LCMSMS MRM LINEARITY LIMITS

[shemesh199]

Hello all,

More recently I have come across a unique small molecule compound. True the sensitivity on our QQQ is there, we still have near 3/1 S/N near 1 ng/mL. The problem is , when running a wide calibration curve (with Internal Standard of course) from say 1ng/mL , 5 , 10, 25, 50, 100, 250, 500, 1000, 2000 ng/mL I observe a loss of linearity at the low end or high end, and the curve looks very quadratic.

It is quite troublesome to see the least after being used to perfectly linear response from other compounds.

I noticed that if i process the curve from 1-100 it is perfectly linear, similarly if I process from 50-2000 it is linear.

Its almost like you need two separate curves, one for the low end and one for the high end.

My question is simple, is it appropriate to run a wide range of standards, then when running an unknown sample for example at higher concentration you can simply drop the lower standards, similarly, if the concentration is found lower, can you drop the higher standards?

Is this practice acceptable? If not, what do you guys suggest? Has anybody come across a similar situation?

I am preparing for a rodent PK study, and probably we need the sensitivity.

I have fiddled with this method quite some time, and found it very hard to increase linearity, probably one negative factor is it takes lots of FV and CE and the compound is a larger small molecule. Also the internal standard is a structural analog and not isotope.

Thanks guys!!!

[Camisotro]

Hello all,

More recently I have come across a unique small molecule compound. True the sensitivity on our QQQ is there, we still have near 3/1 S/N near 1 ng/mL. The problem is , when running a wide calibration curve (with Internal Standard of course) from say 1ng/mL , 5 , 10, 25, 50, 100, 250, 500, 1000, 2000 ng/mL I observe a loss of linearity at the low end or high end, and the curve looks very quadratic.

It is quite troublesome to see the least after being used to perfectly linear response from other compounds.

I noticed that if i process the curve from 1-100 it is perfectly linear, similarly if I process from 50-2000 it is linear.

Its almost like you need two separate curves, one for the low end and one for the high end.

My question is simple, is it appropriate to run a wide range of standards, then when running an unknown sample for example at higher concentration you can simply drop the lower standards, similarly, if the concentration is found lower, can you drop the higher standards?

Is this practice acceptable? If not, what do you guys suggest? Has anybody come across a similar situation?

I am preparing for a rodent PK study, and probably we need the sensitivity.

I have fiddled with this method quite some time, and found it very hard to increase linearity, probably one negative factor is it takes lots of FV and CE and the compound is a larger small molecule. Also the internal standard is a structural analog and not isotope.

Thanks guys!!!

You could cap your calibration curve at a place where it's still relatively linear, and simply run your samples at two different dilution levels.

Or alternately: Run two batches. One batch calibrates 1-100 and runs samples, all using your usual injection volume. The second batch calibrates 50-2000 and re-runs the same sample vials, all using 1/50th the usual injection volume.

Also worth noting, if you run your 50-2000 calibration with 1/50th the injection volume and it's still not linear, then you have some other problem going on.

[Ruth]

Are your compound and IS co-eluting? Maybe some ion suppression from the analyte on the IS or the other way round can explain what you are seeing?

If the porblem does not get solved, I would suggest to use the 1-100 ng/ml linear curve. If you have an unknown sample containing higher concentrations, dilute and rerun it.

Hope this helps!

[shemesh199]

Thanks for the quick reply, actually the compound requires quite high fragmentor voltage and collision energy for our system to gain sensitivity. I am not sure if this could be a contributing factor.

The compounds do not coelute, and the internal standard is a structural analog and not isotopically labeled.

One contributing factor might be that the analyte elutes off the column rather early near 2 minutes, i heard unretained compounds might coelute and cause ion supression , but this occurs when using the pure standard in meoh reconstituted in mobile phase itself, so i dont believe there are any matrix effects.

I guess this is a compound with a really narrow dynamic linear range as opposed to most of the other compounds.

In the meantime I will attempt to cut the injection volume on the higher standards to see if that improves.

[catlover]

I am also playing with LC/MSMS and experienced similar problem before.
My case was that peak area suddenly drops in certain concentration and increase again after that concentration, the reason is that signal is saturated in high concentration detection.
You can lower CE to remain the linearity for high concentration, that's the simplest way to solve this problem.
And I think if your calibration range is width enough, you can just omit the 1000 and 2000 ng/mL because you can dilute your sample if it is out of calibration range.

[shemesh199]

Thats a good idea.

You know at first I thought there was some weird ion supression or enhancement going on, so I played with the gradient shifting the RT drastically to another region of interest, with no effect.

I was thinking that the reason for the nonlinearity must have to do with the high fragmentor voltage actually as opposed to collision energy.

To be more detailed the agilent 6410QQQ that we have, for most small molecules FV is optimal around 110-140. In this case I feel we are pushing the source to the limit by having this setting at 190FV. The collision energy is set to around 35 which seems high but okay to me. Do you think the CE has more of an affect in this case versus the FV, also the injection volume is around 10uL, probably i could lower the CE which would lose response, but i could increase the injection volume?

I am thinking since the compound we are looking at is on the upper limit of being classified as a small molecular weight compounds (X~800D), this might be necessary in this case. And actually upon direct syringe infusion into the ion source, we do observe drastic loss of both precursor and product ion as the FV is lowered to a more acceptable range.

Does anybody have this sort of issue before?, Somehow I feel this is directly affecting the linearity.

The trend we observe, is that the linear dynamic range is shorter than usual, at higher concentrations the curve shifts upward with more detector response, at lower the opposite effect, its almost like processing two curves which are cut off somewhere in the middle. The best fit to incorporate all of calibrators must be a quadratic.

I have attached a photo, depicting levels 50, 100, 250, 500, 750, 1000, 1250, and 1500 ng/mL.




I usually do cut this curve in half and make the dilution accordingly, but I wanted to really have an idea of what is causing this phenomena.

Kind regards

[Loekie]

It looks like the compound is sticking somewhere in an absolute amount and that has relatively more effect on the bottom end of the curve than on the top end, hence the smiley curve. It could be sticking to your vials, autosampler syringe, tubing etc. It can but it does not have to be the mass spec.

you said it is a bigger small molecule. Bigger molecules usually need higher voltages and more energy to fragment. this could explain the higher voltages that you see.

Regards

[James_Ball]

I have two compounds I run Sulforaphane and Sulforaphane Glucosinolate. Running similar molar concentrations the Sulforaphane will give a nice linear curve while the Glucosinolate gives a quadratic curve. In fact Sulforaphane Glucosinolate gives a quadratic curve even if you calibrate using 3,6,12,30, and 60 ppm as the concentrations with the higher concentrations having responses that fall off from what would be linear. It could simply be that as you increase concentration you don't ionize completely, that is what seems to happen with mine.