The Predictive Value of Staging Systems and Inflammation Scores for Patients with Combined Hepatocellular Cholangiocarcinoma After Surgical Resection: a Retrospective Study

Consecutive patients with pathologically confirmed cHCC-CC after surgical resection performed between June 2000 and January 2017 at the Department of Hepatobiliary and Pancreatic Surgery of Sun Yat-Sen University Cancer Center were enrolled in the present study. The following exclusion criteria were considered: (1) other treatments, including transarterial chemoembolization, radiofrequency treatment, and liver transplantation for patients with cHCC-CC prior to surgical resection; (2) inadequate renal function (serum creatinine and blood urea nitrogen levels higher than the upper limits of normal); (3) severe coagulopathy (prothrombin activity < 40% or platelet count < 40,000/mm3); (4) Child–Pugh C liver function or evidence of hepatocellular decompensation, including refractory ascites, esophageal or gastric variceal bleeding, and hepatic encephalopathy; (5) obstructive jaundice; (6) other concurrent primary tumors; (7) pathological confirmed subtype of cHCC-CC with stem cell features; or (8) lost to follow-up.

A total of 99 patients were included in the present study. For each patient, a computed tomography (CT) or magnetic resonance imaging (MRI) examination was reviewed by two independent radiologists with more than 10 years of experience. The imaging examination was performed using a previously reported procedure.17 Patients were allocated into imaging HCC-dominant (IHD) and imaging ICC-dominant (IID) groups based on their radiological characteristics. IHD tumors showed high attention in the arterial phase and declining attention in the portal and delayed phase,22,23 while IID tumors showed delayed enhancement in both the portal and delayed phases. The following characteristics were also present in IID tumors: rim enhancement capsular retraction and focal dilatation of the intrahepatic ducts around the tumor.24,25 Similarly, the pathological examination was reviewed by two professional pathologists separately. Patients were also divided into pathological HCC-dominant (PHD) and pathological ICC-dominant (PID) groups based on pathological performance and immunohistochemistry (IHC). The pathological features of HCC included trabecular growth pattern, bile production, and intercellular bile canaliculi, while classical gland formation or mucin production was considered to be features of ICC. Further IHC studies were conducted when there were difficulties of making pathological diagnoses. Tumor cells with Heppar1, GPC-3, CK8/18, CD10, and alpha-fetoprotein (AFP) positivity were classified to be HCC, while tumor cells with CK7, CK19, and carcinoembryonic antigen (CEA) positivity were considered to be ICC. In the present study, 50% was adopted as the practical and user-friendly cutoff value to divide patients into PHD group and PID group, which was similar with previous studies.17,25 The PHD or PID group included patients for whom the proportion of HCC or ICC cells was over 50%, respectively.

All clinical and pathological data for diagnosis were retrieved from medical records archived at Sun Yat-Sen University Cancer Center. The following clinical and pathological data were collected and analyzed: age, gender, white blood cell (WBC) count, neutrophil count, lymphocyte count, hemoglobin (HGB), platelet (PLT) count, alanine transaminase (ALT), aspartate aminotransferase (AST), albumin (ALB), total bilirubin (TBIL), indirect bilirubin (IBIL), hepatitis B surface antigen (HBsAg), serum levels of AFP, CEA, carbohydrate antigen 19-9 (CA19-9), lymph node (LN) metastasis, microvascular invasion, nerve tract invasion, macrovascular invasion, satellite nodule, and tumor diameter. Macrovascular invasion was defined as the presence of a thrombus adjacent to the tumor in the portal system or the hepatic vein system with a vague boundary confirmed by at least two imaging modalities.26 The inflammation-based prognostic scores were determined as described in Table 1. All the prognostic value of staging systems and inflammation-based prognostic scores were entered in the present study.

Patients were followed-up at least every 2 months during the first year and every 3 months thereafter. The AFP test, CEA test, CA19-9 test, liver ultrasonography, CT, and MRI were selectively performed as needed. OS was defined as the duration from the date of operation until death or the last follow-up. PFS was defined as the duration from the date of operation until the date when tumor progression was diagnosed or the last follow-up. The last follow-up was completed on October 1, 2017. The median follow-up period was 537 days for the study cohort.

SPSS version 22 software (SPSS Inc., Chicago, IL, USA) was used to analyze the data. The optimal cutoff values for NLR and PLR were determined using time-dependent receiver operating characteristic (ROC) curve analysis, which was performed using the package “survivalROC” in R version 3.2.5 (The R Foundation for Statistical Computing, Vienna, Austria. http://​www.​r-project.​org). The laboratory threshold was used as the cutoff value for other clinical data. Categorical variables were compared using the chi-square test and Fisher’s exact test. Continuous variables were compared using the two-tailed unpaired t test or Mann–Whitney U-test.

