Introduction

Hepatocellular carcinoma (HCC) constitutes the third leading cause of cancer-related mortality worldwide.¹ Approximately 44–62% of HCC patients have portal vein tumor thrombosis (PVTT) and are diagnosed as being in an advanced stage with an extremely poor prognosis.^2–4 PVTT promotes the intrahepatic dissemination of tumor cells and increases the risk of distant metastasis. Furthermore, the elevated portal venous pressure resulting from PVTT contributes to the progressive impairment of liver function and exacerbates ascites formation.⁵

Systemic therapies, including tyrosine kinase inhibitors (TKIs) and programmed cell death-1 (PD-1) inhibitors, constitute the cornerstone of treatment strategies for HCC patients with advanced stage.^6,7 Lenvatinib, one of the TKIs, has attracted growing attention and application in HCC owing to its dual anti-angiogenic and immunomodulatory effects.⁸ The combination of lenvatinib and PD-1 inhibitors may exert a synergistic anti-tumor effect.⁹ Although the IMbrave 150 study (atezolizumab plus bevacizumab) and the Keynote 524 study (lenvatinib plus pembrolizumab) achieved a benefit for advanced HCC, the prognosis for HCC patients with PVTT still remains poor after systematic therapies.^10–12 This suggests that more effective treatment strategies for PVTT are needed.

Transcatheter arterial chemoembolization (TACE) and hepatic arterial infusion chemotherapy (HAIC) are the two most prevalent forms of hepatic arterial interventional therapies (HAIT) and are widely used for treating advanced HCC patients in Asia-Pacific region.^13,14 HAIT can change the tumor immune microenvironment to improve the efficacy of systemic therapies in addition to performing local anti-tumor treatment.^15,16 A multicenter study demonstrated that the combination of TACE and TKIs plus PD-1 inhibitors for HCC with PVTT led to longer overall survival (OS) and progression-free survival (PFS) than TKIs plus PD-1 inhibitors.¹⁷ Another multicenter study comparing camrelizumab plus rivoceranib with or without HAIC for HCC patients with PVTT showed that the addition of HAIC to camrelizumab plus rivoceranib significantly improve median OS (18.7 months vs 11.0 months, p < 0.001) and PFS (10.0 months vs 5.6 months, p < 0.001).¹⁸

Supported by growing evidence of clinical benefits,^19,20 the combination of HAIT and systemic therapies has emerged as a promising treatment pattern for HCC patients with PVTT. Hence, we conducted this study to evaluate the efficacy and safety of HAIT combined with lenvatinib and PD-1 inhibitors for HCC patients with PVTT in a real-world setting. Additionally, we developed a nomogram model incorporating PVTT types to predict patients’ 6-month PFS and provide decision support for clinicians.

Materials and Methods

Study Population

This study included patients diagnosed with advanced HCC with PVTT between June 2020 and December 2023. These patients received either a treatment regimen consisting of HAIT in combination with lenvatinib and PD-1 inhibitors (H-L-P group) or a regimen involving lenvatinib plus PD-1 inhibitors (L-P group). The inclusion criteria were as follows: (1) HCC diagnosed radiologically or pathologically in accordance with the practice guidelines of the American Association for the Study of Liver Diseases (AASLD); (2) PVTT was diagnosed by intravenous contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI); (3) Child-Pugh stage A or B; (3) the score of Eastern Cooperative Oncology Group (ECOG) was within 1. The exclusion criteria included: (1) Patients with ruptured liver cancer; (2) Patients who received less than two cycles of treatment; (3) Patients who received a previous treatment; (4) Patients who had a history of other malignancies.

Treatment Procedures

HAIT included TACE or HAIC, the choice of HAIT was determined through communication between the patients and their attending physician. However, the general guidelines are as follows: TACE is not recommended for patients with the Vp4 type of PVTT, while HAIC is not advisable for patients with Child-Pugh stage B liver function. The basic procedures for HAIT were conducted as follows: Selinger’s technique was employed to percutaneously puncture the right femoral artery, a 5F catheter sheath was placed and matching catheters were then inserted into the celiac and superior mesenteric arteries. Digital subtraction angiography (DSA) was performed to ascertain the location, size, and feeding artery of the tumor. For TACE, emulsions (approximately 50 mg) of lipiodol and lobaplatin (at a ratio of 1:6) were administrated into the tumor’s feeding artery through a microcatheter for chemoembolization. For HAIC, after catheterizing the feeding artery of tumor, the microcatheter was connected to the arterial infusion pump, and patients were admitted to the ward for subsequent infusion chemotherapy. In our center, HAIC used the modified FOLFOX chemotherapy regimen: oxaliplatin, 85 mg/m² for 2 hours on day 1; leucovorin, 200mg/m² for 3 hours on day 1; 5-fluorouracil, 2400 mg/m² for 46 hours. Every 3 to 4 weeks, TACE or HAIC was given again until the tumor progressed, or the toxicities were intolerable.

