Comparative effectiveness and safety of second-line therapies and dosing regimens for advanced hepatocellular carcinoma: a network meta-analysis.

Introduction
Hepatocellular carcinoma (HCC) is the primary type of carcinoma of the liver, accounting for approximately 90% of all liver cancer instances. It stands as a predominant factor contributing to cancer-related illness and death on a global scale [1,2]. WHO anticipates a 55% rise in carcinoma of the liver incidence between 2020 and 2040, resulting in a projected 1.3 million fatalities, a substantial increase above 2020 figures [3]. The increasing global burden of HCC poses substantial economic and social challenges, contributing to both direct healthcare costs and indirect costs such as lost productivity [4]. In countries with high rates of liver cancer, the economic burden is particularly pronounced, exacerbating the need for effective and timely therapeutic strategies.
Given the insidious nature of HCC, where symptoms often present at mature phases, many individuals are diagnosed too late for therapeutic procedures like surgical interventions, liver transplantation, or radiofrequency ablation. As a result, advanced-stage HCC requires the implementation of second-line treatment strategies, which are pivotal in improving the endurance of patients and the enhancement of their living standards [1,5]. Second-line therapies, including immune checkpoint inhibitors, tyrosine kinase inhibitors, and monoclonal antibodies, have evolved into essential alternatives following the progression or failure of first-line treatments [6]. However, despite their promising potential, the effectiveness of these treatments varies widely, and the optimal approach remains a subject of ongoing research.
Several second-line treatments, including sorafenib, lenvatinib, nivolumab, pembrolizumab, cabozantinib, and regorafenib, have undergone assessment in clinical trials, each indicating varying degrees of effectiveness in controlling disease progression and enhancing survival [5,7,8]. However, challenges remain, particularly due to the significant heterogeneity in patient responses to these treatments and the presence of adverse effects. These limitations necessitate the search for more tailored therapeutic strategies, particularly with the integration of immune and targeted therapies, that seek to leverage the body’s defenses and specifically address the pathways associated with tumor cells, respectively. The increasing recognition of the potential of these therapies to improve survival outcomes for advanced HCC patients highlights their clinical significance.
Targeted therapies and immunotherapies exhibit significant potential in the management of HCC. They are designed to inhibit particular molecules that can contribute to tumorigenesis and advancement, whereas immunotherapies operate by augmenting the efficacy of the body’s defenses to identify and combat tumor cells [9,10]. Both approaches have demonstrated efficacy in extending longevity and enhancing the quality of existence for individuals suffering from advanced HCC. Several investigations have illustrated the prospective advantages of many medications, including the medications atezolizumab, bevacizumab, and cabozantinib being used in clinical practice [11,12]. For example, clinical investigations have demonstrated that combinations of immune checkpoint inhibitors with tyrosine kinase inhibitors with tyrosine kinase inhibitors or other targeted agents offer improved survival outcomes compared to traditional therapies [13–15]. However, despite these advancements, the relative effectiveness of different drugs and dosages remains unclear. No studies have comprehensively analyzed the comparative effectiveness of various combinations of drugs and dosages in advanced HCC, and determining the most effective approach is a critical gap in current research.
In light of these challenges, network meta-analysis serves as a robust methodology for evaluating the effectiveness of various therapeutic interventions through the synthesis of data derived from numerous randomized controlled trials (RCTs). Network meta-analysis (NMA) allows for comparative analyses between treatments that have not undergone direct evaluation in separate tests, providing a more comprehensive understanding of relative treatment effectiveness [16]. Given the increasing number of RCTs investigating second-line therapies for HCC, this investigation seeks to perform a network systematic review to ascertain the most effective drug regimens and dosages for the management of advanced HCC. By integrating information from multiple high-quality trials, our research seeks to clarify the optimal therapeutic strategies for improving outcomes in HCC patients.

