Introduction
Hepatocellular carcinoma (HCC) is the major type of primary liver cancer and the third leading cause of cancer-related death worldwide [
1]. Although early-stage tumor can be curable by surgical resection, ablation, or liver transplantation [
2], the vast majority of patients had advanced unresectable disease at time of initial diagnosis with a relatively poor prognosis owing to the absence of early clinical symptoms and effective screening methods. Hepatitis B virus (HBV) is the leading cause of incident cases of HCC and deaths worldwide (33%), followed by alcohol (30%), hepatitis C virus (HCV) (21%) and other reasons (16%) [
3,
4].
The previous standard first-line systemic treatments for HCC were only lenvatinib and sorafenib [
5‐
10]. However, targeted agents only conferred limited survival benefits [
11‐
13]. In addition, the efficacy of sorafenib in patients with HBV-related HCC was revealed to be inferior to that in patients without HBV infection [
11,
13]. Recently, immunotherapy is changing treatment strategies for many malignant tumors, and increasing evidence suggests that patients with HCC may benefit from these new therapies [
14,
15]. Single-agent immune checkpoint inhibitors (ICIs) represented by programmed cell death 1(PD-1) and programmed cell death ligand 1 (PD-L1) inhibitors have been recently evaluated in HCC patients, and the results of clinical trials were disappointed [
16‐
18]. A combination of ICIs and vascular endothelial growth factor (VEGF) inhibitors might promote an immune permissive environment and enhance ICI response [
19,
20]. Therefore, immune-based combinations have been more striking [
21]. In 2020, the IMbrave-150 trial demonstrated for the first-time that atezolizumab combined with bevacizumab is superior to sorafenib in the treatment of unresectable HCC and obtained clinically meaningful improvement in overall survival (OS) and progression-free survival (PFS), leading to its global approval [
22]. Similar result has also been found in ORIENT-32 trial [
23]. COSMIC-312 study reported that atezolizumab plus cabozantinib achieved a lack of improvement in OS compared to sorafenib [
24]. In 2022, European society of medical oncology (ESMO) congress which updated the latest progress of first-line treatment regimens for HCC published primary results from Leap-002 and SHR-1210-III-310 studies [
25,
26]. Leap-002 as a multicenter phase III study did not meet pre-specified statistical significance for primary endpoints of OS and PFS between lenvatinib plus pembrolizumab and lenvatinib in advanced HCC. Correspondingly, SHR-1210-III-310 study showed positive findings, that the combination of camrelizumab and apatinib in patients with advanced liver cancer demonstrated significant clinical benefits in terms of OS and PFS at the common primary endpoint.
Although some studies have displayed that combination therapy achieved better survival benefits than alone [
27,
28]. Leap-002 and SHR-1210-III-310 study, global multicenter phase III clinical trials reported in ESMO 2022 have not been included in the previous research. Considering a large number of immunotherapy studies and new combination therapies for HCC exhibiting different clinical outcomes, we conducted this systematic review and meta-analysis aimed to overcome the limitations of individual research to better estimate the efficacy and safety of ICIs-combined anti-angiogenic therapy in treatment of unresectable HCC. Simultaneously, whether subgroups provided better OS and PFS outcomes were also explored to screen advantageous populations and determine the best therapy regimen.
Materials and methods
This systematic review and meta-analysis were in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). This study was registered with the International Prospective Register of Systematic Reviews (PROSPERO CRD42023400568).
Data source and search strategy
The databases, including Cochrane Library, PubMed, Embase and Web of Science were searched for eligible studies. The search time was from inception to November 2022. The main search therapy-related retrieval fields included (anti-angiogenic OR molecular targeted therapy OR targeted therapy) AND (PD-1 inhibitors OR programmed death ligand 1 OR PD-L1 inhibitors OR programmed death 1 receptor OR immunotherapy OR immune checkpoint inhibitors). The disease-related retrieval fields included hepatocellular carcinoma OR liver cell carcinoma OR Liver cancer. In addition, the reference lists of all relevant articles as well as conference abstracts published in main international oncological meetings (such as American Society of Clinical Oncology (ASCO), ASCO gastrointestinal cancer Symposium (ASCO-GI) and ESMO) were also searched to identify additional relevant studies.
