Validation: how to validate accuracy

[shaun78]

You are indeed correct in your comments above. This is not the most appropriate way of doing things and more conventional ways are definately preferred.

Those need to be followed if possible. Sure, if you can, you need to determine the responce factors and use those to evaluate accuracy. Then you would have a more reasonable basis for inferring accuracy form linearity.

Sometimes this can not be done and you are simply stuck. It seems like the original poster is in one of these instances. It is the best the OP can do and one is absolutely not allowed to validate an impurities method without attempting to show accuracy.

[alicee!]

hi al..
i'm working on similar method validation, and have no impurity standards.

it helps to know that it could be acceptable to do a linearity with the main peak when impurities are not available.
i'd like to read references that suggest this too.

further to this, if possible i agree establishing the correction factor for impurity response when the impurity is available could be done. in my case there are multiple small impurity peaks and that would be difficult to do for each one of these impurities.

going to establishing linearity.. if the impurities fall in the range of 0.3% to around 2%, would it be sufficient to show a linearity from say, 0.1-4%. would inclusion of the 100% response.. in the linearity evaluation be considered necessary? what are the ranges to be considered?

also, i would like to ask this..
from a 100% solution response.. if i go down to 10%, 1% and further 0.1%.. the response shows a loss in dilutional recovery..
meaning if 100% gives an area value of 100.. the 1% shows 0.8 (and not 1) and further 0.1% shows 0.05 (and not 0.1). This loss is observed to increase when the amounts go to the lower range.. starting say 4%.
Is this usually observed for chromatographic methods.. why? and is this considered to be impacting the accuracy?

in such cases, can area% of peaks be reported?

thanks..

[morse]

Dear HM and IMH,

I'd like to know what you'd do if you were in my shoes.

1) the impuriable is known and specified.
2) it is not in commerce, not stable and very hard to synthetise.

Though, if you are to declare the amount of it in a DS making use of a chromatographic method you need to be sure your method is properly validated (actually, I've been asked to do it).

If you think Shaun approach is not correct, what would you suggest?

Alice, I'll try to have a look at yur questions in a while... Tried to take a look at your post but need to pay more attention.

Morse

[HW Mueller]

I certainly would not assume accuracy. I don´t know whether I can help as I never had anything to do with this sort of quality control. In my situation I would see to it that my analyte (here called API?) is accuratly detemined and just note that there are unidentified peaks present.
This may help??: When confronted with dubvious propositions I changed the job.

[Bruce Hamilton]

There are some reasons why impurities can't be easily analysed, however, in my experience the requirement to analyse impurities accurately will increase as an API or formulated product progresses through to approval.

In early phase development, it's recognised that process changes could change impurities, hence there is more tolerance for surrogates or similarity, but once the process is is finalised, regulators expect impurities to be identified and quantified. The acceptable limits drop if genotoxic/carcinogenic impurities could be present.

I'd be very surprised if regulators would accept surrogate or % area wrt API for a late phase drug - unless the procedures/assumptions had been clearly demonstrated to be valid.

For mid-late phase submissions, assumptions about response similarity ( without independent verification ), would require serious justification - such as a risk analysis, with accompanying documentation, that only closely-related impurities would be present ( due to synthetic route, chromatographic cleanup etc. ).

For example, if an impurity is highly unstable, why would it still be present in the product?. If an impurity is unstable, you should be looking at/for the degradation products.

The original poster really needs to be discussing what is acceptable with their regulatory compliance/quality assurance teams.

Bruce Hamilton

[morse]

Thank you very much, Bruce.

Actually, the impurity I am dealing about is very difficult to synthetize.

Of course I have talked to QA managers here regarding to this issue and their point of view is similiar to yours (I doubted I could have been different!).

Though, there is a technical problem to sort out. And this kind of issue seems to be taken into consideration by regulators as it is mentioned in ICH Q2(R1).

"In cases where it is impossible to obtain samples of certain impurities and/on degradation products, it is considered acceptable to compare results obtained by an independent procedure."

As I have already said, I haven't come across a chromatographic method that allows me to analyze the aforementioned impurity yet.

I would like to know if someone in his own experience has managed to solve a similiar problem.

Thank you.

Morse

[lmh]

Morse, please, please, don't take what I wrote as a criticism of you; it wasn't meant that way at all.

Mostly, I'm annoyed with ICH for their apparently haphazard use of the word "accurate". The offending paragraph would have been more honest had they written something along the lines of: "if it is impossible to quantify in conventional molar or mass units (for instance because the detector response is unknown and no standard can be obtained), then it is acceptable to set up an assay in arbitrary units, provided the following conditions of precision and inter/intra-lab repeatability are met...."

I really think it helps to distinguish between accuracy and precision, as we were all taught at school, and as ICH themselves divide their document. But if ICH envisage a situation where there's no known "correct" value, and only one assay available (which I presume is the situation they're addressing in the poorly-written paragraph), then it becomes impossible to look at accuracy. Accuracy is error between the population mean of measurements and a "correct" value. Really, they're fudging the issue, and saying in the absence of accuracy, precision will do. Which is fine, but why not be open and straightforward about it?

After all, provided a drug contains not more of a contaminant than it contained during safety trials, it ought to be safe, irrespective of the actual molar amount of impurity.

[morse]

Imh,

Never though you were offending me.
On the contrary, thank you for expressing your point of view..

Actually,how can I disagree with you?

Yet, I need to find a way to accomplish this task in a "regular" way!

Morse

[HW Mueller]

Hm, the statement:
". . . . . it is considered acceptable to compare results obtained by an independent procedure"
doesn´t seem to be equivocal to me. I read this to say that if HPLC doesn´t work try another method, alltogether. Some of these other methods have been mentioned over and over again.

Also, a person like me who worked outside this type of QC has the possibility of defining his standards, I can´t see how an agency could say if you can´t get a standard, then use arbitrary units, etc.

[grzesiek]

some comments:
- one option is to do prep chromatography if impurity is hard to synthetize
- procent recovery and from linearity are possible ways of accuracy testing, when doing from linearity you first need to establish response function (area/concentration for example)
- the question is - do you expect this impurity in your final product??

[Bruce Hamilton]

You've talked to your QA people, so now talk to your manager/bean counters. Point them towards the QA people, and stand back to watch the fun.

Some choices - could be others,

1. Have QA perform risk assessment, identify what will be acceptable, and what must be achieved. I assume that because it's been identified, they will want some of the material at some stage, unless they plan on eliminating it.

2. Pay a contract lab or internal chemists to synthesize and characterise prepared impurity ( eg LC-MS, nmr, FTIR, elemental composition or perhaps HR-MS). Note that they probably only need to be a "research" lab, not a cGxP service, but the characterisation should be cGxP ( probably after you receive it would be cost effective ). There are plenty of labs globally that prepare such molecules - for a price, China and India are now in the game so the price may be a little lower.

3. Preparative HPLC ( that would be my choice - if you pay me enough I'll even attempt it, but I'm sure there are other labs nearer and cheaper ). You will need to confirm impurity identity, probably to cGxP - as above, using a separate technique. I'm sure your QA will advise on minimum needs.

4. Pay a CRO to perform all the work, and move onto something else. Doing an inadequate job because of insufficient resources will come back to bite you.

Please keep having fun,

Bruce Hamilton

[morse]

Thank you all, mates!

Morse