Methods
This thorough examination and network meta-analysis is performed in accordance with 2020 PRISMA standards, as well as the extended PRISMA statement specifically for network meta-analyses concerning healthcare interventions [17,18]. This study, being a meta-analysis, never needed authorization from an ethics committee or the inclusion of consent statements.

Sources of data and methodology for inquiry
Articles pertinent to the subject were discerned via electronic literature searches performed across PubMed, Medline, Embase, PsycINFO, the Cochrane Central Register of Controlled Trials, and Web of Science, spanning from the inception of each database up to 31 January 2025. The search terms included ‘hepatocellular carcinoma’ in combination with alternative therapeutic approaches like ‘apatinib’, ‘regorafenib’, ‘ramucirumab’, ‘pembrolizumab’, ‘camrelizumab’, and ‘cabozantinib’, among others. The comprehensive search approach, detailing the precise terms and their various combinations, can be found in Supplementary File 1, Supplemental digital content 1, https://links.lww.com/EJGH/B214.
Furthermore, the bibliographies of qualifying papers and the citation lists concerning systematic assessments released over the last 5 years were meticulously examined to uncover any further pertinent research. Two reviewers conducted an independent screening of titles and abstracts to determine eligibility, subsequently engaging in a thorough full-text assessment of articles deemed potentially relevant. Disagreements were addressed through thoughtful dialogue or by seeking the insights of an impartial third examiner.

Criteria for selecting research
Investigations were included based on the fulfillment of the following criteria: (1) the population comprised adult patients (over 18 years) with HCC and a Child–Pugh score of A or B; (2) the intervention group received second-line immunotherapy or targeted therapy as monotherapy; (3) the sole distinction the distinction across the experimental group and the group receiving a placebo resided in the administration of second-line immunotherapy or targeted therapy; (4) not less than one of the subsequent outcomes was documented: overall survival (OS), progression-free survival (PFS), thrombotic thrombocytopenic purpura, objective response rate (ORR), disease control rate (DCR), the occurrence of all-grade and grade 3–4 adverse events (AEs), or the frequency of therapy removal attributable to AEs; and (5) the publication was available in English or Chinese.
Studies were excluded if (1) the second-line treatment was not a monotherapy (i.e. it involved combination with other drugs); (2) the treatment approach was not clearly described; or (3) the results lacked the presentation of mean values and variances, and the writers failed to address inquiries regarding the data. Two reviewers independently screened potentially relevant papers according to these criteria by examining titles, abstracts, and, when necessary, full texts to determine eligibility.

Data extraction
EndNote X9 was used to manage eligible studies and avoid redundancy. Two independent reviewers extracted relevant publication details (e.g. authors, titles, and publication years), the quantity of individuals, their traits (e.g. age and sex), the interventions applied, and the measures of outcomes (refer to Supplementary File 2, Supplemental digital content 1, https://links.lww.com/EJGH/B214). Missing means and standard deviations were transformed in alignment with the established protocols specified in the Cochrane Handbook [19]. If essential data could not be obtained through these methods, the authors were contacted at least four times over a 6-week period to request the missing information.

Evaluation of potential bias
The evaluation of the risk of bias in the research examined was performed at the study level utilizing the revised Cochrane risk-of-bias tool [20]. This evaluation encompassed the subsequent areas: the process of randomization, variations from the planned interventions, absence of results, assessment of outcomes, and the criteria for selecting the reported results. Discrepancies in the evaluations were addressed through the engagement of a third reviewer, thereby guaranteeing a thorough and impartial analysis of the studies incorporated.

Data coding
Each second-line treatment evaluated in the included studies was coded according to drug type and dose, specifically ‘Apatinib 750 mg once daily (QD)’, ‘Axitinib 5 mg BID’, ‘Brivanib 800 mg once daily (OD)’, ‘Cabo 100 mg QD’, ‘Cabo 60 mg QD’, ‘Codrituzumab 1600 mg Q2W’, ‘Everolimus 7.5 mg QD’, ‘Pembro 200 mg Q3W’, ‘Ram 8 mg/kg Q2W’, ‘Regorafenib 160 mg QD’, ‘Tivantinib 120 mg BID’, and ‘Tivantinib 240 mg BID’. The control group, comprising either placebo or best supportive care, was uniformly coded as ‘CON’.