Lenvatinib and PD-1 inhibitors were administrated within 1 week depending on the recovery of liver function after the initial HAIT. Lenvatinib was taken orally daily (weight ≥ 60 kg: 12 mg; weight < 60 kg: 8 mg). PD-1 inhibitors (including camrelizumab, tislelizumab, pembrolizumab and sintilimab) were administered intravenously over 60 minutes at a dose of 200 mg every 3 weeks.

Data Collection and Follow Up

The data from the Laboratory, including alpha-fetoprotein (AFP), hepatitis B virus (HBV) load and liver function were recorded prior to the administration of initial HAIT. The data of the imaging, including tumor number and size, extrahepatic metastasis, and the type of PVTT were recorded through intravenous contrast-enhanced CT or MRI. The primary endpoint was PFS, which was defined as the duration from the admission to disease progression or death occurred from any cause. OS was defined as the duration from admission to death occurred from any cause. Tumor responses, including complete response (CR), partial response (PR), stable disease (SD) and progressive disease (PD), were evaluated every 2–3 cycles using the Modified Response Evaluation Criteria in Solid Tumors (mRECIST). The objective response rate (ORR) was defined as the proportion of patients who achieved CR or PR, and the disease control rate (DCR) was defined as the proportion of patients who achieved CR, PR or SD. Treatment-related adverse events (TRAEs) were evaluated by the National Cancer Institute Standard for Common Terminology for Adverse Events (version 5.0). The follow-up deadline was set for July, 2024.

Statistical Analyses

Statistical analyses were performed using the R software (version 4.4.2) and SPSS (version 26.0). Complete-case analysis was used to handle missing data. Categorical variables were analyzed by the chi-square (χ2) test and Fisher’s exact test. P values < 0.05 were considered to be statistically significant. 1:1 propensity score matching (PSM) was performed with a caliper width of 0.25. Covariates including age, sex, ECOG score, ALBI (Albumin-bilirubin) grade, Child-Pugh class, HBV load, AFP, tumor size, tumor number, PVTT, hepatic vein tumor thrombus (HVTT), tumor metastasis, and PD-1 inhibitors were used in the matching model. Univariate and multivariate analyses were based on Cox proportional analysis, factors with P values less than 0.1 in univariate analysis will be further included in multivariate analysis. Kaplan–Meier analysis and the Log rank test were employed to compare the outcomes of the two groups. The results of Cox regression analysis were used to construct nomogram model. Using the ROC curve to assess the prognostic accuracy of the nomogram. Calibration curve was used to illustrate the accuracy of the nomogram, while the clinical practicability was evaluated through analysis of the decision curve.

Results

Characteristics of Patients

Following the assessment of inclusion and exclusion criteria, 208 patients were enrolled in this study. Among them, 120 patients received HAIT combined with lenvatinib and PD-1 inhibitors (H-L-P group) and 88 patients received lenvatinib plus PD-1 inhibitors (L-P group). The screening process is illustrated in Figure 1. Detailed characteristics of patients before and after PSM were shown in Table 1. Most patients in both groups were male and had AFP levels ≥ 400 ng/mL, multiple lesions, and absence of HVTT. All patients had hepatitis B virus infection and PVTT. Twenty-six patients (12.5%) had vp1/2 type of PVTT and 182 patients (87.5%) had vp3/4 type of PVTT. Compared to the H-L-P group, metastasis was more frequently observed (p = 0.004) and the liver function was poorer (p = 0.049) in L-P group. Following a 1:1 PSM, the baseline characteristics between the two groups were well-matched.

Table 1 Baseline Characteristics of Patients Before and After PSM

Figure 1 Flow chart of patient screening.

Survival Outcomes and Tumor Response

The median follow-up time was 32 months. In the entire cohort, patients in the H-L-P group exhibited significantly longer median OS of 22 months (95% CI: 16.23–27.00) and median PFS of 10 months (95% CI: 8.20–12.00) compared to those in the L-P group, who had median OS of 12 months (95% CI: 10.00–15.00) and median PFS of 4 months (95% CI: 4.00–6.00). The H-L-P group showcased significantly higher OS rates at 6, 12, and 24 months (95.8%, 68.3%, 45.9%) compared to the L-P group (89.8%, 49.6%, 4.9%) (p < 0.001). Similarly, the H-L-P group showcased significantly higher PFS rates at 6, 12, and 24 months (70.8%, 38.2%, 27.2%) compared to the L-P group (26.1%, 6.2%, 3.1%) (p < 0.001).

