API with 2 isomers: one above and one below LOQ

[GregK]

We've got a method where the API has a major and minor component. Recently, the API source and the reference standard source have changed their manufacturing which has resulted in a significant decrease in the minor component. (Used to be about 2-3%, and is now 0.2-0.3%.) We quantitate by LC external standard, typically by adding the two isomers' area in both the standard and sample and utilizing the sum of the area and the sum of their purities.

The question we are wrestling with is what do we do when the minor isomer is below LOQ? Specifically when it is below LOQ in the reference standard. Do we include the area that is below LOQ? Do we include the purity of the minor isomer in the calculation of the standard concentration if we don't include the area?

Apologies if this is confusing... Feel free to ask questions and I'll try my best to answer. Thanks in advance.

[DR]

If it's below the Limit Of Quantitation, you do not quantitate it. (read that first bit aloud - should reduce the tendency to wrestle a lot).

At this point, whether you choose to rewrite your sample and standard preps to get this new and improved API to spit out >LOQ areas for the minor isomer is a function of how much sleep your QA department is losing over it and your appetite for revalidation of method(s).

So long as your customers are seeing a net increase in the result and are not that interested in the isomeric purity of the API, they probably will not care so long as they are aware of the changes in the API mfg process (which they should be as it's a DMF listed item and all).

[GregK]

Right, I get it, if it's below LOQ, I can't quantitate it . What about for a system suitability requirement? If there's a resolution requirement between the two peaks, would it generally be acceptable to use a peak below LOQ (but well above LOD) to calculate R?

The other concern is when the standard's minor isomer is below LOQ, but the sample's is above LOQ. Do we use the combined purity to determine the standard concentration or just use the major isomer?

[DR]

As long as the CDS can integrate a second peak with some consistency, it's OK to have resolution for part of SST results.

RE: calculations - if your standard solution's secondary isomer is <LOQ, you cannot use that to calculate sample results.
Since it's a known, you can use the main isomer peak to calculate concentration of both isomers, assuming you've documented any difference in response factor that may exist (which depends on type of isomer).
Of course, this statement assumes that you are not tied to a compendial method that has not yet been updated to reflect the new manufacturing process.
Something similar to your situation happened when Mallinckrodt cleaned up their Levorphanol synthesis procedure years ago. They improved it so much that they added a new code to their catalog, started sending the USP much cleaner material for RS use and really confounded the (very small) market for the stuff. There was a lot of "where did this peak go? Is my LC broken?" and such.

So, if you're in that boat, come up with a cogent explanation for your change in calculation strategy, re-calculate some of your old runs with the new method to demonstrate that you get the same results, write a nice report and move along. The side benefit of this is that if your 2 isomers are equivalent with respect to efficacy, you can base your purity on the sum of the isomers' areas (assuming they have the same response factor).

If I'm off track with any of this, please provide more information (like whether you're in pharma, whether you're using a compendial method...).

[GregK]

DR - I appreciate the thoughtful and thorough response. I'm never quite sure how much detail to share, as I don't want to get in any legal hot water about company secrets or something. To provide a few more details, I am in pharma, this is a USP material, but it is our finished product method that we're concerned about. The API is ivermectin. The situation we're in sounds a lot like the Levorphanol situation you describe, only the USP reference standard hasn't "caught up" with the newer manufacturing. The secondary standard we use (from Sigma) has, and the incoming API has, so we're juggling standards that are below LOQ with stability that is above and new product that is below.

(Edited to add):
I've done some back of the envelope calculations and I'm not seeing much difference in the assay result, so I think it doesn't matter too much for that aspect. We're just trying to convince our QA folks that we can calculate the required resolution using a minor component that is below LOQ.

[DR]

Resolution is a somewhat colligative property (only in the sense that you can overload the column and effectively reduce resolution), but as long as you can consistently integrate the small peak, there should be no issue with figuring resolution as it is a function of peak widths, not areas.
Lower amounts of the minor isomer will probably improve your resolution!

If your QA department is still clutching pearls over the improvements in your API's purity, they should ask for a repeat of the intermediate precision portion of your original method validation, using the new API.
I think that would be reasonable given that the method was originally validated for a different isomer/impurity profile.

[vmu]

I've done some back of the envelope calculations and I'm not seeing much difference in the assay result, so I think it doesn't matter too much for that aspect. We're just trying to convince our QA folks that we can calculate the required resolution using a minor component that is below LOQ.

The two components under discussion are not isomers. They are homologues. The only reason to strive to separate them and to have their resolution as a part of SST is the necessity to control their ratio as described in USP (in other words, to control the fraction of the minor component; this implies that the check of the signal-to-noise ratio must be a part of SST as well). The standard to be used for SST must be suitable for this purpose. It must contain sufficient amounts of the substances needed for the resolution check. Your current standard contains insufficient amount of the minor component rendering the standard unsuitable.

This questions the competence of the laboratory choosing the standard, the competence of the USP authors who don't specify the disregard limit for the minor component in their method, and the competence of the standard manufacturer that does not take into account all the aspects of the intended use of the standard. You may try to convince your QA dept. to calculate Rs with the minor component below LOQ and above LOD. However, when you come across with a new batch of the standard with the minor component below your LOD, you will have to search for another standard.

Another option will be to try to lower your method's LOQ and LOD, e.g. by some minor changes such as decreasing the data acquisition rate or using VWD instead of DAD (both variants lower the noise). You can also increase the injection volume; the current USP chapter 621 Chromatography permits this (if linearity and the other SST are still satisfactory).