Transarterial Radioembolization Versus Chemoembolization for Unresectable Large Single Hepatocellular Carcinoma

Abstract

Objective

To compare oncologic outcomes and safety between transarterial radioembolization (TARE) and transarterial chemoembolization (TACE) in patients with unresectable large solitary hepatocellular carcinomas (HCCs; ≥6 cm).

Materials and Methods

This single-center retrospective study included patients with a solitary HCC ≥6 cm who underwent initial treatment with TARE or TACE between January 2010 and December 2023. Patients who were initially scheduled to undergo TARE but were subsequently treated with TACE because of a high lung shunt fraction (LSF) were included in the TACE group. The primary outcome was overall survival (OS), and secondary outcomes included progression-free survival (PFS), local PFS, tumor response, and safety. Inverse probability of treatment weighting (IPTW) was used to adjust for baseline differences in the survival analyses.

Results

In total, 306 patients were included in this study (TACE, n = 187; TARE, n = 119). Before IPTW adjustment, the TARE group demonstrated significantly longer OS than the TACE group (median, 79.6 vs. 46.5 months; P = 0.004). Patients who were ineligible for TARE because of extensive lung shunting demonstrated particularly poor survival (median OS, 20.3 months). After IPTW adjustment, TARE remained associated with longer OS (median, 79.6 vs. 53.4 months; P = 0.033) and local PFS (median, 25.1 vs. 9.5 months; P = 0.001). Median PFS remained numerically longer with TARE (9.8 vs. 7.8 months), but the difference did not reach statistical significance (P = 0.066). In comparison with TACE, TARE was associated with less frequent serious adverse events (10.1% vs. 26.7%; P < 0.01) and a markedly lower rate of prolonged hospitalization (>7 days) (0.8% vs. 22.5%; P < 0.01).

Conclusion

In patients with unresectable large solitary HCCs, TARE was associated with longer OS, improved local tumor control, and fewer serious adverse events than TACE. However, the survival advantage should be interpreted with caution, because biological selection related to LSF eligibility may have contributed to the observed difference.

INTRODUCTION

Hepatocellular carcinoma (HCC) is one of the leading causes of cancer-related mortality worldwide. Surgical resection offers the greatest chance of long-term survival in patients with solitary tumors, making it the treatment of choice whenever technically feasible [1]. However, many patients with large solitary HCCs are not surgical candidates because of insufficient future liver remnants, impaired liver function, portal hypertension, unfavorable tumor location, or major medical comorbidities. For these patients, locoregional therapies play a central role, and both transarterial chemoembolization (TACE) and transarterial radioembolization (TARE) are recommended in the Barcelona Clinic Liver Cancer (BCLC) treatment strategy [1].

TACE has long been the standard locoregional treatment for unresectable HCC, and has achieved favorable outcomes in carefully selected patients through repeated superselective embolization [2]. Nevertheless, treatment efficacy decreases with an increase in tumor size because complete tumor necrosis is difficult to achieve and repeated procedures are frequently required [3,4]. In contrast, TARE delivers high-dose internal radiation while producing minimal embolic effects, providing better tolerability and more durable local tumor control. Randomized trials and recent meta-analyses have demonstrated prolonged time to progression, improved quality of life, and favorable survival outcomes after TARE in comparison with TACE in selected patient populations [5,6,7,8].

Despite these encouraging results, important gaps in the evidence remain. Most of the previous comparative studies included patients with early-stage tumors, multifocal disease, or heterogeneous BCLC stages [5,6,7], whereas data specifically evaluating unresectable large solitary HCCs are limited. More importantly, interpretation of comparative outcomes is complicated by the treatment selection bias inherent to TARE. Patients with excessive lung shunting are generally excluded from TARE because of the risk of radiation pneumonitis [9], and are instead treated with TACE [10]. Since a high lung shunt fraction (LSF) has also been associated with aggressive tumor biology and poor prognosis [11], any survival advantage observed with TARE may partly reflect differences in baseline tumor biology rather than the therapeutic effects of the procedure itself. This potential source of bias has received little attention in previous comparative studies. Therefore, the aim of this study was to compare oncologic outcomes and safety between TARE and TACE in patients with unresectable large solitary HCCs (≥6 cm), with consideration of the influence of lung shunt–related treatment selection when interpreting survival outcomes.

