Transarterial chemoembolization combined with tyrosine kinase inhibitors and programmed death-1 inhibitors in recurrent intermediate and advanced hepatocellular carcinoma: a single-centre retrospective cohort study

Abstract

Background

We aimed to evaluate the efficacy and safety of transarterial chemoembolization (TACE) plus tyrosine-kinase inhibitors (TKIs) with or without programmed death-1 (PD-1) inhibitors in recurrent intermediate-advanced hepatocellular carcinoma (HCC).

Materials and methods

In this retrospective, single-center cohort study (January 2016-January 2024), 39 consecutive patients with recurrent HCC (BCLC B/C) received TACE combined with (1) regorafenib (n = 14), (2) lenvatinib + PD-1 inhibitor (n = 12), or (3) regorafenib + PD-1 inhibitor (n = 13). Overall survival (OS) and progression-free survival (PFS) were primary outcomes. Objective response rate (ORR) and disease control rate (DCR) were secondary. Multivariable Cox regression and propensity score matching analyses were performed. A post-hoc power analysis was conducted to quantify statistical power.

Results

Median follow-up was 36.5 months. Median OS and PFS for the entire cohort were 42.9 months (95% CI 27.1–58.7) and 16.5 months (95% CI 13.6–19.5), respectively. No significant differences in OS (P = 0.164) or PFS (P = 0.830) were observed among the three groups. ORR was 12.8% and DCR 51.3%. BCLC stage B (vs. C) was an independent favorable prognostic factor for OS (HR 0.03, P = 0.01) and PFS (HR 0.04, P = 0.038). Grade ≥ 3 adverse events occurred in 2 (5.1%) patients. The median number of TACE sessions was 4 (IQR 3–6) for Group A, 5 (IQR 3–7) for Group B, and 4 (IQR 3–5) for Group C (P = 0.72).

Conclusion

TACE plus TKIs with or without PD-1 inhibitors showed acceptable efficacy and safety in recurrent intermediate-advanced HCC. The addition of PD-1 inhibitors did not demonstrate additional survival benefit in this small cohort. However, the present sample size was underpowered to detect a clinically meaningful but modest survival difference. Prospective randomized trials are warranted.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12876-026-05070-3.

Introduction

According to GLOBOCAN 2022, liver cancer was the sixth most common cancer and the third leading cause of cancer-related death worldwide [1]. Although early-stage HCC can be cured by radical treatments, nearly 70% of patients in China are first diagnosed at intermediate or advanced stages and thus lose the chance for curative therapy [2]. Even after potentially curative resection or ablation, recurrence rates remain high, making recurrent HCC a clinically distinct and challenging population [2].

Recurrent lesions often arise in the background of chronic liver injury, display more aggressive biology, and exhibit heterogeneous responses to repeat locoregional or systemic therapy. Therefore, evaluating treatment strategies specifically in this subgroup is essential.

According to the Barcelona Clinic Liver Cancer (BCLC) staging system and several clinical studies, systemic therapies including tyrosine kinase inhibitors (TKIs), anti-VEGF agents, and programmed death-1 (PD-1) inhibitors are recommended for advanced stage HCC [3–5]. Sorafenib and lenvatinib were approved as first-line TKIs, whereas regorafenib was approved as second-line treatment after failure of first-line therapy [5]. However, primary or acquired resistance to TKIs inevitably leads to progression and recurrence [5]. PD-1 inhibitors have shown promising efficacy in advanced HCC [6, 7], and current guidelines already endorse combination regimens of PD-1 inhibitors with anti-VEGF and/or TKIs as first-line options [5].

Transarterial chemoembolization (TACE) remains the standard treatment for intermediate stage HCC and is also widely used for advanced HCC in China [2]. Nevertheless, the efficacy of TACE alone is relatively poor [5]. Repeated TACE induces tumor hypoxia, up-regulates hypoxia-inducible factor 1-α, and increases local VEGF expression, which promotes tumor progression and metastasis [8]. Conversely, TACE modulates the tumor micro-environment and enhances the response to PD-1 inhibitors [9]. Pre-clinical and early clinical studies have verified the synergistic effect of TACE combined with TKIs plus PD-1 inhibitors in HCC [10–12]. Yet, to date, no study has specifically evaluated TACE plus TKIs with or without PD-1 inhibitors in the recurrent intermediate-advanced HCC setting.

