Between January 2013 and November 2018, 773 patients with HCC underwent hepatectomy at the Department of Hepatopancreatobiliary Surgery, The First Affiliated Hospital of Fujian Medical University (Fuzhou, China). Patients who had previously received locoregional therapy, such as hepatectomy (n=63), transarterial chemoembolization (TACE; n=11), radiofrequency ablation (RFA; n=4), Gamma Knife^® treatment (n=1) or TACE combined with RFA (n=4), were excluded. Patients were also excluded if the tumour had been found to rupture (n=25), invade peripheral organs or metastasize (n=2) during the surgery. Patients whose data were incomplete or those who were lost to follow-up (n=32) were also excluded from this study. Ultimately, a total of 631 patients (age range, 20–85 years) were included in the current study, 25 (3.96%) of whom were allocated to the HCC with macroscopic BDTT group, and 606 (96.04%) of whom were allocated to the group without BDTT. The clinicopathological characteristics and survival rates of the two groups were compared. To decrease possible confounding and selection biases of the baseline characteristics in the two groups, 1:1 matching variables [age, sex, body mass index, presence of comorbidity, α-fetoprotein (AFP), hepatitis B virus, hepatitis C virus, indocyanine green retention rate at 15 min, Child-Pugh classification (13), maximum tumour size, number of tumours, cirrhosis, macrovascular invasion, microvascular invasion, tumour differentiation, American Joint Committee on Cancer (AJCC) stage (14) and resection margins] were applied using PSM. As a result, 50 patients were divided into two groups: The BDTT group (n=25) and the group without BDTT (n=25). The clinicopathological characteristics and survival rates of the two groups were compared once more after PSM and the prognostic risk factors of HCC patients who underwent liver resection were assessed by the Cox proportional hazards model after excluding other influencing factors. In addition, the curative effects on BDTT in patients with HCC who underwent tumour thrombectomy and bile duct resection were compared. The current study was approved by the Institutional Review Board of The First Affiliated Hospital of Fujian Medical University and performed in accordance with the ethical guidelines of the institute.

Primary tumours were evaluated by contrast-enhanced thin-slice computed tomography (CT) or magnetic resonance imaging (MRI) and MR cholangiopancreatography. Liver function and whether another comorbidity was present were also assessed. Liver function was evaluated using the Child-Pugh classification system. An attempt was made to convert Child-Pugh class B disease into class A disease via appropriate preoperative treatment. To improve liver function and treat cholangitis, it was necessary to perform biliary drainage by percutaneous transhepatic biliary drainage (PTBD) or to use an endoscopic retrograde cholangiopancreatography in certain patients. The serum total bilirubin (TBIL) level of patients with obstructive jaundice after appropriate preoperative management needed to decrease to <2.0 mg/dl or by >50% before major hepatectomy was performed (normal range, 0.1-1.0 mg/dl).

The surgical strategy was determined preoperatively at the department specialist meeting (similar to a multidisciplinary team conference). Sometimes, adjustments were necessary during surgery to select a reasonable procedure based on intraoperative exploration. The basic policy for the surgical treatment of HCC in The First Affiliated Hospital of Fujian Medical University is that if the remaining liver function is sufficient, there is an inclination to perform an anatomical resection; non-anatomical resection is recommended only when the tumour is small, the position is superficial or the remaining liver function is insufficient. Anatomical resection involved resection of the tumour with its related portal vein branches and corresponding hepatic region. The surgical treatment strategy was complete resection of the primary HCC and BDTT, trying not to remove the extrahepatic bile duct, similar to the ‘peeling off technique’ (10). The extrahepatic bile duct was resected only when the BDTT could not be completely peeled off or would have violated the bile duct wall.

After the specimens were stored and fixed in 10% formalin at room temperature for 12 h, they were embedded in paraffin and cut into 5-µm-thick sections. Hematoxylin and eosin staining (performed for 10 min at room temperature) and a light microscope (Nikon Corporation; magnification, ×100, ×200 and ×400; analyzed using QImaging MicroPublisher 3.3 with Real-Time Viewing; Teledyne QImaging) were used. The numbers, sizes (maximum tumour diameter) and locations of the tumours were recorded during a macroscopic examination of the resected specimens by a pathologist. Moreover, the microscopic examination included determination of the histological differentiation, invasion of the bile duct (the size and location of BDTT), microvasculature and macrovasculature, the specimen margin and the presence of cirrhosis. The histological differentiation of HCC was assigned according to the Edmondson-Steiner system (15). Tumour stage was classified in accordance with the criteria of the AJCC (8th edition). HCC with macroscopic BDTT represents a tumour thrombus found in the common hepatic duct or the first to second branches of the intrahepatic bile duct. Classification of HCC with macroscopic BDTT was in accordance with the criteria of the Ueda classification (3).

