Locoregional therapy combined with targeted therapy and immunotherapy for hepatocellular carcinoma with portal vein tumor thrombosis: a systematic review and meta-analysis

Abstract

Hepatocellular carcinoma (HCC) with portal vein tumor thrombosis (PVTT) represents a challenging clinical scenario with poor prognosis. This study aimed to evaluate the efficacy and safety of combining locoregional therapy with targeted therapy and immunotherapy (triple therapy) in unresectable HCC with PVTT. We conducted a Bayesian network meta-analysis of ten studies involving 1,241 HCC-PVTT patients, comparing triple therapy with targeted therapy plus immunotherapy in terms of overall survival (OS), progression-free survival (PFS), tumor response, and treatment-related adverse events. The results demonstrated that triple therapy significantly improved OS and PFS compared to targeted therapy plus immunotherapy alone. For OS, hepatic arterial infusion chemotherapy (HAIC)-based combination showed the greatest benefit (HR 0.48, 95% CI 0.32–0.74), followed by radiotherapy-based (HR 0.53, 95% CI 0.30–0.91) and transarterial chemoembolization-based combinations (HR 0.65, 95% CI 0.45–0.94). For PFS, radiotherapy-based triple therapy demonstrated the most pronounced benefit (HR 0.43, 95% CI 0.30–0.63), followed by HAIC-based combination (HR 0.49, 95% CI 0.34–0.72). While the addition of locoregional therapies increased the incidence of grade 3–4 adverse events (73.5% vs. 39.4%, p < 0.001), the safety profile remained clinically manageable. In conclusion, triple combination therapies represent a promising approach for unresectable HCC with PVTT that requires validation through large-scale randomized controlled trials to establish optimal treatment regimens.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-23027-6.

Introduction

Globally, hepatocellular carcinoma (HCC) ranks as the sixth most common cancer and the third leading cause of cancer-related mortality^1,2. Due to the anatomical characteristics of the liver and the highly aggressive biological nature of HCC, the tumor often invades the portal venous system, forming portal vein tumor thrombosis (PVTT), with an incidence ranging from 10% to 40%^3–5. Patients with PVTT not only experience intrahepatic and extrahepatic metastases but also suffer from portal hypertension, malignant ascites, and liver failure, consequently leading to rapid clinical deterioration and poor prognosis, with a median overall survival (OS) of only 2.7-4.0 months without intervention⁶.

The Barcelona Clinic Liver Cancer (BCLC) staging system classifies HCC with PVTT as advanced HCC (BCLC stage C), for which systemic therapy is the primary treatment recommendation⁷. Unfortunately, patients with major trunk PVTT (Vp3-Vp4) achieve disappointingly short median OS of only 4.3–7.2 months with sorafenib or lenvatinib monotherapy^8–11. In recent years, immune checkpoint inhibitors (ICIs) have emerged as promising treatment options for advanced HCC. Multiple phase III trials have demonstrated survival benefits when ICIs are combined with targeted therapy, including multi-targeted tyrosine kinase inhibitors (TKIs) or anti-angiogenic/anti-VEGF agents^12–15, establishing this combination as the recommended first-line treatment for advanced HCC. Nevertheless, overall efficacy remains unsatisfactory, with an objective response rate (ORR) of 25–35%^13–16. The therapeutic challenge is particularly pronounced in the PVTT subgroup. In the IMbrave150 study, atezolizumab plus bevacizumab achieved a median OS of only 7.6 months in the Vp4 PVTT subgroup¹⁷. Several retrospective studies have reported similarly modest outcomes, with median OS around 10 months in PVTT populations^18,19.

Locoregional therapies, including transarterial chemoembolization (TACE), hepatic arterial infusion chemotherapy (HAIC), external beam radiotherapy, transarterial radioembolization, and endovascular brachytherapy, have long been integral components of HCC management. Recent evidence suggests that triple therapy regimens (targeted therapy + immunotherapy + locoregional therapy), may improve outcomes. In the TALENTop study, approximately 60% of HCC patients with PVTT progressed after induction with atezolizumab plus bevacizumab. Without subsequent treatment, median OS was only 5.7 months, whereas the addition of locoregional therapy extended median OS to 16.2 months²⁰. Furthermore, in PVTT cohorts, combining locoregional therapy with targeted therapy and immunotherapy has been associated with ORRs of 50–60% and the disease control rates (DCR) of 80–90%, alongside prolonged survival^21–30. However, current evidence remains insufficient to establish the efficacy and safety of triple therapy in HCC-PVTT patients due to small sample sizes, heterogeneous results, and the lack of large prospective randomized controlled trials (RCTs). Real-world studies have also yielded conflicting results, with some indicating no survival benefit from adding locoregional therapy to systemic therapy in patients with main portal vein invasion^31–35. These observations imply that the indiscriminate application of triple therapy may not benefit all patients and could potentially increase toxicity and economic burden. Given the differing mechanisms, efficacy, and toxicity profiles of locoregional modalities, the optimal locoregional partner for targeted therapy plus immunotherapy remains undefined.

Currently, there is no consensus regarding the optimal management of HCC with PVTT. Treatment strategies vary regionally: Western countries favor systemic therapy, while the Asia-Pacific region advocates more aggressive approaches³⁶. The benefit of adding locoregional therapy to targeted therapy plus immunotherapy in unresectable HCC with PVTT remains unclear because patients with major trunk PVTT are rarely enrolled in clinical trials, and direct comparisons among regimens are lacking. To address these gaps, we conducted a comprehensive literature review and Bayesian network meta-analysis (NMA) to evaluate the efficacy and safety of adding locoregional therapy to targeted therapy plus immunotherapy in unresectable HCC with PVTT, aiming to provide valuable insights for clinical decision-making.

Materials and methods

This meta-analysis study adhered to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta Analyses) guidelines and was registered in the PROSPERO database (No. CRD42024577128). Ethical approval was not required as only published literature data were used.

Literature search

A systematic literature search was conducted in databases including PubMed, Embase, Web of Science, Cochrane Library and ClinicalTrials.gov, with the search time ranging from database inception to April 30, 2025. Search strategies were based on combinations of MeSH (Medical Subject Headings)/Emtree terms and keywords. Detailed search strategies are presented in Table S1. Reference lists of relevant reviews and meta-analyses were manually screened for comprehensiveness of the literature. EndNote software was used to remove duplicates.

