Treatment of advanced gallbladder cancer: A SEER‐based study

Abstract Purpose The treatment of advanced gallbladder cancer (GBC) remains controversial. Therefore, the purpose of this study was to explore treatment choices for advanced GBC. Methods We identified four different treatments from the surveillance, epidemiology, and end results (SEER) database: surgery, chemotherapy (CT), surgery and chemotherapy (Surgery + CT), and no surgery/no chemotherapy (No surgery/No CT). Kaplan‐Meier method and Cox proportional hazards regression method were used to determine the risk factors for overall survival (OS) and cancer‐specific survival (CSS). In addition, patients in AJCC stages III and IV stage were matched with 1:1 propensity score matching (PSM) for diagnosis age, race, marital status, histological type, tumor grade, and treatment pattern to decrease the possibility of selection bias. Results A total of 288 AJCC stage III patients and 4239 AJCC stage IV patients with advanced GBC were identified from the SEER database between 2004 and 2015. Treatment pattern was an independent risk factor for patients with advanced GBC. For all patient, AJCC stage III patients and AJCC stage IV patients, “Surgery + CT” treatment minimized the OS and CSS in advanced GBC patients. In addition, after the PSM analysis, the “Surgery + CT” treatment still significantly decreased patient OS and CSS. Conclusions “Surgery + CT” treatment can provide survival benefits for patients with advanced GBC. In addition, “Surgery + CT” treatment was not fully utilized and may further improve the survival rate of GBC patients.

The median survival time of patients with GBC is less than 1 year, the overall survival (OS) is approximately 17.8%-21.7%, and the 5-year OS is only 5%. 1,5 GBC treatments include surgery, chemotherapy (CT), radiotherapy (RT), and other immunotherapy. [6][7][8] Although GBC has high invasiveness and metastasis, surgical resection remains recognized as the best treatment. 9,10 The 5-year survival rate of early T1 GBC patients is as high as 95%-100%. However, for patients with T3 stage and T4 stage, the 5-year survival rate is only 23% and 12%, respectively. 11 Both in China and abroad, most diagnosed GBC patients are in moderate and advanced stages. The treatment of GBC still confuses many physicians, even experienced surgeons. The treatment of patients with advanced GBC remains especially controversial. In our study, we used data from the surveillance, epidemiology, and end results (SEER) cancer registration database to explore the treatment options for patients with advanced GBC.

| Patients selection
The SEER database is currently the largest publicly available cancer database, covering approximately 28% of the US population. 12 All cases are from the SEER Program (www.seer. cancer.gov) SEER*Stat database released in November 2017: version 8.3.5; SEER 18 Regs Custom Data (with additional treatment field), Nov 2017 Sub (1973-2015 varying) database. The SEER database contains information about patient demographics and cancer characteristics, such as sex, age at diagnosis, year of diagnosis, race, marital status, tumor grade and stage, histological type, treatment, and patient survival time.
Using the "Primary Site-labeled" variable, we selected tumor cases from the primary site of the gallbladder diagnosed between 2004 and 2015. The study included only patients with American Joint Committee on Cancer (AJCC) stage III and IV cancer. According to the 2004 AJCC staging principle, stage III is defined as "T4M0, any N," and stage IV is defined as "M1, any T or any N." Only patients above 18 years of age were included in this study. Patients with any of the following criteria were also excluded: unknown treatment, not the first tumor, unknown survival time, and unknown marital status. Finally, 4527 eligible patients diagnosed with GBC remained.

| Study variables
Definition and information about the variables of sex, diagnosis age, year of diagnosis, race, marital status, tumor grade, histological type, and survival time can be found in the SEER database. OS and cancer-specific survival (CSS) were the primary study endpoints. For OS, death from any cause was considered as an event, and the survivor was regarded as censored. For the CSS analysis, deaths caused by GBC were considered events, and deaths from other causes or survivors were censored.
For the diagnosis age, we divided all patients into three groups: less than 60 years old, 60-80 years old, and older than 80 years old.
For marital status, patients were divided into a Married group, an Unmarried group, and an Unknown marital status group. Unmarried patients include Single, Separated, Divorced, and Widowed.
For race, patients were divided into a Non-Hispanic White group, a Non-Hispanic Black group, a Hispanic group, and an Others group.
The ICD-0-3 site/histology validation list was used to distinguish adenocarcinoma, squamous cell carcinoma, and other histological types.
Grade was defined by the following codes: well differentiated (Grade I); moderately differentiated (Grade II); poorly differentiated (Grade III); undifferentiated (Grade IV); and unknown grade.