Specially trained researchers used outpatient records and telephone calls to follow-up with the patients. After discharge from the hospital, AFP tests, ultrasound (US) and contrast-enhanced CT or MRI were performed at least every 3 months during the follow-up period. If recurrence or metastasis was observed, follow-up treatments, such as TACE, RFA, microwave ablation, reoperation and targeted drug administration, such as Sorafenib (400 mg bid), Rivatinib (8–12 mg qd) or Regorafenib (160 mg qd), were performed, as determined by the patient's condition. Survival time was calculated from the date of surgery to the date of last contact, death or collection of survival information. The follow-up deadline was May 2019.

PSM was performed with a multiple factor logistic regression model, and a calliper of 0.10 of the standard deviation of the logit was imposed. Continuous variables are reported as the mean ± standard deviation, and categorical variables are reported as the count and percentage. Continuous variables were compared using Student's t-test or the Mann-Whitney U test as appropriate. Categorical variables were compared using the χ² test or Fisher's exact test if necessary. Patient survival rates were analysed by the Kaplan-Meier method, and survival curves were compared between groups using the log-rank test. Patients who succumbed within 30 days of surgery were excluded from the survival analysis. Moreover, patients who underwent non-curative surgical resection were excluded from the recurrence analysis. Univariate and multivariate analyses were performed using Cox proportional hazards regression analyses. The variables found to have prognostic significance in the univariate analysis were entered into the Cox multivariate proportional hazards regression analysis to identify factors that independently predicted the HCC prognosis. P<0.05 was considered to indicate a statistically significant difference. SPSS statistics software (version 24.0; IBM Corp.) was used for all statistical analyses.

The clinical features of the 25 HCC patients with macroscopic BDTT are shown in Table I. The common bile duct (CBD) was the most frequent location of BDTT in this study (15 patients; 60%), followed by the left hepatic duct (5 patients; 20%), the right hepatic duct (3 patients; 12%) and the second-order branch of the intrahepatic bile duct (2 patients; 8%). According to the Ueda classification, BDTT was classified as type 1 in 2 patients (8%), type 2 in 8 patients (32%) and type 3 in 15 patients (60%). No type 4 cases were included in this study. Among these patients, 11 (44%) were diagnosed with obstructive jaundice on admission. A subsequent biliary drainage procedure was performed using PTBD guided by US in 8 patients (32%), and 2 patients (8%) underwent PTBD of the right hepatic duct and then PTBD of the left hepatic duct again on postoperative days 19 and 27. Another 3 patients (12%) did not have biliary drainage as their TBIL increased in the short term and did not increase markedly before surgery. The median duration from biliary drainage to surgical treatment was 23 days (range, 13–42 days). A total of 17 patients (68%) had TBIL levels <2.0 mg/dl at the time of surgery, and TBIL levels decreased by 14.3±6.4 mg/dl at the time of preoperative biliary drainage. However, preoperative TBIL in the BDTT group was significantly higher than that in the without BDTT group (1.98±1.72 vs. 1.00±0.62 mg/dl; P=0.010; data not shown).

The operative procedures used in the BDTT group are shown in Table I. The primary tumours were removed by right hemihepatectomy in 11 patients (44%), including an extended right hemihepatectomy in 1 patient (4%), by left hemihepatectomy in 9 patients (36%), by central hepatectomy in 3 patients (12%) and by right posterior sectionectomy in 2 patients (8%). Caudate lobectomy was performed in 8 patients (32%). The BDTT was removed by thrombectomy either through choledochotomy or the cut end of the bile duct in 11 patients (44%), and extrahepatic bile duct resection was performed in 4 patients (16%). In the remaining 10 patients (40%), BDTTs were resectioned en bloc together with the primary tumour. Compared with those in the group without BDTT, the average operative time and intraoperative blood loss in the BDTT group were significantly greater (249.67±80.54 vs. 166.75±55.73 min, P<0.001; and 966.67±917.16 vs. 352.86±335.72 ml, P=0.012, respectively). The perioperative mortality rate was not significantly different between the two groups (4.0 vs. 1.16%; P=0.287) (data not shown).

