Question on MS/MS techniques

[Rndirk]

MS/MS can depend on how you have the experiment setup. Does it always perform the analysis in mass order? If you tell it to perform a transition of 357/150, then 200/57, then 400/310, you will always see the results in that order since it will send only one mass at a time through the first quad, into the cell then through the second quad to the detector. Then it will process the second set and then the third. If you make the dwell 1 second, then all of those ion pairs will arrive at the detector one second apart which is 0.0167 minutes difference. If you scan for 20 ion pairs with such settings then the first pair will elute 0.33 minutes before the last pair, if you use 100ms dwell then it will be 0.033 minutes separation, so it will definitely not hold true for MS/MS systems.

I don't get why the MS would separate anything in the dimension of retention time. I was under the impression it measures the signal of a transition of a period in time. It puts this point on the retention time axis. (whether it's the time it enters the MS or it hits the detector). Depending on dwell it switches to another transition and measures a point. This process doesn't make one thing elute before another one right?

Or in other words. Suppose 2 compounds have different transitions but elute at exactly the same time = enter the MS at the same time. They will be separated by the MS but not along a retention time dimension.

EDIT: I saw the light => the peaks do not get separated but it looks like it because it can miss the real maximum of the peak if the dwell time is too long compared to the peak width. Excuse me for my tunnel vision in this topic!

[lmh]

so, as I see it, there are four very different things going on:

(1) spectra can be biased because the overall intensity of ions entering the system is changing (because we're in the up-slope or down-slope of a chromatographic peak) during the time that the instrument takes to carry out a scan. This is spectral skewing.

Given that we know how fast the instrument scans, and how fast the intensity is changing, we could, if we wanted, make an attempt at compensating for it. In general you'd look at the average across a peak, or the spectrum at the peak, either of which compensates.

If you're distinguishing compounds that differ only quantitatively in their qualifier ions, and the qualifier ions are sufficiently different in mass, and the scan is slow enough, and the difference in fragmentation between the things you need to distinguish is very small, and you collect your spectrum from the extreme up-slope or down-slope, then spectral skewing will cause you problems. Otherwise it's a non-event.

(2) spectra from deuterated products can show different relative intensities of peaks for their fragments when compared to non-deuterated products, because the extra mass can influence the rate of reaction. Isotopic effects of this sort are something we only discuss in hushed voices when we think no one is listening, as they rather undermine the concept of the internal standard (which we want to be chemically identical to the analyte). We really want to believe that this is a non-event. In practice it doesn't matter if the internal standard favours a particular fragmentation because the same thing should happen in standards and samples, so it really is a non-event.

(3) reported retention times might be biased because the instrument is looking for different masses at different times. The nature of the bias depends entirely on how the instrument manufacturer treats reporting of time. If they report, for a given mass, the actual time that the quadrupole was filtering that mass, then there is no systematic bias in retention times (you've moved the sampling-points left and right along the curve, according to the scan speed and direction, but the points are still reported correctly). If (more likely) the system returns spectra associated with a single time-point, then there is a bias, whose maximum size depends on the scan-time (the actual data were collected left or right of the reported sampling-point). If you have two compounds that differ in retention time by less than the time-difference between two successive scans, then you are in any case completely doomed. Unless you have inadequate sampling-rates, this is a non-event.

(4) genuine retention times of deuterated standards can differ, very very slightly, from those of the undeuterated analyte. The difference is usually tiny, and will only mess up your internal standard calculation if, simultaneously, an interfering compound is changing dramatically its intensity, so this is generally a non-event.

There is one thing that definitely isn't going on: we are not, in a quadrupole instrument, seeing any separation of masses based on their different time of flight through the system. Ions pass through mass spec systems really fast compared to chromatographic time-scales.

[James_Ball]

so, as I see it, there are four very different things going on:

(1) spectra can be biased because the overall intensity of ions entering the system is changing (because we're in the up-slope or down-slope of a chromatographic peak) during the time that the instrument takes to carry out a scan. This is spectral skewing.

Given that we know how fast the instrument scans, and how fast the intensity is changing, we could, if we wanted, make an attempt at compensating for it. In general you'd look at the average across a peak, or the spectrum at the peak, either of which compensates.

If you're distinguishing compounds that differ only quantitatively in their qualifier ions, and the qualifier ions are sufficiently different in mass, and the scan is slow enough, and the difference in fragmentation between the things you need to distinguish is very small, and you collect your spectrum from the extreme up-slope or down-slope, then spectral skewing will cause you problems. Otherwise it's a non-event.

(2) spectra from deuterated products can show different relative intensities of peaks for their fragments when compared to non-deuterated products, because the extra mass can influence the rate of reaction. Isotopic effects of this sort are something we only discuss in hushed voices when we think no one is listening, as they rather undermine the concept of the internal standard (which we want to be chemically identical to the analyte). We really want to believe that this is a non-event. In practice it doesn't matter if the internal standard favours a particular fragmentation because the same thing should happen in standards and samples, so it really is a non-event.

