Frequency and Clinicopathological Features of Distant Metastasis of Bladder Cancer: A SEER-Based Study

Background: bladder cancer (BCa) is a common malignancy in the urinary system. But the hematogenous metastatic pattern of it was poorly explored. The aim of this study was to provide a better understanding of the prognosis of the different distant metastatic pattern from stage IV BCa patients and support for making a suitable clinical decision. Methods: The Surveillance, Epidemiology and End Result database (SEER) provided data for this study include BCa from 2010 to 2015. Kaplan–Meier method was used to evaluate the survival prognosis of patients of metastatic BCa. Cox proportional hazards regression model was utilized to analyzed risk factors. All statistical tests were two-sided. Results: At the time of diagnosis, a total of 6808 eligible patients at stage IV were extracted from SEER database. Patients who suffered metastasis to either one of the four sites occupied 25.31% (1723/6808) of BCa. Bone was the most common distant metastatic site of BCa (1225 cases), and brain metastases had the worst prognosis whose mean survival was 6.282 months. The results of univariate survival analysis presented that diagnostic age, race, gender, primary site surgery, tumor size, T stage, N stage, primary tumor site, histology, marital status and metastatic number were independent prognostic factors affecting overall survival (OS) (P<0.05). On multivariable Cox regression, squamous cell carcinoma was an independent risk factor affecting the overall survival (P < 0.05). The nomogram model was constructed to show the 1-, 3- and 5-year survival rates of patients. Conclusion: In analysis of single metastatic sites, patients with brain metastasis had the worst overall survival and lung metastasis had the best outcomes than other three distant metastases. Knowledge of these differences in metastatic patterns is helpful for clinicians to make better pre-treatment evaluation and clinical decisions. gender, insurance status and marital status were not signicantly different between patients with and without liver metastasis. Features for patients with lung metastasis were comparable to those with bone metastasis, including unmarried predominant and T3 and N1 with the lowest percentage. Patients with lung metastasis were older than those without. The difference from other metastases is that the metastatic rate in black patients was much higher than p = 0.015. The ureteric orice was the primary site that accounts for the highest percentage of lung metastasis. Whatever the metastasis site was, patients underwent primary tumor surgery had a lower percentage than those not.

We gather the individuals in the "Any Medicaid", "Insured", and "Insured/No speci cs" groups as "Insured groups" for insurance status. As result, there was for two groups: "Insured" and "Uninsured".
Because of the confused de nition of "Unmarried or domestic partner", this item was removed when we include marital status. "Unmarried" groups contained single, widowed, and separated/divorced.
According to the ICD-O-3, we selected the highest number of types of histology, and grouped as "Non-Papillary Transitional cell carcinoma", "Papillary transitional cell carcinoma", "Squamous cell carcinoma" and "Small cell carcinoma".
Non-surgical group performed only autopsy without surgery of primary site.
Distant metastatic sites were composed of bone, brain, liver and lung. Meanwhile, survival status, follow-up time and causes of death were also drawn from the SEER database.

Statistical analysis
Mean and interquartile range (IQR) were generated for diagnostic age. Pearson's chi-square tests were utilized to compare the distribution of clinicalpathogical and demographic characteristics. Kaplan-Meier method was used in univariate analysis of the above variables and calculation of the survival curves of overall survival (OS). The evaluation of the survival between different groups was implemented in the log-rank test. The variables in multivariate analysis must be consistent with P-value < 0.05 of OS in the log-rank test. Multivariate Cox proportional hazards regression models were carried out to assess the independent risk factors of bladder cancer.
All above statistical analyses were performed by using the SPSS 20.0 software package (IBM Corporation, Armonk, NY, USA). All statistical tests were set at two-sided, and P < 0.05 was examed statistically signi cant.

