Because BC is prone to relapse, easily progresses and has high treatment costs, it is particularly imperative to identify cancer biomarkers that can reflect the biological functions of BC. Konety et al. posited that the ideal bladder cancer marker would not only be safe, economical and stable but also have low false-positive and false-negative rates . Meanwhile, diagnostic biomarkers can take many forms and may be lipids, proteins or DNA in this study, and their expression levels are highly correlated with the risk of bladder cancer .
In the current research, the datasets of the same PL570 and PL96 platforms were merged and standardized, and then the limma package was used to find DEGs on these two platforms combined with the GSE13507 dataset. After the intersection of the DEGs of the three datasets, two common upregulated genes (TOP2A and ISG15) and 11 downregulated DEGs were found. Thereafter, GO and KEGG enrichment analyses of downregulated DEGs were carried out, revealing that the DEGs were associated with a decline in muscle system contraction. Next, eight downregulated genes identified through the PPI network, together with the above two upregulated genes, were selected as hub genes. To verify the diagnostic validity, the transcriptional level, mutation rate and protein expression of hub genes in BC tissues were compared with those of normal tissues in TCGA and GTEx datasets, and satisfactory results were found. To validate their clinical applicability, the hub genes were analysed by random forest analysis and combined ROC analysis, and TOP2A was found to have conspicuous diagnostic advantages. In random forest analysis, TOP2A had the highest MDG value, which meant that its separate diagnostic value was the most significant. Among combinations of two indexes, TOP2A and CNN1 were the highest-performing pair. Among three-gene combinations, TOP2A combined with CNN1 plus ISG15 or ISG15 plus ACTG2 had the largest AUC.
TOP2A, also known as topoisomerase II α, is an enzyme that changes the topological state of DNA by breaking double-stranded DNA during mitosis and participates in the process of DNA transcription and replication in cell proliferation [25–26]. Generally, cancer is a disease characterized by malignant proliferation and death regulation disorder of tumour cells. Therefore, abnormal TOP2A overexpression is closely related to cancer. Previously, its overexpression was reported in ovarian cancer, gastric cancer, colorectal cancer and other cancer tissues [27–29]. In the present study, the expression of TOP2A in BC and normal tissues was compared in the GEPIA, Oncomine, HPA and R software platforms, and it was found that the expression of TOP2A in distinct stages of bladder cancer was significantly different from that in normal tissues. GSEA also showed mitotic phase and Rho GTPase enrichment. These results were consistent with those of Gao et al [30–31]. Moreover, similar findings were found in mouse bladder cancer models . On this basis, Zeng  further demonstrated that the invasiveness, proliferation and migration ability of TOP2A knockout BC cells were markedly restrained by xenograft tumour formation in nude mice and other tests. Regarding the clinical application of TOP2A for BC, Kim’s report is the only one to date regarding urinary cell-free DNA . His research showed that the expression level of TOP2A in BC patients was higher than that in normal persons and haematuria patients. In addition, the areas under the ROC curves of TOP2A for BC, non-muscle-invasive BC and muscle-invasive BC were 0.741, 0.701 and 0.838, respectively.
Although few studies have been performed on BC, CNN1 was found to be expressed at low levels in an array of cancer tissues [35–37]. Menéndez  reported that CNN1 can inhibit tumour formation by mediating a variety of angiogenic factors. Scratch and wound healing tests also showed that the invasion and migration ability of CNN1-overexpressing lung squamous cell carcinoma cells decreased significantly . ISG15 overexpression has been reported to promote distant metastasis of mouse liver cells. However, in fact, its effects on different tumour cells are contradictory [40–42] and still need to be further clarified by scholars.
Finally, it is worth pointing out that our research may have some limitations. First, DEGs were detected in resected tissues using microarray or sequencing technology. Unfortunately, it is unreasonable for these two techniques to be widely used in clinical practice. The search must continue for an ideal technology that can find genetic differences in blood or urine samples economically and quickly. Another intrinsic limitation is that the specific cancer-promoting mechanisms of the screening biomarkers are still obscure; thus, further in vivo and in vitro experiments are necessary. Last but not least, there were relatively few clinical diagnostic tests for screening gene biomarkers. In Kim’s study on the diagnostic function of TOP2A in bladder cancer, the sample size of 83 cancer patients still proved to be insufficient. External validation in diverse and much larger-scale populations with longer follow-up periods may be beneficial in further studies.
In summary, the purpose of the present study was to identify the hub genes involved in the development of BC through a comprehensive bioinformatics analysis. The diagnostic effectiveness of the screened genes was verified by expression, mutation rate and clinical application. Our findings indicated that TOP2A had prominent diagnostic value for BC. Nevertheless, larger populations and biological trials are needed to verify this finding.