Prognostic factors and nomogram for the overall survival of bladder cancer bone metastasis: A SEER-based study

Abstract

Bone metastasis has a poor prognosis in patients with bladder cancer (BC). This study aimed to construct a prognostic nomogram for predicting the overall survival of patients with bone-metastatic BC (BMBC). The Surveillance, Epidemiology, and End Results database was used to recruit patients with BMBC between 2010 and 2018. Univariate and multivariate analyses were performed to screen for prognostic factors and construct a nomogram. Harrell concordance index, receiver operating characteristic curve, and calibration curve were used to verify the prognostic nomograms. All statistical analyses and chart formation were performed using SPSS 23.0 and R software 4.1.2. A total of 1361 patients diagnosed with BMBC were identified in the Surveillance, Epidemiology, and End Results database. Six independent prognostic factors, including marital status, histological type, T stage, other metastases, surgery, and chemotherapy, were identified and included in the nomogram construction. Among them, chemotherapy contributed the most to the prognosis in the nomogram. The concordance index of the nomogram was 0.745 and 0.753 in the training and validation groups, respectively, and all values of the area under the curve were >0.77. The calibration curves showed perfect consistency between the observed and predicted survival rates. The prognostic nomogram developed in this study is expected to become an accurate and individualized tool for predicting overall survival in patients with BMBC and providing guidance for appropriate treatment or care.

1. Introduction

Bladder cancer (BC) has become the most common urinary system malignant tumor and the 4th-most common malignancy in male patients, with over 83,000 new cases and over 17,000 deaths annually in the USA.^[1] The treatment of BC faces challenges due to its high recurrence and metastasis rates. Approximately 25% of BC patients will develop metastatic disease during the course of the cancer, with a poor prognosis due to the fact that it invades the muscular layer easily.^[2] The 5-year survival rate is 95.8% among BC patients diagnosed in situ, while it is only 4.6% for metastatic disease.^[3] Lymph nodes are the most common metastatic site in BC, but a recent study showed that bones could be considered the most common site for distant metastases in BC, and up to 42.9% of metastatic patients present with bone metastases (BM).^[4] Compared to patients with BM from other genitourinary cancers, those with bone-metastatic BC (BMBC) have the worst prognosis.^[5] The 1-year survival rate was reported to be as low as 21% among patients with BMBC.^[6]

Accurate prediction of the prognosis of patients with BMBC can help clinicians choose an appropriate treatment and further optimize regimens. With the advantage of accurate prediction and visualization, nomograms have been widely applied to predict the survival rate of patients with oncological diseases. Nomograms for BC patients have been constructed in previous studies.^[7–9] However, these nomograms only predict the survival and risk of developing BM in patients with BC, which cannot be used to predict the survival time of patients with BCBM. The aim of our study was to identify prognostic factors for patients with confirmed BCBM and to provide a nomogram to predict overall survival (OS) in patients with BMBC.

2. Materials and methods

2.1. Data source and patient selection

The Surveillance, Epidemiology, and End Results (SEER) database, sponsored by the National Cancer Institute, records cancer incidence and survival data from 18 population-based cancer registries that cover approximately 28% of the US population.^[10] The latest data on BMBC patients were extracted from the SEER database using SEER Stat software 8.3.6 (https://seer.cancer.gov/seerstat/). We screened BC patients diagnosed with BM between January 2010 and December 2018 because the sites of metastases were not recorded before 2010. Exclusion criteria were: patients <18 years of age, with or without unknown bone metastasis, unknown survival time, unknown vital status, and not the first malignant primary tumor. Due to public availability and anonymized patient information, this study was exempted from obtaining approval from the institutional review board.

2.2. Data collection and endpoint

We collected the following variables from the selected cohorts: age (≤40 years, 41–60 years, 61–80 years, and ≥81 years), sex (female or male), race (White, Black, or other), marital status (married, unmarried, and unknown), histological type (transitional cell carcinoma (TCC), squamous cell carcinoma (SCC), adenocarcinoma, and other), grade (I, II, III, IV, or unknown), T stage (T0, T1, T2, T3, T4, TX, or unknown), N stage (N0, N1, N2, N3, NX, or unknown), number of extra-bone organ (brain, liver, and lung) metastases (0,1, 2, or 3), surgery (Yes or No), radiotherapy (Yes or No), and chemotherapy (Yes or No). We then collected data on survival months and vital status as outcome variables. The main endpoint was OS, which was defined as the time from diagnosis until death for any reason. Finally, the patients from the selected cohorts were randomly divided (7:3) into the training and validation groups.

