Standards solutions preparation

[ricopalazo]

Dear all,
I've to choose a way to do the quantification of an API during an analysis.
1. 1 unique solution dilute at three levels of concentration
2. 3 solutions at three different levels of concentration (different weight for each solution).

What way fit best to a GMP (pharmaceutical) environnement?

Basic questions are often the most important^^

Thanks.
Rico

[Blazer]

I'm not aware of any guidance document that states one way is preferred over another or that one way is forbidden, if that's what you're asking. As long as you are consistent in what path you choose, I don't see that it makes a huge difference.

[JGK]

I would choose the first way primarily for logistical reasons.

The second way requires 3 weights which may be pushing the limits of your balance unless you increase volume for the lower concentrations.

Also, and I'm speaking from a GLP perspective here, it is ususal to check the preparation of the primary "weighed" STD by preparing in duplicate and comparing the unit responses of the replicate preparations.

The first method requires only one comparison check, the second would require 3.

[Jumpshooter]

I have always written out in the analytical method to "prepare two primary standards that are 120% of the normal concentration of target analyte by weight (in duplicate, two separate weighings required); then serially dilute the primary to make a "100% level", then dilute the primary to obtain an "80% of normal level" for the active ingredient (analyte). This yields three calibration points representing 120%, 100%, and 80% normal of the analyte concentration; assay these standards and contruct a regression line; acceptance criterion is R-square greater than or equal to 0.95 (which means that greater than or equal to 95% of the variance in the peak area response is explained differences in analyte concentration--with the rest being written off as random error); then perform the same serial dilution of the duplicate preparation, inject them as "unknowns", and quantify them against the cal curve; the resultant values for the unknowns should be between 95% to 105% of the true value".

Hope this helps.

Jumpshooter