Transarterial chemoembolization (TACE) with or without immune checkpoint inhibitors in unresectable hepatocellular carcinoma: a meta-analysis of randomized clinical trials

Abstract

Background

Transarterial chemoembolization (TACE) is the standard of care for intermediate-stage unresectable hepatocellular carcinoma (HCC), but progression-free survival (PFS) with TACE alone remains limited. Recent randomized clinical trials have evaluated the addition of immune checkpoint inhibitors (ICIs) to TACE. We conducted a meta-analysis of randomized trials to assess the efficacy and safety of TACE plus ICIs versus TACE alone in unresectable HCC.

Materials and methods

A systematic search of PubMed, Embase, Cochrane Library, and American Society of Clinical Oncology and European Society for Medical Oncology databases was carried out through 15 December 2025 and last updated on 3 June 2026 to identify randomized clinical trials comparing TACE with or without ICIs in unresectable HCC. Outcomes included PFS, objective response rate (ORR), overall survival (OS), and safety. Random-effects meta-analyses were conducted using R software (version 4.5.2, R Foundation for Statistical Computing, Vienna, Austria).

Results

Six randomized clinical trials (EMERALD-1, EMERALD-3, Jia et al., LEAP-012, TALENTACE, and CARES-005) involving 2439 patients were included. Compared with TACE alone, combination therapy improved PFS [hazard ratio (HR) 0.63, 95% confidence interval (CI) 0.52-0.77, P < 0.01] and ORR (risk ratio 1.45, 95% CI 1.31-1.61, P < 0.001). OS benefit was borderline (HR 0.85, 95% CI 0.73-0.99, P = 0.04), although data remain immature. Combination therapy was associated with higher rates of any-grade and grade ≥3 adverse events, serious adverse events, and treatment discontinuations.

Conclusion

Adding ICIs to TACE improved PFS and response rates compared with TACE alone in unresectable HCC, at the cost of increased toxicity. OS benefit was borderline and should be interpreted with caution; longer follow-up is required.

Highlights

• •TACE plus ICIs s improved PFS in unresectable HCC (HR 0.63, 95% CI 0.52-0.77, P < 0.01).
• •Combined therapy improved ORR (RR 1.45, 95% CI 1.31-1.61, P < 0.001).
• •Grade ≥3 AEs were more frequent with the combination approach (RR 2.20; 95% CI 1.72-2.81; P < 0.001).
• •Treatment discontinuations were also more frequent in the combination arm (RR 5.02; 95% CI 1.43-17.68; P < 0.012).
• •A borderline improvement in OS was observed with combination therapy (HR 0.85, 95% CI 0.73-0.99, P = 0.04), limited by immature time-to-event data.

Introduction

Liver cancer is the sixth most common malignancy worldwide and the third leading cause of cancer-related death.¹ Approximately 80% of cases are diagnosed at an unresectable stage, resulting in a poor prognosis. The five-year overall survival (OS) is only about 9.6% with conservative management.²

The 2025 Barcelona Clinic Liver Cancer (BCLC) update refines the definition of intermediate-stage hepatocellular carcinoma (HCC; BCLC stage B) as characterized by >3 nodules or ≤3 nodules with at least one lesion >3 cm. This stage represents a heterogeneous population that can be stratified into three subgroups: patients eligible for liver transplantation under extended criteria, those with preserved liver function suitable for locoregional therapies such as transarterial chemoembolization (TACE), and patients with diffuse or extensive disease requiring systemic therapy. The update also emphasizes individualized assessment of TACE suitability and timely transition to systemic therapy in cases of untreatable progression, reinforcing the concept of treatment stage migration.³

Randomized evidence established TACE as a survival-improving therapy in unresectable HCC,⁴ with early trials demonstrating improved 2-year survival and objective response rates (ORRs) of ∼35%.⁵ Subsequent studies have reported median progression-free survival (PFS) of ∼5 to 8 months with TACE alone.^6,7