Study selection
Potential trials, with the exception of reviews (including meta-analysis), editorials, fundamental studies, animal studies, comments and case reports, were eligible to be included in this meta-analysis if all of the following criteria apply: (1) prospective phase III randomized controlled trials (RCTs); (2) diagnosis of unresectable HCC; (3) comparison with PD-1/PD-L1 inhibitors plus anti-angiogenic drugs and anti-angiogenic therapy alone; (4) clinical outcomes of the study including OS, PFS, objective response rate (ORR), disease control rate (DCR) and treatment-related adverse events (TRAEs); (5) English as study language.
Data extraction and quality assessment
The following contents were extracted for each eligible study: (1) study general information (study name, first author, publication year, trial phase, study design, sample size); (2) basic information about the patients (age, male, etiology, geographical region); (3) interventions and control group. The main outcomes are PFS, OS, ORR, DCR and TRAEs. Both Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST 1.1) and hepatocellular carcinoma-specific modified RECIST (mRECIST) criteria were used in the study. The risk of bias to verify methodological quality was evaluated based on the Cochrane Collaboration’s tool for randomized control trials by the Review Manager 5.4 (RevMan5.4).
Statistical analysis
The statistical analysis was conducted using Stata 14.0 and RevMan5.4. The pooled hazard ratios (HRs) and 95% confidence interval (CI) for OS and PFS were calculated, as well as the pooled odds ratios (ORs) and 95% CI for ORR, DCR, any grade TRAEs and grade 3–5 TRAEs. Heterogeneity among studies was quantified by the
I2test, and
I2 > 50% and
p < 0.05 was considered statistically significant heterogeneity [
29]. When heterogeneity was significant, a random-effects model was used to calculate the pooled HR and OR; otherwise, the fixed-effects model was adopted.
Through subgroup analysis, publication bias assessment and sensitivity analysis, the origin of the heterogeneity was further explored. Begg’s test and Egger’s test were conducted to evaluate publication bias. The publication bias was absent with
p > 0.05 in Begg’s test and Egger’s test [
30]. We performed a sensitivity analysis by removing each study to observe changes in pooled HR. Region (Asia vs non-Asia), macrovascular invasion (MVI) or extra hepatic spread (EHS) (presence vs absence), alphafetoprotein (AFP) level (< 400 vs ≥ 400 ng per milliliter), etiology of HCC (HBV vs HCV vs Non-viral) and Barcelona Clinic Liver Cancer (BCLC) stage (B vs C) were considered in subgroup analysis.
Discussion
Unresectable HCC accounts for approximately 75–85% of primary liver cancers, and treatment options are limited due to poor prognosis [
31]. Finding appropriate treatment is necessary to improve patient survival [
32]. Combination immunotherapy had a higher chance of being the most effective therapy than targeted monotherapy [
22]. Recently, several ICI combination strategies for unresectable HCC have reported the encouraging results [
22,
23,
26], but other results have been disappointing [
24,
25]. Therefore, screening advantageous populations and determining the best combination therapy regimen has become a major challenge for HCC immunotherapy. We sought to find biomarkers or specific populations associated with immunoefficacy to stratify patients with HCC, distinguish between responders and non-responders, and recommend alternative therapies for patients who are not expected to respond to immunotherapy to avoid unnecessary toxicity. We conducted a meta-analysis which included five randomized controlled phase III trials of first-line therapies for unresectable HCC to show significantly better OS, PFS, ORR and DCR outcomes with PD-1/PD-L1 inhibitors in combination with anti-angiogenic drugs compared with anti-angiogenic drugs alone. Moreover, heterogeneities were revealed among the included studies for both OS and PFS. Subgroup analyses were performed to assess differences in outcomes and screen out the dominant population.
