Can single working std point be used for GLP study?

[flora0975]

If all the samples need to be diluted into one concentration, can I make a single point working std (same concentration to the sample testing concentration)? or I still need to prepare 5 points std which cover the testing range. It is an GLP study.

Thanks.

[Consumer Products Guy]

Make a five point linearity study for your GLP test method validation. typically, target, 80% target, 90%, 11%, and 120% can be used.

After that, make your standard and target samples of concentrations near that 100% target concentration, that's what we do for both GLP and for cGMP.

You'll also need to do the typical precision, accuracy, robustness, etc. Here we also need to do a protocol before starting (get it signed) and a validation report, our QA department demands that.

[flora0975]

The method is transfered by outside. It didn't mentioned std curve, only mentioned working std solution.
I think we can validate this method with 5 point std. In the real QC study, can we just use single point?

Thanks.

Make a five point linearity study for your GLP test method validation. typically, target, 80% target, 90%, 11%, and 120% can be used.

After that, make your standard and target samples of concentrations near that 100% target concentration, that's what we do for both GLP and for cGMP.

You'll also need to do the typical precision, accuracy, robustness, etc. Here we also need to do a protocol before starting (get it signed) and a validation report, our QA department demands that.

[whyallthefuss]

I am very interested in this subject as well.

[JGK]

I've validated quite a few single working STD methods for GLP studies.

Consumer Products Guy (CPG) is right, in the validation you have to prove that the WS can be used to accurately measure across the desired range. CPG suggested 80-120% relative to WS conc (I used to prefer 50 - 150%).

If you show acceptable measurement accuracy in the validation there is no issue with using single point calibration.

[unmgvar]

remember one of the golden rules in GLP,
Nothing is assumed, and taken for granted.

what is the scope of the GLP study?
is it simply to have a secondary lab for backup, is it to show greater robustness of the method? what is your SOP saying? what does the working protocol saying?


for example in the pharma world:
If the method is coming from a "tech transfer" from another lab/company, then you do need to ask them if they have done a proper validation of the method like Consumer Products Guy says, and see that the method is GLP compliant with the requirements such as linearity, robustness, accuracy etc...

if they have done it then you can use the method they have given you.
but to be on the safe side, like they do in pharma I suggest you to do what we call a "validation transfer".
best is to do in parallel the same samples in your and their labs and see that you get very close results.
also redo accuracy and robustness tests on your instruments as well since they might differ from those in the original lab, and most important the lab chemists are different

[Consumer Products Guy]

Right, test method transfer. We typically use three samples, each prepared in triplicte. One high, one low.

And one target.

[danko]

But 5 point linearity calibration in one lab (f. x. where the method originates from) on one certain instrument is far from justification of the same linearity on another instrument and in another lab.
Running several samples in both labs, in parallel, does not document consistency either. Because if these several samples represent 100 % concentration, or content or whatever, then there is still the possibility of ending with the same result no matter what the real concentration is.

Going into more detail, one can easily imagine different slopes on different days, or with different eluents for instance. Besides, when one point calibration is performed the assumption is that the (linear) standard curve intercepts the zero point. Now, how many of you guys have seen the latter?

In summary: Multiple (at least 3) points calibration bracketing the working range, is a valid requirement, if serious work is mentioned in the scope.

Best Regards

[unmgvar]

Danko,

People have been transferring methods for quite a while, and the greater problems are generally that of human error and not instruments.
tech transfer that i have been charge of required 4% difference top and it is not as easy sometime to get, mainly due to the sample prep and less the instrument.
thanks to physics behaving the same all over the world a method is yes linear on different systems from different vendors unless there are some very excentric differences in the specs of the modules
for example I had a case last week of a method moved from a UV with a 10nm slit to a PDA that had only a 8 or 16 slit. the sample is made of 6 peaks where only one was actually at Lambda max on the set WL all the others were in the middle or less of a spectra down slope. yes the picture came out very different, and no it was not checked for linearity when we asked

you are correct to speak of a set of 3 calibration curves, that is how we do it during validation.
right again the slope never goes through 0 but it not more then Y intercept then the effect for 100% zone values are minimal and so we can yes go to single point, because we are in a linear zone, thanks to physics

Running several samples in both labs, in parallel, does not document consistency either. Because if these several samples represent 100 % concentration, or content or whatever, then there is still the possibility of ending with the same result no matter what the real concentration is.

again if we do look at things in perspectives of risk assessment taking into account that physics behaves the same all over the world, running samples in parallel is one of the best approaches in order to perform a tech transfer without the need to do a full revalidation. again all a matter of the margin of error you set yourself, and the regulated pharma is pretty requiring.

in the end most errors come from the human side, and the most important is to know what is wanted, what is the scope of the work.

[danko]

Unmgvar,

Sorry I haven’t had the time to address your comments until now. But here you are:

tech transfer that i have been charge of required 4% difference top and it is not as easy sometime to get

At the companies I’ve worked for that number used to be 2% and I don’t recall too many difficulties living up to it. Maybe because of the calibration procedures?

thanks to physics behaving the same all over the world a method is yes linear on different systems from different vendors unless there are some very excentric differences in the specs of the modules

Lows of physics are the same, but conditions certainly not. It could be temperature, it could be flow cell volume or even flow cell light path, it could be a polluted chemical used to prepare the mobile phase and yes different hardware makes and even models render different detection behaviour. All these things and more, can cause, and typically will cause different linear range and especially standard curve slope. Your example illustrates the issue quite well and there are many more examples of the kind.

you are correct to speak of a set of 3 calibration curves, that is how we do it during validation.

Actually I was talking about 3 point (level) calibration and not only during the validation, but every time a calibration is performed. The minimum 3 levels are needed in order to create a cure with a relevant slope. During the validation requires are typically 3 curves, 5 levels.

right again the slope never goes through 0 but it not more then Y intercept then the effect for 100% zone values are minimal and so we can yes go to single point, because we are in a linear zone, thanks to physics

The effect for the 100% zone maybe minimal, but that’s what you don’t know – i.e. whether or not the samples (unknowns) contain 100 % the same amount as the standard, hence the term “unknownsâ€

[unmgvar]

Since the matter of the thread is more important than single limited opinions
I guess that flora0975 will have to decide what to do depending on her needs and only she knows according to her SOPs and protocols what is best suited for her