Introduction

Hepatocellular carcinoma (HCC) is the world’s third common cause of cancer-related deaths [1]. Hepatitis B virus (HBV) infection is the most common etiology of HCC [2]. At present, there are few radical therapies for HBV-related HCC, including liver transplantation, resection and ablation [3,4,5]. Among them, liver transplantation is the most effective treatment for HCC, but it is severely limited by high costs and donor shortage [6]. Therefore, the most effective treatments are still resection and ablation [7, 8]. Unfortunately, the 3-year recurrence rate after resection or ablation is more than 50%, and the 5-year overall survival (OS) rate after resection ranges from 40% to 60% [9,10,11].

Fortunately, with the deepening of research on tumor immune microenvironment, researches have confirmed that anti-PD-1 could remove the immunosuppression of antitumor T cells and increase the sensitivity of its mediated killing effect, finally activate the antitumor immune response [12]. Anti-PD-1 has achieved unprecedented success in treating HCC, but it still faces major challenges. Encouragingly, a novel study found that IFN enhanced antitumor activity of anti-PD-1-based immunotherapy in patients with unresectable HCC [12]. Authors further built immunocompetent orthotopic and spontaneous HCC models and found IFN synergized with anti-PD-1 led to significant enrichment of cytotoxic CD27+ CD8+ T cells. Mechanistically, IFN suppressed HIF1α signaling by inhibiting FosB transcription in HCC cells, resulting in reduced glucose consumption capacity and consequentially establishing a high-glucose microenvironment that fostered transcription of the T-cell costimulatory molecule Cd27 via mTOR–FOXM1 signaling in infiltrating CD8+ T cells, thereby potentiating the PD-1 blockade-induced immune response [12]. Therefore, IFN may play a crucial role in preventing recurrence and improving OS of patients with HBV-related HCC after curative surgery.

However, it was found that different clinical trials had different results through literature search, and even the opposite results were obtained from several previous meta-analyses [10, 11, 13]. Therefore, there is always a question that whether IFN can improve survival and reduce recurrence of patients with HBV-related HCC after curative surgery. Encouragingly, new randomized controlled trial (RCT) and retrospective cohort study (RCS) have been published in recent years, and the previous analytical results need to be updated now. Hence, we conducted this comprehensive meta-analysis to re-evaluate the efficacy of IFN on recurrence and OS of patients with HBV-related HCC after curative surgery.

Materials and methods

The protocol of this meta-analysis was registered in PROSPERO (CRD42022381867).

Search strategy

We searched MEDLINE/PubMed, Cochrane Library, EMBASE, Web of Science and CNKI databases from inception to November 2022. The terms were “hepatocellular carcinoma OR liver cancer OR liver neoplasm OR hepatic carcinoma OR HCC”, “resection OR surgery OR hepatectomy OR ablation OR TACE OR transcatheter arterial chemoembolization”, “HBV OR hepatitis B” and “interferon OR IFN” in English or Chinese. Conference proceedings at the International Liver Congress and the Liver Meeting were also searched manually. Two authors independently evaluated the retrieved studies. If there was any disagreement, we would consult a third author for quality assurance.

Inclusion and exclusion criteria

The inclusion criteria included the following: a. all eligible patients had HBV-related HCC; b. the outcomes included at least one of OS and recurrence; c. curative surgery included surgical resection or ablation or transcatheter arterial chemoembolization (TACE). d. the sample size of each group > 30; e. the follow-up time > 12 months.

Exclusion criteria: (a) metastatic liver cancer or recurrent liver cancer; (b) the follow-up time < 12 months; (c) repetitive articles written in different languages; (d) HCC not associated with HBV; (e) single-arm studies.

Outcomes

The OS and recurrence rate were the primary outcomes. The secondary outcome was disease-free survival (DFS).