| Statistical analysis
Pearson's chi-square analysis was used to analyze and evaluate the different clinical characteristics between different treatment patterns. The Kaplan-Meier curve was used to estimate the OS and CSS in different groups, and the differences between the curves were analyzed by log-rank test. Univariate and multivariate Cox regression models were performed to estimate the hazard ratios (HR) and 95% confidence interval (CI) to analyze the independent prognostic factors associated with OS and CSS in GBC patients.
According to AJCC stage, the patients were divided into AJCC stage III and IV groups. 1:1 propensity score matching (PSM) was to reduce the selection bia of the two groups of baseline variables, including age, race, marital status, histological type, grade, and treatment pattern seven variables. After PSM, the clinicopathological features of the patients were reevaluated according to AJCC stage. All statistical analyses were conducted with the Statistical Package for the Social Sciences software (version 24.0; IBM Corporation). A P value ≤ .05 was considered statistically significant.

| Demographic and clinical characteristics
From 2004 to 2015, our study cohort included a total of 4527 eligible GBC patients. Among them, 1575 patients with "No surgery/No CT," 938 patients with "Surgery", 1222 patients with "CT" and 792 patients with "Surgery + CT". The demographic and clinical characteristics of GBC patients with different treatment patterns are shown in Table  1. Male patients accounted for 30.7%, and female patients accounted for 69.3%. Moreover, there were 288 patients with AJCC stage III and 4239 with AJCC stage IV. Chisquare test showed significant differences in some variables and treatment patterns, including diagnosis age, sex, race, marital status, histological type, tumor grade, and AJCC stage (All P < .05).

| Trend in different treatment patterns
As shown in Table 2, the proportion of patients who accepted the "No surgery/No CT" or "Surgery + CT" models remained relatively stable between 2004 and 2015. Simultaneously, the number and proportion of patients receiving the "CT" model increased each year. The proportion of the "Surgery" treatment was significantly lower in patients with advanced GBC compared with the increase in "CT" mode ( Figure 1).

| Identification of prognostic factors of OS and CSS in patients with advanced GBC
Univariate and multivariate Cox regression were used to analyze the factors associated with OS and CSS in patients with advanced GBC. Before matching, as shown in Table 3, the age at diagnosis, histological type, tumor grade, AJCC stage, and treatment pattern affected the OS and CSS in patients with advanced GBC. Multivariate Cox regression showed that "surgery" (vs "No surgery/No CT"; HR = 0.60, 95% CI 0.55-0.66, P < .001), "CT" (vs "No surgery/No CT"; HR = 0.44, 95% CI 0.40-0.48, P < .001), and "Surgery + CT" (vs "No surgery/No CT"; HR = 0.29, 95% CI 0.26-0.33, P < .001) were associated with OS ( Figure 2A). Similarly, in terms of CSS, multivariate Cox regression analysis also indicated that treatment pattern was a prognostic factor for patients with advanced GBC ("surgery" vs "No surgery/No CT"; HR = 0.65, 95% CI 0.57-0.74, P < .001; "CT" vs "No surgery/No CT"; HR = 0.44, 95% CI 0.39-0.50, P < .001; "Surgery + CT" vs "No surgery/No CT"; HR = 0.33, 95% CI 0.29-0.39, P < .001) ( Figure 2B). To better characterize the influence of treatment pattern on OS and CSS in patients with advanced GBC, we performed AJCC stage stratification on all patient parameters on the basis of multivariate analysis. We found that the treatment pattern was also an independent risk factor for patients with AJCC stage III and AJCC stage IV (Table  4). "Surgery," "CT," and "Surgery + CT" improve the AJCC stage III and AJCC stage IV patient OS and CSS. In particular, "Surgery + CT" could significantly improve OS and CSS in both AJCC stage III (HR = 0.36, 95% CI 0.22-0.60, P < .001) and AJCC stage IV (HR = 0.33, 95% CI 0.29-0.39, P < .001) groups ( Figure 3).

| Identification of prognostic factors of OS and CSS in 1:1 PSM sample
To better balance the patients with AJCC stage III or AJCC stage IV GBC, we performed a 1:1 PSM for variables such as diagnosis age, race, marital status, tumor grade, and treatment pattern to decrease the selection bias and further examine the Year of diagnosis relationship between treatment patterns and OS and CSS, as assessed with the Cox regression model. First, we performed univariate and multivariate Cox regression analysis of all patients after PSM, and found that only diagnosis age and treatment pattern were independent risk factors (Table S1). "Surgery + CT" significantly improved the OS (vs "No surgery/No CT"; HR = 0.28, 95% CI 0.21-0.37, P < .001) and CSS (vs "No surgery/No CT"; HR = 0.30, 95% CI 0.22-0.43, P < .001) ( Figure S1).
Moreover, we performed AJCC stage stratification on patient parameters after PSM on the basis of multivariate analysis. We found that the treatment pattern was also an independent risk factor for AJCC stage IV patients (Table S2), and "Surgery + CT" significantly improved the OS (vs "No surgery/No CT"; HR = 0.23, 95% CI 0.15-0.35, P < .001) and CSS (vs "No surgery/No CT"; HR = 0.31, 95% CI 0.19-0.52, P < .001) of AJCC stage IV patients after PSM ( Figure S2).

