Survival Benefit of Combined Systemic and Locoregional Therapy in Hepatocellular Carcinoma with Portal Vein Tumor Thrombus: A Propensity Score-Matched Analysis

Abstract

Introduction

The prognosis of hepatocellular carcinoma (HCC) with portal vein tumor thrombus (PVTT) remains dismal. Although systemic therapy is the standard of care, its effectiveness is limited. This study aimed to compare the efficacy and safety of transarterial chemoembolization (TACE) or hepatic arterial infusion chemotherapy (HAIC) combined with systemic therapy versus systemic therapy alone in patients with HCC and PVTT.

Methods

We retrospectively analyzed 478 patients newly diagnosed with HCC and PVTT between January 2021 and December 2024. Propensity score matching (PSM) was used to balance baseline characteristics. Outcomes compared between the combination therapy group (TACE/HAIC plus targeted therapy and immunotherapy, n = 374) and the systemic therapy group (n = 104) included overall survival (OS), progression-free survival (PFS), tumor response, and adverse events.

Results

After PSM (184 vs. 102 patients), the combination therapy group showed significantly longer median OS (15.7 vs. 5.9 months; hazard ratio [HR] = 0.524, 95% confidence interval [CI]: 0.391–0.702; p < 0.001) and PFS (7.0 vs. 3.6 months, HR = 0.732, 95% CI: 0.558–0.959; p = 0.024). The disease control rate was also higher in the combination therapy group (43.5% vs. 27.5%, p = 0.007). Subgroup analyses revealed pronounced survival benefits in patients with Vp4 PVTT and those with Child-Pugh B liver function. Although adverse events were more frequent in the combination therapy group, the incidence of grade 3–4 toxicities was generally comparable between the two groups.

Conclusion

In HCC patients with PVTT, combining TACE or HAIC with systemic therapy significantly improves survival outcomes compared to systemic therapy alone, with acceptable safety. This multimodal approach offers a promising treatment strategy, particularly for patients with advanced PVTT or impaired liver function.

Plain Language Summary

This study looked at patients with a serious type of liver cancer called hepatocellular carcinoma (HCC), which had also spread into the main blood vessel of the liver (called portal vein tumor thrombus, or PVTT). The goal was to see if adding a local treatment directed at the liver tumor to the standard medication therapy could help patients live longer. We studied 478 patients; 374 received the combination of local and medication therapy, while 104 received medication therapy alone. To ensure a fair comparison, we used a statistical method to match patients in both groups with similar characteristics. We found that patients who received both the local treatment and medication lived significantly longer. Their median overall survival was 15.7 months, compared to 5.9 months for those on medication alone. The time before their cancer worsened was also longer (7.0 months vs. 3.6 months). Patients with more advanced disease in the main portal vein (called Vp4) and those with moderately reduced liver function (Child-Pugh B) saw the greatest benefit from the combined approach. Although side effects like low blood counts and liver enzyme changes were more common with the combined treatment, they were generally manageable. This study suggests that for patients with this advanced form of liver cancer, combining a local therapy with standard medications is a more effective strategy than using medications alone.

Introduction

Hepatocellular carcinoma (HCC) represents a significant global health burden, with high incidence and mortality rates, particularly in Asia [1]. The prognosis for patients with advanced HCC is further worsened by the development of portal vein tumor thrombus (PVTT), which is associated with limited treatment options, rapid disease progression, and poor survival outcomes [2]. The presence of PVTT, characterized by tumor invasion into the portal venous system, not only signifies advanced disease but also often precludes patients from curative surgical interventions and is associated with a higher risk of intrahepatic spread and liver failure, underscoring the critical need for effective non-surgical strategies in this distinct subpopulation [3].

According to current international guidelines, systemic therapy, primarily consisting of anti-angiogenic agents combined with immune checkpoint inhibitors (ICIs), is the standard first-line treatment for advanced HCC (Barcelona Clinic Liver Cancer [BCLC] stage C) [4]. Recent studies have shown that combining transarterial chemoembolization (TACE) with anti-angiogenic agents and ICIs significantly improves tumor response and increases surgical conversion rates in patients with unresectable intermediate-to-advanced HCC [5], particularly in those with high tumor burden or disease beyond the Milan criteria [6]. Mechanistic studies suggest that this combination enhances the tumor immune microenvironment, promoting increased T-cell infiltration and the formation of tertiary lymphoid structures [7]. However, adjuvant TACE has not demonstrated a survival benefit in early-stage HCC following curative resection [8], underscoring the importance of appropriate patient selection. These findings support the integration of locoregional and systemic therapies in advanced HCC.

In this context, both TACE and hepatic arterial infusion chemotherapy (HAIC) represent cornerstone locoregional approaches that can be rationally combined with systemic therapy. Despite their technical differences − TACE aims to embolize tumor-feeding arteries while delivering high-dose chemotherapy, whereas HAIC provides sustained, high-concentration hepatic arterial chemotherapy − they share the fundamental therapeutic goal of achieving potent local tumor control within the liver. This is particularly relevant for PVTT as both modalities can directly target the primary liver tumor and, in some cases, the thrombus itself, potentially complementing the systemic effects of immunotherapy and anti-angiogenic drugs.

Nonetheless, the current evidence base is limited by heterogeneous study designs, small sample sizes, and a lack of focused comparative analyses targeting the PVTT subpopulation. Therefore, it remains unclear whether the addition of locoregional therapy to systemic treatment provides a survival advantage over systemic therapy alone in this high-risk group.

To address this gap, the present study evaluated the efficacy and safety of the combination of TACE or HAIC and systemic therapy compared with systemic therapy alone in patients with HCC and PVTT. By utilizing a retrospective cohort design supported by propensity score matching (PSM) to minimize selection bias and enhance comparability, we assessed key outcomes, including overall survival (OS), progression-free survival (PFS), objective response rate (ORR), disease control rate (DCR), conversion-to-resection rate, and treatment-related adverse events. This study aims to contribute higher-quality evidence to inform clinical decision-making in this challenging patient population.

Methods

Study Design

This retrospective study analyzed clinical records of patients diagnosed with HCC and PVTT who received treatment at the First Affiliated Hospital, Zhejiang University School of Medicine, from January 2021 to December 2024. The study protocol was approved by the hospital’s Ethics Committee (Approval No. IIT20250396B-R1), and the study was conducted in compliance with the ethical principles of the Declaration of Helsinki (1975, revised in 1983).

