Survival outcomes of combined hepatocellular‐cholangiocarcinoma compared with intrahepatic cholangiocarcinoma: A SEER population‐based cohort study

Abstract Background Combined hepatocellular‐cholangiocarcinoma (CHC) is a heterogeneous group of primary liver cancers characterized by the coexistence of both hepatic and biliary cellular contents. The aim of this study was to compare CHC and intrahepatic cholangiocarcinoma (ICC) and investigate the treatment and survival of patients with CHC. Methods Data on CHC and ICC, including clinicopathological characteristics, treatments, and survival outcomes were extracted from the SEER database between 2004 and 2016. Univariate and multivariate analyses of all data were performed to identify the risk factors associated with survival outcomes. The overall survival (OS) rates of CHC patients who underwent hepatic resection (HR) or liver transplantation (LT) were also assessed before and after propensity score matching. Results A total of 1066 consecutive patients who had been diagnosed with CHC (n = 286) or ICC (n = 780) were identified. The mean age of the CHC cohort was 60.8±10.7 years old. Among the CHC group, a large proportion of the patients were men and of White ethnicity (73.1% and 71.3%, respectively). The majority of tumors were poorly differentiated (37.8%), while the most common AJCC stage at presentation was stage I (31.4%). Multivariable analysis of all CHC patients revealed that only tumor size, M1 stage, AJCC stage IIIC, AJCC stage IV, surgery, and chemotherapy were significantly associated with OS. The OS was comparable with the ICC in the initial 36 months and better in the subsequent follow‐up after treatment. Surgery was associated with better survival outcomes, whether in the early or advanced stages. Regarding the specific types of surgery, the OS rates were similar in selected patients following HR or LT. Conclusion In patients with CHC, surgical intervention resulted in better long‐term survival outcomes than nonsurgical treatments. The OS rate of CHC patients compared with that of ICC patients was discriminated before and after a 3‐year follow‐up.


| INTRODUCTION
Combined hepatocellular-cholangiocarcinoma (CHC) is a rare subtype of primary liver cancer with biphenotypic characteristics of both hepatocytic and cholangiocytic differentiation, accounting for 0.4%-14.2% of primary liver malignancies. [1][2][3] As the name implies, CHC was currently viewed as a constitution with the dual clinicopathological features of hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC). 4,5 Due to its rarity and dual nature, the disease is not yet well understood. The ambiguous biological features of CHC may hinder the formulation of treatment protocols and seriously affect the prognosis of patients. Previous studies have reported that CHC and ICC have comparable survival outcomes. 6 Molecular analysis suggested that CHC and ICC shared common altered carcinogenic pathways and were genetically closer compared to HCC, 7 while other genomic profiling uncovered genetic similarities between CHC and HCC. 8 To date, there have been limited investigations of patients with CHC. As one subtype of the uncommon liver cancers, the demographics, clinical features, survival outcomes, risk factors, and therapeutic landscapes especially the role of radical resection remained largely unknown. Therefore, the purpose of this large population study based on the SEER program was to clearly define the clinicopathologic characteristics and to evaluate the survival outcomes as well as different therapeutic regimens.

| Database
Patients pathologically diagnosed with CHC and ICC were identified from the Surveillance, Epidemiology, and End Result cancer database (SEER) between 2004 and 2016. A total of 1066 patients were included in this retrospective study. Of these, 286 patients were confirmed to have CHC and 780 patients had ICC. The data on demographics, American Joint Committee on Cancer (AJCC) stages, histopathological characteristics, treatments, and survival outcomes were collected from the database. The patients with incomplete information regarding the clinical variables were not included.

| Statistical analysis
Continuous variables were described as means (±SD) or medians (interquartile range) and analyzed using the Student's t test or the Mann-Whitney U test, as appropriate. Categorical data are shown as number (percentage) and were assessed using the chi-square (χ²) test or Fisher's exact test, as appropriate. The overall survival (OS) and cancer-specific survival (CSS) curves were calculated using the Kaplan-Meier method and estimated using the log-rank test. The univariate and multivariate analyses of all data were performed using the Cox proportional hazards regression model. A propensity score matching (PSM) method using logistic regression was utilized to correct the case-mix in order to overcome the selection biases of baseline covariates between groups. 9,10 The cohorts were matched for propensity scores at a 1:1 ratio with a caliper of 0.02. The confounding factors included in the model were age, race, tumor size, and marital status. Before PSM, quantitative data were analyzed by the student's t test or the Mann-Whitney U test, while qualitative data were compared using the χ 2 test or Fisher's exact probability test, as appropriate. After PSM, the matched groups were compared using a paired t-test or Wilcoxon signed test for continuous variables and the McNemar test for categorical data. A two-sided p value of less than 0.05 was considered as statistically significant. All the analyses were conducted using SPSS version 26.0 and R software version 3.6.2.

