chemotherapy in metastatic prostate cancer and may guide treatment decisions in the clinic [26]. In addition, methylation of other genes, including olfactory 4 (OLFM4)[27], MTSS1[7] and myeloid ecotropic viral insertion site 2 (MEIS2)[28], also plays an important role in the occurrence and development of prostate cancer. The methylation pattern of prostate cancer can be used as a potential reasonable choice for disease classification and prognosis evaluation. At present, DNA methylation sites related to the prognosis of prostate cancer have been found, including GPR7 (known as neuropeptides B and W receptor 1, NPBWR1), ABHD9 (known as epoxide hydrolase 3, EPHX3), an expressed sequence tag on chromosome 3 (Chr3-EST)[29], paired like homeodomain 2 (PITX2)[30] and zinc finger protein 154 (ZNF154)[31]. The clinical significance of these genes with different methylation patterns in tumor classification, survival and prognosis has yet to be examined in a large prostate cancer dataset. In this study, we tried to explore a classification method that integrates several DNA methylation markers to help clinicians evaluate the therapeutic effect and prognosis of prostate cancer patients and choose treatment strategies.
In general, we used a large dataset from the TCGA and GEO databases to screen methylation sites at the prognostic level by means of bioinformatics. We used the LASSO algorithm to eliminate the methylation sites with small differences; we retained the 21 methylation sites with the most significant differences, and used their corresponding gene expression data to construct a recurrence prediction model for patients with prostate cancer. We observed methylation models in terms of immunity, pathway, and genomic variation.
Flammiger et al [17] detected Treg cells in prostate cancer tissues by FOXP3 immunohistochemistry and found that prostate-specific antigen (PSA) recurrence-free survival was reduced in patients with high levels of Treg cells, which was associated with more advanced tumor stages and a higher Ki67 labeling index. Erlandsson et al [16] found that the higher the degree of tumor infiltration of M2 macrophages was, the greater the risk of death in patients with prostate cancer; the high degree of infiltration of M2 macrophages and other inhibitory cells (such as Tregs) may support the immunosuppressive microenvironment. The role of B cells in tumors is not clear. Many studies have reported that tumor-infiltrating B cells may promote tumor progression. However, other studies have shown that tumor infiltrating B cells may be associated with a good prognosis and an enhanced response to chemotherapy[32]. WooJR et al[18] found that the degree of B lymphocyte infiltration in prostate cancer tissues was higher than that in adjacent tissues, and the degree of B lymphocyte infiltration in patients with a high risk of recurrence was higher than that in patients with a low risk of recurrence. The role of B cell immunity in prostate cancer is unclear, which indicates opportunities for research and potential treatment.
According to the gene enrichment analysis, we determined that the related metastatic gene sets of prostate cancer are enriched in cluster 3 subtypes, and it can be inferred that the genes that distinguish subtypes play an important role in the metastasis of prostate cancer. In addition, our results suggest that the SUMOylation of DNA methyltransferase may play a vital role in tumor metastasis.
Our results show that patients in cluster 3 have a higher burden of somatic mutations in TP53, TTN and KMT2D. p53 is a tumor suppressor that responds to DNA damage[33] by regulating the cell cycle checkpoint pathway and tends to mutate in advanced, recurrent and metastatic prostate cancer [34]. Missense mutation of titin (TTN) is correlated with favorable prognosis in lung squamous cell carcinoma [35] In multiple tumor cohorts, the progression-free survival time or overall survival time of patients with TTN mutations is higher than that of patients with wild-type TTN, which is not consistent with our results. The role of TTN in prostate cancer needs to be further determined. KMT2D is a histone H3 lysine 4 (H3K4) methyltransferase that inhibits the growth and metastasis of bladder cancer cells by regulating tumor suppressor genes such as TP53[36]. Our study provides new ideas regarding the potential mechanism by which the high TP53, TTN and KMT2D mutation burden in cluster 3 subtypes may promote the development of prostate cancer through increased DNA damage and dysregulation of tumor suppressor genes.
Androgen receptors are members of the nuclear receptor superfamily that play a key role in prostate cancer. The upregulation of AR may be the cause of the development of prostate cancer into castration-resistant prostate cancer and the related poor prognosis [37]. Studies show that metadherin (MTDH) plays a role in the progression of various tumors [38]. In prostate cancer, MTDH acts as an oncogene, regulating the growth and metastasis of prostate cancer cells [39].
Finally, the effectiveness and universality of the methylation model were verified and evaluated. Our study shows that classification based on DNA methylation is an independent factor for predicting tumor recurrence in patients with prostate cancer. Our model can provide clinicians with a judgment of prognosis in patients with prostate cancer and provide them with guidance for choosing treatment strategies. However, our research has some limitations. Our methylation model was established using data from prostate cancer. Further studies are needed to determine whether it is suitable for different subtypes of prostate cancer, such as castration-resistant prostate cancer and metastatic prostate cancer.