Inclusion and exclusion criteria

Eligible studies must meet the following criteria: (1) the study design was either a prospective RCT or a retrospective cohort study employing propensity score matching (PSM) to reduce bias; (2) the study population consisted exclusively of HCC patients with PVTT, or, for broader advanced HCC cohorts, baseline, treatment, and outcome data for the PVTT subgroup were reported separately; (3) the study involved a comparative analysis of at least two treatment regimens, with at least one regimen comprising triple therapy: a targeted therapy agent, an immunotherapy agent, and a locoregional modality (e.g., TACE, HAIC, or radiotherapy). In this study, “targeted therapy” refers to multikinase TKIs (e.g., sorafenib, lenvatinib, regorafenib) or anti‑VEGF/anti‑angiogenic agents (e.g., bevacizumab, apatinib), and “immunotherapy” refers to ICIs targeting PD‑1/PD‑L1 or CTLA‑4; and (4) studies provided comprehensive details on treatment regimens, baseline characteristics, and outcome measures (e.g., survival, tumor response, adverse events). When multiple studies reported on the same patient cohort, the study with the most comprehensive data was selected. Exclusion criteria included: (1) Studies in which patients with PVTT constitute only a portion of the cohort and PVTT-specific data cannot be extracted; (2) Letters, case reports, animal studies, editorials, and conference reports; (3) Studies with flawed designs or incomplete data; and (4) Non-English publications.

Data extraction

A standardized data extraction template was used to systematically extract the following information from all included studies: (1) Study characteristics, including author, publication year, region, interventions, study design features (randomization, blinding, allocation concealment), and follow-up information; (2) Patient baseline characteristics, including sample size, gender, age, Eastern Cooperative Oncology Group (ECOG) performance status, BCLC stage, Child-Pugh classification, and PVTT classification; (3) Outcome measures for each treatment group, including detailed efficacy, safety, and other relevant clinical results.

The primary efficacy endpoint was OS. Secondary efficacy outcomes included PFS, ORR, and DCR. ORR included patients exhibiting complete response (CR) and partial response (PR), while DCR included those with CR, PR, and stable disease (SD). Efficacy assessment was based on the Modified Response Evaluation Criteria in Solid Tumors (mRECIST). Safety outcomes were primarily evaluated according to treatment-related adverse events (TRAEs), especially grade 3 or higher TRAEs classified according to the Common Terminology Criteria for Adverse Events (CTCAE). Continuous survival outcomes were measured using hazard ratios (HRs) with 95% confidence intervals (CIs), while dichotomous outcomes were measured using risk ratios (RRs) with 95% CIs. When HRs were not directly reported in the original studies, survival data were extracted from Kaplan-Meier curves using Engauge Digitizer 4.1 software according to the methodology described by Tierney et al.³⁸. All literature screening and data extraction were conducted independently by two authors, with any discrepancies or uncertainties resolved through consultation with a third author to achieve consensus.

Statistical analysis

This NMA was conducted within a Bayesian framework, using the gemtc package (version 1.0–2) in RStudio software (version 4.5.0) to activate JAGS software for executing the Markov Chain Monte Carlo (MCMC) sampling simulations and computations. The model parameters were set as: n.adapt = 50,000, n.iter = 100,000, thin = 1. Model convergence was assessed using multiple diagnostic methods, including trace plots, density plots, and Brooks-Gelman-Rubin diagnostic plots. (Figure S2-S3) Both fixed-effects and random-effects models were fitted and compared using the deviance information criterion (DIC), with the model demonstrating better fit selected for final analysis. Forest plots were constructed to visualize treatment effects and their corresponding 95% CIs relative to the reference treatment. The rank probabilities and cumulative ranking probability plots were generated to evaluate the probability of each treatment achieving specific ranks in terms of efficacy. The surface under the cumulative ranking curve (SUCRA) was calculated to provide a single summary measure for ranking different interventions, with higher SUCRA values (ranging from 0 to 1) indicating a greater probability of being among the most effective treatments. Heterogeneity was assessed with I², with values > 50% considered indicative of substantial heterogeneity.

Quality assessment

Two authors independently assessed methodological quality. For RCTs, the Cochrane Risk of Bias tool (ROB 2.0) was employed, while for non-RCTs, a revised Newcastle-Ottawa Scale (NOS) was used. Publication bias was assessed through visual inspection of funnel plots and statistical testing using Egger’s regression test, implemented with the netmeta package. Asymmetry in funnel plots or significant Egger’s test results (p < 0.05) were considered indicative of potential publication bias.

Results

Overview of the selected studies and patient characteristics

Following our established search strategy, we initially retrieved 504 potentially relevant studies from the database searches. After deduplication, 151 non-duplicate studies remained for screening. Subsequent title screening led to the exclusion of 97 studies due to irrelevance to our research objectives. After comprehensive abstract and full-text screening, a total of 10 studies ultimately met the inclusion criteria and were included in the final analysis^27,39–47. Figure 1 shows the flowchart of selection process.

Fig. 1.

Flow chart of selection process and search results.

The included studies, published between 2023 and 2025, were all conducted in China and enrolled exclusively Chinese adult patients diagnosed with HCC. Among the 10 studies, 9 were retrospective cohort studies employing PSM methods, and 1 was a prospective RCT. A total of 1,241 patients were included: 558 received targeted therapy plus immunotherapy alone, 588 received triple therapy (targeted therapy + immunotherapy + locoregional therapy), 60 received TACE combined with targeted therapy, and 35 received HAIC monotherapy. Among the triple therapy patients, 293 received TACE-based combinations, 176 received HAIC-based combinations, and 119 received radiotherapy-based combinations. The majority of patients were male (91.2%) and classified as Child-Pugh A (80.0%) with an ECOG performance status of 0–1 (95.3%). Detailed clinical characteristics are shown in Table 1. For patients undergoing TACE, the commonly used chemotherapeutic agents included platinum compounds, doxorubicin/epirubicin, mitomycin, and 5-fluorouracil, along with absorbable gelatin sponge and/or iodized oil as embolization materials. HAIC was predominantly administered using oxaliplatin-based FOLFOX regimens (fluorouracil, leucovorin, and oxaliplatin), delivered continuously via hepatic arterial microcatheters. External beam radiotherapy was performed using either intensity-modulated radiation therapy (IMRT) or stereotactic body radiation therapy (SBRT) techniques. The gross tumor volume (GTV) primarily encompassed the PVTT and, when tumor burden and liver function permitted, adjacent primary hepatic lesions. The biological equivalent dose (BED, assuming α/β = 10 Gy) ranged from 53.1 to 86.4 Gy across different fractionation schemes.