Analysis of the time-dependent ROC curves and comparisons were performed using R software with the “survival ROC” package and “survival ROC.C” package.

Over a 17-year period, between 2000 and 2017, 99 patients who were newly diagnosed with cHCC-CC and received surgical operation as the initial therapy in our hospital were included in the present study. The median age of all patients was 50 years (range 25–76 years). Most of the patients in the entire study cohort (73 patients, 73.7%) were men. All patients had a good reserve liver function with Child–Pugh A. Approximately, 80.8% of the enrolled patients were positive for HBsAg. Most of the patients were absent of LN metastasis and satellite nodules. The proportions of patients in the study cohort with microvascular invasion, perineural and lymphatic invasion, and macrovascular invasion were 74.7, 93.9, and 91.9%, respectively. The NLR and PLR scores were divided into two groups: < 2.75 and ≥ 2.75, and < 68.95 and ≥ 68.95, respectively. The baseline characteristics of patients are shown in Table 2. There were 72 (72.7%) patients in the IHD group and 27 (27.3%) patients in the IID group, whereas 61 (61.6%) patients were sorted into the PHD group and 38 (38.4%) patients were sorted into the PID group. The discrepancy between imaging and pathological examination was 35.4% (35/99).

The median OS time was 1680 days, and the 1-, 2- , and 3-year OS rates were 82.6, 66.3, and 59.6%, respectively. The correlations between the staging systems and OS are shown in Fig. 1. A higher HCC-TNM stage (P = 0.008, Fig. 1a) and BCLC stage (P = 0.048, Fig. 1b) were associated with a reduced OS, while there were no significant differences of OS stratified by ICC-TNM (P = 0.096, Fig. 1c), CLIP scores (P = 0.375, Fig. 1d), and Okuda staging (P = 0.943, Fig. 1e). Furthermore, the elevated NLR (P = 0.014, Fig. 1f) was associated with a reduced OS, while the differences of OS stratified by PLR (P = 0.149, Fig. 1g), mGPS (P = 0.694, Fig. 1h), and PNI (P = 0.214, Fig. 1i) were not significant.

In univariate analysis, age, gender, WBC, neutrophil, lymphocyte, HGB, PLT, ALT, AST, ALB, TBIL, IBIL, AFP, CEA, CA19-9, satellite nodule, microvascular invasion, nerve tract invasion, PLR, mGPS, PNI, CLIP score, and Okuda staging were not related to OS (P > 0.05). However, tumor diameter, LN metastasis, macrovascular invasion, NLR, and the HCC-TNM, ICC, and BCLC staging systems were significantly associated with OS (Table 3). These seven risk factors were entered into multivariate Cox regression analysis. After the stepwise removal of variables, HCC-TNM (hazard ratio (HR) = 1.224, 95%CI = 1.022–1.464, P = 0.028) and tumor diameter (HR = 1.122, 95%CI = 1.015–1.240, P = 0.024) remained significant predictors for OS (Table 4).

The median PFS time was 292 days. The 1-, 2-, and 3-year PFS rates were 52.2, 38.1, and 31.5%, respectively. The correlations between the staging systems and PFS are shown in Fig. 2. Univariate analysis revealed that tumor diameter, LN metastasis, microvascular invasion, macrovascular invasion, and the HCC-TNM, ICC, BCLC, CLIP, and Okuda staging systems were all associated with PFS (P < 0.05, Table 3). After the stepwise removal of variables, multivariate analysis revealed that HCC-TNM (HR = 1.247, 95%CI = 1.084–1.434, P = 0.002) and tumor diameter (HR = 1.117, 95%CI = 1.057–1.203, P = 0.003) remained significant predictors for PFS (Table 4).

The prognostic value of the staging systems and inflammation-based scores was compared by analyzing the area under ROC curves (AUC) at the 1-, 2-, and 3-year follow-up (Fig. 3). As shown in Table 5, the HCC-TNM staging system had a higher AUC value at the 1-, 2-, and 3-year follow-up for both the OS and PFS predictions compared with other staging systems or inflammation-based scores. In addition, an evaluation and comparison of the predictive power were also performed in subgroup analyses for all of the staging systems and inflammation-based scores. The results showed that the HCC-TNM staging system had good predictive power for OS and PFS with relatively high values of AUC in patients of all subgroups, including the IHD, IID, PHD, and PID groups. Moreover, a higher HCC-TNM stage was associated with a reduced OS and PFS in patients in the HCC-dominant groups (P < 0.050, Fig. 4a–g). However, in OS and PFS analyses of patients of the ICC-dominant groups, significant differences of OS stratified by the HCC-TNM stage were only observed in patients of the IID group (P < 0.001, Fig. 4b). There were no significant association between the HCC-TNM stage and PFS in patients of the IID group (P = 0.210, Fig. 4f) and PID group (P = 0.281, Fig. 4h).