Method Transfer Problems

[LabRat]

Hi,

I think we spend too much time redeveloping and revalidating methods that fail transfer or have problems in Stability or QC. Each time the problem seems different, but endless problems. Any thoughts or suggestions?

I read the post regarding stability indicating purity and assay with the same method. Has anyone done this? How do you calibrate? Will I have even more transfer problems? Thanks for any help.

[tom_mizukami]

Hi Labrat,

I think all of your questions are related to how you design your methods as opposed to how you validate or transfer them. I think all HPLC methods should be DESIGNED to run on any instrument in a company that can pass its instrument qualification.

Now that all of the HPLC vendors that sell into pharmaceutical companies also sell qualification services this is not that hard to do. Vendors used to spec their instruments in way that had no relationship to how they are used, like with dry flow cells, etc. Now you can parse the instrument OQ/PQ specs for values that make more sense.

In your example of a pharmaceutical lab wanting to do assay and impurity with the same stability indicating method let say your QC were using both Agilent VWD and DAD detectors. The VWD has an OQ ASTM nosie spec of 0.04 mAU and the DAD 0.05 mAU. I would start with the higher noise spec and say you need LOQ at 10 x noise so this would be 0.5 mAU and say this needs to correspond to 0.05% Label Claim. This puts 100% LC at 1AU and you will probably need to validate out to 120% or 1.2 AU.

This type of explicit design will ensure that your methods are conservative and have the minimum number of problems on transfer. Of course you will need to integrate your instrument and method validation and instrument selection processes.

So if you want to validate an impurity to assay level method you will be using an appreciable portion of the linear range of a UV detector so you will probably need to pay attention to the background absorbance of your mobile phase. Most people would probably calibrate with relative response factors based on the active at a single wavelength if possible. Since your method will be spannning more than three orders of magnitude you will need to use a weighted linear regression or force the calibration curve through zero or just use a single point. Plot the specific response versus level to see what is going on.

There have been other good threads that have covered this ground you might want to try to search the archives. If none of this is what you were after try to post single specific problems and I'm sure one of the experts will try to help you. Good luck.

[Consumer Products Guy]

I basically agree with Tom. We've never had any issues with HPLC test method transfers to either our QC or to contract manufacturers. We do the three columns of different lot numbers, try different instruments, different operators as well, in the validations. We haven't had to travel either.