Data extraction and quality assessment

Two reviewers independently extracted the data and assessed the quality of each study. The following data were extracted: author names, study type, dates, follow-up period, setting of study, interventions, characteristics of patients and outcomes. The number of survivors or patients with recurrence was computed from the Kaplan–Meier curves, if not reported in the text. The risk of bias tool suggested by the Cochrane Handbook for Systematic Reviews of Interventions was used to adjudicate the methodological quality of RCTs [14]. The Newcastle–Ottawa Scale was used to assess the methodological quality of non-RCTs [15].

Statistical analysis

Stata software ver.12 was used to conduct statistical analysis. The Q test and I2 statistic were used to assess the heterogeneity of the effects. Significant heterogeneity was defined as p < 0.1 and I2 > 50%, and the random effect model was used, otherwise, the fixed effect model was used. When significant heterogeneity existed, a subgroup analysis was carried out according to operation selection and whether to use NAs or not. The relative risk and their 95% confidence interval were calculated. Publication bias was assessed qualitatively by Funnel plot, and statistically using Egger’s and Begg's test. Sensitivity analysis was conducted by excluding a single study and recalculating the pooled estimates. p < 0.05 was considered to be significant (p values were two-sided).

Results

Characteristics of studies

As shown in the flow diagram (Fig. 1), 1236 clinical studies were identified and finally a total of 10 studies were finalized based on the predefined inclusion and exclusion criteria.

Fig. 1
figure 1

Flow diagram

Full size image

There were 2062 patients in the 10 studies [16,17,18,19,20,21,22,23,24,25], which included 6 RCTs and 4 RCSs. Among them, 814 patients were treated with IFN and 1248 patients were treated without IFN. The length of follow-up ranged from 12 to 96 months. The ages of the patients ranged from 18 to 79 years. IFN combination with nucleoside analogues (NAs) were used in one study compared with NAs alone. The curative surgery included resection in 4 studies, TACE in 3 studies, resection combination with TACE in 1 study and resection or other in 2 studies. The Child–Pugh was A or B in 4 studies, while other 6 studies did not mention Child–Pugh Grade. Two articles reported the 8-year OS, and 8 years were the longest available follow-up time. Characteristics of studies is shown in Table 1.

Table 1 Characteristics of studies
Full size table

Primary outcome

OS

There were 10 articles (2026 patients), 9 studies (1834 patients), 8 studies (1784 patients), 7 studies (1646 patients) and 2 studies (474 patients) compared the 1-, 2-, 3-, 5- and 8-year OS, respectively. Results showed that IFN significantly improved the 1-, 2-, 3- and 5-year OS (RR = 1.09, 95% CI, 1.03–1.16, p = 0.006, I2 = 82.8%; RR = 1.27, 95% CI, 1.04–1.55, p = 0.021, I2 = 95.9%; RR = 1.22, 95% CI, 1.02–1.46, p = 0.026, I2 = 86.3%; RR = 1.25, 95% CI, 1.03–1.50, p = 0.021, I2 = 51.7%, respectively) (Fig. 2a–d), but IFN could not improve the 8-year OS (RR = 0.87, 95% CI, 0.37–2.09, p = 0.763, I2 = 61.3%) (Fig. 2e).

Fig. 2
figure 2

Forest plots of OS. a 1-year OS. b 2-year OS. c 3-year OS. d 5-year OS. e 8-year OS

Full size image

Recurrence rate

There were 4 studies (438 patients), 4 studies (340 patients), 3 studies (356 patients) and 4 studies (604 patients) compared the 1-, 2-, 3- and 5-year recurrence rates, respectively. The result showed IFN could not reduce 1-year recurrence rate (RR = 0.73, 95% CI, 0.36–1.49, p = 0.383, I2 = 71.7%) (Fig. 3a), but it significantly reduced 2-, 3- and 5-year recurrence rates (RR = 0.72, 95% CI, 0.58–0.90, p = 0.003, I2 = 33.1%; RR = 0.82, 95% CI, 0.69–0.97, p = 0.025, I2 = 0.0%; RR = 0.83, 95% CI, 0.74–0.94, p = 0.002, I2 = 28.8%) (Fig. 3b–d).

