Specificity in Validation of a Limit Test

[Noser222]

I'm working on a method validation of a reverse phase/ion-pair method method of analysis of EDTA in a natural carbohydrate polymer. The method converts the EDTA to the iron(iii) form and then it's a simple isocratic run with a water/methol with acetate buffer at pH 4 with tetrabutylammonium bromide.

Our specification for EDTA is < 0.1 %(w/w) and our three batches showed << 0.01%, so I'm revising the method to be a limit test where we run samples and compare to a standard prepared at 0.01% and report < 0.01%(w/w) as the concentration. Based on the ICH guideline, the table says to evaluate specificity and detection limit. Detection limit is straightforward.

For specificity, I don't have a true blank sample, where EDTA is undetectable, rather it is present in all samples. I can show resolution > 2 from another peak in the chromatogram. I can also show by spiking that there are no interferences that would cause a result of > 0.1% or even > 0.01%. Is peak purity necessary?

Thanks for any replies.

[Noser222]

bump

[Bruce Hamilton]

The suggested use of peak purity isn't mandatory, it's merely offered as a way of helping confirm the peak is only the analyte.

What is required is demonstration that the test is specific for EDTA, and is not subject to interference from other components, impurities, related substances, or degradation products.

If you spike with a small amount of EDTA, and the peak spectrum doesn't change and peak purty is OK, you still have to demonstrate that, of the chemicals likely to be present, only EDTA will give that spectrum at that retention time. Peak purity is of very limited utility if peaks with similar UV properties exactly co-elute.

The problem may that if you always have a peak at that retention time in your product, how can you confirm that the peak is EDTA?. I would perform the spiking above, and also look to another method to measure EDTA ( could still be HPLC, but a different complex ), and show that the results correlate with quantitation of your peak by your method.

Please keep having fun,

Bruce Hamilton