Efficacy of regorafenib in the treatment of advanced hepatocellular carcinoma: A systematic review and meta-analysis

Abstract

BACKGROUND

Regorafenib is approved as a second-line treatment for advanced hepatocellular carcinoma (HCC), but its comparative efficacy remains under evaluation.

AIM

To evaluate the efficacy and safety of regorafenib vs other second-line therapies in advanced HCC.

METHODS

This systematic review and meta-analysis were conducted in accordance with PRISMA guidelines. A comprehensive search of PubMed, EMBASE, Web of Science, and the Cochrane Library was performed on June 6, 2025. Studies were included if they reported at least one relevant clinical outcome: Overall survival, progression-free survival, objective response rate, or disease control rate. Data was extracted independently by two reviewers. Quality was assessed using the Cochrane Risk of Bias 2.0 tool for randomized controlled trials and the Newcastle-Ottawa Scale for cohort studies. Pooled effect estimates were calculated using random- or fixed-effects models depending on the degree of heterogeneity. Sensitivity analyses and Egger’s test were performed to evaluate the robustness of the results and potential publication bias.

RESULTS

Nine studies met inclusion criteria. Regorafenib significantly improved overall survival compared to controls [weighted mean difference = 2.54 months; 95% confidence interval (CI): 0.26-4.81; P < 0.05], but no significant benefit was observed for progression-free survival (weighted mean difference = 1.04; 95%CI: -1.27 to 3.36). The pooled objective response rate showed no overall difference, though regorafenib was inferior to nivolumab in subgroup analysis (odds ratio = 0.34; 95%CI: 0.20-0.58). Disease control rate did not differ significantly. No publication bias was detected.

CONCLUSION

Regorafenib offers a survival advantage in advanced HCC but does not significantly improve tumor response rates compared to alternative therapies.

Key Words: Regorafenib; Hepatocellular carcinoma; Second-line therapy; Overall survival; Meta-analysis

Core Tip: This study presents a comprehensive synthesis of current evidence regarding regorafenib’s clinical performance as a second-line treatment in patients with advanced hepatocellular carcinoma. Utilizing data from nine eligible studies, our meta-analysis provides an updated assessment of survival outcomes and tumor response metrics. Notably, regorafenib demonstrated a statistically significant improvement in overall survival compared to control groups, while objective tumor response rates did not show superiority, particularly when compared to immunotherapies such as nivolumab. Our results highlight regorafenib’s survival benefit and its role within a competitive therapeutic landscape.

INTRODUCTION

Hepatocellular carcinoma (HCC) represents the most common primary malignancy of the liver and constitutes a significant global health burden. Liver cancer is the sixth most commonly diagnosed cancer and the third leading cause of cancer-related mortality worldwide, with HCC accounting for approximately 75%-85% of all primary liver cancers[1,2]. The majority of HCC cases are diagnosed at an advanced stage, at which point curative options such as surgical resection, liver transplantation, or local ablative therapies are no longer feasible. As a result, systemic therapy remains the cornerstone of treatment for patients with advanced HCC. The therapeutic landscape for advanced HCC has evolved significantly over the past decade. Sorafenib, a multi-kinase inhibitor targeting vascular endothelial growth factor receptor, platelet-derived growth factor receptor, and RAF kinases, was the first agent to demonstrate survival benefit in this population and had been considered the standard of care for over a decade[3,4]. However, the efficacy of sorafenib is modest, and disease progression during or after sorafenib therapy is common, necessitating the development of effective second-line treatment strategies.