2.3. Statistical analysis

Descriptive statistics were used for demographic information. Survival estimation and comparison among different variables were performed using Kaplan–Meier analysis, and the parameters included mean survival time, median survival time, and a 95% confidence interval (CI). The log-rank test was used to compare the significance of the survival curves. Variables determined to be significant in univariate and multivariate Cox regression analyses were used to generate the nomogram. The hazard ratios (HRs) and corresponding 95% CIs for all strata for each variable were calculated. Harrell concordance index, receiver operating characteristic curve, and calibration curves were used to estimate the predictive performance of the nomogram. All statistical analyses and chart formation were performed using SPSS Statistics version 23.0 (IBM, Armonk, NY) and R software version 4.1.2 (https://www.r-project.org/). Statistical significance was set at P < .05.

3. Results

3.1. Demographic baseline characteristics

A total of 1361 eligible patients were included in this study via a predesigned screening process (see Fig. 1). In the entire group, the majority of the categorical variables were 61 to 80 years old (58.34%), male (74.5%), White (84.06%), married (47.32%), TCC (77%), grade IV (42.69%), unknown T stage (38.57%), unknown N stage (38.57%), no other metastases (57.45%), surgery performed (68.26%), no radiotherapy (68.77%), and no chemotherapy (56.21%) (Table 1).

Table 1 - Baseline characteristics of BMBC patients.

Variables		Overall N (%)	Training group N (%)	Validation group N (%)
		1361 (100)	952 (100)	409 (100)
Age	≤40	14 (1.03)	9 (0.95)	5 (1.22)
	41–60	316 (23.22)	217 (22.79)	99 (24.21)
	61–80	794 (58.34)	553 (58.09)	241 (58.92)
	≥81	237 (17.41)	173 (18.17)	64 (15.65)
Gender	Female	347 (25.5)	249 (26.16)	98 (23.96)
	Male	1014 (74.5)	703 (73.84)	311 (76.04)
Race	White	1144 (84.06)	801 (84.14)	343 (83.86)
	Black	141 (10.36)	98 (10.29)	43 (10.51)
	Other	76 (5.58)	53 (5.57)	23 (5.62)
Marital	Married	644 (47.32)	454 (47.69)	190 (46.45)
	Unmarried	643 (47.24)	446 (46.85)	197 (48.17)
	Unknown	74 (5.44)	52 (5.46)	22 (5.38)
Histological type	TCC	1048 (77)	720 (75.63)	328 (80.2)
	SCC	40 (2.94)	28 (2.94)	12 (2.93)
	AC	48 (3.53)	40 (4.2)	8 (1.96)
	Other	225 (16.53)	164 (17.23)	61 (14.91)
Grade	I	6 (0.44)	5 (0.53)	1 (0.24)
	II	38 (2.79)	24 (2.52)	14 (3.42)
	III	205 (15.06)	148 (15.55)	57 (13.94)
	IV	581 (42.69)	408 (42.86)	173 (42.3)
	Unknown	531 (39.02)	367 (38.55)	164 (40.1)
T stage	T0	14 (1.03)	10 (1.05)	4 (0.98)
	T1	132 (9.7)	100 (10.5)	32 (7.82)
	T2	301 (22.12)	211 (22.16)	90 (22)
	T3	54 (3.97)	37 (3.89)	17 (4.16)
	T4	152 (11.17)	102 (10.71)	50 (12.22)
	TX	183 (13.45)	126 (13.24)	57 (13.94)
	Unknown	525 (38.57)	366 (38.45)	159 (38.88)
N stage	N0	451 (33.14)	307 (32.25)	144 (35.21)
	N1	73 (5.36)	51 (5.36)	22 (5.38)
	N2	144 (10.58)	110 (11.55)	34 (8.31)
	N3	41 (3.01)	30 (3.15)	11 (2.69)
	NX	127 (9.33)	88 (9.24)	39 (9.54)
	Unknown	525 (38.57)	366 (38.45)	159 (38.88)
Other metastases	0	740 (57.45)	509 (53.47)	231 (56.48)
	1	402 (31.21)	284 (29.83)	118 (28.85)
	2	133 (10.33)	96 (10.08)	37 (9.05)
	3	13 (1.01)	9 (0.95)	4 (0.98)
	Unknown	73 (5.36)	54 (5.67)	19 (4.65)
Surgery	Yes	929 (68.26)	664 (69.75)	265 (64.79)
	No	432 (31.74)	288 (30.25)	144 (35.21)
Radiotherapy	Yes	425 (31.23)	302 (31.72)	123 (30.07)
	No	936 (68.77)	650 (68.28)	286 (69.93)
Chemotherapy	Yes	596 (43.79)	414 (43.49)	182 (44.5)
	No	765 (56.21)	538 (56.51)	227 (55.5)

Other metastases, number of other organs metastasis (liver, brain, and lung).