HCC was diagnosed based on radiological criteria established by the European Association for the Study of the Liver (EASL) and the American Association for the Study of Liver Diseases (AASLD). The severity of PVTT was classified using the Liver Cancer Study Group of Japan (LCSGJ) system: Vp1 denotes thrombus confined to segmental or tertiary portal vein branches; Vp2 indicates involvement of a second-order portal vein branch; Vp3 refers to thrombus in a first-order branch; and Vp4 represents involvement of the main portal trunk and/or contralateral/bilateral first-order branches. PVTT diagnosis and classification were determined through cross-sectional imaging (contrast-enhanced computed tomography or magnetic resonance imaging) and confirmed by two independent senior abdominal radiologists with over 5 years of experience, who were blinded to the clinical outcomes. In the case of disagreement, a third expert radiologist made the final determination. All PVTT grading assessments were completed prior to PSM. While patients across all LCSGJ grades (Vp1–Vp4) were initially considered, those with Vp1 PVTT represented a relatively small proportion (<0.5%) of the final analytical cohort, reflecting the typical distribution of more advanced PVTT in patients referred to our tertiary care center for combination therapy consideration. Tumor burden score (TBS), a continuous quantitative metric integrating both tumor size and number, was calculated using the following formula:

$${\text{TBS}}^2={\left(\text{maximum}\text{tumor}\text{diameter}\right)}^2+{\left(\text{number}\text{of}\text{tumors}\right)}^2.$$

This calculation was based on imaging assessments. TBS values were stratified into three categories based on previously validated cutoff values: low (TBS ≤3.83), moderate (3.83 < TBS ≤ 8.54), and high (TBS >8.54). These categories were used to stratify patients for subgroup analyses. Liver function was also assessed using the albumin-bilirubin (ALBI) grade. The ALBI score was calculated as follows: (log₁₀ bilirubin [µmol/L] × 0.66) + (albumin [g/L] × −0.085). Based on the ALBI score, patients were classified into three grades: grade 1 (score ≤−2.60), grade 2 (score >−2.60 and ≤−1.39), and grade 3 (score >−1.39). For regression analyses, the ALBI grade was dichotomized as grade 1 versus grade 2/3.

Inclusion criteria were as follows: (a) newly diagnosed HCC with PVTT and no prior cancer treatment (including surgery, ablation, TACE/HAIC, radiotherapy, or systemic therapy); (b) initial treatment consisting of either systemic therapy alone (targeted therapy combined with immunotherapy, given concurrently or sequentially) or combination therapy (TACE/HAIC plus systemic treatment); (c) age ≥ 18 years; (d) Child-Pugh A or B liver function; (e) Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1; and (f) adequate hematologic function, defined as white blood cell count >3 × 10⁹/L, absolute neutrophil count >1.5 × 10⁹/L, platelet count >100 × 10⁹/L, and hemoglobin >85 g/L. Exclusion criteria were as follows: (a) severe comorbid cardiac, pulmonary, or renal diseases (defined in online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000551935); (b) previous diagnosis of a second primary malignancy; (c) initial management with surgical resection; (d) liver transplantation during the study period; (e) monotherapy with either targeted therapy or immunotherapy; (f) receipt of radiotherapy at any time during treatment; (g) incomplete clinical data; and (h) loss to follow-up.

Treatment Protocol

All patients included in the study received both targeted therapy and immunotherapy. The following targeted agents were administered: bevacizumab (7.5–15 mg/kg intravenously every 3 weeks), lenvatinib (12 mg/day orally for bodyweight ≥60 kg or 8 mg/day for bodyweight <60 kg), sorafenib (400 mg orally twice daily), regorafenib (80–160 mg orally once daily for the first 21 days of each 28-day cycle), donafenib (0.2 g orally twice daily), and apatinib (250–500 mg orally once daily). Immunotherapy involved programmed cell death protein 1 (PD-1) or programmed death-ligand 1 (PD-L1) inhibitors, including sintilimab (200 mg every 3 weeks), atezolizumab (1,200 mg every 3 weeks), durvalumab (1,500 mg every 4 weeks), envafolimab (150 mg subcutaneously every week or 450 mg subcutaneously every 3 weeks), camrelizumab (200 mg every 3 weeks), pembrolizumab (200 mg every 3 weeks), toripalimab (240 mg every 3 weeks), and tislelizumab (200 mg every 3 weeks) (online suppl. Table 2). The selection of specific agents was guided by factors such as drug cost, patient assistance programs, regional insurance reimbursement policies, and individual patient preferences.

Patients in the combination therapy group received at least one session of locoregional therapy (TACE or HAIC) in addition to systemic therapy. The choice between TACE and HAIC was determined through multidisciplinary team discussion based on established institutional guidelines, primarily considering tumor burden characteristics and portal vein patency. TACE was generally preferred for patients with: (1) predominantly nodular or massive tumor morphology; (2) preserved portal vein flow (LCSGJ Vp1–Vp2); and (3) adequate liver function reserve. In contrast, HAIC was typically selected for patients with: (1) infiltrative or diffuse tumor growth patterns; (2) more extensive portal vein invasion (LCSGJ Vp3–Vp4); and (3) compromised portal venous flow, where HAIC’s lower embolic effect was considered advantageous. Patients were strictly limited to receiving either TACE or HAIC throughout their treatment course, with no crossover or sequential application of both modalities permitted in this study cohort. TACE was performed as previously described. In brief, a chemoembolic emulsion composed of 30 mg/m² epirubicin, 200 mg/m² carboplatin, and 4 mg/m² mitomycin C was mixed with 2–5 mL lipiodol and infused into the tumor-feeding artery. Additional lipiodol was incrementally administered until stasis of blood flow was achieved, with a maximum total volume of 20 mL. Repeat TACE procedures were scheduled at 3- to 4-week intervals, based on tumor response and liver function tolerance.

Alternatively, selected patients received HAIC according to the institutional protocol of the Department of Hepatobiliary and Pancreatic Surgery at the First Affiliated Hospital, Zhejiang University School of Medicine. A microcatheter was selectively positioned in the tumor-feeding artery, followed by continuous infusion of an FOLFOX-based regimen comprising the following: 85 mg/m² oxaliplatin infused over 2 h; 400 mg/m² leucovorin infused over 2 h; a 400 mg/m² bolus of 5-fluorouracil; and finally, a continuous infusion of 2,400 mg/m² 5-fluorouracil over 46 h. Treatment cycles were repeated every 3–4 weeks, depending on treatment response and patient tolerance.

Salvage liver resection following tumor downstaging was determined through multidisciplinary evaluation involving hepatobiliary surgeons, diagnostic and interventional radiologists, anesthesiologists, and other relevant specialists. All surgical procedures were performed by attending hepatobiliary surgeons with more than 15 years of experience in liver resection. The choice of surgical approach (robot-assisted, laparoscopic, or open radical hepatectomy) was determined based on tumor location, size, vascular involvement, and anatomical feasibility. The management of PVTT was determined by the extent and location of the thrombus. En bloc resection was performed when PVTT was confined to the resection margin. For PVTT extending into the main portal vein beyond the intended transection plane, the thrombus was carefully extracted through the opened portal vein stump under direct visualization and controlled surgical conditions.

Response Assessment and Follow-Up

Baseline evaluations, including serum alpha-fetoprotein (AFP), liver function tests, complete blood count, coagulation profile, and contrast-enhanced CT or MRI of the abdomen, were completed within 4 weeks before treatment initiation. After treatment commencement, follow-up assessments consisting of the same laboratory tests and cross-sectional imaging studies were performed every 8–12 weeks or earlier if clinically indicated. To ensure consistency in response evaluation, the same imaging modality (either CT or MRI) used at baseline was maintained throughout the follow-up period for each patient, unless contraindicated due to safety concerns (e.g., renal impairment precluding contrast-enhanced CT) or technical limitations. All images were independently reviewed by two attending physicians, and any discrepancies were resolved through consensus with a senior consultant. Tumor response was assessed using the modified Response Evaluation Criteria in Solid Tumors (mRECIST) for HCC as the primary endpoint [9]. A sensitivity analysis using RECIST v1.1 was also conducted. Responses were classified as complete response, partial response, stable disease (SD), or progressive disease. The ORR was defined as the proportion of patients achieving CR or PR. The DCR was calculated as the sum of patients achieving complete response, partial response, or SD. While the standard follow-up interval was 8–12 weeks, earlier imaging was performed in cases of clinical deterioration or ambiguous laboratory findings to minimize misclassification of disease status, particularly for distinguishing between SD and early progression.