Regorafenib, an oral multi-kinase inhibitor structurally related to sorafenib, has emerged as a promising second-line agent. It targets a broader spectrum of kinases involved in tumor angiogenesis (e.g., vascular endothelial growth factor receptor 1-3, tyrosine kinase with immunoglobulin and epidermal growth factor homology domains 2), oncogenesis (e.g., KIT, RET, RAF-1, BRAF), and the tumor microenvironment (e.g., platelet-derived growth factor receptor, fibroblast growth factor receptor). The efficacy of regorafenib was first established in the RESORCE trial, a pivotal phase III randomized controlled trial (RCT) demonstrating that regorafenib significantly prolonged overall survival (OS) compared to placebo in patients with advanced HCC who had progressed on sorafenib and maintained adequate liver function (Child-Pugh A)[5-7]. Although widely adopted, evidence on the comparative effectiveness and safety of regorafenib vs alternative second-line treatments remains inconsistent. These discrepancies may reflect differences in patient selection, baseline liver function, regional treatment patterns, and study design[8-10].

Given the accumulating yet heterogeneous body of evidence, a comprehensive synthesis of the available data is warranted to clarify the therapeutic value of regorafenib in advanced HCC. A systematic review and meta-analysis allows for quantitative evaluation of efficacy outcomes, including OS, progression-free survival (PFS), objective response rate (ORR), and disease control rate (DCR). This type of analysis is essential for informing evidence-based clinical decisions and refining second-line treatment strategies. The objective of this study is to determine whether regorafenib provides a significant survival or tumor response benefit compared to other approved or commonly used second-line therapies in patients with advanced HCC.

MATERIALS AND METHODS

Search strategy

This systematic review was conducted in accordance with the PRISMA guidelines[11]. A comprehensive literature search was performed across four major electronic databases: PubMed, EMBASE, Web of Science, and the Cochrane Library. The search was carried out on June 6, 2025, and no restrictions were placed on the date of publication to ensure the inclusion of all potentially relevant studies. The full search strings for each database are provided in Supplementary Table 1. To minimize selection bias, no language restrictions were applied. The reference lists of all included articles and relevant review papers were manually screened to identify additional eligible studies. Duplicate records were removed using EndNote reference management software, and titles and abstracts were independently screened by two reviewers for relevance. Full-text articles were then assessed for eligibility according to the predefined inclusion and exclusion criteria.

Inclusion criteria and exclusion criteria

Studies were considered eligible for inclusion in the meta-analysis if they met all of the following criteria: (1) Population: Studies involving adult patients (≥ 18 years) diagnosed with advanced, unresectable, or metastatic HCC, confirmed by histological, cytological, or imaging criteria, were included; (2) Intervention: Studies that evaluated regorafenib as monotherapy or in combination with other agents for the treatment of advanced HCC; (3) Outcomes: Studies were required to report at least one of the following clinical outcomes: OS, PFS, ORR, and DCR; and (4) Study design: RCTs, prospective or retrospective cohort studies, and single-arm clinical trials were included.

Studies were excluded if they met any of the following criteria: (1) Population: Studies involving patients with early-stage HCC, recurrent HCC after curative resection, or mixed tumor types (e.g., cholangiocarcinoma) without separate data on advanced HCC were excluded; (2) Intervention: Studies that did not include regorafenib as a treatment modality, or those evaluating regorafenib in combination with non-standard experimental agents without control groups, were excluded; (3) Outcomes: Studies that did not report relevant efficacy or safety outcomes, or lacked sufficient data to estimate effect sizes or confidence intervals, were excluded; (4) Study design: Case reports, case series with fewer than 10 patients, narrative reviews, editorials, expert opinions, meeting abstracts, and conference proceedings were excluded; and (4) Duplicates: Duplicate publications or studies with overlapping data were excluded; in such cases, the most comprehensive and recent version was retained for analysis.

Literature screening and data extraction

Two independent reviewers screened the literature and extracted data in accordance with the predefined inclusion and exclusion criteria. Initial screening was conducted by reviewing titles and abstracts to exclude clearly irrelevant studies. Full texts of potentially eligible articles were then reviewed to determine final inclusion. Relevant data were extracted using a standardized data collection form and cross-checked for consistency. In cases of disagreement, a third reviewer was consulted to resolve discrepancies through discussion and consensus. The extracted information included the first author’s name, year of publication, study design, sample size, patient demographics (age and sex), details of the intervention (regorafenib dosage and administration), and reported clinical outcomes such as OS, PFS, ORR, and DCR.

