Trends of Clinical Outcomes of Patients with Advanced Hepatocellular Carcinoma Treated with First-Line Sorafenib in Randomized Controlled Trials

Abstract

Background: Sorafenib has consistently served as the control arm in multiple randomized clinical trials (RCTs) evaluating novel therapies for advanced hepatocellular carcinoma (HCC) for more than a decade. Analyzing trends in clinical outcomes of patients treated with sorafenib for the same indication over time offers the opportunity for unique insight into the evolution of clinical trial conduct and potential non-drug factors impacting outcomes.

Methods: We identified RCTs in patients with treatment-naïve advanced HCC where sorafenib was compared to another systemic therapy or placebo. We extracted trial-level demographic, clinicopathologic, and outcome data (overall survival [OS], progression-free survival [PFS], objective response rate [ORR], and duration of therapy). Sample-weighted linear regression was used to identify temporal trends with significance set at p ≤ 0.05.

Results: Sixteen RCTs (9 phase III and 7 phase II) enrolling 4,086 patients treated with sorafenib were included in the analysis. Included trials enrolled patients from 2005 to 2019. OS has significantly improved by 4.5 months from 2005 to 2019 (p = 0.048) over time. Thirteen studies provided data on PFS using Response Evaluation Criteria in Solid Tumors (RECIST) 1.1, with no significant change over time (p = 0.69). ORR assessed by RECIST 1.1 has significantly improved by 6.0% over time (p = 0.003). Median duration of therapy with sorafenib has decreased by 53% since the enrollment of the first clinical trial in 2005, from 23.1 weeks to 12.2 weeks (p = 0.0037). There was no significant change in patient demographics were identified over time to explain the OS findings.

Conclusion: The median OS of patients with advanced HCC treated with sorafenib has improved significantly over 15 years. At the same time, the median duration of therapy with sorafenib has decreased. The reason for these findings was not explained by changing demographics of patients enrolled in these trials and has implications for ongoing clinical trials.

Keywords: Cancer outcomes; Clinical trials; Hepatocellular carcinoma; Sorafenib.

Fig. 1

Median OS and year of enrollment. Scatterplot showing reported median OS in each included trial in months and the first date of enrollment. Since the publication of the first trial to demonstrate an improvement in OS with treatment with sorafenib (SHARP), a significant positive trend in improving OS has emerged in subsequent publications. OS has been improving by 0.34 months per year since the initial study (p = 0.048, R² = 0.38). Markers are weighted by sample size and trendline represents the weighted linear regression fit.

Fig. 2

Median PFS and year of enrollment. Scatterplot showing reported median PFS by RECIST 1.1 and the first date of enrollment. No significant linear trend was noted between PFS by RECIST 1.1 and date of first enrollment (p = 0.69, R² = 0.021). Markers are weighted by sample size and trendline represents the weighted linear regression line of best fit.

Fig. 3

ORRs (by RECIST 1.1) and year of enrollment. Scatterplot showing reported ORRs by RECIST 1.1 and the first date of trial enrollment. ORRs have been significantly increasing over time, at a rate of 0.46% per year (p = 0.003, R² = 0.50). This trend was driven by results seen on phase III trials (p = 0.027, R² = 0.654), no significant linear trend was identified on a subset analysis of phase II trials only. Markers are weighted by sample size and trendline represents the weighted linear regression fit.

Fig. 4

Median duration of therapy and year of enrollment. Scatterplot showing the median reported durations of therapy with sorafenib and the first date of enrollment. Thirteen studies including 3,021 patients reported duration of therapy. Since the initial study, the median duration of therapy has decreased by 53% (p = 0.0037, R² = 0.669). Markers are weighted by sample size and trendline represents the weighted linear regression fit.