(3) reported retention times might be biased because the instrument is looking for different masses at different times. The nature of the bias depends entirely on how the instrument manufacturer treats reporting of time. If they report, for a given mass, the actual time that the quadrupole was filtering that mass, then there is no systematic bias in retention times (you've moved the sampling-points left and right along the curve, according to the scan speed and direction, but the points are still reported correctly). If (more likely) the system returns spectra associated with a single time-point, then there is a bias, whose maximum size depends on the scan-time (the actual data were collected left or right of the reported sampling-point). If you have two compounds that differ in retention time by less than the time-difference between two successive scans, then you are in any case completely doomed. Unless you have inadequate sampling-rates, this is a non-event.

(4) genuine retention times of deuterated standards can differ, very very slightly, from those of the undeuterated analyte. The difference is usually tiny, and will only mess up your internal standard calculation if, simultaneously, an interfering compound is changing dramatically its intensity, so this is generally a non-event.

There is one thing that definitely isn't going on: we are not, in a quadrupole instrument, seeing any separation of masses based on their different time of flight through the system. Ions pass through mass spec systems really fast compared to chromatographic time-scales.

Well explained. The conclusion was most of what my point was about, any retention time difference that appears on the chromatogram is due to chromatographic differences not spectral.

With how most instruments work, they will scan the mass range requested multiple times, average those scans, then save that as a single scan to the chromatogram. This is what the instrument manufacturers do to try and smooth out the data before it is reported. If you set the scan speed so fast that you get single passes for each scan then you zoom in on the peak, either TIC or EIC, the peak will look as fuzzy as a piece of Velcro. There is so much data manipulation going on in the background to make the results look presentable it is no wonder we see slight deviations in retention time for the maxima of each mass.

[MAForensics]

Definitely learning a lot here.

Unfortunately an infusion is going to be tough to get due to the time constraints; our instruments are pretty much under constant use. But it's a possibility.

A different detector isn't going to happen, we don't have the necessary equipment or the time available.

My previous understanding was that a non-deuterated drug could never appear before it's deuterated analogue due to the mass difference in the quadrupole. I see this is not entirely accurate (though I stand by the idea that a higher m/z will reach the detector first in a quadrupole MS), so I'll be reading more in spectral skew and the isotope effect.

Essentially all of this comes about because we had an instance where a false positive occurred for 6-acetylmorphine. The chromatography/response was poor, but it quantitated above our LOD and the monitored transition was in the correct ratio. However, the drug eluted before the internal standard, which to our understanding was not possible even though we were (apparently) wrong in our reasoning. As it turns out, naloxone shares similar transitions to 6-acetylmorphine, but elutes before 6AM.

This issue was caught and corrected for, but given that heroin abusers are frequently given Narcan (naloxone) to combat overdose, it poses a significant issue for us and thus we increased the attention to RT.

On the GC/MS in SIM mode we have found it to be a generally acceptable truth that the drug will always elute subsequent to it's deuterated internal standard, so we can use the RRT fairly reliably. But with our other LC/MS/MS assays we are seeing greater variability in RT.

This requires us to examine the XIC in any case in which the non-deuterated ions show an RT less than that of the deuterated ions, which is time consuming. Hence me trying to understand exactly why we are seeing this happening, and if it's anything to be concerned about.

Really, it's all about being lazy.

[mhr311]

Common in HRGC-HRMS analysis for Dioxin. PCBs for example. The labelled (in this case 13C) standards always elute slightly quicker then the native species. On sector instruments at 8KV energy, with dynodes set close to 20KV, residence time is slight. The analysis would be in question if the unlabelled species eluted earlier then the labelled compound.

[MMJ88]

My previous understanding was that a non-deuterated drug could never appear before it's deuterated analogue due to the mass difference in the quadrupole. I see this is not entirely accurate (though I stand by the idea that a higher m/z will reach the detector first in a quadrupole MS), so I'll be reading more in spectral skew and the isotope effect.

LMH is correct in that even if this effect is possible, it would be on a time scale much removed from that of chromatographic retention times.

Moving on then, it is possible that you are seeing actual differences in chromatography, but the most likely explanation is simply the dwell/scan settings of the detector. Luckily this can be easily checked, I just did it with some data I obtained this morning. When I have multiple transitions being monitored, my software (Masshunter, forgot which revision) scans them from high to low, and if you zoom in enough on the data to see the individual samples this is quite apparent. See below:



If you recreate the same thing with your data, you may see that is the cause for the RT difference. I guess if you zoom right in to this level on top of your analyte and IS peaks, and there are multiple samples between the maxima of each, you can write this effect off.