For patients who subsequently underwent surgical resection, follow-up was conducted at 1 month postoperation, every 3 months during the first 2 years, and every 6 months thereafter. Adverse events were recorded at each visit and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE), version 5.0. Survival data were obtained through medical record review and telephonic follow-up. The data cutoff for the final analysis was May 4, 2025.

Statistical Analysis

To reduce potential confounding, a PSM was performed. Propensity scores were estimated using a logistic regression model that included clinically relevant covariates: age, gender, etiology, AFP level, Child-Pugh grade, ECOG performance status, TBS, and PVTT classification. A 1:2 nearest neighbor matching algorithm was employed, with a caliper width set to 0.2 of the standard deviation of the logit of the propensity score to prevent poor matches. Matching was performed without replacement to ensure the statistical independence of matched pairs. The balance of baseline covariates before and after matching was assessed using the standardized mean difference (SMD), an effect size measure independent of sample size. For continuous variables, SMD was calculated as the difference in group mean values divided by the pooled standard deviation; for dichotomous variables, SMD was derived based on the difference in proportions. Covariate balance was considered excellent if the post-matching SMD was <0.1, while an SMD between 0.1 and 0.2 was deemed acceptable, representing a negligible difference unlikely to introduce confounding effects. Table 1 shows the efficacy of the PSM procedure. Prior to analysis, the extent of missing data was evaluated. Key variables included age, gender, etiology, Child-Pugh grade, ECOG status, PVTT classification, and availability of complete data (0% missing). For continuous laboratory values (e.g., AFP), missing data accounted for less than 5% of cases. Given the minimal and random nature of the missing data, complete case analysis was employed without imputation.

Table 1.

Baseline characteristics of HCC patients with portal vein tumor thrombus

Variables	Before matching		p value	SMD	After matching		p value	SMD
	systemic therapy (n = 104)	combination therapy (n = 374)			systemic therapy (n = 102)	combination therapy (n = 184)
Age, years	56.4±10.3	56.7±10.9	0.640	0.028	56.3±10.4	56.4±11.1	0.437	0.009
Gender, n (%)	​	​	0.669	0.045	​	​	0.682	0.052
Male	94 (90.4)	343 (91.7)	​		93 (91.2)	165 (89.7)	​
Female	10 (9.6)	31 (8.3)	​		9 (8.8)	19 (10.3)	​
ECOG, n (%)	​	​	0.157	0.157	​	​	0.767	0.035
0	37 (35.6)	162 (43.3)	​		37 (36.3)	70 (38.0)	​
1	67 (64.4)	212 (56.7)	​		65 (63.7)	114 (62.0)	​
HBV/HCV infection, n (%)		​	0.729	0.038	​	​	0.780	0.036
Present	67 (64.4)	234 (62.6)	​		66 (64.7)	116 (63.0)	​
Absent	37 (35.6)	140 (37.4)	​		36 (35.3)	68 (37.0)	​
TNM, n (%)	​	​	0.070	0.200	​	​	0.685	0.050
IIIB and IVA	54 (51.9)	231 (61.8)	​		54 (52.9)	102 (55.4)	​
IVB	50 (48.1)	143 (38.2)	​		48 (47.1)	82 (44.6)	​
TBS, n (%)	​	​	0.474	0.081	​	​	0.889	0.019
Low and Moderate	44 (42.3)	173 (46.3)	​		43 (42.2)	76 (41.3)	​
High	60 (57.7)	201 (53.7)	​		59 (57.8)	108 (58.7)	​
Tumor distribution, n (%)		​	0.291	0.117	​	​	0.778	0.035
Unilobar	58 (55.8)	230 (61.5)	​		57 (55.9)	106 (57.6)	​
Bilobar	46 (44.2)	144 (38.5)	​		45 (44.1)	78 (42.4)	​
AFP, n (%)	​	​	0.645	0.049	​	​	0.820	0.026
Abnormal	23 (22.1)	75 (20.1)	​		21 (20.6)	40 (21.7)	​
Normal	81 (77.9)	299 (79.9)	​		81 (79.4)	144 (78.3)	​
Child-Pugh, n (%)	​	​	<0.001	0.524	​	​	0.362	0.111
A	58 (55.8)	294 (78.6)	​		57 (55.9)	113 (61.4)	​
B	46 (44.2)	80 (21.4)	​		45 (44.1)	71 (38.6)	​
PVTT, n (%)	​	​	0.096	0.200	​	​	0.682	0.052
Vp1 and Vp2	9 (8.7)	56 (15.0)	​		9 (8.8)	19 (10.3)	​
Vp3 and Vp4	95 (91.3)	318 (85.0)	​		93 (91.2)	165 (89.7)	​
Targeted therapy, n (%)		​	<0.001	0.452	​	​	0.250	0.143
Bevacizumab	53 (51.0)	110 (29.4)	​		51 (50.0)	79 (42.9)	​
TKI	51 (49.0)	264 (70.6)	​		51 (50.0)	105 (57.1)	​

Values are presented as numbers (%).

p values were calculated using a two-sided χ² test.

SMDs were calculated as described in the Statistical Analysis section. For dichotomous variables, SMD is reported for the first listed category (e.g., male, ECOG 0, Child-Pugh A); normal AFP was defined as ≤20 ng/mL.

HCC, hepatocellular carcinoma; SMD, standardized mean difference; ECOG, Eastern Cooperative Oncology Group; HBV, hepatitis B virus; HCV, hepatitis C virus; TNM, tumor, node, and metastasis; TBS, tumor burden score; AFP, alpha-fetoprotein; PVTT, portal vein tumor thrombus; Vp, vena porta; TKI, tyrosine kinase inhibitor.

Continuous variables were analyzed using an independent or paired t test, as appropriate, while categorical variables were compared using Pearson’s chi-square test. Intergroup differences in clinical parameters were evaluated using either Mann-Whitney U test or chi-square test. PFS was defined as the time from the initiation of systemic therapy to the first radiographic evidence of disease progression (according to mRECIST) or death from any cause. Patients who underwent curative intent (R0) resection before progression were censored on the date of surgery. Those lost to follow-up or alive without progression at the data cutoff were censored at the date of their last tumor assessment. OS was defined as the time from the start of treatment to death from any cause. Kaplan-Meier curves were generated to illustrate PFS and OS, and comparisons between groups were performed using the log-rank test.

Variables for multivariate Cox regression were selected through a two-step process. First, all clinically relevant covariates (age, gender, etiology, AFP level, Child-Pugh grade, ECOG status, tumor burden, PVTT classification, and treatment modality) were included in univariate analysis. Variables demonstrating association with outcomes in univariate analysis (p < 0.05) were subsequently entered into the multivariate model. The proportional hazards assumption was formally tested using Schoenfeld residuals, with no significant violations detected (global test p > 0.05). In cases where time-varying effects were suggested, we employed stratified analysis or introduced time-dependent covariates to address potential non-proportionality.

