Systemic treatment patterns and clinical outcomes in hepatocellular carcinoma: A multicenter real-world study from China (STAR-HCC)

Abstract

Background & Aims

Systemic therapy for hepatocellular carcinoma (HCC) has evolved rapidly with the introduction of tyrosine kinase inhibitors (TKIs), immune checkpoint inhibitors (ICIs), anti-vascular endothelial growth factor (anti-VEGF) antibodies, and their combinations. This multicenter study aimed to characterize real-world systemic treatment patterns and clinical outcomes in treatment-naïve patients with HCC in China.

Methods

Patients initiating systemic therapy for HCC between 2017 and 2023 across five centers were retrospectively enrolled. Four systemic treatment categories were analyzed: TKI monotherapy, ICIs monotherapy, TKIs plus ICIs, and anti-VEGF plus ICIs. Baseline clinical characteristics, treatment patterns across lines, survival outcomes, tumor response, and adverse events (AEs) were analyzed.

Results

A total of 4,250 patients were included (median age, 53 years), predominantly male (88.7%) with hepatitis B virus infection (86.3%); 81.3% had Barcelona Clinic Liver Cancer stage C disease. In the first-line setting, 92.3% of patients received systemic therapy combined with transarterial therapy. Median overall survival was 20.7 months (95% CI, 19.1–22.3), and median progression-free survival was 9.7 months (95% CI, 8.2–11.2). Baseline characteristics differed across systemic treatment groups, including demographic and disease characteristics and treatment patterns involving transarterial therapy, and were accounted for using inverse probability of treatment weighting and multivariable Cox regression. Median overall survival was 17.0, 21.4, 21.6, and 22.4 months in the TKI monotherapy, ICIs monotherapy, TKI plus ICIs, and anti-VEGF plus ICIs groups, respectively. Grade ≥3 AEs occurred in 28.3% of patients.

Conclusions

In this multicenter cohort, ICIs-based systemic regimens combined with transarterial therapy were frequently applied in patients with HCC and associated with favorable survival outcomes and acceptable safety profiles. These findings provide contemporary real-world insights into systemic treatment patterns and outcomes.

Impact and implications

The therapeutic landscape of HCC has evolved rapidly, whereas treatment patterns in routine clinical practice may differ substantially from those represented in randomized clinical trials, particularly in regions with a high burden of hepatitis B-related disease. This large multicenter real-world study delineates contemporary systemic treatment strategies in China and describes clinical outcomes associated with the use of immune checkpoint inhibitor–containing regimens, particularly when integrated with transarterial therapy, in routine clinical practice. These findings provide additional real-world context regarding multidisciplinary treatment approaches for HCC and may help inform future prospective evaluation of systemic and locoregional treatment integration. Although limited by its observational design, this study generates hypotheses for future prospective studies evaluating multimodal treatment strategies and patient selection for immunotherapy-based approaches in HCC.

Graphical abstract

Highlights

• •Systemic therapy combined with transarterial therapy (TAT-Sys) predominated in this real-world cohort.
• •TKI plus ICIs combinations were the most commonly used regimens across treatment lines.
• •ICIs-based regimens and TAT-integrated strategies were associated with favorable real-world survival outcomes.
• •Severe adverse events remained manageable in this TAT-Sys–predominant cohort.

Introduction

Liver cancer is the sixth most common malignancy worldwide and remains associated with high mortality and poor prognosis.¹ Hepatocellular carcinoma (HCC), the predominant type of liver cancer, is closely associated with chronic HBV infection, particularly in the Asia-Pacific region.² More than 50% of patients with HCC are diagnosed at an advanced stage and are therefore ineligible for curative treatments.³ Systemic therapy has become the standard treatment for late-stage HCC, as endorsed by major guidelines, including those of the Barcelona Clinic Liver Cancer (BCLC) group, the American Society of Clinical Oncology, and the China Liver Cancer guidelines.[4], [5], [6]

Systemic therapies for HCC, including tyrosine kinase inhibitors (TKIs), immune checkpoint inhibitors (ICIs), anti-vascular endothelial growth factor (anti-VEGF) monoclonal antibodies, and their combinations, have advanced rapidly worldwide in recent years. Following the introduction of sorafenib, multiple first- and second-line systemic regimens have demonstrated clinical benefit in advanced HCC.[7], [8], [9], [10], [11], [12], [13], [14], [15] Despite advances in systemic therapy, the integration of transarterial therapy (TAT) with systemic agents (TAT-Sys) has been increasingly explored as a strategy to improve tumor control in HCC.^16,17 Phase III trials, including TALENTACE, EMERALD-1, and EMERALD-3, have evaluated these multimodal strategies and reported encouraging activity in unresectable HCC.[18], [19], [20] Notably, although the phase III LEAP-012 trial improved progression-free survival (PFS), no overall survival (OS) benefit was observed according to recent updates, and further development of this strategy has been discontinued.²¹ Taken together, the clinical role of different TAT-Sys strategies remains to be fully defined despite their increasing incorporation into real-world clinical practice.²²

The real-world treatment landscape of HCC in China may differ substantially from that of many Western countries and some other Asian regions.[23], [24], [25] In China, HBV infection remains the predominant underlying etiology of HCC, and patients frequently present with high intrahepatic tumor burden despite preserved liver function, contributing to the frequent use of TAT in multidisciplinary treatment strategies.[26], [27], [28] In addition, differences in drug approval timelines, reimbursement policies, multidisciplinary treatment practices, and the extensive use of TAT have contributed to distinct treatment strategies in routine clinical practice. In particular, TAT-Sys is commonly adopted for patients with intermediate- to advanced-stage HCC in China. In this treatment framework, systemic therapy generally serves as the foundation of treatment, and TAT is selectively incorporated for intrahepatic tumor control, whereas such approaches are less routinely used in many Western practice settings.

