Method Validation - Stress Testing

[BMac1984]

Hi All,
I am working in a cosmetic / pharmaceutical laboratory for a contract manufacturing company. Well I inherited the method validation protocols, and currently the FDA is here doing an investigation. We pretty much follow ICH guidelines and do the typical LOD/LOQ, Linearity, Accuracy, Precision, Ruggedness, Repeatability, and Selectivity/Specificity. But a question arose about our stress testing / forced degradation. This is how our protocol was written in summary. Stress your standard, sample and placebo in acid, base, oxidation (peroxide), UV-Light, and Heat. Compare the peak areas of the control to that of each stress sample. A comparison of less than 5 % difference is acceptable (I do understand we should try and shoot for a 10 - 20 % degradation). And this may have been his first issue. Our problem is we only have UV detectors on our LCs. I understand with a PDA detector it would be very easy to do comparisons and ratios. I guess my question is if anybody has some suggestions as to how to go about doing this with only a UV detector? The lead chemist for the FDA mentiond it is 'accepted' to analyze at two wavelengths and do comparisions. But in response I said how would I still be sure that there is still no interference at my wavelength of interest. At this point any suggestions would help? All of our upper management looks like this right now....
Thank You For Any Quick Suggestions!
Regards
Brian

[HW Mueller]

This question,

". . . . how would I still be sure that there is still no interference at my wavelength of interest",

applies to a PDA as well.
Just wondering: Are there still any UV detectors on the market with which one can not easily run a spectrum? Now, one can even run (not so easily) a spectrum with a filter UV detector if it has different UV filters.

[pkenny]

The assumption is that if there are two compounds present at the same retention time then they would have different UV spectra. So, if you measure stressed and unstressed samples at two different wavelengths and determine the peak area ratio (for example, A215/A254) then the ratio will be the same for stressed and unstressed samples if the peak is pure.

It isn't the strongest assumption, but you have to do what you have to do with the equipment on hand. It is an accepted practice.

Good luck

[HW Mueller]

Who accepts that?

[krickos]

Who accepts that?

Appearantly FDA

The lead chemist for the FDA mentiond it is 'accepted' to analyze at two wavelengths and do comparisions.

I am personally also a bit surprised. Especially when reading this out of ICH Q3B: "When identification of a degradation product is not feasible, a summary of the laboratory studies demonstrating the unsucessful efforts to identify it should be included in the registration application"

Doubt simply stating we only have a UV detector would do in that case, so very surprised to see a laboratory doing forced degradation without having support by PDA, MS detectors to rule out the potential of overlapping peak.
If your a small company you can use contract laboratories I suppose.

Or have I totally misunderstood the question?

[BMac1984]

Thanks pkenny, I guess I was having trouble with the 'assumption', you know what kind of trouble we can get into when we start to assume things.

As far as the acceptance of this. Like I mentioned it was stated by the lead chemist of the FDA. I had trouble believing him too, but he is the FDA, and we all know that maybe the next person to do the audit will say this is not acceptable. Yes he did mention that a PDA or MS is absolutely the better way to go. But mentioned this was how it was done pre-PDA, so it is still accepted. This company is in no position to purchase a PDA or MS in the near future, so I guess this is how it will be done for now. For what it is worth.

Thanks Again,
Brian

[HW Mueller]

From the FDA statements given here I don´t see that they say it is ok to make an assumption that one can not make. On the contrary, ". . . .a summary of the laboratory studies demonstrating the unsucessful efforts to identify it should be included in the registration application . . . . , indicates to me that one has to indicate that one does not know what he has. Such statements do not indicate to me, at all, that one can assume a peak is homogeneous when the spectrum is ok, etc., etc.

[pkenny]

I think these are two different concerns - specificity vs. identification.

In the original post the concern is that there could be a degradation product peak hidden under the main peak. My answer was to how a UV detector can be used to help determine that. In order for that method to work - an assumption has to be made that the degradation product and main product have different UV spectra. (If they do have the same UV spectra, then even a PDA detector would not help) This is a method that was used prior to PDA detectors and LCMS. It certainly is not as strong as using a PDA or LCMS, but can be used.

Your posts are addressing identifying a degradation product peak in a chromatogram. If you have a degradation product peak in your samples that is above a certain level, then yes, you have to identify it. By identify, I mean determine what its structure is so that its potential toxicity can be examined. If you cannot identify it, then you would need to explain and rationalize why not.