The objective of this exploratory study was to describe the real-world outcomes of three distinct combination strategies used in our center for recurrent intermediate and advanced HCC: TACE+Regorafenib (standard second-line systemic therapy), TACE+lenvatinib + PD-1 inhibitor (first-line TKI plus immunotherapy), and TACE+regorafenib + PD-1 inhibitor (second-line TKI plus immunotherapy). We explicitly acknowledge that these groups differ in TKI type, treatment line, and immunotherapy exposure, precluding any isolated attribution of effects to PD-1 inhibitors. We hypothesized that the addition of PD-1 inhibitors would prolong OS and PFS compared with TACE + TKI alone, but recognized that the retrospective design and small sample size limit causal inference.

Materials and methods

Study design and setting

We retrospectively reviewed consecutive patients with recurrent intermediate and advanced HCC treated at the Zhenjiang First People’s Hospital between January 2016 and January 2024. The study protocol was approved by the institutional ethics committee of the Affiliated People’s Hospital of Jiangsu University (SQK-2025235-W) and conducted in accordance with the Declaration of Helsinki. Zhenjiang First People’s Hospital is the clinical service name of the Affiliated People’s Hospital of Jiangsu University.

Patient selection

Inclusion criteria: (1) Age ≥ 18 years; (2) Histologically or clinically confirmed HCC with BCLC stage B or C; (3) ECOG performance status 0–2; (4) Child-Pugh class A or B; (5) At least one measurable lesion per modified RECIST (mRECIST) [13]; (6) Recurrence after prior curative-intent hepatic resection or thermal ablation, confirmed by contrast-enhanced imaging (CT/MRI) or histopathology, with an interval of at least 2 months from the prior curative treatment; (7) Received combination therapy of TACE plus TKIs with or without PD-1 inhibitors.

Exclusion criteria: (1) Incomplete clinical data; (2) Severe comorbidities (brain, cardiovascular, or renal failure); (3) Concurrent malignancies; (4) Disease progression after non-curative systemic or locoregional therapy (rather than recurrence after curative-intent therapy).

Treatment allocation and protocols

Treatment allocation was determined by a multidisciplinary team (MDT) comprising interventional radiologists, hepatobiliary surgeons, medical oncologists, and hepatologists. The MDT reviewed each case weekly based on the following criteria: (1) BCLC stage and tumor burden per the 2022 BCLC strategy; (2) prior systemic therapy line and history (first-line vs. post-sorafenib/lenvatinib failure); (3) Child-Pugh score and ECOG performance status; (4) patient financial capacity and local reimbursement policy for PD-1 inhibitors and TKIs; and (5) patient preference following informed discussion. Decisions were made according to the institutional consensus aligned with the Chinese Society of Clinical Oncology (CSCO) guidelines for HCC (2024 edition) and the 2022 BCLC strategy.

Group A: TACE+Regorafenib (second-line); Group B: TACE+Lenvatinib + PD-1 inhibitor (first-line); Group C: TACE+Regorafenib + PD-1 inhibitor (second-line).

TACE procedure

Conventional TACE (cTACE) was performed by interventional radiologists with ≥ 10 years of experience using a selective or super-selective approach via femoral arterial access. The chemotherapy regimen consisted of epirubicin 40–50 mg mixed with lipiodol (5–15 mL), followed by embolization with polyvinyl alcohol (PVA) particles (300–500 μm) or gelatin sponge particles until near-stasis was achieved. Drug-eluting bead TACE (DEB-TACE) was not used in this cohort. The median lipiodol dose per session was 8 mL (IQR 5–12 mL). TACE was repeated every 4–6 weeks if viable tumor was detected on contrast-enhanced CT or MRI. Discontinuation criteria included:

(1) Child-Pugh C; (2) ECOG > 2; (3) TACE refractoriness, defined as progression after ≥ 2 consecutive TACE sessions within the same arterial territory or the appearance of new extrahepatic lesions; (4) patient refusal.

TKIs and PD-1 inhibitors administration

Regorafenib was started at 160 mg daily (3 weeks on/1 week off) and lenvatinib at 8–12 mg daily according to body weight. PD-1 inhibitors were infused every 3 weeks. The immune checkpoint inhibitors used included sintilimab (n = 6), camrelizumab (n = 13), and tislelizumab (n = 6). Dose modifications followed standard guidelines.