Quality assessment

For RCTs, the Cochrane Risk of Bias 2.0 tool was employed[12]. This tool assesses bias across five domains: (1) The randomization process; (2) Deviations from intended interventions; (3) Missing outcome data; (4) Measurement of the outcome; and (5) Selection of the reported result. Each domain was judged as having a low risk of bias, some concerns, or high risk of bias, and an overall risk-of-bias judgment was assigned for each study accordingly.

For observational cohort studies, the Newcastle-Ottawa Scale (NOS) was used to assess study quality[13]. The NOS evaluates studies based on three broad perspectives: (1) Selection of study groups (maximum of 4 points); (2) Comparability of groups (maximum of 2 points); and (3) Ascertainment of outcomes (maximum of 3 points). A total score ranging from 0 to 9 was assigned to each study, with higher scores indicating better methodological quality. Studies scoring ≥ 7 were considered to be of high quality, scores of 5-6 were considered moderate quality, and scores < 5 were classified as low quality.

The risk of bias and methodological quality of all included studies were independently assessed by two reviewers. The results of the evaluations were cross-checked, and any discrepancies were resolved through discussion and consensus. If disagreements could not be reconciled, a third reviewer was consulted for adjudication. All quality assessment results were incorporated into the final analysis and considered in the interpretation of findings to account for potential sources of bias and study heterogeneity.

Statistical analysis

All statistical analyses were conducted using Review Manager version 5.4 (Cochrane Collaboration) and Stata version 18.0 (StataCorp, College Station, TX, United States). For studies reporting median OS or PFS with ranges or interquartile intervals, the values were converted to mean ± SD using established formulas (Hozo et al[14], 2005) to enable pooling of continuous data as weighted mean differences (WMDs). Time to progression data were handled in the same manner to ensure consistency across survival outcomes. Heterogeneity among studies was evaluated using Cochran’s Q test and quantified by the I² statistic, which estimates the percentage of total variation due to heterogeneity rather than chance. An I² value ≤ 50% and a P value ≥ 0.10 were considered to indicate low or acceptable heterogeneity, whereas an I² > 50% or P < 0.10 suggested substantial heterogeneity. When heterogeneity was low (I² ≤ 50%), a fixed-effects model (Mantel-Haenszel method) was applied, assuming that all studies estimated a common underlying effect size. When heterogeneity was moderate to high (I² > 50%), a random-effects model (DerSimonian-Laird method) was used, which accounts for between-study variability and provides more conservative estimates. To explore sources of heterogeneity, subgroup analyses were performed when sufficient data were available, based on factors such as study design, geographic region, line of treatment, or type of control intervention. Additionally, sensitivity analyses were conducted by sequentially removing one study at a time to assess the robustness and stability of the pooled estimates. Publication bias was evaluated using Egger’s regression test, with a P value < 0.05 suggesting potential bias. A two-sided P value < 0.05 was considered statistically significant for all analyses unless otherwise specified.

RESULTS

Search results and study selection

A total of 1049 records were identified through database and registry searches, including 1021 records from databases (PubMed, EMBASE, Web of Science, and the Cochrane Library) and 28 from trial registries. Before screening, 764 records were removed due to duplication (n = 335), automatic exclusion by filtering tools (n = 253), or other reasons (n = 176). The remaining 285 records were screened by title and abstract, of which 251 were excluded as irrelevant. Full-text retrieval was attempted for 34 articles, but 3 could not be accessed. Among the 31 full-text articles assessed for eligibility, 22 were excluded for the following reasons: Reviews (n = 9), sequentially published analyses (n = 5), insufficient data (n = 5), and absence of control groups in clinical trials (n = 3). In addition, 39 reports were identified from supplementary searches (websites, organizational reports, and citation tracking). After excluding 32 reports that could not be retrieved, 7 were assessed for eligibility, among which 3 were reviews and 4 lacked sufficient data. Finally, data from 9 studies met the inclusion criteria and were incorporated into the meta-analysis[15-23] (Figure 1). No additional reports were identified beyond these included studies.