Given the increasing complexity of systemic and multimodal treatment strategies, real-world treatment patterns for HCC have become increasingly heterogeneous. Treatment selection is influenced by multiple factors, including clinical guidelines, drug approval status, insurance coverage, integration with TAT, and individual patient characteristics. Therefore, in this large-scale multicenter study, we aimed to characterize real-world treatment patterns, clinical outcomes, and safety related to systemic therapy–based strategies in treatment-naïve patients with HCC in China.

Materials and methods

Study population

Patients with HCC were retrospectively identified from five tertiary referral hospitals in China, four of which were university-affiliated academic medical centers, between 1 January 2017 and 31 December 2023. The year 2017 was selected as the starting point because it marked the beginning of the transition from sorafenib as the only widely available systemic therapy for unresectable HCC. Inclusion criteria were as follows: (1) age ≥18 years; (2) histologically or clinically confirmed HCC, as defined by the diagnostic criteria of EASL and AASLD^29,30; (3) receipt of first-line systemic therapy for HCC without prior anticancer treatment or liver transplantation; and (4) availability of complete baseline clinical documentation and follow-up records after treatment initiation. The index date was defined as the date of treatment initiation. Patients who did not meet the inclusion criteria were excluded. This study was approved by the Institutional Review Board and Institutional Ethics Committee of Sun Yat-sen University Cancer Center (Guangzhou, China; Approval No. B2025-237-01) and was conducted in accordance with the Declaration of Helsinki and the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. The requirement for informed consent was waived by the ethics committee because of the retrospective design and use of anonymized data.

Data collection

Patient data were retrospectively extracted from electronic health records across participating centers. The following variables were collected: demographic characteristics (age, sex, etiology); performance status (Eastern Cooperative Oncology Group-performance status [ECOG-PS]); liver function parameters (Child–Pugh class and score, albumin–bilirubin [ALBI] grade); tumor characteristics (BCLC stage, serum alpha-fetoprotein [AFP], serum protein induced by vitamin K absence or antagonist-II [PIVKA-II], tumor size, number of tumors, and presence of macrovascular invasion or extrahepatic spread); systemic treatment regimens across first- to third-line therapy; combined TAT (transarterial chemoembolization [TACE], hepatic arterial infusion chemotherapy [HAIC], or HAIC plus TACE); and subsequent anticancer treatments, including resection, ablation, and radiotherapy. Treatment outcomes, including toxicity, tumor response, and survival, were retrospectively collected from inpatient records, outpatient visits, laboratory examinations, and follow-up documentation (including telephone-based follow-up when available).

Treatment, follow-up, and outcomes

Treatment decisions, including the use of TAT, systemic drug therapy, and combination regimens, were determined through multidisciplinary team (MDT) discussions in routine clinical practice across participating centers. Detailed information on treatment categorization, including systemic therapy regimens and TAT procedures, is provided in the Supplementary Methods. All patients were followed up at intervals of 2–3 months. At each scheduled visit, a physical examination, blood tests (including serum tumor markers and liver function tests), and contrast-enhanced abdominal computed tomography or magnetic resonance imaging scans were performed. Survival status was verified using the National Population Death Register Information System. Tumor response was assessed according to the Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST v.1.1) and its modified version (mRECIST v1.1)³¹, including complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD). The objective response rate (ORR) was defined as the proportion of patients achieving CR and PR, while the disease control rate was defined as the proportion of patients achieving CR, PR, or SD. Safety was evaluated throughout the treatment period, and adverse events (AEs) were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 5.0. The clinical pathway of patients with HCC from initial diagnosis through follow-up is illustrated in Fig. S1.

The primary outcomes were OS and PFS in the overall cohort and according to treatment patterns. OS was defined as the time from treatment initiation to death from any cause or the end of follow-up (last follow-up date: 1 March 2025). PFS was defined as the time from treatment initiation to the first documented disease progression or death from any cause, whichever occurred first. Secondary endpoints included first-line systemic treatment patterns, distributions of second- and third-line systemic regimens, safety, and clinical characteristics associated with first-line survival. Additional analyses included comparisons of survival, tumor response, and safety across different systemic treatment patterns and regimens.

Statistics analysis

Categorical variables were summarized as frequencies and percentages. Group comparisons were performed using the χ² test or Fisher’s exact test, as appropriate. Survival probabilities were estimated using the Kaplan–Meier method and compared with the log-rank test. Median survival times and corresponding 95% CIs were calculated. Univariable and multivariable Cox proportional hazards models were used to identify prognostic factors associated with OS and PFS, with results presented as HRs and 95% CIs.

