Chromatography Forum

Hi everyone,

I'm hoping to get help on following topic:

I'm analysing pyrrolzidine alkaloids via LC-MS/MS.

I have prepared following samples:
- Solvent standard
- matrix matched standard
- sample blank
- sample fortified at 0.05 mg/kg

For some analytes (e.g. Atropine), the fortified sample has a peak area that is half is big as for the unfortified blank sample:


This doesn't make sense. I know that everything was injected properly and no mistake was made during sample preparation, because for 10 of 15 analytes, the sample is behaving as expected (showing sufficient recoveries etc.)

This is only occuring for peppermint tea.

Do you have any ideas for me?

Now you hit my field of research since i am mainly dealing with phytochemistry and am focused on pyrrolizidine alkaloids. So first mistake Atropin is of course not a pyrrolizidine alkaloid its a tropane alkaloid.....you be forgiven. Since i noticed from your picture that you are german i guess you used the published method of the Federal Risk Assessment Institute(BfR) for plant material. The method is not validated for tropane alkaloids so don't use them, sometimes the coeuleted substances can drastically effect the ionization effiency. See if you can get a matrix matched standard calibration that makes sense, but most likely you need to improve sample preparation, treat it like they treat black tea in the method.

Otherwise is it a possibility for you to use a better separation system? GC could do the trick(there is an excellent method published by Kempf) or a better lc program should deliver reproducable results.

Also check with an internal standard in your sample and replicates if your lc system might be the problem and doesn't inject it as good as you might think...

I don't have experience with these compounds in particular but I do have experience with situations like these in LC-MS/MS in complex matrices: multi-component methods where the signal for just a couple of analytes goes down when the concentration is obviously higher.

Assuming there is not a problem with your sample prep or putting a vial in the wrong position or something... It is likely due to matrix effects.

There's no easy solution. But I think it would make sense to work in a lower concentration range (dilute everything). Judging by the peak areas you can probably afford to dilute 10-100 times. Matrix effects are less severe in lower concentrations. Another thing you could try is to lower the Q1 and/or Q3 peak width. This makes the detection more selective (narrowing the m/z around the peak), but it also lowers the overall signal.

Like the poster above me said, a better chromatographic separation or sample prep (cleanup) should help. The separation between your compounds of interest might be OK, but there's a forest of peaks that you don't see (you're not measuring them) and they can affect the ionization of your compounds.

So first mistake Atropin is of course not a pyrrolizidine alkaloid its a tropane alkaloid.....you be forgiven.

- sorry, of course atropine is a tropane alkaloid

Since i noticed from your picture that you are german i guess you used the published method of the Federal Risk Assessment Institute(BfR) for plant material. The method is not validated for tropane alkaloids so don't use them, sometimes the coeuleted substances can drastically effect the ionization effiency.

- I'm using a modified method based on BfR PA-Tea

See if you can get a matrix matched standard calibration that makes sense, but most likely you need to improve sample preparation, treat it like they treat black tea in the method.

- 10 of 15 TAs show sufficient recoverys (of the same sample work up!) and indicate that sample are prepared and injected correctly

Otherwise is it a possibility for you to use a better separation system? GC could do the trick(there is an excellent method published by Kempf) or a better lc program should deliver reproducable results.

- assuming there is an coeluting interference, how do you explain that the interference influences the spiked and unspiked sample with a different impact?

Also check with an internal standard in your sample and replicates if your lc system might be the problem and doesn't inject it as good as you might think...

- internal standard shows sufficient recovery with good RSDs
- I've injected the sample 6 times in a row to check for possible carry over. RSDs are low

Assuming there is not a problem with your sample prep or putting a vial in the wrong position or something... It is likely due to matrix effects.

- see above, these kind of mistakes can be excluded

There's no easy solution. But I think it would make sense to work in a lower concentration range (dilute everything). Judging by the peak areas you can probably afford to dilute 10-100 times. Matrix effects are less severe in lower concentrations. Another thing you could try is to lower the Q1 and/or Q3 peak width. This makes the detection more selective (narrowing the m/z around the peak), but it also lowers the overall signal.

- I've already diluted samples 1:10, the effect is the same
- matrix effects were compensated
- using a different column also leads to the same efffect

The separation between your compounds of interest might be OK, but there's a forest of peaks that you don't see (you're not measuring them) and they can affect the ionization of your compounds.

- as said above: how can it come that this effects a sample with spike in a different way than a sample without spike?

- 10 of 15 TAs show sufficient recoverys (of the same sample work up!) and indicate that sample are prepared and injected correctly

- assuming there is an coeluting interference, how do you explain that the interference influences the spiked and unspiked sample with a different impact?

I assume you are using ESI as a ionization method? it is quite common that the coeulting substances interfere more with higher concentration than lower, even higher concentration of difficult substances like alkaloids can itself lead to problems with the ionization effiency. I never tested that for tropane alkaloids, but please give me the settings of your ESI and i test some parameters for atropin.

There's no easy solution. But I think it would make sense to work in a lower concentration range (dilute everything). Judging by the peak areas you can probably afford to dilute 10-100 times. Matrix effects are less severe in lower concentrations. Another thing you could try is to lower the Q1 and/or Q3 peak width. This makes the detection more selective (narrowing the m/z around the peak), but it also lowers the overall signal.

- I've already diluted samples 1:10, the effect is the same
- matrix effects were compensated
- using a different column also leads to the same efffect

The separation between your compounds of interest might be OK, but there's a forest of peaks that you don't see (you're not measuring them) and they can affect the ionization of your compounds.

- as said above: how can it come that this effects a sample with spike in a different way than a sample without spike?

1.) You will never get rid of matrix effects lowering your ion signal, even the sales rep tell you how great MS/MS is in avoiding matrix effects....the MS/MS will just filter out wrong ions coming from the matrix. The separation of those molecules/ions happens AFTER the ionization takes place in the ESI-chamber so be advised not to think that you can get rid of matrix effects. Also don't trust the method of the BfR in context of matrix effects, phytochemistry is extremely diverse and a peppermint tea matrix will even vary from batch to batch so do a proper cleanup (maybe even change to SPE-CX instead of SPE-RP).

Another possibility normally to avoid matrix effects is very easy in theory, switch to ESI- instead of ESI+, there are less molecules making negative ions then positive ones. But be advised that your Tropane alkaloids could be also be in this group. Its just a cheap trick to change the ionization settings and see if the coeulting things still happen.

Or switch to APCI, or you might even consider GC-MS. Alkaloids are nicely analyzed by a GC-MS setup, a vast variety of tropane alkaloids are already in the NIST database) and there are good methods published to use them. With a good GC setup/sample prep your lod/loq should reach the LC-MS/MS equivalent, since alkaloids are often little bitches in the ESI

Hi,

just to keep you informed. The problem was induced by inhomogenous sample material. By sieving for particle size and using only small particle sizes, all recoveries and fortifications behaved as expected.

It is likely that rhizomes of PA metabolizing plants couldn't be homogenized as properly as other plant materials and were weighed in in different amounts