Evidence-Based Management of Hepatocellular Carcinoma: Systematic Review and Meta-analysis of Randomized Controlled Trials (2002-2020)

Abstract

Background & aims: Hepatocellular carcinoma (HCC) is a leading cause of cancer-related mortality, with a rapidly changing landscape of treatments. In the past 20 years, numerous randomized controlled trials (RCTs) have aimed at improving outcomes across disease stages. We aimed to analyze the current evidence and identify potential factors influencing response to therapies.

Methods: We conducted a systematic review of phase III RCTs (2002-2020) across disease stages. A meta-analysis was designed to examine the relationship between etiology and outcome after systemic therapies with either tyrosine-kinase inhibitor (TKI)/antiangiogenic or immune checkpoint inhibitor (ICI) therapy.

Results: Out of 10,100 studies identified, 76 were phase III RCTs. Among them, a rigorous screening algorithm identified 49 with high quality including a total of 22,113 patients undergoing adjuvant (n = 7) and primary treatment for early (n = 2), intermediate (n = 7), and advanced (first-line, n = 21; second-line, n = 12) stages of disease. Nine of these trials were positive, 6 treatments have been adopted in guidelines (sorafenib [2 RCTs], lenvatinib, atezolizumab+bevacizumab, regorafenib, cabozantinib and ramucirumab), but 2 were not (adjuvant CIK cells and sorafenib plus hepatic arterial infusion with FOLFOX). Meta-analysis of 8 trials including 3739 patients revealed ICI therapy to be significantly more effective in patients with viral hepatitis compared with nonviral-related HCC, whereas no differences related to etiology were observed in patients treated with TKI/anti-vascular endothelial growth factor.

Conclusions: Among 49 high-quality RCTs conducted in HCC during 2002-2020, 9 resulted in positive results. A meta-analysis of systemic therapies suggests that immunotherapies may be more effective in viral etiologies.

Keywords: Hepatocellular carcinoma; Liver cancer; Randomized controlled trials; Systematic review.

Figure 1.

Study flowchart. Description of reasons for including/excluding RCTs from the current systematic review. Forty-nine phase III randomized studies were finally included and 3 meta-analysis conducted: systemic therapies based on etiology, Sorafenib + TACE vs sorafenib (Supplementary Figure 1), and TARE-Y90 vs sorafenib (Supplementary Figure 2). CNI, calcineurin inhibitor; HCC, hepatocellular carcinoma; ICI, immune checkpoint inhibitor; LT, liver transplantation; OS, overall survival; RCT, randomized controlled trial; RFA, radiofrequency ablation; TACE, transarterial chemoembolization; TAE, transarterial embolization; TARE, transarterial radioembolization; TKI, tyrosine-kinase inhibitor; VEGF, vascular endothelial growth factor; Y90, yttrium-90.

Figure 2.

Hazard ratios (HRs) for overall survival of 30 high-quality trials in advanced hepatocellular carcinoma. First-line trials are separated into those applying placebo or sorafenib as a control arm. Positive trials are shown in green, noninferior in orange, and negative trials in red. The orange shaded area defines the upper boundary for positive results with a noninferiority design (HR from 1 to 1.08). The HR for overall survival is represented by a circle (for trials that excluded patients with main portal vein invasion [Vp4]) or a triangle (Vp4 accepted). The size of the shape corresponds to the number of randomized patients. *Trials enriched for high Met expression. **α-Fetoprotein ≥400 ng/mL was required for enrollment.

Figure 3.

Meta-analysis of drug class efficacy based on underlying etiology. Trials are divided into immune checkpoint inhibitor (ICI) and tyrosine-kinase inhibitor (TKI)/anti–vascular endothelial growth factor (VEGF) trials, and hazard ratios (HRs) and 95% confidence intervals (CI) per etiology are colored in red (viral) and green (nonviral). (A) The pooled HR of death in each study was calculated by using a random-effects model, and the variance of the distribution of true effect sizes was estimated by means of the DerSimonian-Laird method. The dashed vertical lines show the pooled estimate of HRs for each etiology. The P of heterogeneity defines significant differences in overall survival effect according to etiology for ICI but not for TKI/monoclonal antibody. All trials listed in the ICI meta-analysis are first-line trials except for KEYNOTE-240. (B) Sensitivity analyses were performed for the ICI and TKI/anti-VEGF trials. For each comparison, the meta-analysis was repeated by applying a fixed-effects model, adjustments of τ² estimation (Hartung-Knapp), or alternative methods for estimating τ² (maximum likelihood and empirical Bayes). Diamonds represent the pooled estimate for viral and nonviral etiologies, and lines represent the 95% CIs.