Processing method

[chemist91]

Hello to everyone,

When I began using processing methods (software - millenium, empower) I was told that the way you process your standards it is the same way you have to process your samples. In other words your standards and samples need to be process using the same processing method- no excuse.

I have a friend that is telling me that he uses different processing methods for his standard and his samples, because they are hard to integrate since they are impurities. And because they put a value in the calculation of the samples (RF average) that came from the standard calculation should be ok.
For me this is not right because in the sample he is using a different integration parameters than the standards and if he uses the processing method of the sample on the standard his results will be different.

Can somebody told me if it is correct (GMP) to do what he is doing? Maybe something has change in the past 13 yrs and I didn't heard about it. Somebody knows if FDA will accept something like that?. For me that is data manipulation.

Thanks,

[tom jupille]

That's a thorny area.

First of all, GMP is much more about consistency than correctness. If this is documented in a validated method, and the method is followed, then it is OK from a GMP perspective.

Second, so long as the processing method is specified in advance, then it's not manipulation (in my mind, at least).

Third, if the peaks are not fully resolved chromatographically, then no amount of integration processing will ensure accurate results. The best you can hope for is to be wrong the same way every time.

All that said, any difference in handling, integration, or processing between calibrators and samples adds potential errors. If I were a reviewer looking at a validation package, I would want to see extensive justification for the difference.

[chemist91]

Thanks for the answer,

And you right about GMP, as far you can justify why you are doing things and you are consistant you are OK.

So far my understanding is that they are validating the methods (so it is not validated yet). However they are running some impurities hard to integrate but not imposibble, but it takes them less time and effort to do it that way (I guess).
It is not a "specific processing method" that they use..it is my understanding that everyone process the way they want.

My point is (as you mention) if QA or any reviewer is taking a look to that package they will see different processing methods for the standards and the samples.

I believe they don't have any written explanation for that. He just told me that since they are impurities it has to be some flexibility because they are hard to integrate. However that it is not a justification for me.

[Jackus]

We follow rule that calibration and sample must be calculated with the same integration parameters. Quite stupid, because sometimes is not possible to correctly integrate peak of standard in one-peak chromatogram and peak in complex mixture with one integration parameters.

[chemist91]

That has been the rule that I had been following all my life in several diferent companies.

However, I don't understand how you can do other way if you are using standards to quantify your samples. In other words the results of our samples depend on the area of our standards.

I agree sometimes it is hard but never impossible to do a good integration, however I think it is not such thing as a perfect integration especially when you have impurities.

Thanks for the answer

[Dan]

I agree with Tom that the cGMPs are more about consistency. I also agree in that I don't see what your colleagues are doing as 'data manipulation'.

That part about consistency is a big part of why people require the samples and standards to be processed the same. There can be exceptions. A simple one is that the standard chromatogram has a shorter run time than the sample chromatogram. Another is difficulty in integrating smaller peaks.

Older CDS software may have problems in getting integration parameters that work for both large and small peaks. The newer software (like Empower) shouldn't have that problem.

I would ask three questions to your colleagues:
1) Are they sure that the integration parameters they use in the sample chromatogram cannot be used in the standard chromatogram? It would seem that the integration parameters should work in both as they may not have any influence on the peak of the standard(s).
2) Is the difficulty in integration a software problem or an indication that the method is not as good as it can (or should) be? Maybe they need better sensitivity (or higher concentration, etc.) to get larger peaks.
3) Can they transfer this method to another site or lab that may use a different CDS?

The answer to number 3 can be the deal breaker.

Regards,
Dan

[Jamie]

I do not work in a GMP regulated Firm. Yet I do QC and my chromatograms were processed with Turbochrome software. Sometimes I do have to manipulate the integration parameters in order to get accurate results.

Some of the things that happen which needed different integration is how the baseline was drawn in the sample, where the peak starts and ends integration. If the sample peak comes at the tail of a huge peak, then you have to integrate that differently.

Jamie

[chemist91]

As far as I know they have new equipment and the latest one. They are also using Empower2. I do use on my lab Empower2 too and I never have any trouble integrating any peaks, however it does sometimes take time to find the best "processing method".

In this case of my friend what I don't like is that they use their standars to quantify their samples. So my point is if they process their std's with one method and the samples with other. Who assures that the %assay will be correct. I can proccess my std's a way that my areas are lower or higher and then make pass my samples. That is what I was meaning with data manipulation.

Thanks for the answer. I still trying to change his mind but he said that is the way they are doing it because they don't know another way. However I will say that all the chist in that lab have less than 3 yrs of experience and maybe they are not familiar with the software.

Regards,

Maria

[HW Mueller]

Seems to me that some people should ask themselves whether they want a number that pleases regulators or whether they want results which are close to the correct values.
Obviously, HPLC requires calibration. If you change the "rules" you have to recalibrate.

[DR]

Except for the part where Dan picks on older CDS software, I agree with him (a good CDS will not fix bad chromatography).

If you can do it consistently, if it is defendable, if it errs on the side of caution (not putting a regulator type in the position of thinking "I would have made that RS peak area much larger.") then you can probably get by with it. In the long run, you have to realize 2 things. 1) It's a PITA to have to explain manual integrations to reviewers and 2) It's not the best use of time (to have to spend a lot of time manually integrating, then more explaining it). If you routinely use manual integrations, you should be asking yourself if it's really necessary (by what % of the original answer did that manual integration change that result?). If it is, you may want to revisit your means of method development. Better methods will result in less need for manual integration.

[Rande]

Below is a reply I made to a similar discussion under the "Data Systems" Forum.

See:
http://www.sepsci.com/chromforum/viewtopic.php?t=5625

Rande

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

I fully agree that you should have clearly defined and documented integration methods for your standards (especially for system suitability).

If your method is not reproducable enough to integrate your standards, how can you trust your results?

You should have "suggested" integration methods for each product. Often there will be differences; lot to lot variation or samples from accelerated stability studies, that require modified integration methods. My procedure is to clearly document the revised integration method, then use that method for all replicates of the sample. Sometimes there will be a noisy baseline for one replicate injection that requires "manual intervention" with the integration. A brief memo is included with the assay results to show the revised integration method.

Just a tip - I like to create a screen view on the computer with a "Blow Up" of the integrated chromatogram and the integration table. I then COPY the image on the computer screen by using the "print screen" key and PASTE it into a word document (along with appropriate discrriptions and assay information). This allows me to make a clearly understandable one page memo that I then sign and include with the assay data. I make no claim that this is the official (part 11 compliant) records. The complete assay run is stored by the HPLC software - this is just a brief memo to show the integrated peaks and "integration event table".

Make sure you have good scientific justification for everything and it does not look like you are trying to hide something.