Problems with internal standards (negative polarity)

[jorainer]

Looking through the EICs for internal standards we (@SorenFjel and I) spotted some problems:

1) Different ion suppression in each matrix. Example internal standard L-Alanine (13C3, 99; 15N, 99), m/z for the [M-H]- adduct: 92.04695. We find two features for that m/z in our negative polarity data (see below). The colors: RBC (red), venous (orange), plasma (blue), capillary (purple), pool (green).

@BeppePaglia, can the difference in retention time be due to ion suppression? If that is the case, many of the differences we observe between matrices could in fact also be due to ion suppression, i.e. the matching of the peaks across samples would simply fail badly and it would look like the peak, in this case L-Alanine would be completely missing in RBC and capillary samples while the left feature in the plot above would look to be missing in all other samples.

2) The abundances for many internal standards are very different between abundances. Could be that there is something in the sample prep, but I would expect the concentration of the internal standards to be somewhat comparable between matrices. Example is the internal standard L-Cysine (13C6, 99; 15N2, 99), expected m/z ([M-H]- adduct): 247.03022, expected retention time 210 seconds.

The signal in the plasma samples is much larger than in all other matrices. Is there anything that could explain this effect?

We need here input from all of you @BeppePaglia , @cvolani, @vveri and @SiggiSmara

[vveri]

My insights:

1. I do not believe this difference in RT is due to ion suppression since chromatography (which determines the RT) occurs prior to ionization. First of all, I believe you are maybe working with a too high upper range for the second case (92.05382). Compared to the theoretical value, it gives you a 74ppm error, which is too high for a TOF and could lead to the extraction of others compounds with similar m/z (the others values you are using are in the range of 30ppm). Anyway, just something to keep in mind. Coming back to the question, in the first plates run in untargeted, @SiggiSmara and @BeppePaglia run together the same mix of labeled amino acids only diluted in ACN. I extracted here in the raw data and even with lower error you still have the same result.

This means, to me, that they are definitely 2 distinct compounds. Which one is Alanine, becomes another question. Anyway, my guess is that you maybe have ion suppression effects that do not enable the visualization of Alanine in all the sample types.

2. As far as I know, the IS Amino Acid Mix is added in the end of the sample prep, so it should be the same concentration for all (@SiggiSmara and @BeppePaglia can confirm that). Considering this is true, I believe we are again dealing with ion suppression. Plasma is the "cleanest" sample from this group and therefore would have less interference at the ionization process, making it more efficient and producing a higher signal.

[BeppePaglia]

Hi Everybody.

The abundances of some metabolite (IS in this case) can be different among matrices due to ion suppression. Ion suppression cannot influence strogly the retention time ( even though metabolites at lower abundancies can have different retention times). Consider also that low abundancies metabolites might have higher error in ppm. It can also happen that some matrices have a higher content of metabolites and this can cause overload of the column an drift in retention times. In the case of alanine, looking at the extracted ion cromatogram of @vveri https://github.com/vveri, it seems that those are two different compounds. I would guess that one is IS of alanine and the other is a fragment of internal standard of phenylalanine, but I can be wrong.

Comparing samples from different matrices can be complicated, and for sure we can have problem with allignement of lower abbundancies metabolites. My suggestion is to increase a bit the threshould and focus on metabolites with higher signal.

Hope this can help you. I am here if you have other questions.

Beppe

Il giorno gio 28 mar 2019 alle ore 18:31 vveri notifications@github.com ha scritto:

My insights:

1. I do not believe this difference in RT is due to ion suppression since chromatography (which determines the RT) occurs prior to ionization. First of all, I believe you are maybe working with a too high upper range for the second case (92.05382). Compared to the theoretical value, it gives you a 74ppm error, which is too high for a TOF and could lead to the extraction of others compounds with similar m/z (the others values you are using are in the range of 30ppm). Anyway, just something to keep in mind. Coming back to the question, in the first plates run in untargeted, @SiggiSmara https://github.com/SiggiSmara and @BeppePaglia https://github.com/BeppePaglia run together the same mix of labeled amino acids only diluted in ACN. I extracted here in the raw data and even with lower error you still have the same result.

[image: Untitled] https://user-images.githubusercontent.com/43552685/55176667-d94f6180-5181-11e9-9da5-eefdc1401778.png

This means, to me, that they are definitely 2 distinct compounds. Which one is Alanine, becomes another question. Anyway, my guess is that you maybe have ion suppression effects that do not enable the visualization of Alanine in all the sample types.

1. As far as I know, the IS Amino Acid Mix is added in the end of the sample prep, so it should be the same concentration for all ( @SiggiSmara https://github.com/SiggiSmara and @BeppePaglia https://github.com/BeppePaglia can confirm that). Considering this is true, I believe we are again dealing with ion suppression. Plasma is the "cleanest" sample from this group and therefore would have less interference at the ionization process, making it more efficient and producing a higher signal.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/EuracBiomedicalResearch/VAMS_vs_intraveinous/issues/34#issuecomment-477686070, or mute the thread https://github.com/notifications/unsubscribe-auth/Aj8mP2SP_ksLYLAAUjkN-3Z7GyHa4Apxks5vbPZVgaJpZM4cQI9q .

[jorainer]

Thanks for your feedback @vveri and @BeppePaglia !

[cvolani]

Sorry.. just coming up into my mind! Maybe a useful/important info on IS. IS were not spiked before injection into the LC-MS. IS are present in the extraction mixture, so differences between the matrices might also be due to different extraction yields (matrix-dependent) of the IS.

[jorainer]

Thanks for the input @cvolani . We might have to have a small chat on that also sometime. For us/now it simply means at present that we will definitely not use the IS for normalization.