Residual Solvents Validation Question

[GoldLeader]

Hello all,

I am validating a residual solvent method for toluene, for which the spec may be NMT 890 ppm. The range I'm validating is 450 ppm - 1050 ppm (about 50 - 120%). My question: What to do about Limit of Quantitation? At 450 ppm our S/N is WAY more than 10. Do we need to run samples down until we reach S/N of 10, or would the lower end of our range suffice as the LOQ? We have an SOP in place so I have to follow that but I'm trying to get a feel for what is the common, accepted practice.

Thanks,
Jeff

[chromatographer1]

The common most accepted practise (I am not trying to be cute or smart-alec) is to follow your SOP.

If you don't follow your SOP then questions may arise, why did you not follow it?

Save yourself trouble and questions by following your SOP. If your SOP says to determine the LOQ by finding the level where the signal is 10X the noise, then do exactly that. Just keep diliuting your standard until you get there. Your dilutions need not be linear, from 450 ppm go to 100 ppm then 10 ppm, or whatever.

Remembering exactly why you deviated from your SOP a year or 10 years from now may not be easy. The less you do that requires explanation, the better. IMHO.

The added benefit is that if in the future someone wants to know "can you see 10 ppm accurately" you can give an answer, or if the FDA decides that the limit for toluene should be lowered, you will know without question if you can or cannot use the method for the new proposed limit.

best wishes,

Rod

[GoldLeader]

Rod-

Thanks for your input. I'm currently following the SOP but I'm also in a position to suggest changes to the document if they can be justified. Without going into too many details, the SOP would require 15 - 24 extra injections and with a GC / Headspace cycle time of about 80 minutes it takes a bloody long time to do this single experiment. I'm more or less asking if it acceptable to "define" LOQ based on the range study that has already been run.

-Jeff

[Bruce Hamilton]

Unfortunately, even though it often seems like it, chromatography in regulatory environments isn't Alice in Wonderland territory.

Lower Limit of Quantitation (LLOQ, often minus the "lower" - but there could also be an "upper" limit in some methods ), can't be redefined to match whatever you would like it to mean.

If you want to avoid performing a LLOQ determination, you will have to get approval that the method can be reported as suitable over the range X to Y, without reference to LLOQ. That's probably the responibility of your Quality Assurance group.

As Rod said, it's often preferable to have historical method performance information in case future needs arise that could require knowing the LLOQ.

Please keep having fun,

Bruce Hamilton

[chromatographer1]

Bruce's answer was excellent.

If I may diverge from the initial topic but to comment on Jeff's proper and just motivation for his question, (I apologize Jeff if I seem to criticize you, it is not my intention)

One way to reduce the cost of method validation is to develop a method that takes a minimum of time. Equilibrium of toluene from water or dimethylacetamide or DMSO only takes 6 minutes if the sample and vial size are optimized, and 10-12 minutes if you use a 22mL vial. For example, I once developed a headspace method to do the 5 USP <467> analytes in five minutes isothermally, and could do many solvents (toluene was one of them) in less than 10 minutes, from 1 ppm to 1% levels with the same method.

My point here is to encourage those in the industry to develop faster and cheaper methods of headspace analysis. It CAN be done.

Even faster methods can be developed today with a little technology. For example, having a "backflush" or a "early vent to waste" is very cost effective. I don't work for Agilent, but that is one reason they have a vent, a type of Deans switch, I believe, in their new GC models. Other vendors may have the same option as well.

In my opinion and experience, headspace methods should not take 80 minutes.

best wishes,

Rod

[Bruce Hamilton]

.....
My point here is to encourage those in the industry to develop faster and cheaper methods of headspace analysis. It CAN be done.
....
In my opinion and experience, headspace methods should not take 80 minutes.
Rod

My perception is that regulatory agencies are retirement homes for school prefects and similar folk. They never promoted efficiency or convenience, they enforced tidyness and strict adherence to rules.

In theory, the current emphasis on risk-based assessment could favour method innovation, but I suspect most incumbents are extremely risk-adverse.

Life's too short to spend time trying to retrain all the people who approve methods, so perhaps analysts just modify procedures that will provide the minimum energy path to approval.

Please keep having fun,

Bruce Hamilton

[Dan]

Jeff,

I agree with Rod that you need to follow the SOP.

From what you said, it seems that the SOP may be too restrictive. If that is the case, then change it! However, that won't help you now as the change will be after the fact.

