Bladder cancer (BLCA) is the most common urinary tract malignancy worldwide, and its incidence rate is increasing gradually. According to global cancer statistics, an estimated 81,180 new BLCA cases and almost 17,100 BLCA deaths occurred in the United States in 2022 (1). Although age-standardized morbidity and mortality are declining over the past 20 years, the number of BLCA events is increasing globally, and the burden of BLCA is likely to rise in the future due to trends in population aging and environmental pollution (2, 3). Ninety-five percent of BLCA cases are urothelial carcinomas, including forms of divergent differentiation (4). Urothelial cancer is a highly diverse disease, ranging from low-grade tumors that are less threatening to high-grade tumors that can be fatal (5, 6). The grade of urothelial carcinoma is particularly important in non-invasive diseases, especially papillary tumors, which are closely associated with the recurrence and invasive behavior of BLCA. High-grade bladder cancer (HGBC) is aggressive with a high risk of recurrence, progression, and metastasis and a poor prognosis, whereas low-grade bladder cancer (LGBC) has a relatively good prognosis (7). Radical cystectomy (RC) and pelvic lymph node dissection (PLND) are usually recommended for high-grade non-muscle-invasive bladder cancer (NMIBC) with a high risk of tumor progression and muscle-invasive bladder cancer (MIBC) (4, 8, 9).

In a Canadian multi-medical center follow-up study, 33% of patients undergoing RC recurred and the 5-year overall survival (OS) rate was <60% (10). Therefore, there is a need for separate analyses to identify the prognostic factors most associated with OS in HGBC patients after RC and to individualize the management of postoperative patients to improve the effectiveness of surgical treatment. However, the development of predictive models for OS in HGBC patients after RC has not been performed so far.

Nomograms serve as useful statistical models to integrate relevant factors to predict individual tumor prognosis, advancing the development of personalized medicine and assisting physicians in developing treatment plans (11). In our study, we obtained the clinical and pathological characteristics of HGBC patients undergoing RC from 2004 to 2015 from the Surveillance, Epidemiology, and End Results (SEER) database and identified the risk factors to build a practical nomogram that helps predict overall survival time. In addition, we evaluated the performance of the nomogram and validated the applicability of the nomogram internally and externally separately to make the prediction model more convincing.

BLCA is one of the major global health problems with a high morbidity and mortality rate. Although various treatments have been developed, RC with PLND remains the standard treatment method for MIBC and high-risk NMIBC. Based on the histological differences among cancer cells, bladder urothelial carcinoma was divided into high-grade and low-grade urothelial carcinoma of the bladder. Among high-risk NMIBC and MIBC, most of them are high-grade urothelial carcinoma, and high-grade bladder urothelial carcinoma has a high risk of recurrence and metastasis (7). Predictive models for OS for HGBC patients undergoing RC have never been available before. Therefore, it is necessary to explore the prognostic factors that affect the survival benefit of patients undergoing surgery in order to facilitate follow-up counseling and follow-up programs for HGBC patients undergoing RC.

Our study selected 4,541 HGBC patients from the SEER database, which were randomized to the development cohort (n = 3,178) and the validation cohort (n = 1,363) in a 7:3 ratio, and we also selected patients from our hospital (n = 218) as an external validation cohort. Based on the multivariate Cox regression analysis, T stage, positive lymph nodes, age, chemotherapy, RLNE, and primary tumor size were identified as independent prognostic predictors of OS. On the basis of these factors, we developed a nomogram to predict the OS of HGBC patients undergoing RC for 1, 3, and 5 years, and we further evaluated the performance of the nomogram. The findings suggest that T stage is the most important factor for the model. Several studies have shown that the depth of bladder infiltration significantly affects the prognosis of BLCA, as its aggressiveness and progressiveness increase significantly with the progression of T stage (15–18). The fact of more positive lymph nodes has been a disadvantage for patients undergoing RC, which is consistent with previous reports (8). Additionally, in this study, we found that the number of RLNE was significantly associated with OS of HGBC patients undergoing RC. With more RLNE, the prognosis is better (19–21). It suggested that lymph node dissection not only removes undetected micrometastases probably but also provides pathological information about regional lymph nodes to help clinicians develop subsequent treatment plans for this population (22). Age is also a prognostic factor, which may be related to the fact that elderly patients are more likely to have other chronic diseases. In the meantime, consistent with previous studies, chemotherapy helped improve OS (23, 24). In addition, our study also confirmed that primary tumor size was also associated with different oncology outcomes (25–27). Moreover, the results of the C-index and AUC proved that our nomogram had good concordance. In fact, the results of DCA confirm the clinical usefulness of our nomogram. In a word, the variables used to construct the prognostic model are statistically reliable in this study.

In recent years, many researchers have contributed to the development of postoperative predictive models for bladder cancer (26, 28–31). Although those models have been well validated externally, they may not be universally applicable. Because some of the incorporated variables are often difficult to obtain, such as lymphovascular infiltration, and are relatively complex to calculate, so further generalization of the prediction model is hindered. In addition, Yang et al. constructed a prognostic model to assess the cancer-specific survival of BLCA patients treated with RC (26). The final model included five variables: T stage, marital status, N stage, tumor size, and chemotherapy. However, the model lacked the factor of the number of RLNE and ignored the effect of individual PLND. Compared with previous prognostic models for BLCA patients treated with RC, we constructed an original predictive model for HGBC patients receiving RC.

However, this study has some limitations. First, our study was based on the SEER database and, therefore, lacked potentially important factors, such as preoperative laboratory results, lymphovascular infiltration, comorbidities, socioeconomic status, and Charlson comorbidity index, which can cause a significant impact on the prognosis of patients. Also, we excluded patients who have unknown information on many variables, which could lead to the appearance of selection bias. In addition, the SEER database lacks specific information on chemotherapy and surgical treatments (open or robot-assisted radical cystectomy, type of urinary diversion, and approach for urinary diversion). The factor of chemotherapy in our model represented that the patients have received chemotherapy, whether it is neoadjuvant or adjuvant chemotherapy. Moreover, the sequence of chemotherapy, the choice of chemotherapeutic drugs, and the surgical methods may have different prognostic effects (32–34). Previous strong evidence manifested that compared with open radical cystectomy, robot-assisted radical cystectomy had more advantages in reducing the blood transfusion rate by 50% and resulted in a statistically significant increase in days alive and out of the hospital over 90 days (35–38). However, a large number of patients and a population-based design can enhance our model and reduce potential confounding effects. In addition, some new therapeutic approaches such as immunotherapy and molecularly targeted therapies are also important variables (39, 40), but there is a lack of relevant information in the SEER database. Since our study is retrospective, the results should be treated with caution. Lastly, even though we used an external validation cohort to validate the model, further prospective studies are still needed to validate our conclusions.

In short, this study used a large number of cases with demographic information, clinicopathological factors, and treatment characteristics to construct and validate a nomogram for predicting the OS of HGBC patients after RC. This nomogram can help clinicians make follow-up plans and subsequent treatment options to improve patient prognosis.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding authors.

This study was approved by the Institutional Review Board of the First Affiliated Hospital of Nanchang University. All data from the SEER database were open access.

TC proposed the research idea. YLi collected the data and wrote the manuscript. TC and YLi completed the statistical analysis. LC, YLu and BF reviewed the research framework. All authors contributed to the article and approved the submitted version.