MATERIALS AND METHODS

Patients

This single-center retrospective study was approved by the Institutional Review Board of Seoul National University Hospital, and the need for patient consent was waived owing to the retrospective study design (IRB No. 2412-092-1596). The study included all patients with large solitary HCCs who were treated with either TACE or TARE from January 2010 to December 2023 and met the following criteria: 1) TACE or TARE as the initial treatment modality, 2) maximum tumor diameter 6 cm or larger, 3) age ≥18 years, 4) Eastern Cooperative Oncology Group performance status ≤1, 5) Child–Pugh score A or B7. The exclusion criteria were as follows: 1) patients who received prior treatments, such as liver resection, ablation, or systemic therapy, 2) vascular invasion or extrahepatic metastasis, and 3) having another malignant disease or a history of malignancy within 2 years from the beginning of treatment. All patients were diagnosed with HCC on the basis of the American Association for the Study of Liver Diseases criteria [12].

Some patients were initially scheduled to undergo TARE; however, TARE could not be performed because of the high estimated lung dose based on planning angiography and macroaggregated albumin scans. These patients subsequently underwent TACE and were included in the TACE group for the analysis.

Treatments

TACE was performed by eight interventional radiologists under the supervision of a senior faculty member with more than 10 years of experience. Both conventional TACE (cTACE) and drug-eluting bead-TACE (DEB-TACE) were performed on the basis of availability and operator preference. For cTACE, 50 mg of doxorubicin (Ildong, Seoul, Korea) and 10 mL of Lipiodol (Guerbet, Roissy, France) were used as the maximal doses. The tumor-feeding arteries were embolized using gelatin sponge particles. For DEB-TACE, drug-eluting beads (DC-Bead; Boston Scientific, Marlborough, MA, USA) sized 100–300 µm or 300–500 µm were used. A vial of DC-Bead was loaded with 25–50 mg of doxorubicin, and one or two vials were used.

TARE was performed by three interventional radiologists with over six years of experience. Yttrium-90-labeled microspheres (TheraSphere; Boston Scientific or SIR-Spheres; Sirtex, Woburn, MA, USA) were administered as selectively as possible when feasible, and a boosted dose was frequently used [13,14]. The choice of the microsphere type was determined on the basis of the interventional radiologist’s preference. Large HCCs frequently derive their blood supply from extrahepatic collateral arteries; therefore, radioactive microspheres were infused through these collateral vessels as well [15,16]. Dosimetry was retrospectively performed using the commercially available software Simplicit⁹⁰Y (Mirada Medical, Oxford, UK). The mean absorbed dose (Mad) to the treated area was calculated using single-compartment dosimetry, and the tumor absorbed dose (TAD) of the primary target tumor was calculated using multicompartment dosimetry.

Outcomes and Assessment

The primary outcome of this study was overall survival (OS), measured from the date of initial treatment to either the date of death from any cause or the last follow-up. Survival data were sourced from hospital medical records and the national database maintained by the Ministry of the Interior and Safety of Korea; sourcing was performed using resident registration numbers. The secondary outcomes included progression-free survival (PFS), local PFS, best treatment response, and safety. Local PFS was defined as the time from treatment initiation to the first recorded progression of the primary target tumor, whereas overall PFS was defined as the progression of HCC at any location, including extrahepatic sites, or death from any cause, whichever occurred first. Best treatment response was evaluated utilizing the modified Response Evaluation Criteria in Solid Tumors (mRECIST) guideline [17]. In cases involving repeated TACE or TARE, the best treatment response was assessed after the repeated treatment. When a treatment modality different from the initial modality was subsequently administered, the best treatment response was evaluated using computed tomography (CT) or magnetic resonance imaging (MRI) examinations performed before the subsequent treatment.

Serious adverse events (SAEs) were defined as Common Terminology Criteria for Adverse Events (CTCAE, version 5.0) grade ≥3 events occurring within 90 days of treatment [18]. If any subsequent treatment was performed within 90 days, SAEs were recorded immediately before subsequent therapy. In the TARE group, radiation pneumonitis, radioembolization-induced liver disease (REILD), and benign biliary strictures were additionally monitored for up to 6 months. Thirty-day mortality included all deaths within 30 days of initial treatment regardless of the cause of death.

Statistical Analysis

Continuous variables were presented as mean and standard deviation, while categorical variables were reported as frequencies and percentages. Categorical variables were compared using the chi-squared test or Fisher’s exact test, as appropriate. Laboratory variables were dichotomized according to the commonly used normal limits and established liver function assessments. Laboratory tumor markers were dichotomized at 200 units, consistent with their use as prognostic thresholds [19,20].