Figure 1 PRISMA flow diagram of study selection process. This flowchart outlines the selection of studies included in the meta-analysis. A total of 1088 records were identified from databases and other sources. After removing duplicates and ineligible records, 9 studies met the inclusion criteria and were included in the final analysis.

Study characteristics

Nine studies were included in this meta-analysis, comprising two RCTs and seven retrospective studies. All studies evaluated the use of regorafenib as second-line therapy in patients with advanced HCC. Sample sizes ranged from 36 to 379 in the treatment groups and from 28 to 331 in the control groups. The mean or median patient age varied from approximately 58-68 years. Most study populations were predominantly male. While regorafenib was consistently used in the treatment arms, control interventions varied and included placebo, cabozantinib, nivolumab, and best supportive care. Despite some variability in study design and comparator regimens, all included studies reported relevant clinical outcomes necessary for meta-analysis, including survival metrics and adverse event profiles (Table 1).

Table 1 Characteristics of studies included in the meta-analysis.

Ref.	Study design	Age, years (T: C)	Sex (M/F, T: C)	Sample size (T: C)	Treatment group	Control group	Child-Pugh classification
Lee et al[15], 2024	Retrospective cohort study	63: 59	120/17: 45/7	137: 52	Regorafenib	Nivolumab	Regorafenib group: Child-Pugh B (24.1%); nivolumab group: Child-Pugh B (42.3%)
Adhoute et al[16], 2022	Retrospective, multicenter	68: 68	53/5: 24/4	58: 28	Regorafenib	Cabozantinib	Combined A/B proportion approximately 24%-25%
Casadei-Gardini et al[19], 2021	Retrospective study	Not reported	222/46: 264/67	278: 331	Regorafenib	Cabozantinib	Not reported
Iavarone et al[18], 2021	Retrospective study	60: 61	27/9: 38/7	36: 45	Regorafenib	BSC	Not reported
Kuo et al[17], 2021	Retrospective study	63.4 ± 10.7: 62.2 ± 10.1	44/14: 23/9	58: 32	Regorafenib	Nivolumab	Not reported
Choi et al[21], 2020	Retrospective study	58.5 ± 9.4: 56.9 ± 10.0	202/21: 125/25	223: 150	Regorafenib	Nivolumab	Child-Pugh A: 132 patients; Child-Pugh B: 71 patients
Lee et al[20], 2020	Retrospective cohort study	62: 61	83/54: 39/9	102: 48	Regorafenib	Nivolumab	Not reported
Finn et al[22], 2018	RCT	64: 62	333/46: 171/23	379: 194	Regorafenib	Placebo	Child-Pugh A only
Bruix et al[23], 2017	RCT	64: 62	333/46: 171/23	379: 194	Regorafenib	Placebo	Child-Pugh A only

RCT: Randomized controlled trial; T: Treatment group; C: Control group; M: Male; F: Female; BSC: Best supportive care.

Results of quality assessment

The two RCTs were assessed using the Cochrane Risk of Bias 2.0 tool. Both studies demonstrated a low risk of bias across most domains. However, one trial (Finn et al[22], 2018) showed a high risk of bias in the domain of allocation concealment, while all other domains, including random sequence generation, blinding, outcome assessment, attrition, and reporting, were judged to be at low risk (Figure 2). For the seven retrospective cohort studies, quality was evaluated using the NOS. The total scores ranged from 8 to 9 out of a possible 9, indicating moderate to high methodological quality. Most studies received full scores in the domains of representativeness, ascertainment of exposure, and adequacy of follow-up. Comparability of cohorts was generally well addressed, with most studies controlling for key confounders through design or analysis. Minor limitations were observed in a few studies with respect to the length of follow-up or documentation of baseline outcome absence (Table 2). Overall, the included studies were considered to be of sufficient quality to support the validity of the meta-analysis findings.