Propensity score methods, including inverse probability of treatment weighting (IPTW) and propensity score matching (PSM), were applied to adjust for potential confounding factors. In the IPTW analysis, stabilized weights were calculated based on predicted probabilities derived from the propensity score model. In the PSM analysis, matched control groups were constructed using nearest-neighbor matching. Associations between the TAT-Sys and systemic therapy alone (Sys-only) groups were evaluated using multivariable Cox regression models with additional IPTW- and PSM-adjusted sensitivity analyses.

A random survival forest (RSF) model was used to evaluate the relative importance of prognostic variables for survival outcomes. Variable importance was assessed using permutation importance.

All statistical analyses were performed using R software (version 4.2.1, R Foundation for Statistical Computing, Vienna, Austria), and figures were generated with GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA) and OriginPro 2021.

Results

Patient characteristics

Between January 2017 and December 2023, a total of 13,602 hospitalized patients with systemic therapy records were screened across five medical centers. Among them, 4,250 patients who met the inclusion criteria were enrolled in the study (Fig. S2). The median follow-up duration was 17.1 months (mean: 20.2 months). The median age of the cohort was 53 years (range, 18–89 years). Most patients were male (88.7%; n = 3,771) and had HBV infection (86.3%; n = 3,668). At baseline, the majority had BCLC stage C disease (81.3%; n = 3,453), Child–Pugh class A liver function (81.9%; n = 3,481), ALBI grade 1 or 2 (96.8%; n = 4,114), and ECOG-PS of 0 or 1 (92.5%; n = 3,931). Serum AFP levels ≥400 ng/ml were reported in 36.7% (n = 1,561), and serum PIVKA-II levels ≥40 mAU/ml were observed in 69.9% (n = 2,970). Tumor size ≥100 mm was present in 45.0% (n = 1,911), multiple lesions in 68.9% (n = 2,930), macrovascular invasion in 53.2% (n = 2,261), extrahepatic spread in 33.9% (n = 1,441), and concurrent macrovascular invasion and extrahepatic spread in 20.0% (n = 851) (Table 1). Over time, the cohort shifted toward older age, whereas the distributions of Child–Pugh class and BCLC stage remained largely stable, with BCLC stage C consistently predominating (Fig. S3).

Table 1.

Patient demographic and clinical characteristics.

Characteristics	Total (N = 4,250)
Age (years)
Median (range)	53.0 (18.0–89.0)
<55	2,285 (53.8)
≥55	1,965 (46.2)
Sex, n (%)
Male	3,771 (88.7)
Female	479 (11.3)
Follow-up duration (months)
Median	17.1
Mean	20.2
Hepatitis B virus infection, n (%)
Yes	3,668 (86.3)
No	582 (13.7)
ECOG-PS, n (%)
0	2,096 (49.3)
1	1,835 (43.2)
2	294 (6.9)
3-4	25 (0.6)
Child–Pugh class and score, n (%)
Child–Pugh A
CPA5	2,163 (50.9)
CPA6	1,318 (31.0)
Child–Pugh B
CPB7	508 (12.0)
CPB8	185 (4.4)
CPB9	56 (1.3)
Child–Pugh C
CPC ≥10	20 (0.3)
ALBI grade, n (%)
1	1,555 (36.6)
2	2,559 (60.2)
3	136 (3.2)
BCLC stage, n (%)
A	341 (8.2)
B	421 (9.9)
C	3,453 (81.3)
D	29 (0.7)
Serum AFP, n (%)
<400 (ng/ml)	2,689 (63.3)
≥400 (ng/ml)	1,561 (36.7)
Serum PIVKA-II, n (%)
<40 (mAU/ml)	1,280 (30.1)
≥40 (mAU/ml)	2,970 (69.9)
Tumor size (mm)
<100	2,339 (55.0)
≥100	1,911 (45.0)
Number of tumors, n (%)
Single	1,320 (31.1)
Multiple	2,930 (68.9)
Presence of macrovascular invasion/extrahepatic spread, n (%)
Macrovascular invasion	2,261 (53.2)
Extrahepatic spread	1,441 (33.9)
Both	851 (20.0)
Combined with transarterial therapy, n (%)
None	329 (7.7)
HAIC	1,882 (44.3)
TACE	895 (21.1)
HAIC plus TACE	1,144 (26.9)

AFP, alpha-fetoprotein; ALBI, albumin–bilirubin; BCLC, Barcelona Clinic Liver Cancer; ECOG-PS, Eastern Cooperative Oncology Group-performance status; HAIC, hepatic arterial infusion chemotherapy; HCC, hepatocellular carcinoma; PIVKA-II, protein induced by vitamin K absence or antagonist-II; SE, standard error; TACE, transarterial chemoembolization.

Systemic treatment patterns

Systemic treatment patterns were analyzed according to systemic regimens, use of combined TAT, and treatment line. Systemic agents were categorized into the following classes: TKI, ICIs, and anti-VEGF monoclonal antibodies. Four systemic treatment categories were identified: TKI monotherapy (TKI), ICIs monotherapy (ICIs), TKI plus ICIs combination (TKI + ICIs), and anti-VEGF antibody plus ICIs combination (VEGF + ICIs). In the first-line setting, 7.7% of patients received Sys-only, whereas the remaining patients received TAT-Sys, particularly TACE (21.1%), HAIC (44.3%), or HAIC plus TACE (26.9%) (Table 1). The proportion of HAIC-based treatment patterns increased over time, with systemic therapy combined with HAIC plus TACE rising from none in 2017 to approximately 30% in 2023 (Fig. S3D).

