Chromatography Forum

This is a general question regarding accuracy assessment during method validation.

ICH states in Section 4.2 of the Q2 (R1) guideline that, "accuracy should be assessed on samples spiked with known amounts of impurities". I define "should" in this statement to mean "obligation or duty". A colleague of mine defines it to mean "probable/probably", and has used this position to validate impurity methods using solutions containing only the impurities - no drug substance is present. The rationale is the impurities are present in significant amounts and it is not possible to achieve the concentrations required.

I contend this is not validation as it is performed outside of the actual sample matrix, that the spiked test solutions should be corrected for the impurity amount determined using control solutions.

I can not find any clarification in the ICH documentation. How do others interpret this guideline? Any supporting documentation. What do you do when impurities are present?

I know the FDA's recent Method validation guidance document of July 2015 clearly states in lines 66-70 that "In general, FDA’s guidance documents do not establish legally enforceable responsibilities. Instead, guidances describe the Agency’s current thinking on a topic and should be viewed only as recommendations, unless specific regulatory or statutory requirements are cited. The use of the word should in Agency guidances means that something is suggested or recommended, but not required.

Thanks,
Robotjock

I would support your position.
Of course, ICH Q2 is quite vague in many terms - "should"/"might"/"could"/... This is because it's a general guideline (remember, it's a guideline, not a law) and it's supposed to be applicable to pretty much all the analytical methods out there. But this doesn't mean you're not supposed to still perform scientifically sound work! When it comes to word interpretation and nitpicking in order to support your case, something's wrong.

The basic idea of proving accuracy this way is determining recovery - "recovery" means recovery from anything that's present in your actual sample solution. That's why the best way to do it is to spike actual samples. If this is not possible - we also had situations where the available samples already contained quite high levels of the impurities, so when spiking them the concentrations were irregularly high - I'd expect you to still try to resemble the sample as close as possible. By including any matrix compound avialable. For pharmaceutical products, this usually means spiking a placebo solution which contains any excipient present in the product.
"Spiking" the solvent only, with nothing else present - that's a nice interpretation of "recovery". I'd suspect to get into real trouble if anyone with a sound knowledge of method validation checks this...

I would support your position.
Of course, ICH Q2 is quite vague in many terms - "should"/"might"/"could"/... This is because it's a general guideline (remember, it's a guideline, not a law) and it's supposed to be applicable to pretty much all the analytical methods out there. But this doesn't mean you're not supposed to still perform scientifically sound work! When it comes to word interpretation and nitpicking in order to support your case, something's wrong.

The basic idea of proving accuracy this way is determining recovery - "recovery" means recovery from anything that's present in your actual sample solution. That's why the best way to do it is to spike actual samples. If this is not possible - we also had situations where the available samples already contained quite high levels of the impurities, so when spiking them the concentrations were irregularly high - I'd expect you to still try to resemble the sample as close as possible. By including any matrix compound avialable. For pharmaceutical products, this usually means spiking a placebo solution which contains any excipient present in the product.
"Spiking" the solvent only, with nothing else present - that's a nice interpretation of "recovery". I'd suspect to get into real trouble if anyone with a sound knowledge of method validation checks this...

I agree. In the cases where your active drug substance is not possible to obtain in a very pure state (one can try additional recrystalizations for instance), you do "unspiked samples" and use the mean to correct your spiked samples.