Chromatography Forum

Hi, I have to validate a residual solvent method- i have also to design the protocol. I went to ICH guidelines and found that linearity should cover the range from the reporting level of the impurity to 120 % of the specification therefore, linearity should be performed from the QL to 120 % of the limit. So if the limit was 5000ppm then this would be from QL to 6000 PPM

Accuracy assessed on samples spiked with known amount of impurity at 3 levels/ 3 replicates. Therefore the drug substance should be spiked at 3 levels. (Maximum 120 %?)

Robustness should examine different columns and different temperature and flow rate?

Since residual solvents are volatile it seems pointless to look at solution stability for longer than 24 hours.

Please can someone let me know if I am going in the right direction

Thanks

Liv

You are going in the right direction.

I would not concern myself with solution stability more than the time that is required to perform the analysis, perhaps 30 minutes.

Robustness? I would look at three different brands of the same column, or 3 of the same brand on 2 or 3 different GCs using at least two analysts.

I would vary the flow rate by ±20% and injection size by the same amount.

I would expect my results to vary by a defined amount using the dimensions specified above.

I hope this is helpful.

Rod

Hi

Regarding the stability of solutions I would actually consider examining the stock standard solution depending on what sample solvent you use and how often you expect to run the method.

If water is used as sample solvent it is like Rod stated usually not necessary to study the stability as the solvents typically reasonbly fast evaporate over time.
But as you indicate a pharmaceutical analysis you maybe use DMF. DMSO or DMMA as sample solvents and as these solvents typically retains solvents better the stock solution may be stable for a week or more if stored at 2-8°C. This could be worthwile examining if you intend to run the method often and/or if the solvents in question is less fun to handle from a SHE perspective.
The QC lab doing the routine testing will "love" you if they do not have to redo the stock standard every second day


Cheers Chris

Hello,

For validation of residual solvent ICH guidelines ask you to do different things. What we usually do where i work is test precision (100%), accuracy (50%-100-150%), linearity (5 to 10%- up to 300%), specificity , sensitivity (DL-QL) and no with headspace analysis, stability is not tested. Only thing with stability is that we check that calibration standard throughout the run are stable within 85-115%. For linearity point if you try to evaluate it from the QL you could get headaches. QL is usually very small. We usually do around 5-10% of limit for each solvent. There is no need to do linearity so low. For robustness we usually test 2 columns on different systems.


If you need more info please reply and ill be glad to answer back.

Willy the GC

Willy,

What are the headaches if linearity is evaluated from QL to 120% specification limit?

I have done residual solvent linearity this way for some time, I got a good linear regression line and 0.99+ correlation coefficient. It seems fine for me.

Terry

Willy,

What are the headaches if linearity is evaluated from QL to 120% specification limit?

I have done residual solvent linearity this way for some time, I got a good linear regression line and 0.99+ correlation coefficient. It seems fine for me.

Terry

What I think Willy is trying to say and what I also agree with in general is that, sure the linearity many times is accepteble down to the DL, but from a Range perspective the accuracy and precision usually detoriates quite a bit from the QL from time to time. So practically there is not always a need to push the linearity invastigation to far down in concentrations.


I have one/two other comment regarding how far down in concentration you want to push your validation.

For organic impurities it is somewhat easier as ICH Q3a+b quite cleary tells what the expect, but for residual solvents (ICH Q3C) it may become more complicated especially on the global market. Class 3 solvents rarely complicates things, but class 2 solvents tend to. Especially in EU and Japan regulators tend to sharpen the limit below the ICH Q3C limit based on the batch data submited. Just had a discussion with another company where EU dropped the toluene limit from 890ppm to 100ppm for one of their drug substances.

Also if your submiting or want to change a specification limit for a class 2/1 solvent in Europe, EU (EMEA) has published an annex to the ICHQ3C guideline with clearification on expections. For example what data/discussions is needed to rule out a class 2 solvent from the specification.
This can also have impact on how far you want to push the validation. Unfourtunatly my PC is currently not working as expected so I can not provide a link at this time but you can find it on http://emea.europa.eu/, continue to human medicines and look in quality.
The annex came into operation in January 2005 and is labled:
CPMP/QWP/450/03
EMEA/CVMP/511/03
in the top right corner.


Cheers Chris

Krickos,

Thanks for your sharing your wonderful insight

Regards

Terry

Hi all

Just updating a link. As suspected EMEA had move the position paper on class 1 & 2 solvents.

http://emea.europa.eu/index/indexh1.htm

Alternatively:
http://emea.europa.eu
continue to human medicines and look in "CHMP working parties" on the folders to the left. Then click "position papers" and look for:
CPMP/QWP/450/03 Position Paper on Specifications for class 1 and class 2 residual solvents in active substances (Released for consultation April 2003)

Cheers Chris

(Two hours left of work then 4 weeks vacation )

Hi is it possible to vary injection volume by +- 20 % using chemstation software and 6890 gc it does not seem possible
The injector seems to be taking sample size as % of syringe volume
Can anybody help?
Thanks
Liv

How about we try to make this a little easier?

What is the point of your residual solvents testing? Is it the same as the USP/EP?

If yes, then you only need to be able to state that you have less than a specified amount. You do not really need to know the exact concentration in ppm.

If this approach can be taken, it sounds like we have a limits test on our hands. In which case all you need to look at is specificity and precision at the desired level.

In most of what I have seen, this is all people want to know in the first place. As long as they are below the limit, they don't care what the number is exactly. All too often I have seen people work up full analytical validations just to they can state NMT 3500 ppm on a C of A.