Deduct blank baseline from chromatography result?

[zlb215]

Hi;
I joined a new analysis team, and they deduct blank baseline from sample picture, then integrate and print it as final result.
Could this comply with CFR or other FDA regulations?If not, why do the software have such function?

[krickos]

Hi zlb215

Well personally I am quite reluctant to that as a routine approach especially if both raw data and "modified" raw data is not easily availble and gone throught during review (basic CFR/EU GMP requirement).

To start from the top so to speak. ICH Q2 R1 (analytical validation), essentially points to the pharmacopieas when it comes to system suitability test, so essentially for all releated substances methods signal to noise testing is mandatory (USP/Ph Eur) regardless if it is a generic or new product/API. Can not see that you can use modified data to comply with signal to noise testing to start with in my opinion. But of course SSTs can be done the normal way.

On the other hand, if validation/development has failed to sufficiently remove interfering peaks/baseline disturbances that are repeateble in blank runs throughout the analytical sequence, I can see the the use of using an "automated" function to remove those peaks before integration rather than manually recalculate peaks for each chromatogram.
However this would require some very clear routines around working with this and injections of blanks throughout the sequence. Also (as I do assists in audits sometimes), I would expect the unmodified raw data to be easily accesible during an audit to be able to review/verify that the SOPs/methods are followed accordingly as this is fundementaly different from disregarding peaks under the report/disregard limit.

Looking at the regulatory requirements: Regardless if you look at EU (Eudralex Volume 4, part II APIs §6.60) or CFR§211.194, a complete record of data generated should be included, including calculations/conversion factors.
I would say the situation you describe falls within this and you have to be able to show how you have done it for each and every batch.

Intresting question, I hope more provide feedback

[krickos]

Hi again

Audit trail: Well I left out ER/ES systems intentionally, but as you mention it, well audit trial is one thing, the GMP requirement that a second person should review and sign (manually or by ES) is another. Regardless of audit trial or not a reviewer have to review and sign any conversion/transformation/calculation of raw data, how is up to you.

Regarding "clear routines": Yes in the analytical method or appropiate SOP/routine whichever suits your organisation best, should include a clear directives how to use this option, not only software wise but also number of blank runs, where to put them in sequence etc to ensure a proper use of this function in the software.

"Actually, We print the deducted report as final result, I don't think there will be a complete data of deducting procedure.Will this be a mistake during FDA inspection? "

I am afraid that it could be a potential issue with FDA if you can not trace back the treatment of the original raw data. I suggest you sit down with your local QA/QC people and discuss this, traceability may exist through the software if it is at least ER compliant (CFR part 11). The tricky part is to make it availble to an auditor.

Signal to noise: First it seems that I have to back up a little. As per 2nd suppplement 2006 it seems the general requirement is removed from <621>, USP32 does not contain it.

As for gradients, well Ph Eur just narrowed down the measure area in the blank to 5 times the width at half height, internally we have adjusted the measurement area (smaller) with "troublesome" methods or where we have early eluters with small impurities in blank close by.