Critical Pair Separation

[wanda50]

I have a sticky problem. I need to separate a critical pair, but have to follow USP guidelines for accepted changes to a method. Actually, I have 1 critical pair and an isomer. I was able to separate the isomers by changing the column. Original column was a Prodigy C18, 4.6 x 150 mm, 5 um at 30C. (I've tried gradients and many columns.)

MP A: 5/95 ACN/H2O, 0.1% TFA MP B: 50/50 ACN/H2O, 0.1% TFA
Gradient: 0 to 100% B over 20 min
column: Eclipse XDB C18, 4.6 x 150 mm, 3 um at 30C
Injection 10 uL. Assay concentration is 0.5 mg/mL. Diluent is MP A.

the chromatogram shows the main components. The first 3 compounds are overloaded and are related subs. I did a quick linearity, and all related subs and assay have a R2 >0.99.

Any suggestions? I think I'm having a brain cramp day.


[/img]

[Rob Burgess]

Hi Wanda,

I don't want to keep promoting the point, but have you tried modelling the chromatography (assuming you have access to something like DryLab). I would hazard a guess that this would be the quickest way to a solution.

Failing that, look to make changes that will have the biggest impact on the selectivity of the separation i.e. column and organic mobile phase changes. Also I am surprised that chnaging the gradient condiutions has not worked thus far.

PS. I don't truly understand the rationale for "acceptable" changes as given in the USP. Surely changing the column is quite a drastic change - how can that be advocated I wonder?

[wanda50]

Gradient changes with this MP system are difficult due to the %of the lesser component at any given time. It may come down to totally redoing the method.

No modeling program available. This company doesn't believe in it for plant support folks (I'm in an API company where I am Big D in R&D).

thanks for the thoughts.

[danko]

Hi Wanda,

The column temperature is one of the parameters you can try to modify (to 35 degrees?). Sometimes column temperature change can affect the selectivity hugely.

Also, the TFA concentration can affect the selectivity and the retention factor. Try for instance 0.05 % TFA and se what happens.

Last but not least, your gradient seems quite steep to me. Try 10 to 80 % B over 20 min.

Best Regards

[Uwe Neue]

What you have does not look too bad...

For the type of problem described, I would evaluate two packings with a significant difference in ligand density. For example, Symmetry C18 with a ligand density close to 3.5 micromoles/m^2 and Atlantis T3, with a ligand density around 1.5 micromoles/m^2.

[wrxtotoro]

I agree with Uwe. The chromatogram doesn't look too bad at all so the separation should be easy to manipulate.
Personally I don't have much good experience with the Eclipse XDB. Why don't you try changing to some better (IMHO) C18 column such as Xbridge, Discovery or Ascentis Express? Both Xbridge and Discovery has 3 and 5 um particles I believe. I'd perfer to change the column before I start messing with other chromatographic parameters. The USP usually just states C18 as L1 packing so as long as you are using a C18 column. Technically you are not deviating from the USP at all and you might not even need a full validation.

[wanda50]

I calculated resolution, and it is actually 1.5 - not bad. I will attempt Components 1-3 at related subs concentration rather than assay concentration. I hope the peak doesn't get retained longer.

Yeah, I went for the L1 substitution, too. Calculating exact smaller concentration component and figure out the % variability is a royal pain.

Thanks for the column suggestions. I used what I had (don't we all?). For some wierd reason our compounds do well with Zorbax columns. Don't understand it, would rather use ones that I am familiar with, but you gotta go with what works.

The reason I am so picky is that this will go to our Quality department for release testing. I have to hand them bullet proof methods since there are analysts all stages of training. There is nothing worse than having to go back and redo their chromatograph reports if they have misnamed something.

[Rob Burgess]

I'd make an stp about getting in modelling software if I were you!

For the investment, made it will save you ooodles of time and effort and will give you a much better method to transfer at the end of it.... sometimes managers with purse strings need to be banged over the head to see the light!

[danko]

Hi Wanda,

Yes, the existing resolution is acceptable, but when it comes to robustness it could cause trouble.

And now to the column – it was suggested that you should find a “better columnâ€

[wrxtotoro]

For normal method development, I'd agree that modifing chromatographic conditons (e.g. temperature, gradient or even buffer conc/pH) to improve separation is the way to go (with the shown chromatogram).
However if the task is to use an existing USP method and avoid full method validation, there's really limited modification you can induce in the method. I used to work in a generic pharmaceutical company so transferring a USP method in-house (to dodge full validation) was a daily task for me. First thing I'd try is to change to a smaller particle size to improve peak shape and resolution. However, sometimes if the USP says "150x4.6 mm column with L1 packing material at 5 µm particle size", changing to a 2.5-3 µm particles may not be so ideal anymore. (The more changes you make, the more parts you will need to validate. Brandname employee never have this headache!! ) In that case, changing column would be the easiest as a column with higher carbon loading usually gives better separation. Don't get me wrong. Eclipse XDB is a fine column for some applicaitons and I developed methods with it a number of times. But for the OP's situation, it might be easier to stuck 12 different columns (Well, with some educated guess of course. If you just pick columns by trial and error, you might really end up trying all the columns in your drawer!!) in column switchers and let 2 LC run overnight with the original method. The next morning you should have a usable method ready.

P.S. If your superior doesn't mind changing particle size, you should really try the Ascentis Express C18. (No, I don't work for Sigma or Supelco ) I just developed a method with it the other day and the peak shape and separation was just amazing when compare to some other C18. I don't have enough experience to comment on the durability but so far so good!!

[wanda50]

Thanks everyone! I can tell from the replies that I wasn't really having a brain cramp, just had hit the "developer's wall". I did indeed try to choose columns with similar carbon loads, etc. The Eclipse was not one of my choices - I was desparate! So, I was surprised!

The other part that i did forget to share is that these are bases, and finding a robust method is rather like trying to get a good souffle during a birthday party for 10 year olds. One of my dreams is to get away from all irregular non-end capped column methods, but sometimes that is what works.

thanks again for your thoughts. I am going to try TFA at 0.05% and a bit of temperature variation.

[Uwe Neue]

Wanda,

Carbon load is not the right way of looking at this. You need to look at the surface coverage in micromoles/m^2. Decent suppliers will give you this value on top of the carbon load.

As suggested by me above, if you select two columns with a significant difference in C18 surface coverage, you have a chance to be sucessful.

[wanda50]

Thanks, Uwe for setting me straight. I was trying to get out of the lab, and I guess I rushed too fast.

You are so right about the surface area vs carbon load. Selecting a column with the same carbon load absolutely does not work for our molecules.

I have one column to try and a pH experiment, and I hope to report baqck with some good news in a day or so.

Thanks!

[wrxtotoro]

I appologize for providing some misleading information above. I was always under the impression that
Retention/Selectivity ∞ Surface Coverage ∞ (Carbon Load/Surface Area)
so if the surface area of both packing materials are comparable,
Retention/Selectivity ∞ Carbon Load.

I guess I still have a lot to learn.

[wanda50]

thanks to everyone for their advice. I was definitely having a brain cramp day when I originally posted.

After pH, similar surface area columns, and temperature changes, the original chromatogram that I posted it the best choice. And, a quick and dirty linearity showed me all components are linear. So, I am ready to rock and roll.

Thanks again!