All statistical tests were two sided, and a p value of <0.05 was considered statistically significant. All statistical analyses were conducted using SPSS software (version 27.0) and R software (version 4.2.1). PSM was performed using SPSS, while survival analyses were conducted in R using the survival package (version 3.3.1) to test the proportional hazards assumption and fit survival regression models. Visualizations were generated using the survminer package (version 0.4.9) and ggplot2 (version 3.4.4).

Results

Patient Characteristics

A total of 478 patients were included in this study. Among them, 104 patients received systemic therapy alone (targeted agents and immunotherapy), while 374 patients underwent combination therapy comprising systemic therapy in addition to TACE or HAIC. The median follow-up duration was 7.9 months in the systemic therapy group and 16.2 months in the combination therapy group. The patient selection process is illustrated in Figure 1. Significant differences were observed between the groups in terms of Child-Pugh class and the types of targeted agents used. After PSM, 102 and 184 patients remained in the systemic therapy and combination therapy groups, respectively. Following matching, baseline characteristics, including TNM stage, tumor burden score (TBS), tumor distribution, and AFP levels, were well balanced (all p > 0.2; Table 1). Before matching, patients in the combination therapy group received a total of 794 cycles of TACE/HAIC, with a median of 1 cycle per patient (interquartile range: 1–2; range: 1–9 cycles). The treatment strategy was primarily based on a “treat-and-assess” approach, where subsequent cycles were contingent upon tumor response and liver function tolerance. The median of 1 cycle observed in the overall cohort reflects that a substantial proportion of patients did not proceed to additional locoregional treatments, primarily due to disease progression (20.7%) or treatment-related adverse events (11.9%) after the initial session, rather than a pre-planned single-session therapeutic strategy. In contrast, patients who exhibited objective response or SD typically received multiple cycles as scheduled.

Fig. 1.

Flow diagram of HCC patients with PVTT who underwent combination therapy or systemic therapy. HCC, hepatocellular carcinoma; PVTT, portal vein tumor thrombus; ECOG, Eastern Cooperative Oncology Group; TACE, transcatheter arterial chemoembolization; HAIC, hepatic artery infusion chemotherapy.

Tumor Responses

Table 2 summarizes the tumor response outcomes before and after PSM. Based on mRECIST criteria, the combination therapy group demonstrated significantly higher rates of SD (27.5% vs. 14.4%; p = 0.006) and DCR (47.1% vs. 26.9%; p < 0.001) compared to the systemic therapy group. These benefits remained consistent following PSM, with higher SD (26.1% vs. 14.7%; p = 0.026) and DCR (43.5% vs. 27.5%; p = 0.007) observed in the combination therapy group (shown in Table 2; Fig. 2). After matching, 24 patients (13.0%) in the combination therapy group underwent salvage liver resection following successful tumor downstaging, compared to 10 patients (9.8%) in the systemic therapy group (shown in Fig. 3).

Table 2.

Treatment efficacy was evaluated by mRECIST criteria before and after PSM

​	Before matching		p value	OR	After matching		p value	OR
	systemic therapy (n = 104)	combination therapy (n = 374)			systemic therapy (n = 102)	combination therapy (n = 184)
CR, n (%)	0 (0)	1 (0.3)	–	–	0 (0)	0 (0)	–	–
PR, n (%)	13 (12.5)	72 (19.3)	0.148	0.60	13 (12.7)	32 (17.4)	0.179	0.69
SD, n (%)	15 (14.4)	103 (27.5)	0.006	0.44	15 (14.7)	48 (26.1)	0.026	0.49
PD, n (%)	76 (73.1)	198 (52.9)	<0.001	2.41	74 (72.5)	104 (56.5)	0.007	2.03
Objective response, n (%)	13 (12.5)	73 (19.5)	0.133	0.59	13 (12.7)	32 (17.4)	0.179	0.69
Disease control, n (%)	28 (26.9)	176 (47.1)	<0.001	0.41	28 (27.5)	80 (43.5)	0.007	0.49

Summary of best response.

Values are presented as n (%).

p values were calculated using a two-sided χ² test.

mRECIST, modified response evaluation criteria in solid tumors; PSM, propensity score matching; OR, odds ratio.

Fig. 2.

Bar chart depicting the response of HCC using the mRECIST criteria in the combination therapy and systemic therapy groups after PSM. Ns, not significant; PR, partial response; SD, stable disease; PD, progressive disease; HCC, hepatocellular carcinoma.

Fig. 3.

Bar chart depicting conversion resection rate for HCC with PVTT in the combination therapy and systemic therapy groups after PSM. Ns, not significant; HCC, hepatocellular carcinoma; PVTT, portal vein tumor thrombus; PSM, propensity score matching.

Survival Outcomes

Based on pre-matching long-term mortality data, 212 of 374 patients (56.7%) in the combination therapy group and 81 of 104 (77.9%) in the systemic therapy group died during the follow-up period (shown in Fig. 4a, b). After matching, mortality remained lower in the combination therapy group (105 of 184, 57.1%) compared to that in the systemic therapy group (79 of 102, 77.5%) (shown in Fig. 4c, d). Univariate and multivariate analyses for PFS and OS are detailed in the online supplementary material (Tables 3–6). Multivariate Cox regression identified combination therapy as an independent prognostic factor for both OS and PFS.

Fig. 4.

Kaplan-Meier survival curves comparing OS and PFS among patients who underwent combination therapy vs. systemic therapy before (a, b) and after (c, d) PSM. As log-rank and Cox regression p values were nearly identical, only the Cox regression p values are displayed for clarity. OS, overall survival; PSM, propensity score matching; HR, hazard ratio; PFS, progression-free survival.

Table 3.

Univariate Cox regression analyses of the prognostic factors for OS

Variables	Group	n	OS univariate		p value
			HR	95% CI
Age	>60	148	0.873	0.679–1.123	0.290
	≤60	330
Gender	Male	437	0.749	0.519–1.082	0.124
	Female	41
ECOG	0	199	0.850	0.671–1.076	0.176
	1	279
HBV/HCV infection	Present	301	1.091	0.861–1.382	0.471
	Absent	177
Imaging cirrhosis	Present	381	0.802	0.610–1.054	0.113
	Absent	97
Child-Pugh	A	352	0.546	0.426–0.700	<0.001
	B	126
ALBI	Grade 1	206	0.701	0.554–0.887	0.003
	Grade 2/3	272
PLT	Normal	379	0.987	0.745–1.308	0.929
	Abnormal	99
AFP	≤20	98	0.958	0.723–1.268	0.763
	>20	380
Tumor involvement	Hemiliver	288	0.736	0.583–0.928	0.010
	Both hemiliver	190
TBS	Low and moderate	216	0.855	0.678–1.078	0.185
	High	262
PVTT	Vp1-2	65	0.710	0.491–1.024	0.067
	Vp3-4	413
Distant metastasis	Present	193	1.329	1.054–1.675	0.016
	Absent	285
Treatment	Combination therapy	374	0.474	0.366–0.612	<0.001
	Systemic therapy	104
Targeted therapy	Bevacizumab	163	1.041	0.809–1.340	0.754
	TKIs	315
Immunotherapy	PD-1	413	1.290	0.897–1.853	0.169
	PD-L1	65

OS, overall survival; HR, hazard ratio; CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; HBV, hepatitis B virus; HCV, hepatitis C virus; Child-Pugh, Child-Pugh score; ALBI, albumin-bilirubin index; PLT, platelets; AFP, alpha-fetoprotein; TBS, tumor burden score; PVTT, portal vein tumor thrombus; TKIs, tyrosine kinase inhibitors; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1.

