Liver cancer is one of the leading causes of cancer-related deaths worldwide, with hepatocellular carcinoma (HCC) being the most common histological type, accounting for 75–85% of all reported cases (1). Several patients with HCC develop metastatic disease that cannot be effectively treated with current curative therapies and have poor prognosis despite using standard therapeutic regimens, such as sorafenib or lenvatinib, resulting in a median survival time (MST) of only 6.5–13.6 months (2, 3). Immune checkpoint inhibitors (ICIs), particularly programmed cell death checkpoint inhibitors (anti-programmed cell death-1 [PD1]/programmed cell death ligand 1 [PD-L1]), provide an effective treatment option for various carcinomas, including HCC (4–6). Several studies have focused on administering anti-PD1/PD-L1 regimens to patients with metastatic HCC (7–11). However, findings from several studies have generated controversy, underscoring the limited impact of anti-PD1 alone on HCC (7, 9, 10, 12). Previous studies have reported better outcomes with the addition of targeted therapy to immunotherapy; however, the response (21–32%) remains limited (10, 13, 14). Therefore, there is an urgent need to explore novel treatment combinations, particularly those involving the utilization of immune sensitizers.

Evidence has demonstrated the efficacy of radiotherapy in HCC (15–19). Studies conducted on other malignancies have observed that radiation upregulates PD-L1 expression in tumor cells. Preclinical data have also shown that radiation augments the antitumor effect of immune checkpoint inhibitors in a murine HCC model (20). Therefore, combining radiation with anti-PD1/PD-L1 may increase the response rate and improve treatment outcomes. However, despite these potentially impressive theoretical advantages, large cohort data regarding the safety and efficacy of clinical practice are lacking. To better understand the activity of this new treatment combination in unselected patients with advanced and recurrent HCC, we conducted a real-world study.

Our study showed a promising response rate, favorable survival rate, rare radiation field relapse, and tolerable AEs. Additionally, this study indicated that combined radiation and anti-PD1 was an efficient and tolerable treatment for these patients.

Our results were compared with those of previous studies ( Supplementary Table 1 ). Notably, previous studies have utilized other treatments as first- or second-line therapies, whereas in our study, more than 40% of patients received the current therapy as third-line or subsequent treatments. In comparison to classical multi-kinase inhibitors, such as sorafenib (9.2–11%),lenvatinib (24.1%) and regorafenib (2, 7, 14, 22) (23), our study demonstrated a significantly higher response rate(52.8% according to RECIST, 75.0% according to mRECIST). Moreover, when compared with anti-PD1 therapies alone (7, 10, 14, 22, 24, 25) (9), our study exhibited a clear advantage. Additionally, we achieved the highest disease control rate (97.5% according to RECIST, 98.1% according to mRECIST) compared with previous studies (55–75.5%) (7, 10, 14, 22, 24, 25). Durable responses and prolonged survival times were observed in this study. Compared with single multi-kinase inhibitors, anti-angiogenic therapy, and immunotherapy options—which were also employed during subsequent-line therapy—our results demonstrated a significantly extended median PFS (mPFS) (12.6 months vs. 2.1–4.9 months) (2, 8, 23, 25–28). The mPFS achieved in our study was better than first-line systemic therapies using sorafenib or lenvatinib (4.3–7.4 months) and r (2, 14). Moreover, our 1-year PFS rate (51.5%) surpassed that of the KEYNOTE 220 and KEYNOTE-224 trial using pembrolizumab (28%) and the CheckMate-459 trial using nivolumab (22%) (7, 28). The MST (17.0 months) in our study outperformed those of previous studies with multi-kinase inhibitors (8.5–14.7 months), cabozantinib (10.2 months), pembrolizumab (12.9–13.9 months), durvalumab (13.8 months), and camrelizumab (15.1 months) (9, 24, 25, 27). These comparisons demonstrate that the PFS post-immunotherapy alone is limited and that the addition of radiation can enhance the antitumor effect. Our study included some patients with progressive disease after immunotherapy, and we found that the addition of radiotherapy might reverse immunotherapy resistance subsequently achieving profitable PFS.

