ABSTRACT
Quantitative analytical methods such as high-performance liquid chromatography use a calibration relationship between the measured response and analyte concentration to quantify the amount of analyte in a test sample. The relationship is normally established during method development through fitting a regression model to test results of reference standards of known amounts of analyte. Depending on the true nature of the calibration relationship, fewer or more reference standard concentrations may need to be tested in each run along with the test sample, in routine use. For example, if the relationship is linear, potentially two reference standards at different concentrations would suffice. However, if the true relationship is nonlinear, more reference standards are needed to capture the curvature of the doseresponse curve. Therefore, a linear calibration relationship is operationally desirable. When the calibration relationship is nonlinear, approximating it with a linear model will cause bias in the measured analyte concentration, thus affecting the performance of the analytical method. Suffice it to say, it is important to evaluate assay linearity. In fact, validation of analytical method linearity is a regulatory requirement (CLSI 2003; ICH Q2(R1) 2005; USP <1225> 1989). ICH Q2(R1) clearly states, “A linear relationship should be evaluated across the range of the analytical procedure.”
