Hepatocellular carcinoma (HCC) is a highly invasive cancer and has a median overall survival (mOS) of 2–4 months (1, 2). Recently, programmed death 1 (PD-1)/programmed death-ligand 1 (PD-L1) inhibitors opened up new perspectives in the therapy of advanced HCC patients (3, 4). However, their efficacy alone is limited in advanced HCC patients.

The combination therapy of PD-1/PD-L1 inhibitors with antiangiogenic drugs, such as pembrolizumab plus lenvatinib (5), sintilimab plus bevacizumab (6), and camrelizumab plus apatinib (7), attained good clinical results in advanced HCC patients and improved overall survival (OS) over single-agent therapy. At the same time, nivolumab combined with radiotherapy had also demonstrated significant efficacy in advanced HCC (8). In addition, Zhong et al. found that PD-1/PD-L1 inhibitors in combination with radiotherapy and antiangiogenic therapy could improve the mOS to 22.4 months and the objective response rate (ORR) to 40% with low toxicity in advanced HCC patients (9). Radiotherapy and antiangiogenic agents can not only promote T cell infiltration into tumors, but they can also improve responsiveness to immunotherapy (10–12). As a result, based on these findings, triple therapy appears to be a safe and promising treatment strategy for HCC. However, predicting the efficacy of triple therapy remains a challenge.

The PD-L1 expression in a tumor is an important factor in the cancer immune cycle (13). The binding of PD-L1/PD-1 could prompt immune evasion and suppress T cell activation (14, 15). In the current studies, the PD-L1 expression could be detected using tissue biopsy and liquid biopsy (circulating tumor cells, CTCs) (13, 16). However, because the spatial-temporal heterogeneity of PD-L1 expression is inevitable, the information needed for the selection of appropriate therapy is limited to a single biopsy at a specific point (17–19). CTCs exhibit the advantages of continuous assessment, low invasiveness, and interrogation of the overall tumor burden, rather than a specific area (20, 21). Therefore, CTCs appear to be a more promising technique to assess the dynamic expression of PD-L1 in tumors and offer a better understanding of tumor heterogeneity compared to tissue biopsy.

Currently, multiple studies showed that PD-L1⁺ CTCs could be a prognostic indicator in a variety of cancers including HCC (22, 23). However, there was a controversy about the link between PD-L1⁺ CTCs and clinical efficacy (21, 24–26). The triple therapy had improved the prognosis of HCC patients significantly (9). We aimed to confirm whether PD-L1⁺ CTCs could be a predictive biomarker in HCC patients treated with triple therapy.

In our study, we performed multiple assessments on the PD-L1 expression on CTCs of HCC patients treated with triple therapy using the CytoSorter™ BioScanner system. The sensitivity and specificity of this approach were clinically validated in different solid cancers including HCC, breast cancer, lung cancer, and pancreatic cancer, with 61.9%-95.4% sensitivity and 76.6%-91.3% specificity (27–30). Our data showed that HCC patients with < 2 PD-L1⁺ CTCs were related to a better OS and ORR than those with ≥ 2 PD-L1⁺ CTCs. Moreover, we demonstrated that the decrease of PD-L1⁺ CTC counts at T1 after the beginning of triple therapy was significantly related to the ORR, which proved the necessity of continuous assessment of PD-L1⁺ CTCs.

The mOS of advanced HCC patients was only 2–4 months (1, 2). The emergence of PD-1/PD-L1 inhibitors completely revolutionized the treatment for advanced HCC patients. Although patients in our study had a poor tumor burden, the ORR of 36.1%, DCR of 91.5%, mPFS of 8.6 months, and mOS of 20.1 months were encouraging.

