Impurities HPLC method-Sample prep

[Nataliagp]

Hello
I´m new in the forum.
First sorry, my english is very bad.
I have an HPLC reversed-phase method for quantified impurities for tablet dosage form, samples and standard solution are filter with nylon membrane. So I need proof that filters do not affect impurities recovery.
The samples were not spiking with impurities, becouse I does not have enough quantity of them.
Then sample (1, 2, 3 etc ml of sample) were filtered and analyzed by HPLC method plus a sample was not filtered. Results, %impurities versus ml of filtered sample, were graphicated and analyzed.
I need help with criteria to stablish minimun volume of sample that must be filtered.
I think that minimum volume is that fulfill accuracy criteria and are withing precission of the method.
What is the best statistical criteria to evaluated this results?

[grzesiek]

I do not understand what you are talking about

You want to check if filtering the sample affects impurties recovery analysing samples with no impurities present? How do you want to do that? Guess?

[Uwe Neue]

My interpretation of the question is as follows:
you have samples of various volumes that have been filtered, and you have one sample that has not been filtered.

I would simply plot the results of the filtered samples divided by the results of the unfiltered sample versus the filtration volume. These values should scatter around 1, and if there is adsorption to the filter, the results should be less than 1 for a low filtration volume. Now you create criteria for the tolerance of a departure from 1 that you are willing to live with, or that get buried in the noise of the experiment.

[Nataliagp]

Samples containing impurities from drug substance and process.

Thanks very much!!!

[grzesiek]

"and if there is adsorption to the filter, the results should be less than 1 for a low filtration volume." - why only for low volume?

[Don_Hilton]

RE question immediately above:

The effect of adsorption would show more with the low volume because the quatity of matial to be adsorbed would be low and the surface area approximately the same as with the larger volumes. Once the filter is saturated, it can not take any more from the solution.

[Uwe Neue]

Filters do not have a lot of surface area, so I assume that as more sample is pumped through the filter, more of the surface will be covered, and more will break through.

You are correct that it is possible that there is a constant loss. One would see it in the same plot.

[grzesiek]

I agree, I asked the question to clarify that if adsorption occurs one would see the loss for all volumes and so for all volumes one would see results less than 1

as for the originalquestion, Uwe gave you some good advice, you can use t test to see if the number is statistically different from 1

the question remains (as I was not able to fully grasp the core of your problem) wheter you do the filtration or not and want to validate the method or just show that there is no loss of analyte(s)

[Nataliagp]

I want to validate the method but before starts I must show that there is no loss of analyte(s) by filtration. Samples must be filtered becouse some excipients are insoluble.
@grzesiek: I hope that this answered your question.

Impurities are in low levels (0.1-0.3%), will be t student test good enough? what number of sample prep will be need for 95% confidence interval?
Thanks for all advices

[Don_Hilton]

The number of replicates and standard deviation determine the confidence interval. I would suggest that you start with 5 to 7 injections and see what you get for a standard deviation. Then look at the t-table and see how many injections you need to obtain the necessary confidence interval.

[HW Mueller]

One can think of two more experiments to check on adhesion to the filter:
1. Do a centifugation and compare the supernatant with the filtration
2. Wash the filter (after filtering the sample) with a solvent known to solvate the impurities, analyse the wash.

[grzesiek]

"will be t student test good enough?" - what do you mean by this question?

[JA]

One can think of two more experiments to check on adhesion to the filter:
1. Do a centifugation and compare the supernatant with the filtration
2. Wash the filter (after filtering the sample) with a solvent known to solvate the impurities, analyse the wash.

I can see the value in using these complementary sample prep approaches in order to validate a chosen methodology, however, is it possible that a centrifuge is more advantageous and could replace the current filtration step altogether?