This retrospective study included HCC patients who underwent radical resection at the Hepatic Surgery Center, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology between January 2019 and December 2021. This study was in line with the requirements of the Helsinki Declaration and was approved by the hospital’s Ethics Committee (TJ-IRB20230127). Due to the retrospective nature of this study, the committee abandoned the requirement of informed consent for all patients.

The inclusion criteria for this study were as follows: (1) age ≥ 18 years; (2) histopathology confirmed HCC; (3) all patients underwent radical hepatectomy according to the Chinese guidelines for diagnosis and treatment of HCC; (4) not receiving any anti-tumor treatment before radical hepatectomy; (5) Child-Pugh grade A or B7. The exclusion criteria for this study were as follows: (1) number of tumors ≥ 4 or extrahepatic metastasis; (2) spontaneous rupture of tumor; (3) PAT with other adjuvant strategies, such as TKIs or anti-PD-1 antibodies monotherapy, TACE; (4) loss to follow-up.

The demographic and clinical features were extracted from the electronic medical record system in our hospital. Vascular invasion was defined as concomitant PVTT or MVI. PVTT mainly included the following types in this study: type I (vp1): tumor thrombus invading the tertiary branch of the portal vein; type II (vp2): tumor thrombus invading the secondary branch of the portal vein; type III (vp3): tumor thrombus invading the primary branch (left or right branch) of the portal vein (24). MVI was defined as nests of cancer cells in the lumen of blood vessels lined by endothelial cells observed microscopically, usually detectable in adjacent liver tissue (25, 26). Satellite nodules were defined as small tumor foci that appear macroscopically or microscopically within the peritumoral liver tissue. The distance between the tumor foci and the primary tumor was < 2 cm. Satellite nodules are considered as intrahepatic micrometastases that occurred on the basis of MVI. They could be classified as MVI, when it was challenging to pathologically distinguish satellite nodules and MVI within the pericancerous liver tissues (26). In this study, HRRFs were defined as including at least one of the following factors: tumor diameter > 5 cm, multiple tumors, satellite nodules, or vascular invasion.

All enrolled patients underwent radical hepatectomy, which was defined as: (1) complete resection of tumor nodules found by preoperative imaging or intraoperative exploration, and no cancer cells or residual cancer tissues were detected at the surgical margin of the submitted specimens; (2) no gross tumor thrombus in the main portal vein, hepatic vein, or bile duct, and without intrahepatic or extrahepatic metastasis; (3) no recurrence of HCC within two months after surgery, and for patients with baseline alpha-fetoprotein (AFP) positive, AFP-levels decreased to the normal reference range (26, 27). All operations were performed by the same experienced surgeon with the assistance of other members of the medical team.

All patients received a careful preoperative evaluation. The Eastern Cooperative Oncology Group performance status score assessed the patient’s general condition, and necessary examinations were performed to evaluate the functioning of the vital organs, such as the patients’ heart, lungs and kidneys. The tumor status (location and stage) was judged by color Doppler ultrasound, contrast-enhanced computed tomography (CT), or magnetic resonance imaging (MRI). Indocyanine green retention for 15 min and Child-Pugh grading estimated the liver function reserve. Wide or narrow surgical margin was defined as the shortest distance from the tumor margin to the hepatectomy plane ≥ 1 cm (wide margin) or < 1 cm (narrow margin) (28). Major hepatectomy was defined as the resection of 3 or more liver segments, and minor hepatectomy was defined as the resection of 1 or 2 liver segments (29).

HCC patients with HRRFs would be recommended adjuvant therapies after radical hepatectomy, and the decision ultimately depended on the patient’s own wishes (26). Patients who did not receive adjuvant therapy after hepatectomy were routinely managed according to the guidelines (26, 30). For patients who chose PAT with TKIs and anti-PD-1 antibodies, the TKIs included sorafenib, lenvatinib, donafenib, regorafenib, and apatinib. The anti-PD-1 antibodies included pembrolizumab, sintilimab, camrelizumab, toripalimab, and tislelizumab. TKIs and anti-PD-1 antibodies were administered according to the recommended dosages from four weeks after surgery until HCC recurrence or serious adverse events, or until the patient withdrew automatically. Generally, 3 weeks was taken as one course, and patients in PAT group received at least 3 courses of treatment. Intermittent or reduced dosage was allowed during treatment to reduce drug-related toxicities. Adverse events were classified according to the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0.

