Friday, 1 March 2013

Help on: Validation Shows that Method is Not Linear Through Zero

Do you have any problems relating to analytical chemistry for pharmaceuticals or training? Send your questions to the MTS helpdesk using our contact form.

"I am currently validating an assay method as per ICH guidance. I have a problem with the linearity determination. The method has a single concentration of calibration standard and therefore I believe that I have to show that the line goes through the origin but when I run the linearity standards at the range in ICH, i.e. 80% to 120% of the assay concentration, the intercept of the resulting line is not close to zero. Can you suggest what I should do?" 
"If an analytical method involves the use of calibration standards which are prepared at a single concentration (more than one standard solution is commonly prepared) then the actual calibration line generated for the analysis is drawn between the calibration standard concentration response and zero. This method of calibration will only work if the line does go through zero and therefore an important part of method validation is to confirm that this is the case.

A typical assay method will be designed such that the typical response of the analyte is at a concentration level where the response is not subject to random errors, such as injection repeatability and integration in the case of an HPLC method, and therefore is significantly greater than zero.

When validating an assay method the range to be considered is usually ± 20% of the expected sample concentration, expressed as 80 to 120% in the ICH validation guidance Q2(R1). This means that it seems sensible to investigate linearity using standards prepared over this range. However, when running the method in routine analysis the linear range of the method is from 0 to 120%. To assess whether the calibration line goes through the origin involves an extrapolation of the line from 80% down to zero. With this approach it is quite likely that the line derived from the linearity study does not go through zero but the method actually is linear through zero. 
I recommend that an extra linearity standard is included in the linearity study at a concentration equivalent to 40%. This represents the actual linear range of the method and reduces the extent of the extrapolation required.”

No comments:

Post a Comment