Table 1.

Characteristics of included studies.

Author, year	Country	Study design	Interventions	Participants	Sex (M/F)	Child-Pugh (A/B)	ECOG (0/1/2)	Type of PVTT (Cheng’s classification I-IV or Vp1-4)#	Extrahepatic metastasis Y/N	ORR (%)	DCR (%)	mOS (months)	mPFS (months)
Zhang ZH, 2023	China	retrospective PSM	TACE + EVBT stent + Lenvatinib + PD-1	32	29/3	30/2	0/1–32; 2 − 0	Vp4	3/29	55.3*	70.2*	17.7	17.0
			TACE + EVBT stent + Lenvatinib	32	30/2	30/2	0/1–29; 2–3	Vp4	2/30	17.5*	30.0*	12.0	8.0
Xia WL, 2023	China	retrospective PSM	TACE + Apatinib + PD-1	28	25/3	26/2	7/21/0	18/4/6/0	13/15	53.6	82.1	14.6	6.9
			TACE + Apatinib	28	26/2	26/2	7/21/0	21/4/3/0	14/14	17.9	75.0	8.5	4.0
Zhang JX, 2024	China	retrospective IPTW	TACE + TKI + ICI	106	94/12	106/0	65/41/0	24/33/49/0	28/78	50.9	76.4	19.1	9.1
			TKI + ICI	109	98/11	109/0	66/43/0	25/35/49/0	29/80	28.4	57.8	12.7	5.0
Mei JH, 2025	China	retrospective PSM	irradiation stent + TACE + TKI + ICI	127	118/9	69/58	41/66/20	0/0/0/127	0/127	58.3	75.6	11.9	5.2
			TKI + ICI	220	204/16	121/99	70/115/35	0/0/0/220	0/220	29.1	54.1	8.6	3.0
Li YY, 2024	China	retrospective PSM	HAIC + Apatinib + Camrelizumab	83	76/7	80/3	81/2/0	NA	51/32	42.2	94.0	18.7	10.0
			Apatinib + Camrelizumab	83	77/6	75/8	82/1/0	NA	41/42	42.2	78.3	11.0	5.6
Fu YZ, 2023	China	retrospective PSM	HAIC + Lenvatinib + PD-1	58	54/4	57/1	0/1–58	0/10/18/30	21/37	61.8*	86.5*	22.8	10.7
			Lenvatinib + PD-1	47	44/3	41/6	0/1–47	0/7/13/27	20/27	20.8*	56.6*	13.5	5.4
Tang SY, 2025	China	retrospective PSM	HAIC + Lenvatinib + Tislelizumab	35	34/1	28/7	23/12/0	0/0/0/35	13/22	77.1*	94.3	23.2	6.6
			HAIC	35	34/1	33/2	24/11/0	0/0/0/35	13/22	42.9*	65.7	6.9	2.4
Tang CP, 2023	China	retrospective PSM	RT + ICI + Bevacizumab	47	41/6	42/5	15/32/0	4/21/18/4	28/19	48.9	97.9	not reach	9.6
			ICI + Bevacizumab	47	40/7	39/8	13/34/0	5/15/22/5	23/24	27.7	83.0	not reach	5.4
Hu Y, 2023	China	RCT	SBRT + Camrelizumab + Apatinib	40	33/7	33/7	16/24/0	0/11/23/6	19/21	47.5	72.5	12.7	4.6
			Camrelizumab + Apatinib	20	18/2	14/6	9/11/0	0/9/6/5	13/7	20.0	40.0	8.6	2.5
Ma JN, 2025	China	retrospective PSM	RT + TKI + PD-1	32	28/4	18/14	19/13/0	0/12/13/7	7/25	78.1	93.8	15.6	8.1
			TKI + PD-1	32	29/3	16/16	21/11/0	0/13/11/8	9/23	56.3	81.3	8.2	5.2

M, male; F, female; ECOG, Eastern Cooperative Oncology Group; ORR, overall response rate; DCR, disease control rate; OS, overall survival; PFS, progression-free survival; PSM, propensity score-matched; IPTW, inverse probability of treatment weighting; TACE, transarterial chemoembolization; RT, radiotherapy; EVBT, endovascular brachytherapy; HAIC, hepatic arterial infusion chemotherapy; PD-1, Programmed death-1 inhibitor; ICI, immune checkpoint inhibitor (e.g., atezolizumab, camrelizumab, sintilimab, tislelizumab, pembrolizumab, nivolumab, toripalimab); TKI, tyrosine kinase inhibitor (e.g., lenvatinib, sorafenib, apatinib, regorafenib); NA, not available.

^#PVTT was reclassified based on Cheng’s classification (type I, tumor thrombus involving segmental branches of portal vein or above; type II, tumor thrombus involving right/left portal vein; type III, tumor thrombus involving main portal vein; and type IV, tumor thrombus involving superior mesenteric vein) or classification proposed by Liver Cancer Study Group of Japan (Vp1, tumour thrombus distal but not in second-order branches of the main portal vein; Vp2, tumour thrombus in second-order branches; Vp3, tumour thrombus in first-order branches; and Vp4, tumour thrombus in the main trunk, contralateral to the trunk, or both).

^*Treatment response for intrahepatic tumors in all patients.

Quality assessment

For the 9 retrospective cohort studies, quality was evaluated using the NOS, with detailed scores presented in Table S2. All retrospective studies demonstrated high methodological quality, with NOS scores ranging from 8 to 9 points. The single prospective RCT was assessed using the RoB 2.0 and was rated as having low risk of bias across all domains. Assessment of publication bias revealed no significant concerns, as evidenced by approximately symmetrical funnel plot morphology and non-significant results from both Egger’s (p > 0.05) and Begg’s tests (p > 0.05). (Figure S1)

Outcome of Bayesian network meta-analysis

Survival outcome (PFS and OS)

In the preliminary analysis, both random-effects and fixed-effects models were fitted and compared using DIC. Given the better model fit and expected heterogeneity across studies, the random-effects model was selected for all subsequent analyses.