The clinicopathological characteristics of the two groups before and after PSM are shown in Table II. The proportion of females and AFP levels in the BDTT group seemed to be higher, but these results did not reach statistical significance. Fewer patients had a Child-Pugh A classification in the BDTT group than in the group without BDTT just before surgery (P=0.002), which corresponds to the aforementioned result that the preoperative TBIL level in the BDTT group was significantly higher than that in the group without BDTT. The numbers of tumours were not significantly different between the two groups, while the tumour diameter in the BDTT group was significantly larger than that in the group without BDTT (P=0.019). In the BDTT group, 5 patients (20%) had macrovascular invasion and 15 patients (60%) had microvascular invasion, with both values being significantly higher than those in the group without BDTT (P=0.002 and P<0.001, respectively). The Edmondson-Steiner grade and AJCC stage in the BDTT group were also significantly higher than those in the group without BDTT (P=0.003 and P=0.001, respectively). However, the presence of cirrhosis was significantly lower in the BDTT group than in the group without BDTT (P=0.001).

The imbalance in clinicopathological characteristics between the two groups indicates that the data between the two groups were incomparable before PSM. To balance this difference, the clinicopathological indices were set as matching covariates for PSM. The aforementioned variables were not significantly different between these two groups after PSM (P>0.05), indicating that the two sets of data were comparable (Table II).

Although the most common location of BDTT was the CBD (60%) in the present study, BDTT rarely invaded the wall of the CBD histologically. The underlying epithelium of the resected large bile ducts was well preserved based on the histological examination of the 4 patients who underwent extrahepatic bile duct resection in this study. One of the postoperative histopathological findings was that BDTT did not usually adhere to the CBD. The cancer cells in BDTT were flaky and solid, the cells had obvious atypia and pleomorphism, and pathological mitosis was not uncommon (Fig. 1A). Although the tumour cells were scattered and haemosiderin deposition was evident, suggesting old bleeding, biliary epithelial cells were still continuous and intact, and were protected by profibrotic reactions (Fig. 1B).

Patients who died perioperatively were excluded from the survival analysis. The median follow-up duration was 32 months (range, 1–70 months). The cumulative 1-, 3- and 5-year overall survival (OS) rates in the group without BDTT (88.61, 69.01 and 51.16%, respectively) were significantly higher than those in the BDTT group (75.00, 38.67 and 17.68%, respectively) (P<0.001; Fig. 2A). The cumulative 1-, 3- and 5-year recurrence-free survival (RFS) rates in the group without BDTT (69.32, 43.01 and 29.40%, respectively) were significantly higher than those in the BDTT group (45.00, 25.00, and 6.67%, respectively) (P=0.003; Fig. 2B). However, no significant differences in the cumulative 1-, 3- and 5-year OS or RFS rates were identified between the two groups after PSM (OS: 87.62, 57.07 and 41.61%, respectively, vs. 75.00, 38.67 and 17.68%, respectively; P=0.249; Fig. 2C; RFS: 61.09, 41.01 and 30.76%, respectively, vs. 45.00, 25.00 and 6.67%, respectively; P=0.121; Fig. 2D). These results illustrate that covariates other than BDTT also affect the RFS and OS rates.

In the BDTT group, the median OS and RFS times of patients who underwent tumour thrombectomy and bile duct resection were not significantly different (OS: 36 vs. 30 months, respectively; P=0.891; Fig. 3A; RFS: 11 vs. 18 months, respectively; P=0.787; Fig. 3B). A total of 17 HCC (68.0%) patients with BDTT experienced recurrence during the follow-up; intrahepatic recurrence was the most common type [12/17 (70.59%)], followed by combined intrahepatic and extrahepatic recurrence [3/17 (17.65%)] and extrahepatic recurrence [2/17 (11.76%)], both of which were in the lungs (data not shown). None of the patients who underwent tumour thrombectomy exhibited peritoneal dissemination, and only one of these patients experienced bile duct recurrence and underwent re-resection.

Univariate and multivariate analyses were performed by Cox proportional hazards regression in 49 patients who underwent PSM. The univariate analysis revealed that macrovascular invasion (HR, 4.613; P=0.001), microvascular invasion (HR, 2.942; P=0.007), tumour differentiation (HR, 2.302; P=0.031) and resection margins (HR, 5.715; P=0.001), rather than BDTT (HR, 1.558; P=0.254), significantly affected OS (Table III). These potential risk factors were further examined by multivariate Cox proportional hazards regression analysis. The results revealed that macrovascular invasion (HR, 3.220; 95% CI, 1.184-8.756; P=0.022) was the only independent predictor of OS (Table III).