Data collection and follow-up

Baseline characteristics included demographics, liver function, tumor burden, and prior treatment history. Follow-up contrast-enhanced CT or MRI was performed every 6–9 weeks. Tumor responses were evaluated by two independent radiologists using mRECIST [13]. AEs were graded using CTCAE v5.0.

Statistical analysis

Continuous variables were presented as median (IQR) and categorical variables as n (%). Between-group comparisons used Kruskal-Wallis or χ² tests. Survival curves were generated by Kaplan-Meier and compared by log-rank test. Multivariable Cox regression included variables with P < 0.10 in univariate analysis. To avoid overfitting, multivariable models were constructed using a backward selection procedure with a maximum of 1 variable per 5 events, ensuring at least 5 events per variable.

Treatment group (Group B vs. Group A; Group C vs. Group A) was included as a covariate in the multivariable models. Proportional hazards assumptions were tested using Schoenfeld residuals and log-minus-log plots; no significant violations were observed (all P > 0.05). Propensity score matching analyses (1:1 nearest-neighbor matching, caliper 0.2) were performed to compare TACE + TKI vs. TACE + TKI+PD-1. Two-sided P < 0.05 was considered significant. Analyses were performed with SPSS 25.0.

A post-hoc power analysis was performed using G*Power 3.1. Assuming a clinically meaningful hazard ratio of 0.60 for OS with α = 0.05 (two-sided), 80% power would require approximately 120 patients (40 per group). With only 39 patients, this study had approximately 25% power to detect such a difference. For PFS (HR 0.60), approximately 90 patients (30 per group) would be needed; our study had approximately 35% power.

Results

Patient characteristics

Thirty-nine patients were enrolled (Fig. 1). Baseline characteristics were balanced across groups except for prior systemic therapy line (Table 1). All patients had recurrence after prior curative-intent therapy: hepatic resection (n = 28, 71.8%) or thermal ablation (n = 11, 28.2%). Early recurrence (< 2 years) occurred in 23 patients (59.0%) and late recurrence (≥ 2 years) in 16 patients (41.0%).

Fig. 1.

Flow chart

Table 1.

Baseline characteristics of the patients

Characteristics	Total (N = 39)	Group A (n = 14)	Group B (n = 12)	Group C (n = 13)	P-value
Age, years, median (IQR)	62 (54–68)	61 (53–67)	63 (55–70)	62 (54–68)	0.89
< 60	17 (43.6)	6 (42.9)	5 (41.7)	6 (46.2)	> 0.99
≥ 60	22 (56.4)	8 (57.1)	7 (58.3)	7 (53.8)
Sex
Male	30	11	9	10	> 0.99
Female	9	3	3	3
ECOG PS
0	10	3	3	4	0.78
1	24	9	8	7
2	5	2	1	2
Child-Pugh Class
A	31	10	10	11	0.55
B	8	4	2	2
Extrahepatic metastases
Yes	7	2	3	2	> 0.99
No	32	12	9	11
BCLC stage
B	28	10	8	10	> 0.99
C	11	4	4	3
Tumor distribution
Single	10	4	3	3	> 0.99
Multiple	29	10	9	10
Max tumor size, cm
< 10 cm	35	12	11	12	> 0.99
≥ 10 cm	4	2	1	1
Up-to-seven criteria
Within	35	12	11	12	0.89
Beyond	4	2	1	1
Liver cirrhosis
Yes	32	11	10	11	> 0.99
No	7	3	2	2
Etiology
Hepatitis B	31	11	9	11	0.81
Hepatitis C	5	2	2	1
No viral hepatitis	3	1	1	1
AFP, ng/mL
< 400	27	10	8	9	0.78
≥ 400	12	4	4	4
Prior curative treatment
Resection	28	10	8	10	> 0.99
Ablation	11	4	4	3
Recurrence timing
Early (< 2 years)	23	8	7	8	> 0.99
Late (≥ 2 years)	16	6	5	5
Prior systemic therapy line
First-line	12	0	12	0	< 0.001
Second-line	27	14	0	13

ECOG PS Eastern Cooperative Oncology Group performance status, BCLC Barcelona Clinic Liver Cancer, TACE transarterial chemoembolization, PD-1 programmed death-1, AFP alpha-fetoprotein

Group A: TACE+Regorafenib;

Group B: TACE+Lenvatinib + PD-1 inhibitor;

Group C: TACE+Regorafenib + PD-1 inhibitor

The immune checkpoint inhibitors used included sintilimab (n = 6), camrelizumab (n = 13) and tislelizumab (n = 6). χ² test indicated no significant imbalance in agent distribution across Groups B and C (P = 0.65).