In advanced HCC (BCLC stage C), systemic therapy, particularly immune checkpoint inhibitors (ICIs) and antiangiogenic agents, has substantially improved clinical outcomes. ICIs combined with antiangiogenic agents or dual-ICI regimens [anti-PD-1/PD-(L)1 plus anti-CTLA-4] have demonstrated superior OS and response rates compared with tyrosine kinase inhibitors.8, 9, 10

There is a strong biological rationale for combining TACE with ICIs. TACE induces tumor antigen release and upregulates PD-L1, promoting immune cell infiltration into the tumor microenvironment and potentially enhancing tumor control.¹¹ Additionally, antiangiogenic agents contribute to normalization of tumor vasculature, which facilitates immune cell trafficking.¹²

Recently, several randomized clinical trials (RCTs) have evaluated combination approaches as first-line treatment for intermediate-stage unresectable HCC, including EMERALD-1 (durvalumab plus bevacizumab plus TACE),¹³ EMERALD-3 (durvalumab plus tremelimumab plus TACE),¹⁴ LEAP-012 (lenvatinib plus pembrolizumab plus TACE),¹⁵ TALENTACE (atezolizumab plus bevacizumab with on-demand TACE),¹⁶ CARES-005, and Jia et al.^17,18 (camrelizumab plus rivoracenib plus TACE).

In this systematic review and meta-analysis, we evaluated the efficacy and safety of TACE combined with ICIs, with or without antiangiogenic agents, in patients with unresectable HCC. We assessed tumor response, survival, and safety outcomes and explored treatment effects across clinically relevant patient subgroups.

Materials and methods

This systematic review and meta-analysis was conducted in accordance with the Cochrane Handbook for Systematic Reviews of Interventions and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.^19,20 The study protocol was registered in PROSPERO with the identifier CRD420251033775.

Search strategy and data collection

A comprehensive systematic search was conducted from database inception through 3 June 2026 in PubMed, the Cochrane Library, Embase, the European Society for Medical Oncology database, and the American Society of Clinical Oncology Meeting Library. Search terms included: ‘transarterial’, ‘chemoembolisation’, ‘local treatment’, ‘TACE’, ‘unresectable’, ‘hepatocellular carcinoma’, ‘HCC’, and ‘systemic therapy’. No limits regarding publication date or study filters were applied. The complete search strategy is provided in Supplementary Table S2, available at https://doi.org/10.1016/j.esmogo.2026.100371.

Two reviewers (H.K. and A.L.) independently screened titles and abstracts, evaluated full texts for eligibility, extracted data, and assessed study quality. Discrepancies were resolved by consensus or, when necessary, by consultation with a third reviewer (A.B.).

Eligibility criteria

We included RCTs evaluating the efficacy and safety of TACE combined with ICIs, with or without concomitant antiangiogenics, compared with TACE alone in patients with unresectable HCC, and that reported at least one outcome of interest. No restrictions were applied regarding publication year, patient demographics, or sample size. Conference abstracts and presentations were eligible.

Exclusion criteria were real-world evidence studies, nonrandomized trials, reviews, case reports, case series, case–control studies, preclinical research, ongoing trials without available results, studies with comparator arms that included treatments other than TACE alone, and full texts unavailable in English.

Intervention

The intervention of interest was TACE in combination with ICIs.

Outcomes

The primary outcome was PFS. Secondary outcomes included ORR, OS, the incidence of adverse events (AEs) of any grade, grade ≥3 AEs, treatment discontinuation due to AEs, and treatment-related AEs leading to death. Exploratory subgroup analyses for PFS were carried out according to BCLC stage (A, B, and C), Child–Pugh class A, etiology [hepatitis B virus (HBV), hepatitis C virus (HCV), and nonviral], and α-fetoprotein (AFP) level (>400 versus ≤400 ng/ml), whenever these data were reported by the included trials.