First-line treatment strategies varied according to systemic therapy regimen, and the distribution of TAT modalities among patients receiving each systemic regimen is presented in Fig. S4. The overall treatment sequence across therapy lines is illustrated in Fig. 1A. The TKI + ICIs regimen was the most frequently administered systemic therapy across first-line (2,822/4,250), second-line (665/1,060), and third-line (162/239) therapies. Most patients did not proceed to second-line (3,190/4,250) or third-line (821/1,060) systemic therapy. The detailed distribution of specific systemic therapy regimens by treatment line is summarized in Table S2. Lenvatinib-based combinations with ICIs were the predominant first-line regimens. In subsequent lines of treatment, lenvatinib-based regimens also accounted for a substantial proportion of treatment strategies. Regorafenib-based regimens were also frequently administered in both second- and third-line settings. Treatment strategies evolved across therapy lines, with a shift from TKI monotherapy toward ICIs-based regimens. After 2022, the VEGF + ICIs regimen accounted for a substantial proportion of first-line treatments (Fig. 1B).

Fig. 1.

Regimens distribution (A) and monthly numbers of patients receiving first-line (B), second-line (C), and third-line (D) systemic therapy. Regimens are shown in blue, pink, red, and yellow for tyrosine kinase inhibitor monotherapy (TKI), immune checkpoints inhibitor monotherapy (ICIs), TKI plus ICIs combinations (TKI + ICIs), and anti-VEGF antibody plus ICIs combinations (VEGF + ICIs), respectively. VEGF, vascular endothelial growth factor.

Survival outcomes

The median OS in the entire cohort was 20.7 months (95% CI: 19.1–22.3). By the final follow-up, 2,527 patients (59.5%) had died. The 12-, 24-, and 36-month OS rates were 69.8% (95% CI: 68.4–71.2), 45.4% (95% CI: 43.8–47.0), and 34.0% (95% CI: 32.3–35.7), respectively. Disease progression was documented in 3,225 patients (75.9%). The median PFS was 9.7 months (95% CI: 8.2–11.2), with 6- and 12-month PFS rates of 65.3% (95% CI: 63.9–66.7) and 43.2% (95% CI: 41.6–44.7), respectively (Fig. 2A,B).

Fig. 2.

Kaplan–Meier survival curves for overall survival (OS) and progression-free survival (PFS) in the study cohort. (A,B) OS and PFS in the overall cohort. (C,D) OS and PFS according to treatment strategy (TAT-Sys vs. Sys-only). (E,F) OS and PFS according to systemic treatment regimens. Survival curves were estimated using the Kaplan–Meier method and compared using the log-rank test. ICIs, immune checkpoint inhibitors; TAT-Sys, transarterial therapy combined with systemic therapy; TKI, tyrosine kinase inhibitor; VEGF, vascular endothelial growth factor.

Patients receiving first-line TAT-Sys strategies had longer OS (21.2 [20.3–22.3] months vs. 16.1 [13.6–19.4] months, p <0.001, HR = 0.76 [95% CI: 0.66–0.87]) and PFS (10.0 [9.6–10.6] months vs. 7.2 [6.2–8.6] months, p <0.001, HR = 0.78 [95% CI: 0.69–0.89]) than those receiving Sys-only therapy (Fig. 2C,D). Subgroup analyses showed generally consistent survival trends across most subgroups, with numerically greater survival differences observed among patients receiving combination systemic regimens, particularly TKI + ICIs or VEGF + ICIs (Fig. S5). With respect to systemic treatment regimens, median OS and PFS were 17.0 (95% CI 15.9–19.0) and 5.7 (95% CI 5.0–6.6) months with TKI monotherapy; 21.4 (18.4–25.4) and 7.8 (6.4–9.2) months with ICIs; 21.6 (20.6–22.9) and 10.6 (10.2–11.2) months with TKI + ICIs; and 22.4 (19.0–29.6) and 11.9 (10.5–14.3) months with VEGF + ICIs, respectively (Fig. 2E,F). These findings remained consistent after IPTW adjustment (Fig. S6). Clinical outcomes according to specific first-line ICIs-based combination regimens are summarized in Table S5. Among the commonly used regimens, bevacizumab plus atezolizumab, bevacizumab plus sintilimab, rivoceranib plus camrelizumab, and lenvatinib plus ICIs combinations showed numerically favorable survival outcomes in this real-world cohort.

A survival pathway analysis demonstrated distinct survival distributions across hierarchical treatment subgroups stratified by BCLC stage, TAT patterns, and systemic treatment regimens. Patients receiving ICIs-based regimens alongside HAIC-based TAT tended to show longer median OS across several treatment subgroups in both BCLC stage A and stage C disease (Fig. 3). When stratified by line of systemic therapy, patients who failed first-line TKI monotherapy demonstrated numerically longer median OS when treated with either second-line TKI monotherapy or TKI plus ICI combination therapy (Fig. S7). In addition, patients treated during 2020–2023 showed longer OS than those treated during 2017–2019 (Fig. S8), consistent with the temporal evolution of treatment patterns within the cohort. Survival analyses further demonstrated progressively poorer OS among patients with impaired baseline liver function according to Child–Pugh score and ALBI grade stratification (Fig. S9).