What you have is a test method that is a limit test (with a limit of NMT 890 ppm). In the pharmacuteical departments where I have worked, we always worked to have the LOQ of a limit test method be NMT 1/2 of the specification limit and that the S/N should be NLT 10 (NOT that the S/N must be 10!). So, if the specification was NMT 890 ppm, then we should have a method with an LOQ of NMT 445 ppm with a S/N >=10. If we validated and found the method to have an LOQ of 445 ppm with a S/N of 25, then that was fine. We did not need to find a lower LOQ as the method is doing its job.

So, the SOP should not state that the S/N at the LOQ must be 10! It should state that the S/N at the LOQ must be greater than or equal to 10.

Regards,
Dan

[Bruce Hamilton]

This is trivial, and probably not worth pursuing, but my understanding of the LoQ is from Section 7 of ICH Q2...

" The quantitation limit of an individual analytical procedure is the lowest amount of analyte in a sample which can be quantitatively determined with suitable precision and accuracy. The quantitation limit is a parameter of quantitative assays for low levels of compounds in sample matrices, and is used particularly for the determination of impurities and/or degradation products. "

" The quantitation limit and the method used for determining the quantitation limit should be presented. The limit should be subsequently validated by the analysis of a suitable number of samples known to be near or prepared at the quantitation limit. " ( 7.4 )

As I noted earlier, if you don't determine the LoQ, as they recommend, presumably using one of the protocols shown in ICH Q2, then an auditor could question why you are calling a standard " the LoQ " when you haven't actually determined the " Limit ".

If you redefine the LoQ signal-to-noise ratio to a very large value, then you have also defined " suitable precision and accuracy ". To later redefine the LOQ, without significant changes to a method, could lead to potential grief with an auditor.

If your SOP says determine the LOQ, it presumably also expects you to use one of the typical protocols to determine the value.

My suggestion remains, do not define your lowest standard as a LoQ, but just as a standard whose precision and accuracy greatly exceeds the requirements of a LoQ. However, you still have to get prior agreement from your Quality Assurance people.

Please keep having fun,

Bruce Hamilton

[Paulhurleyuk]

Temporarily ignoring the issues over procedures for a minute, Instead of doing serial dilutions to find the point where your S/N is <=10, this can be estimated forma blank injection and your calibration curve.

Calculate the noise from a blank injection around the area of your peak of insterest, and then use that to find what concentration gives a peak height 10 times as big. Then just confirm at that concentration.

Just a thought

Paul.

[GoldLeader]

Thanks for all your comments. And Rod, No offense taken at all. I'm on the forum to learn. I completely agree with the "shorten the method" approach, but as we all know in a regulated environment, sometimes you have to play the cards you're dealt (I have little latitude to change the method at this point...). At this point it seems safer to go the long route and take a few extra days to validate to avoid any questions from the auditors later. Thanks again.

Jeff

[Dan]

A note of clarification for my previous post:

My comments were for the specific instance of a Limit Test for which there is a bit more leeway allowed than if it were an assay or purity test method. For the Limit Test, you only are required to have an LOQ that is below your specification limit. So, your LOQ does not necessarily have to be the lowest level that your method can detect; you just need an LOQ that is less than the specification.

If this were an assay or impurity method, then you want the actual LOQ that the method can deliver. You then determine the LOQ as Bruce has noted.

Sorry for any confusion. The Limit Test is a special case.

Regards,
Dan

[chromatographer1]

Your comments are solid and correct too, Jeff. You have to play the hand you are dealt.

In future methods, if you are permitted to develop new parameters, try to find a better way and do the same validation you would do with a traditional starting point method.

These old starting parameters (1gm of sample, 6mL of dissolution solvent, 60 minutes of heating) were originally used when most methods were to evaluate chlorinated solvents and hydrocarbons. What is interesting (and I have found articles which confirm my findings on headspace) is that one can get the same results in 1/10th the time if one uses 1/10th the sample and solvent, with the same sensitivity but with better reproducibility.

Don't take my word for it, try it yourself next time or in your spare time, do a little experiment using 200µL stds containing 100-2000 ng of solvent analytes. You can use even smaller samples. And try using 6mL vials not 22mL. Heat them for 10 minutes, not 80. And DON'T use uncoated BUTYL RUBBER septa. Use teflon coated silicone rubber septa.

And remember, if you cannot seal consistently you cannot do headspace accurately or reproducilbly.

I found that headspace was more accurate and reproducible than I was in making standards and samples. So I had to come up with more careful and exact ways of preparing samples to show the actual error using headspace. The biggest error I found in headspace analysis was in inconsistent sealing of vials due to vial imperfections and poor crimpling technique. The next biggest error was in using inappropriate septa.

So I hope you and your company are able to develop cheaper and better methods in the future. I KNOW it can be done.

with all best wishes,

Rod