Quality assessment

Study quality and risk of bias were evaluated using the Cochrane Risk of Bias 2 (RoB 2) tool.²¹ Two reviewers (H.K. and A.L.) independently carried out the assessments, resolving disagreements through discussion after reviewing the full study report. Each RCT was categorized as low risk of bias, some concerns, or high risk of bias across five domains: bias arising from the randomization process, bias due to deviations from intended interventions, bias due to missing outcome data, bias in outcome measurement, and bias in selection of the reported results.

Statistical analysis

All analyses were conducted using R software (version 4.5.2 - R Foundation for Statistical Computing, Vienna, Austria) with the meta and metafor packages. Random-effects models using Mantel–Haenszel and inverse variance weighting were applied for dichotomous and time-to-event outcomes, respectively. For binary outcomes, effect estimates were calculated based on the number of events per total population and expressed as risk ratios (RRs) with corresponding 95% confidence intervals (CIs). Time-to-event outcomes, including OS and PFS, were synthesized using hazard ratios (HRs) and associated 95% CIs reported in each study. Statistical heterogeneity was assessed using the I² statistic, with values >50% considered indicative of substantial heterogeneity.

Results

Our initial search identified 2434 records. After removing duplicates and screening titles and abstracts, 34 studies were assessed in full. Six phase II-III RCTs met the eligibility criteria and were included in the final analysis (Figure 1).

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)flow diagram of study screening and selection. Blue vertical boxes indicate each stage of the screening, and the horizontal boxes present more detailed information about the process, including the steps carried out in each stage. The search was last updated on 3 June 2026. ∗Records were identified from databases and conference registries. ∗∗Records were excluded after title and abstract screening. ASCO, American Society of Clinical Oncology; ESMO, European Society for Medical Oncology.

Baseline characteristics

A total of 2439 patients with unresectable HCC were included across the six RCTs. The median age ranged from 57 to 67 years. Most patients were male (1528 patients reported; 62.6%). HBV was the predominant underlying etiology (1505 patients; 61.7%), followed by nonviral causes (832 patients; 34.1%). Regarding tumor staging, BCLC stage B accounted for most patients (1439; 58.9%), followed by BCLC stage A (586; 24%) and BCLC stage C (407; 16.7%). Tumor burden score >7 at baseline was also common across the included trials (1337; 54.8%).

All six studies included TACE alone as the control arm. In the experimental arms, TACE was combined with different systemic regimens: camrelizumab plus rivoracenib (CARES-005 and Jia et al; n = 100 and n = 214, respectively), durvalumab plus bevacizumab (EMERALD-1; n = 204), durvalumab plus tremelimumab plus lenvatinib (EMERALD-3; n = 293), pembrolizumab plus lenvatinib (LEAP-012; n = 237), and atezolizumab plus bevacizumab (TALENTACE; n = 171). Detailed study characteristics are summarized in Table 1.

Table 1.

Baseline characteristics of the randomized trials included in the meta-analysis comparing TACE alone with TACE-based combination therapy in unresectable hepatocellular carcinoma