But with our other LC/MS/MS assays we are seeing greater variability in RT.

'tis the nature of the beast in LC (vs. GC), in my experience.

[Rndirk]

But with our other LC/MS/MS assays we are seeing greater variability in RT.

'tis the nature of the beast in LC (vs. GC), in my experience.

Seconded!

MAForensics maybe it's an idea to incorporate 6-acetylmorphine in your calibration, or to run it every sequence between your samples. While exact retention times can shift more in LC compared to GC, the elution order is more stable. But in this particular case I would be tempted to have a conformation.

Does it also shares the ion ratios?

[Peter Apps]

My previous understanding was that a non-deuterated drug could never appear before it's deuterated analogue due to the mass difference in the quadrupole. I see this is not entirely accurate (though I stand by the idea that a higher m/z will reach the detector first in a quadrupole MS), so I'll be reading more in spectral skew and the isotope effect.

LMH is correct in that even if this effect is possible, it would be on a time scale much removed from that of chromatographic retention times.

Moving on then, it is possible that you are seeing actual differences in chromatography, but the most likely explanation is simply the dwell/scan settings of the detector. Luckily this can be easily checked, I just did it with some data I obtained this morning. When I have multiple transitions being monitored, my software (Masshunter, forgot which revision) scans them from high to low, and if you zoom in enough on the data to see the individual samples this is quite apparent. See below:



If you recreate the same thing with your data, you may see that is the cause for the RT difference. I guess if you zoom right in to this level on top of your analyte and IS peaks, and there are multiple samples between the maxima of each, you can write this effect off.

Thanks MMJ88, that was what I was trying to explain. A couple of pictures are worth at least a thousand words. It is also important to note that the time differences (0.005 min) are not on a chromatographic scale.

Peter

[James_Ball]

My previous understanding was that a non-deuterated drug could never appear before it's deuterated analogue due to the mass difference in the quadrupole. I see this is not entirely accurate (though I stand by the idea that a higher m/z will reach the detector first in a quadrupole MS), so I'll be reading more in spectral skew and the isotope effect.

LMH is correct in that even if this effect is possible, it would be on a time scale much removed from that of chromatographic retention times.

Moving on then, it is possible that you are seeing actual differences in chromatography, but the most likely explanation is simply the dwell/scan settings of the detector. Luckily this can be easily checked, I just did it with some data I obtained this morning. When I have multiple transitions being monitored, my software (Masshunter, forgot which revision) scans them from high to low, and if you zoom in enough on the data to see the individual samples this is quite apparent. See below:



If you recreate the same thing with your data, you may see that is the cause for the RT difference. I guess if you zoom right in to this level on top of your analyte and IS peaks, and there are multiple samples between the maxima of each, you can write this effect off.

Thanks MMJ88, that was what I was trying to explain. A couple of pictures are worth at least a thousand words. It is also important to note that the time differences (0.005 min) are not on a chromatographic scale.

Peter

This is why in a recent EPA group meeting on new criteria for SIM analysis when applied to current methods, it was decided that to call a hit a positive, the Quant ion and two qualifiers have to have peak maxima within a three scan window. This is to allow for the fact that the instruments are not stable enough to generate all masses with the exact same ratio of abundance on each scan they take. Easy to see this when you put a single quad into manual tune and monitor PFTBA in scan mode. The ratio of m/z69 set as 100%, will generate a slightly different ratio of m/z 219 and m/z 502 with each scan.

[Peter Apps]

Thanks James - when the regulators start to take notice it must be a real thing. Usually a real thing that the workers have known about for at least a decade !

Peter

[carlo.annaratone]

MS/MS has a dwell time of tes of milliseconds, definitely not enough for skewing peak intensities unless one is aquiring tens of functions at a time

[Peter Apps]

MS/MS has a dwell time of tes of milliseconds, definitely not enough for skewing peak intensities unless one is aquiring tens of functions at a time

It's not the dwell time that causes skew, but the times between when the intensities of different ions are measured.

Peter

[James_Ball]

MS/MS has a dwell time of tes of milliseconds, definitely not enough for skewing peak intensities unless one is aquiring tens of functions at a time

One MRM transition would show no skew, but 100 MRMs at 10msec dwell would give one second between the time of the first scan and the beginning of the second scan (same for 10 MRMs with 100msec dwell). If the column is producing a sharp enough peak then the first mass transition would be read on the up slope, the middle mass transition measured at the top of the peak and the last mass transition would be read on the down slope. This would bias the transition read at the middle of the peak high relative to the ones on either side. The only way to reduce the bias would be to have the instrument scanning about 5-10 replicates per second and have very broad peaks that are 10-15 seconds wide.

Both super fast replicates and super wide peaks are not optimal so we just live with the bias.