ZIC-HILIC Columns

[chrisw]

Just a post from someone who is completely new to this type of column.

I'm trying to replicate the paper by Oertel et al determining six aminoglycosides in serum, but only for two of those entities.

I have replicated the two mobile phases in the paper and followed the gradient listed and get either really broad peaks or nothing at all.

The mobile phases consist of acetonitrile:2mM ammonium acetate:formic acid with Phase A being 5:95:0.2 and Phase B being 95:5:0.2

I'm sure my mass transitions and source settings are correct, because when I swap over to a C18 column and re-inject, I can see the aminoglycosides in the solvent front.

Any suggestions where I may have gone wrong either in preparation or equilibration would be greatly appreciated.

Chris

[HW Mueller]

I am also still learnig about HILIC and particularily Zic-HILIC regarding prolines (isocratic methods, mentioned before). There also is a tendency to get relatively broad peaks which, therefore, have a tendency to disappear or have fronting/tailing. One reason for this may be that I am apparently operating near an overload of the column due to being presently limited to UV. Thus the questions: What do you consider broad peaks here? How much of the glycosides do you inject (moles)? Have you injected different amounts under conditions where you see peaks? Have you done experiments to optimize pH, Ionic strength, organic modifyer?
There is an interesting article: Pack and Risley, LC.GC North America, 24 #8, 776(2006). But : Presently I am at odds with some of this presentation and am trying to alleviate this.

[Uwe Neue]

I hope you are running a gradient from B to A, i.e. from high acetonitrile to low acetonitrile.

[chrisw]

Firstly, thanks HW and Uwe for your replies. I'll address each of your questions starting with the easiest.

Uwe, I followed the paper exactly, with a gradient starting at 95% MeCN and bottoming out at 15% MeCN before returning to 95%. I adjusted the length of the holding periods because the column I have is exaclty the same except for the lenght which is 150 mm instead of 100 mm.

HW (Hope you're not offended by the abbreviation), the peaks are up to 3 minutes in width, and are split, with no real shape to them. I'm injecting 1000 ng/ml standard using 10 uL injection volume, so overloading doesn't appear to be an issue here. I haven't done any investigation with regards to pH or ionic strength, though I must admit I thought that 2mM was a little bit skinny. I was just trying to reproduce the assay. Any suggestions on which direction I should head?

The other problem is the requirement for the sample to be in high organic. Aminoglycosides absolutely hate acetonitrile. I put gentamicin in 50% acetonitrile with buffer and while it improved the peak shape slightly, the peak area was a lot smaller. But at this stage I just want good chromatography, and I'll worry about the solubility issue later.

[Uwe Neue]

Read your initial post carefully then you will understand my question about the solvent composition of the gradient.

You say that you have split peaks. This indicates that you are injecting the sample dissolved in a strong solvent, which is water. Try to dilute your sample with as much organic as you can, best to the composition of the starting mobile phase. This may create some solubility problems. So the sample solvent is something that one must be careful with in HILIC. One gets the same type of problems as one gets when injecting DMSO in RP.

A small amount of sample may very well the right thing to do. I used to work with sugars with this technique, so I am aware of the issues. Aminoglycosides are neither worse nor better. I thought that you detect with MS, so you should have plenty of sensitivity. I used to work successfully with this technique with a refractometer. You should have at least some 10^6 more sensitivity...

[Tobias Jonsson]

I have heard several that are having difficulties reproducing the separations in the papers you are referring to [J. Chromatogr. A 1058 (2004) 197-201 and J. Pharm. Biomed. Anal. 35 (2004) 633-638], and I believe it is due to the gradients employed, and the fact that retention in HILIC is dependent of the amount of water adsorbed to the stationary phase, which in turn depends on the concentration of water in the eluent.

In these papers there is a fast linear gradient from 5% to 86% buffer (in ACN) in 1 min and the total equilibration time (3.8 min) is shorter than the hold at final conditions (5.4 min). Thus the column will never have time to equilibrate to initial conditions, and the actual conditions in the column (including within the stationary phase!) at the time of injection will reach a “dynamicâ€

[HW Mueller]

From your newest explanation I would agree that it´s either a sample solvent incompatibility or an equilibrium problem or both. (Particularily if you go from high water and/or ion content to a lower one, equilibration seems to be slow). Does it have to be done via a gradient?
(On the HW: Thats my fault as I usually don´t sign contributions, my name is Hans)

[chrisw]

Uwe, Hans, Tobias

Thank you very much for your contributions. You have given me plenty of things to think about and try, for which I'm grateful.