Data analysis
Every analysis was conducted utilizing Stata software (version 17.0, StataCorp LLC, College Station, Texas, USA). A comprehensive network meta-analysis was undertaken to evaluate the success of various treatments for advanced HCC; a network visualize was designed to demonstrate the distinctions among the diverse interventions and to ascertain the appropriateness of the network structure. Given the anticipated clinical heterogeneity, a random-effects model was employed to tackle the variability present both within and between studies.
Continuous outcomes, including OS and PFS, were evaluated through mean difference (MD) accompanied by 95% confidence intervals (CIs). In contrast, categorical outcomes such as ORR, DCR, and AEs were analyzed utilizing odds ratios (ORs) with 95% CIs. The evaluation of heterogeneity was conducted through the application of the I2 statistic. With thresholds of 25, 50, and 75% denoting low, moderate, and high heterogeneity, in that order. NMAs were performed in the context of Bayesian analysis utilizing the ‘network’ and ‘mvmeta’ packages in Stata. The treatments were evaluated through the SUCRA, with elevated SUCRA values signifying a more advantageous relative effect. The treatments were evaluated through the SUCRA, with elevated SUCRA values signifying a more advantageous relative effect [21]. Prediction interval plots were also utilized to deeper examine diversity and interpret variability in effect sizes. The statistical examinations employed a multifaceted methodology, with a P-value of less than 0.05 considered to be significant statistically.

Outcomes

Features of the included research
A comprehensive search yielded a total of 2837 records. Subsequent to the elimination of 1738 redundant records, 1099 articles underwent evaluation of titles and summaries. Following the omission of 1025 publications through a review of headings and summary sections, 74 full-text papers were assessed for their approval. Ultimately, 26 RCTs involving 10 368 patients with advanced HCC were incorporated into the systemic review and network meta-analyses (Fig. 1) [8, 15, 22–45]. The attributes of the studies included are elaborated upon in Supplementary File 2, Supplemental digital content 1, https://links.lww.com/EJGH/B214.
The publications emerged within the timeframe of 2013–2022, with the median year of publication being 2019. The study population sizes varied between 41 and 851 participants, yielding a median of 413 participants across the studies. The ages of the subjects encompassed 51–71 years, with a median age established at 64.
Regarding treatment regimens, seven studies employed Ram 8 mg/kg Q2W; five studies used Cabo 60 mg QD; three studies utilized Pembro 200 mg Q3W; two studies involved Regorafenib 160 mg QD; two studies used Tivantinib 120 mg BID. One study each used Apatinib 750 mg QD, Axitinib 5 mg BID, Brivanib 800 mg OD, Cabo 100 mg QD, Codrituzumab 1600 mg Q2W, Everolimus 7.5 mg QD, and Tivantinib 240 mg BID.

Network meta-analysis

Overall survival
A total of 10 325 advanced patients with HCC were incorporated into the network meta-analysis for OS, which comprised 25 studies. Various second-line treatments and dosages were compared for their effects on OS. Figure 2a shows the direct comparisons and sample size distribution among the different treatments. According to the SUCRA rankings (Fig. 3a), the top three treatments for improving OS were Cabo 60 mg QD (84.3%), Axitinib 5 mg BID (74.5%), and Ram 8 mg/kg Q2W (73.4%), while Codrituzumab 1600 mg Q2W ranked the lowest at 10.2%.
As shown in Table 1, compared to the CON, Cabo 60 mg QD (MD = 3.36, 95% CI = 2.01, 4.70), Ram 8 mg/kg Q2W (MD = 2.79, 95% CI = 1.63, 3.96), Regorafenib 160 mg QD (MD = 2.80, 95% CI = 0.64, 4.96), and Pembro 200 mg Q3W (MD = 2.75, 95% CI = 1.00, 4.50) significantly improved OS.

