This retrospective study enrolled HCC patients with macro-BDTT or macro-PVTT who had undergone curative hemihepatectomy between September 2014 and July 2018 at four major cancer institutions: the Fujian Provincial Hospital (Fuzhou, China), West China Hospital of Sichuan University (Chengdu, China), First Affiliated Hospital of Fujian Medical University (Fuzhou, China), and Zhongshan Hospital of Xiamen University (Xiamen, China). The study protocol was approved by the Institutional Review Board of Fujian Provincial Hospital (approval number: K2017-058-04). Written informed consent was obtained from all participants or their legal guardians. Clinical and pathological data were retrospectively obtained from a prospectively maintained database.

All the patient images and specimens were separately reviewed by two experienced radiologists and two experienced pathologists in each participating hospital, and the diagnosis of HCC with macro-BDTT or macro-PVTT was confirmed by preoperative imaging and histopathological examination. Macro-BDTT was classified as B3 BDTT, indicating that the tumor thrombus invaded the initial branches of the bile duct, and B4 BDTT, indicating the invasion of the common hepatic duct (26). Macro-PVTT diagnosis indicated that the tumor thrombus was in the main trunk and the first branches of the portal vein (27). However, patients with PVTT involving the main trunk were excluded owing to a lack of recommendations for surgery (9, 10). In addition, the criteria for curative resection were as follows: the margin was free, as shown by histology, and the serum alpha-fetoprotein (AFP) level and postoperative radiographic examinations, including magnetic resonance imaging (MRI) and computed tomography (CT), showed no signs of tumor 3 months postoperatively (28–30).

The inclusion criteria were as follows (1): patients aged 18–75 years with good operative tolerance (2), HCC with macro-BDTT or macro-PVTT confirmed via imaging and histopathological analyses (3), underwent curative hemihepatectomy, and (4) no extrahepatic or distant metastases. The exclusion criteria were as follows (1): combined with hepatic vein tumor thrombus (6 patients) (2), main trunk/contralateral branch PVTT (17 patients) (3), both combined with PVTT and BDTT (26 patients) (4), non-curative resection (24 patients) (5), hepato-cholangiocarcinoma (18 patients) (6), anticancer treatment prior to the surgery (4 patients) (7), history of other cancers (1 patient), and (8) incomplete data (16 patients) ( Figure 1 ).

All individuals had undergone preoperative serological and imaging examinations. General information (such as sex, age, disease, and treatment history) was obtained from all participants. Additionally, data on other examinations including laboratory tests for liver function (serum albumin, total bilirubin, alanine aminotransferase, and aspartate aminotransferase), tests for hepatitis B virus infection and serum AFP levels, imaging examinations (abdominal ultrasonography, contrast-enhanced MRI or CT), surgical and postoperative therapy, postoperative histopathological examinations was collected. Child-Pugh classification was assessed according to the study by Pugh (31). The albumin bilirubin (ALBI) score and grading system was used for this study as outlined by Johnson (32): three grades were defined based on the following thresholds: ALBI grade 1 (G1) ≤ -2.60, ALBI grade 2 (G2) > -2.60 and ≤ -1.39, and ALBI grade 3 (G3) > -1.39.

Intraoperative ultrasonography was used routinely to determine the tumor margin and transection plane and define the relationship between the tumor and major vasculatures. After ligating the hepatic artery and portal vein individually, the ipsilateral hemihepatic and first branch of the bile duct or portal vein containing the tumor thrombus were resected integrally. For patients with PVTT, the blood inflow of the liver was occluded using Pringle’s maneuver at a site distal to the tumor thrombus. For patients with BDTT, extrahepatic bile duct resection (BDR) was performed if the tumor thrombus was poorly removed intraoperatively or extrahepatic bile duct invasion was detected.

Postoperative days 1, 3, and 5 involved liver function and routine blood tests. The postoperative complications were recorded using the Clavien–Dindo classification (33). The degree of liver cirrhosis, tumor number, tumor size, tumor differentiation, and type of tumor thrombus were determined by pathological examinations. Tumor size was defined as the maximum diameter of the largest tumor in the resected specimen. The histologic grade of tumor differentiation was assigned according to the Edmonson–Steiner classification (34).

Preventive transarterial chemoembolization (TACE) was recommended for all patients 4‒8 weeks after surgery. Follow-up was performed every 3 months for the first year and every 6 months thereafter until death. Further, the following tests were performed: serum AFP level test, liver function test, and contrast-enhanced CT or MRI. If there was confirmation of recurrence during the follow-up visits, the optimal treatment, which could include ablation, surgical resection, TACE, systemic therapy, or a combination of these approaches, was carried out in accordance with the clinical practice guidelines for HCC, multidisciplinary team’s decision for HCC management and the patient’s general condition.

The OS and recurrence-free survival (RFS) rates were the primary and secondary endpoints, respectively, of this study. OS was defined as the period from the date of hepatectomy to the date of death due to any cause or the date of the last follow-up, and RFS was defined as the period from the date of hepatectomy to the date of the first locoregional recurrence, distant metastasis, or last follow-up. All patients were followed-up until death or the conclusion of the study in September 2023.

PSM was performed to overcome deviations caused by baseline imbalance. A logistic regression model was used to calculate propensity scores. Propensity scores were calculated on the basis of clinically important variables and factors in relation to survival, including sex, age, serum albumin, alanine aminotransferase, and aspartate aminotransferase levels, hepatitis B surface antigen status, liver cirrhosis, AFP level, tumor size, and tumor number. PSM analyses were performed using the MatchIt R package (version 4.5.5) with a caliper width set equal to 0.2 standard deviations of the logit of the propensity score (35), and a 1:1 nearest-neighbor matching approach between the BDTT and PVTT groups.

Categorical variables were expressed as counts and percentages and compared using the chi-squared test or Fisher’s exact test, as appropriate. To show the degree of adjustment for PSM, we calculated standardized mean differences (SMD) for all variables before and after PSM using the stddiff R package (version 3.1) to show the size of differences in patient characteristics between two groups, where 0.8, 0.5, and 0.2 were considered large, medium, and small differences, respectively (36). The Kaplan‒Meier method was used to calculate the recurrence and survival rates, and the log-rank test was used for comparison. In the propensity-matched cohort, survival curves were compared using a stratified log-rank test by matched pairs, and we calculated hazard ratios (HR) via the Cox proportional hazards model with a robust variance estimator to account for the clustering within matched pairs. Additionally, the prognostic factors for RFS and OS before PSM were analyzed, and the variables with P values < 0.05 in univariate analyses, type of tumor thrombus, and factors significant to prognosis (such as sex, age, and ALBI grade) were further entered into the multivariate Cox proportional hazards model, using the enter method to identify independently significant factors. After PSM, non-matched factors (ALBI grade, Edmonson-Steiner grade, type of hemihepatectomy, and type of tumor thrombus) were separately analyzed by univariate Cox regression proportional hazards model to determine adjusted P values and HR. Statistical significance was set at two-tailed P values < 0.05. All statistical analyses were performed using the R software (version 4.3.1).

A total of 112 HCC patients with macro-BDTT or macro-PVTT who underwent hemihepatectomy at the 4 centers were included. Among them, 101 (90.2%) were males and 11 (9.8%) were females. Ninety-six patients (85.7%) were diagnosed with hepatitis B virus infection, and 62 were diagnosed with cirrhosis; the baseline characteristics of the total study population are presented in Table 1 . Before PSM, there were 65 patients in the BDTT group and 47 patients in the PVTT group. The BDTT group comprised 12 patients with B3 BDTT and 53 patients with B4 BDTT, of whom 5 underwent extrahepatic BDR. After PSM, 42 patient pairs were selected from each group. The PSM-BDTT group comprised 11 patients with B3 BDTT and 31 with B4 BDTT, of whom 3 underwent extrahepatic BDR. The clinical baseline characteristics between the two groups were more comparable and balanced after PSM, with SMD < 0.4 for all variables ( Table 2 ).

With a median follow-up of 31 months (range, 3‒139 months), 80 patients experienced recurrence (71.4%) and 74 died (66.1%). The median RFS and OS in this study were 10.6 months (95% confidence interval [CI], 7‒14 months) and 32 months (95% CI, 24‒40 months), respectively. The total 1-, 3-, and 5-year RFS rates were 44.1%, 32.1%, and 29.5%, respectively, and the total 1-, 3-, and 5-year OS rates were 84.8%, 45.6%, and 33.1%, respectively.

Before PSM, the 1-, 3-, and 5-year RFS rates in the BDTT group (51.8%, 40.9%, and 36.7%, respectively) were significantly higher than those in the PVTT group (37.7%, 19.9%, and 19.9%, respectively; P = 0.043, Figure 2A ). Further, the 1-, 3-, and 5-year OS rates were significantly higher in the BDTT group (92.3%, 56.3%, and 42.6%, respectively) than in the PVTT group (74.5%, 30.7%, and 19.9%, respectively; P = 0.008, Figure 2B ). There were 38 deaths (38/65, 58.5%) in the BDTT group and 36 (36/47, 76.6%) in the PVTT group.

After PSM, 42 patient pairs were selected. Better RFS and OS were observed in the PSM-BDTT group (P = 0.037 and P = 0.004, respectively). The 1-, 3-, and 5-year RFS rates were 52.4%, 42.9%, and 42.9%, respectively, in the PSM-BDTT group and 33.3%, 23.8%, and 23.8%, respectively, in the PSM-PVTT group. The 1-, 3-, and 5-year OS rates were 92.9%, 59.5%, and 52.1%, respectively, in the PSM-BDTT group and 73.8%, 33.3%, and 20.2%, respectively, in the PSM-PVTT group ( Figures 2C, D ).

Before PSM, 43 patients (43/65, 66.2%) in the BDTT group and 37 (37/47, 78.7%) in the PVTT group suffered tumor recurrence (P = 0.146). After PSM, 24 patients in the PSM-BDTT group developed tumor recurrence, including 16 with intrahepatic recurrence alone and 8 with extrahepatic recurrence. Thirty-two patients in the PSM-PVTT group developed tumor recurrence, including 28 with intrahepatic recurrence alone and 4 with extrahepatic recurrence. Compared with the PSM-BDTT group, patients in the PSM-PVTT group had a higher likelihood of developing intrahepatic recurrence alone (P = 0.032).

Before PSM, univariate analysis indicated that PVTT and large HCC were risk factors for poor RFS (HR = 1.577; 95% CI, 1.013–2.454, P = 0.043 and HR = 1.668; 95% CI, 1.031–2.697, P =0.037, respectively) and OS (HR = 1.859; 95% CI, 1.174–2.943, P = 0.008 and HR = 1.677; 95% CI, 1.020–2.756, P = 0.042, respectively) ( Table 3 ). Multivariate analysis indicated that PVTT (HR = 1.835; 95% CI, 1.121–3.005, P = 0.016) was a risk factor for poor OS, whereas large HCC (HR = 1.553; 95% CI, 1.050–2.536; P = 0.039) was a risk factor for poor RFS ( Table 4 ).

After PSM, univariate analysis revealed that PVTT represented a risk factor for both poor RFS (HR = 1.749; 95% CI, 1.034–2.958; P = 0.037) and poor OS (HR = 2.245; 95% CI, 1.286–3.921; P = 0.004) ( Supplementary Table 1 ).