AC = adenocarcinoma, BMBC = bone-metastatic bladder cancer, SCC = squamous cell carcinoma, TCC = transitional cell carcinoma, TX = unstaged.

Figure 1.: Flow chart of patient identification and selection.

3.2. Kaplan–Meier survival analysis for different variables

The mean OS time for 952 BMBC patients was 9.127 months (95% CI: 7.994–10.26 months), and the median OS time was 4 months (95% CI: 3.481–4.519 months). Age (P = .006), marital status (P < .001), histological type (P = .01), grade (P < .001), stage T disease (P < .001), other metastases (P < .001), surgery (P < .001), and chemotherapy (P < .001) were significantly associated with OS. The risk tables show the number of surviving patients at each time point; the 1-, 3-, and 5-year OS were further calculated to 18.3%, 2.8%, and 0.6%, respectively (see Figs. 2 and 3).

Figure 2.: Kaplan–Meier survival curves of BMBC patients (n = 952). BMBC = bone-metastatic bladder cancer.

Figure 3.: Kaplan–Meier survival curves of BMBC patients according to different variables. BMBC = bone-metastatic bladder cancer.

3.3. Prognostic factors for BMBC patients

We identified 6 prognostic factors using multivariable Cox regression analyses (Table 2). The variables in the Kaplan–Meier analysis with a P value ≤ 0.1 were included in the multivariate analysis to avoid omitting possible prognostic factors. Marital status (Unmarried, HR = 1.243, 95% CI: 1.077–1.436, P = .003), histological type (SCC, HR = 1.493, 95% CI: 1.002–2.224, P = .049), stage T (TX, HR = 2.147, 95% CI: 1.100–4.191, P = .025), number of other metastases (1, HR = 1.419, 95% CI: 1.212–1.662, P < .001; 2, HR = 1.867, 95% CI: 1.474–2.363, P < .001; 3, HR = 2.779, 95% CI: 1.425–5.420, P = .003), surgery (Yes, HR = 0.709, 95% CI:0.589–0.854, P < .001), and chemotherapy (Yes, HR = 0.326, 95% CI: 0.280–0.379,P < .001) were identified as independent prognostic factors of OS.

Table 2 - Results of multivariate Cox regression analyses.

	Kaplan–Meier survival analysis	Multivariable analysis
Variables	P	HR (95% CI)	P
Age	.006
≤40		Reference
41–60		1.300 (0.596–2.925)	.52
61–80		1.339 (0.606–2.927)	.47
≥81		1.235 (0.552–2.734)	.608
Marital	<.001
Married		Reference
Unmarried		1.243 (1.077–1.436)	.003
Unknown		0.939 (0.684–1.290)	.699
Histologic type	.01
TCC		Reference
SCC		1.493 (1.002–2.224)	.049
AC		1.158 (0.809–1.657)	.423
Other		1.082 (0.892–1.312)	.425
Grade	<.001
I		Reference
II		1.129 (0.409–3.118)	.815
III		1.674 (0.660–4.246)	.278
IV		1.562 (0.621–3.928)	.343
Unknown		1.615 (0.640–4.079)	.311
T	<.001
T0		Reference
T1		1.539 (0.777–3.049)	.216
T2		1.838 (0.938–3.603)	.076
T3		1.174 (0.568–2.429)	.664
T4		1.952 (0.990–3.847)	.053
TX		2.147 (1.100–4.191)	.025
Unknown		1.534 (0.789–2.983)	.208
Other metastases	<.001
0		Reference
1		1.419 (1.212–1.662)	<.001
2		1.867 (1.474–2.363)	<.001
3		2.779 (1.425–5.420)	.003
Unknown		1.239 (0.906–1.694)	.179
Surgery	<.001
No		Reference
Yes		0.709 (0.589–0.854)	<.001
Radiotherapy	.089
No		Reference
Yes		0.960 (0.825–1.118)	.6
Chemotherapy	<.001
No		Reference
Yes		0.326 (0.280–0.379)	<.001

Other metastases, number of other organs metastasis (liver, brain, and lung).

AC = adenocarcinoma, CI = confidence interval, HR = hazard ratio, SCC = squamous cell carcinoma, TCC = transitional cell carcinoma.