All patients were followed up regularly after radical hepatectomy to monitor recurrence, survival status and drug-related toxicities. In the first year of the surgery, blood biochemical (mainly including liver function and AFP), abdominal ultrasound or contrast-enhanced CT/MRI, chest radiograph, or lung CT were performed every two months. In the second year, routine reviews were conducted every 3-4 months and then every half a year. Recurrence was diagnosed according to the typical imaging findings of HCC and/or persistently elevated serum AFP-levels (26). Early recurrence was defined as HCC recurrence within two years after radical resection (31).

Recurrence-free survival (RFS) was defined as the interval between receiving radical hepatectomy and the first diagnosis of recurrence, or the last follow-up. Overall survival (OS) was defined as the period from surgery to the date of death or the last follow-up. The last follow-up date for this study was on August 1, 2022.

Statistical analysis in this study was performed using R software version 4.2.0 (http://www.R-project.org). Continuous variables with normality were presented as mean ± standard deviation (SD), and those without normality were reported as medians and interquartile ranges (IQR). Comparisons between continuous variables were performed using the independent samples t-test or Mann-Whitney U test, as deemed appropriate. Categorical variables were described as numbers (n) or percentages (%) and compared using the appropriate Chi-square test or Fisher’s exact test. A 1:1 propensity score matching (PSM) was performed to adjust for confounding factors between the two groups. The continuous propensity scores from 0 to 1 were generated by binary logistic regression with selected variables. Nearest-neighbor matching between the PAT and non-PAT groups was done to choose patients for subsequent analyses. The pairs on the propensity-score logit were matched to within a range of 0.2 of SD. Survival curves were created and compared between groups using the Kaplan-Meier method and the log-rank test. Independent prognostic factors for RFS and OS were identified by univariable and multivariable Cox regression analyses. Variables with P < 0.05 in the univariable Cox regression were entered into multivariable Cox regression for further analysis. Subgroup survival analyses were conducted using univariable Cox regression, stratified by different clinical variables (gender, age, AFP, number of tumors, tumor diameter, satellite nodules, grade, and vascular invasion), and forest plots were drawn with hazard ratio (HR) and 95% confidence interval (CI). The differences between groups with a two-tailed P < 0.05 were considered statistically significant.

A total of 315 HCC patients underwent hepatectomy between January 2019 and December 2021, of whom 65 cases were excluded, including intrahepatic/extrahepatic metastasis (n= 8), spontaneous rupture of HCC (n= 12), previous received hepatectomy, local or systemic treatments (n= 23), PAT with other adjuvant treatments (n= 13), and loss to follow-up (n= 9). Finally, 250 eligible patients were included in this study ( Figure 1 ). The clinicopathological characteristics of enrolled patients being depicted in Table S1 . In these patients, age was 54.3 ± 11.6 years, 89.2% cases were male, 16.8% cases were with multiple tumors, 51.2% cases were with tumor diameter > 5 cm, 25.2% cases were with satellite nodules, and 30.8% cases were with vascular invasion.

As shown in Table 1 , among the 174 HCC patients with HRRFs before PSM, 47 (27%) were assigned to the PAT group, and 127 (73%) were assigned to the non-PAT group. Significant differences in four variables (AFP, satellite nodules, vascular invasion, and extent of resection) (all P < 0.05) were observed upon comparing the baseline characteristics of the two groups. Considering that these confounding factors may interfere with the comparative analysis of survival outcomes between the two groups, PSM was conducted in this study. There were 47 pairs of patients with HRRFs in the PAT and non-PAT groups after PSM, with no significant difference between the two groups (all P > 0.05).

The median follow-up for the 250 HCC patients was 22.4 (IQR: 14.3-34.1) months, including 84 (33.6%) recurrences and 39 (15.6%) deaths; and the 90-day mortality was 0 ( Table S1 ). As shown in Figures 2A, B , the RFS and OS of patients without HRRFs were significantly longer than those with HRRFs (all P < 0.001). Moreover, before PSM, there was no significant difference in RFS and OS between PAT and non-PAT groups of HCC patients with HRRFs. However, the OS of the PAT group of patients tended to be longer than those of the non-PAT group (all P > 0.05, Figure S1 ). This observation might be attributed to the imbalance of baseline characteristics of the two groups of patients. In contrast, the PAT group patients had higher tumor malignancy (AFP ≥ 400 ng/mL, satellite nodules, and vascular invasion) than those in the non-PAT group.

After PSM, there were 17 patients (36.2%), 35 patients (74.5%) recurrences (P < 0.001), and 4 patients (8.5%), 23 patients (48.9%) deaths (P < 0.001) in the in PAT and non-PAT groups, respectively. Patients in the PAT group had significantly longer RFS and OS than those in the non-PAT group (all P < 0.05). The 1- and 2-year RFS rates of the patients in the two groups were 82.1% vs. 40.0% (P < 0.001) and 54.2% vs. 25.1% (P = 0.012), respectively; and the corresponding 1- and 2-year OS rates were 95.4% vs. 69.8% (P = 0.001) and 84.3% vs. 55.5% (P = 0.014), respectively ( Figure 3 , Table 2 ).

