Optimizing Drug Response Study Design in Patient-Derived Tumor Xenografts.

Area of Science:

• Preclinical oncology research
• Experimental design and statistics

Background:

• Patient-derived tumor xenograft (PDX) models are crucial for evaluating anticancer agents.
• The 1 mouse per patient per drug (1×1×1) design is practical for large-scale efficacy studies but requires optimization.
• Statistical power is essential for reliable preclinical study outcomes.

Purpose of the Study:

• To evaluate modifiable parameters for increasing the statistical power of the 1×1×1 PDX design.
• To investigate the relationship between statistical power, effect size, inter-mouse variation, and tumor measurement frequency.
• To determine the minimum number of mice per group for achieving 80% power in PDX efficacy studies.

Main Methods:

• Analysis of consolidated PDX experiments and real-world study data.
• Investigation of statistical power under varying treatment effect sizes, inter-mouse variation, and measurement frequencies.
• Calculation of minimum mice per group required for 80% power at an alpha level of 0.05.

Main Results:

• The 1×1×1 design can detect large effect sizes at significance levels of 0.2 or 0.05.
• Minimum mice required for 80% power (alpha=0.05) were 21 per group for small effect sizes.
• Minimum mice required for 80% power (alpha=0.05) were 5 per group for medium effect sizes.

Conclusions:

• The 1×1×1 PDX design is feasible for preclinical drug efficacy studies.
• Optimizing parameters like measurement frequency and accounting for variation can enhance statistical power.
• This research provides evidence-based recommendations for sample size determination in PDX studies.