3.4. Prognostic nomogram for OS

A novel nomogram was constructed based on Cox proportional hazards regression models (Fig. 4). The nomogram illustrated that chemotherapy had the largest contribution to prognosis, followed by other metastases, T stage, histological type, surgery, and marital status. Each subgroup of these variables was assigned a score of 0 to 100 on a point scale. By adding up the total score and positioning it on the bottom scales, we can easily draw a straight line down to determine the estimated probability of survival at each time point (6 months, 1 year, and 2 years).

Figure 4.: Nomogram for predicting OS of BMBC patients. BMBC = bone-metastatic bladder cancer, OS = overall survival.

3.5. Validation and calibration of the nomogram

The verification results showed the better discriminatory ability of our nomogram. The C-indices of the nomogram were 0.745 and 0.753 in the training and validation groups, respectively. All areas under the curve values of the receiver operating characteristic were >0.77 in the training and validation groups (Fig. 5). The calibration plots presented an excellent fitting degree between the nomogram prediction and actual observation for OS in the training group (Fig. 6A) and an acceptable agreement between the nomogram prediction and actual observation for OS in the validation group (Fig. 6B), which guaranteed the repeatability and reliability of the established model.

Figure 5.: ROC curve of the prognostic nomogram. (A–C) Validated by the training group. (D–F) Validated by the validation group. ROC = receiver operating characteristic.

Figure 6.: Calibration curves of the prognostic nomogram. (A) Validated by the training group. (B) Validated by the validation group.

4. Discussion

Currently, the incidence of BM in BC patients is between 1.39% and 5.5%, which is a synchronous metastatic disease, indicating that the tumor enters a late stage.^[6] BM from BC remains a partially known field and is difficult to manage owing to the limited data available and the lack of prognostic prediction. Therefore, we sought to develop a prognostic nomogram for patients with BMBC based on a large population database.

The risk factors for BM spread in BC have been well investigated by previous studies and included age, race, marital status, tumor differentiation grade, T stage, N stage, other organ metastasis (liver, brain, and lung), histological type, and primary site.^[6,11] However, few studies have focused on prognostic factors for patients with BMBC. In our study, we identified marital status, histological type, T stage, other metastases, surgery, and chemotherapy as independent prognostic factors for OS in patients with BMBC by applying univariate and subsequent multivariable analyses.

In our nomogram, chemotherapy contributed the most to the prognosis. Although targeted therapy and immunotherapy are promising, chemotherapy still presents maximal survival benefits for patients with BMBC. The 2018 National Comprehensive Cancer Network (NCCN) guidelines recommend platinum-based chemotherapy as the standard of care for BC patients with metastatic disease, with an OS of 9 to 15 months.^[12] Radiotherapy is an effective option for pain relief caused by BM, prevention of local symptoms, and progression of urological malignancies.^[13] However, it did not yield a significant survival benefit for patients with BMBC in our study; thus, it could not be used as a factor for modeling the nomogram. However, the effect of surgery on the prognosis of metastatic BC remains controversial. A systematic review showed that surgical resection of metastases might improve BC control and survival in highly selected patients.^[14] Dong et al also believed that surgery (radical cystectomy and metastasectomy) might lead to survival benefits for these patients.^[4] A recent study indicated that surgery at the primary tumor site was associated with improved survival in patients with metastases who received standard chemotherapy, but this effect disappeared in patients affected by 2 or more metastatic sites.^[15]

The metastatic number of other organs was the second-largest contributor to prognosis in our nomogram. Previous studies have indicated that different distant metastatic sites and multiple site metastases are independent prognostic factors for OS in metastatic BC patients.^[4,9] Once the distant metastatic disease develops in BC patients, it is conventionally viewed as incurable. The greater the number of metastases to other organs, the worse the prognosis for BMBC.

The T stage of the disease was the third largest contributor to the prognosis. A higher T stage in patients with malignant tumors may indicate a larger tumor volume, a wider range of adjacent tissues, and a higher degree of involvement. As shown in our nomogram, TX had the worst prognosis among all T stages. Similar results were found in another study, which reported that the T stage was independently associated with OS in BC patients and that a higher T stage (>T2, HR = 2.04, 95% CI: 1.50–2.77, P < .001) would lead to a higher risk of death.^[16]

The histological type of BC was used to construct a nomogram. The most predominant histological phenotype is TCC, which constitutes approximately 80% of BC.^[17] Bladder SCC accounts for only 2 to 5% of BCs, but it has the characteristics of a high malignancy degree and a high recurrence rate,^[18] and progresses faster than TCC in BC patients at stage III or IV.^[19] In our study, we concluded that SCC was also associated with a worse prognosis.