I sometimes wish that publications like J of Chrom would have a laboratory of their own that would test methods before publishing to determine reproducibility, or that there is a requirement for publication that methods have completed robustness tests.

I remember the first column I tried when developing this method was an old Restek PFPP column. It worked brilliantly. Bought a new column and tried the exact conditions and it didn't work. Obviously some phase modification had occured from previous samples. Talk about delusions of adequecy.....

Has anyone else had come across this phenomena before?

[HW Mueller]

You may want to read Hung, et al, J Am Pharm Assoc, 44(1), 30 (2004, on "Deficiencies of Product Labeling Directions for the Preparation of Radiopharaceuticals". This includes difficulties with repeating the quality control with TLC, mostly. A highly simplified explanation: The phrase, "time is money".
Also, many times instructions seem to be written out of a faulty memory, some important details left out on purpose, or just sloppy work is described (also, many referees seem to review papers while whatching TV or ....?).

You have witnessed in this forum that many times it is possible to give a meningfull hint only after several back-questions (this chain not being an exception either). Many equivocacies about buffer preparation are evident here, or ~all of us have assumed some things to be self evident when they are not. This type of thing enters into publications as well.

[Mary Carson]

"I sometimes wish that publications like J of Chrom would have a laboratory of their own that would test methods before publishing to determine reproducibility, or that there is a requirement for publication that methods have completed robustness tests."

Some journals (J AOAC Int) are heading in that direction.

In the meantime, we tried HILIC for aminoglycosides a couple years back--very promising initial results. Some of our first conditions were:
Waters Atlantis HILIC 50x3mm, 0.5mL/min 90+10, ACN + 0.2 M NH4OAc, pH 4, to 20 + 80 over 5 min (later modified to only go down to 40% ACN), samples and standards injected in 90% ACN or MeOH. Detected by ESI on an ion trap. Had the best sensitivity I personally had ever seen on the instrument. Nice peaks for spectinomycin, dihydrostrep, couple of the gentamicins; not quite so nice for neomycin or paromycin (really bad tailing). BTW, anything much less than 90% initial ACN concentration meant almost no retention of spec or DHS...

I used an ion-pair SPE extraction (it's published in the afore-mentioned J. Chrom) which did leave the extracts in mostly organic. I think I was doing only 5 uL injections, too.

I think someone in the lab might have tried ZIC (I know I ordered the column), but I'm not sure of the conditions they used in order to get it to work (or if it worked...). I'm pretty sure they were not able to use the Oertel method unmodified.

[chhubert]

Hi Mary,

Did you have any successful application on aminoglycosides analysis of real samples by HILIC-ITMS? Is it robust enough c.f. traditional method using ion-pairing LC-MSMS?

[Mary Carson]

We ran some fortified fish tissue samples through the ion-pair SPE extraction (which I just realized I haven't published yet, but it is available on the USDA/FSIS website), then analyzed standards and extracts by HILIC LC-MS/MS. I won't make any claims about great quantitation (we were using an ion trap, and I think data-dependent analysis, after all), but we could confirm all residues at 0.1 ppm, and the control tissues failed to confirm. That was nice, as carryover with neomycin is nearly as bad a problem as all the other neomycin problems. Just haven't had a chance to go back a properly validate the method....only really spent 2-3 weeks working with the AGs.

However, if I were doing AGs now, and hadn't already contaminated my LC-MS/MS with TFA, HFBA, etc., I would spend some serious time evaluating HILIC to try to make it work. The perfluorinated acids do tend to kill your sensitivity in other analyses, and can be very difficult to get rid of. That being said, I believe FSIS has dedicated an ion trap to AG analysis, and has been successfully using the HFBA ion-pair method for a couple of years now to identify and confirm residues picked in micro assays.

[Mary Carson]

Chris, I just looked at your post again and realized your ionic strength is WAY TOO LOW. You probably need to bump your ammon. formate concentration up by a factor of a hundred or so.

[chrisw]

Hi Mary,

Thanks for your input, it is greatly appreciated.

The ionic strength that I used was what was quoted in the paper. I suspected that it was a little too low.

So I'm wondering if the concentration that is quoted in the paper is a typographical error? Who knows.....

Thanks for helping me clarify that point.