Fig. 3.

Survival pathway analysis of first-line treatment patterns in patients with HCC stratified by BCLC stage. The figure presents median overall survival (OS) across hierarchical treatment subgroups defined by BCLC stage, transarterial therapy patterns, and systemic treatment regimens. Subgroups with <10 cases were excluded from the analysis. Median OS was estimated using the Kaplan–Meier method. BCLC, Barcelona Clinic Liver Cancer; HAIC, hepatic arterial infusion chemotherapy; HCC, hepatocellular carcinoma; ICIs, immune checkpoint inhibitors; TACE, transarterial chemoembolization; TKI, tyrosine kinase inhibitor; VEGF, vascular endothelial growth factor.

Progression and response evaluation

During post-baseline assessments, 47.1% (2,000/4,250) of patients developed either new intrahepatic lesions or extrahepatic disease progression. In the TAT-Sys group, new intrahepatic lesions developed in 33.7% (1,321/3,921) of patients and extrahepatic progression occurred in 30.1% (1,181/3,921). By contrast, the corresponding rates were 42.9% (141/329; p = 0.001) and 36.8% (121/329; p = 0.014) in patients receiving Sys monotherapy. According to RECIST v1.1, disease progression occurred in 53.7% (577/1,075) of patients with enlargement of target lesions, compared with 44.8% (1,423/3,175) of those showing reduction in lesion size. Similar findings were observed using mRECIST v1.1 criteria (Fig. S10). The TAT-Sys group demonstrated higher ORRs than the Sys-only group both before and after IPTW adjustment (RECIST v1.1: 32.6% vs. 17.6% pre-IPTW and 32.3% vs. 21.2% post-IPTW; mRECIST v1.1: 50.9% vs. 29.8% pre-IPTW and 50.6% vs. 34.3% post-IPTW) (Table S6). Subsequent surgical resection after initial treatment was more frequently observed in the TAT-Sys group than in the Sys-only group (15.3% [600/3,921] vs. 3.0% [10/329], p <0.001). Clinical response according to systemic regimen type is summarized in Table S7.

Prognostic factors associated with survival

In the RSF model, subsequent therapy, tumor response according to RECIST/mRECIST v1.1, Child–Pugh class, and BCLC stage were identified as major variables associated with OS and PFS (Fig. S11). Univariable and multivariable Cox regression analyses of first-line OS and PFS were subsequently performed for the entire cohort (Fig. S12 and Table 2). Patients with lower ECOG-PS scores and preserved liver function (Child–Pugh class A and ALBI grade 1) demonstrated improved survival. By contrast, elevated serum AFP levels (≥400 ng/ml), tumor size ≥10 cm, multiple lesions, macrovascular invasion, and extrahepatic spread were associated with significantly poorer OS and PFS. These findings were consistent with established adverse prognostic factors in HCC. Regarding treatment patterns, strategies incorporating TAT (HAIC, TACE, or their combination) were associated with longer OS, while HAIC-based regimens were additionally associated with prolonged PFS. Among systemic therapies, ICIs-based regimens (including ICIs monotherapy, TKI + ICIs combinations, and anti-VEGF + ICIs combinations) were associated with longer OS and PFS than TKI monotherapy.

Table 2.

Multivariable Cox regression analysis of first-line overall survival and progression-free survival.

Variable	N	OS		PFS
		HR (95% CI)	p values	HR (95% CI)	p values
Sex
Female	479	Ref.		Ref.
Male	3,771	1.09 (0.96–1.23)	0.208	1.00 (0.90–1.12)	0.966
Age
<55	2,285	Ref.		Ref.
≥55	1,965	1.00 (0.92–1.09)	0.961	0.90 (0.84–0.97)	0.005
HBV infection
No	582	Ref.		Ref.
Yes	3,668	0.94 (0.84–1.05)	0.278	0.99 (0.89–1.10)	0.836
ECOG-PS
0	2,096	Ref.		Ref.
1	1,835	1.00 (0.91–1.09)	0.932	0.97 (0.89–1.05)	0.403
2	294	1.24 (1.02–1.51)	0.027	1.02 (0.85–1.22)	0.824
3	25	1.78 (1.03–3.09)	0.039	1.06 (0.63–1.80)	0.817
Child–Pugh class
A	3,481	Ref.		Ref.
B	749	1.26 (1.09–1.45)	0.001	1.22 (1.07–1.38)	0.003
C	20	5.03 (2.70–9.36)	<0.001	3.31 (1.85–5.94)	<0.001
ALBI grade
1	1,555	Ref.		Ref.
2	2,559	1.21 (1.10–1.33)	<0.001	1.07 (0.99–1.16)	0.087
3	136	1.08 (0.85–1.36)	0.548	0.98 (0.78–1.22)	0.839
AFP, ng/ml
<400	2,689	Ref.		Ref.
≥400	1,561	1.18 (1.09–1.29)	<0.001	1.14 (1.06–1.22)	<0.001
PIVKA-II, mAU/ml
<40	1,280	Ref.		Ref.
≥40	2,970	0.98 (0.89–1.07)	0.590	1.04 (0.96–1.12)	0.339
Tumor size, mm
<100	2,339	Ref.		Ref.
≥100	1,911	1.29 (1.19–1.41)	<0.001	1.21 (1.12–1.30)	<0.001
Number of tumors
Single	1,320	Ref.		Ref.
Multiple	2,930	1.57 (1.43–1.72)	<0.001	1.55 (1.43–1.67)	<0.001
Extrahepatic spread
Absent	2,809	Ref.		Ref.
Present	1,441	1.69 (1.56–1.84)	<0.001	1.61 (1.50–1.74)	<0.001
Macrovascular invasion
Absent	1,989	Ref.		Ref.
Present	2,261	1.40 (1.28–1.52)	<0.001	1.14 (1.06–1.23)	<0.001
Combined with transarterial therapy
None	329	Ref.		Ref.
HAIC	1,882	0.72 (0.62–0.84)	<0.001	0.78 (0.68–0.89)	<0.001
TACE	895	0.85 (0.72–0.99)	0.039	0.94 (0.82–1.09)	0.433
HAIC + TACE	1,144	0.69 (0.59–0.81)	<0.001	0.74 (0.64–0.86)	<0.001
Systemic therapy regimen
TKI	613	Ref.		Ref.
ICIs	350	0.82 (0.70–0.97)	0.019	0.77 (0.67–0.90)	<0.001
TKI + ICIs	2,822	0.70 (0.63–0.79)	<0.001	0.61 (0.55–0.68)	<0.001
VEGF + ICIs	465	0.76 (0.64–0.90)	0.002	0.56 (0.48–0.65)	<0.001

