Low Accuracy Results - HPLC specific

[Cindy]

We are currently validating a method and had accuracy failures at the upper end. Specifically, we are validating a method for two strengths of product (one double the other) and the accuracy failed at the 100% level for the stronger strength. We thought the issue was related to placebo interaction and/or overloading of the system, but the additional data we have gathered discounts this theory. Our current investigation is aimed at running the accuracy study from 30% of the lower strength to 120% of the stronger product (dissolution method). We again saw failures at the 100% and 120% for the stronger product, but only on one HPLC. When the same samples were run on two different HPLCs they were OK. We swapped out columns and still the failures on the one HPLC. The results for the 30% - 100% of the lower strength product look OK. We are investigating the suspect HPLC, but I was hoping that someone could shed some light on what the issue might be. Right now we are looking at the lamp and detector cell (dirty?), but are really perplexed as to why we are only having issues at the higher concentrations. Any thoughts?

[Bryan Evans]

Have you looked into maybe the lamp intensity? Perhaps you have
an older lamp on the HPLC (with the good results) - and a more intense lamp on the instrument that is failing.

What do the peak apex look like on the failing instrument?
Is there a sharp peak at the top of the more concentrated peak - or does it flatten out at the top?

[danko]

Hi Cindy,

You don’t mention the mobile phase. Did you swap it as well as the column? High absorbing mobile phase as well as an old lamp (low energy) can reduce the detector’s dynamic range.
Another very important check you should conduct is, to verify the flow-cell dimensions. Are they the same on all 3 HPLC systems?
A larger flow-cell will increase the sensitivity but will prompt you to reduce the load at the high end.

I’m assuming that the results/values that you are not satisfied with, are lower than expected.

Best Regards

[Cindy]

In response to both Bryan and Danko:

The lamp on the 'bad' HPLC is old. We ran the same experiment on the same HPLC today using a new lamp. The results are consistent with what was seen before (about 94%).

The mobile phase we are using is 50:50 0.02M Phosphate Buffer, pH 2.5: Acetonitrile. The same batch of mobile phase were used on all HPLCs so I don't anticipate it is an issue.

There is no correlation between peak shape vs concentration vs HPLC system.

I will follow-up on the cell size, but I suspect they are all the same.

Thanks for your help. Any other thoughts?

Cindy

[Uwe Neue]

Following up on the thought of low light intensity: it could be caused by dirty windows on the flow cell as well...

[DR]

Are all LCs involved the same type (brand, model etc.)?

Not all flow cells are created equal. As previously suggested, there could be dirt involved. If it isn't dirt, you may have to consider diluting the dissolution samples for the higher strength tablet samples (or lowering injection volumes).

Another question - how's the standard r² through 120% of the stronger dose's API concentration?

If the r² for standards is good but for the samples it's bad, it's time to start asking about dissolution parameters. Maybe you have not established "sink conditions" for the API and your result is correctly low because you need a larger dissolution volume (solubility issue).

To get a really accurate diagnosis of the problem, more information will be needed (LCs, disso parameters, sample prep., standard curve range & results etc).

[Consumer Products Guy]

I guess I'm wondering why one standard isn't used, with the sample preparation for the higher level simply using one-half the sample weight as used for the low standard? That's what we would do, so both types of sample solutions would contain the same analyte level. For your specific situation, I agree that it sounds like a lamp/linearity issue.

[Dan]

Cindy,

Others have covered the instrument-related issues. I suggest you also look at the accuracy validation experiment itself. DR touched upon this with the mention of sink conditions.

I have been involved in many validations for dissolution methods and the accuracy experiment is always a tough one. The question is how to spike the API and placebo materials into the media in the dissolution vessel. The new USP chapter covers this (I don't have the USP around, I think it's USP<1092>).

How did you do the experiment? Did you add the API to the vessel before or after adding the dissolution media? Either way can be a problem, especially for low solubility APIs. You can lose some of the spiked API due to non-wetting or loss by sticking to vessel walls and the shaft.

You do not even need to use the dissolution apparatus for the accuracy/recovery experiment since you are not validating the dissolution of the product or the dissolution apparatus, you are validating the analysis meethod itself. That new USP Chapter mentions this option. You can perform the experiment outside of the dissolution vessel just using the API, placebo material and the dissolution media.

Of course, I just wrote all this and then realized that it would not explain the problem occuring on the one HPLC only (unless there were different samples used). So, follow up on that flow cell and maybe the injection volume that others mentioned.

Regards,
Dan