Study	Location	Treatment regimen	Number of patients	Median age, years (range)	Male, n (%)	Etiology, n (%)			Macrovascular invasion, n (%)	BCLC stage, n (%)			Prior TACE, n (%)	Tumor burden at baseline, n (%)
						HBV	HCV	Nonviral		A	B	C		≤7	>7
CARES-005 (NCT04559607)	China	TACE + CAM + RIVO	100	58.5 (51.5-65.0)	87 (87)	80 (80)	1 (1)	19 (19)	42 (42)	14 (14)	44 (44)	42 (42)	25 (25)	11 (11)	89 (89)
		TACE alone	100	57.0 (51.0-65.5)	87 (87)	79 (79)	6 (6)	15 (15)	44 (44)	14 (14)	42 (42)	44 (44)	24 (24)	10 (10)	90 (90)
EMERALD-1 (NCT03778957)	Multicenter	TACE + DURVA + BEVA	204	64.5 (58.0-73.0)	162 (79)	75 (37)	42 (21)	86 (42)	16 (8)	51 (25)	117 (57)	35 (17)	0	97 (48)	106 (52)
		TACE alone	205	66.0 (59.0-71.0)	163 (80)	74 (36)	54 (26)	76 (37)	13 (6)	49 (24)	122 (60)	31 (15)	0	102 (50)	103 (50)
EMERALD-3 (NCT05301842)	Multicenter	TACE + DURVA + TREME + LENVA	293	67.0 (25.0-86.0)	N/A	131 (44.7)	39 (13.3)	120 (41.0)	29 (9.9)a	51 (17.4)	187 (63.8)	55 (18.8)	51 (17.4)	115 (39.2)	178 (60.8)
		TACE alone	292	65.0 (34.0-85.0)	N/A	138 (47.3)	37 (12.7)	109 (37.3)	28 (9.6)a	69 (23.6)	171 (58.6)	50 (17.1)	50 (17.1)	115 (39.4)	177 (60.6)
Jia et al. (2026) (NCT05320692)	Multicenter	TACE + CAM + RIVO	214	61.0 (30.0-82.0)	183 (85.5)	180 (84.1)	19 (8.9)	12 (5.6)	N/A	55 (25.7)	137 (64.0)	22 (10.3)	N/A	N/A	N/A
		TACE alone	209	61.0 (30.0-83.0)	172 (82.3)	173 (82.8)	19 (9.1)	14 (6.7)	N/A	57 (27.3)	133 (63.6)	19 (9.1)	N/A	N/A	N/A
LEAP-012 (NCT04246177)	Multicenter	TACE + PEMBRO + LENVA	237	65.0 (57.0-72.0)	192 (81)	153 (65)	42 (18)	161 (68)	0	80 (34)	135 (57)	21 (9)	0	112 (47)b	125 (53)c
		TACE alone	243	66.0 (59.0-73.0)	206 (85)	144 (59)	39 (16)	187 (77)	0	68 (28)	146 (60)	29 (12)	0	116 (48)b	127 (52)c
TALENTACE (NCT04712643)	China and Japan	TACE + ATEZO + BEVA	171	62.0 (30.0-89.0)	136 (79.5)	143 (83.6)	8 (4.7)	16 (9.4)	N/A	36 (21.1)	100 (58.5)	35 (20.5)	0	0	171 (100)d
		TACE alone	171	60.0 (21-90.0)	140 (81.9)	135 (78.9)	5 (2.9)	17 (9.9)	N/A	42 (24.6)	105 (61.4)	24 (14)	0	0	171 (100)d

ATEZO, atezolizumab; BCLC, Barcelona Clinic Liver Cancer; BEVA, bevacizumab; CAM, camrelizumab; DURVA, durvalumab; HBV, hepatitis B virus; HCV, hepatitis C virus; LENVA, lenvatinib; N/A, not available; PEMBRO, pembrolizumab; RIVO, rivoceranib; TACE, transarterial chemoembolization; TREME, tremelilumab.

^aOnly Vp1 or Vp2 reported.

^bTumor burden score ≤6.

^cTumor burden score >6.

^dTumor burden score ≥6.

Efficacy endpoints

In the meta-analysis of PFS, the pooled results demonstrated a significant benefit for the combination of TACE with ICIs (HR 0.63, 95% CI 0.52-0.77, P < 0.01). Substantial heterogeneity was observed among trials (I² = 66.0%) (Figure 2A). The pooled analysis of ORR also favored combination therapy (RR 1.45, 95% CI 1.31-1.61, P < 0.001) (Figure 2B), with low heterogeneity across studies (I² = 7.9%). Pooled OS showed a borderline benefit with combination therapy (HR 0.85, 95% CI 0.73-0.99, P = 0.04) (Figure 2C).

Figure 2.

Efficacy outcomes of patients with unresectable hepatocellular carcinoma treated with TACE plus ICIs versus TACE alone. RRs or HRs for each trial are represented by squares, with size reflecting study weight, and horizontal lines indicate the 95% CIs. The diamonds represent the estimated overall effect of the meta-analysis based on random effects. Studies include EMERALD-1, EMERALD-3, Jia et al., LEAP-012, TALENTACE, and CARES-005.