OS, PFS, and local PFS were obtained using the Kaplan–Meier method and compared between the two treatments by using the log-rank test. Inverse probability of treatment weighting (IPTW) was employed to reduce potential confounding factors and minimize imbalances in baseline characteristics between treatment strategies in the survival analyses. The following covariates were initially considered independent variables: sex, age, tumor size, prothrombin time (PT) (international normalized ratio), PT (%), Child–Pugh class, albumin–bilirubin grade, albumin, bilirubin, aspartate aminotransferase, alanine aminotransferase, creatinine, alpha-fetoprotein (AFP), and protein induced by vitamin K absence (PIVKA)-II levels, platelet count, hepatitis B virus and hepatitis C virus infections, and alcohol consumption. Baseline covariate balance was assessed using standardized mean differences (SMDs) in addition to statistical significance, and variables with an absolute SMD greater than 0.2 before weighting were included in the propensity-score model. Propensity scores were estimated using a logistic regression model, with treatment assignment as the dependent variable. Subsequently, weighted log-rank tests incorporating IPTW-derived weights were performed to evaluate the statistical differences in OS, PFS, and local PFS between the groups.

Additionally, as a sensitivity analysis, multivariable Cox proportional-hazard regression was performed to compare OS between the TARE and TACE groups. Because both the albumin–bilirubin grade and Child–Pugh class are composite measures derived from serum albumin and bilirubin levels, only the albumin–bilirubin grade was entered into the multivariable model to minimize multicollinearity.

For all analyses, a two-sided P-value <0.05 was considered statistically significant. All statistical analyses were conducted using the R software, version 4.5.2 (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Baseline Characteristics

This study included 306 patients, of which 187 (median age, 70 years; interquartile range [IQR], 60–77 years; 157 men) and 119 (median age, 70 years; IQR, 63–78 years; 98 men) patients received TACE and TARE, respectively, as the initial treatment modality. In the TACE group, 39 patients (20.9%) were initially scheduled to undergo TARE but ultimately received TACE because of a high estimated lung dose. The baseline characteristics are summarized in Table 1. The annual numbers and proportions of patients undergoing TACE and TARE are shown in Figure 1. Among the 187 patients who underwent TACE, 153 received cTACE and 34 received DEB-TACE.

Table 1. Baseline characteristics before and after IPTW.