HRs and 95% CIs were derived from multivariable Cox proportional hazards regression analyses. AFP, alpha-fetoprotein; ALBI, albumin–bilirubin; ECOG-PS, Eastern Cooperative Oncology Group-performance status; HAIC, hepatic arterial infusion chemotherapy; ICIs, immune checkpoints inhibitors; OS, overall survival; PIVKA-II, protein induced by vitamin K absence or antagonist-II; PFS, progression-free survival; TACE, transarterial chemoembolization; TKI, tyrosine kinase inhibitor; VEGF, anti-VEGF antibody.

Safety

Treatment-related AEs (TRAEs) for the overall cohort are summarized in Table 3. The most frequently reported AEs were hepatotoxicities (91.3%; n = 3,880) and hematologic toxicities (61.9%; n = 2,631). TRAEs of grade ≥3 occurred in 28.3% of patients, among whom hematologic (15.4%; n = 654) was the most common grade 3/4 adverse event. Grade 3/4 hepatotoxicity occurred in 11.5% of patients.

Table 3.

Adverse events appeared after treatment in the cohort.

Adverse events	Any grade	Grade 3/4
Hematologic toxicities	2,631 (61.9)	654 (15.4)
Leukopenia	916 (22.0)	175 (4.1)
Neutropenia	1,821 (42.9)	476 (11.2)
Reduced hemoglobin level	1,366 (32.1)	37 (0.9)
Thrombocytopenia	369 (8.7)	142 (3.3)
Cardiovascular disorders
Hypertension	73 (1.7)	42 (1.0)
Dermatology/skin
Anaphylaxis	193 (4.5)	34 (0.8)
Gastrointestinal events	1,308 (30.7)	101 (2.4)
Abdominal pain	1,100 (25.9)	29 (0.6)
Abdominal distension	128 (3.0)	18 (0.4)
Anorexia	90 (2.1)	1 (0.0)
Diarrhea	108 (2.5)	9 (0.2)
Nausea	198 (4.7)	4 (0.1)
Hemorrhage	50 (1.2)	46 (1.1)
Hepatotoxicities	3,880 (91.3)	490 (11.5)
Elevated AST	3,624 (85.3)	281 (6.6)
Elevated ALT	2,569 (60.5)	98 (2.3)
Elevated TBIL	1,349 (31.7)	202 (4.8)
Hypoalbuminemia	3,540 (83.3)	20 (0.5)
Renal/genitourinary
Elevated CRE	297 (7.0)	8 (0.2)

Data are presented as n (%). Grade 3/4 referred to combination of grade 3 and grade 4. ALT, alanine aminotransferase; AST, aspartate transferase; CRE, creatinine; TBIL, total bilirubin.

Compared with the Sys-only group, patients receiving TAT-Sys therapy experienced a higher incidence of any-grade AEs, particularly hepatotoxicities and hematologic toxicities, whereas rates of grade 3/4 AEs were generally comparable between groups (Table S8). AE profiles according to different systemic regimens are presented in Fig. S13. Subgroup analyses stratified by baseline Child–Pugh class further demonstrated that severe hepatotoxicity remained generally comparable between treatment groups across different liver function subgroups, despite a higher incidence of any-grade hepatotoxicities in the TAT-Sys group (Table S9).

