Analysis of histamine release assays using the Bootstrap.

Area of Science:

• Pharmacology
• Biostatistics
• Experimental Biology

Background:

• Bioassay data is often presented as quotients for comparison.
• Evaluating transformed data statistically, rather than raw data, is crucial for accurate interpretation.
• Compound 48/80's histamine release from mast cells serves as a model system.

Purpose of the Study:

• To determine a valid statistical method for evaluating transformed bioassay data.
• To assess the impact of treating control groups as constants versus variables with experimental error.
• To explore the utility of resampling methods for analyzing small sample size bioassay data.

Main Methods:

• Dose-response experiments using compound 48/80 and mast cells (n=24 replicates).
• Descriptive statistics, Ryan-Joiner test, and normal probability plots for data normality assessment.
• Parametric analysis and resampling statistics (bootstrap, Monte Carlo) for standard error and confidence interval estimation.

Main Results:

• Parametric analysis showed bias when control groups were treated as errorless constants.
• Resampling methods, including bootstrap, confirmed data normality and provided reliable confidence intervals.
• Effective analysis of transformed data (ratio estimates) was achieved with small sample sizes (n=4-6) using resampling.

Conclusions:

• Resampling techniques offer a statistically robust approach to analyzing transformed bioassay data.
• Ignoring control group variation introduces bias, which resampling methods mitigate.
• Small sample sizes (n=4-6) are sufficient for reliable statistical analysis using resampling methods in bioassays.