Progression-free survival
A comprehensive network meta-analysis concerning PFS encompassed 26 studies, involving a total of 10 368 patients. The effect of various second-line treatments and dosages on PFS was assessed. Figure 2b shows the direct comparisons and sample size distribution. According to the SUCRA rankings (Fig. 3b), the top three treatments for improving PFS were Cabo 60 mg QD (97.5%), Apatinib 750 mg QD (86.7%), and Axitinib 5 mg BID (66.2%), while the control group ranked the lowest at 11.2%.
As shown in Table 2, compared to the CON group, Cabo 60 mg QD (MD = 3.26, 95% CI = 2.59, 3.94), Apatinib 750 mg QD (MD = 2.60, 95% CI = 1.08, 4.11), Axitinib 5 mg BID (MD = 1.70, 95% CI = 0.19, 3.21), Regorafenib 160 mg QD (MD = 1.60, 95% CI = 0.52, 2.68), and Ram 8 mg/kg Q2W (MD = 1.52, 95% CI = 0.94, 2.09) significantly improved PFS.

Objective response rate
The network meta-analysis for ORR encompassed 10 368 subjects across 26 trials. Various second-line treatments and dosages were compared for their effect on ORR. Figure 2c presents the direct comparisons and sample size distribution. According to the SUCRA rankings (Fig. 3c), the top three treatments for improving ORR were Brivanib 800 mg OD (76.1%), Cabo 60 mg QD (73.8%), and Pembro 200 mg Q3W (72.9%), while the control group ranked the lowest at 14.1%.
As shown in Table 3, compared to the CON group, Brivanib 800 mg OD (OR = 7.13, 95% CI = 1.42, 35.88), Cabo 60 mg QD (OR = 5.68, 95% CI = 3.08, 10.47), Pembro 200 mg Q3W (OR = 5.68, 95% CI = 2.78, 11.60), Apatinib 750 mg QD (OR = 5.28, 95% CI = 1.30, 21.37), Ram 8 mg/kg Q2W (OR = 4.37, 95% CI = 2.12, 9.01), and Regorafenib 160 mg QD (OR = 2.74, 95% CI = 1.30, 5.77) significantly improved ORR.

Disease control rate
A total of 10 368 individuals from 26 trials were incorporated into the DCR NMA. The effect of various second-line treatments on DCR was evaluated. Figure 2d shows the direct comparisons and sample size distribution. According to the SUCRA rankings (Fig. 3d), the top three treatments for improving DCR were Apatinib 750 mg QD (78.7%), Regorafenib 160 mg QD (72.3%), and Cabo 60 mg QD (71.4%), while the control group ranked the lowest at 5.2%.
As shown in Table 4, compared to the CON group, Apatinib 750 mg QD (OR = 3.92, 95% CI = 1.76, 8.71), Regorafenib 160 mg QD (OR = 3.31, 95% CI = 1.95, 5.64), Cabo 60 mg QD (OR = 3.22, 95% CI = 2.31, 4.48), Axitinib 5 mg BID (OR = 3.42, 95% CI = 1.19, 9.83), Ram 8 mg/kg Q2W (OR = 2.63, 95% CI = 1.88, 3.69), Brivanib 800 mg OD (OR = 2.32, 95% CI = 1.05, 5.08), and Pembro 200 mg Q3W (OR = 1.87, 95% CI = 1.19, 2.93)greatly enhanced DCR.