Figure 2 Risk of bias assessment for randomized controlled trials using the Cochrane RoB 2.0 tool. The figure presents the domain-specific and overall risk of bias judgments for the two included randomized controlled trials. One trial exhibited high risk in allocation concealment, while both demonstrated low risk in other domains such as blinding, outcome measurement, and attrition.

Table 2 The quality assessment according to Newcastle-Ottawa Scale of each retrospective study.

Ref.	Representativeness of the exposed cohort	Selection of the non-exposed cohort	Ascertainment of exposure	Demonstration that outcome of interest was not present at start of study	Comparability of cohorts on the basis of the design or analysis	Assessment of outcome	Was follow-up long enough	Adequacy of follow up of cohorts	Total score
Lee et al[15], 2024	★	★	★	★	★	★	★	★	8
Adhoute et al[16], 2022	★	★	★	★	★★	★	★	★	9
Casadei-Gardini et al[19], 2021		★	★	★	★★	★	★	★	8
Iavarone et al[18], 2021	★	★	★	★	★★	★		★	8
Kuo et al[17], 2021	★	★	★	★	★★	★	★	★	9
Choi et al[21], 2020	★	★	★	★	★★	★	★	★	9
Lee et al[20], 2020	★	★	★	★	★★	★		★	8

Effect of regorafenib on OS in patients with advanced HCC

A total of seven studies reported data on OS for patients receiving regorafenib compared with control interventions. Given the presence of significant heterogeneity among the included studies (I² = 63.2%, P = 0.012), a random-effects model was applied to pool the results. The meta-analysis demonstrated that regorafenib was associated with a significantly prolonged OS compared to the control group [WMD = 2.54; 95% confidence interval (CI): 0.26-4.81; P < 0.05] (Figure 3A). To evaluate the robustness of the pooled estimate, sensitivity analysis was conducted by sequentially excluding each individual study. The results remained consistent across iterations, indicating that no single study had a disproportionate influence on the overall effect size (Figure 3B).

Figure 3 Overall survival comparison between regorafenib and control treatments and sensitivity analysis of the pooled effect. A: Forest plot of overall survival (OS) comparing regorafenib vs control therapies in patients with advanced hepatocellular carcinoma. A random-effects model was used due to moderate heterogeneity (I² = 63.2%). Regorafenib was associated with a statistically significant improvement in OS (weighted mean difference = 2.54 months; 95% confidence interval: 0.26-4.81; P < 0.05); B: Sensitivity analysis of pooled OS results by sequential study exclusion. The analysis demonstrates the robustness of the OS effect estimate, showing consistent results with the omission of each individual study. CI: Confidence interval.

Effect of regorafenib on PFS in patients with advanced HCC

Five studies reported data on PFS in patients treated with regorafenib vs control interventions. Due to substantial heterogeneity among studies (I² = 95.5%, P < 0.001), a random-effects model was used to pool the data. The combined results showed no significant difference in PFS between the regorafenib and control groups (WMD = 1.04; 95%CI: -1.27 to 3.36) (Figure 4A). To test the robustness of the findings, a sensitivity analysis was performed by removing each study one at a time. The overall results remained stable, indicating that no single study had a significant influence on the pooled estimate (Figure 4B).

Figure 4 Progression-free survival comparison between regorafenib and control treatments and sensitivity analysis of the pooled estimate. A: Forest plot of progression-free survival (PFS) comparing regorafenib vs control therapies. A random-effects model was applied due to substantial heterogeneity (I² = 95.5%). No statistically significant difference in PFS was observed between groups (weighted mean difference = 1.04; 95% confidence interval: -1.27 to 3.36; P > 0.05); B: Sensitivity analysis of pooled PFS results by sequential study exclusion. The results remained stable across iterations, confirming that no individual study had a major impact on the overall pooled effect. CI: Confidence interval.

