calculating peak tailing

[ACarp]

Maybe someone can help!

I have a HPLC separation of enantiomers - not quite baseline resolved but pretty darn good. Certainly good enough to drop a line and get a good and reproducible isomeric ratio results. The "active" component is the second peak and is usually ~3:1 over the first peak. Forced degradation studies show its stability indicating (both peaks). We use this method to assay and we combine the peak areas of the two peaks for this analysis. So we use the same method for assay and isomeric ratio. All is well and the method works great.

I'm transferring this to someone who says thay require a tailing factor of < 1.25. (!!) . So I have a double peak - peak shape looks good on the front end of the first peak and the tail end of the second peak. But theres enough overlap to make it difficult to do the measurement at 5% peak height. But the peak shapes look really good. I'm forced to do manual calculations becasue I don't have access to sys suit software. But the eventual testing lab will.

Assuming that we have done everything we can to separate the two peaks, any ideas on how to demonstrate adequate peak tailing fot two non-resolved well shaped peaks? Will software take care of this?

Thanks!

[tom jupille]

First off, where on earth did they get 1.25 as a criterion for tailing? The FDA's "guidance for reviewers" suggests that TF should be less than 2 (which is, admittedly, pretty bad!).

Second, one of the problems with tailing factor (or the ASTM asymmetry factor, for that matter) is that it is essentially a one-point measurement taken on a complex distribution. There is no way to relate the value at 5% to the value at any other % without making some assumptions about the exact form of the distribution (sum of Gaussians will give different results from exponentially-modified Gaussian will give different results from any othe distribution function you choose.

So:your colleague (client?) is asking for a value that has no regulatory precedent, and you're summing the areas anyway (so that resolution between them is really a non-issue), and you're going to use a manual calculation anyway. If this were my problem I would simply use the fron half of the first peak and the back half of the second peak -- and cross your fingers, because 1.25 is a hard target to hit!

[grzesiek]

"I'm transferring this to someone who says thay require a tailing factor of < 1.25. (!!) ." - so he's a moron, what can you do?

anyway, what was this peak tailing criterion developed for anyway? - to get accurate and reproducible results, that's all, if you can get accurate and reproducibile results with tailing over 3, then for me you are ok

as Tom correctly pointed, guidance calls for less than 2, and you know what? it is still guidance not a value set in stone

[danko]

In my humble opinion, the tailing factor is not as useless as it seams. And tailing of 2 may be just guidance, but it’s there for a reason. Apart from effects such as unspecific interactions etc. there is the risk of not separated peak/s causing this tailing and that is bad. Btw tailing of 2 represents quite a grim looking peak, resembling a never ending story

Best Regards

[tom jupille]

danko, I agree with you 100% about the problems caused by tailing. My objection to the tailing factor is that it is a grossly inadequate measure of tailing. It's better than nothing, but not much better! Its major virtue is that it can be measured and calculated by hand.

Just as an example, here is an overlay of the bottom half of two peaks with the same TF. These are simulations with different distribution isotherms for secondary interactions.



That said, yes, a tailing factor of 2 is a horrible looking peak. In our Method Development and Troubleshooting courses, I usually recommend 1.5 as the point at which tailing becomes a concern, with 2.0 as an "absolute" limit. Even then, if the resolution and quantitation are shown to be satisfactory (that's what validation is all about!), then I could be persuaded to accept a larger value.

[grzesiek]

"That said, yes, a tailing factor of 2 is a horrible looking peak" - true, over 2 does not look very nice, but looking nice is not the goal as was being said

"there is the risk of not separated peak/s causing this tailing and that is bad." - the risk is always danko, peak with Tf of 1.1 can't be said to be pure only depending on tailing

All that being said I have yet to develop method with Tf of more or close to 2.0

[danko]

true, over 2 does not look very nice, but looking nice is not the goal as was being said

The goal of various SST parameters is to ensure trustworthy results and thus no reference to aesthetics involved in this particular context.

the risk is always danko, peak with Tf of 1.1 can't be said to be pure only depending on tailing

I never said that it was the only indicator, but it certainly is one of them. And, in my mind, ignoring it is even worse than relying exclusively on it.

I’ve seen situations where people are replacing columns and/or altering conditions in order to make a composite peak resemble a single/pure one. Longer story…….

Best Regards

[grzesiek]

"And, in my mind, ignoring it is even worse than relying exclusively on it. " - I'm on the opposite site, ignoring is better than relying, there are better ways

[ACarp]

Thanks for the comments. And I like the front end and back end idea Tom!

I don't think you can ignore tailing, but I think that the 1.25 requirement is a bit steep for a new method. It's one thing if after experience with the method and column you routinely get 1.1 and you set 1.25 is an alert limit to tell you that your column is getting old. But to arbitrarily set 1.25?