Our nomogram showed that marital status had a small contribution to predicting the OS of patients with BMBC. We determined that married BMBC patients had significant survival benefits compared to unmarried patients (including single, domestic partner, divorced, separated, and widowed), which revealed that a good marital status was associated with a better prognosis. As shown in previous studies, patients who were unmarried or from poor neighborhoods were not only less likely to receive chemotherapy but also to have worse survival.^[9,20] Favorable financial conditions and emotional support from spouses contributed to cultivating treatment adherence and regular follow-ups in married patients. However, the effect of marital status on the outcomes of BC patients was not stable and susceptible to other factors, such as sex, and the outcome was addressed.^[21]

In addition to the newly discovered prognostic factors, we studied several other factors and analyzed their possible reasons. Advanced age has been considered a prognostic factor for poor survival.^[2,22–24] Interestingly, increasing age is associated with decreased rates of metastases to multiple organs.^[11,25] Furthermore, Janisch et al^[26] reported a null association between age <50 versus >50 years and the risk of BC death. Recently, a study failed to show an association between age and response to survival outcomes from preoperative chemotherapy and suggested that preoperative chemotherapy should be considered regardless of the patient’s age. Cisplatin-based chemotherapy constitutes an integral part of muscle-invasive BC treatment in all eligible patients, but the number of elderly patients eligible for cisplatin decreases largely due to impaired renal function.^[27] Thus, in view of the greatest contribution of chemotherapy to the survival of BMBC patients, the worse OS of older patients may be mainly related to the absence of chemotherapy but not to age itself. In previous studies, Black patients showed higher disease stages, a higher risk of BM, and worse survival rates than White patients.^[28–31] Contrary to these reports, a study by DeDeugd et al indicated that there was no difference in OS in Blacks with BC compared with Whites with BC after adjusting for important covariates such as age, tumor stage, tumor grade, treatment rendered, health insurance, and tobacco history.^[32] Notably, the incidence of tobacco use was higher among White patients (35.2%) than among Black patients (23.2%), which could translate into higher mortality rates.^[32] Thus, these results may signify a more complex relationship between race and BMBC outcome.

To the best of our knowledge, this is the first nomogram for predicting the survival of BC patients with BM based on a large database with long-term follow-up. The OS curve declined much more rapidly before the 2-year cutoff compared with a slow downward trend in patient survival after 2 years (Fig. 2). Furthermore, the 1-year OS rapidly decreased to 18.3%, while the median OS of BMBC was barely 4 months. Thus, we believe that predicting the 6-month, 1-year, and 2-year OS of these patients is of clinical value in treatment planning systems. Both clinicians and patients can perform individualized survival predictions through this easy-to-use scoring system. Identifying subgroups of patients at different risks for poor survival might contribute to treatment or care options.

However, several limitations of our study should be considered with caution. First, this nomogram was limited by the retrospective nature of data collection and the failure to incorporate some potential prognostic parameters, such as smoking status, performance status, comorbidities, and time to metastasis. Second, information on the types of surgery, chemotherapy, radiotherapy, and systemic therapy is lacking. Third, data on marital status, grade, T stage, N stage, and other metastases were partly unknown. Furthermore, information on skeletal-related events is considered one of the most important prognostic factors for urinary malignancies. The lack of such data lowered the prediction accuracy of the model. Finally, the verification of the clinical prediction model requires a larger external dataset.

5. Conclusions

In conclusion, we determined that marital status, histological type, T stage, other metastases, surgery, and chemotherapy are independent prognostic factors for OS in patients with BMBC. A novel prognostic nomogram developed in our study is expected to become an accurate and individualized tool for clinicians estimating the survival of BMBC patients and identifying subgroups of patients who need specific treatment or care.

Acknowledgment

We would like to thank Editage (www.editage.cn) for English language editing.

Author contributions

Conceptualization: Yu Huang, Chengxin Xie, Dong Yin.

Data curation: Yu Huang, Chengxin Xie, Dong Yin.

Formal analysis: Chengxin Xie, Qinglong Li.

Investigation: Chengxin Xie.

Methodology: Chengxin Xie, Xiao Huang.

Project administration: Qinglong Li, Xiao Huang.

Resources: Qinglong Li.

Supervision: Xiao Huang, Wenwen Huang.

Validation: Chengxin Xie.

Visualization: Wenwen Huang.

Writing – original draft: Yu Huang, Chengxin Xie.

Writing – review & editing: Chengxin Xie, Dong Yin.