Owing to the retrospective nature of the study, drug selection was mainly dictated by local reimbursement policy and physician preference rather than biomarker guidance. Thus, the potential influence of different agents could not be completely excluded.

Tumor response

No complete responses were observed. Partial response occurred in 5 patients (12.8%), stable disease in 15 (38.5%), and progressive disease in 19 (48.7%). ORR and DCR did not differ significantly among groups (Table 2).

Table 2.

Tumor response

Response	Total (n = 39)	Group A (n = 14)	Group B (n = 12)	Group C (n = 13)	P-value
CR	0	0	0	0	-
PR	5 (12.8)	2 (14.3)	2 (16.7)	1 (7.7)	0.78
SD	15 (38.5)	6 (42.9)	4 (33.3)	5 (38.5)	0.91
PD	19 (48.7)	6 (42.9)	6 (50.0)	7 (53.8)	0.89
ORR	12.8%	14.3%	16.7%	7.7%	0.78
DCR	51.3%	57.1%	50.0%	46.2%	0.83

CR complete response, PR partial response, SD stable disease, PD progressive disease, ORR objective response rate, DCR disease control rate

Group A: TACE+Regorafenib;

Group B: TACE+Lenvatinib + PD-1 inhibitor;

Group C: TACE+Regorafenib + PD-1 inhibitor

Survival outcomes

Median OS was 42.9 months (95% CI 27.1–58.7) and median PFS 16.5 months (95% CI 13.6–19.5). Kaplan-Meier curves showed no significant differences in OS (P = 0.164) or PFS (P = 0.830) among the three groups (Fig. 2). The Kaplan-Meier curves almost overlapped during the entire follow-up period, visually reinforcing the absence of measurable benefit from PD-1 inhibitors in this small cohort. After propensity score matching (n = 24), OS remained similar (P = 0.211) (Supplementary Table 1).

Fig. 2.

a Overall Survival. b Progression Free Survival

Prognostic factors

Multi-variable analysis identified BCLC stage B (vs. C) as an independent favorable prognostic factor for both OS (HR 0.03, P = 0.01) and PFS (HR 0.04, P = 0.038) (Table 3). After adjusting for BCLC stage, Child-Pugh class, and other significant covariates, neither Group B (HR 1.42, 95% CI 0.48–4.21; P = 0.52) nor Group C (HR 1.18, 95% CI 0.41–3.39; P = 0.76) was significantly associated with OS compared with Group A. Similarly, no significant PFS differences were observed for Group B (HR 1.08, 95% CI 0.42–2.78; P = 0.88) or Group C (HR 0.95, 95% CI 0.37–2.44; P = 0.91).

Table 3.

