Integration parameters- locked after validation?

[scpercy]

All-
A quality rep recently opened a corrective action on one of my labs regarding ALL of our HPLC methods- to verify that the integration paramaters currently used are the same as those used in validation. This came as a shock to me, since we purposely do NOT control integration parameters, since so many things can change and need optimization over the life of a method. We specify a retention time range, tailing, etc. in our system suitability, and then simply state that the entire run must be processed with the same integration parameters (no tweaks for individual chromatograms.) The processing parameters are printed with every analysis.
Am I crazy, does everyone else lock down integration parameters? Is this necessary for compliance to FDA/GMP etc?
Thanks-

[PJ8]

I hope not! Otherwise how would we cope with the small variations in retention times, or changes in blank peaks/baseline shapes seen throughout the lifetime of a method as it is used on different systems and columns.

Certainly this has never been brought up by CDQA during my 2 years of analysis (and we have just filed an NDA.) The veteran (20 years with the company) members of my team have also never heard the need to lock integration parameters after validation.

[Dan]

scpercy,

I have never heard of such a thing. Nor have I heard of anyone ever requesting that and I have worked in a GMP regulated industry for nearly 19 years.

I would ask the quality rep. for clarification. Perhaps the person was only referring to individual runs. There should be a way to lock down the chromatographic parameters for an individual run either after the processing is complete or after the data review is complete.

However, there is no such requirement for the integration parameters to match exactly with those used in the method validation. How would you do that in the instance where a different CDS has been used? You would never be able to replace (or upgrade) your CDS or do a method transfer to another site that uses a different CDS.

Regards,
Dan

[DR]

I'll be brief on this:

[Peter Apps]

I am not a huge fan of eveything in a lab being regulated, but in this case QC might have a point.

It is possible to generate dramatically different peak areas from a given chromatogram by changing integration parameters, especially if resolution is borderline, baselines drift, or peaks tail.

For well behaved symmetrical peaks on a flat basline peak area is robust to integration parameters within wide margins, for some other situations a small chage in parameter can make the integrator miss a valley for example.

Why not do a post hoc robustness on integration using existing data files ? - vary the parameters by whatever and demonstrate that it does not make any difference to the results.

Peter

[PJ8]

It makes total sense to keep processing parameters the same for a single run. You'd have to have a really good reason to use two processing methods on the same run.

However to insist on not changing the process parameters for every run of the method from now to forever is daft. Identifing peaks would be a nightmare since you'd have to set wide windows to ensure you covered for possible future peak drifting.

I agree with Peter in that you need to keep integration as similar as possible between run, but surely being part of being a repsonsible analyst is checking processed results to ensure all the integration is acceptable. Inflexible rules in place of common analytical sense makes our job much harder.

[rc_12321]

One should not use different processing methods to process the injections of the same set.But it is rediculous to say that one has to follow same processing methods which have been used during method validation.I think one has to apply same integration parameters i.e one processing method from where actual quantiation process starts in the sequence.

However use of different processing methods needs good justification from regulatory and compliance point of view.One should keep in the mind that some of the warning letters issued by FDA included remarks on bad chromatographic practices also .Howevr locking integration parameters does not make any sense as it prevents permissions from changing parameters that best suits one's chromatograms .

[Consumer Products Guy]

Sounds like your QA (like ours) with limited knowledge or hands-on experience, is trying to justify their existence. Our test procedures state "integration parameters can be "fine-tuned" to optimize area measurements". Sorry one size does not fit all, unless you're selling scarves.

[HW Mueller]

But scarves have a lot of stuff in them that need QC, probably including HPLC! There have been some tragic cases here (Germany) of skin cancer being produced by blouse imports.

Reconsidering, I think I am not getting paranoid, but rather despondent.

As an example: What are you (the ones who would lock integration parameters) going to to if you have to ananlyze samples with highly variable matrices (Blood of patients.....) and suddenly some samples show overlapping peaks and your software places the baseline totally wrong, or the overlapping is not resolvable within your tolerance, etc, etc. etc. ?

Maybe it should also be pointed out that in some samples one might remove the cause of parameter drifts before integration parameters need changing.

DR, don´t you have a larger flag so that you can spell it out? (But then Tom might have to get active).

[HW Mueller]

Sorry CP Guy, just noticed that you hung that last sentence on size. Thought, on first reading, that you recommended that some people should sell scarves instead of doing analytics.

[rc_12321]

Sounds like your QA (like ours) with limited knowledge or hands-on experience, is trying to justify their existence. Our test procedures state "integration parameters can be "fine-tuned" to optimize area measurements". Sorry one size does not fit all, unless you're selling scarves.

Sorry, i do not agree with you.I have been working in validation group.

What do you mean by fine tuned ? Does your procedure say use one processing method for standard and another one for sample ?Then the result is biased .

What i am saying is that one has to use one and only one processing method to process the entire set that falls under one systemsuitability criteria.It may be some what tedious to apply this to tests like RS where many known and unknown peaks bother you ,but use of single processing method for the single peak tests like dissolution ,assay and some limit tests do not pose any problem. We are following this.

[Consumer Products Guy]

If one (e.g. Chemstation) locks in a threshold of 6 (for example), you may find in a few months you need to change to 5 for the peak to be recognized and counted. If you don't change the parameters, the peak will not be counted, and your result would be zero, requiring OOS investigation, having administrator adjust conditions, write a point of view for the file, etc. And retention times can drift a little as well. Our consumer products contain fragrances and components that stuff like "Bayer aspirin" doesn't, so our retention time windows need to be set a little tighter so that the correct peak is quantitated. Our operators are trained enough and skilled enough that I don't need to oversee every data processing. Yeah, I can imagine myself in a cell between Manson and McVeigh: "I'm in for mass murder, yeah me too - what about you?" "I changed threshold from 6 to 5"....

[Consumer Products Guy]

To clarify, we never use different parameters for standards than for samples; all stuff during the same sequence is integrated and calculated the same. Our ChemStore stores all the keystrokes and parameters, let the auditors interpret.

[DR]

DR, don´t you have a larger flag so that you can spell it out? (But then Tom might have to get active).

I have this:

[tom jupille]

On re-reading the posts, I think no one is seriously suggesting that integration parameters should be changed within a "run" ("one system suitability criteria"). In other words, a set samples should be run under conditions/parameters identical to that used for the calibrators.

I believe the original question dealt with day-to-day differences. In general, you are allowed to adjust chromatography conditions (parameters) as necessary to meet system suitability, but if the conditions are modified, the method must be revalidated. The question is "how large must a change be to be considered a modification?".

Given that the FDA allows reasonable default ranges of adjustment for other chromatography parameters (see Attachement A of ORA-LAB.5.4.5 http://ww.fda.gov/ora/science_ref/lm/vo ... _04_05.pdf ), I think it would be hard to justify a "no adjustment" policy on integration parameters. The magic words here are "as necessary to meet system suitability", which implies that you do, in fact, have meaningful system suitability criteria in place.