After PSM, the OS and PFS of the H-L-P group were also longer than those in the L-P group. The median OS for the H-L-P group and L-P group were 19 months (95% CI: 14.10–21.00) and 14 months (95% CI: 11.00–15.00), while the median PFS was 10 months (95% CI: 8.00–12.00) and 4 months (95% CI: 4.00–6.00). Moreover, the H-L-P group also demonstrated superior OS rates at 6, 12, and 24 months (94.6%, 63.5%, 39.2%) compared to the L-P group (90.5%, 52.3%, 5.8%) (p < 0.001), and superior PFS rates at 6, 12, and 24 months (67.6%, 36.2%, 22.2%) compared to the L-P group (27.0%, 7.4%, 3.7%) (p < 0.001). Survival curves between the two groups before and after PSM were shown in Figure 2. In addition, subgroup analysis demonstrated that the H-L-P group resulted in trends of lower risk of progression and death compared to L-P group across nearly all subgroups (Figure 3).

Figure 2 Survival analysis for OS and PFS between the two groups before PSM (A and B) and after PSM (C and D).

Figure 3 Subgroup analysis for OS (A) and PFS (B) between the two groups.

The best tumor responses before and after PSM were shown in Table 2. Before PSM, the ORR for the H-L-P group and the L-P group was 37.5%, 18.2%, respectively (p < 0.001), while the DCR was 75.8%, 48.9% (p < 0.001). After matching, the H-L-P group also demonstrated superior ORR and DCR at 37.8% and 78.4%, respectively, compared to the L-P group at 16.2% and 47.3%.

Table 2 Best Tumor Response Before and After PSM

Univariate and Multivariate Analysis for Survival

The results of univariate and multivariate analysis for OS and PFS are shown in Table 3. Factors with P values less than 0.1 in univariate analysis will be further included in multivariate analysis. Univariate analysis revealed that AFP, tumor size, tumor number, type of PVTT, hepatic vein tumor thrombus (HVTT), metastasis, and treatment regimen were associated with OS. Ultimately, tumor size ≤ 10 cm, single tumor lesion, Vp1/2 type of PVTT, and receiving H-L-P treatment regimen were found to be strongly associated with favorable OS. Similarly, the same factors were identified as independent prognostic factors of PFS.

Table 3 Univariate and Multivariate Analysis of Overall Survival and Progression-Free Survival

Development of a Prognostic Nomogram Model

Based on the results of univariate and multivariate analyses, we constructed a prognostic nomogram model that incorporates AFP levels, tumor size, tumor number, PVTT types, presence of HVTT, presence of metastasis and treatment regimen to predict the 6-month PFS of HCC patients with PVTT in this study (Figure 4A). The ROC curve depicted in Figure 4B demonstrated the good performance of the nomogram in predicting 6-month PFS, and the AUC value at 6-month was 0.84 (95% CI: 0.78–0.90). Additionally, the calibration curves (Figure 4C) demonstrated good concordance between the predicted and observed outcomes. The decision curves in Figure 4D demonstrated the nomogram had great clinical application in predicting 6-month PFS.

Figure 4 The development and validation of the prognostic nomogram model (A) The prognostic nomogram model for predicting 6-month PFS; (B) The ROC curve for predicting 6-month PFS; (C) The calibration curve for predicting 6-month PFS; (D) The decision curve for predicting 6-month PFS.

Safety

The treatment-related adverse events (TRAEs) details were provided in Table 4. All TRAEs were controlled, and there were no treatment-related deaths. The three most prevalent TRAEs in the H-L-P group were elevated aspartate aminotransferase (AST), elevated alanine aminotransferase (ALT), and vomiting. Any grade of TRAEs, including abdominal pain (21.7% vs 9.1%, p = 0.015), nausea (34.2% vs 8.0%, p < 0.001), vomiting (37.5% vs 13.6%, p < 0.001), anorexia (27.5% vs 15.9%, p = 0.048), elevated ALT (45% vs 17.0%, p < 0.001), elevated AST (47.5% vs 27.3%, p = 0.003), and hyperbilirubinemia (23.3% vs 4.5%, p < 0.001), was more frequently reported in the H-L-P group. No significant differences were observed between the two groups in other TRAEs such as hypertension, fatigue, and pruritus. Additionally, grade 3 or 4 TRAEs were comparable between the H-L-P group and the L-P group.

Table 4 Treatment-Related Adverse Events

Discussion

PVTT in HCC patients has long been recognized as a prominent factor associated with unfavorable outcomes.^5,21 There are notable disparities in the treatment strategies for PVTT between Eastern and Western countries. Systemic therapies, such as sorafenib and lenvatinib, are recommended as the first-line treatment option in Western countries.^22,23 However, in addition to systemic therapies, Eastern countries emphasize that locoregional therapies (such as radiotherapy, TACE, and HAIC) can provide additional survival benefits for HCC patients with PVTT.^24–26 In recent years, combining locoregional and systemic therapies has shown remarkable potential for treating advanced HCC.^19,27 Hence, we investigated the efficacy and safety of HAIT in combination with lenvatinib and PD-1 inhibitors for HCC patients with PVTT in a real-world setting.