		Before IPTW					After IPTW
		Total (n = 306)	TACE (n = 187)	TARE (n = 119)	SMD	P	TACE (n = 187)	TARE (n = 119)	SMD	P
Sex		306	187	119	-0.043	0.714
	Woman	51 (16.7)	30 (16.04)	21 (17.65)
	Man	255 (83.3)	157 (83.96)	98 (82.35)
Age, yr		306	187	119
	Mean ± SD	68.3 ± 12.1	67.7 ± 13.0	69.1 ± 10.6	0.120	0.453
	Median (min, max)	69.9 (16.7, 91.7)	69.8 (16.7, 91.7)	70.0 (38.0, 89.0)
Tumor size		306	187	119	0.178	0.133
	>80 mm	195 (63.7)	113 (60.4)	82 (68.9)
	≤80 mm	111 (36.3)	74 (39.6)	37 (31.1)
Albumin		306	187	119	-0.368	0.002			0.014	0.911
	<4 g/dL	165 (53.9)	114 (61.0)	51 (42.9)			95.7 (53.5)	62.4 (52.8)
	≥4 g/dL	141 (46.1)	73 (39.0)	68 (57.1)			83.3 (46.5)	55.8 (47.2)
	Bilirubin	305	187	118	-0.188	0.119
	>1.2 mg/dL	40 (13.1)	29 (15.5)	11 (9.3)
	≤1.2 mg/dL	265 (86.9)	158 (84.5)	107 (90.7)
PT (INR)		306	187	119	-0.301	0.013			0.016	0.906
	>1.1	69 (22.6)	51 (27.3)	18 (15.1)			41.7 (23.3)	26.8 (22.6)
	≤1.1	237 (77.5)	136 (72.7)	101 (84.9)			137.3 (76.7)	91.5 (77.4)
PT, %		306	187	119
	Mean ± SD	95.9 ± 13.8	93.9 ± 13.9	99.0 ± 13.1	0.372	0.002	95.64 ± 13.61	95.99 ± 12.90	0.026	0.831
	Median (min, max)	96.0 (57.0, 143.0)	95.0 (57.0, 130.0)	97.0 (69.0, 143.0)
CP class		306	187	119	-0.146	0.220
	A6, B7	65 (21.2)	44 (23.5)	21 (17.6)
	A5	241 (78.8)	143 (76.5)	98 (82.4)
ALBI grade		306	187	119	-0.361	0.003			0.008	0.950
	2	141 (46.1)	99 (52.9)	42 (35.3)			82.0 (45.8)	53.7 (45.4)
	1	165 (53.9)	88 (47.1)	77 (64.7)			96.9 (54.2)	64.5 (54.6)
AST		306	187	119	-0.332	0.005			0.005	0.965
	≥40 IU/L	177 (57.8)	120 (64.2)	57 (47.9)			104.4 (58.3)	68.7 (58.1)
	<40 IU/L	129 (42.2)	67 (35.8)	62 (52.1)			74.6 (41.7)	49.6 (41.9)
ALT		306	187	119	-0.284	0.017			0	1.000
	≥40 IU/L	126 (41.2)	87 (46.5)	39 (32.8)			73.8 (41.2)	48.7 (41.2)
	<40 IU/L	180 (58.8)	100 (53.5)	80 (67.2)			105.2 (58.8)	69.5 (58.8)
Creatinine		306	187	119	-0.206	0.089
	≥1.2 mg/dL	38 (12.4)	28 (15.0)	10 (8.4)
	<1.2 mg/dL	268 (87.6)	159 (85.0)	109 (91.6)
AFP		306	187	119	0.377	0.002			0.015	0.906
	≥200 ng/mL	110 (35.9)	80 (42.8)	30 (25.2)			63.4 (35.4)	41.0 (34.7)
	<200 ng/mL	196 (64.1)	107 (57.2)	89 (74.8)			115.5 (64.6)	77.2 (65.3)
PIVKA		299	180	119	-0.101	0.396
	≥200 mAU/mL	239 (79.9)	141 (78.3)	98 (82.4)
	<200 mAU/mL	60 (20.1)	39 (21.7)	21 (17.7)
Platelet		306	187	119	-0.287	0.018			0.029	0.830
	<130 x103/µL	53 (17.3)	40 (21.4)	13 (10.9)			30.4 (17.0)	18.8 (15.9)
	≥130 x103/µL	253 (82.7)	147 (78.6)	106 (89.1)			148.5 (83.0)	99.4 (84.1)
HBV		306	187	119	-0.056	0.635
	Negative	98 (32.0)	58 (31.0)	40 (33.6)
	Positive	208 (68.0)
HCV		306				0.383
	Negative	292 (95.4)
	Positive	14 (4.6)
Alcohol		306				0.579
	Absent	254 (83.0)
	Present	52 (17.0)

Data are presented as number of patients (%) unless specified otherwise.

IPTW = inverse probability of treatment weighting, TACE = transarterial chemoembolization, TARE = transarterial radioembolization, SMD = standardized mean difference, SD = standard deviation, PT = prothrombin time, INR = international normalized ratio, CP = Child–Pugh, ALBI = albumin bilirubin, AST = aspartate aminotransferase, ALT = alanine aminotransferase, AFP = alpha-fetoprotein, PIVKA = protein induced by vitamin K absence, HBV = hepatitis B virus, HCV = hepatitis C virus

Fig. 1. Yearly proportions (A) and annual numbers (B) of patients receiving TACE as the initial treatment plan (TACE_initial), TACE in TARE-ineligible patients (TACE_{high LSF}), and TARE between 2010 and 2023. TACE = transarterial chemoembolization, TARE = transarterial radioembolization, LSF = lung shunt fraction.

Table 2 provides an overview of TARE treatment. The median Mad was 231 Gy (IQR = 169–318 Gy), and the median TAD on post-treatment PET/CT was 582 Gy (IQR = 374–778 Gy).

Table 2. Radioembolization data of the 119 patients.

Characteristic		Data
Glass		90 (75.6)
Resin		29 (24.4)
Number of treated vessels*		3 (2–4)
	1	16 (13.4)
	2	35 (29.4)
	3	28 (23.5)
	4	20 (16.8)
	5 or more	20 (16.8)
Treated vessel level
	Lobar	14 (11.8)
	Lobar + segmental	8 (6.7)
	Lobar + subsegmental	11 (9.2)
	Lobar + segmental + subsegmental	9 (7.6)
	Segmental	33 (27.7)
	Segmental + subsegmental	43 (36.1)
	Subsegmental	1 (0.8)
Lung shunt fraction, %		4.1 (2.2–6.6)
Lung dose, Gy		8.8 (4.9–17.1)
Total radiation activity delivered, GBq		4.7 (3.3–6.2)
Total liver volume, mL		1,518 (1,268–1,856)
Treated liver volume, mL		877 (672–1,213)
Tumor volume, mL		366 (167–640)
mAD, Gy		231 (169–318)
Pre-treatment TAD on SPECT/CT, Gy		416 (278–588)
Post-treatment TAD on PET/CT, Gy		582 (374–778)

Data are presented as number of patients (%) or median (interquartile range).

^*The full range was 1–9.

mAD = mean absorbed dose to the treated volume, TAD = tumor absorbed dose, SPECT = single-photon emission computed tomography

Survival Outcomes

Before IPTW adjustment, the TARE group demonstrated a significantly longer OS, local PFS, and PFS than the TACE group (Table 3, Fig. 2). Median OS was 79.6 months in the TARE group and 46.5 months in the TACE group (P = 0.004), while median local PFS and PFS were 28.1 versus 9.3 months (P = 0.001) and 11.7 versus 7.1 months (P = 0.017), respectively. Patients who were initially considered for TARE but ultimately underwent TACE because of a high estimated lung dose (TACE_{high LSF} subgroup) had particularly poor outcomes, with median OS, local PFS, and PFS of 20.3, 5.5, and 4.0 months, respectively (Table 3, Fig. 3).

Table 3. OS, local PFS, PFS before and after IPTW.

	Before IPTW						After IPTW
	TARE (n = 119)	TACE (n = 187)	P *	TACEinitial†(n = 148)	TACEhigh LSF†(n = 39)	P ‡	TARE	TACE	P
Median OS	79.6 (NA–NA)	46.5 (29.6–63.4)	0.004	56.0 (35.4–76.6)	20.3 (13.1–27.5)	0.075	79.6 (69.3–NA)	53.4 (40.1–75.6)	0.033
Median local PFS	28.1 (16.1–40.1)	9.3 (6.9–11.7)	0.001	10.2 (7.9–12.4)	5.5 (2.6–8.4)	0.572	25.1 (14.2–NA)	9.5 (7.8–13.6)	0.001
Median PFS	11.7 (7.5–15.8)	7.1 (5.5–8.7)	0.017	8.8 (6.8–10.8)	4.0 (3.3–4.7)	0.005	9.8 (7.9–20.7)	7.8 (6.4–10.2)	0.066

Data are presented as months with the 95% confidence interval in parentheses.

^*For comparing TARE (n = 119) and TACE (n = 187) over the entire follow-up period using the log-rank text, ^‡For comparing TACE_initial (n = 148) and TACE_{high LSF} (n = 39) over the entire follow-up period using the log-rank test, ^†TACE_initial and TACE_{high LSF} indicate those who underwent TACE as initial treatment plan and patients who were initially planned for TARE but ultimately received TACE due to a high estimated lung dose, respectively.

OS = overall survival, PFS = progression-free survival, IPTW = inverse probability of treatment weighting, TACE = transarterial chemoembolization, LSF = lung shunt fraction, TARE = transarterial radioembolization, NA = not applicable

Fig. 2. Kaplan–Meier curves of (A) OS, (B) local PFS, and (C) PFS in patients with hepatocellular carcinoma treated with TACE and TARE before IPTW. OS = overall survival, PFS = progression-free survival, TACE = transarterial chemoembolization, TARE = transarterial radioembolization, IPTW = inverse probability of treatment weighting.

Fig. 3. Kaplan–Meier curves of (A) OS, (B) local PFS, and (C) PFS in patients with hepatocellular carcinoma treated with TACE as initial treatment plan (TACE_initial), TACE in TARE-ineligible patients because of high lung shunt fraction (TACE_{high LSF}), and TARE before IPTW. OS = overall survival, PFS = progression-free survival, TACE = transarterial chemoembolization, TARE = transarterial radioembolization, IPTW = inverse probability of treatment weighting.

After IPTW adjustment, TARE was associated with superior OS and local PFS (Fig. 4). The median OS was 79.6 months for TARE versus 53.4 months for TACE (P = 0.033), and the median local PFS was 25.1 versus 9.5 months, respectively (P = 0.001) (Table 3).

Fig. 4. IPTW–adjusted Kaplan–Meier curves of (A) OS, (B) local PFS, and (C) PFS comparing TARE and TACE. The numbers at risk for each group are weighted numbers. IPTW = inverse probability of treatment weighting, OS = overall survival, PFS = progression-free survival, TARE = transarterial radioembolization, TACE = transarterial chemoembolization.

In the multivariable Cox proportional-hazard regression analysis, TARE remained independently associated with improved OS than with TACE (hazard ratio [HR], 0.624; 95% confidence interval [CI], 0.422–0.924; P = 0.018) (Table 4). Age >70 years, creatinine level >1.2 mg/Dl, AFP level >200 ng/Ml, and PIVKA-II level >200 Mau/Ml were also independently associated with poor OS.

Table 4. Multivariable Cox proportional-hazard regression analysis for overall survival.

Variable		Univariable analysis		Multivariable analysis
		HR (95% CI)	P	HR (95% CI)	P
Main variable
	TARE (vs. TACE)	0.580 (0.401–0.840 )	0.004	0.624 (0.422–0.924)	0.018
Adjustment variables
	Sex, man	1.540 (0.953–2.489)	0.078	1.447 (0.883–2.372)	0.142
	Age >70 years	2.221 (1.610–3.063)	<0.001	2.697 (1.867–3.896)	<0.001
	Tumor size >8 cm	1.684 (1.195–2.371)	0.003	1.460 (0.965–2.210)	0.074
	Albumin <4.0 g/dL	1.799 (1.273–2.542)	<0.001
	Total bilirubin >1.2 mg/dL	1.187 (0.749–1.880)	0.466
	PT INR >1.1	1.297 (0.875–1.922)	0.195
	Child–Pugh class A6/B7	1.436 (1.000–2.063)	0.050
	ALBI grade 2	1.571 (1.150–2.147)	0.005	1.259 (0.913–1.736)	0.161
	AST >40 IU/L	1.352 (0.979–1.867)	0.067	1.192 (0.806–1.765)	0.379
	ALT >40 IU/L	1.195 (0.869–1.645)	0.273
	Creatinine >1.2 mg/dL	2.372 (1.599–3.519)	<0.001	1.825 (1.176–2.834)	0.007
	AFP >200 ng/mL	1.742 (1.271–2.389)	<0.001	2.108 (1.497–2.970)	<0.001
	PIVKA >200 mAU/mL	1.535 (1.000–2.355)	0.050	1.601 (1.008–2.541)	0.046
	Platelet <130 ×103/µL	1.160 (0.762–1.767)	0.489
	HBV	1.036 (0.742–1.447)	0.833
	HCV	0.866 (0.382–1.962)	0.731
	Alcohol	0.756 (0.504–1.133)	0.175

HR = hazard ratio, CI = confidence interval, TARE = transarterial radioembolization, TACE = transarterial chemoembolization, PT = prothrombin time, INR = international normalized ratio, ALBI = albumin bilirubin, AST = aspartate aminotransferase, ALT = alanine aminotransferase, AFP = alpha-fetoprotein, PIVKA = protein induced by vitamin K absence, HBV = hepatitis B virus, HCV = hepatitis C virus

Best Tumor Response

The best tumor response in the TARE group was complete response (CR) in 35 patients (29.4%), partial response (PR) in 69 patients (58.0%), and stable disease (SD) in 15 patients (12.6%). The best tumor responses in the TACE group were CR in 77 patients (41.2%), PR in 98 patients (52.4%), SD in four patients (2.1%), progressive disease in four patients (2.1%), and no follow-up imaging in four patients (2.1%).

In the TARE group, 26 (21.8%) patients underwent subsequent resection 2–7 months after TARE. Immediately before resection, the best tumor responses were CR (n = 9) and PR (n = 17) in these 26 patients. In the TACE group, 13 patients (7.0%) underwent resection, and the best tumor responses were CR (n = 5) and PR (n = 8) immediately before resection. The proportion of patients who subsequently underwent surgical resection was significantly higher in the TARE group than in the TACE group (P < 0.001).

Safety Assessment

The TARE group showed a significantly lower incidence of treatment-related SAEs than the TACE group (Table 5). SAEs occurred in 12 patients (10.1%) and 50 patients (26.7%) in the TARE and TACE groups, respectively. Forty-two patients (22.5%) in the TACE group required long hospitalization (>7 days) after treatment, but only one (0.8%) in the TARE group required hospitalization over 7 days.

Table 5. Thirty-day mortality and SAEs until 3 months after treatment or before subsequent treatment.

		TACE (n = 187)	TARE (n = 119)	P
30-day mortality		1 (0.5)	0 (0)	1.00
Hospital stay >7 days after treatment		42 (22.5)	1 (0.8)	<0.01
No. of patients with at least one SAE		50 (26.7)	12 (10.1)	<0.01
	Total no. of SAEs	60	14
	Postembolization syndrome	47	3
	Hepatic encephalopathy	3
	Bacteremia	3
	Acute kidney injury	2
	Urinary tract infection	1
	Myocardial infarction	1
	Jaundice	1
	Liver abscess	1
	Pneumonia		1
	Radiation pneumonitis		4
	Benign biliary stricture		5
	Radioembolization-induced liver disease		1
	Death	1

Data are presented as number of patients (%).

SAE = serious adverse event, TACE = transarterial chemoembolization, TARE = transarterial radioembolization

In the TARE group, symptomatic radiation pneumonitis, benign biliary stricture, and REILD occurred in four (3.4%), five (4.2%), and one (0.8%) patients, respectively. Four patients developed symptomatic radiation pneumonitis 2–5 months after the treatment. All patients were treated with corticosteroids, and their symptoms improved subsequently. Five patients developed benign biliary strictures 1–6 months after treatment. One patient underwent percutaneous transhepatic biliary drainage (PTBD) three months after treatment and remained dependent on PTBD until death two years later. The remaining four patients underwent endoscopic retrograde cholangiopancreatography with biliary stent placement. One patient was lost to follow-up 6 months after treatment. Two patients underwent curative surgical resection at 6 and 13 months after treatment, at which time the biliary stents were removed. One patient underwent periodic biliary stent exchange until death from tumor progression 3 years after treatment. One patient developed REILD, presenting with ascites one month after treatment and deterioration of liver function to Child–Pugh class B. The patient subsequently died of coronavirus disease 6 months after treatment.

DISCUSSION

This single-center retrospective study compared clinical outcomes between TARE and TACE in patients with unresectable large solitary HCCs (≥6 cm). The principal findings were that TARE was associated with a significantly longer OS and local PFS than TACE after IPTW adjustment, while also demonstrating a substantially more favorable safety profile. However, interpretation of the survival advantage requires caution because patients who were ineligible for TARE due to a high LSF and subsequently underwent TACE exhibited particularly poor outcomes, suggesting that biological selection may have contributed to the observed differences.

After IPTW adjustment, TARE was associated with a significantly longer OS than TACE (79.6 vs. 53.4 months, P = 0.033). Before IPTW adjustment, the survival difference was even greater (79.6 vs. 46.5 months, P = 0.004). However, these findings should be interpreted in the context of lung shunt–based treatment selection. In this study, 39 patients initially considered for TARE ultimately underwent TACE because of a high estimated lung dose. These patients demonstrated particularly poor survival, with a median OS of only 20.3 months. Similarly, previous studies have shown that patients with a high LSF have unfavorable outcomes regardless of the treatment modality, suggesting that a high LSF may serve as a surrogate marker for aggressive tumor biology [10,11,21]. Extensive arterioportal shunting, large dysmorphic intratumoral vessels, and advanced tumor vascular remodeling may contribute to elevated lung shunting and a poor prognosis. Therefore, the observed survival advantage of TARE may reflect not only the treatment effect, but also the biological selection associated with TARE eligibility.

However, the potential genuine therapeutic benefits of TARE should not be ignored. A recent meta-analysis demonstrated a significant reduction in mortality with TARE than with TACE (pooled HR, 0.68; 95% CI, 0.55–0.86) [8]. Our findings are consistent with these data and suggest that TARE may provide meaningful survival benefits in patients with low lung shunting. Future prospective studies that account for lung shunt status, e.g., by randomizing patients who have been proven eligible for both treatments, are necessary to better define the true survival benefit attributable to the treatment itself.

The superiority of TARE for local tumor control seemed more robust and less susceptible to selection bias than the OS findings. Even after IPTW adjustment, the median local PFS was 25.1 months in the TARE group and 9.5 months in the TACE group (P = 0.001). These results are consistent with those of previous randomized trials and meta-analyses demonstrating prolonged tumor control with TARE [6,7,8]. Salem et al. [6] similarly reported a significantly prolonged time to progression favoring TARE (>26 months) over cTACE (6.8 months) in mainly BCLC stage A small tumors. The TRACE trial demonstrated a median time to progression of 17.1 months with TARE in comparison with 9.5 months with DEB-TACE (P = 0.002), along with improved OS at 30.2 months versus 15.6 months in mainly BCLC stage B tumors [7]. A recent meta-analysis demonstrated that TARE significantly reduced the hazard of death in comparison with TACE (HR, 0.68; 95% CI, 0.55–0.86; P < 0.001) and provided superior PFS (HR, 0.54; 95% CI, 0.44–0.67; P < 0.001) [8]. The superior local tumor control observed with TARE may be explained by the delivery of ablative radiation doses to the tumor, resulting in more durable tumor destruction than that achieved with the combined ischemic and chemotherapeutic effects of TACE.

An intriguing finding of this study was the apparent paradox between radiological CR and clinical outcomes. Although TACE achieved a higher CR rate than TARE (41.2% vs. 29.4%), TARE resulted in a significantly higher rate of subsequent surgical resection (21.8% vs. 7.0%, P < 0.001) and superior local PFS. A similar discordance between radiologic response and long-term clinical outcomes was reported in the TRACE trial [7], wherein objective response rates were comparable between TARE and DEB-TACE despite the significantly improved progression-related outcomes with TARE. This discrepancy is likely due to the fundamentally different response kinetics of the two treatment modalities. In the TACE group, repeated treatment sessions were performed until radiologic CR was achieved, enabling many patients to achieve CR after multiple procedures. In contrast, the tumor response after TARE is typically more gradual but more durable. Consequently, many patients were classified as having a PR at the time when additional treatments, such as surgical resection or salvage TACE, were performed before radiologic CR could be documented, thereby lowering the observed CR rate despite favorable tumor control. Furthermore, TARE effectively promoted contralateral liver hypertrophy [22]. This hypertrophic effect can convert patients who are initially unresectable because of an insufficient future liver remnant into candidates for curative hepatectomy, providing a plausible explanation for the significantly higher rate of subsequent surgical resection observed in the TARE group. These findings suggest that radiologic CR alone may underestimate the clinical benefit of TARE and should be interpreted alongside clinically meaningful outcomes, such as local tumor control, conversion to resection, and long-term survival.

The safety advantages of TARE were substantial and clinically meaningful. Among patients who underwent TACE, 26.7% experienced at least one SAE, while the corresponding proportion among those who underwent TARE was 10.1% (P < 0.01), and 22.5% of patients who underwent TACE required hospitalization for more than 7 days, in comparison with only 0.8% of those who underwent TARE (P < 0.01). Post-embolization syndrome accounted for the majority of SAEs in the TACE group (47 of 60 total SAEs), whereas it was rare in the TARE group (3 of 14). These findings are consistent with the widely reported favorable tolerability profile of TARE [6,7,8,23,24], which is due to its minimal embolic effect. Importantly, TARE-specific complications, including radiation pneumonitis (3.4%), benign biliary stricture (4.2%), and REILD (0.8%), were infrequent and manageable, reinforcing the safety of radiation-based approaches in this population.

This study had several limitations. First, the retrospective design introduced potential selection bias, although IPTW adjustment was applied and the baseline covariates were well-balanced after weighting. Patients expected to benefit from TARE may have been preferentially selected based on physician judgment and the higher cost of radioembolization. Second, both glass and resin microspheres were used with varying dosimetry approaches; however, this reflects the inherent heterogeneity in real-world practice. Third, the 14-year study period (2010–2023) spans an era of evolving TARE techniques, including the adoption of personalized dosimetry, boosted dosing, and superselective administration, which may have improved the outcomes for patients treated in later years [25]. Fourth, during the early study period (2010–2012), only TACE was performed, and TARE was not available. As a result, patients from this era, including those with a high LSF and biologically aggressive tumors, were allocated exclusively to the TACE group. This temporal and biological selection may have biased the comparison in favor of TARE despite the IPTW adjustment. Fifth, both cTACE and DEB-TACE were performed in the TACE group; however, no definitive evidence suggests the superiority of one over the other. This heterogeneity should be acknowledged. Finally, the effect of subsequent therapies, including resection, systemic therapy, and additional locoregional treatments, on OS could not be fully controlled.

In conclusion, in patients with unresectable large solitary HCCs, TARE was associated with longer OS, superior local tumor control, and fewer treatment-related SAEs than TACE. However, the observed benefit should be interpreted with caution because the eligibility for TARE is influenced by the LSF, and patients with a high LSF demonstrated particularly poor outcomes.

Availability of Data and Material

The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.