Variance weighting functions in radioimmunoassay calibration

Area of Science:

• Biochemistry
• Analytical Chemistry
• Immunology

Background:

• Radioimmunoassay (RIA) calibration typically assumes counting rate is solely a function of ligand dose.
• Previous research suggests counting rate variance also correlates with dose, but individual assays lack precision for variance estimation.
• Accurate RIA calibration requires accounting for dose-dependent counting rate variances.

Purpose of the Study:

• To develop and validate a method for characterizing the mean-variance relationship of counting rates across multiple radioimmunoassays.
• To implement a variance weighting function for improved RIA calibration curve fitting.
• To enhance the precision and reliability of RIA results by optimizing data utilization.

Main Methods:

• Accumulation of data from multiple radioimmunoassays (cortisol, testosterone, growth hormone, triiodothyronine).
• Statistical analysis to characterize the relationship between mean and variance of counting rates.
• Development of empirical weights based on estimated counting variances.
• Fitting calibration curves using weighted least squares regression.

Main Results:

• An asymmetric rising ogive characterized the variance-mean relationship in cortisol assays.
• A rectangular hyperbola adequately described this relationship in testosterone assays.
• Straight lines and rising exponential curves characterized the relationships in growth hormone and triiodothyronine assays, respectively.
• Weighted least squares fitting improved calibration by utilizing all observations proportionally to their estimated variance.

Conclusions:

• A robust method for estimating counting rate variance in RIA has been established by pooling data from multiple assays.
• The proposed variance weighting function enhances RIA calibration accuracy and data interpretation.
• This approach ensures all assay observations contribute optimally to the final calibration curve, improving overall assay performance.