Table 6.

Multivariate Cox regression analyses of the prognostic factors for PFS

Variables	PFS multivariate		p value
	HR	95% CI
Child-Pugh (A, B)	0.731	0.579–0.924	0.009
Tumor involvement (hemiliver, both hemiliver)	0.774	0.625–0.958	0.019
Distant metastasis (present, absent)	1.228	0.996–1.515	0.055
Treatment (combination therapy, systemic therapy)	0.677	0.531–0.864	0.002

PFS, progression-free survival; HR, hazard ratio; CI, confidence interval; Child-Pugh, Child-Turcotte-Pugh score.

Table 4.

Multivariate Cox regression analyses of the prognostic factors for OS

Variables	OS multivariate		p value
	HR	95% CI
Child-Pugh (A, B)	0.633	0.474–0.846	0.002
ALBI (grade 1, grade 2/3)	0.923	0.701–1.215	0.567
Tumor involvement (hemiliver, both hemiliver)	0.748	0.589–0.950	0.017
Distant metastasis (present, absent)	1.199	0.947–1.518	0.131
Treatment (combination therapy, systemic therapy)	0.495	0.381–0.644	<0.001

OS, overall survival; HR, hazard ratio; CI, confidence interval; Child-Pugh, Child-Turcotte-Pugh score; ALBI, albumin-bilirubin index.

The 1-, 2-, and 3-year OS rates in the combination therapy group were 52.1%, 18.4%, and 5.3%, respectively, compared to 30.8%, 13.5%, and 3.8%, respectively, in the systemic therapy group (all p < 0.001). The median OS was significantly longer with combination therapy (17.2 months) than with systemic therapy alone (5.8 months; hazard ratio [HR] = 0.472, 95% confidence interval [CI] = 0.365–0.611; p < 0.001; shown in Fig. 4a). Similarly, the median PFS was prolonged in the combination therapy group (7.6 months) compared to that in the systemic therapy group (3.6 months). The 1- and 2-year PFS rates were 32.4% and 9.9%, respectively, in the combination therapy group, versus 23.1% and 7.7%, respectively, in the systemic therapy group (HR = 0.656; 95% CI = 0.516–0.835, p < 0.001; shown in Fig. 4b).

PSM subgroup analyses were conducted to assess the differential efficacy of combination therapy versus systemic therapy alone across clinically relevant patient strata. Among patients with low and intermediate tumor burden (n = 119), combination therapy significantly improved both median OS (15.7 vs. 5.0 months; HR = 0.452, 95% CI: 0.288–0.710; p < 0.001; shown in Fig. 5a) and PFS (6.7 vs. 3.5 months; HR = 0.631, 95% CI: 0.419–0.950; p = 0.027; shown in Fig. 5b). In the high tumor burden subgroup (n = 167), combination therapy maintained a significant OS advantage (15.9 vs. 6.9 months; HR = 0.577, 95% CI: 0.392–0.849; p < 0.001; shown in Fig. 5c), although PFS did not differ significantly between the groups (7.0 vs. 3.7 months; HR = 0.796, 95% CI: 0.553–1.146; p = 0.220; shown in Fig. 5d).

Fig. 5.

Kaplan-Meier survival curves comparing OS and PFS among different TBS subgroups of patients after PSM (low and moderate (a, b), high (c, d)). As log-rank and Cox regression p values were nearly identical, only the Cox regression p values are displayed for clarity. OS, overall survival; PSM, propensity score matching; HR, hazard ratio; PFS, progression-free survival; TBS, tumor burden score.

Stratification by PVTT extent revealed that patients with Vp4 disease (n = 138) experienced substantial benefit from combination therapy, with longer median OS (13.5 vs. 4.0 months; HR = 0.410, 95% CI: 0.275–0.612; p < 0.001; shown in Fig. 6c) and PFS (6.75 vs. 2.9 months; HR = 0.624, 95% CI: 0.432–0.903; p = 0.012; shown in Fig. 6d). Conversely, in patients with Vp1–3 PVTT (n = 148), no statistically significant differences were observed in OS (16.4 vs. 13.2 months; HR = 0.772, 95% CI: 0.491–1.215; p = 0.264; shown in Fig. 6a) or PFS (7.3 vs. 4.9 months; HR = 0.952, 95% CI: 0.626–1.447; p = 0.816; shown in Fig. 6b).

Fig. 6.

Kaplan-Meier survival curves comparing OS and PFS among different PVTT subgroups of patients after PSM (Vp1-3 (a, b), Vp4 (c, d)). As log-rank and Cox regression p values were nearly identical, only the Cox regression p values are displayed for clarity. OS, overall survival; PSM, propensity score matching; HR, hazard ratio; PFS, progression-free survival; PVTT, portal vein tumor thrombus.

Liver function, as assessed by Child-Pugh classification, also influenced treatment efficacy. Among Child-Pugh B patients (n = 116), combination therapy was associated with significantly longer OS (11.0 vs. 2.9 months; HR = 0.378, 95% CI: 0.246–0.581; p < 0.001; shown in Fig. 7c) and PFS (median 5.5 vs. 2.6 months; HR = 0.534, 95% CI: 0.355–0.805; p = 0.003; shown in Fig. 7d). However, in Child-Pugh A patients (n = 170), the observed differences in OS (median 16.4 vs. 8.7 months; HR = 0.668, 95% CI: 0.445–1.001; p = 0.051; shown in Fig. 7a) and PFS (median 7.9 vs. 5.2 months; HR = 0.956, 95% CI: 0.660–1.383; p = 0.810; shown in Fig. 7b) did not reach statistical significance.

Fig. 7.

Kaplan-Meier survival curves comparing OS and PFS among different Child-Pugh subgroups of patients after PSM (Child-Pugh A (a, b), Child-Pugh B (c, d)). As log-rank and Cox regression p values were nearly identical, only the Cox regression p values are displayed for clarity. OS, overall survival; PSM, propensity score matching; HR, hazard ratio; PFS, progression-free survival.

