What is zero?

[Consumer Products Guy]

OK, I've got a question (I'm a separations scientist much more than a cGMP expert). For a cGMP validated test procedure, sure, we'd all like the formulation blank to be absolutely blank. However, in real life, with modern computerized systems, one can always zoom in on the formulation blank chromatograms and adjust the inegration parameters, where oftentimes tiny, but real peaks might be measurable. What criteria do you all use to justify that such tiny peaks are negligible. Do you say if the tiny peaks are 100 times smaller than the active API? 500 times smaller? Is there USP/cGMP/ICH guidance on this, or is this just another "gray" region?

Sure, under normal integrtion the data station shows no peaks in the formulation blank, but if I set it sensitive enough: you know the issue. Thanks.

[HW Mueller]

A nonregulated view: You define what you neglect via S/N. But if you consistently have something there that exceeds the S you know that it isn´t clean (zero).

[Consumer Products Guy]

A nonregulated view: You define what you neglect via S/N. But if you consistently have something there that exceeds the S you know that it isn´t clean (zero).

HW: thanks for the response, but I ask for clarification (from HW and others). Say I've got a sunscreen product with 15% octylmethoxycinnamate active which gives me a huge peak. The formulation blank contains a fragrance component that gives a tiny signal at that retention time and wavelength (let's say 30:1 signal to noise for the tiny peak), but if I calculate that tiny peak as if it were OMC, would calculate as 0.005%. Obviously 0.005% would be negligible in determining the real OMC content, but the question is: does my formulation blank need to be actually ZERO (if that's ever attainable), or should I just set my integration parameters to exclude such tiny peaks, to provide real-world quantitation?

[unmgvar]

if you go according to ICH then any and all compounds and expients need to be chromatographically separated one to each other and should specifically not interfere with your main compound.
so if you have a fragrance at about the same RT then that of your compound you need to improve separation so that it is sufficiently separated.
with sufficient separation you can in your method then state to disregard that fragrance peak.

[Anthony_Ng]

Let me share my experience to you. (May/May not be useful)

I was working in a commercial testing lab for 1.5 year, checking for Heavy Metals of, say toys, clothes and accessories with ICP-MS for suppliers before they ship the goods to overseas.

First we check the Limit of detection (LOD) and instrument detection limit. Hence, according to different extraction methods (e.g. microwave digestion, artifical sweat/saliva extraction, etc.), we run the so-called "sample blank" or "matrix blank".

Sometimes we see there are signals on the specific heavy metals in the blank. We then check if the signals are noise, above the LOD, carryover or not. And we run customers samples, subtract the signals from the sample blank and calculate the result.

If we see the signal are very close to the LOD, our boss will say go ahead and draft the result. If not, he will suspect something wrong somewhere, maybe from matrix, reagent or the instrument. Usually it takes very long time to find out the cause.

What's more, the newer generation instrument has lower detection limit. As we will buy some new instrument every year (e.g. I saw 3 generations Agilent GC-MS in the lab). We can only say the "zero" by old instrument is not relly "zero" as the new technology come.

[HW Mueller]

OK, what I tried to say is that in my opinion one can not say something is zero when one knows that it is not. Even a consistent peak of, for instance N/S of around 2 is not zero. Considering your definition of N/S or from the size of your analyte peak you might be able to neglect it, but you can not state that it is zero in the blank. Mathematically, that would repeat what I sayed before: It is not scientifically sound to force a curve through zero. If you do it, but clearly tell that you "forced", it might be acceptable, doesn´t make to much sense, though. This is an opinion, what regulators say is irrelevant to me.

[LC_labrat]

From a GMP perspective a 0.005% peak would not effect an active of 100.0% purity. You can set integration to anything that is greater than 0.05%, thus it won't effect your final result. Anything less than that is negligible.

[Dan]

CPG,

As I think you have realized, it is not that the ICH specificity guideline states that you have zero interference. The intent of the guideline is that you have no interference with the analysis of your analyte.

As HW Mueller and LC_Labrat and you point out, you are not looking to have zero interference, as you may never have that and you need to define how much interference is allowable.

In the regulated labs where I have worked, we used either a vlaue of 0.1 or 0.5% as the limit of allowable interference. The limit is relative to a 100% analyte peak.

So, as LC_Labrat noted, your fragarnce peak is 0.005%, then it is not an interference in the analysis of your OMC peak.

Of course, you need to base your interference limit on such factors as your method capability (LOD, LOQ, etc.) and any other pertinent factors such as what your QA group would let you use.

Regards,
Dan