Hi all,

I have a question about how to perform the accuracy determination during a HPLC validation. I know your first reaction would be to point me at FDA, ICh and so on, but I've got them all read and it is still unclear.

The thing is, they always assume you're quantifying an API in a drug product. So here it's OK, you calibrate your system with the 100% API and you do the accuracy with a drug product containing the 100% API.

But I am validating a method where I quantify a chemical alone, not in a drug product or any formulation. And there is no other method than the one I amtrying to validate (ie no "reference" method or whatever). So, I run the calibration, linearity, precision and so on using my 100% standard. And what about accuracy ? If I have read well, I must make sure that, when I quantify my 100% standard, using my calibration (done with that same 100% standard), the result is 100% ? If not I'd surely kill myself!! There must be something I'm missing there... or is it just that in that very case I can only assume accuracy from the results of linearity and precision ?

Could someone please help me understand, or am I really to dumb for that ?

Thanks in advance.

This seem more like a question about how much do you have to (need to) do to characterize your (reference) standard.

How pure is your "chemical"? Was it synthesized or purchased? Does it have a C of A. Can the manufacturer/vendor provide any purity tests? Does it chromatograph as a single peak? Have you analyzed for percent moisture? Etc. etc.

It makes no sense to calibrate with the exact same sample that you are going to analyze.

OK, I'll try to be a bit clearer.

We synthesize organic compounds which usually are not commercially available elsewhere. What would be my "reference" is the purest compound obtained in our lab, which usually comes out as a single peak, eventually melts in a 1°C range, and so on ; and which I assume to be at 100%. My problem is the validation of the HPLC method that I will use to quantify the purity of other samples of this molecule.

I would tend to say that, in this case, I can only assume the method is accurate once I have checked that it is linear and precise, but I would like to have other peoples' opinions on this matter.

Dominique (aka "anonymous" in my first post, sorry for that!)

Because you have the only available standard, performing accuracy studies on the HPLC validation seem like a waste of. I will pass on a word caution however (experience talking here), make very sure that your compound is as pure as you think it is. Use multiple methods of analysis and detection if possible. Your method validation is only as good as your standard.

Some anecdotal advise here: a company I previously worked for synthesized some otherwise unobtainable compounds. These compounds were then used for calibration of an HPLC method determining their levels in a variety of products. When these compounds became available commercially, we discovered that our original standards were not as pure as we thought and the company was faced with revising product specifications downward. Not a happy scenario. The lesson: verify the purity of your standards as accurately as possible.

remember chromatograqphy is not the only analysis-method:
-titrimetry
-graphymetry
etc,etc
you can use these to check purity of the standards...

I agree with the previous anonymous,

You can also try with:

- GC or GC/MS

- HPLC/MS

I hate to beat a dead horse here but also do not forget about NMR analysis. While it may not be the most sensitive technique for purity analysis I have seen some cases where it forestalled some nasty surprises that the chromatography did not detect.

We have also faced this issue. One approach we took was to show the specificity of the HPLC method by force-degrading the sample or using a lot of material that was known to be impure. Demonstrating that the UV-HPLC method would be capable of resolving away any impurities if they were present left us confident that the material really was pure, and not just eluting as a single band. However, we also sent the sample out for testing, including elemental analysis and LC/MS/MS. Our HPLC method development plan was quite rigorous, examining several gradient, temperature, stationary phase, and pH. All of this evidence taken together provided us with enough documentation to be able to state that the method was specific and accurate. We then issued our own C of A for the material and used it as the reference material for subsequent lots. We haven't had an FDA audit, but hopefully, we're covering our bases.

For accuracy forAPI/chemical ,one should very the results of the HPLC method with with other alternative methods like, DSC, Titration and compare the recoveries

I agree with everything above, but the question is still : how do I do ?

Let us assume I have checked by all possible means that my standard is 100%, or 99.6%, or any other value. Since I am doing external calibration, I am using this standard, with the given value, to calibrate my method, right ? Then again, how could I not find 100% recovery when I assay my reference against itself, or ANY OTHER SAMPLE ASSAYED AGAINST MY REFERENCE !!

It seems to me, in the end, that I need TWO samples of KNOWN (or assumed) assays, obtained by TWO INDEPENDENT METHODS !!

Am I going wrong again ?

The ICH states that accuracy for a drug substance (API) may be determined by ONE of the following:

a) running your method using a standard of KNOWN purity (i.e., USP standard, etc.)

b) comparison of your method results vs. another analytical method that has a known accuracy (i.e. it is stated and defined)

c) "accuracy may be inferred once precision, linearity, and specificity have been established."

Taken from ICH Harmonised Tripartite Guidline, "Validation of analytical Procedures: Methodology", Reccommended for Adoption at Step 4 of the ICH Process on 6 Nov 1996 by the ICH Steering Committee

This sound to me like you are finshed according to c), as long as you have acceptable linearity, precision, and specificity. By the way, the reference above also lists guidelines for drug product (formulation).

I am in process to determine standard solution stability for one of my assays. What is the practice on evaluatin criteria for the solution to be stable? Another question is, if I prep a new standard solution to quantify against, then I am adding another variability. Can some one tell me, other practices that may have been used to minimize the varability to determine standard solution stability. I am thinking of using area % (ratio of main peak response/total peak response) in the injections and compare that at different time points. Any comments?

Thanks

I am in process to determine standard solution stability for one of my assays. What is the practice on evaluatin criteria for the solution to be stable? Another question is, if I prep a new standard solution to quantify against, then I am adding another variability. Can some one tell me, other practices that may have been used to minimize the varability to determine standard solution stability. I am thinking of using area % (ratio of main peak response/total peak response) in the injections and compare that at different time points. Any comments?

Thanks

Here is how I have validated the accuracy of an analytical method using a sample of unknown purity:

First, we did a linearity study using at least five points, plotting amount injected versus peak area of the main analyte. Then, we performed a precision study by analyzing three amounts of the sample in triplicate, within the range of the linearity study. Then, we plugged the peak area data generated from the precision study into the equation of the line generated in the linearity study, and calculated the percent recovery. The calculated amount injected had to be within a certain percent of theoretical. We also used to equation of the line and the residual sum of squares to calculate the LOD and LOQ, and verified them with actual injections in that range. By expressing everything in terms of amount injected, not amount of analyte, we could show that the method was accurate. You might be able to tell the purity of your sample by looking at the intercept of your standard curve.

I think that you are forgetting aobut random errors in the analysis. No analysis will be 100.00% accurate, but will be 100% +/- some percent. You need to determine what the analysis method variation is and determine your recovery that way

A second method if always a good choice to look at purity. LC/MS has already been mentioned as a possible method. You might also try DSC. You might try "Hi-Lo" chromatography to look for and (semi-)quantitate impurities. If you have established that your chromatography yields a single peak for the analyte, coulometry will yield an absolute measure of purity.