Patient characteristics and metastasis pattern
A total of 6808 patients with distant metastasis were extracted from the SEER database, including 4878 males (71.65%) and 1930 females (28.35%). The median age of this population was 70 years old. At the time of diagnosis, 1225 patients suffered from bone metastasis and 1123 patients suffered from lung metastasis. These two sites were the most common sites in single metastasis. Table 1 showed the demographics and metastasis pattern features of BCa patients. NOS, not otherwise speci ed. a The comparison between patients with and without bone metastasis. b The comparison between patients with and without brain metastasis. c The comparison between patients with and without liver metastasis. d The comparison between patients with and without lung metastasis The mean age of patients with bone metastasis was 69 years old, but there was no signi cant between patients with and without bone metastasis. Males presented a higher percentage of bone metastasis than females. Urachus had the lowest percentage in the primary tumor sites. Squamous cell carcinoma possessed lower bone metastatic percentage than any other type. Notably, tumor in T3 stage had the lowest bone metastatic rate, as well in N1 stage.
Unmarried patients appeared to tend to a higher bone metastasis percentage. Obviously, patients with bone metastasis were more liable to operation. Insurance status and race showed no statistical signi cance in the metastasis of bone.
The same as the bone metastasis, patients with T3 stage or N1 stage had the lowest percentage of brain metastasis in T stage and N stage, respectively.
Patients undergoing surgery had a lower brain metastatic rate, as we expected. It is worth mentioning that no brain metastasis was observed in the ureteric ori ce and squamous cell carcinoma. No signi cant difference was concluded in the age of diagnosis, gender, race, and insurance and marital status.
As for liver metastasis, the mean age of patients with metastasis was 1.5 years older than those without. Interestingly, patients with the primary tumor in urachus had the highest percentage of liver metastasis. Just like the bone and brain metastasis, T3 stage and N1 stage still occupied the lowest percentage of liver metastasis. The patients in surgery without liver metastasis had higher percentage than those with liver metastasis. Race, gender, insurance status and marital status were not signi cantly different between patients with and without liver metastasis.
Features for patients with lung metastasis were comparable to those with bone metastasis, including unmarried predominant and T3 and N1 with the lowest percentage. Patients with lung metastasis were older than those without. The difference from other metastases is that the metastatic rate in black patients was much higher than others, p = 0.015. The ureteric ori ce was the primary site that accounts for the highest percentage of lung metastasis. Whatever the metastasis site was, patients underwent primary tumor surgery had a lower percentage than those not.

Distribution of patients with distant metastases
As is illustrated in Fig. 1, the distribution and divergences of the metastatic site were obvious. Bone was the most common metastasis (752cases), followed by lung (627cases), liver (313cases) and brain (31cases). Most patients (1723 cases) were single site metastasis, followed by two sites (523 cases), three sites (120 cases), and four sites (9 cases). Patients with bone and lung metastases were the most in two sites metastases, and bone and liver and lung were the most common metastatic types in three sites.

Pathological feature of distant metastases
The 4 most common pathological types were selected from the SEER database for representing the relationship between the pathology and metastatic sites.
As shown in Fig. 2, the percentage of exclusive bone was higher to bones and others in types of transitional cell carcinoma, papillary transitional cell carcinoma and squamous cell carcinoma. The same phenomenon recurred in the lung metastasis. As for liver metastasis, exclusive metastasis surpassed liver and other metastases no matter the histological type. The percentage of exclusive brain metastasis and brain and others were equal in small cell carcinoma.

Univariate survival analysis of distant metastases
As is showed in Table 2, all four metastatic sites presented statistically signi cant (P < 0.001) using univariate survival. On Kaplan-Meier analysis, the differences between patients with and without distant metastasis were statistically signi cant on OS (Fig. 3). Patients with lung metastasis possessed the longest mean survival time (MST) and brain metastasis experienced the lowest MST.   The construction of the nomogram model was based on the results of multivariable Cox regression and clinical feasibility (Fig. 4). We selected age, primary surgery, tumor size, T/N stage, histology and metastatic sites as in uence factors integrating into nomogram model. Every factor has its own corresponding score for the prediction of the survival possibility. For instance, patients with surgery scores 0 points, those without scores 44 points. The total point for all factors is 450. Adding up corresponding scores, it is accessible for us predicted 1-year survival rate, 3-year survival rate and 5-year survival rate, individually.