All ten studies provided PFS data with HRs and 95% CIs. A network plot summarized direct and indirect comparisons among these regimens. (Fig. 2A) Forest plot analysis demonstrated that all triple therapy regimens significantly improved PFS compared to targeted therapy plus immunotherapy alone (TKI + ICI): RT + TKI + ICI showed the greatest benefit (HR 0.43, 95% CI 0.30–0.63), followed by HAIC + TKI + ICI (HR 0.49, 95% CI 0.34–0.72), and TACE + TKI + ICI (HR 0.60, 95% CI 0.42–0.85). In contrast, TACE combined with TKI showed no significant PFS benefit compared to TKI + ICI (HR 1.10, 95% CI 0.67–1.90). (Fig. 2B) The SUCRA ranking further confirmed RT + TKI + ICI as optimal combination (SUCRA 0.9213), followed by HAIC + TKI + ICI (SUCRA 0.8154) and TACE + TKI + ICI (SUCRA 0.6445). Treatment ranking analyses depicted the likelihood of efficacy comparisons, and the cumulative ranking probability plots are shown in Fig. 2C.

Fig. 2.

Meta-analysis results of progression-free survival (PFS). (A) Network plot of PFS; (B) forest plot of PFS for different regimens vs. targeted therapy plus immunotherapy; (C) cumulative ranking probability plot of PFS. TKI, tyrosine kinase inhibitor; ICI, immune checkpoint inhibitor; TACE, transarterial chemoembolization; HAIC, hepatic arterial infusion chemotherapy; RT, radiotherapy. Note: In Figs. 2, 3, 4 and 5, ‘TKI’ denotes targeted therapy and may indicate either a multikinase TKI (e.g., lenvatinib, sorafenib, apatinib, regorafenib) or an anti‑VEGF/anti‑angiogenic agent (e.g., bevacizumab).

Nine studies reported OS data. (Fig. 3A) All triple therapy regimens significantly improved OS compared to TKI + ICI: HAIC + TKI + ICI demonstrated the greatest OS benefit (HR 0.48, 95% CI 0.32–0.74), followed by RT + TKI + ICI (HR 0.53, 95% CI 0.30–0.91), and TACE + TKI + ICI (HR 0.65, 95% CI 0.45–0.94). The TACE + TKI and TKI + ICI showed no significant difference in OS (HR 1.50, 95% CI 0.78–2.90). (Fig. 3B) The SUCRA ranking analysis indicated that HAIC + TKI + ICI was most likely to be associated with the longest OS (SUCRA 0.8950), followed by RT + TKI + ICI (SUCRA 0.8220) and TACE + TKI + ICI (SUCRA 0.6584). (Fig. 3C) These findings consistently demonstrate the superior efficacy of triple therapy regimens over targeted therapy plus immunotherapy alone in HCC patients with PVTT.

Fig. 3.

Meta-analysis results of overall survival (OS). (A) Network plot of OS; (B) forest plot of OS for different regimens vs. targeted therapy plus immunotherapy; (C) cumulative ranking probability plot of OS.

Tumor response (ORR and DCR)

Ten studies reported ORR and DCR outcomes. (Fig. 4A) Network meta-analysis revealed that triple therapy regimens showed higher tumor response rates compared to systemic therapy alone, though most differences did not reach statistical significance. In terms of ORR, TACE + TKI + ICI (RR 2.09, 95% CI 0.98–4.37), HAIC + TKI + ICI (RR 1.60, 95% CI 0.77–3.64), and RT + TKI + ICI (RR 1.71, 95% CI 0.92–3.49) all showed numerically favorable but non-significant RRs relative to TKI + ICI. (Fig. 4B; Table 2) Notably, TACE + TKI + ICI significantly outperformed TACE + TKI (RR 3.08, 95% CI 1.28–8.01), which was the only statistically significant comparison, indicating significant benefit from adding ICIs. (Table 2)

Fig. 4.

Meta-analysis results of objective response rate (ORR) and disease control rate (DCR). (A) Network plot of ORR and DCR; (B) forest plot of ORR for different regimens vs. targeted therapy plus immunotherapy; (C) forest plot of DCR for different regimens vs. targeted therapy plus immunotherapy.

Table 2.

Network comparisons of outcomes among different treatments.

ORR
TACE + TKI + ICI
1.31 (0.42, 3.65)	HAIC + TKI + ICI
1.22 (0.43, 3.15)	0.93 (0.34, 2.61)	RT + TACE + TKI
3.08 (1.28, 8.01)	2.38 (0.62, 10.5)	2.54 (0.7, 10.56)	TACE + TKI
2.09 (0.98, 4.37)	1.6 (0.77, 3.64)	1.71 (0.92, 3.49)	0.68 (0.2, 2.12)	TKI + ICI
2.38 (0.49, 10.5)	1.82 (0.61, 5.46)	1.95 (0.44, 8.73)	0.77 (0.12, 4.27)	1.14 (0.29, 4.18)	HAIC
DCR
TACE + TKI + ICI
1.02 (0.58, 1.7)	HAIC + TKI + ICI
1.08 (0.61, 1.68)	1.05 (0.62, 1.69)	RT + TACE + TKI
1.38 (0.93, 2.28)	1.34 (0.73, 2.96)	1.28 (0.73, 2.83)	TACE + TKI
1.33 (0.91, 1.94)	1.29 (0.92, 1.96)	1.23 (0.94, 1.82)	0.97 (0.5, 1.63)	TKI + ICI
1.47 (0.66, 3.12)	1.44 (0.83, 2.54)	1.37 (0.67, 3)	1.07 (0.4, 2.4)	1.11 (0.55, 2.14)	HAIC

Comparisons should be read from left to right. Cells marked in bold are significant (RR > 1).

TACE, transarterial chemoembolization; TKI, tyrosine kinase inhibitor (e.g., lenvatinib, sorafenib, apatinib, regorafenib) plus bevacizumab; ICI, immune checkpoint inhibitor (e.g., atezolizumab, camrelizumab, sintilimab, tislelizumab, pembrolizumab, nivolumab, toripalimab); HAIC, hepatic arterial infusion chemotherapy; RT, external beam radiotherapy.

