Chromatography Forum

Dear All,

We validated a GC method of a drug substance, containing 6 OVIs, 2 years ago ( by using Trace 2000 GC ). Nowadays, we are testing this drug substance on the same GC, but we can not get same peak area from standart with the validation results for all OVIs, some of them give same peak area but others are about 2 times greater than the validation results.

Can we use this method? or must we revalidate the method? or what can we do else?

Thanks for your hepls in advance.

If your standard substance has a different amount of solvent now from the amount originally calculated it may not be the method that is the cause.

Your substance may be breaking down and producing the solvent increase.

Alternately, the substance itself may not have been homogeneous in residual solvent content. One bottle may contain the substance with more or with less residual solvent.

If your standard that you refer to is not the drug substance but a mix of solvents external to the drug, an external std, then the differences in what you are seeing may be an error from the original validation or from the standard at hand.

It is not likely that a detector would change as much as you indicated for only part of the solvent mix.

If you cannot duplicate the results of your original validation, then I would say your method is not validated due to a lack of ruggedness.

Could you give a bit more detail to the situation?

best wishes,

Rod

Dear Rod,

First of all, I would like to thank for your interest.

We have met the problem in external std., which has been prepared freshly, using same brand Referance Stds at the same concentrations ( in the past and now).

Areas of TRACE 2000 GC

OVI______________~RT______________~Peak Area
(in ext. Std.)__past____now_______past_______now
MeOH.............7,583......7,610.........4338660......5294288
EtOH..............10,373....10,445.......5811965......10380016
AcON.............12,206....12,298.......20208062.....35394817
MeCl2............14,579....14,712.......3076915.......3420816
n-Hxn............18,109....18,275.......22226917.....43006369
Diisopropyl....19,813....20,022.......84549309.....160953170
ether

Also reproducibility test was performed in an other lab. by using HP GC.
~Peak Area
past______now
85.2..........84,8
154.2........160,6
497.3........471,8
37.6.........50,2
286.7.......491,3
1465,9.....1876,6

In our lab. , there are problems with areas of Ethanol, n-hexan and Diisopropyl ether, but n_Hexan and Diisopropyl ether in the other lab.

When method validation was performed, linearity test was done between 10-150% level of target concentration. It seems; Ethanol, n-hexan and Diisopropyl ether are out of linearity range.

(Note: Repaetability result and reproducibility result are in the acceptance criteria.)

It appears the problem is probably in sample/std preparation.

Your analytes are volatile and errors can easily be made.

I would have to be familiar with your entire methodology to trouble shoot.

Sorry I cannot be of additional assistance.

best wishes,

Rod

This could be in inlet problem. The main component has a higher MW than the others, so to get repeatable transfer of sampl eto column might be tricky. What are your GC conditions - in detail please.

Peter

The method details are below;

Inlet ( split )
Temp.: 220 C
Split flow: 30 ml/min
Split ratio: 20:1

Carrier (Nitrogen)
Flow: 1.5 ml/min. (constant flow)

Dedector
Temp.: 270
Air flow: 350 ml/min
Hydrogen flow: 35 ml/min
Make up gas flow: 35 ml/ dk

Standart solution is prepared at limit concentration in DMSO. Headspace technique is used.

What column are you using - length, diameter, stationary phase ?

What do you mean by headspace technique ?

Are you injecting samples manually or automatically. If automatically, which autosampler are you using ?

Peter

The differences you see are extreme, almost double the recovery seen now as opposed to then. Are the conditions the same ? Headspace recoveries can be affected greatly by conditions if not optimised, was this done to validate the method ? Is your current method right and the fault was with the old method/results ?
Thermo describe a headspace method for recovery of organics in Pharmaceuticals by TraceGC/HS2000 Headspace Application Note 528 have you read it ?

Unless you run the same headspace with the different GCs I am not surprised they give different results. Are you running the same conditions on both GCs ? I ask because with the Trace I would not be using FID make up, and I would run the Trace injector and detector 20C lower than the figures given (and that were used on the Agilent). I mention this particularly as there is no requirement to run the Trace injector at 220C for this type of analysis I would have thought 180-200C max.
I hope you are/have always run an automated Headspace method (both GCs) as manual headspace could easily explain these differences.

Can it be the make-up gas type which was caused the problem?

May be, we have used different make-up gas in the past from now. You can chose make-up gas as helium or nitrogen or none for Trace GC.

simyaci2501

You appear to have an inconsistent headspace methodology. You have not given additional details as Peter requested.

Since you are forcing us to guess I will say:

If you are using manual headspace, it is very likely that your technique or your ambient conditions are slightly variable. This can account for your differences in peak area and between labs.

If you are using automated headspace, the most likely problem would lie in the sealing of the vials. It is a common cause for errors in headspace analysis and it manifests itself in different analytes having different relative responses, exactly as you have described as the problem you are seeing.

This can also be a problem in manual headspace analysis.

best wishes in finding the cause and resolution of your problems.

Rod

Additional details as Peter requested are below in blue;

1. What column are you using - length, diameter, stationary phase ?
The column is DB-624, 30x0.53 and 3 u.

2. What do you mean by headspace technique ?
Are you injecting samples manually or automatically. If automatically, which autosampler are you using ?
We are using automated headspace which is HS2000 for Trace GC.

We have injected standart solution 6 times consecutively. The % RSD of 6 consecutive injections is less than 5%. So I don't think the problem is the sealing of the vials.

Explanations for the differing peak areas from those obtained before:

1. original work was in error in preparation of standards.

2. the volume of sample (carrier solvent) in the vials has changed although the amount of analytes spiked into the sample has not, so the partition coefficient has changed. This can greatly affect hydrocarbon partition if the solvent is non-polar (I believe you are using DMSO) which is the case. Alcohols and other polar analytes would be affected to a much smaller degree.

The only other variable I can think of is that splitting of the sample is undergoing a discrimination due to column installation position.

best wishes,

Rod

You have inconsistent peak areas on two different instruments, and good repeatability of peak areas on one instrument. This points to the problem being in the preparation of the samples and/or standards.

There could be a problem with the conditions of headspace sampling, but you have still not told us what they are.

Peter

In particular please check the vial size and the amount of solvent placed in the vials. Did you originally use 1mL of DMSO in a std vial and later used 0.25mL?

This will affect the relative partition of the alcohols versus the alkanes.

A leaky vial can also discriminate between analytes. (not likely to this degree, but possible)

Most likely this is a human error, and that happens to the best of us.

Please let us know what you uncover. It is an opportunity for us to learn as well as you.

best wishes,

Rod

Hi,
I agree with previous comments. I want to emphasize on headspace conditions, such as vial temperature, injector or sample loop temperature etc.
Also HP (Agilent) and Trace HS2000 (Thermo) are very different headspace designs, it's no surprise to see different results. I believe headspace sampling details are not included in the method and/or method validation documents.
best luck


thanks