CI, confidence interval; HR, hazard ratio; ICI, immune checkpoint inhibitor; IV, inverse variance method; MH, Mantel–Haenszel method; PFS, progression-free survival; ORR, objective response rate; OS, overall survival; RR, risk ratio; TACE, transarterial chemoembolization.

In subgroup analyses (Figure 3), the PFS benefit was significant in patients with BCLC stage A disease (HR 0.77, 95% CI 0.68-0.87, P = 0.01), BCLC stage B disease (HR 0.71, 95% CI 0.55-0.92, P = 0.02), Child–Pugh class A (HR 0.74, 95% CI 0.54-0.99, P = 0.05), HBV-related disease (HR 0.67, 95% CI 0.52-0.86, P = 0.01), nonviral etiology (HR 0.76, 95% CI 0.59-0.99, P = 0.04), AFP >400 ng/ml (HR 0.69, 95% CI 0.51-0.94, P = 0.03), and AFP ≤400 ng/ml (HR 0.72, 95% CI 0.61-0.86, P = 0.01). In contrast, no statistically significant benefit was observed in patients with BCLC stage C disease (HR 0.59, 95% CI 0.30-1.17, P = 0.10) or HCV-related HCC (HR 0.88, 95% CI 0.46-1.68, P = 0.61).

Figure 3.

Subgroup analyses for PFS in unresectable hepatocellular carcinoma. HR for each subgroup is represented by squares, and horizontal lines indicate the 95% CIs. Diamonds represent the subgroup-specific and overall pooled effect estimates from random-effects meta-analysis. Subgroups derived from included randomized controlled trials (EMERALD-1, EMERALD-3, Jia et al., LEAP-012, TALENTACE, and CARES-005).

BCLC, Barcelona Clinic Liver Cancer; CI, confidence interval; HBV, hepatitis B virus; HCV, hepatitis C virus; HR, hazard ratio; PFS, progression-free survival.

Safety endpoints

Both any-grade AEs (RR 1.14, 95% CI 1.04-1.25, P = 0.007) and grade ≥3 AEs (RR 2.20, 95% CI 1.72-2.81, P < 0.001) occurred more frequently in the combination therapy arm. Treatment discontinuation (RR 5.02, 95% CI 1.43-17.68, P < 0.012) and treatment-related AEs leading to death (RR 2.45, 95% CI 1.16-5.17, P = 0.019) were also more frequent in the combination arm (Figure 4).

Figure 4.

Safety outcomes of patients with unresectable hepatocellular carcinoma treated with TACE plus ICIs versus TACE alone. RRs for each trial are represented by a square, and the horizontal line crossing the square indicates the 95% CI. The diamonds represent the estimated overall effect of the meta-analysis based on random effects. Studies include EMERALD-1, EMERALD-3, Jia et al., LEAP-012, TALENTACE, and CARES-005.

CI, confidence interval; ICI, immune checkpoint inhibitor; MH, Mantel–Haenszel method; RR, risk ratio; TACE, transarterial chemoembolization.

Quality assessment

EMERALD-1 and LEAP-012 met all predefined RoB 2 criteria and were classified as having a low overall risk of bias. In contrast, CARES-005, EMERALD-3, Jia et al., and TALENTACE were open-label, leading to concerns in the domain of deviations from intended interventions. This resulted in a judgment of some concerns for CARES-005, EMERALD-3, and Jia et al., whereas TALENTACE, in which TACE was administered on an on-demand basis, was considered at high risk for this domain and assigned an overall rating of some concerns (Supplementary Table S3, available at https://doi.org/10.1016/j.esmogo.2026.100371). Leave-one-out sensitivity analyses showed that excluding each study one at a time did not meaningfully change the pooled results for PFS or ORR (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmogo.2026.100371). Other sensitivity analyses also showed that excluding unpublished trials, excluding Asian-only trials, and excluding TACE on-demand trials did not meaningfully change the pooled results for PFS or ORR (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmogo.2026.100371). Treatment-related adverse events, including immune-related adverse events and TACE-related adverse events, are summarized in Supplementary Table S4, available at https://doi.org/10.1016/j.esmogo.2026.100371.

