Internal Standards in LC-MS and LC-MS/MS

[james little]

I have had a lot of problems getting good unlabeled internal standards (IS) in LCMS work. Mainly using to attempt to overcome instrument instability in time.

Problem is that the unlabeled IS response doesn't always drift at same rate as analyte. Labeled IS not always available.

Found an interesting discusision at

http://www.boomer.org/pkin/PK04/PK2004313.html

Anyone else have any insights?

[JA]

Luckily, I haven't had to quantify much in my time with a mass spectrometer.

I want to openly ask though, in cases where a heavy isotope is unavailable, is it general practice to apply some sort of correction factor for the differing ESI responses for equal quantities of standard and target analyte? It is evident when comparing a HILIC method to typical reverse phase conditions that analyte response can increase in the MS due to the more volatile eluent. I'm going to assume that if a gradient were used in an analysis employing MS-quantification that a response factor (and hence a small quantity of pure standard and target compound) is required. Comments?

[MG]

JA, I've never seen this done. It is assumed that for small changes in the amount of sample injected (i.e. from autosampler variability or evaporation from the sample), that if ISTD signal changes by e.g. 10%, then analyte signal will also change by 10%. Thus the ratio is the same. Where you can get into trouble with an ISTD eluting at a different time than the analyte, is if matrix ion suppression is significantly different at the different retention times.

[JA]

That's pretty much one of the factors I'm considering. Effectively, suppression of ionisation for the earlier eluting component in a RP gradient because of the water-rich eluent.

I guess what I'm really questioning is how we quantify the target analyte against the ISTD when they are fundamentally different species. Sure I might have built a calibration curve of the ISTD but how does that relate to my analyte signal? If I were to be performing this task in a HPLC-UV method I would undoubtedly be using a correction factor. This isn't the case with mass spectrometrists? My points are firstly that even if the ISTD and target analyte were separately introduced into the MS in identical eluent composition we might observe a different response due to nonidentical ionisation potentials. Do people check this in their search for a suitable ISTD? Secondly, even if the ionisation potentials were equal, they might differ in different eluent compositions, ie. two separated peaks eluting in a gradient.

I have to admit that at time of writing I haven't previously considered investigating my second point, perhaps by infusing a chosen sample prepared in differing solution composition.

I might be talking rubbish too?

[meillid]

Hi, JA,

I believe that I didn't follow you thinking very well. Why do you need to build "a calibration curve of the ISTD " ? I only build calibration curve of the ratio of ISTD and analyte.

[JA]

So, it turns out I was right to put a dum dum disclaimer at the end of my post

I know that the regular use of internal standards is to plot the ratio of analyte:IS vs [analyte] but I had posted in the context of where an amount of pure analyte is unavailable and so this couldn't be done. I take it then that this is never the case.

Continuing down the wrong road from my message of 18th Jan I then thought if pure analyte is not available one might make a calibration curve of ISTD vs [ISTD] in the actual test samples and assume the analyte responds identically to the ISTD to read off [ISTD] = [analyte]. After all the standard is typically an analogue.

[JNF]

Another approach is to use a calibration curve with spiked blanks e.g. use a placebo or blank that matches your sample and spike it with a standard. There will never be a "bullet" proof solution to your problem without using a isotopically labelled compound but using spiked blanks minimise the differences in ionisation yields you are observing.

[james little]

Sometimes I've even had trouble with using calibrators pepared from purchased blank plasma. When I added the internal standard and the standard to t=0 samples from the study, their response was 30-40% different than the purchased plasma. Tried to match the purchased plasma (type of rat, gender, anticoagulant, etc.).

Found in these cases using 25 ul of the study sample with 75 ul of the purchased plasma minimized the problem..