Method Validation - Accuracy

[syx]

Dear Members,

In cases where it is impossible to obtain samples of all drug product components, we need to add known quantities of the analyte to the drug product.
Question:
How much percentage of the analyte can be added to drug product? Is it 100% analyte to 100% drug product concentration?

Best regards,
SYX

[adam]

If I understand you plan to take drug product and then spike more drug substance into it. Then subtract the unspiked result from the spiked result; and use this to calculate recovery.

This is not a great way to do it. If you spike a large amount in, you will be well out of the range that you would actually encounter. Conversely, if you spike a small amount in, you will ge high variability because you are determining a small value from the difference of 2 large values (error propogation).

Is it possible to get at least some of the excipients. It seems to me it would be better to use a synthetically prepared excipient solution even if it doesn't contain everything you are looking for.

I guess if I had to do it as you are proposing I would spike something like 20%, 30%, and 40%. Basically I am trying to balance between spiking too much and too little: for the reasons stated above.

[Mark Tracy]

What both of you are talking about is the method of standard additions. This is a well known and (sometimes) accepted procedure, but it is tedious and may not solve the problem at hand. This technique is primarily used in MS, GC or atomic absorption spectrometry because the response factors may be modulated by the presence of the matrix. This is almost never true for HPLC-UV. The problem in HPLC-UV is usually an offset due to an unresolved interference, and unfortunately, standard additions does no good for this situation. I know because I have tried.

Standard additions is not totally useless. You can use it to demonstrate linearity in the presence of the matrix.

Your best bet is to use a diode array detector and determine if the spectral match between a pure standard and your peak in matrix is sufficiently good to exclude the presence of an interference. Pay close attention to the baseline and see if the times and background at peak start and peak end agree with the pure standard.

[syx]

Your best bet is to use a diode array detector and determine if the spectral match between a pure standard and your peak in matrix is sufficiently good to exclude the presence of an interference. Pay close attention to the baseline and see if the times and background at peak start and peak end agree with the pure standard.

What if I cannot get the matrix? I only have the drug product, without clear description about the excipient inside.

[Mark Tracy]

I understand your problem. What I mean is to compare the pure standard drug against the formulated drug and look for any measurable difference in the peak.

You have three technical problems:
1- qualitatively is the peak pure even in matrix? Diode array is good here, and you need to prove this anyway.
2- is the slope of the calibration line dependent on the matrix? standard addition can help here, even with its limitations.
3- is the intercept of the calibration line dependent on the matrix? Without a genuine blank matrix, you can depend only on indirect measures. Again diode array can help.

If that is not good enough, get someone higher in your organization to get you more resources. Time on the mass spec. Persuade the sample supplier to give you a real blank matrix. Send it out to a qualitatitive analysis lab.