Work

Assessing surrogacy using restricted mean survival time ratio for overall survival in liver cancer: a narrative review

Year: 2023

Type: review

Abstract: BACKGROUND AND OBJECTIVE: The application of immunotherapy in cancers, including liver cancer, has been increasing. However, non-proportional hazard (NPH) is often observed in cancer immunotherapy trials. In presence of violation of proportional hazard (PH) assumption, restricted mean survival time (RMST) ratio was proposed as an alternative to hazard ratio (HR) for evaluating the treatment effects of such trials. To shorten the total study duration, an intermediate endpoint with shorter follow-up such as progression-free survival (PFS) is used as the primary endpoint. Our aim is to evaluate the applicability of RMST ratio in addition to the HR in assessing the level of PFS serving as a surrogacy of overall survival (OS). METHODS: Phase II or phase III hepatocellular carcinoma (HCC) immunotherapy studies that were published between January 2013 and August 2022 were identified via the search in PubMed. Weighted least-square regression (WLSR) was applied to analyze the trial level data with the sample size of study being set as the weight. The evaluation was conducted twice with RMST ratio and HR being applied in respective evaluation to examine the level of PFS as a surrogacy for OS. KEY CONTENT AND FINDINGS: Based on the results of eight included trials, the R-square values of WLSR with either HR or RMST ratio being applied were 0.31 and 0.16 separately, indicating a moderate and low correlation between PFS and OS respectively. CONCLUSIONS: In this study, our results demonstrated the potential of RMST ratio in addition to HR for evaluating the level of surrogacy in immunotherapy trials. Furthermore, including more large scale and homogeneous studies into the research may help better understand the level of surrogacy in liver cancer.

Source: Chinese Clinical Oncology

DOI: 10.21037/cco-23-48

Authors: Tiffany H. Leung, James C. Ho, Xiaofei Wang, Herbert Pang

Institutions: University of Hong Kong, Duke University

Language: English

Topic: Statistical Methods in Clinical Trials

Subfield: Statistics and Probability

Field: Mathematics

Domain: Physical Sciences

SDG: Good health and well-being

Open Access status: diamond

Funders: American Society of Clinical Oncology, Health and Medical Research Fund