Association of liver cirrhosis with the prognosis of patients with intrahepatic cholangiocarcinoma: a SEER-based cohort study

The association of liver cirrhosis with the prognosis of intrahepatic cholangiocarcinoma (ICC) remains controversial. We aimed to evaluate the relationship between liver cirrhosis (brosis score) and the long-term prognosis of patients with ICC. SEER 18 registry from 2004 to 2015 was queried for this study. Propensity score matching (PSM) was performed to eliminate possible bias. In addition, multivariable analysis was utilized to adjust for potential confounders. The interaction test was performed to explore the impact of each stratied factor on the relationship between FS and patient survival. Overall survival (OS) and disease-specic survival (DSS) were the major endpoints.


Abstract Background
The association of liver cirrhosis with the prognosis of intrahepatic cholangiocarcinoma (ICC) remains controversial. We aimed to evaluate the relationship between liver cirrhosis ( brosis score) and the longterm prognosis of patients with ICC. Methods SEER 18 registry from 2004 to 2015 was queried for this study. Propensity score matching (PSM) was performed to eliminate possible bias. In addition, multivariable analysis was utilized to adjust for potential confounders. The interaction test was performed to explore the impact of each strati ed factor on the relationship between FS and patient survival. Overall survival (OS) and disease-speci c survival (DSS) were the major endpoints.

Conclusion
Our outcomes indicated that brosis score is an independent risk factor for both overall and tumorspeci c survival of ICC patients.

Background
Intrahepatic cholangiocarcinoma (ICC), the second most common primary liver tumor, shows increasing incidence rates over the past decades worldwide. [1] Liver resection (LR) is the mainstay treatment with curative intent for ICC and is associated with improved long-term survival in selected patients. [1][2][3][4] However, the prognosis after LR for ICC remains poor, and risk factors related to survival after LR have not been clearly elucidated. [5][6][7][8] For patients with hepatocellular carcinoma (HCC), liver cirrhosis (usually resulted from hepatitis B or C viral infection) has been shown to be one of the risk factors associated with poor prognosis after LR. [9] However, several studies showed controversial outcomes about the exact relationship of liver cirrhosis with the prognosis of ICC after LR. [10][11][12] In the present study, utilizing the Surveillance, Epidemiology, and End Results (SEER) database, we aimed to explore the independent relationship between liver cirrhosis ( brosis score) and the long-term prognosis of patients with ICC receiving surgical or non-surgical treatment.

Patients And Methods
Study population

Statistical analysis
The primary endpoint in this study was overall survival (OS), de ned as the period from the date of diagnosis to death of any cause. Disease-speci c survival (DSS) was a secondary endpoint, which was de ned as the interval from diagnosis to death from ICC. Categorical and continuous data were compared by utilizing the χ2 test and the Student t test, respectively. The survival curves were determined by the Kaplan-Meier method and differences between groups were compared by log-rank test. Multivariable Cox proportional hazards regression models were used to analyze risk factors for survival outcomes. The interaction test was performed to explore the impact of each strati ed factor on the relationship between FS and patient survival. P value for interaction less than 0.05 means interaction exists between that factor and the relationship.
Propensity score matching (PSM) was performed on the variables of age, race, sex, marriage status, year of diagnosis, tumor grade, treatment methods, lymph node status and AJCC-TNM stage. Patients were matched with the closest estimated propensity score within 0.01 of the standard deviation of the logittransformed propensity score. After PSM selection, a new cohort including two comparable groups were obtained and compared by univariable Cox regression. P value less than 0.05 was deemed statistically signi cant. All statistical analyses were performed by R (http://www.R-project.org) and EmpowerStats software (www.empowerstats.com, X&Y solutions, Inc. Boston MA).

Multivariable analyses
A total of 540 patients with available prognostic data were included in survival analyses. In the total cohort, the mean OS times for patients with LFS and HFS were 33.1 months and 25.3 months, respectively. The mean DSS times for patients with LFS and HFS were 33.4 and 26.7 months, respectively. Overall, patients with LFS had longer OS (P = 0.001) and DSS (P = 0.017) compared to patients with HFS ( Fig. 2A and 2B). Table 2, in the multivariable adjusted cohort (OS: n = 540; DSS: n = 417), after adjusting potential confounding factors, patients with HFS had worse OS (HR, 1.43; 95% CI, 1.10 to 1.85; P = 0.007) and DSS (HR, 1.46; 95% CI, 1.08 to 1.97; P = 0.013) compared to patients with LFS.

As shown in
Long-term outcomes in PSM-adjusted population As shown in Table 1, in the matched cohort, the potential prognostic variables became well-balanced for most baseline characteristics. In the PSM cohort, patients with HFS showed worse OS (P = 0.014) and DSS (P = 0.022) compared to patients with LFS ( Fig. 2C and 2D). In the propensity-matched cohort, the univariable analysis showed that patients with HFS still had worse OS (HR, 1.50; 95% CI, 1.08 to 2.09; P = 0.016) and DSS (HR, 1.54; 95% CI, 1.05 to 2.26; P = 0.026) compared to patients with LFS (Table 2).

Strati ed Analyses
In the total cohort, in multivariable analyses strati ed by clinicopathologic features (Fig. 3 and Fig. 4), patients with HFS were found to have signi cantly worse OS and DSS compared with those with LFS across all the subgroups (all P values for interaction > 0.05).

Discussion
In the present study, by utilizing conventional multivariable analyses and PSM methods, we demonstrated that patients with patients with F5-6 brosis had signi cantly worse long-term survival than patients with F0-4 brosis. In addition, the role of FS in the survival of ICC patients was consistent across all subgroups. Several publications have explored the clinical signi cance of liver cirrhosis for prognosis in patients with ICC after hepatectomy. [10][11][12] Actually, due to the advanced cirrhosis, a large proportion of patients (approximately 65%) could not be treated with surgery. In this study, we enrolled patients with ICC receiving both surgical (including LR, LT and LTD) and non-surgical treatment to reach more representative conclusions. Additionally, we utilized tumor-speci c survival as a major endpoint and found that patients with F5-6 brosis still had a worse survival.
Liver cirrhosis have been demonstrated to be a risk factor for long-term survival in patients with HCC. [9,13] However, owing to the limited number of studies, data related to survival results and prognostic factors of cirrhotic ICC is still insu cient. Recently, Jeong et al. found that the presence of cirrhosis did not signi cantly in uence the long-term prognosis of patients with ICC. [10] However, this study did not provide a convincing evidence due to the limited sample size (included only 106 resected patients with ICC). Study of Li et al. highlighted that cirrhosis was an independent factor associated with poor prognosis in patients with ICC receiving surgery, [11] while they did not perform strati ed analyses based on some signi cant factors such as tumor stage, tumor size and lymph lode status. In addition, they only used OS as the endpoint without DSS.
Liver cirrhosis have been shown to be a risk factor for early tumor recurrence in patients with HCC, which is likely because of multicentric de novo carcinogenesis in the remnant part of the liver. [14] In patients with ICC, the prognostic differences between different FS groups may be attributable to the same reason.
In addition, difference in treatment strategies for recurrent tumor may also account for differences in long-term survivals and patients with advanced cirrhosis had less chance for reoperations.
[15] However, further studies are needed to explore the mechanisms related to the prognostic differences.
The present study had several limitations. First, liver function, performance status, presence of comorbidities are also signi cant factors associated with long-term survival in ICC patients. However, the SEER database does not provide these data, thus we cannot evaluate the in uence of these risk factors in multivariable analyses. Second, details on the etiology of ICC were extremely limited and we could not know the causes of liver cirrhosis of the included patents (such as the proportion of HBV or HCV infection). Third, we cannot acquire the data about pre-or post-operative treatment (such as transcatheter arterial chemoembolization), thus the in uence of adjuvant treatments was not considered in the analytic process. Fourth, the ndings of the current study should be interpreted cautiously because a number of patients were excluded from our main analyses due to having an unknown covariates in the SEER database.
In conclusion, our outcomes indicated that FS 0-4 patients had an overall and disease-speci c survival bene t for ICC patients. ICC patients with cirrhosis should be followed-up carefully after treatment. In addition, further studies are still needed to uncover the mechanisms related to the prognostic differences between HFS and LFS groups.    Figure 1 Flowchart representing selection process of patients included in this study. SEER, Surveillance, Epidemiology, and End Results; FS, brosis score; AJCC, American Joint Committee on Cancer.