To further explore the potential value of PAT for improving RFS and OS, subgroup analyses were performed. The results indicated that PAT could significantly improve RFS in HCC patients with multiple tumors (HR: 0.30, 95% CI: 0.09-0.98, P = 0.046), tumor diameter > 5 cm (HR: 0.45, 95% CI: 0.24-0.86, P = 0.015), satellite nodules (HR: 0.32, 95% CI: 0.15-0.70, P = 0.004) or vascular invasion (HR: 0.36, 95% CI: 0.17-0.76, P = 0.007) ( Figure 4A ).

In addition, PAT significantly improved OS in HCC patients with tumor diameter > 5 cm (HR: 0.16, 95% CI: 0.05-0.53, P = 0.003), satellite nodules (HR: 0.13, 95% CI: 0.03-0.56, P = 0.007) or vascular invasion (HR: 0.15, 95% CI: 0.03-0.67, P = 0.013) ( Figure 4B ).

PAT significantly prolonged the RFS (P = 0.034), and relatively extended the OS (P = 0.053) for HCC patients with multiple tumors. The 1-year RFS rate of patients in the PAT and non-PAT groups was 75% and 28.8%, respectively (P = 0.015); and the 1- and 2-year OS rates were 100% vs. 59.2% (P = 0.006) and 66.7% vs. 32.9% (P = 0.227), respectively ( Figures 5A, B , Table 2 ).

PAT significantly improved the RFS and OS for HCC patients with tumor diameter > 5 cm (P= 0.012 and P < 0.001, respectively). The 1- and 2-year RFS rates of patients in the two groups were 78.7% vs. 38.2% (P < 0.001) and 46.5% vs. 20.9% (P = 0.048), respectively; while the 1- and 2-year OS rates were 96.7% vs. 61.8% (P < 0.001) and 82.2% vs. 48.7% (P = 0.018), respectively ( Figures 5C, D , Table 2 ).

The RFS and OS of HCC patients with satellite nodules in the PAT group were significantly longer than those in the non-PAT group (P = 0.003 and P = 0.001, respectively). The 1- and 2-year RFS rates of patients in the two groups were 82.6% vs. 25% (P < 0.001) and 32.8% vs. 15% (P = 0.220), respectively. The 1- and 2-year OS were 100% vs. 58.3% (P < 0.001) and 71.1% vs. 35.9% (P = 0.093), respectively ( Figures 6A, B , Table 2 ).

The RFS and OS of patients in the PAT group were significantly longer than those of HCC patients with vascular invasion in the non-PAT group (P = 0.005 and P = 0.004, respectively). The 1- and 2-year RFS rates were 80.2% vs. 40% (P < 0.001) and 62.2% vs. 32% (P = 0.028), respectively. The 1- and 2-year OS rates were 93.5% vs. 71.4% (P = 0.022) and 93.5% vs. 56.5% (P < 0.001), respectively ( Figures 6C, D , Table 2 ).

Multivariable Cox regression analysis indicated two variables to closely associated with RFS in HCC patients with HRRFs: AFP ≥ 400 ng/mL (HR: 2.28, 95% CI: 1.27-4.11, P = 0.006) and PAT (HR: 0.33, 95% CI: 0.18-0.60, P < 0.001) ( Table S2 ). Moreover, three factors were strongly related to OS in HCC patients with HRRFs: AFP ≥ 400 ng/mL (HR: 2.91, 95% CI: 1.19-7.11, P = 0.019), satellite nodules (HR: 2.76, 95% CI: 1.05-7.22, P = 0.039), and PAT (HR: 0.18, 95% CI: 0.06-0.54, P = 0.002) ( Table S3 ). The above results significantly established PAT as a favorable factor in improving the RFS and OS of HCC patients with HRRFs.

The occurrence of adverse events in 47 patients receiving PAT was summarized in Table 3 . No grade 4/5 toxicities or serious adverse events were observed, and common grade 1-3 toxicities included pruritus (44.7%), hypertension (42.6%), dermatitis (34.0%), and proteinuria (31.9%). Grade 3 toxicities included dermatitis (6.4%), diarrhea (6.4%), pruritus (2.1%), hypertension (2.1%), hand-foot skin reaction (2.1%), proteinuria (2.1%), and thrombocytopenia (2.1%).