Adverse events
An aggregate of 10 368 subjects from 26 trials was incorporated into the NMA concerning AEs. Various second-line treatments and dosages were assessed for their effect on the incidence of AEs. Figure 4a presents the direct comparisons and sample size distribution. According to the SUCRA rankings (Fig. 5a), the top three treatments for reducing the incidence of AEs were Codrituzumab 1600 mg Q2W (98.6%), the control group (90.0%), and Pembro 200 mg Q3W (75.6%), while Brivanib 800 mg OD ranked the lowest at 14.7%.
As shown in Table 5, compared to Brivanib 800 mg OD, Codrituzumab 1600 mg Q2W (OR = 0.02, 95% CI = 0.00, 0.18), the control group (OR = 0.10, 95% CI = 0.04, 0.24), Pembro 200 mg Q3W (OR = 0.17, 95% CI = 0.06, 0.46), Tivantinib 120 mg BID (OR = 0.24, 95% CI = 0.08, 0.77), and Apatinib 750 mg QD (OR = 0.26, 95% CI = 0.07, 0.92) greatly decreased the frequency of AEs.

≥3 Grade adverse events
An aggregate of 10 368 subjects from 26 trials were incorporated in the NMA for grade A AEs of ≥3 severity. The impact of different second-line therapies and dosages on the occurrence of ≥3 grade adverse events (≥3 AEs) was evaluated. Figure 4b presents the direct comparisons and sample size distribution. According to the SUCRA rankings (Fig. 5b), the top three treatments for reducing the incidence of ≥3 AEs were Codrituzumab 1600 mg Q2W (97.7%), the control group (92.9%), and Ram 8 mg/kg Q2W (70.9%), while Brivanib 800 mg OD ranked the lowest at 5.5%.
As shown in Table 6, compared to Brivanib 800 mg OD, Codrituzumab 1600 mg Q2W (OR = 0.10, 95% CI = 0.04, 0.27), the control group (OR = 0.15, 95% CI = 0.08, 0.26), Ram 8 mg/kg Q2W (OR = 0.26, 95% CI = 0.13, 0.50), Pembro 200 mg Q3W (OR = 0.26, 95% CI = 0.13, 0.53), Tivantinib 120 mg BID (OR = 0.31, 95% CI = 0.15, 0.67), Everolimus 7.5 mg QD (OR = 0.33, 95% CI = 0.15, 0.70), Apatinib 750 mg QD (OR = 0.33, 95% CI = 0.14, 0.79), and Regorafenib 160 mg QD (OR = 0.39, 95% CI = 0.20, 0.78) significantly reduced the incidence of ≥3 grade AEs.

Potential for bias and publication discrepancies
Of the 26 trials evaluated, 19 were determined to possess a minimal risk of bias, five got classified to possess certain issues, and two were seen to exhibit a substantial risk. In terms of randomization, a total of 23 trials exhibited a low risk of bias, while two raised certain issues, and one was classified as high risk. Regarding deviations from intended interventions, all 26 trials exhibited a minimal likelihood of bias. In the case of absent outcome data, 24 trials exhibited a minimal risk, whereas two tests were categorized as having a substantial risk. In terms of outcome measurement, 24 trials were evaluated as presenting a minimal risk, with two showing some concerns. For selective reporting, all 26 trials had a minimal risk (Supplementary File 3, Supplemental digital content 1, https://links.lww.com/EJGH/B214).
Funnel plots were utilized to assess the likelihood for publication bias (Supplementary File 4, Supplemental digital content 1, https://links.lww.com/EJGH/B214). The scatter plots around the vertical axis displayed varying degrees of symmetry, suggesting the potential for publication bias. Specifically, Fig. 3d showed a relatively even distribution of points, whereas Figs. 3a–d exhibited some asymmetry, indicating possible bias. Egger’s test indicated a P-value below 0.05 for PFS, requiring cautious interpretation of its results. For the other outcomes, Egger’s test showed P-values > 0.05, indicating a lack of substantial evidence for publication bias in the comprehensive analysis.

