Chromatography Forum

In one of our fragranced products (a wash-off product), supervisor noticed a tiny bump he said "could be integrated" if zoomed-in and integrator events changed. Supervisor said all peaks must be integrated, even if only observed when zoomed-in, and ALL such peaks have to be resolved fully (like 1.5) from the active pharmaceutical ingredient. Well, fragrances can have up to 200 ingredients, and I contend that most fragrances could have tiny "impurity" peaks if one really hunts low for them. He's terrified that the FDA will come in some day and find this out. Are FDA auditors really that anal and vindictive?

OK, supervisor says we should just dilute samples and standards further so thatany such "impurities" are hidden in the HPLC baseline and do not meet 3:1 signal to noise, and re-validate the method. But that really does absolutely nothing to improve the signal to noise of the API v. the fragrance component. Does this make any sense to anyone, like is this an accepted technique? And I assume that I must revalidate the test procedure if I choose to inject 2.5 ul instead of 5 ul (I estimate if the peak was halved that we couldn't see it at all), that's what supervisor said.

So I responded to schedule in which month I should do this, as other work would be suspended during that. Our other fragrances in same matrix do not have the same tiny component in their fragrances; we sure would hate to have a separate validated test procedure for just one fragrance, would cause confusion at QC locations. Comments appreciated, thanks.

Hi

Well strickly speaking I would say your supervisor is right regarding that all peaks should initially be investigated, howver once it has been established which peak area/height is less than the LOQ, disregard limit (Ph Eur monographs and ICH Q3A+B guidelines), (or S/N 10:1 rather than 3:1), those peaks could be disregarded as they normally would not be included in total impurity content (if specified reported as less than LOQ). This is according to ICH Q3A and B guidelines, see end of both guidelines that contains examples.

So I do not really see a point with revalidating the procedure and I am confident that auditiors will recognize the difference between a complex fragrance and lets say a more or less 100% pure drug substance.

I have enomous problems with "hiding" anything.

I have enomous problems with "hiding" anything.

Forgot to comment on that, yes I agree with Mueller, try justifing that for an auditor.

Here, during specificity we inject all raw materials separately and in the same concentration as the product to identify them. If none are withing 1.5 resolution of the active peak we move on with validation. If there is a peak that is at the same retention time, it must be less than 1% of the peak area for the active in the product. I have had this problem many times with fragrance, dyes and flavours! As long as you prove the peak is less than 1% of the active peak area then you should be fine. When QC is testing for release, only the peaks of interest need to be integrated. There can be tons of other peaks, but as long as they are separated from your active peak and have been previously identified, you should be fine. I don't know if it's proper, but if we have an extra peak show up from contamination etc, we don't integrate it if it is separated from the active.
That said - if this is a method for stability, or impurities, then they probably all should be labelled and integrated I think.

If there is a peak that is at the same retention time, it must be less than 1% of the peak area for the active in the product. I have had this problem many times with fragrance, dyes and flavours! As long as you prove the peak is less than 1% of the active peak area then you should be fine.

Cody84 - do you have a reference for the above practice, like from ICH, USP? In this instance the fragrance component for us is likely 1/500 the area of the active peak.

One never would even notice it while using traditionally-accepted integration or scaling, and definitely would have been lost if one was just using an integrator and no after-run processing.

What if you inject an LOQ solution which will give you the height/area to use as justification that any peak smaller should not be quantitated or reported.

I can't seem to find the reference today. I will try again tomorrow, I know we have it in our method validation SOP and that SOP is based on ICH...so hopefully I can find it.

Hi

From ICH Q3B:
1.3 Scope of the guideline
This guideline addresses only those impurities in new drug products classified as degradation products of the drug substance or reaction products of the drug substance with an excipient and/or immediate container closure system (collectively referred to as “degradation products” in this guideline). Generally, impurities present in the new drug substance need not be monitored or specified in the new drug product unless they are also degradation products (see ICH Q6A guideline on specifications).Impurities arising from excipients present in the new drug product or extracted or leached from the container closure system are not covered by this guideline..........

So the above relates to what Cody says, still that is for known peaks, for unknowns (not traced to excipient or uncertain if degradation product or not...) the post I made is valid. And yes injecting a LOQ soulution is not a bad idea.

I don't think the above section covers my situation. Even though I know the identity of the tiny fragrance component, it is not resolved as to requirements from the active peak. So the tiny peak doesn't meet the criteria of being resolved away enough from active.

In a perfect world, the testing procedure would have an statement like: "integrate all unknown peaks >=0.05%" if 0.1% was the disregard limit.
We usually scale our purity chromatograms that a 0.5% standard would be full scale.
Having a 1% placebo or matrix or blank peak underneath the main peak is certainly inacceptable. In that case any assay values are useless as you don't know wether its 94.5% or 95.5% - thats the difference between not approving or approving a batch.

In a perfect world, the testing procedure would have an statement like: "integrate all unknown peaks >=0.05%" if 0.1% was the disregard limit.

Alex - do you have an official reference for the above? Like: is that acceptable for cGMP ?

ICH Q3B, page 7
Reporting Threshold A limit above which an impurity needs to be reported.

That is valid for pharmaceuticals.

Thanks, I'll try to get a copy of that.

But I believe my issue will be that the one tiny fragrance peak (present in just one of my 27 fragrances tested with the procedure) does not have the resolution from the active peak as detailed by regulations.

In fact, since my supervisor is saying that any "peak", no matter how small, under or near the active peak renders the sample unable to meet cGMP specificity or resolution parameters, that it's likely that by that definition that NONE of our assays would meet that, as they all have very tiny fragrance components eluting everywhere. Even our QA says that those peaks are so small so are negligible, says FDA auditors would understand that, but my supervisor says company can't take that risk....

Someone must have encountered this before on cough syrup, lozenges, mouthwash, liquid soaps, anti-dandruff formulations, etc. It's essentially the "what is zero" question in the modern world. Please help if you can.

http://www.ich.org/products/guidelines/ ... lines.html

You should have written procedures or SOPs that state which is the minimum peak size (0.01% would be sensible).
On the other hand you could dilute your sample to a concentration in which the required sensitivity is given, i.e. 0.05% impurities are just above the qualification limit. That has nothing to do with "hiding", thats just adjusting your method parameters to regulatory needs.
BTW there is no "minimum resolution from the active peak as detailed by regulations" AFAIK. 1.5 are often mentioned but can be insufficient in some cases and to strict in others.
The real question is: does the fragance peak influence the assay oft the active? If there is no or just little effect in recovery then I wouldn't care. Does it influence precision?
Another try could be to switch to more specific wavelengths.