Measuring parallelism, linearity, and relative potency in bioassay and immunoassay data

Area of Science:

• Biostatistics
• Pharmacology
• Statistical Modeling

Background:

• Determining parallelism between dose-response data sets is crucial for biological applications.
• The standard extra-sum-of-squares method with an F-test statistic is commonly used but has limitations.

Purpose of the Study:

• To develop a more reliable metric for assessing parallelism between dose-response curves.
• To address the shortcomings of the conventional F-test-based parallelism metric.

Main Methods:

• Modification of the extra-sum-of-squares principle.
• Development of a chi-square test metric applied directly to the extra-sum-of-squares statistic.
• Investigation of curve model choice, noise, and weighting effects.

Main Results:

• The proposed chi-square metric directly and reliably corresponds to parallelism.
• The conventional F-test metric can vary in opposition to actual parallelism.
• Asymmetric models, like the asymmetric five-parameter logistic function, are necessary for asymmetric data.

Conclusions:

• The chi-square test-based metric offers a more appropriate and reliable measure of parallelism.
• Model selection and data quality (noise, weighting) significantly impact parallelism assessment and relative potency.