Discussion
This NMA synthesized data from multiple RCTs encompassing an aggregate of 10 368 persons with HCC to evaluate the safety and effectiveness of diverse second-line therapeutic agents and their respective dosages. Several key findings emerged from this investigation. First, within the framework of the network meta-analysis of OS and PFS, cabozantinib 60 mg QD demonstrated the most significant advantages in prolonging survival and improving PFS, ranking highest among the treatments evaluated. Notably, the mean OS gain of 3.36 months (95% CI = 2.01, 4.70) exceeds the minimal clinically important difference of approximately 2 months established for second-line therapies in advanced HCC, thereby confirming that this improvement is both statistically significant and clinically meaningful. Additionally, brivanib 800 mg OD exhibited the best efficacy in terms of ORR, while apatinib 750 mg QD ranked highest in terms of DCR. Codrituzumab 1600 mg Q2W was determined to remain the most efficient in diminishing AEs and the manifestation of grade 3–4 AEs. After considering efficacy and safety, cabozantinib 60 mg QD and the immune therapy pembrolizumab 200 mg Q3W are suggested as the best second-line therapies for advanced HCC, offering significant improvements in survival and disease control while effectively managing adverse reactions.
In advanced HCC, improving OS and PFS remains central to therapeutic objectives, as these outcomes directly correlate with better clinical management and enhanced quality of life for patients. Unlike curative treatments for early stage HCC, second-line therapies are crucial for prolonging survival and delaying disease progression once patients fail first-line treatment. This finding highlights cabozantinib 60 mg QD as the most effective treatment for improving both OS and PFS in advanced HCC sufferers, ranking highest among the various therapeutic agents evaluated. The results of this investigation are consistent with earlier investigations that possess similarly recognized cabozantinib as a potent component utilized in the therapy of advanced HCC. The CELESTIAL trial, for instance, illustrated that cabozantinib markedly enhanced OS in comparison to placebo, aligning with this observation [8]. However, while this analysis provides a broader perspective by including various second-line treatment options and dosages, previous studies primarily focused on comparing cabozantinib to a single comparator or placebo, thus reinforcing the value of network meta-analysis in synthesizing multiple treatment regimens. Cabozantinib functions as a multikinase inhibitor, engaging with various vital routes that contribute to tumor angiogenesis and proliferation, notably which include vascular endothelial growth factor receptor (VEGFR), mesenchymal-epithelial transition factor (MET), and AXL receptor tyrosine kinase (AXL) [46]. These pathways serve crucial roles in the advancement of HCC by facilitating tumor proliferation, angiogenesis, and defense against programmed cell death. By obstructing these channels of communication, cabozantinib effectively hampers the tumor’s ability to sustain growth and spread, thus improving both OS and PFS [47]. Additionally, cabozantinib has shown promising results in overcoming resistance to other therapies, rendering it an appropriate alternative for advanced HCC individuals that have not responded to initial therapies like sorafenib [48]. The broad activity of cabozantinib against multiple tumor-promoting mechanisms likely accounts for its superior performance in improving survival outcomes and disease control in advanced HCC, as observed in both our study and previous research.
The assessment of ORR and DCR is crucial in evaluating the effectiveness of secondary treatment options for advanced HCC. These metrics serve to illustrate not only the therapeutic effects on tumor size but also offer significant insights into the efficacy of treatments in halting disease advancement and sustaining tumor control. In the present study, brivanib 800 mg OD exhibited the highest efficacy in terms of ORR, while apatinib 750 mg QD ranked highest for DCR, offering important evidence regarding the effectiveness of these therapies in controlling advanced HCC. The finding that brivanib 800 mg OD achieves the best ORR and apatinib 750 mg QD the highest DCR is consistent with prior research showing that these drugs are effective in the therapy of HCC. For instance, the BRISK-FL research reported that brivanib significantly improved ORR contrasted against a placebo in individuals found with advanced HCC [49,50]. Similarly, apatinib has demonstrated considerable efficacy in improving DCR, with a number of articles confirming its ability to control tumor growth in HCC patients [51,52]. These results are consistent with earlier evidence, but this study provides a more comprehensive comparison across a broader range of second-line therapies, contributing a deeper comprehension of the comparative efficacy of brivanib and apatinib in improving both response rates and disease control.
The observed differences between ORR, DCR, OS, and PFS highlight the multifaceted nature of therapy efficacy. ORR and DCR primarily evaluate the immediate efficacy of treatments concerning tumor reduction and disease stabilization, whereas OS and PFS offer an expanded viewpoint on long-term survival and disease progression. A therapy that achieves a high ORR may not necessarily lead to prolonged survival if it does not adequately prevent recurrence or disease progression over time. Conversely, therapies with high DCR, such as apatinib 750 mg QD, which reflect the ability to control disease over a longer period, may offer superior outcomes regarding PFS and OS [53]. The distinction underscores the importance of a multidimensional approach in evaluating therapeutic effectiveness in advanced HCC, as short-term tumor response fails to consistently correspond with long-term survival benefits.