Pooled and subgroup analysis of ORR between regorafenib and control treatments

ORR data were reported in five included studies comparing regorafenib with various control therapies in patients with advanced HCC. Substantial heterogeneity was observed across studies (I² = 79.7%, P < 0.001), warranting the use of a random-effects model for data synthesis. The pooled analysis demonstrated no statistically significant difference in ORR between the regorafenib and control groups (OR = 0.56; 95%CI: 0.21-1.53; P > 0.05), indicating that, in general, regorafenib did not achieve a higher tumor response rate compared with the control interventions assessed. To investigate potential sources of heterogeneity, a subgroup analysis was performed focusing on studies comparing regorafenib with nivolumab, an immune checkpoint inhibitor. In this subgroup, heterogeneity was not observed (I² = 0.0%, P = 0.826), suggesting consistency among the included studies. The pooled results revealed that regorafenib was associated with a significantly lower ORR than nivolumab (OR = 0.34; 95%CI: 0.20-0.58; P < 0.05), highlighting a possible superiority of immune checkpoint blockade in eliciting objective tumor responses in this patient population (Figure 5). These findings suggest that while regorafenib may offer clinical benefit in selected contexts, its anti-tumor activity, as measured by radiographic response, may be less pronounced than that of immunotherapeutic agents such as nivolumab.

Figure 5 Forest plot of objective response rate for regorafenib vs control treatments, including subgroup analysis with nivolumab. The overall analysis showed no significant difference in objective response rate (odds ratio = 0.56; 95% confidence interval: 0.21-1.53; P > 0.05; I² = 79.7%). However, subgroup analysis demonstrated a significantly lower objective response rate with regorafenib compared to nivolumab (odds ratio = 0.34; 95% confidence interval: 0.20-0.58; P < 0.05; I² = 0.0%). Subgroup analyses were conducted based on the type of comparator: Regorafenib vs nivolumab (n = 4 studies); regorafenib vs placebo (n = 1 study). The square markers in the forest plot have been scaled according to the inverse-variance weight of each study, where larger squares indicate studies with greater statistical weight. CI: Confidence interval.

Comparison of DCR between regorafenib and control therapies in advanced HCC

Five studies reported data on DCR among patients receiving regorafenib compared to control treatments. Statistical analysis revealed no significant heterogeneity among the included studies (I² = 0.0%, P = 0.498), indicating consistency in the direction and magnitude of effect sizes across studies. Consequently, a fixed-effects model was applied to synthesize the data. The pooled results showed no statistically significant difference in DCR between the regorafenib and control groups (OR = 1.09; 95%CI: 0.81-1.45; P > 0.05), suggesting that the likelihood of achieving disease stabilization or partial response was comparable across treatment strategies (Figure 6). These findings indicate that regorafenib provides a similar disease control benefit as alternative therapies in the second-line setting for advanced HCC. The absence of heterogeneity supports the reliability of this conclusion across different study populations and clinical contexts.

Figure 6 Forest plot of disease control rate comparing regorafenib and control therapies. Using a fixed-effects model (I² = 0.0%), the pooled analysis indicated no statistically significant difference in disease control rate between treatment groups (odds ratio = 1.09; 95% confidence interval: 0.81-1.45; P > 0.05), suggesting comparable disease stabilization efficacy. CI: Confidence interval.

Publication bias

Given that fewer than 10 studies were included in each meta-analysis, Egger’s linear regression test was applied to assess potential publication bias, as funnel plot asymmetry tests are considered less reliable with limited study numbers. The results of Egger’s test showed no statistically significant evidence of publication bias across any of the pooled analyses (P > 0.05 for all), suggesting that the meta-analytic findings were unlikely to be influenced by selective reporting.

