Robust Modestly Weighted Log-Rank Tests

Area of Science:

• Biostatistics
• Clinical Trial Methodology
• Immunotherapy Research

Background:

• The log-rank test is standard for immuno-oncology trials but struggles with non-proportional hazards.
• Checkpoint inhibitors can lead to delayed or non-proportional survival benefits, challenging traditional analysis.
• Existing weighted log-rank tests have limitations, including potential counterintuitive results or reduced power in specific scenarios.

Purpose of the Study:

• To develop a novel statistical test for immuno-oncology confirmatory trials.
• To enhance the power and robustness of survival analysis, especially with non-proportional hazards.
• To provide a reliable primary analysis method for checkpoint inhibitor studies.

Main Methods:

• Proposed a new test statistic integrating standard log-rank, modestly weighted log-rank, and MaxCombo test principles.
• The novel statistic maximizes the standard log-rank and a modestly weighted log-rank statistic.
• Evaluated the test's performance using simulation studies and a real-world case study.

Main Results:

• The proposed test demonstrates efficiency and robustness in simulation and case studies.
• It maintains statistical power in scenarios with delayed separation of survival curves.
• It minimizes power loss compared to the standard log-rank test in worst-case scenarios.

Conclusions:

• The novel integrated test statistic offers a robust alternative for primary analysis in immuno-oncology.
• It effectively addresses the challenges posed by non-proportional hazards and delayed treatment effects.
• This approach has the potential to improve the reliability of clinical trial results in immuno-oncology.