BC is one of the most common tumours in the urinary system. The global incidence rate of BC is increasing every year. It is one of the 10 most common cancers in men 10 and is a non-negligible hazard to human health. BC prognosis-related biomarkers have an important guiding role in diagnosis, prognosis and corresponding treatment methods. Therefore, it is one of the hot spots to find effective BC prognostic biomarkers. This study aimed to screen differentially expressed genes and construct a prognostic evaluation model of BC patients. The DEGs were obtained by R analysis of the GEO and TCGA databases in our study. Univariate Cox analysis was used for preliminary screening, and the results showed that 126 genes had significant differences that were related to the prognosis of BC. Seven genes (GRHL2, ANXA1, APOL1, SETBP1, NR2F1, KLRB1 and PLAC9) were identified to be independent prognostic factors for the prognosis of BC patients via LASSO and multivariate Cox regression analyses. These genes were used to establish a prognostic risk score model for BC patients to evaluate the OS of BC patients. The Kaplan-Meier curve and ROC curve demonstrated that the 7-gene signature prognostic model had a good prediction performance. The 7-gene signature risk score was confirmed to be independent of other clinicopathological features of BC patients by univariate Cox and multivariate Cox regression analyses. Further, the risk score and clinicopathological risk factors (age, lymphovascular invasion, lymphatic metastasis, metastasis) were used to establish a prognostic nomogram to predict the prognosis of patients at 1, 3, 5 and 8 years. The results of the calibration curve and DCA confirmed the reliability of the nomogram, which showed that the nomogram could help clinicians predict patient prognosis and provide guidance for patient rehabilitation assessment and treatment decision-making.
At present, although many genes have been identified as effective indicators for predicting the prognosis of BC patients 11–14, their predictive performance is relatively poor due to the limited number of screened genes as well as the lack of correlation with clinicopathological data. Compared with a single clinical gene biomarker, the integration of multiple related genes into a prognostic evaluation model can effectively improve the accuracy of prediction 15–17. Similarly, the prognostic evaluation model of the 7-gene signature has displayed satisfactory predictive performance in this study. Among the 7 genes in the signature, PLAC9, SETBP1, ANXA1 and NR2F1 were negatively correlated with OS in BC patients. PLAC9 is a placental protein that is upregulated in embryonic expression (weeks 8 to 9) 18. Ouyang C et al. reported that overexpression of PLAC9 induces G2 and M phase arrest in cell division and inhibits cell growth 19, but it has not been reported in bladder cancer. It might be necessary to further explore the specific role of PLAC9 in bladder cancer in the future. SETBP1 has been established as an important diagnostic marker for bone marrow malignancies, and its expression is closely related to the prognosis of patients. Studies have shown that it plays a key role in tumour invasion and rapid evolution 20. Additionally, SETBP1 inhibits the expression of tumour protein phosphatase 2A (PP2A) 21 and regulates cell proliferation 22, suggesting that SETBP1 may also play a regulatory role in tumorigenesis and progression. ANXA1 functions to regulate a variety of cell biological behaviours, including membrane aggregation, phagocytosis, cell proliferation and tumorigenesis 23. Reports have shown that ANXA1 is related to the recurrence and drug resistance of BC, and its expression level was positively correlated with T phase 24. ANXA1 was shown to play an important role in the development of high-grade BC 25. Our findings also demonstrated that high expression of ANXA1 is related to the occurrence and development of bladder cancer and affects the prognosis of BC patients. NR2F1 regulates the progression of cell differentiation, cancer progression, and central and peripheral neurogenesis 26. Kikuchi-Koike R et al. indicated that NR2F1 is highly connected with the high-risk recurrence of breast cancer 27. High NR2F1 expression has a strong effect on increasing lung metastatic potential. Additionally, the overexpression of NR2F1 was reported to enhance invasion and metastasis in salivary adenoid cystic carcinoma 28. At present, the role of NR2F1 in BC has not been reported. In this study, we showed for the first time that NR2F1 is highly expressed in BC and is closely related to the prognosis of BC patients. However, the specific regulation of NR2F1 in BC still needs further exploration. The other three genes, GRHL2, APOL1 and KLRB1, were positively correlated with the OS of BC patients. GRHL2 enhances the effect of carcinogenesis and promotes tumour cell proliferation. It can also promote the growth of oral squamous cell carcinoma, colorectal cancer and hepatocellular carcinoma 29. However, more reports have shown that GRHL2 can inhibit epithelial-mesenchymal transition to inhibit cell migration 30 and cancer metastasis 31. Additionally, the overexpression of GRHL2 can effectively inhibit tumour invasion and migration in gastric and breast cancer 30, 32. In this study, we confirmed that GRHL2 is significantly expressed at low levels in BC and has a positive regulatory effect on the prognosis of BC patients. GRHL2 may be an effective prognostic marker for BC. APOL1 is mainly active in the kidney 33, and its high expression can promote the occurrence of kidney disease 34. In the study of Gutiérrez OM et al, the mortality rate of patients with a high-risk APOL1 genotype was significantly lower than that of patients with low-risk genotypes in the prognostic review cohort of chronic kidney disease 35. However, there are few reports about the effect of APOL1 on BC, and further research is still needed. KLRB1 is involved in the coding of NKRP1A/LLT1, which triggers the activation of T cells and B cells and can be used in the treatment of cancer 36, and its interaction with LLT1 can inhibit natural killing (NK) cell cytotoxicity 37. Many studies have shown that a high level of LLT1 expression can significantly restrain NK cells and enhance tumour immune escape, as knockdown of LLT1 enhances NK cell-mediated lysis of tumour tissues 38–40. These results combined with our results demonstrate that KLRB1 is expressed at low levels in BC and is positively correlated with OS in BC patients. Because KLRB1 is capable of eliciting spontaneous antitumour immune responses, it may be a promising potential target for BC immunotherapy. Further experiments are needed to verify our hypothesis.
To better apply this model for predicting the OS of BC patients, the risk score was combined with other independent prognostic factors (age, lymphovascular invasion, lymphatic metastasis, metastasis) to establish a nomogram to predict the OS of BC patients at 1 year, 3 years, 5 years, and 8 years. The C-index, calibration curve and DCA results showed the good performance and clinically usefulness of the nomogram. These results showed that the nomogram established by combining these independent prognostic factors was an effective and accurate tool for predicting the OS of BC patients. Additionally, this nomogram played an important role in both short-term and long-term observation studies, so it might be helpful for patient consultations, clinical decision-making and determining follow-up arrangements for BC patients.
However, our research still has some limitations. First, most of the patients in the TCGA database were white or black. The racial diversity was limited, so the results may be biased towards these groups. Second, most of the selected genes were not previously reported in BC or lacked research in BC; therefore, further experimental validation was the direction of our research. Finally, the nomogram for predicting the OS of BC patients, which was based on the TCGA database, was only established and validated in a single data set, lacking further validation with external BC data. Therefore, corrections according to future verification studies in multiple different data sets are required.