Discussion

This systematic review and meta-analysis of randomized studies explored TACE plus ICIs with or without antiangiogenic agents compared with TACE alone in patients with unresectable HCC. Across six RCTs including 2439 patients, we observed a consistent improvement in PFS (HR 0.63, 95% CI 0.52-0.77, P < 0.01) and ORR (RR 1.45, 95% CI 1.31-1.61, P < 0.001) with the addition of ICIs to TACE. A borderline improvement in OS was observed with combination therapy (HR 0.85, 95% CI 0.73-0.99, P = 0.04). The magnitude of benefit may be influenced by immature OS data in several studies and by the methodological challenges associated with pooling time-to-event estimates across trials, including differences in follow-up duration, number of events, subsequent therapies after progression, potential nonproportional hazards, and clinical heterogeneity across trials in patient populations, locoregional treatment strategies, and systemic therapy backbones.

Subgroup analyses suggested that the PFS benefit of combination therapy was consistent across several clinically relevant subgroups, including patients with BCLC stage A/B disease and AFP levels ≤400 ng/ml and >400 ng/ml. A significant benefit was also observed in both HBV-related and nonviral HCC, although the magnitude of effect appeared greater in HBV-related disease, potentially reflecting differences in tumor immunobiology.²² In contrast, no statistically significant benefit was observed in patients with BCLC stage C disease or HCV-related HCC; however, these findings should be interpreted with caution given the wide CIs and limited sample sizes within these subgroups. Overall, these findings support the broad applicability of combination therapy across clinically relevant patient subgroups while highlighting the inherent heterogeneity of unresectable HCC.

TACE remains a cornerstone locoregional therapy for patients with intermediate-stage HCC who are not candidates for curative approaches. By delivering chemotherapeutic and embolic agents directly through the hepatic artery, TACE induces ischemic necrosis and cytotoxic tumor death.²³ Despite its widespread use, its effectiveness is limited by insufficient drug penetration, the risk of liver decompensation, and the development of treatment resistance, partly driven by an immunosuppressive and heterogeneous tumor microenvironment. These limitations have driven efforts to explore therapeutic strategies that enhance or complement the effects of TACE.²⁴

ICIs have substantially transformed the therapeutic landscape of advanced HCC. The IMbrave150 trial highlighted the superiority of atezolizumab combined with bevacizumab over sorafenib for first-line treatment of unresectable disease.²⁵ This was followed by positive first-line data from HIMALAYA, which demonstrated the superiority of durvalumab plus tremelilumab over sorafenib.²⁶ Similarly, CheckMate 9DW showed that nivolumab plus ipilimumab outperformed lenvatinib or sorafenib.¹⁰ Additionally, CARES-310 revealed the superiority of camrelizumab plus rivoracenib over sorafenib,²⁷ and ORIENT-32 demonstrated that sintilimab plus bevacizumab was superior to sorafenib.²⁸

The efficacy of immunotherapy is closely linked to the immune profile of the tumor microenvironment, with hot tumors showing greater responsiveness than cold tumors.²⁹ TACE may enhance immunotherapy activity by promoting antigen release, immunogenic cell death, and PD-L1 upregulation, thereby potentially converting an immunosuppressive microenvironment into a more immunoresponsive state.³⁰ To date, preclinical and translational evidence has provided a biological basis for this combination. However, current clinical data do not clearly differentiate between a synergistic effect and a mainly additive benefit,²³ and this should be explored for a better therapeutic sequencing decision. A combination is considered synergistic when its observed effect exceeds the effect expected from each therapy separately. However, in clinical trials, superiority alone is not sufficient to prove synergy, which requires a better understanding of variability in drug response and new biomarkers.^31,32