Discussion
BCa is the most common cancer in the urinary system followed the prostate cancer in males [1]. According to a population-based retrospective study, the mortality rate uctuated slightly after the diagnosis of metastatic BCa without amelioration over the past few years [8]. From 1993 to 2000, the 5-year relative survival of AJCC stage IV BCa was only 17.4% based on the National Cancer DataBase [13]. Consequently, it is necessary to analyze the relationship between different metastatic pattern and overall survival in AJCC stage IV BCa patients In our study, we found that: (1) bone and lung were the most common metastatic sites; (2) male patients outnumber females in stage IV; (3) squamous cell carcinoma was the poorest prognostic histological types; (4) T3 had better survival outcomes than any T stage; (5) on multivariate analysis, the primary tumor surgery, tumor size and histology were independent prognostic factors impacting the patients with distant metastasis.
The results of our study were aligned with previous research, bone and lung were the most common site of distant metastasis for BCa [9]. As for bone metastasis, the coexistence of both osteolytic and osteoblastic lesions can be observed [14]. If the bone metastatic lesions do not be treated properly, some skeletal-related events (SREs) would be occurred, which can make people debilitative [15,16]. Those events include pathologic fractures, spinal cord compression, radiation therapy to bone, surgery to bone, and hypercalcemia of malignancy [17]. Kinnane et al. also concluded that the great mass of patients with bone metastasis would suffer one SRE at least without bone-targeted therapies [18]. From these above, bone-targeted treatments are inevitable in case of the occurrence of SREs. For lung, there are numerous studies revealed the association between lung metastasis and survival. Interestingly, no survival advantages were presented in metastasectomy [9]. Meanwhile, Luzzi et al. discovered that diameter of the metastatic lesions < 3 cm was an independent prognostic factor that possessed higher 5-year survival in lung metastasis [19].
On multivariate analysis, primary tumor site was an independent prognostic factor of OS (P < 0.05), with patients who suffered from tumor in ureteric ori ce obtaining the shortest overall survival. This conclusion is consistent with the outcomes of Eric et al, which discovered that the patients with primary tumors involving the ureteral ori ce had 12.9% recurrence rate of upper urinary tract. [20]. This phenomenon might be caused by the loss of the anti-re ux mechanism after resected the ori ce, and thus the tumor cells seeded to the upper urinary tract [21]. However, Mano et al reported their outcomes in 65 patients, which revealed only one patient was diagnosed with recurrent tumor in the upper urinary tract after transurethral resection of bladder tumor from ureteral ori ce, and their conclusion suggested that ureteral ori ce resection was not very relevant with an increased risk of tumor recurrence in the upper urinary tract [22]. These outcomes disagree with the results of Eric et al. and us, which might be caused by the small number of patients.
Consistent with the results of other studies, primary surgery was a bene t for the survival of patients with locally advanced disease [23]. Nevertheless, only surgery does not curative enough for locally advanced and metastatic BCa patients [24]. As a chemosensitive tumor, adjuvant therapy as chemotherapy always be proceeded within the perioperative period. Herr et al. reported 80 underwent radicalcystectomy after chemotherapy, 24 of the 80 cases (30%) presented no active cancer [25]. Currently, open radical cystectomy (ORC) plus pelvic lymphadenectomy is the gold standard for high-grade invasive BCa and bene ts local cancer control [26]. Meanwhile, laparoscopic radical cystectomy (LRC) is an alternative treatment for those patients, which can reduce morbidity in the perioperative period [27]. Robot-assisted radical cystectomy (RARC) has many considerable advantages, but there are some studies revealed that LRC and RARC have no obvious advantages on ORC in controlling postoperative complications [28,29].
Notably, urachus is the rarest site of BCa in our study, which is consistent with what we now know from early literature [30,31]. Adenocarcinomas is the most common pathological type of urachus malignancy [32]. In some earlier reports, because of the lack of effective treatment protocol, the 5-year survival rate of locally advanced urachus cancers was 43% [31]. With the progress in surgical techniques and adjuvant therapies in recent years, the 5-year survival rate has risen to 50% [33,34]. Even so, urachal cancer exhibited a better prognosis than other types of bladder cancer whatever the order of initiation of tumor formation [35,36].
To our knowledge, this is the rst study focusing on the hematogenous metastatic pattern of BCa patients based on SEER database. Inevitably, there are obvious limitations due to the limited information of SEER database and the retrospective nature of this kind of study. First of all, the metastatic data to above 4 sites were provided since 2010 and follow-up time is not very long. Secondly, comparing to those patients with synchronous metastasis, metachronous metastasis may possess larger quantity. Besides, information on other metastatic sites is bland, such as upper urinary tract and adrenal gland. Moreover, laboratory parameters including alkali phosphatase, alanine transaminase could not acquire from SEER for assessing the in uence of metastatic sites.

Conclusion
In conclusion, patients with brain metastasis have the worst OS in single metastases. In contrast, lung metastasis possess the better survival outcome than the other three distant metastasis. To our knowledge, BCa patients with AJCC stage IV is under a poor prognosis and need a proper pre-treatment evaluation urgently. Knowledge of these differences in metastatic patterns is helpful for clinicians to make better pre-treatment evaluation and clinical decisions.

Declarations
Ethics approval and consent to participate All procedures performed in studies involving human participants were in accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. We signed the 'Surveillance, Epidemiology, and End Results Program Data Use Agreement' in accordance with the requirement of using SEER database. Approval was waived by the local ethics committee, as SEER data is publicly available and de-identi ed. Figure 1 Venn diagram of the distribution of distant metastatic sites of bladder cancer. There were four types of metastatic sites in 6808 patients.

Figure 2
The distribution of metastases in a single versus multiple concomitant sites, strati ed by pathology categories: bone (A), brain (B), liver (C), and lung (D).