Regarding validation of an HPLC method for a pharmaceutical product, if the "recovery" portion will use solutions containing analyte + sample matrix (i.e. excipients) in diluent, is it acceptable to perform the linearity portion using solutions which omit the sample matrix (i.e. contain only the main analyte and sample diluent)? If so, how do you verify linearity in the presence of the sample matrix without essentially repeating the linearity studies-do you assume linearity if recoveries are acceptable?

Method validation is taken as a whole. Just as precision of standard injections implies that samples (with excipients) will be constrained by the same instrument precision. You would measure a method precision to deconvolute the sample preparation precision (glassware, weights, volumes, recovery, etc..) from the instrument precision.

Now to linearity. Linearity is a function of the detector in the RANGE that your analyte will be measured (e.g. 50% to 150% target levels). Accuracy then deconvolutes the contribution or interference of excipients in the chromatography from the drug only samples. Accuracy must of course be coupled with method precision which ensures that your recovery of analytes is acceptable (typically above 98%).

So, you can merge these studies for the sake of time by spiking placebo, comparing to drug only solutions (used for linearity and instrument precision.

Thanks for your response.

I am still trying to grasp this, though: Is the linearity supposed to demonstrate relationship between analyte only and detector response as an independent parameter, or is it supposed to demonstrate the relationship of the analyte to the detector response in the presence of matrix components as expected in the sample?

I don't believe the response of an analyte dissolved to analytical level in system diluent or mobile phase, without sample matrix, can be assumed to represent the linearity of the compound when analyzed in the presence of sample matrix. So then, why bother demonstrating linearity without matrix components if there is a possibility of it differing from that observed WITH the matrix components?

What continues to gives me trouble is what is found in the ICH guideline ("Text on Validation of Analytical Procedures") which defines linearity in this way: "The linearity of an analytical procedure is its ability (within a given range) to obtain test results which are directly proportional to the concentration (amount) of analyte in the sample".

... it's that "in the sample" that has me stuck-