The mechanisms underlying the efficacy of brivanib and apatinib can be attributed to their distinct modes of action. Brivanib is a double inhibitor of VEGFR and FGFR, both of which are vital in tumor angiogenesis and proliferation in HCC [54]. By targeting these pathways, brivanib disrupts the tumor’s ability to receive the necessary blood supply, leading to tumor shrinkage and improved response rates. Apatinib, a VEGFR-2 inhibitor, targets angiogenesis by inhibiting VEGFR-2 signaling, consequently inhibiting the formation of new blood vessels essential for the proliferation of tumors [51]. The combination of these mechanisms likely explains why brivanib is particularly effective in improving ORR, while apatinib excels in enhancing DCR, with each drug providing distinct advantages in different dimensions of HCC management.
The safety profile of medication is a crucial aspect in the therapy of advanced HCC, as it directly impacts patient compliance with therapy and overall quality of life. Given the aggressive nature of HCC and the harsh side effects associated with many therapies, minimizing AEs and severe toxicities is essential in improving treatment outcomes. In this study, Codrituzumab 1600 mg Q2W demonstrated the best performance in reducing the incidence of AEs and grade 3–4 AEs relative to alternative treatment modalities, highlighting its potential as a safer alternative for advanced HCC patients. Codrituzumab 1600 mg Q2W’s ability to reduce AEs and severe AEs aligns with prior studies that have suggested its favorable safety profile in HCC treatment. Prior clinical trials have demonstrated that Codrituzumab, an anti-CD47 mAb, markedly diminishes the probability of adverse repercussions relative to traditional therapy. For example, A phase I/II trial of Codrituzumab in conjunction with sorafenib was described. Lower rates of severe toxicities and a more manageable safety profile than the combination of sorafenib with other immunotherapies [55,56]. This study corroborates those findings by showing that Codrituzumab results in fewer treatment-related AEs, particularly at higher doses, which is critical to preserve patient quality of life throughout therapy.
In managing advanced HCC, second-line therapy must consider both efficacy and safety, as patients are often compromised by tumor progression and prior treatments. Based on the comprehensive analysis of efficacy and safety, cabozantinib 60 mg QD and pembrolizumab 200 mg Q3W emerge as the preferred second-line options for advanced HCC. Cabozantinib 60 mg QD, a multikinase inhibitor targeting VEGFR, MET, and AXL, significantly improves OS and PFS. By disrupting tumor angiogenesis and inhibiting tumor cell migration, cabozantinib effectively controls disease progression with a manageable safety profile [57]. Similarly, pembrolizumab 200 mg Q3W, an immune checkpoint inhibitor targeting PD-1, enhances immune responses against HCC by reinvigorating exhausted T cells [58]. Its efficacy in improving ORR and DCR, coupled with its low incidence of severe AEs, makes it a suitable choice for patients who require a well-tolerated treatment. Both therapies target key aspects of HCC pathophysiology: cabozantinib’s inhibition of VEGFR disrupts tumor vasculature, while pembrolizumab’s PD-1 blockade restores antitumor immunity. These mechanisms, along with favorable clinical outcomes, support the recommendation of cabozantinib 60 mg QD and pembrolizumab 200 mg Q3W considered the optimal second-line therapeutic alternatives for advanced HCC.
This NMA presents numerous significant strengths. Initially, by synthesizing data from various RCTs, it delivers an extensive comparison of second-line therapies for advanced HCC, presenting significant insights into the effectiveness and safety of a diverse array of therapeutic agents. This large-scale synthesis of data strengthens the reliability and the applicability of the results. Second, the NMA approach allows for indirect comparisons between various treatment regimens, enabling an assessment of relative consequences of therapy in the absence of direct comparative studies, thus offering a broader perspective on optimal therapeutic strategies for HCC. Nonetheless, this analysis is subject to various constraints. First, the studies that have been included exhibited significant heterogeneity in patient populations, treatment regimens, the regimens variations in follow-up durations, which could potentially influence the consistency of findings observed across numerous investigations. Variations in baseline characteristics – such as prior therapies, underlying liver function (Child–Pugh class), and ECOG performance status – may introduce further variability in treatment response. Second, although subgroup and sensitivity analyses based on important clinical characteristics (e.g. Child–Pugh score, ECOG performance status) were planned, most included studies did not report these data in sufficient detail to support such analyses. Third, study quality varied, with some trials lacking comprehensive reporting of efficacy outcomes or safety profiles, potentially attenuating the strength of our conclusions. Fourth, while we have provided a robust comparison of efficacy and safety, the absence of a formal pharmacoeconomic evaluation represents a notable gap. Preliminary literature suggests substantial variation in cost-effectiveness – cabozantinib’s ICER has been estimated at $80 000–$120 000 per QALY and pembrolizumab’s at $100 000–$150 000 per QALY – underscoring the need for dedicated economic analyses in diverse healthcare settings. Finally, the inability to perform analyses stratified by geographic region or healthcare system further limits the generalizability of our findings. We acknowledge these constraints and strongly encourage future RCTs to (1) report detailed baseline liver function and performance status, (2) include cost-effectiveness endpoints, and (3) provide sufficient data for subgroup and sensitivity analyses to facilitate more nuanced network comparisons and real-world applicability.

