Design-Outcome Concordance in Targeted-Therapy trials in Non-Small cell lung cancer: a Meta-Research study

Area of Science:

• Oncology
• Clinical Trials
• Biostatistics

Background:

• Biomarker-selected non-small cell lung cancer (NSCLC) benefits significantly from targeted therapies.
• Phase III randomized controlled trials (RCTs) are crucial for evaluating these treatments.

Purpose of the Study:

• To assess the agreement between design assumptions and observed treatment effects in oncogene-driven NSCLC.
• To evaluate the performance of targeted agents versus standard regimens in superiority trials.

Main Methods:

• Systematic review of phase III RCTs (2005-2025) in biomarker-selected NSCLC.
• Analysis of progression-free survival using hazard ratios (HRs).
• Included meta-analyses, retrodesign analysis, and restricted mean survival time (RMST).

Main Results:

• 45 studies analyzed; observed HR (0.47) was lower than expected (0.64).
• Median power increased from 0.78 to 1.00 when using observed HRs.
• 84% of RCTs reported statistically significant results; 19% violated proportional hazards assumption.

Conclusions:

• Observed treatment effects in NSCLC trials often surpass design expectations.
• Recommends reporting retrodesign metrics, minimal clinically important difference, and RMST.
• Highlights the need for robust statistical reporting in targeted therapy trials.