DISCUSSION

The present meta-analysis indicates that regorafenib confers a significant survival benefit in advanced HCC patients who have failed prior therapy. Pooled results showed that regorafenib significantly improves OS compared to control treatments, with a weighted mean OS prolongation of approximately 2.5 months in favor of regorafenib (WMD = 2.54 months, 95%CI: 0.26-4.81; P < 0.05). This modest but significant improvement in OS underscores regorafenib’s efficacy in extending patient survival beyond what is achieved with placebo or other second-line options. In contrast, PFS was not significantly improved with regorafenib in the pooled analysis (WMD = 1.04 months, 95%CI: -1.27 to 3.36), suggesting that regorafenib does not substantially delay radiographic tumor progression relative to comparators. The divergence between the clear OS benefit and the minimal PFS gain may indicate that regorafenib’s survival advantage stems from factors beyond just tumor growth delay - for example, disease stabilization or post-progression outcomes - rather than dramatically prolonging the time to progression. Tumor response outcomes in this meta-analysis further highlight regorafenib’s disease-stabilizing effect[24,25]. The ORR - the proportion of patients achieving a significant tumor size reduction - was not significantly different overall between regorafenib and the pooled controls (OR = 0.56, 95%CI: 0.21-1.53; P > 0.05). This implies that regorafenib produces objective tumor responses in a comparable fraction of patients as other second-line treatments when all studies are considered together. Likewise, the DCR - which encompasses patients with stable disease or better - was similar between regorafenib and control groups (OR = 1.09, 95%CI: 0.81-1.45), indicating no significant difference in the overall likelihood of achieving disease control.

Notably, subgroup analyses revealed an important nuance in treatment response. When regorafenib was specifically compared against the immune checkpoint inhibitor nivolumab, regorafenib demonstrated a significantly lower ORR (OR = 0.34, 95%CI: 0.20-0.58). This indicates that patients receiving nivolumab were substantially more likely to experience an objective tumor response than those receiving regorafenib in the second-line setting. The finding aligns with clinical expectations: Immunotherapy can induce deep and durable tumor regressions in a subset of HCC patients, whereas multi-kinase inhibitors like regorafenib more often yield stable disease. Despite this disparity in objective response, it is worth noting that disease control (including stable disease) remained high with regorafenib. Regorafenib-treated patients often achieved tumor growth stabilization, which likely contributed to the prolonged OS even though frank tumor shrinkage was less frequent compared to nivolumab[26,27].

These results are broadly consistent with trends in the literature on second-line therapies for advanced HCC and reflect known efficacy profiles of regorafenib and newer agents. The OS benefit observed with regorafenib in this analysis mirrors what has been reported in pivotal clinical trials, which established regorafenib’s role after prior treatment failure. In those trials, regorafenib demonstrated a median OS improvement of approximately 2-3 months over placebo, consistent with the pooled result in this study[23,28]. This magnitude of OS gain is comparable to that observed with other second-line targeted therapies such as cabozantinib and ramucirumab. Network meta-analyses have suggested similar efficacy between regorafenib and cabozantinib, with overlapping confidence intervals for OS and PFS[29]. The lack of significant difference in PFS in the present analysis is likely influenced by the inclusion of studies with active comparators, which naturally diminish the detectable treatment effect compared to placebo-controlled settings. Moreover, the fact that OS benefits are observed despite modest PFS gains suggests that downstream therapies, post-progression management, and quality of disease control may contribute to prolonged survival[30]. The superior ORR observed with nivolumab is in line with known patterns in immunotherapy trials, where checkpoint inhibitors produce higher response rates but not necessarily superior OS in unselected populations. Single-agent immunotherapies in second-line HCC have demonstrated ORRs in the range of 15%-20%, compared to approximately 10% with tyrosine kinase inhibitors like regorafenib. However, confirmatory phase III trials for immunotherapies have often failed to reach statistical significance for OS endpoints[31]. This meta-analysis therefore reflects a familiar clinical trade-off: Regorafenib offers modest and consistent survival benefit, whereas immunotherapy may produce dramatic responses in fewer patients but with less predictable survival advantage across the entire population. The findings also fit within the evolving treatment landscape of HCC. Historically, most patients in second-line settings had progressed on sorafenib. However, with the increasing use of immune checkpoint inhibitor combinations in first-line therapy (e.g., atezolizumab-bevacizumab), the applicability of regorafenib after first-line immunotherapy is still under investigation. Nevertheless, regorafenib remains a valuable therapeutic option where immunotherapy is contraindicated, unavailable, or not tolerated[32,33].