Compared with treatments of anti-PD-1/PD-L1 regimens combined with various of other systemic therapies involving cytotoxic T lymphocyte-associated antigen-4 antibody, multi-kinase or anti-angiogenesis agents, anti-PD-1/PD-L1 therapies plus radiation have demonstrated a higher tumor response rate (52.8% vs. 24–32% according to RECIST, 75.0% vs. 30.0–46.0% according to mRECIST), better disease control (98.1% vs. 75% according to mRECIST), and longer PFS (12.6 months vs. 2.2–6.9 months) (7, 8, 10, 14, 22, 25) (29). However, this superiority does not extend to MST (17.0 months vs. 12.0–22.8 months). Two factors account for this difference: First, there is an imbalance in the treatment history between the patients in our study and those in other studies. As previously mentioned, more than 80% of patients in our study had already received multiple lines of therapy, with 17% having undergone anti-PD-1/PD-L1 therapy. Consequently, limited salvage therapeutic regimens were available after disease progression. Second, our study enrolled more patients with adverse prognostic factors, such as PVTT (63.9%), main trunk of portal vein tumor thrombosis (Vp4) (27.8%), hepatocirrhosis (75.0%), viral hepatitis (87.1%), and higher AFP levels: 39.8% of patients with AFP ≥ 1,000 ng/mL, 50% of patients with AFP ≥ 400 ng/mL in our study, and 38% of patients with AFP ≥ 400 ng/mL in the IMbrave150 study (14).

Previous studies have consistently demonstrated that PVTT is a negative prognostic factor for HCC (30). However, we observed that some of the other studies listed PVTT, especially Vp4, as an exclusion criterion (29), whereas our study enrolled more than 60% of patients with PVTT and nearly one-third with Vp4. Nevertheless, our combined therapy approach showed more promising survival outcomes compared with historical studies, even with the disadvantage of these baseline characteristics (2, 8, 9, 28). Importantly, we found no significant difference in survival between patients with and without PVTT. The median survival of patients with PVTT was superior to that reported in previous studies utilizing targeted therapies, immunotherapies, or other combinations (31). Even patients with Vp4 disease showed favorable MST (14.9 months) and mPFS (9.2 months), compared with the Vp4 subgroup analysis in the IMbrave150 study (MST 7.6 months, mPFS 5.4 months), which used atezolizumab and bevacizumab (32). Two factors might have contributed to the phenomenon. First, previous studies have demonstrated that PVTT has an unsatisfactory prognosis due to the increasing difficulty of performing operations and interventional therapy, without considering radiation. Second, patients with PVTT might be sensitive to radiotherapy due to the abundant oxygen supply from diffusive blood circulation at this site.

Several previous studies on PD-1/PD-L1 antibodies have excluded patients with viral hepatitis due to the potential for virus reactivation during treatment. However, viral hepatitis accounts for > 60% of HCC cases in China, and excluding it from the study would reduce its representativeness (33). Meanwhile, virus reactivation after anti-PD1 therapies is approximately 2%–5% (34, 35), which is relatively acceptable according to historical reports. In a phase III trial, KEYNOTE-394, 78.7% of patients with HBV infection were enrolled, and immunotherapy-induced hepatitis occurred in only 1.7% of the patients (36). Therefore, in real-world practice, we did not prohibit ICI delivery to patients with hepatitis. Theoretically, patients with hepatitis might benefit more from PD1/PD-L1 antibodies based on the phenomenon that high levels of T cells in HBV-infected patients may enhance the sensitivities of PD-1 antibodies, and PD-1 antibodies could also evoke the immune system to defend against the virus (37). In contrast, other studies have shown that the virus-related HCC microenvironment is more immunosuppressive and exhausting than the non-viral microenvironment (38). However, these two theories have not been proven in clinical practice. We found that the survival rates in patients with hepatitis were equivalent to those of other patients, consistent with the results of other studies (10, 13, 28, 36). Overall, our study safely administered immunotherapy to patients with HBV/HCV infection with the protection of antiviral therapies.