Theoretically, PD-L1 expression is the most direct marker for predicting the efficacy of PD-1/PD-L1 inhibitors, but unlike other malignancies, HCC is often accompanied by hepatitis or cirrhosis, which makes the tumor microenvironment of HCC more complex. The more complex classification of PD-L1 in HCC tissues and higher levels of spatial and cellular heterogeneity may affect the reliability and reproducibility of PD-L1 as a predictor of ICIs efficacy [
33]. In the CheckMate 040 study [
34], the ORR of PD-L1-positive patients was 26% in PD-L1-positive patients and 19% in negative patients, suggesting that negative expression of PD-L1 on tumor cells had no significant difference in the ORR against PD-1 therapy compared with PD-L1-positive patients. A phase II clinical trial of pembrolizumab in patients with unresectable advanced HCC suggested that there was no significant correlation between PD-L1 positivity and treatment response [
35]. However, Zhou et al. reported a meta-analysis study showing that positive PD-L1 expression is better associated with ORR in patients with advanced liver cancer treated with anti-PD-1/PD-L1 [
36]. Therefore, the expression of PD-L1 is currently controversial in predicting the efficacy of HCC. All included clinical trials in this meta-analysis lacked clinical outcomes in PD-L1-positive people. Therefore, PD-L1 expression was not included in the subgroup analysis in this study. Subgroup analyses were conducted according to the baseline characteristics of patients (geographical region, presence of MVI or EHS, AFP level, etiology and BCLC Stage).
In subgroup analysis, COSMIC-312 [
24], IMbrave150 [
22], SHR-1210-III-310 [
26] and Leap-002 [
25] were stratified for OS according to etiology (HBV, HCV and non-viral). The HRs of OS and PFS were stratified based on the factors of HBV status (positive vs negative) in the ORIENT-32 study [
23]. Therefore, ORIENT-32 study was only included in the HBV subgroup. Subgroup analysis for PFS included COSMIC-312, IMbrave150 and SHR-1210-III-310. Leap-002 was not included due to the lack of data for each interested subgroup for PFS. It is found that combination therapy has no significant impact on the reduction of the risk of death compared with anti-angiogenic drugs in HCV patients. In the non-virus subgroup, there did not appear to be a difference between the combination therapy and anti-angiogenic treatment for OS and PFS. When considering only HBV-infected patients, combination therapies of all studies were confirmed to substantially reduce the risk of death compared to monotherapy, and the heterogeneity decreased substantially (
I2 = 0%). Additionally, most patients in ORIENT-32 (94%) and SHR-1210-III-310 (76.8%) studies had HBV-related HCC, compared with less than 50% of participants in the Leap-002, IMbrave 150 and COSMIC-312 studies. This may be the reason for the better clinical outcomes of the ORIENT-32 and SHR-1210-III-310 studies. Basic studies have discovered that chronic HBV infection results in virus-specific T cell exhaustion and the PD-1/PD-L1 axis is a crucial inhibitor of HBV-specific CD8 + T cell activity [
37]. Therefore, PD-1/PD-L1 inhibitors blocking could partially restore effective HBV-specific T-cell responses to viral proteins, which could theoretically affect the efficacy of ICIs [
38,
39]. By contrast, non-viral HCC as a heterogeneous population that includes hepatic steatosis might be less responsive to immunotherapy compared with other etiologies of HCC [
40]. HCV patients have wide geographical variations that exhibit different regional characteristics, such as metabolic syndrome and alcohol consumption, as well as anti-cancer treatments which might influence survival through both hepatic and extra hepatic effects or through follow-up therapy [
41,
42]. Unlike previous meta-analyses that have not highlighted the characteristics of population, the present meta-analysis exclusively focused on differences in efficacy in subgroups of HBV, HCV and non-viral patients. Observations were extended to new combinations of therapeutic that were not covered in previous work.
COSMIC-312 [
24], IMbrave150 [
22], SHR-1210-III-310 [
26] and Leap-002 [
25] were included in subgroup analysis based on geographical region for the primary endpoints (OS). ORIENT-32 [
23] study which was done for the Chinese population was only included in the Asian subgroup. Because of the lack of information for each subgroup of interest for PFS, Leap-002 was not included. The findings showed that combination therapy was significantly superior to monotherapy for OS in Asia, whereas there was no advantage benefit in patients with HCC of non-Asian population. For PFS, that combination therapy was significantly superior to monotherapy in both Asian and non-Asian population, however, pooled HR value was lower in Asia. In Africa and East Asia, the largest proportion of the population is attributable to be cause by HBV (60%); however, only 20% of cases in the Western world can be attributed to HBV infection, and chronic HCV is the most common potential liver disease etiology [
4,
43]. Therefore, the perfect clinical outcomes of OS and PFS in the Asian population were also attributed to HBV infection being the dominant immunotherapy population. ORIENT-32 and SHR-1210-III-310 studies which had more patients in Asia (100% and 83.0%) differed from those in IMbrave150 (40%) study, COSMIC-312 (28%) study and Leap-002 (30.6%) study. This may induce the ORIENT-32 and SHR-1210-III-310 studies to have a lower HR value with positive outcomes.