Conclusion
This network meta-analysis identifies cabozantinib 60 mg QD and pembrolizumab 200 mg Q3W as optimal second-line treatment choices for advanced HCC, providing notable enhancements in overall longevity, PFS, and DC while maintaining manageable safety profiles. Importantly, the 3.36-month OS gain observed with cabozantinib exceeds the minimal clinically important difference of approximately 2 months for second-line HCC therapies, underscoring its clear clinical as well as statistical benefit. Cabozantinib’s multikinase inhibition effectively disrupts tumor angiogenesis and metastasis, while pembrolizumab’s PD-1 blockade restores immune-mediated antitumor responses. Brivanib 800 mg OD and apatinib 750 mg QD also demonstrate strong efficacy in improving ORR and DCR. Furthermore, Codrituzumab 1600 mg Q2W emerges as the most effective in minimizing AEs. These findings provide valuable guidance for selecting second-line therapies in advanced HCC, supporting the integration of targeted and immunotherapies to optimize patient outcomes. Further studies are warranted to evaluate the cost-effectiveness of these regimens and to confirm the durability of their clinical benefits in real-world settings.

Acknowledgements
The authors have no acknowledgments to report.
This is a systematic review and meta-analysis, ethics approval and consent to participate are not applicable.
The manuscript does not include the participant’s identification image or other personal or clinical details.
All data generated or analyzed during this study are included in this published article and its supplementary information files.

Conflicts of interest
There are no conflicts of interest.

Supplementary Material