Predictors of PFS and OS

	Univariate analysis			Multivariable analysis
	HR	95% CI	P-value	HR	95% CI	P-value
PFS analysis
ECOG PS (2 vs. 1 vs. 0)	0.19	0.03–1.31	0.092	-	-	-
Etiology (HBV vs. HCV)	46.26	1.10–193.11	0.044	11.11	1.49–82.90	0.019
Cirrhosis (Yes vs. No)	0.11	0.02–0.66	0.016	0.67	0.19–2.37	0.531
Child-Pugh (A vs. B)	2.65	0.65–10.76	0.173	-	-	-
BCLC stage (B vs. C)	0.002	0.00–0.12	0.003	0.04	0.01–0.38	0.038
Up-to-seven (> 7 vs. ≤7)	24.05	0.71–817.45	0.077	-	-	-
Portal invasion (Yes vs. No)	4.273	0.46–39.74	0.202	-	-	-
Extrahepatic spread (Yes vs. No)	0.03	0.00–0.35	0.001	0.17	0.02–1.94	0.155
Max tumor size (> 10 vs. ≤10)	0.39	0.03–4.37	0.443	-	-	-
AFP (> 400 vs. ≤400)	0.98	0.32–3.04	0.975	-	-	-
Treatment group
Group B vs. A	0.92	0.36–2.35	0.86
Group C vs. A	1.05	0.41–2.68	0.92
OS analysis
ECOG PS (2 vs. 1 vs. 0)	0.73	0.04–12.81	0.727	-	-	-
Etiology (HBV vs. HCV)	5.53	1.20–25.53	0.038	1.30	0.48–3.55	0.61
Cirrhosis (Yes vs. No)	2.20	0.10–27.83	0.081	-	-	-
Child-Pugh (A vs. B)	0.40	0.03–1.20	0.077	-	-	-
BCLC stage (B vs. C)	0.01	0.00–0.33	0.012	0.03	0.00–0.38	0.01
Up-to-seven (> 7 vs. ≤7)	3.17	0.13–4.44	0.476	-	-	-
Portal invasion (Yes vs. No)	0.68	0.10–4.44	0.682	-	-	-
Extrahepatic spread (Yes vs. No)	0.19	0.00–9.38	0.412	-	-	-
Tumor distribution (Multiple vs. Single)	8.88	1.80–41.80	0.001	3.52	0.91–13.63	0.067
Max tumor size (> 10 vs. ≤10)	6.97	1.80–27.07	0.004	9.03	0.76–19.93	0.100
AFP (> 400 vs. ≤400)	0.85	0.12–6.31	0.875	-	-	-
Treatment group
Group B vs. A	1.35	0.48–3.82	0.56
Group C vs. A	1.12	0.40–3.15	0.83

HR hazard ratio, CI confidence interval, PFS progression-free survival, OS overall survival, BCLC Barcelona Clinic Liver Cancer, ECOG EasternCooperative Oncology Group, AFP alpha-fetoprotein

Group A: TACE+Regorafenib;

Group B: TACE+Lenvatinib + PD-1 inhibitor;

Group C: TACE+Regorafenib + PD-1 inhibitor

Safety

Post-TACE complications included nausea/vomiting (30.7%), pain (58.9%), and fever (51.2%). Grade ≥ 3 AEs occurred in 2 patients (5.1%; Grade 4 gastrointestinal bleeding in Group A and Grade 4 hypersensitivity in Group C). The most common TKI-related AEs were hand-foot skin reaction (41.0%; all Grade 1–2) and hypertension (17.9%; all Grade 1–2). Only one patient permanently discontinued PD-1 inhibitor due to grade 3 rash (Table 4). The median duration of PD-1 inhibitor exposure was 4.2 months (IQR 2.1–8.5) in Group B and 3.8 months (IQR 1.9–7.2) in Group C.

Table 4.

Postoperative complications and adverse events

	Total (n = 39)	Group A (n = 14)	Group B (n = 12)	Group C (n = 13)
Postoperative complications
Nausea and vomiting	12 (30.7%)	4 (28.6%)	4 (33.3%)	4 (30.8%)
Pain	23 (58.9%)	8 (57.1%)	7 (58.3%)	8 (61.5%)
Fever	20 (51.2%)	7 (50.0%)	6 (50.0%)	7 (53.8%)
Adverse events
Grade 3/4	2 (5.1%)	1 (7.1%)	0 (0%)	1 (7.7%)
Hand-foot skin reaction	16 (41.0%)	6 (42.9%)	5 (41.7%)	5 (38.5%)
Gastrointestinal ulcers	3 (7.7%)	1 (7.1%)	1 (8.3%)	1 (7.7%)
Diarrhea	3 (7.7%)	1 (7.1%)	1 (8.3%)	1 (7.7%)
Gastrointestinal reaction	6 (15.4%)	2 (14.3%)	2 (16.7%)	2 (15.4%)
Fatigue	8 (20.5%)	3 (21.4%)	2 (16.7%)	3 (23.1%)
Hypothyroidism	6 (15.4%)	2 (14.3%)	2 (16.7%)	2 (15.4%)
Hypertension	7 (17.9%)	2 (14.3%)	3(25.0)%	2 (15.4%)
RCCEP	7 (17.9%)	2 (14.3%)	3 (25.0%)	2 (15.4%)

RCCEP reactive cutaneous capillary endothelial proliferation

Group A: TACE+Regorafenib; Group B: TACE+Lenvatinib + PD-1 inhibitor

Group C: TACE+Regorafenib + PD-1 inhibitor

Discontinuation reasons for PD-1 inhibitors were: disease progression (n = 18), completion of planned therapy (n = 4), adverse events (n = 1), and death (n = 2).