Fig. 3
figure 3

Forest plots of recurrence rates. a 1-year recurrence rate. b 2-year recurrence rate. c 3-year recurrence rate. d 5-year recurrence rate

Full size image

Secondary outcome

DFS

There were 5 studies (1328 patients), 4 studies (1100 patients), 4 studies (1112 patients) and 4 studies (1112 patients) compared the 1-, 2-, 3- and 5-year DFS, respectively. The result showed IFN significantly improved 1-, 2- and 3-year DFS (RR = 1.13, 95% CI, 1.05–1.23, p = 0.002, I2 = 0.0%; RR = 1.27, 95% CI, 1.12–1.44, p = 0.000, I2 = 0.0%; RR = 1.20, 95% CI, 1.01–1.42, p = 0.041, I2 = 0.0%, respectively) (Fig. 4a–c), but it could not improve 5-year DFS (RR = 1.17, 95% CI, 0.85–1.60, p = 0.336, I2 = 0.0%) (Fig. 4d).

Fig. 4
figure 4

Forest plots of DFS. a 1-year DFS. b 2-year DFS. c 3-year DFS. d 5-year DFS

Full size image

Subgroup analysis

Stratified analyses by operation selection

Studies were further grouped according to operation selection. Long-term OS was defined as the survival time exceeded 2 years, short-term OS was defined as the survival time < 2 years, early recurrence meant the tumor recurred within 2 years and late recurrence meant the tumor recurred after 2 years [26, 27]. In this meta-analysis, patients receiving IFN showed higher long-term OS and lower late recurrence rate in resection combined with TACE subgroup, higher long-term OS and lower early and late recurrence rates in TACE subgroup, higher short-term OS in resection subgroup (Tables 2, 3).

Table 2 OS based on operation selection
Full size table
Table 3 Recurrence based on operation selection
Full size table

Stratified analyses based on whether to use NAs or not

Studies were grouped according to whether to use NAs or not. Results showed that IFN significantly improved the 1-, 2-, 3- and 5-year OS in the subgroup without NAs, but there were no statistical significances in the subgroups with NAs. IFN could not significantly improve the 8-year OS in both subgroups (Table 4).

Table 4 OS based on whether to use NAs or not
Full size table

Publication bias and sensitivity analysis

No evidence of publication bias was detected by funnel plots and Begg’s and Egger’s test. The statistical significances were not altered by removing one study and re-analysing the data of the remaining studies, which meant the data were comparatively stable and credible.

Discussion

Postoperative recurrence significantly affect the outcome of patients with HBV-related HCC [28, 29], IFN is hopeful to solve this problem, however, there is a controversy in the application of IFN [30, 31]. Hence, we conducted this meta-analysis and obtained valuable data that should provide valuable references and recommendations for the clinical application of IFN.

In this meta-analysis, IFN statistically reduced recurrence rates of patients with HBV-related HCC at 2, 3 and 5 years, and significantly improved the DFS at 1, 2 and 3 years, finally improved the OS at 1, 2, 3 and 5 years. However, the result showed IFN did not improve the OS at 8 years. Therefore, we suppose that the efficacy advantage of IFN may gradually weaken as time goes on. However, only two studies reported the 8-year OS, more clinical researches with longer observation time are needed to further support our hypothesis. On the other hand, researches were mainly conducted in Asia, there were only few studies including non-Asian population but not met the criteria. A study from Italy showed IFN could reduce late recurrence in patients with HCC after resection [32], which was consistent with our conclusions. Another study from Italy showed the time to recurrence was lower in patients with SVR by IFN compared with those with active HCV infection, but there was no difference in patients with SVR by IFN-free or IFN-based strategies [33], which was also consistent with our subgroup analysis.