| DISCUSSION
Currently, the treatment of patients with advanced GBC remains controversial. Radical surgical resection is the only possible cure treatment for patients with GBC. However, most patients with GBC are at an advanced stage at the time of discovery, thus limiting the opportunity for radical resection; even in the moderate stage, the prognosis of patients undergoing radical resection is highly unsatisfactory. Therefore, clinicians have been exploring the application and combination of adjuvant treatments, including RT, CT, and other treatments, to improve the prognosis of patients with GBC.
Radical surgical resection remains the most important treatment for improving the survival rate of patients with GBC. 13 Surgeons have long been pessimistic about the treatment of advanced GBC. In recent years, owing to the development of GBC radical surgery, the long-term survival rate has significantly improved. Nakamura et al 14 have reported that in 33 GBC patients with Nevin V stage, 13 patients underwent extended radical resection, and the 1-year and 3-year survival rates were 46% and 23%, respectively, whereas the 1-year survival rate of 20 patients without resection was only 15%. Matsumoto et al 15 reported that the average survival time of 15 patients undergoing extended radical resection was 26 months, whereas that of patients who did not undergo resection was only 10 months.
Patients in stage IV are generally considered unable to undergo surgical resection, 1 but many clinical studies have supported more aggressive surgical treatment of patients with advanced GBC. 16, 17 Kang's 17 study has shown that radical surgery in stage IV GBC patients can prolong survival time. Christina et al 18 have further confirmed this conclusion, suggesting that radical surgery can be performed in stage IV patients as long as the lesion is local and can reach the R0 margin. Studies from Japan also suggest that if the tumor is relatively limited and strictly screened, even if the lesion is large and has invaded adjacent organs, stage IV patients are expected to achieve long-term survive after radical resection. [19][20][21][22] However, increased surgical complications and mortality have hampered the adoption of these radical surgical approaches as a standard treatment for GBC. 23 Similarly, big data research from Japan does not support radical surgical resection in patients in stage IV, and some studies have indicated that radical surgery does not improve prognosis in patients in stage IV. 24,25 According to the 7th edition of AJCC guidelines, patients with stage T4 are usually considered unresectable and should be treated with palliative care. 26 Groot et al 27  Although surgical radical resection of the GBC is currently extensively performed, the rate of radical resection is only 25%-30%. After radical resection, nearly half of patients still have a risk of recurrence of GBC. Therefore, to decrease the postoperative recurrence of GBC patients and improve the prognosis of patients with advanced disease, some patients are given CT treatment. At present, gemcitabine combined with oxaliplatin (GEMOX) or cisplatin and tegafur combined with oxaliplatin (SOX) are widely used and recognized as effective chemotherapy regimens for GBC patients. 33,34 In this large population-based study, we used the SEER database to analyze the best treatment options for patients with advanced GBC. Through univariate and multivariate Cox survival regression analysis, in all patients, AJCC stage III patients and AJCC stage IV patients, we found that the "Surgery + CT" treatment significantly increase the OS (vs "No surgery/No CT"; HR = 0.29, 95% CI 0.26-0.33, P < .001) and CSS (vs "No surgery/No CT"; HR = 0.33, 95% CI 0.29-0.39, P < .001). The results of 1:1 PSM analysis also showed that the "Surgery + CT" treatment significantly decreased the risk of death in patients with advanced GBC. In addition, the proportion of patients with "Surgery + CT" remained relatively stable over the past 12 years, and "Surgery + CT" may not currently be fully utilized. The combination of surgery and CT may improve the survival rate of patients with advanced GBC. This study has several limitations. First, it was a retrospective study and thus had clear inherent limitations. Second, the SEER database lacked information about the physical condition and complications of patients, and older patients may be more likely to choose conservative treatment. In addition, the sequence of surgery and CT, as well as the specific regimen of CT was unknown. Nonetheless, the study remains convincing given the large demographics.

| CONCLUSIONS
We found that the "Surgery + CT" treatment model provided greater survival benefits for patients with advanced GBC. Because this was a retrospective analysis, further prospective studies are needed to provide verification.