The median number of TACE sessions was 4 (IQR 3–6) for Group A, 5 (IQR 3–7) for Group B, and 4 (IQR 3–5) for Group C (P = 0.72). A total of 28 patients (71.8%) received subsequent salvage therapies (e.g., re-resection, ablation, or radiotherapy) at progression.

Discussion

Our retrospective cohort demonstrated that TACE combined with TKIs, with or without PD-1 inhibitors, achieved a median OS of 42.9 months and median PFS of 16.5 months in patients with recurrent intermediate-advanced HCC. These survival figures appear favorable compared with some historical data [10–12]. However, we observed no statistically significant difference in OS or PFS among the three treatment groups, indicating that we were unable to detect an additional survival benefit from PD-1 inhibitors in this specific setting.

It is critical to emphasize that this was an exploratory, real-world comparison of three distinct clinical regimens, not a formal assessment of PD-1 inhibitor addition alone. The three groups differed simultaneously in TKI type (lenvatinib vs. regorafenib), treatment line (first-line vs. second-line), and immunotherapy exposure. Therefore, the absence of significant differences cannot be interpreted as evidence of equivalent efficacy or lack of benefit from PD-1 inhibitors. Rather, it reflects the severe underpowering of this small cohort to detect clinically meaningful but modest survival differences.

Our median OS of 42.9 months is numerically higher than the 26.8 months reported by Han et al. for TACE plus regorafenib after sorafenib failure [14]. The discrepancy likely reflects patient selection bias-our cohort comprised more Child-Pugh A patients (79.5%) and a higher proportion of BCLC-B stage (71.8%). Conversely, our ORR (12.8%) and DCR (51.3%) were modest compared with the higher ORR reported in studies combining TACE with lenvatinib plus pembrolizumab [15, 16]. Heterogeneous imaging schedules and differences in tumor burden assessment might have contributed to this difference.

The median OS of 42.9 months is numerically higher than frontline systemic therapy benchmarks (e.g., IMbrave150: 19.2 months; LEAP-002: 26.3 months). This discrepancy likely reflects several sources of selection bias: (1) our cohort was enriched with Child-Pugh A patients (79.5%) and BCLC-B stage (71.8%), representing a better-prognosis population; (2) all patients had recurrent disease after prior curative therapy, which may select for slower tumor biology; (3) the repeated TACE sessions (median 4 per patient, range 1–9) provided ongoing locoregional control; and (4) 28 of 39 patients (71.8%) received subsequent salvage therapies at progression, which may have prolonged survival beyond the initial protocol. Direct comparisons to historical controls should therefore be interpreted with extreme caution due to these favorable selection characteristics.

Grade ≥ 3 adverse events occurred in only 5.1% of patients. Hand-foot skin reaction (41.0%) and hypertension (17.9%) were the most common TKI-related toxicities, though all were Grade 1–2. Only one patient permanently discontinued PD-1 inhibitor because of grade 3 rash. Grade 3/4 AEs were managed by temporary drug interruption (regorafenib held for 7 days; PD-1 permanently discontinued in one patient). Grade 2 hand-foot skin reaction was treated with topical urea-based cream and dose reduction to 80 mg regorafenib. Hypertension was controlled with amlodipine without TKI interruption. Post-procedural symptoms (pain, fever, nausea) were transient and managed conservatively, confirming the overall tolerability of this combination approach.

TACE induces acute ischemia followed by reperfusion injury, releasing tumor-associated antigens and pro-inflammatory cytokines [8]. TKIs inhibit VEGF-mediated re-vascularization and might transiently normalize tumor vessels, thereby improving T-cell trafficking [17]. These mechanisms theoretically support the addition of PD-1 inhibitors. Nevertheless, excessive anti-angiogenic blockade could prune vessels excessively, creating a hypoxic, immunosuppressive niche [18]. In our cohort, the lack of incremental benefit suggested that the balance between vascular normalization and hypoxia was not optimally achieved.

