Hello

I took a USP method for related substances (Monograph Dexamethasone) and modified the method to analyze impurities in pharmaceutical solutions (containing isopropanole). I dilute 1 g of this solution (1 mg/g) up to 50.0 mL and inject 20 µl of this solution onto the following chromatographic system:

Phenyl-Column (5 µm, 250 mm) according to the chapter in the monograph.

Mobile phase: 380 mL Acetonitrile are mixed with 680 mL of a solution of ammonium formiate (1.32 g/L). The pH of the mixture is brought to pH 3.5 with formic acid.

Flow: 1.75 mL/min (yields in a run time of 25 minutes, which is suitable for the use in routine analysis).

Detection: UV (240 nm)

The chromatogram shows a peak directly after the main peak of the Dexamethasone. If the concentration of this compound is very small (what it should be !), the peak is swallowed after about 50 injections on this column.

The deeper the pH value, the better the separation of these two peaks.

Is a pH value of 3.5 of the mobile phase to deep for such a column ?
Or is it possible to go deeper with the pH value to force the separation ?
Has the compound ammonium formiate the special property to "destroy" RP-columns ?

Thank you very much for every answer
Florian Waser, Switzerland

What happens if you wash the column thoroughly? Does the peak reappear?
Is the buffer and phenyl column recommended for the chromatography of the matrix? We have used dexa as internal standard on C-18 columns and water/methanol mobile phase in the analysis of cortisol in plasma/serum (no buffer, as cortisol and dexamethasone do not readily ionize, it is possible, though, that we might have had less problems with the matrix if a buffer was used?).

I washed the column many times, the separation is given for two or three additional injections only.
Do you think the use of a buffer in addition to the ammonium formiate might be a solution ?

Ammonium formiate is a buffer at pH=3.5, you may go to a somewhat higher conc. for better buffering. If I understand your info correctly, washing improves the situation slightly. If this is correct the first thought is that the matrix is fouling your column and that you don´t or can´t wash it correctly. I am not familiar with phenyl column, but I would do this with a C-18 anyway.

H W Mueller,

Maybe you would do it with a C-18, but then you are not conform the USP as described above. Or can you just change from stationary phase?

If so, where are the references that prove that it is allowed?

If you are following a USP method, you are not supposed to change the packing type.

Sounds like a discussion among physicians that I witnessed. They decided to continue with a treatment that harmed patients, because it was "officially" accepted and was thus legally saver than a much better method.
Can´t you devise a working method, do the USP stuff on a few samples, and report both?

Is that all?

Prepare six times a sample solution and standard solution for a self-developed method, do the same for the USP method. And proove that the results are the same by statistical way?

Is that all what it takes to change a method? Don't you have to proove linearity, accuracy, specificity, robustness,...?

All phenyl columns are not the same. Check the USP publication on Chromatographic reagents to determine the specific brand of column used in the monograph

It is nice to do all this stuff Anon June 5 mentiones, but if you do not make sure that the method realy works . . . . why do you do it? To satisfy a law? "Accuracy" usually seems to be confused with precision. I can´t recall any "official" methods which require to give correct results. One can only be sure to approach correct results by doing more than one type of analysis.

Here I have to disagree.
In pharmaceutical industry a method should work and has to be validated. Validation should show the fitness of a method for the intended purpose. So theoretically it should work. Auditors and
registrators want to see a nice validation document and results within the specification, so thats what is required.
At least EP methods "can be assumed to be validated". That doesnt mean that it is working in a special lab under special conditions.
Accuracy and the precisions are defined in the ICH guides on validation. There are other methods described for accuracy testing (e.g. recovery)that usually work well.
So with a new method a new validation is necessary. imaging one compares a not working USP method with a well working own method. Most likely results differ by statistic or rational evaluation. How to prove that your method is better (more accurat for instance)?

How to prove...? More than one method as stated above.
I have never seen an "official" chromatography method with instructions that will assure that the method works with all possible REAL samples.
They only seem to try to make sure the apparatus functions as it should, nothing there to absolutely eliminate overlapping, for instance. Furthermore, there are some people (for instance, Kaiser of the Bad Dürkheim Chromatography Institute) that have opined that "official" methods severely hinder advancement of chem. analysis.

Hi,

For accuracy : recovery testing can be performed in a way that eliminates the need for a second method (as long as you regard weighting of standards, adding to placebo and perform sample preparation not as a alternative method).
Some official methods have system suitability criteria. If these criteria are met one could assume the method as working.
If one develops a method on its own and wants to use it in a second lab one has to do a method transfer, even if equipment is similar and all details of the method are known. If he wants to use a official method no method transfer is required, even though information on the is rudimentary.
From my point of view getting an official method working and do a validation (what you would like to do when you have to sign for results) on one hand and develop and validate a new method isn't very much difference in terms of expediture. And a new method is often likely to be faster and more robust.

How does recovery testing tell you that a method gives the correct figures for a real unknown sample? The worst case known to me is that someone thought that his HPLC method gave him the right values for one analyte until someone showed, with GC, that his supposedly homogeneous peak was composed of at least 100 different compounds.
Certainly the "official" rules seem to have blunted critical selfappraisal.

We were talking about pharmaceutical analysis. If your drug substance peak contains 100 components then you are really in trouble and analytics is one of your smaller problems.
Remember that many pharmacopoeal methods (Assay and related substances) use non-selective methods like UV or titration or TLC.
Who knows a RP-method to separate hydrocortisone, cortisone and prednisolon (and maybe prednisone) without THF as solvent?

Two substances under one peak can already represent an analytical failure. Not knowing whether this is the case is an analytical failure.

We did HPLC of hydrocortisone in plasma/serum with either three columns in series or one ultrafiltration plus two columns in series, using various H2O-MeOH mobile phases. Some other steroids were involved, probably not interesting for pharmaceutical analysis.