Currently, many biomarkers such as alpha fetoprotein (AFP), alkaline phosphatase (ALP), and PD-L1 expression had been assessed in previous studies with promising outcomes, particularly PD-L1 expression that had been proven to be linked to the ORR of PD-1/PD-L1 inhibitors in various cancers including HCC (21, 22, 31, 32). The triple therapy improved the prognosis of HCC patients significantly (9), although the correlation between PD-L1⁺ CTCs and their prognosis remains unclear. In a meta-analysis of patients with cancer receiving immune checkpoint inhibitors, the expression of PD-L1 on CTC was not found to be associated with the patient’s prognosis (20). However, Sonja et al. demonstrated that urothelial carcinoma patients receiving immunotherapy with < 2 PD-L1⁺ CTCs had a better OS than those with ≥ 2 PD-L1⁺ CTCs (p = 0.008) (23). In contrast, in a report of pembrolizumab in melanoma by Khattak et al., the presence of PD-L1⁺ CTC was associated with longer PFS and OS (33). Ikeda et al. also reported the association of PD-L1⁺ CTC with longer PFS in lung cancer patients receiving nivolumab (21). Moreover, in a study of HCC patients receiving PD-1/PD-L1 inhibitors, patients without PD-L1⁺ CTCs had a worse ORR than patients with PD-L1⁺ CTCs (22). It did not further explore the predictive value of a specific number of PD-L1⁺ CTCs. In this study, we found that a cut-off value of 2 PD-L1+ CTCs had the highest capability of prediction for ORR and patients with < 2 PD-L1+ CTCs were related to a better ORR and mOS. Thence, the PD-L1+ CTCs may be related to the mechanism of tumor escape (34).

Subsequently, we conducted multivariate Cox regression analysis and found that PD-L1⁺ CTCs were proven to be an independent predictive biomarker factor of OS. Furthermore, we investigated the dynamic change in PD-L1⁺ CTC counts and noted that the PD-L1⁺ CTC counts declined at T1 in the responder group and increased in the nonresponder group (p < 0.05). This revealed that both < 2 PD-L1⁺ CTCs at T0 and dynamic decrease in PD-L1⁺ CTC counts at T1 could be used to assess prognosis. However, we found that one or less PD-L1⁺ CTCs at T1 were not related to a better ORR, mPFS, and mOS. These findings may be resulting from the small sample size (n = 47). In addition, the PD-L1⁺ CTCs at T2 remained unchanged in both responders and nonresponders upon treatment (p > 0.05). This indicated that the unchanged PD-L1⁺ CTC counts at the later stage of triple therapy were not a predictor of efficacy but rather a maintenance of long-term efficacy.

There are currently no biomarkers that can be used to guide the use of PD-1 inhibitors in combination with radiation and antiangiogenic therapy and reflect therapeutic response. These findings highlight the importance of dynamic evaluation of PD-L1 expression in identifying HCC patients who are suitable for triple therapy and supporting physicians to alter treatments as soon as possible for HCC patients who are unlikely to respond.

Of course, there were certain limitations to the study. An important limitation of the study was the use of different antiangiogenic agents as well as different PD-1 inhibitors. Substantial heterogeneity in the treatment regimen might have influenced the interpretation of our findings. In addition, this was a small sample, single-arm trial, and not all of the patients completed the CTC test at T1 and T2. As a result, bigger sample size is required to determine whether these findings and the ROC curve cutoff values are solely appropriate for present patients.

In conclusion, our study showed that HCC patients with < 2 PD-L1⁺ CTCs at baseline had a better ORR and OS receiving triple therapy. Also, < 2 PD-L1⁺ CTCs were a positive independent prognostic factor for OS. Furthermore, this study demonstrated that the response was more likely to be achieved in patients with a dynamic decrease in PD-L1⁺ CTC counts at 1 month. These results support PD-L1⁺ CTCs as a predictive biomarker for HCC patients and provide a benchmark for future trials.

The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

The studies involving human participants were reviewed and approved by Clinical Trials Ethics Committee of the Affiliated Hospital of Southwest Medical University. The patients/participants provided their written informed consent to participate in this study.

KS, LG, KH, MR, JZ, XY, WH, TG, KX, YL, JW, JC, ZW, LH, HZ, HYL, JT, XL, YY, HLL, YX, ZT, XT, XF, SC, BY, HJ, and LZ collected the data. YH and BL designed the research study. KS, LG, KH, BL and YH wrote the manuscript and analyzed the data. All authors approved the final version of the manuscript. All authors contributed to the article and approved the submitted version.

This work was supported by a grant from Project of Science and Technology Department of Sichuan Province (2020JDTD0036).

The authors thank Hangzhou Watson Biotech, Inc. (Hangzhou, China) for technical help in CTC detection. The authors also thank the patients and families who made this study possible and the investigators and the clinical study teams who participated in this study.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.