Recent high-level evidence has reshaped the landscape of TACE-based combination therapy. In the phase 3 LEAP-012 trial (Lancet 2025), TACE plus lenvatinib and pembrolizumab significantly improved PFS compared with TACE plus placebo (median 14.6 vs. 10.0 months; HR 0.66, P = 0.002) in intermediate-stage HCC [16]. Similarly, the EMERALD-1 trial (Lancet 2025) demonstrated that TACE plus durvalumab and bevacizumab improved PFS versus TACE plus placebo (median 15.0 vs. 8.2 months; HR 0.77, P = 0.032) [19]. The phase 2 CARES-005 (CHANCE2005) trial reported that TACE plus camrelizumab and rivoceranib achieved a median PFS of 10.8 months versus 3.2 months with TACE alone (HR 0.34, P < 0.001) [20]. Furthermore, the nationwide CHANCE2202 target trial emulation showed that TACE combined with immune checkpoint inhibitors and anti-angiogenic agents improved OS (HR 0.57) and PFS (HR 0.70) compared with TACE monotherapy in intermediate-stage HCC [21]. While these studies establish the benefit of TACE-ICI-anti-angiogenic triple therapy in intermediate-stage HCC, evidence specifically in the recurrent intermediate-advanced setting—where patients have already undergone curative-intent therapy and harbor more aggressive tumor biology-remains limited. Our findings suggest that in this specific recurrent population, the benefit of adding PD-1 inhibitors to TACE + TKI may be less pronounced, possibly due to immune evasion mechanisms established after prior treatment and the high proportion of BCLC-B patients who may derive adequate locoregional control from TACE alone.

Prospective, randomized trials specifically dedicated to recurrent intermediate-advanced HCC are urgently needed. Adaptive platform designs comparing TACE-based regimens, enriched by predictive biomarkers such as ctDNA kinetics or tumor mutational burden, would provide definitive evidence. Until then, treatment decisions should be individualized based on tumor burden, liver function, and patient preference.

Our study had several limitations inherent to its retrospective, single-center design. First, treatment allocation was influenced by physician preference, financial considerations, and governmental reimbursement policies, introducing selection bias. Although propensity score matching was attempted, the present sample size was underpowered to detect a clinically meaningful but modest survival difference and could not correct for the fundamental differences in treatment line (first-line lenvatinib in Group B vs. second-line regorafenib in Groups A and C). Second, heterogeneous PD-1 therapies confounded outcomes, though the small sample size precluded formal sensitivity analyses by agent. Third, the absence of a TACE + lenvatinib group without PD-1 inhibitor precluded direct assessment of lenvatinib’s contribution in Group B. This limitation fundamentally restricts our ability to isolate the specific effect of PD-1 inhibition from that of first-line lenvatinib. Fourth, PD-(L)1 expression and circulating biomarkers were not routinely measured, preventing subgroup analyses. Fifth, because prior treatment line was perfectly collinear with treatment group, it could not be adjusted as an independent covariate, representing a structural confounding that PSM and multivariable analysis cannot fully resolve in a cohort this small. Finally, the single-center design and retrospective nature limit generalizability. These findings should be considered hypothesis-generating descriptive data rather than comparative effectiveness evidence.

Conclusion

In this real-world cohort of recurrent intermediate-advanced HCC, TACE combined with TKIs with or without PD-1 inhibitors yielded favorable survival and acceptable safety. The addition of PD-1 inhibitors did not demonstrate additional survival benefit in this small cohort. These hypothesis-generating findings require prospective validation.

Abbreviations

TACE

Transarterial chemoembolization

TKIs

Tyrosine-kinase inhibitors

PD-1

Programmed death-1

HCC

Hepatocellular carcinoma

OS

Overall survival

PFS

Progression free survival

ORR

Objective response rate

DCR

Disease control rate

BCLC

Barcelona Clinic Liver Cancer

ECOG PS

Eastern Cooperative Oncology Group performance status

HR

Hazard ratio

CI

Confidence interval

AFP

Alpha-fetoprotein

RCCEP

Reactive cutaneous capillary endothelial proliferation

Authors’ contributions

Wei fan Sui: Conceptualization, Methodology, Software, Data curation, Writing Original draft preparation, Writing- Reviewing and Editing. Yun xin Duan, Ze feng Cai and Jian yun Li: Visualization, Investigation. Jian hua Fu: Supervision.

Funding

The study was supported by Jinshan Doctor Project (2025-JSYZ-33).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study protocol was approved by the institutional Ethics Committee of the Affiliated People’s Hospital of Jiangsu University (SQK-2025235-W) and consent to participate was waived by the Ethics Committee due to the retrospective nature of the study.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.