Discussion

To the best of our knowledge, this study represents the largest multicenter real-world analysis to date examining systemic treatment patterns and clinical outcomes in treatment-naïve patients with HCC. It provides comprehensive real-world evidence on patient characteristics, therapeutic efficacy, and safety profiles across diverse treatment modalities. Because the strict eligibility criteria of interventional trials do not fully capture the heterogeneity of routine clinical practice,³² observational studies provide important complementary insights into treatment delivery, prognostic outcomes, and safety in real-world populations.^33,34

The patient cohort in this study, characterized by a high prevalence of HBV-related advanced HCC, reflects the typical etiological and oncological profile observed in the Asia-Pacific region and is consistent with findings from other real-world studies in similar populations.[35], [36], [37] Compared with highly selected participants in clinical trials, our cohort offers a more extensive view of actual clinical practice. Four predominant systemic treatment regimens were identified among more than 30 recorded combinations, including guideline-recommended therapies such as atezolizumab plus bevacizumab, rivoceranib plus camrelizumab, sorafenib, and lenvatinib, as well as non-guideline regimens, most notably TKI (predominantly lenvatinib) combined with ICIs. TKI + ICIs represented the most frequently administered regimen across treatment lines. This trend may partly reflect the favorable efficacy profile and non-inferior OS demonstrated by lenvatinib compared with sorafenib in the REFLECT trial, after which lenvatinib became widely adopted in routine practice.¹¹ In parallel, the rapid expansion of ICIs-based strategies since 2020 substantially reshaped systemic treatment paradigms for advanced HCC. Following IMbrave150, together with the ORIENT-32 and HEPATORCH studies conducted in China, VEGF + ICIs combinations became established first-line standards of care.^38,39 Subsequently, TKI + ICIs combinations also demonstrated encouraging efficacy in phase III studies, including CARES-310 and APOLLO.^40,41 Although the LEAP-002 trial did not meet its prespecified statistical significance threshold for OS and PFS, lenvatinib plus pembrolizumab demonstrated a trend toward survival benefit in the Asian subgroup.^42,43

In our study, patients receiving TKI + ICIs combinations achieved a median OS of 21.7 months, a median PFS of 10.6 months, and an ORR of 53.3% (per mRECIST v1.1), compared with those in the VEGF + ICIs group (median OS, 22.4 months; PFS, 11.9 months; ORR, 58.5%). These results are consistent with previous reports.^36,44 However, outcomes observed across systemic treatment regimens in our cohort may have been influenced by evolving drug availability and changing standards of care during the study period. Patients receiving newer ICIs-based regimens were more likely to be treated in later years, when supportive care, multidisciplinary management, and access to subsequent therapies had also improved. The longer OS observed among patients treated during 2020–2023 further suggests a potential treatment-era effect. In addition, post-progression treatment patterns may also have influenced OS comparisons across groups. Nevertheless, ICIs-based regimens remained predominant across later treatment lines throughout the study period, without major shifts in subsequent-line therapeutic paradigms. The relative OS and PFS patterns across systemic treatment regimens remained generally consistent in the post-ICI era (2020–2023) analyses. Taken together, these findings support the growing real-world integration of ICIs-based combinations in advanced HCC although limitations arising from cross-regimen comparisons are inevitable.

Survival outcomes and objective response rates in the present cohort appeared numerically higher than those reported in several real-world studies from Western countries and other Asian regions.[23], [24], [25] These differences may partly reflect substantial heterogeneity in patient populations and treatment strategies across regions. In particular, the present cohort predominantly comprised patients with HBV-related HCC and was characterized by frequent integration of TAT into multidisciplinary treatment strategies. By contrast, Western and some other Asian cohorts generally include a higher proportion of non-viral or HCV-related HCC and less frequent incorporation of TAT in advanced-stage disease. Differences in liver disease etiology, intrahepatic tumor burden, preservation of liver function, treatment accessibility, and subsequent treatment patterns may therefore have contributed to variations in survival outcomes across studies. In this context, TAT-Sys therapy as a first-line strategy was associated with more favorable OS and PFS compared with Sys-only therapy, particularly among patients with intermediate-to-advanced disease. These findings remained consistent across sensitivity analyses. Exploratory analyses further suggested lower rates of both new intrahepatic lesions and extrahepatic progression among patients receiving TAT-Sys therapy, supporting the potential contribution of improved local tumor control. In addition, patients treated with TAT-Sys therapy underwent subsequent surgical resection more frequently, suggesting a potential role in treatment intensification and tumor downstaging for selected patients with advanced HCC. Importantly, a subset of patients with BCLC stage A disease were not suitable for upfront surgery because of insufficient future liver remnant, technically challenging tumor location, or high surgical risk. In these patients, MDT-guided transarterial and/or systemic therapies were often adopted to improve resectability and facilitate subsequent resection.

Notably, survival outcomes observed in this real-world cohort were comparable to those reported in several phase III trials evaluating ICIs-based regimens without TAT.^9,12 However, unlike highly selected trial populations, our cohort included patients with broader clinical characteristics, including poorer liver function, larger intrahepatic tumor burden, and extensive vascular invasion or extrahepatic disease, which are often underrepresented in prospective studies. Moreover, several globally adopted dual-ICIs regimens, including nivolumab plus ipilimumab and the STRIDE regimen (durvalumab plus tremelimumab), were not widely available in China during most of the study period. Under these circumstances, TAT-integrated strategies became increasingly incorporated into routine practice to improve intrahepatic tumor control and expand treatment opportunities. Together, these findings support the clinical feasibility of TAT-Sys strategies in the Chinese real-world setting. Recent phase III studies have further highlighted the growing interest in integrating locoregional and systemic therapies for unresectable HCC. In EMERALD-1, TALENTACE, and more recently EMERALD-3, immunotherapy-based combinations with TACE significantly improved PFS.[18], [19], [20] Similarly, LEAP-012 demonstrated improved PFS with lenvatinib plus pembrolizumab combined with TACE, although no OS benefit was observed according to recent updates, and further development of this strategy has been discontinued.²¹ More recently, preliminary results from EMERALD-3 suggested significant PFS improvement with durvalumab plus tremelimumab combined with lenvatinib and TACE, although mature survival data remain awaited.²⁰ Collectively, these findings support continued exploration of multimodal TAT-based strategies for unresectable HCC.