A recent meta-analysis on this topic included only three phase III trials and carried out a restricted safety analysis focused on grade 3-5 AEs.³³ In contrast, our analysis incorporated a broader literature search across multiple databases, included additional randomized trials with a larger population, and provided a more comprehensive evaluation of safety outcomes. Although both analyses demonstrated improved PFS and ORR with combination therapy, our study additionally identified higher rates of any-grade AEs, serious AEs, and treatment discontinuations. It is important to note that pooling AEs across regimens involving different anti-PD-1/PD-(L)1 agents, anti-VEGF/antiangiogenic partners, and TACE schedules produces a composite safety result that is difficult to interpret and should be analyzed with caution. More granular toxicity should be better reported in the trials to guide clinical decision making.

Future research should focus on optimizing treatment sequencing, elucidating mechanisms of resistance, and identifying biomarkers predictive of benefit and toxicity. Ongoing trials will help define the optimal integration of immunotherapy with TACE, including timing strategies and patient selection (NCT04224636, NCT04340193).

Although this meta-analysis focused on strategies involving TACE, the rationale for integrating immunotherapy with locoregional treatments may also apply to methods beyond TACE. Transarterial radioembolization (TARE) and stereotactic body radiotherapy (SBRT) are increasingly being employed in certain patients with HCC, potentially offering additional methods for local tumor management. Biologically, these therapies can also potentially convert an immunosuppressive microenvironment into a more immunoresponsive state.34, 35, 36 This supports the exploration of ICI-based combinations with TARE and SBRT. Nevertheless, the results of this meta-analysis should not be directly applied to other locoregional therapies. Specific prospective studies are necessary to determine the effectiveness, safety, ideal sequencing, and patient selection for the combination of immunotherapy with TARE, SBRT, and other liver-targeted strategies.

This systematic review and meta-analysis has limitations. Two trials (Jia et al and TALENTACE) were available only as conference presentations, and full peer-reviewed publications are awaited. Survival outcomes remain immature in all included RCTs, limiting OS estimates. Some analyses exhibited heterogeneity likely attributable to differences in drug combinations, treatment schedules (on-demand versus fixed-window TACE), trial design, and baseline population characteristics. Furthermore, we did not have access to individual patient data, restricting the depth of subgroup analyses, particularly comparisons between Asian and Western populations, etiology-specific analyses, and evaluation of other clinically relevant factors, nor did we have access to the postprotocol/subsequent systemic therapy, limiting the interpretation of the OS.

Additionally, in the three-arm EMERALD-3 trial, only the comparison between durvalumab plus tremelimumab plus lenvatinib and TACE versus TACE alone was included to avoid double-counting the shared control group. As a result, the potential contribution of the durvalumab plus tremelimumab regimen alone was not captured in the pooled estimates.

Leave-one-out sensitivity analyses were carried out and showed that excluding each study one at a time did not meaningfully change the pooled results for PFS or ORR, suggesting that the findings were not driven by any single trial (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmogo.2026.100371). Additional sensitivity analyses also showed that excluding unpublished trials, excluding Asian-only trials, and excluding TACE on-demand trials did not meaningfully change the pooled results for PFS or ORR (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmogo.2026.100371). Grade 3 or higher adverse events are shown in Supplementary Figure S3, available at https://doi.org/10.1016/j.esmogo.2026.100371.

Despite these limitations, this systematic review and meta-analysis synthesizes the highest level of available evidence from six randomized trials involving 2439 patients, providing a comprehensive evaluation of the efficacy and safety of TACE plus ICIs versus TACE alone in unresectable HCC. Our results support the growing role of combination strategies in intermediate-stage disease, although the optimal regimen, sequencing, and patient selection remain areas for continued investigation.

Conclusion

This systematic review and meta-analysis of randomized trials in unresectable HCC demonstrates that combining TACE with ICIs provides a significant improvement in PFS compared with TACE alone. However, this benefit is associated with increased toxicity, including higher rates of treatment discontinuation due to AEs. More mature data are needed to clarify the impact on OS and long-term outcomes. Future research should focus on optimizing treatment sequencing and refining patient selection to maximize benefit.