For DCR, triple therapy regimens similarly showed non-significant improvements over TKI + ICI: TACE + TKI + ICI (RR 1.33, 95% CI 0.91–1.94), HAIC + TKI + ICI (RR 1.29, 95% CI 0.92–1.96), and RT + TKI + ICI (RR 1.23, 95% CI 0.94–1.82). (Fig. 4C; Table 2)

Subgroup analysis by portal vein invasion extent

A total of five included studies provided relevant data on outcomes in patients with main trunk PVTT, with detailed results presented in Table S3. To address heterogeneity in classification systems across studies, we established correspondence between the Japanese portal vein invasion classification (Vp1-4) and Cheng’s classification system (Type I-IV) based on anatomical extent of invasion. For subgroup analysis, we stratified PVTT into main trunk PVTT (Vp4 or Type III/IV) and non-main trunk PVTT (Vp1-3 or Type I/II).

For patients with main trunk PVTT, triple therapy showed hazard ratios below 1.0 for both PFS and OS endpoints, though the 95% confidence intervals crossed the null value of 1.0, indicating no statistically significant benefit compared to TKI + ICI. (Figure 5A and C) For patients with non-main trunk PVTT, HAIC + TKI + ICI showed the most robust benefits for both PFS (HR 0.43, 95% CI: 0.20–0.85) and OS (HR 0.39, 95% CI: 0.12–1.1). Radiotherapy-based triple therapy also showed significant PFS benefit (HR 0.41, 95% CI: 0.20–0.87) and favorable OS trend (HR 0.65, 95% CI: 0.13–3.2). (Figure 5B and D) These findings suggest that the extent of portal vein invasion may serve as an important factor in treatment selection and patient stratification for future clinical trials.

Fig. 5.

Subgroup analysis results of progression-free survival (PFS) and overall survival (OS) for triple therapy vs. targeted therapy plus immunotherapy stratified by portal vein invasion extent. (A) PFS in main trunk PVTT; (B) PFS in non-main trunk PVTT; (C) OS in main trunk PVTT; (D) OS in non-main trunk PVTT.

Treatment-related adverse events (TRAEs)

All ten studies reported relevant toxicity data, which are summarized in Table S4. The common TRAEs included fever, upper-abdominal pain, nausea/vomiting, and fatigue, which are typically considered manifestations of post-interventional syndrome. Other frequently reported AEs included hypertension, hand-foot syndrome, and loss of appetite. Laboratory abnormalities included hepatotoxicity (elevated transaminase and bilirubin) and hematological toxicity (thrombocytopenia, leukopenia, and anemia). Dermatological toxicity (hand-foot syndrome, rash, pruritus) and hypertension are often associated with specific medications, necessitating particular attention and monitoring. No treatment-related deaths were reported across all studies.

Grade 3/4 TRAEs occurred in 432 of 588 triple therapy patients (73.5%) compared to 220 of 558 patients (39.4%) receiving targeted therapy plus immunotherapy alone, representing a significant increase (χ²=136.27, p < 0.001; RR 1.87, 95% CI 1.67–2.09). Despite substantial heterogeneity across studies due to variations in grading criteria, follow-up periods, and assessment methods, this increased toxicity risk was consistently observed.

Assessment of consistency and heterogeneity

The consistency assumption was evaluated using the node-splitting method. No significant inconsistency was detected between direct and indirect evidence (all p-values > 0.05), indicating that the network meta-analysis assumptions were satisfied. Heterogeneity analysis revealed low heterogeneity across most treatment comparisons (I² = 0.0%), with only one comparison TKI + ICI vs. RT + TKI + ICI showing moderate heterogeneity (I² = 24%). Detailed heterogeneity assessments are presented in Figure S4.

Discussion

Managing unresectable HCC with PVTT remains challenging, as prospective trials often exclude patients with PVTT or macrovascular invasion, resulting in limited evidence for this high‑risk population. While several studies have suggested that triple therapy (targeted therapy + immunotherapy + locoregional therapy) improves efficacy, the majority are retrospective studies with limited sample sizes that provide only low-quality evidence. To our knowledge, this is the first network meta-analysis to compare adding locoregional therapy to targeted therapy plus immunotherapy versus targeted therapy plus immunotherapy alone in unresectable HCC with PVTT. Our findings demonstrate that triple therapy significantly improves both OS and PFS, indicating a clinically meaningful therapeutic approach for these patients. Among the locoregional modalities evaluated, HAIC-based and radiotherapy-based triple regimens demonstrated superior efficacy profiles relative to TACE-based combinations. Nevertheless, given the heterogeneity in efficacy and safety across locoregional approaches, the optimal treatment strategy for this population requires confirmation in high‑quality prospective studies.

TACE has long been the standard treatment for unresectable HCC, exerting anticancer effects through selective arterial occlusion and high-concentration chemotherapy delivery. Recent phase III trials support combining TACE with targeted therapy and immunotherapy for advanced HCC: EMERALD-1 demonstrated statistically significant and clinically meaningful PFS improvement with TACE plus durvalumab and bevacizumab⁴⁸, and LEAP-012 showed similar benefits with TACE plus lenvatinib and pembrolizumab⁴⁹. However, both studies excluded patients with PVTT. TACE application in type III/IV PVTT was previously controversial due to ischemic liver injury concerns. Nevertheless, emerging evidence indicates TACE may be safe and beneficial in carefully selected patients with preserved liver function (Child-Pugh A/B) and adequate collateral circulation^50–53. Recent studies report encouraging outcomes with TACE combined with targeted therapy and immunotherapy in unresectable HCC patients with PVTT: ORRs of 50–70%, DCRs of 80–90%, and median OS of 15–20 months^{21,24–26,30,31,40,41,54–56}.

As an alternative locoregional approach, HAIC has been widely used in Asia and is advocated as the preferred strategy for advanced HCC with PVTT in multiple guidelines^57–59. HAIC delivers high-concentration chemotherapy directly via hepatic artery, potentially achieving superior local control while minimizing systemic toxicity. Several prospective studies have demonstrated impressive outcomes with HAIC-based triple therapy^60–64. In the LTHAIC study, HAIC combined with lenvatinib and toripalimab achieved an ORR of 66.7%, DCR of 88.9%, and median PFS of 10.4 months in high-risk advanced HCC patients, 86.1% of whom had portal vein invasion⁶⁰. The TRIPLET study enrolled BCLC stage C HCC patients, over 70% of whom had PVTT, and demonstrated an ORR of 77.1%, DCR of 97.1%, and median PFS of 10.38 months with the triple combination of camrelizumab, apatinib, and HAIC-FOLFOX⁶². In the Cal Era study, where 86.1% of patients had portal vein invasion, the combination of camrelizumab, lenvatinib, and RALOX-HAIC achieved an ORR of 83.3% with a median PFS of 13.8 months⁶³. For PVTT patients specifically, HAIC-based combinations consistently achieve ORRs of 59–73% and DCRs exceeding 90%^65–67.

Comparative studies have provided evidence of HAIC’s potential superiority over TACE, particularly in patients with high tumor burden and PVTT^68–71. A large retrospective study (n = 748) demonstrated significant advantages of HAIC-based triple therapy over TACE-based triple therapy (ORR: 51.4% vs. 17.5%; PFS: 12.4 vs. 8.2 months; OS: not reached vs. 13.8 months, all p < 0.001)⁷². Similar advantages were observed in recurrent unresectable HCC, where HAIC-based triple therapy showed superior DCR (89.7% vs. 75.0%) and achieved CR in 17.2% of patients compared to none in the TACE-based group⁷³. Consistent findings were observed in PVTT patients, where HAIC-based triple combination therapy significantly outperformed TACE-based treatment, achieving superior ORR (52.9% vs. 27.9%) and prolonged survival (25.0 vs. 19.3 months)⁷⁴. Our meta-analysis corroborates these findings, showing that HAIC-based combinations were most likely to be associated with the longest OS. HAIC’s efficacy may derive from unique pharmacokinetic advantages in PVTT patients. Arterio-portal shunts commonly develop in HCC with PVTT, enabling HAIC to target both primary tumors and PVTT, potentially achieving thrombus regression and recanalization. Based on these advantages, Chinese experts have reached consensus recommending HAIC combined with targeted therapy and immunotherapy for CNLC stage IIIa HCC patients with Vp3/4 PVTT⁷⁵, though individualized selection remains essential.

PVTT appears to demonstrate higher radiosensitivity than primary hepatic lesions, with reported ORRs up to 74% and CR rates of 18.5%^76–79. This characteristic provides rationale for combining radiotherapy with systemic therapies in PVTT patients. An RCT showed significant benefits of adding SBRT to targeted therapy and immunotherapy in PVTT patients (ORR: 47.5% vs. 20.0%, median OS: 12.7 vs. 8.6 months, p < 0.05)²⁹. Multiple studies consistently demonstrate encouraging outcomes with radiotherapy-based triple therapy in HCC-PVTT patients, achieving ORRs of 50–80% and median OS of 10–24 months^45,47,80,81, with superior efficacy versus TACE-based combinations in advanced cases⁸². Our network meta-analysis also confirmed radiotherapy combinations had the highest probability for prolonged PFS. Optimal radiotherapy fractionation requires further investigation. While SBRT theoretically offers advantages through higher fractional doses and reduced normal tissue toxicity, current evidence suggests similar efficacy to IMRT^76,83. Regarding target volume selection, current practice guidelines recommend including both the primary tumor and PVTT when feasible, or PVTT-only when the primary tumor is large or distant from PVTT⁸⁴.

Further research is needed to optimize treatment combinations and sequencing. No consensus exists on optimal agents for triple therapy. While atezolizumab plus bevacizumab demonstrated efficacy in pivotal trials, PD-(L)1 inhibitors plus lenvatinib are more widely adopted due to cost-effectiveness, convenient oral administration, and dosing flexibility. Emerging evidence supports particular advantages of lenvatinib combinations in HCC with PVTT^85,86. A comparative study showed that PD-(L)1 inhibitors plus lenvatinib combined with HAIC provided significant survival benefits compared to atezolizumab plus bevacizumab combined with HAIC in advanced HCC⁸⁷. For patients with PVTT, TACE combined with lenvatinib plus PD-1 inhibitors significantly prolonged OS (21.7 vs. 15.6 months) and PFS (6.3 vs. 3.2 months) compared to TACE combined with sorafenib plus PD-1 inhibitors⁵⁶, underscoring the clinical relevance of TKI selection in combination regimens. Additionally, the STRIDE regimen (tremelimumab plus durvalumab) from the HIMALAYA trial also represents a promising option for combination approaches^88,89. Sequencing optimization represents another critical consideration. Limited evidence suggests timing of systemic therapy relative to locoregional interventions may influence outcomes. For instance, patients receiving TKIs plus ICIs after TACE achieved longer OS than those treated before TACE⁹⁰, while radiotherapy prior to TACE may be superior in HCC with PVTT⁹¹. Currently, the ideal combination regimen and treatment sequence remain unknown, with ongoing trials investigating integrated approaches. (Table 3)

Table 3.

The ongoing trials of locoregional therapy combined with targeted therapy and immunotherapy for HCC with PVTT.

Identifier	Study type	Participants	Enrollment	Interventions	Primary endpoints	Secondary endpoints
TACE + targeted therapy + immunotherapy
ChiCTR2300073121	Comparative non-RCT	BCLC C HCC with PVTT	78	TACE + Lenvatinib + Camrelizumab	ORR	OS, PFS
				TACE + Lenvatinib
ChiCTR2200066830	Phase III RCT	Unresectable HCC with PVTT	218	TACE + Lenvatinib + Sintilimab	TTP	OS, ORR, AEs
				TACE + Lenvatinib
HAIC + targeted therapy + immunotherapy
NCT05198609	Phase III RCT	Unresectable HCC with PVTT	214	HAIC + Apatinib + Camrelizumab	OS	PFS, TTP, DOR, ORR, DCR
				Apatinib + Camrelizumab
NCT05166239	Phase II Non-RCT	BCLC C HCC with PVTT	66	HAIC + Lenvatinib + PD-1	6 m-PFS	OS, PFS, TTP, AEs
				Lenvatinib + PD-1
NCT04947826	Phase II double-blinded RCT	BCLC C HCC with Vp3-4 PVTT	100	HAIC + HLX04 (VEGF Antibody) + HLX10 (PD-1 Antibody)	ORR	PFS, DOR, TTP, OS, AEs
				HAIC + Placebo
NCT04618367	Single-arm	HCC with Vp1-3 PVTT	30	HAIC + Lenvatinib + Sintilimab	6 m-PFS	OS, PFS, ORR, AEs
NCT06210334 (HAI-TL)	Single-arm Phase II	HCC with Vp4 PVTT	54	HAIC + Lenvatinib + Tislelizumab	OS	PFS, TTP, ORR, DCR, AEs
ChiCTR2200063129	Single-arm Phase II	BCLC C HCC with PVTT	70	HAIC + Lenvatinib + Camrelizumab
ChiCTR2100051717	Single-arm Phase II	BCLC C HCC with PVTT	29	HAIC + Lenvatinib + Tislelizumab	ORR	PFS, OS, DOR, AEs
ChiCTR2100051714	Single-arm Phase II	HCC with Vp4 PVTT	32	HAIC + Donafenib + Camrelizumab	ORR	CRR, DCR, PFS, OS, DOR, AEs
ChiCTR2200062213	Single-arm	HCC with Vp4 PVTT	34	HAIC + Apatinib + Camrelizumab	PFS	ORR, DOR, OS, AEs, QoL
Radiotherapy + targeted therapy + immunotherapy
NCT05339581 (iPLENTY-pvtt)	Perspective Non-RCT	HCC with Vp3 PVTT	78	RT + Lenvatinib + PD-1	PVTT related Response and Necrosis Rate	Alpha Fetoprotein Response, PFS, ORR, TTP, DOR
				Lenvatinib + PD-1
ChiCTR2300075975	Phase III RCT	Unresectable HCC with PVTT	156	RT + Atezolizumab + Bevacizumab	OS	PFS, ORR, DCR, DOR, TTP, CRR, QoL, AEs
				Atezolizumab + Bevacizumab
NCT05992220 (ALERT-HCC)	Phase II RCT	HCC with intrahepatic macrovascular invasion	138	RT + Atezolizumab + Bevacizumab	PFS	OS, ORR, AEs, DOR, Tumor marker response
				Atezolizumab + Bevacizumab
NCT05625893 (PORTAL)	Single-arm Phase II	HCC with Vp2-4 PVTT	63	Proton Beam RT + Atezolizumab + Bevacizumab	PFS, AEs	OS, TTP, ORR, DCR
NCT06061445	Single-arm	Stage IIIA HCC with Vp1-3 PVTT	22	RT + TKI + PD-1	OS, ORR	/
NCT05286320	Single-arm Phase Ib/II	HCC with Vp3-4 PVTT	27	SBRT + Lenvatinib + Pembrolizumab	Safety, ORR	PFS, OS
NCT05225116	Single-arm	BCLC C HCC with Vp1-4 PVTT	20	Neoadjuvant RT + Lenvatinib + Sintilimab	Safety	ORR, Imaging-pathology Concordance Rate, PVTT regression rate, OS, RFS
NCT06040177	Single-arm Phase I/II	BCLC C HCC with PVTT	30	SBRT + Lenvatinib + Cardonilizumab	ORR	PFS, DCR, DOR, OS
NCT05530785	Single-arm Phase II	BCLC C HCC with Vp1-3 PVTT	35	RT + Sintilimab + Bevacizumab Biosimilar	ORR	OS, AEs
NCT05908916	Single-arm Phase II	HCC with Vp2-4 PVTT	51	RT + Atezolizumab + Bevacizumab	PFS	/
ChiCTR2100049831	Single-arm Phase II	Unresectable HCC with PVTT	42	RT + Atezolizumab + Bevacizumab	ORR	OS, DCR, PFS, TTP, DOR, CRR, AEs
ChiCTR2300072239	Single-arm	Unresectable HCC with PVTT	30	SBRT + Sintilimab + Bevacizumab	ORR	AEs, OS, DCR, PFS, CRR
ChiCTR2100051537	Single-arm Phase I/II	BCLC C HCC with Type I-III PVTT	43	RT + Donafenib + Sintilimab	PFS	ORR, OS, TTP, AEs
ChiCTR2000040814	Single arm	Unresectable HCC with PVTT	30	RT + Apatinib + Camrelizumab	PFS	ORR, DCR, OS
NCT04649489 (TALENTop)	Phase III RCT	HCC with PVTT/HVTT/IVCTT without extrahepatic spread	456	Induction Atezolizumab + Bevacizumab → Surgery → post-operative Atezolizumab + Bevacizumab	TTF	OS, Time to extrahepatic spread, AEs, RFS, R0 rate, pCR, ORR
				Atezolizumab + Bevacizumab
ChiCTR2300072234	Phase III RCT	Resectable HCC with type II/III PVTT	80	Neoadjuvant RT + Atezolizumab + Bevacizumab → surgery	OS	DFS
				Neoadjuvant RT → surgery
NCT06349317	Single-arm Phase II	Resectable HCC With Vp1-3 PVTT	33	Neoadjuvant RT + Apatinib + Camrelizumab → surgery	1y-EFS	OS, DFS, R0 resection rate, ORR, AEs
NCT06524466	Single-arm Phase II	Resectable HCC with VP1-3 PVTT	35	Neoadjuvant SBRT + Lenvatinib + Pucotenlimab → surgery	ORR, treatment complete rate	PFS, OS, AEs, pCR
Dual locoregional therapy + targeted therapy + immunotherapy
NCT05984511	Phase III RCT	Advanced HCC and Type I/II PVTT	234	TACE + Atezolizumab + Bevacizumab + I-125 Brachytherapy	OS	PFS, ORR, DOR, AEs
				TACE + Atezolizumab + Bevacizumab
NCT06031285	Single-arm Phase II	BCLC C HCC with Vp1-3 PVTT	43	TACE-HAIC + Sinitilimab + Bevacizumab Biosimilar	RFS	/
ChiCTR2300072570	Single-arm Phase II	BCLC C HCC with Vp4 PVTT	70	TACE + Lenvatinib + Sintilimab + SBRT	OS	PFS, ORR, DCR, AEs
ChiCTR2200056181	Single-arm	BCLC C HCC with Vp4 PVTT	38	RT + HAIC + Lenvatinib + Tislelizumab	ORR, PFS	OS, DCR, CRR
ChiCTR2200066334	Single-arm Phase II	HCC with Vp4 PVTT	20	HAIC + SBRT + Lenvatinib + Toripalimab	6 m-OS	PFS, ORR, DOR, DCR, OS, AEs
ChiCTR2200058539	Single-arm Phase II	Resectable HCC with Type I-III PVTT	53	TACE + RT + Apatinib + Camrelizumab	ORR	OS, PFS, CRR, R0 rate, pCR, DCR, AEs
ChiCTR2400080802	Single-arm Phase II	HCC with Vp3-4 PVTT	40	HAIC + SBRT + TKI + PD-1	ORR	AEs, OS, PFS, DCR, TTP
ChiCTR2300078562	Single-arm Prospective	BCLC C HCC with PVTT	50	HAIC + SBRT + Apatinib + Camrelizumab	PFS	OS, ORR, DCR
ChiCTR2400082386	Single-arm Phase II	Unresectable HCC with PVTT	64	TACE + SBRT + TKI + Sintilimab	PFS	OS, ORR, AEs, DCR