| Baseline characteristics
A total of 1066 patients were enrolled in this study between 1975 and 2016. The baseline characteristics of all patients and the comparison of CHC and ICC patients are summarized in Table 1. Among the patients, 286 and 780 patients were in the CHC and ICC cohorts, respectively. The clinicopathological features in Table 1 were significantly different except for the marital status, combined metastasis in the brain, and whether or not the patient received radiation. The CHC group included a larger proportion of men (73.1% vs. 26.9%) and had a majority of White ethnicity (71.3%, 204/286). The mean age of the CHC cohort was 62.8 ± 10.7 years old, which was significantly younger than the ICC cohort (p < 0.001). The patients in the CHC group showed a distinctly higher incidence of elevated AFP levels compared to the patients in the ICC group (53.5% vs. 17.7%, p < 0.001). Notably, the rate of liver fibrosis was significantly higher in CHC patients (12.2% vs. 3.8%, p < 0.001). A large proportion of CHC patients were diagnosed at early stages (stage I and II: 56.6%). Of the patients, 52.8% (151/286) underwent surgery and 42.0% (120/286) underwent chemotherapy in the CHC cohort. Detailed data are presented in Table 1.

| Survival outcomes
Of all the enrolled patients, the OS and cancer-specific survival (CSS) were significantly better in the CHC group than in the ICC group. (Figure 1 Figure 2) In other words, an OS rate estimate of less than 3 years might not be sufficient to evaluate the longterm clinical outcomes in patients with CHC.

| Overall survival of CHC and ICC patients at different stages
The OS rates of CHC patients at any stage were not significantly different from those of ICC patients at the  corresponding stage (p > 0.05). (Figure 3) However, landmark analyses to access the OS probabilities by dividing the entire follow-up period into the initial 3 years and the subsequent years revealed different patterns.

| Treatment patterns and the relative survival outcomes in CHC patients
The treatment patterns of the different stages are summarized in Table 3. Surgery and chemotherapy remained the main treatment options, whereas radiation was rarely performed in the CHC cohort. The majority of early stage patients (stage I and II) underwent surgery and were associated with a better OS than other treatment options in the CHC group (p < 0.001). As for the advanced stages (stages III and IV), surgery also demonstrated more favorable OS outcomes compared with nonsurgical modalities (p < 0.001, Figure 4).

| Overall survival of CHC patients with different surgical types
Of all the patients in the CHC group, 106 patients underwent hepatic resection (HR) and 45 patients underwent liver transplantation (LT). Notably, patients who underwent HR tended to be older (60.58 ± 9.62 vs. 55.87 ± 9.71, p = 0.007) and had a larger tumor size (6.14 ± 3.66 cm vs. 3.12 ± 2.07 cm, p < 0.001). In addition, marital status and race differed significantly between the two groups. Except for these remarkably different characteristics, the remaining baseline demographic data did not show statistical differences (Table 4); after propensity score matching (PSM), 29 patients were identified in each matched cohort, and the baseline parameters were well balanced. (Table 4) The long-term outcomes were compared between the HR and LT groups by evaluating the OS before and after PSM. The OS probabilities were significantly different between the two unmatched cohorts (p = 0.016), while in the matched group, the OS rates were comparable regardless of the surgical type (p = 0.440, Figure 5).