Although combination regimens and TAT-Sys strategies were associated with favorable survival outcomes, safety profiles remained generally manageable. Patients receiving TAT-Sys therapy experienced a higher incidence of any-grade AEs, particularly hepatotoxicities and hematologic toxicities, likely related to transarterial procedures and chemotherapy exposure. However, rates of grade 3/4 AEs were generally comparable between TAT-Sys and Sys-only groups, supporting the feasibility of combining transarterial and systemic therapies under appropriate monitoring. Despite these manageable safety profiles, only a limited proportion of patients proceeded to second-line systemic therapy. Although hepatotoxicities were common, severe grade 3/4 hepatic toxicities were relatively infrequent, suggesting that treatment-related toxicity alone may not fully explain the low transition rate to subsequent therapy. Limited availability of approved second-line agents during the earlier study period (2017–2020), together with disease progression, deterioration of liver function, and challenges in longitudinal follow-up, may also have contributed. Similar observations have been reported in other real-world HCC cohorts,²⁵ underscoring the persistent challenges of treatment sequencing in advanced HCC.

This study has several limitations. First, despite the multicenter design, MDT-based treatment decision-making may have varied across centers and over time, potentially contributing to heterogeneity in treatment selection and outcomes. Accordingly, these treatment strategies should preferably be implemented in experienced, high-volume centers with multidisciplinary expertise in HCC management. Second, although IPTW adjustments were performed, residual confounding and selection bias inherent to the retrospective design cannot be fully excluded. In particular, patients selected for TAT-based multimodal strategies may have differed from those receiving systemic therapy alone in ways not fully captured by retrospective adjustment methods. Third, categorization by treatment class rather than individual agents may have obscured efficacy differences between specific regimens. Fourth, guideline-endorsed dual-ICIs regimens, including nivolumab plus ipilimumab and the STRIDE regimen, were not represented because these therapies had not yet received regulatory approval in China during most of the study period. Finally, the predominance of HBV-related HCC and the frequent use of TAT-Sys may limit the direct generalizability of these findings to regions with different treatment practices.

In conclusion, this large multicenter real-world study delineates the evolving landscape of systemic therapy for HCC in China and provides contemporary evidence regarding the clinical efficacy of combination strategies pairing ICIs-based regimens with transarterial therapy in routine clinical practice. Prospective studies are warranted to further define the optimal integration and sequencing of these approaches.

Abbreviations

AE, adverse event; AFP, alpha-fetoprotein; ALBI, albumin–bilirubin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BCLC, Barcelona Clinic Liver Cancer; CR, complete response; CRE, creatinine; ECOG-PS, Eastern Cooperative Oncology Group-performance status; HAIC, hepatic arterial infusion chemotherapy; HCC, hepatocellular carcinoma; ICIs, immune checkpoint inhibitors; IPTW, inverse probability of treatment weighting; MDT, multidisciplinary team; mRECIST, modified RECIST; ORR, objective response rate; OS, overall survival; PD, progressive disease; PFS, progression-free survival; PIVKA-II, protein induced by vitamin K absence or antagonist-II; PR, partial response; PSM, propensity score matching; RSF, random survival forest; SD, stable disease; Sys-only, systemic therapy alone; TACE, transarterial chemoembolization; TAT, transarterial therapy; TAT-Sys, transarterial therapy combined with systemic therapy; TBIL, total bilirubin; TKI, tyrosine kinase inhibitor; TRAEs, treatment-related adverse events; VEGF, vascular endothelial growth factor.

Authors’ contributions

Conceptualization, study design, and formal analysis: BH, WW. Investigation, methodology, or visualization: BH, YW, SL, RZ. Data collection or material support: BH, WL, ZZheng, JG, ZZuo, MD, FL, CZ, RL, RG, MT. Project administration: WW, RG, LL. Funding acquisition: WW, RG, JM. Original draft or review and editing: BH, WW, JM. Reviewed and approved the manuscript: all authors.

Data availability

The data that support the findings of this study are not publicly available owing to patient privacy and ethical restrictions, but are available from the corresponding author upon reasonable request.

Financial support

This study was supported by the National Science and Technology Major Project of China (2024ZD0520400, 2024ZD0520402); National Natural Science Foundation of China (No. 82303875, No. 82303879); Guangdong Provincial Natural Science Foundation (2024A1515012581); Clinical Trials Project (5010 Project) of Sun Yat-sen University (No. 5010-2023001), and Shenzhen Science and Technology Innovation Committee (JCYJ20250604143712016).

Conflicts of interest

The authors declare no conflicts of interest that pertain to this work.

Please refer to the accompanying ICMJE disclosure forms for further details.

Supplementary data

The following are the Supplementary data to this article: