This retrospective study was conducted using data from patients with HCC and PVTT who were administered SBRT plus lenvatinib and sintilimab, with or without TAE, at Hospital A and Hospital B between June 2018 and December 2022. This study was approved by our Institutional Review Board. The requirement for informed consent was waived owing to the retrospective nature of the study. The study inclusion criteria were as follows: (1) histologically or radiographically confirmed HCC with PVTT; (2) treatment with lenvatinib and sintilimab as the first-line systemic therapy following SBRT; (3) interval between SBRT and TAE ≤ 2 weeks; (4) age ≥18 years; (5) Eastern Co-operative Group (ECOG) performance score 0–1; and (6) Child–Pugh class A or B. The exclusion criteria were as follows: (1) lenvatinib and sintilimab ≤2 cycles; (2) main portal vein completely blocked by tumor thrombus; (3) extrahepatic metastases; (4) other malignancies; (5) incomplete medical information; and (6) loss to follow-up.

Four to six fiducial markers were implanted to determine the SBRT location using computed tomography (CT) simulation images (Accuray Inc., Sunnyvale, CA, USA). The planning target volume expanded the gross tumor volume (GTV) by 0.3–0.5 cm and avoided the organs at risk (OARs). The prescribed dose was 35–56 Gy/5–10 fractions. For dose constraints, the volume of liver receiving high dose radiation was minimized, particularly in cases with scattered intrahepatic satellite lesions. The tolerance doses of the OARs were determined according to the American Association of Physicists in Medicine TG-101 report (17).

Oral lenvatinib was administered within a week of the completion of SBRT at a dose of 8 mg (bodyweight ≤60 kg) or 12 mg (bodyweight >60 kg) once a day. Sintilimab (a programmed death-1 inhibitor) was administered intravenously on the first day of lenvatinib treatment at a dose of 200 mg every 3 weeks. Systemic therapy was not interrupted until disease progression, intolerable side effects, or serious AEs occurred.

TAE was performed according to a previously described method within 2 weeks before or after SBRT initiation (18) In brief, hepatic angiography was performed using a common femoral approach to locate the tumor, and 8–10 mL of lipiodol (Guerbet Pharmaceuticals, Paris, France) was slowly injected to embolize all vessels supplying the target tumor. TAE is commonly used in the following situations: (1) lesions protruding outward or into the liver capsule; (2) intratumoral hemorrhage, and (3) intrahepatic scattered lesions not covered by the target volume. TAE was not performed in the following cases: (1) main portal vein obstruction and high tumor burden, and the possibility of combination therapy leading to the deterioration of liver function; and (2) blood supply to the tumor being poor in the arterial phase of imaging and embolization being ineffective.

The first follow-up was 6–8 weeks after SBRT, and subsequently, every 2–3 months until death or until December 2024. The follow-up included a physical examination, laboratory tests (such as routine blood tests, liver function, and serum tumor markers), and imaging (contrast-enhanced CT or magnetic resonance imaging of the abdomen and lung CT). The primary endpoints were progression-free survival (PFS) and OS; the secondary endpoint was local control (LC). PFS was defined as the duration from the date of SBRT initiation to the date of tumor progression or death from any cause. OS was defined as the time from SBRT initiation to death from any cause or the last follow-up. LC was defined as the duration between the initial date of SBRT and the date of irradiated tumor progression. Local tumor response was assessed using the modified Response Evaluation Criteria in Solid Tumors (19). ORR and disease control rate (DCR) were defined as the rates of complete response (CR) + partial response (PR) and CR + PR + stable disease, respectively. The tumor burden score (TBS) was calculated using the maximum diameter and number of intrahepatic tumors (20, 21). Treatment-related AEs (TRAEs) were graded according to the Common Terminology Criteria for Adverse Events (version 5.0). Additionally, patients were evaluated for radiation-induced liver disease (RILD) (22, 23).

Propensity score matching (PSM) analysis mitigated potential confounders and selection bias. The propensity score was estimated using a multivariate logistic regression model, and 1:1 matches between the two groups were performed using the nearest-neighbor method (caliper width=0.02). Variables included age, sex, virus infection, ECOG performance score, alpha-fetoprotein (AFP), Child–Pugh class, TBS, and PVTT type, and GTV includes only intrahepatic main lesion was considered.

Mann–Whitney–Wilcoxon test (continuous variables) and the chi-square test or Fisher’s exact test (discrete variables) were used to compare baseline characteristics, treatment responses, and TRAEs between the two groups. PFS, OS, and LC were analyzed using the Kaplan–Meier method and compared using the log-rank test and Cox regression model. All data were analyzed using R (version 4.0.3, Vienna, Austria) and Statistical Package for the Social Sciences (SPSS, version 25; IBM Corp., NY, US), and a p-value <0.05 indicated a statistically significant difference.

Between June 2018 and December 2022, 204 of 256 Barcelona Clinic Liver Cancer stage C patients with PVTT who were administered SBRT combined with lenvatinib and sintilimab met the inclusion criteria. Of these, 114 patients (55.9%) who underwent SBRT combined with lenvatinib, sintilimab, and TAE were assigned to the TAE group, while 90 patients (44.1%) treated with SBRT plus lenvatinib and sintilimab alone were included in the non-TAE (NTAE) group (Figure 1). The median ages were 57 years (range, 30–79 years) and 57 years (range, 30–81 years) in the TAE and NTAE groups, respectively. Male patients accounted for 90.4% (103/114) and 80.0% (72/90) of the TAE and NTAE groups, respectively. A TBS ≥8 was observed in 36.0% (41/114) and 45.6% (41/90) of patients in the TAE and NTAE groups, respectively. In the TAE group, PVTT types were classified as I (8.8% [10/114]), II (61.4% [70/114]), and III (29.8% [34/114]), and in 49.1% (56/114) of patients, the GTV was defined solely as the intrahepatic main lesion. In the NTAE group, the PVTT types included I (10.0% [9/90]), II (48.9% [44/90]), III (30.0% [27/90]), and IV (11.1% [10/90]), with 50.0% (45/90) targeting the intrahepatic main lesion alone. Of the 114 patients receiving TAE, the procedure preceded SBRT initiation in 42 (36.8%) and followed it in 72 (63.2%) patients, with all administrations adhering to the predefined 2-week peri-SBRT window. After PSM, the baseline characteristics were well-balanced between the two groups (Table 1).

The ORR and DCR remained consistent before and after PSM (pre-PSM vs. post-PSM: TAE group ORR, 46.5% vs. 40.6%; DCR, 83.3% vs. 79.7%; NTAE group ORR, 42.2% vs. 39.1%; DCR, 79.7% vs. 81.3%). In the post-PSM cohort, CR was achieved in 16.4% (10/64) and PR in 25.0% (16/64) of the patients in the TAE group, compared to 6.2% (4/64) CR and 32.8% (21/64) PR in the NTAE group. No significant differences were observed in the ORR (40.6% [95% CI: 28.3–53.0%] vs. 39.1% [26.8–51.3%], p=0.861) or DCR (79.7% [69.6–89.8%] vs. 81.3% [71.4–91.1%], p=0.819) between the two groups (Table 2).

At the cutoff (December 2024), the median follow-up duration was 59.0 months (95% CI: 39.8–78.2 months; TAE group: 63.0 months [59.8–66.2]; NTAE group: 39.0 months [34.7–43.3]). Disease progression or death occurred in 96.9% (62/64) and 89.1% (57/64) of patients who underwent TAE and NTAE, respectively. The TAE group demonstrated a significantly longer median progression-free survival (mPFS) than the NTAE group (11.0 months [95% CI: 8.4–13.6] vs. 6.0 months [5.0–7.0]; hazard ratio [HR]=0.71, p=0.044), corresponding to a 29% reduction in progression risk (Figure 2a). Median overall survival (mOS) was comparable between the groups (TAE: 19.0 months [14.3–23.7]; NTAE: 18.0 months [14.9–21.1]; HR = 0.90, p=0.606; Figure 2b). The TAE group exhibited a trend toward improved median local control (mLC) (51.0 months [25.7–76.3] vs 36.0 months [24.1–47.9]; HR = 0.54, p=0.066; Figure 2c). The 1-, 2-, and 3-year LC rates were 56.3%, 31.3%, and 17.2%, respectively, in the TAE group, and 48.4%, 23.4%, and 9.4%, respectively, in the NTAE group.

Multivariate analysis (MVA) identified the addition of TAE as an independent prognostic factor for prolonged PFS (HR = 0.52 [95% CI: 0.36–0.76]) and LC (HR = 0.34 [0.16–0.74]). Radiation targeting solely intrahepatic main lesions independently increased risks of disease progression (HR = 1.73 [1.19–2.52]), mortality (HR = 1.62 [1.08–2.44]), and local failure (HR = 2.47 [1.13–5.41]). AFP ≥400 ng/mL was associated with worse OS (HR = 1.73 [1.13–2.64]) and LC (HR = 2.33 [1.06–5.11]) (Table 3).

The TRAEs are summarized in Table 4. We analyzed both any grade and grade ≥3 TRAEs to compare the safety of SBRT combined with lenvatinib and sintilimab with that without TAE. No treatment-related death occurred during the study period. Grade ≥1 TRAEs were reported in 96.9% (124/128) patients. The most common TRAEs in the TAE group were decreased white blood cell count (46.9%), decreased lymphocyte count (34.4%), and rash (32.9%), whereas patients in the NTAE group frequently experienced rashes (48.5%), decreased lymphocyte count (42.2%), and lowered white blood cell count (39.1%). No significant differences were observed in the incidence of TRAEs of any grade or type between the two groups. Grade ≥3 TRAEs occurred in 32 patients (50.0%) in the TAE and NTAE groups. The addition of TAE did not increase the overall incidence of TRAEs. RILD occurred in 12 (18.8%) and 11 patients (17.2%) in the TAE and NTAE groups, respectively.