| DISCUSSION
The current study showed that CHC was associated with better long-term survival outcomes compared with ICC. Surgery and chemotherapy play an increasingly vital role in prolonging the survival. Although the OS rates were comparable at any stage between patients with CHC and ICC, the total OS rates were better in the CHC group. One explanation may be that a large proportion of CHC patients were diagnosed at an early stage. Containing unequivocal components of both biliary and hepatocellular differentiation, CHC, first reported in 1903, was accompanied by growing clinical concerns due to its distinct clinicopathological features. The latest 2019 WHO classification streamlined its previous histological classification system; additionally, the new edition emphasized CHC simply as a mixed tumor with intimate intermingling elements of HCC and ICC. 11 Notably, the proportion of these two different components was not immutably fixed and invariable, and might be correlated with radiographic as well as physiological features. [12][13][14] As a result, the diverse histological characteristics have led to difficulties in accurate diagnosis and posed challenges in therapy.
Till date, the histogenesis of CHC remains controversial, although some related theories have been proposed. 15,16 Recent evidence has shown that hepatic progenitor cells (HPCs), which reside in the biliary ductules, have the potential to differentiate into hepatic or biliary cells. 17 Thus, several investigators have speculated that CHC may be derived from HPCs. 13,18,19 The American Joint Committee on Cancer (AJCC) 8th edition of TNM staging classified CHC and ICC into one category and shared the ICC-specific staging system. 20 The AJCC staging systems have been revised several times; however, until recently, the CHC did not own specific protocol. Tian et al. 21 constructed a risk prediction model based on the clinical parameters to preoperatively discriminate CHC from HCC or ICC and to tailor the optimal treatment.
Are the clinical and pathological features of patients with CHC identical to those of patients with ICC? The  answer in the current large population-based study was "No." Unlike some previous studies, the baseline characteristics regarding the sex ratio, age at diagnosis, tumor grade and stage, and distant metastasis were significantly different between the two groups. Notably, CHC patients were associated with a high incidence of underlying liver cirrhosis and elevated AFP level in comparison with ICC patients, which was in accordance with previous findings. 22 However, subsequent studies found that the background of liver cirrhosis was not sufficient and was a prerequisite condition for the occurrence of CHC. 23 Due to the overlapping imaging features of both HCC and ICC, a misdiagnosis often occurs in CHC patients. 24,25 Li et al. 26 proposed that combining elevation or discordance of the AFP level with incompatible imaging features of HCC may lead to a diagnosis of CHC. The distinction between CHC and HCC or ICC may largely depend on the composition ratio. Surgical resection remains the cornerstone of curative treatment for patients with CHC. In the current study, surgery yielded better survival benefits than any other treatment option in the early stages. As for patients at advanced stages, curative resection was still associated with favorable outcomes compared with non-surgery. However, both CHC and ICC are regarded as more aggressive malignancies with a worse prognosis than HCC. The 3-year and 5-year OS rates were relatively lower than those of HCC reported in previous findings. [27][28][29] Although there was no significant between-group difference in the OS rate, the patients in the ICC group, as compared with those in the CHC group, exhibited a significant reduction in the OS rate after the initial 36 months following treatments. Considering the regional lymphadenopathy features caused by the biliary cellular component, aggressive surgical intervention, including major hepatectomy and lymphadenectomy, may improve the poor prognosis of CHC patients. 12 The role of lymph node dissection in survival benefits remains a matter of disscussion. 30 In our study, a multivariate analysis in the CHC group revealed that the lymph node status was not an independent risk factor for the OS or CSS. Similarly, the survival benefit of LT remains controversial. It is generally believed that transplantation for CHC patients conferred comparable survival benefits to hepatectomy, albeit inferior to HCC, although there are limited data to substantiate the validity. 31 Using a propensity score matching analysis, our study revealed that the long-term outcomes of selected patients with CHC were comparable following HR or LT, which was consistent with previous findings. 32,33 The standard scheme for systemic chemotherapy remains unclear. CHC patients were more frequently treated according to the treatment strategies for HCC or ICC. 34,35 Further investigation is warranted to establish a standard regimen for CHC.
Our study had several limitations that should be acknowledged. First, inherent selection biases were unavoidable owing to the retrospective design. Second, the absence of detailed information, such as underlying liver diseases, tumor markers, pathologic features, and progression-free survival in the SEER database hampered the execution of more specific analyses. Finally, the diagnostic criteria for CHC may be inconsistent because of the varied pathological classifications.
Despite these limitations, our data demonstrated that CHC, as a rare malignancy of the liver, was associated with a similar OS in the initial 36 months, while a better survival in the subsequent follow-up compared with ICC. Furthermore, surgical intervention could significantly  T A B L E 5 (Continued) improve the prognosis in the early stages. For advanced stages, surgery may be the optimal option.