Treatment Safety

Adverse events are summarized in Table 7. Compared to the systemic therapy group, patients receiving combination therapy exhibited significantly higher rates of leukopenia (38.5% vs. 21.2%, p = 0.001), thrombocytopenia (42.3% vs. 30.8%, p = 0.041), anemia (47.9% vs. 32.7%, p = 0.007), elevated ALT/AST (38.4% vs. 25.0%, p = 0.011), vomiting (29.1% vs. 16.3%, p = 0.008), abdominal pain (48.4% vs. 34.6%, p = 0.014), fever (42.1% vs. 23.1%, p = 0.002), fatigue (52.5% vs. 37.5%, p = 0.008), proteinuria (38.7% vs. 27.9%, p = 0.009), and ascites/pleural effusion (29.3% vs. 20.2%, p = 0.008). The incidence of grade 3–4 proteinuria was also significantly higher in the combination therapy group (7.8% vs. 1.9%, p = 0.032). Other grade 3–4 adverse events were more frequent in the combination therapy group, including leukopenia (3.5% vs. 1.0%), thrombocytopenia (8.8% vs. 5.8%), anemia (9.6% vs. 4.8%), elevated ALT/AST (8.3% vs. 5.8%), and abdominal pain (5.3% vs. 2.9%), although these differences were not statistically significant.

Table 7.

Treatment-related adverse events

AEs	Any grade, n (%)		p value	Grade 3–4, n (%)		p value
	systemic therapy (n = 104)	combination therapy (n = 374)		systemic therapy (n = 104)	combination therapy (n = 374)
Hematological AEs, n (%)
Anemia	34 (32.7)	179 (47.9)	0.007	5 (4.8)	36 (9.6)	0.164
Thrombocytopenia	32 (30.8)	158 (42.3)	0.041	6 (5.8)	33 (8.8)	0.418
Leukopenia	22 (21.2)	144 (38.5)	0.001	1 (1.0)	13 (3.5)	0.321
Non-hematological AEs, n (%)
Fatigue	39 (37.5)	196 (52.5)	0.008	4 (3.8)	24 (6.4)	0.478
Abdominal pain	36 (34.6)	181 (48.4)	0.014	3 (2.9)	20 (5.3)	0.438
Hypoalbuminemia	35 (33.7)	181 (48.3)	0.008	0 (0)	0 (0)	1.000
Hypertension	35 (33.7)	139 (37.2)	0.565	11 (10.6)	68 (18.2)	0.073
Proteinuria	29 (27.9)	145 (38.7)	0.009	2 (1.9)	29 (7.8)	0.032
Diarrhoea	29 (27.9)	105 (28.2)	1.000	2 (1.9)	12 (3.2)	0.744
Elevated ALT/AST	26 (25.0)	144 (38.4)	0.011	6 (5.8)	31 (8.3)	0.534
Fever	24 (23.1)	157 (42.1)	0.002	1 (1.0)	8 (2.1)	0.321
Constipation	22 (21.2)	98 (26.1)	0.310	4 (3.8)	7 (1.9)	0.744
Ascites/pleural effusion	21 (20.2)	110 (29.3)	0.008	6 (5.8)	26 (7.0)	0.478
Rash	20 (19.2)	69 (18.5)	0.887	1 (1.0)	10 (2.7)	0.470
Hyperbilirubinemia	19 (18.3)	99 (26.6)	0.095	5 (4.8)	24 (6.4)	0.648
Hypothyroidism	18 (17.3)	65 (17.4)	1.000	0 (0)	0 (0)	1.000
Vomiting	17 (16.3)	109 (29.1)	0.008	2 (1.9)	8 (2.1)	1.000
Gastrointestinal hemorrhage	15 (14.4)	65 (17.3)	0.554	7 (6.7)	35 (9.4)	0.557
Hand-foot skin reaction	5 (4.8)	22 (6.0)	0.813	0 (0)	0 (0)	1.000
Gastrointestinal perforation	2 (1.9)	10 (2.7)	1.000	1 (1.0)	4 (1.1)	1.000

Values are presented as n (%).

p values were calculated using a two-sided χ² test.

AEs, adverse events; ALT, alanine aminotransferase; AST, aspartate aminotransferase.

Despite the increased toxicity profile, these adverse events were generally manageable through standard supportive care and treatment modifications. Dose reductions of targeted agents due to treatment-related adverse events occurred in 28.3% patients in the combination therapy group versus 18.3% patients in the systemic therapy group (p = 0.039), while temporary discontinuation of systemic therapy was required in 15.2% patients versus 8.7% patients in the combination therapy and systemic therapy groups, respectively (p = 0.085). Only 6.0% of patients in the combination therapy group permanently discontinued all anticancer treatments because of toxicity, compared to 3.8% patients in the systemic therapy group (p = 0.418). This manageable safety profile, coupled with the significant improvement in disease control and survival outcomes observed in the combination therapy group, suggests that the enhanced efficacy outweighs the increased but controllable toxicity burden. The higher incidence of hematological and hepatic adverse events in the combination therapy group is consistent with the known toxicity profiles of TACE/HAIC, while the increase in proteinuria may reflect the extended duration and potentially enhanced biological activity of anti-angiogenic therapy in the multimodal setting.

Despite the higher frequency of several adverse events in the combination therapy group, the overall incidence of serious adverse events, defined as events causing death, life-threatening complications, prolonged hospitalization, or substantial disability/incapacity, was comparable between the two groups (all p > 0.05). Given the retrospective design, detailed grading and attribution of adverse events may be limited in accuracy. Therefore, we have chosen not to perform further subgroup analyses of AE severity or treatment discontinuation.

Discussion

This retrospective study demonstrates that combining systemic therapy with locoregional treatment (TACE or HAIC) significantly improves survival outcomes compared to systemic therapy alone in patients with HCC and PVTT. After PSM, the combination therapy group achieved significantly longer median OS (15.7 vs. 5.9 months), extended median PFS (7.0 vs. 3.6 months), and a higher DCR (43.5% vs. 27.5%) than the systemic therapy group. Importantly, a greater proportion of patients in the combination therapy group underwent successful salvage resection following tumor downstaging. These findings support the therapeutic value of integrating systemic and locoregional modalities for this high-risk population.

PVTT is a well-established negative prognostic factor in HCC, associated with aggressive tumor behavior and limited treatment options. Current guidelines recommend systemic therapy as the standard of care for advanced HCC with vascular invasion [10]. However, the efficacy of systemic agents alone remains suboptimal, highlighting the need for more effective approaches. Previous studies have indicated that locoregional therapies, including TACE and HAIC, can achieve meaningful tumor control in the portal venous system [11, 12]. Moreover, emerging evidence suggests that combining these approaches with immunotherapy and antiangiogenic agents may produce synergistic effects, potentially enhancing immune activation and improving vascular normalization [13]. The present findings align with this literature, reinforcing the clinical potential of a multimodal strategy. The marked improvement in disease control, particularly reflected by the significantly higher SD and DCR observed in the combination therapy group, likely stems from the synergistic effect between locoregional and systemic therapies (shown in Table 3). First, TACE/HAIC provides potent local tumor control through direct cytotoxic and ischemic effects, which may stabilize primary liver lesions and PVTT that are less responsive to systemic therapy alone. Concurrently, the locoregional intervention-induced tumor necrosis and antigen release are hypothesized to enhance systemic immune activation by immunotherapy (e.g., ICIs) [14]. Second, antiangiogenic agents promote the normalization of the tumor vasculature, thereby enhancing drug delivery and ameliorating hypoxia-driven treatment resistance [15]. Furthermore, HAIC-administered chemotherapy not only delivers potent cytotoxic agents directly but also alters the tumor microenvironment to facilitate immune cell infiltration [16]. Taken together, these effects constitute a comprehensive, multimodal control mechanism that targets multiple facets of tumor progression and resistance, underlying the superior disease control observed with the combined modality approach.

The observed survival benefit of combination therapy remained robust even after controlling for heterogeneity in systemic treatment. Multivariate analysis adjusting for specific systemic agents (bevacizumab vs. TKI; PD-1 vs. PD-L1 inhibitor) exhibited no significant association with survival (as shown in Tables 3, 5), reinforcing the conclusion that the addition of locoregional therapy improves outcomes independent of the concurrent systemic regimen chosen in clinical practice.

Table 5.

Univariate Cox regression analyses of the prognostic factors for PFS

Variables	Group	n	PFS univariate		p value
			HR	95% CI
Age	>60	148	0.895	0.716–1.118	0.328
	≤60	330
Gender	Male	437	0.772	0.549–1.085	0.136
	Female	41
ECOG	0	199	0.945	0.767–1.165	0.599
	1	279
HBV/HCV infection	Present	301	1.075	0.869–1.330	0.506
	Absent	177
Imaging cirrhosis	Present	381	0.975	0.757–1.255	0.842
	Absent	97
Child-Pugh	A	352	0.669	0.532–0.841	<0.001
	B	126
ALBI	Grade 1	206	0.833	0.677–1.025	0.084
	Grade 2/3	272
PLT	Normal	379	0.947	0.739–1.214	0.670
	Abnormal	99
AFP	≤20	98	1.072	0.838–1.373	0.579
	>20	380
Tumor involvement	Hemiliver	288	0.742	0.602–0.913	0.005
	Both hemiliver	190
TBS	Low and moderate	216	0.978	0.796–1.201	0.829
	High	262
PVTT	Vp1-2	65	0.834	0.611–1.140	0.256
	Vp3-4	413
Distant metastasis	Present	193	1.303	1.059–1.604	0.012
	Absent	285
Treatment	Combination therapy	374	0.659	0.518–0.838	<0.001
	Systemic therapy	104
Targeted therapy	Bevacizumab	163	1.207	0.970–1.502	0.092
	TKI	315
Immunotherapy	PD-1	413	1.132	0.828–1.546	0.437
	PD-L1	65

OS, overall survival; HR, hazard ratio; CI, confidence interval; ECOG, eastern cooperative oncology group; HBV, hepatitis B virus; HCV, hepatitis C virus; Child-Pugh, Child-Turcotte-Pugh score; ALBI, albumin-bilirubin index; PLT, platelets; AFP, alpha-fetoprotein; TBS, tumor burden score; PVTT, portal vein tumor thrombus; TKIs, tyrosine kinase inhibitors; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1.

Beyond the consistency across drug regimens, further subgroup analyses based on disease characteristics yielded clinically relevant observations. In patients with a high tumor burden, combination therapy significantly improved OS (15.9 vs. 6.9 months; p = 0.005). In contrast, no significant difference in PFS was observed between the treatment groups (7.0 vs. 3.7 months; HR = 0.796, 95% CI: 0.553–1.146; p = 0.220). This divergence between OS and PFS may result from multiple factors. Notably, the extended OS despite modest PFS gain suggests a substantial prolongation of post-progression survival in the combination therapy group. This could be explained by the initial combination therapy effectively controlling the overall disease tempo, preserving liver function, and maintaining patients’ performance status. Consequently, these patients may have been better candidates to receive and benefit from subsequent lines of therapy, which could have contributed to the observed long-term survival advantage. Patients with high tumor burden often present with complex disease patterns that challenge accurate radiologic assessment. As a result, standard imaging criteria may fail to capture true disease progression, particularly in the presence of microscopic metastases or infiltrative tumor growth [17]. Furthermore, the immunomodulatory effects of combination therapy may alter disease kinetics. Some patients may exhibit pseudoprogression or delayed tumor response, in which initial radiological progression is followed by subsequent tumor regression or prolonged disease stabilization [18]. Such atypical response patterns, previously observed in studies of immunotherapy in advanced HCC [18], can weaken the correlation between PFS and long-term survival outcomes. Additionally, locoregional therapies may provide clinical benefits not fully reflected by PFS metrics. These include alleviation of tumor-related symptoms, prevention of vascular complications, and preservation or improvement of liver function, all of which may contribute to better overall condition and increased tolerance to subsequent lines of therapy [12, 19].

Second, patients with Vp4 PVTT experienced substantial benefit from combination therapy, exhibiting significantly longer median OS (13.5 vs. 4.0 months; p < 0.001) and PFS (6.75 vs. 2.9 months; p = 0.012). This finding is clinically meaningful, given the particularly poor prognosis typically associated with Vp4 involvement. A noteworthy finding is that the median OS of 5.8 months observed in our systemic therapy cohort was lower than that reported for the Vp4 subgroup in the IMbrave150 trial (7.6 months) [20] and for the TACE-alone group in the IJS 2023 study (10.4 months) [21]. This disparity may be attributed to differences in patient risk profiles and treatment modalities across these studies. Our cohort represented a particularly high-risk population, with 58.7% of patients having Vp4 portal vein invasion and 57.7% of patients presenting with high tumor burden − a “dual-high-risk” profile associated with aggressive disease biology. In contrast, while the IMbrave150 Vp4 subgroup also comprised high-risk patients, all of them received the standardized, first-line “T+A” regimen, whereas only 18.3% of our patients were treated with this combination, reflecting the heterogeneity of real-world practice. Moreover, compared to the localized mechanism of action of TACE, which can provide rapid cytoreduction and direct effects on thrombus vascularity, systemic therapies may exhibit a slower onset of response in high intrahepatic tumor burden setting. These findings highlight the limited survival benefit of systemic therapy alone in patients with both Vp4 disease and high tumor burden, emphasizing the need for more integrated strategies − such as combining systemic agents with locoregional therapies − to improve outcomes in this exceptionally high-risk subgroup. Therefore, the achievement of a 13.5-month median OS in our Vp4 subgroup with combination therapy is particularly striking. For instance, a recent multicenter randomized trial reported a median OS of only 6.3 months for patients receiving sorafenib combined with TACE (Sora-TACE) and 9.9 months with more intensive locoregional intervention using an irradiation stent alongside TACE [22]. In comparison, the 13.5-month median OS observed in our cohort reflects an approximate improvement of 7.2 months over Sora-TACE and 3.6 months over stent-based brachytherapy. These data highlight the superior efficacy of the integrated locoregional and systemic treatment approach.

Third, patients with Child-Pugh B liver function showed significantly improved OS (11.0 vs. 2.9 months; p < 0.001) and PFS (5.5 vs. 2.6 months; p = 0.003) when treated with combination therapy. The importance of baseline liver function was further reinforced by our formal multivariate Cox regression models, where Child-Pugh class B was confirmed as an independent risk factor for both reduced OS and PFS. Notably, even after including this potent prognostic variable, the combination therapy strategy retained its independent association with superior survival outcomes. This statistical reinforcement substantiates our clinical observation: the marked benefit of combined modality treatment in patients with compromised liver function (Child-Pugh B) likely results from its ability to provide rapid and effective local tumor control, thereby alleviating tumor-induced portal hypertension and preserving residual hepatic reserve, rather than merely reflecting the natural course of hepatic decompensation.

In the Child-Pugh A subgroup, comparable baseline characteristics – particularly for extrahepatic spread (47.5% vs. 49.1%) and Vp4 PVTT (51.7% vs. 52.6%) – indicate that the lack of observed benefit from combination therapy is not attributable to conventionally defined, more advanced disease stage. Instead, the outcome likely reflects a higher prevalence of intrinsically resistant tumor biology within this subgroup. Tumors in these patients may harbor more aggressive molecular subtypes or distinct microenvironmental features that diminish the efficacy of the combined modality approach, independent of macroscopically assessed tumor burden. This intrinsic resistance hypothesis is further supported by the contrasting remarkable benefit observed in Child-Pugh B patients. In the context of compromised hepatic reserve, the rapid local tumor control afforded by combination therapy can substantially improve liver function and survival – a therapeutic effect that is less critical and therefore less visible when baseline liver function is already preserved. This interpretation aligns with the established understanding that the survival of patients with advanced HCC depends not only on tumor control but also on their hepatic functional recovery capacity following therapy [23]. The delineation of this potential intrinsic resistance phenotype has direct implications for treatment sequencing and patient selection. Compared to the 47.2% conversion rate reported in our previous study [7], the current cohort achieved a more modest conversion rate of 13.0% following tumor downstaging. This discrepancy underscores that, while combination therapy remains effective for prolonging survival, the consistent achievement of high conversion rates may depend on overcoming such intrinsic resistance through biomarker-guided stratification or novel therapeutic combinations.

Despite the survival advantage, combination therapy was associated with a higher incidence of adverse events, including hematological toxicity, transaminitis, gastrointestinal symptoms, and proteinuria. Notably, grade 3–4 proteinuria occurred more frequently in the combination therapy group (7.8% vs. 1.9%; p = 0.032). However, the overall incidence of serious adverse events remained manageable and did not differ significantly between the two groups (all p > 0.05). The favorable efficacy-safety profile observed in Child-Pugh B patients warrants particular attention. While these patients experienced more frequent hematologic toxicity and liver function abnormalities, these events were predominantly low-grade and reversible with standard supportive measures, without increasing the risk of hepatic decompensation. This suggests that with careful patient selection, close monitoring of liver function, and prompt intervention using established dose modification guidelines (e.g., withholding anti-angiogenic agents for grade 2 proteinuria or grade 3 hematologic toxicity), the survival benefit of combination therapy can be realized without compromising safety, even in patients with compromised hepatic reserve. These findings suggest that although combination therapy increases the burden of toxicity, it remains clinically manageable with appropriate surveillance and supportive care. With careful monitoring and thoughtful patient selection, this treatment approach continues to be a viable option for patients with advanced HCC and PVTT.

Several limitations of this study should be acknowledged. First, the retrospective nature of the study introduces potential selection bias, even though PSM was employed to adjust for baseline imbalances. Second, variability in treatment protocols may have influenced clinical outcomes. Third, the relatively shorter follow-up period for the systemic therapy group could influence long-term survival estimates. However, the shorter follow-up period in the systemic arm reflects the higher early mortality associated with the natural history of advanced HCC. The inclusive enrollment of patients up to late 2024 further contributed to this follow-up imbalance. Nevertheless, the application of a fixed data cutoff and the clear separation of survival curves support the validity of the primary endpoint comparisons. Finally, this was a single-center study, which may limit the external validity of the findings. Furthermore, while our study focused on treatment-naive patients, the therapeutic principle of integrating locoregional control with systemic therapy may be relevant for broader clinical scenarios. For instance, in patients with recurrent HCC and PVTT following prior therapies, a similar combined modality approach could be reconsidered to manage localized disease progression while maintaining systemic efficacy, although the altered liver function and potential treatment resistance in this setting warrant careful evaluation. Moreover, the concept of direct PVTT targeting – central to both TACE and HAIC – logically extends to other localized modalities. External beam radiotherapy or transarterial radioembolization, which can precisely deliver high-dose radiation to the PVTT, represents another promising locoregional strategy that could be rationally combined with systemic agents [24]. Future prospective studies are encouraged to compare the efficacy and safety of different locoregional techniques (TACE, HAIC, and radiotherapy) within the framework of combined modality therapy for advanced HCC with PVTT. Future prospective multicenter trials are necessary to validate these results and refine treatment algorithms for this complex patient population.

Conclusion

Combination therapy involving TACE/HAIC alongside systemic agents significantly improves survival outcomes compared to systemic therapy alone in patients with HCC and PVTT. The observed enhancement in tumor response, particularly among patients with high tumor burden, Vp4 classification, and Child-Pugh B liver function, supports the integration of locoregional modalities into the therapeutic framework for this high-risk population. With appropriate patient selection and multidisciplinary coordination, this treatment strategy holds promise for establishing a new standard of care for patients with advanced HCC and vascular invasion.

Acknowledgments

The authors thank Ms. Chengcheng Li, Xu Zhang, Lixia Cai, and Mingmin Fan for their assistance with information collection. We appreciate the linguistic assistance provided by TopEdit (www.topeditsci.com) during the preparation of this manuscript.

Statement of Ethics

As this study is a retrospective review, it did not involve any prospective data collection from human subjects. The study protocol was reviewed and approved by the Research Ethics Committee of the First Affiliated Hospital of Zhejiang University (Approval No. [2025] IIT Ethics Approval No. 0459). This approval confirms that the study design and data analysis adhere to ethical standards and guidelines. Given the retrospective nature of the research involving the analysis of existing data without compromising patient welfare and rights, the requirement for informed consent was waived by the same Ethics Committee. The utilization of non-identified information ensures the protection of patient confidentiality, in accordance with ethical standards for retrospective research.

Conflict of Interest Statement

The authors declare no conflicts of interest.

Funding Sources

This study was financially supported by grants from National Key Research and Development Program of China (2024YFA1306404), the Clinical Medicine Innovation Team Training Project, Zhejiang University School of Medicine, the Joint Fund for Regional Innovation and Development of National Natural Science Foundation of China (U23A20462), and the Noncommunicable Chronic Diseases-National Science and Technology Major Project of China (2023ZD0502004). The funders had no role in the design, data collection, data analysis, and reporting of this study.

Author Contributions

F.G. and C.G.: conceptualization (equal), data curation (equal), formal analysis (equal), validation (equal), and writing – original draft (equal). Y.C.: data curation (equal), formal analysis (equal), investigation (equal), and validation (equal). Y.S., X.L., X.Z., and S.G.: investigation (equal), supervision (equal), and validation (equal). X.B. and T.L.: conceptualization (equal), investigation (equal), methodology (equal), supervision (equal), and writing – review and editing (equal).

Funding Statement

This study was financially supported by grants from National Key Research and Development Program of China (2024YFA1306404), the Clinical Medicine Innovation Team Training Project, Zhejiang University School of Medicine, the Joint Fund for Regional Innovation and Development of National Natural Science Foundation of China (U23A20462), and the Noncommunicable Chronic Diseases-National Science and Technology Major Project of China (2023ZD0502004). The funders had no role in the design, data collection, data analysis, and reporting of this study.

Data Availability Statement

The data that support the findings of this study are not publicly available due to privacy reasons but are available from the corresponding author, Xueli Bai, upon request.