Because regorafenib is approved as a second-line treatment following sorafenib failure, the inclusion of studies that enrolled patients previously treated with sorafenib, such as Finn et al[22], is consistent with current clinical practice and strengthens the external validity of the findings. Although this may introduce some degree of clinical heterogeneity, sensitivity analysis demonstrated that the exclusion of this study did not materially alter the pooled OS estimate, supporting the robustness of the results. Furthermore, the high risk of bias attributed to Finn et al[22] in the domain of allocation concealment was based on the lack of a clearly described method (e.g., central randomization or sealed envelopes). In accordance with Cochrane recommendations, unclear reporting was conservatively judged as high risk to avoid underestimating potential selection bias. However, this methodological limitation is unlikely to have influenced the overall interpretation of efficacy, given the consistent findings in sensitivity analysis.

Regorafenib is a multi-kinase inhibitor that exerts anti-tumor effects through both direct tumoricidal activity and modulation of the tumor microenvironment[34]. Specifically, it targets vascular endothelial growth factor receptor and platelet-derived growth factor receptor, key mediators of angiogenesis and stromal support. By inhibiting these signaling pathways, regorafenib suppresses neovascularization, normalizes abnormal tumor vasculature, and potentially enhances immune infiltration, thereby contributing to improved tumor control and survival outcomes[35]. The observed discrepancy between OS and PFS is a common phenomenon in trials of anti-angiogenic and immunomodulatory agents. While PFS captures radiographic progression, OS reflects the cumulative benefits of treatment, including effects on micrometastatic disease, host immunity, and post-progression therapies. Therefore, OS may better reflect regorafenib’s multifaceted impact on disease biology, beyond early tumor shrinkage or radiographic stability. The pooled analysis revealed that regorafenib significantly prolonged OS by 2.54 months compared to control therapies. This degree of OS benefit is clinically relevant and aligns with findings from other second-line agents, such as cabozantinib and ramucirumab. Although PFS did not show a statistically significant difference, this result may be influenced by heterogeneity and limited statistical power across the included studies.

This meta-analysis confirms that regorafenib provides a significant survival benefit as a second-line treatment for advanced HCC. Clinicians should consider regorafenib as an effective therapeutic option for patients who have failed prior systemic therapy, particularly where immune checkpoint inhibitors are not suitable. While its tumor shrinkage potential is limited compared to immunotherapy, regorafenib consistently provides disease control and survival extension, making it a reliable component of sequential systemic therapy. This study has several notable strengths. First, it includes a comprehensive analysis of multiple outcome domains, including survival, tumor response, and disease control. Sensitivity analyses confirmed the robustness of the primary findings, and Egger’s test showed no significant publication bias. Additionally, the inclusion of both randomized and observational studies provides a broader view of regorafenib’s real-world effectiveness. However, limitations must also be acknowledged. The number of included studies was relatively small, especially for subgroup analyses. Most studies were retrospective in design, introducing risks of selection bias and confounding. The substantial heterogeneity observed in PFS and ORR outcomes suggests underlying variability in study populations, comparators, and assessment methods. Although subgroup analysis addressed part of this variability, the findings should be interpreted with caution. Furthermore, cross-trial comparisons between regorafenib and immunotherapies are limited by the lack of head-to-head randomized trials. Lastly, with the systemic treatment landscape rapidly evolving, the applicability of these findings to patients receiving immunotherapy-based first-line treatment remains uncertain.

CONCLUSION

In conclusion, regorafenib significantly improves OS in patients with advanced HCC receiving second-line therapy, although no clear benefit was observed for PFS, response rate, or disease control. While immunotherapy may offer higher response rates, regorafenib remains a valuable treatment option.

ACKNOWLEDGEMENTS

We are grateful to everyone involved in this research.