In our investigation, we utilized PSM to reduce the potential bias. After matching, we found that patients in the H-L-P group exhibited significantly better median OS (19 months vs 14 months, p < 0.001) and PFS (10 months vs 4 months, p < 0.001) compared to those in the L-P group. Additionally, the OS and PFS rates at 6, 12, 24 months were consistently higher in the H-L-P group than in the L-P group. Furthermore, the H-L-P group demonstrated superior ORR (37.8% vs 16.2%) and DCR (78.4% vs 47.3%) relative to the L-P group. Subgroup analyses also demonstrated that patients in the H-L-P group achieved better outcomes across most subgroups. A Phase II study demonstrated HAIC combined with camrelizumab and apatinib achieved a median PFS of 10.38 months in patients with advanced HCC, which is consistent with our findings. However, the ORR (77.1%) and DCR (97.1%) of the trial were higher than our study, which might be due to the different proportions of patients with PVTT. There were 87.84% of patients with vp3/4 type of PVTT in the H-L-P group in our study, but the phase II study only included 45.7% of patients who had vp3/4 type of PVTT.²⁸

The excellent benefits of the triple treatment may be attributed to the potential synergistic interactions among HAIT, lenvatinib and PD-1 inhibitors. On the one hand, the local action of HAIT directly targets tumor and can boost the anti-tumor immune response by releasing tumor antigens.²⁹ On the other hand, Lenvatinib not only inhibits tumor microvascular formation following HAIT but also remodels the tumor immune microenvironment, thereby enhancing the efficacy of PD-1 inhibitors.^30,31 Furthermore, chemotherapy drugs in HAIT can induce immunogenic cell death in HCC, thereby synergizing with PD-1 inhibitors to enhance antitumor immunity.³²

In this study, multivariate analyses demonstrated that treatment region and high tumor burden (such as tumor size > 10 cm, multiple tumor lesion, Vp3/4 type of PVTT) were significantly associated with poor survival of HCC patients with PVTT. The combination of HAIT, lenvatinib and PD-1 inhibitors was associated with favorable clinical outcomes, which aligns with the findings reported by Fu²⁰ and Chen.³³ However, the ORR and DCR in Fu’s research were higher than those in this study, which could potentially be attributed to the fact that their cohort also included 11% of patients with the vp0 type of PVTT, who may have a better treatment response. Based on the results of univariate and multivariate analyses, we developed a nomogram model to predict the 6-month PFS of HCC patients with PVTT. This nomogram model demonstrated excellent discrimination and calibration, as validated by the ROC and calibration curves. DCA curve indicated that the nomogram model possessed significant clinical application value. The survival rates for different types of PVTT can vary significantly, and patients with PVTT often present with complex clinical conditions.³⁴ Therefore, emphasizing individual differences is essential for the effective management of PVTT. This model allows for personalized short-term PFS predictions based on patient’s specific condition, thereby providing support for clinical decision-making.

In terms of safety, we observed that the addition of HAIT was associated with an increased incidence of TRAEs. Digestive system symptoms (such as nausea and vomiting) and liver function impairment were more frequently reported in the H-L-P group than the L-P group. However, grade 3 or 4 TRAEs were comparable between the two groups. In addition, all TRAEs were controlled after appropriate therapeutic interventions, and there were no treatment-related deaths. Another study also reported that the combination therapy was associated with an increased incidence of abdominal pain and abnormal elevation of liver enzymes.³⁵ Therefore, we should attach great importance to the management during the HAIT and enhance both symptomatic treatment and liver-protective measures.

This study has some limitations. Firstly, this is a retrospective, single-center study. A prospective multi-center study is needed to verify the reliability of the conclusion. Secondly, our study included a variety of PD-1 inhibitors and two types of HAIT, which may potentially influence the consistency of the results. Thirdly, the nomogram model established in this study has only undergone internal validation and lacks effective external validation. In addition, the relatively small sample size and the limited duration of the follow-up period in our study might potentially restrict the generalizability of the findings.

Conclusion

In conclusion, our study demonstrated that combining HAIT with lenvatinib and PD-1 inhibitors was a safe and promising treatment strategy for HCC patients with PVTT, and the development of the 6-month PFS nomogram model may provide support for personalized treatment and clinical decision-making. These findings may support the clinical integration of HAIT in the multidisciplinary management of advanced HCC with PVTT. However, more large-scale randomized controlled trials are warranted to validate the results.