HCC, hepatocellular carcinoma; PVTT, portal vein tumor thrombus; RCT, randomized controlled trial; TACE, transarterial chemoembolization; HAIC, hepatic arterial infusion chemotherapy; PFS, progression-free survival; OS, overall survival; RFS, recurrence-free survival; DFS, disease-free survival; AEs, adverse events; ORR, objective response rate; DCR, disease control rate; DOR, duration of response; TTP, time to progression; BCLC, Barcelona clinic liver cancer stage; pCR, pathologic complete response; CRR, Conversion resection rate; QoL, quality of life; ICI, immunotherapy checkpoint inhibitor; TKI, tyrosine kinase inhibitor (e.g., lenvatinib, sorafenib, apatinib, regorafenib); PD-1, programmed death-1 inhibitor; VEGF, vascular endothelial growth factor; RT, radiotherapy; SBRT, stereotactic body radiotherapy; HVTT, hepatic vein tumor thrombosis; IVCTT, inferior vena cava tumor thrombosis.

Although triple therapy demonstrates superior efficacy, its safety profile requires careful evaluation. Our meta-analysis showed significantly higher grade 3/4 TRAEs with triple therapy versus targeted therapy plus immunotherapy alone (73.5% vs. 39.4%, p < 0.001). The enhanced toxicity manifests as increased frequency and severity. HAIC-based combinations show elevated AE rates, with particular concern for gastrointestinal bleeding (7.7% incidence), likely resulting from synergistic mucosal damage caused by immunotherapy and anti-angiogenic agents⁹². TACE combined with TKI and ICI demonstrated higher AE rates than systemic TKI plus ICI, primarily from TACE-related complications including liver function injury, pain and nausea/vomiting^93–95. Despite the increased toxicity, most AEs were manageable with appropriate clinical interventions, with no treatment-related deaths reported^93–95. Comprehensive management strategies are essential, encompassing rigorous baseline assessment, robust monitoring systems, and dynamic treatment adjustments.

Triple therapy achieves synergistic benefits through complementary mechanisms. Locoregional interventions rapidly reduce tumor burden while upregulating molecular targets such as vascular endothelial growth factor (VEGF) and epidermal growth factor receptor (EGFR), creating optimal conditions for targeting by anti-angiogenic agents and TKIs. Conversely, targeted therapies enhance the efficacy of locoregional interventions by normalizing tumor vasculature and improving vascular permeability^96–98. These interventions also promote anti-tumor immunity by releasing tumor-associated antigens, augmenting CD8⁺ T-cell responses, diminishing regulatory T-cell functions, and upregulating PD-1/PD-L1 expression, thus transforming the immunosuppressive microenvironment into an immunosupportive one^99–102. Different locoregional therapies offer complementary advantages for specific populations. Emerging evidence supports the clinical potential of intensive multimodal approaches that combine multiple locoregional treatments with targeted therapy and immunotherapy in advanced HCC with PVTT^{23,28,39,103–108}, a promising therapeutic paradigm that warrants further validation in prospective trials.

There are some limitations to this study. First, most included studies were retrospective with relatively small sample sizes, potentially introducing selection bias despite the use of PSM methodology. Second, all studies originated from China, which may limit the generalizability of findings to other populations. This geographic restriction reflects epidemiological differences (higher HCC with PVTT prevalence in Asia due to hepatitis B burden) and treatment paradigm differences (aggressive locoregional approaches in Asia versus systemic therapy preference in Western countries). Third, some survival data were obtained through secondary analysis rather than original sources, potentially introducing additional bias. Methodological limitations include heterogeneity in some paired comparisons and insufficient data for more extensive subgroup analyses. While SUCRA rankings provided treatment hierarchies, inherent methodological limitations require cautious interpretation.

Conclusions

The management of HCC with PVTT remains challenging and controversial. Our meta-analysis indicates that adding locoregional therapy to targeted therapy plus immunotherapy yields superior tumor response and survival compared with targeted therapy plus immunotherapy alone, with manageable toxicity. Notably, among triple therapy combinations, HAIC‑based regimens were associated with superior OS, and radiotherapy‑based regimens with superior PFS, compared with TACE‑based regimens. These findings support a paradigm shift toward multimodal treatment strategies for this complex condition. International multicenter RCTs are needed to establish optimal treatment regimens and sequencing strategies, refine patient selection, and standardize toxicity management. Future research should prioritize investigating predictive biomarkers, molecular mechanisms, and personalized treatment algorithms through international collaboration to validate and optimize combination therapy strategies.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

M.J. and C.C. developed and conceived the manuscript. M.J., C.C., Y.H. and G.L. extracted and analyzed data. C.C., G.L. and H.L. evaluated the data quality. M.J., C.C. and H.L. interpreted the data and prepared figures&tables. M.J. and C.C. drafted the manuscript. All authors reviewed and agreed to the published version of the manuscript.

Funding

National Key R&D Program of China (2023YFC2413900). Medical Health Science and Technology Project of Zhejiang Provincial Health Commission, No. 2025KY092. Zhejiang TCM Science and Technology Project, No. 2025ZR027.

Data availability

All data generated or analyzed during the present study are included in this published article.

Declarations

Competing interests

The authors declare no competing interests.