To further understand the comprehensive efficacy of IFN on HCC, subgroup analysis was conducted. First, some studies have shown that NAs could improve the survival rate of patients with HCC after surgery [34]. To observe whether NAs affected the outcome, studies were grouped according to whether to use NAs or not. Results showed IFN significantly improved the 1-, 2- and 3-year OS in the subgroup without NAs, but there were no statistical significances in the subgroup with NAs. Did this mean that NAs had the same efficacy with IFN? As all we know, researches have shown that sustained virological response (SVR) might be the most important factor affecting the prognosis of patients with hepatitis virus‐related HCC [35,36,37,38]. Previous studies have shown that NAs could significantly reduce the risk of recurrence in patients with HBV‐related HCC and improve OS after curative surgery [39, 40]. However, there was only 1 study compared IFN combined with NAs therapy vs NAs therapy, the data available in this meta-analysis could not completely support this conclusion. More importantly, persistence of cccDNA is the main reason for recurrence of HBV in patients, even years after using NAs [2]. In fact, NAs have no effects of antitumor or immunomodulatory [41], in contrast, IFN can control angiogenesis and tumor immunoregulation, enhance the immunogenicity of tumor, finally directly inhibit the proliferation of tumor cells [16, 42]. Two studies have shown that the prognosis of naive HCC after hepatectomy might be improved through achieving SVR by additional INF [30, 43]. Some other studies even have shown that IFN was superior to NAs in reducing HCC development of CHB patients [44, 45]. Recent research found that IFN liberated T-cell cytotoxic capacities and enhanced antitumor activity of anti-PD-1-based immunotherapy in patients with unresectable HCC [12]. In addition, IFN appeared to be more effective in preventing HCC in Asia [46]. Some reports have demonstrated that even the patients with HCC undergoing INF-free therapy obtained SVR, the rates of early tumor occurrence and recurrence still remained high [30]. Therefore, IFN may have more advantages than NAs in preventing recurrence and improving survival rate of patients with HCC. In recent years, more and more studies have shown that IFN combined with NAs might be a better choice [47, 48]. Therefore, we recommend the clinical application of IFN combined with NAs therapy in reducing recurrence rate and improving OS in patients with HBV‐related HCC after curative surgery.

Second, studies were further grouped according to different operation selection. The result showed the clinical application of IFN in resection combined with TACE subgroup and TACE subgroup had more efficacy advantage, which had higher long-term OS and lower recurrence rate.

At the same time, subgroup analysis indicated that operation selection and whether to use NAs or not were not significant factors affecting heterogeneity. We believe that heterogeneity may be related to the following factors. First, study design. Although the well-designed RCS can achieve the similar effect with RCT, not all RCSs had high quality in this meta-analysis. Second, differences in patient characteristics. The application time of IFN, follow-up time, whether the patients accompanied with cirrhosis and the basic characteristics of tumors, such as the number, size, and location, were not exactly the same, which might influence clinical prognosis of HCC. Third, no study focused on whether IFN and NAs had the same efficacy when all patients reached SVR by IFN or NAs. Due to sample size limitation and incomplete data, we could not further perform subgroup analysis. They were also limitations of this study.

Hence, what should we do next? Our results should further encourage well-designed studies to address these limitations. High quality RCTs with large multicenter which contain appropriate application time of IFN, longer follow-up time, fixed characteristics of patient and tumor should be conducted. More importantly, we should focus on RCTs which contain the comparison between IFN and NAs. In addition, the clinical application of IFN combined with NAs recommended in this meta-analysis is also needed to be further studied. The well-designed RCTs will help us to understand the efficacy of IFN on HCC more accurately and comprehensively, finally guiding clinical application of IFN on treating HCC.

Conclusion

This meta-analysis may provide valuable data to re-evaluate the efficacy of IFN on HBV-related HCC. The results indicated that IFN might significantly reduce recurrence and improve DFS of patients with HBV-related HCC after curative surgery, and finally improve the survival. However, the efficacy advantage of IFN might gradually weaken as time goes on. The clinical application of IFN combined with NAs recommended in this meta-analysis is also needed to be further studied.