HPLC assay validation accuracy question

[mengraul7]

We are trying to validate a method for excipient in the product. The placebo matrix and diluent showed no interference with the peak.

However, everytime when we perform the validation at the 50%, 100%, and 150% level. The 100% is always right on target, having a recovery ~100%. But for 50% level, the %recovery is about 108%, for 150%, is 95% or less. The specification is 95.0% - 105.0%. We tried different injection volumes and concentrations, the results are the same.

What can be the possible cause of this issue?

[mattmullaney]

Hi mengraul7,

Based on these data you list, the instrument itself seems to behave well with regard to injecting samples as the response seems fairly linear over this concentration range.

Please let me make sure I understand properly; there is an apparent bias in the linearity data (slope in particular) obtained with matrix present relative to standards injected, say, in solvent alone, at the same concentration levels?

[HPLC chemist]

It almost seems you have a 'poor' extraction procedure assuming you have a 'decent' specificity and linearity.

[Consumer Products Guy]

...we perform the validation at the 50%, 100%, and 150% level.

My employer only validated assays for actives. That said: we typically validated at the 80%, 100%, and 120% levels.

[mattmullaney]

Hi again, mengraul7,

For the Accuracy study, how does a plot of detector response vs. unit concentration (or unit amount) look--does it appear as a fairly straight, horizontal line?

[mengraul7]

Hi mengraul7,

Based on these data you list, the instrument itself seems to behave well with regard to injecting samples as the response seems fairly linear over this concentration range.

Please let me make sure I understand properly; there is an apparent bias in the linearity data (slope in particular) obtained with matrix present relative to standards injected, say, in solvent alone, at the same concentration levels?

not sure if there is a bias for the matrix. We tried to spike the standards with the placebo as well, it showed the same recovery issue.

[mengraul7]

It almost seems you have a 'poor' extraction procedure assuming you have a 'decent' specificity and linearity.

But there is no extraction for this method. samples were injected directly.

[mengraul7]

Hi again, mengraul7,

For the Accuracy study, how does a plot of detector response vs. unit concentration (or unit amount) look--does it appear as a fairly straight, horizontal line?

yes, it is.

[mengraul7]

...we perform the validation at the 50%, 100%, and 150% level.

My employer only validated assays for actives. That said: we typically validated at the 80%, 100%, and 120% levels.

this inactive ingredients serve as preservatives, which required monitoring throughout stability. And the specification is 70% - 110%, so we are trying to cover a broader range.

[mattmullaney]

Hi Again, mengraul7,

Okay...the linearity of the Accuracy samples as preservatives + placebo (as well as the finished product?) seems to be well (R^2 ~ 1.000), check.

Also, a plot of analytical response vs. concentration (or amount) of each preservative affords a nearly horizontal line, check.

You successfully established the specificity of the preservatives in a spiked matrix (and the finished product?) and Diluent, check. How many different sources of placebo +matrix were evaluated (normally I've seen five independent preparations for these)?

I'm afraid I'm missing something, to me all looks well as per Huber:

https://www.agilent.com/cs/library/prim ... 5140EN.pdf

Are the Accuracy values you report as the 50% and 150% levels in matrix vs. for each preservative in clean solvent?

I think the instrument(s) is/are fine based on what you've done...next would be to consider how the spiked matrix samples were prepared.

[Kenn]

We are trying to validate a method for excipient in the product. The placebo matrix and diluent showed no interference with the peak.

However, everytime when we perform the validation at the 50%, 100%, and 150% level. The 100% is always right on target, having a recovery ~100%. But for 50% level, the %recovery is about 108%, for 150%, is 95% or less. The specification is 95.0% - 105.0%. We tried different injection volumes and concentrations, the results are the same.

What can be the possible cause of this issue?

Is your product basically an aqueous solution? Do you know the water solubility of this exipient? You say at the 150% level recoveries are 95% or less, do you sometimes get markedly less than that?

If samples are aqueous can you fortify them and make the next step in the method be "add x mL methanol to all samples"? Make that x 20% of sample vol then put 20% methanol in your standards?

It also sounds like the problem could be technique because you sound pretty close. I suppose you have prepared a fresh set of standards and diluted all samples regardless of concentration to the same spot on the curve you are getting 100% and run it again?