Grade ≥ 3 AEs occurred in 30.6% of patients, consistent with the findings of a study that prescribed anti-PD-1/PD-L1 regimens alone (22–68%) (7–10, 25, 28). Most AEs involved hematological toxicities that resolved with treatment. Two patients experienced Grade 5 AEs, and one developed severe diarrhea following lenvatinib, and the patient declined any further intervention. That patient received abdominal radiation targeting to intrahepatic lesion and PVTT with dose of 27.5Gy in 5 fractions. Maximum point doses to the intestine (18.6 Gy) and colon (23.8 Gy) over 5 fractions were substantially below tolerance thresholds (35 Gy and 33 Gy respectively). Given that diarrhea is a common treatment-emergent AE among lenvatinib-treated patients, lenvatinib may be the direct cause. The other patient was diagnosed with rib and vertebral metastases after transarterial chemoembolization and sorafenib treatment; subsequently, the patient received concurrent toripalimab, regorafenib, and radiation to bone metastasis at a dose of 40 Gy in 10 fractions. A tracheoesophageal fistula was observed 11 months after radiation therapy and during maintenance therapy with toripalimab and regorafenib. Considering the very reasonable esophageal dose (maximum dose to the esophagus: 33 Gy and V32 < 10 mL with a conventional fraction) and the extended interval between radiation and the AE, radiation was less responsible for the fistula. On the contrary, previous studies reported gastrointestinal perforation in PD-1 inhibitor and regorafenib (39, 40). We inferred that toripalimab or regorafenib might have induced the observed toxicity. Recently, RTOG 1112 demonstrated adding SBRT improved OS (15.8 months) and PFS (9.2 months) in patients with advanced HCC, compared to sorafenib alone without the increment of AEs. Another prospective study proposed by our team also showed satisfied OS (16.5 months) and PFS (6.1 months) with tolerable toxicities after receiving concurrent radiotherapy and sorafenib in advanced HCC patients (41, 42). Furthermore, the combination of radiotherapy and immunotherapy represents an actively investigated synergistic approach in oncology. Radiotherapy has been demonstrated to elicit systemic antitumor immunity, inducing responses in both irradiated lesions and non-irradiated sites – the abscopal effect phenomenon – which appears potentiated by anti-PD-1 therapy (43). Therefore, it is reasonable to infer that combined radiation with either immunotherapy or targeted therapy is safe, while a combination of tri-treatment, including immunotherapy, radiation, and targeted therapy should be administered with caution and tailored to individual patients.

The mechanism of radiotherapy induces an abscopal tumor-specific immune response in both the irradiated and nonirradiated tumors, which is potentiated by PD-1 blockade and the

The strength of our study is multifold. First, our study demonstrated favorable response rates and improved survival despite enrolling patients with negative prognosticators who had already progressed after multiple lines of therapy. Second, most patients achieved satisfactory control of the radiation field and exhibited reliable radiation effects. Third, the combined treatment decision was discussed and made by an MDT, underscoring the advantages of multimodality therapy.

However, our study has some limitations. First, as a retrospective study, we could not obtain data on patients’ PD1 and PD-L1 expressions. The predictive value of tumor cell PD1/PD-L1 expression in terms of response and prognosis remains controversial, as evidenced by contrasting results from the CheckMate 040 and CheckMate 459 studies. Second, our study employed multiple anti-PD1/PD-L1 drugs, which makes it challenging to determine the specific effects of individual drugs. Future studies should use more consistent drug types to reduce variation. Furthermore, a direct comparison to studies involving different therapeutic approaches may introduce inevitable bias due to variations in patient enrollment criteria.