Combination immunotherapy producing better clinical outcomes in patients with HBV-positive patients and Asian patients was discussed. However, there are still some patients who do not benefit from immunotherapy. Exosomes are closely related to viral hepatitis, cirrhosis and HCC. As an important intercellular communication mediator in the tumor immune microenvironment, exosomes may play a unique role in the immune response of HCC, thereby affecting the efficiency of immunotherapy. Exosomes exhibit the dual characteristics of tumor promotion and inhibition. On the one hand, they can mediate immunotherapy resistance by affecting the PD-1/PD-L1 axis or the anti-tumor function of immune cells in the tumor microenvironment. On the other hand, exosomes can carry drugs to downregulate PD-L1 expression on the surface of immune cells to improve the efficacy of ICI [
44]. Unfortunately, however, there were no RCTs on exosome treatment under our search strategy. The relationship between exosome therapy and liver cancer (including HBV-related HCC) will be further explored in the future.
The mRECIST measured only the viable tumor, which is defined as the contrast-enhanced portion of the tumor on hepatic arterial phase images. However, RECIST 1.1 measured the whole lesion, which is not enough to evaluate therapy induced intratumoural necrosis [
45,
46]. ESMO guidelines indicated the application of mRECIST or RECIST 1.1 in patients with HCC treated with anti-angiogenic targeted therapies [
47]. However, National Comprehensive Cancer Network (NCCN) guidelines suggested that mRECIST and RECIST 1.1 are needed to assess tumor response of molecular targeted drugs [
48]. The efficacies (ORR and DCR) of combination therapy compared with anti-angiogenic monotherapy for HCC were assessed according to both RECIST 1.1 and mRECIST in this meta-analysis.
The analysis results of TRAEs showed that compared with anti-angiogenic therapy, the combination therapy appears to have a significantly higher incidence of TRAEs. For these five trials, the most common TRAEs from combination therapies were hypertension, increased alanine aminotransferase, increased aspartate aminotransferase, proteinuria, diarrhea, fatigue, etc. Most of the TRAEs were concentrated in grade 1–2 indicating that the adverse events could be manageable. Grade 3–5 TRAEs occurred more frequently with camrelizumab plus apatinib in the SHR-1210-III-310 study (OR = 3.98; 95% CI: 2.70–5.86). The most common serious TRAEs were hypertension, increased alanine aminotransferase and increased aspartate aminotransferase.
There may be some possible limitations in this meta-analysis. Firstly, the RCTs selected in this meta-analysis involved various types of therapeutic drugs and diverse baseline characteristics, which may cause significant heterogeneities in data analysis in the aspect of the dissimilar clinical therapeutic effects and TRAEs. Therefore, the subgroup analyses were conducted attempting to stratify by baseline characteristics to mitigate the impact of heterogeneities. The efficacy and safety of the combination therapy can be further investigated through network meta-analysis in the future. Secondly, the present study included only five RCTs to compare PD-1/PD-L1 inhibitors combination therapy with anti-angiogenic monotherapy in patients with unresectable HCC. Further clinical trials would provide more reliable data for analysis, which may be included in future studies. Thirdly, there were inadequate cost-effective analyses for HCC in these trials, which might prove to be important for individual therapy. More cost-effective analyses are warranted due to the higher cost of the combination therapy than anti-angiogenic monotherapy. Lastly, there are inadequate mechanism reports of HBV response and resistance of immunotherapy. Future research will be critical for demonstrating the relationship between HBV infection and efficacy of ICIs.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.