Validation of a non native RPC method

[miro2009]

Hello chromatographers,
We have an unusual case where we are quantifying a certain analyte by a non native RPC method, ie. our component of interest is unfortunately very sensitive and the RPC methods completely dissociates it to 2 subunits that appear as 2 distinct peaks, one having a sharp symmetric profile easy to integrate. So we constructed a cal. curve using a standard, that also shows same behaviour, and use the sharp peak for concentration assignment of the cal. curve levels, and accordingly, we only integrate this peak for our unknown samples for quantification. We have gained trust in our method by comparing the results to that of a native validated size exclusion HPLC method and found them very comparable.

As we are forced to use RPC method at certain stages of the production for its better separation of impurities than SE-HPLC, we were asked to validate it. The method passes linearity, precicion and accuracy, but the question, is it acceptable to validate a method that breaks down the main analyte? is there any literature to discuss such specific issues, as ICH guidelines are very generalized?

Thank you

[freemab222]

Are your sure that the component is actually being broken down? Do you have some chemical rational for this belief? If so, then you may be entirely justified in your approach. But it is not wise to assume a breakdown just from two peaks.

Another possibility is that your analyte can exist in two forms in the mobile phase. One example of this is an acid that can exist as the free acid or as a salt. LC systems for such separations typically utilize buffers, but if your buffer is too weak (or absent), you can get the two forms coexisting, giving two peaks.

When the interconversion between forms is slow, compared to the time on the column, you'll get distinct peaks. On a nonpolar column, you can expect a nonpolar form (such as the free acid, in my example) to give a fairly symmetrical peak (assuming a good column without free silanols). However, a polar compound, like a salt, may elute very early and with considerable tailing. (Lots of other possibilities exist.)

When the interconversion between forms is fast, you may get only one peak, or you may get two peaks connected by a "saddle", which corresponds to those molecules that changed form during the chromatogram. Since you don't mention such a symptom, I note this mainly for completeness.

The above symptoms are not unique to acid-salt (and similar) situations. Indeed, an actual break-down can give similar results.

In any event, it is highly questionable to quantify on only one of two peaks unless you are certain of a stoichiometric relationship between peak and analyte, when both are "known" to be due to your sample. How can you be certain that the two observed peaks are always in proportion? You should at least attempt to quantify the second peak as well, taking into account that an asymmetric peak may have a calibration line that intersects the concentration axis at a positive value.