Spiking samples gets overestimated! MS suppression/enhanc?

[Anna01]

Hello all!

I have ran into difficulties with one of my methods, and need some help to figure out what's going on. Simply speaking, when spiking biological samples and calculating the recovery, I find 120-130 % - not 100% as I expected.

Everything else looks fine. Excellent precision and linearity of standards. Linearity of spiked biological samples quite perfect. Separation from known interfering substances is good and adequate. Peak shape is symmetrical, a little narrow perhaps (12s), but the MS acquire at least 16-18 measurements during this time.

The standard curve samples are prepared in a pseudomatrix (charcoal-stripped) - although great care has been taken to remove all charcoal.
The internal standard is a deuterated isotope of the steroid hormone I aim to quantitate.

Samples are prepared by simple dilution in H2O, before it's injected onto an Oasis HLB trap column. After a washing procedure with MeOH:H2O, it's eluted by an ACN:H2O gradient to an C18 column.

Now...the obvious source of errors is the standard curve, but preliminary tests indicate this is ok.

So...my thought are now on a the MS. I monitor two fragments in MRM, both show the same "overestimated recovery". The IS share one daughter ion with the analyte, but mother ion are 4D heavier - and no cross talk occurs.

Is it still possible that some kind of ion enhancement may occur between Q1 and Q3? Could there be something in biological samples that either:
1. increase the ionization of the mother ion of my analyte compared to IS?
2. increase both fragments of my analyte compared to the fragment of IS?

Is this even theoretically possible? Have anyone experienced such phenomena? Should I just forget about this "wild" though, and start over again troubleshooting my standard samples?

Thankful for advice!

Anna
PhD student

[jh1]

Yes, this is a common occurrence with complex matrices, although MS systems tend to suffer less matrix effects than other GC detectors. Try using actual matrix blank created using the same extraction procedure you use for your samples rather than psuedomatrix blank in your standards and see if it fixes your recovery issues.

[yangz00g]

More details needed:

1) the spike value near the center point of your calibration curve?
2) Is your calibration weighted, e.g. 1/x? it's very important in trace biological analysis
3) how is your IS signal in real (decreased?) or fortified matrix?

Remove the highest one or two or lowest standard point (s) to see if there is any big change

Enhancement is not uncommon in biological matrix, but certainly ion suppression dominates.

[Anna01]

To jh1:
I have actually tried preparing my standards in both H2O and MeOH:H2O (1:1). This results in some improvement, but recovery is still about 115-120%.



To yangz00g:
1) the spike value near the center point of your calibration curve?
--> Well, on the low side since almost all real samples are in this level. The intensity equals about 10% of the highest calibration point.

2) Is your calibration weighted, e.g. 1/x? it's very important in trace biological analysis
--> Yes, I apply 1/x weightening (which is a little better than not).

3) how is your IS signal in real (decreased?) or fortified matrix?
--> My IS signal is about 70% in real biological samples compared to prepared in H2O or MeOH:H2O. Is 70% indicating a major problem?
No fortification is involved. The biological matrix is simply diluted with IS and introduced in a rather high volume to the trap column.

4) Remove the highest one or two or lowest standard point (s) to see if there is any big change
--> I have done this too, and thus limited my measuring range already. The lowest calibration point is now about 500 lower than the highest. Before I did this, recovery issues were a little more pronounced than now.

mmm....I am a little lost here! How can I deal with this issue?

[Ken]

What is the sample solvent that you're using? How about the gradient conditions?

Normally, ion enhancement of complex biological matrices are very dependant on whether there is co-elutions of the expected sample peak with the interferences or matrix enhancing factors that co-elutes with the peak.

When co-elution of peak happens, this will of course increase the signal of the peak at that particular retention time.

So, if you can tell us more about your sample extraction as well as the sample solvent and the gradient, it will be very helpful to troubleshoot your problem.

[JGK]

From what I am reading, are you using "External STDs" rather than "Extracted" Stds?

Have you asessed the level of matrix "enhancement"? (Performing blank extractions and spikingat the last step before measurement)

The issue of a high recovery is not an issue when preforming the assays with extracted STDs

http://www.fda.gov/cder/Guidance/4252fnl.pdf is a useful document to follow for bioanalysis

[Anna01]

To Ken:
What is the sample solvent that you're using? How about the gradient conditions?
--> Sample is simply diluted to a final sample solvent of MeOH:H2O 35:65. This is injected onto an Oasis HLB extraction column, and washed with 20% MeOH. A H2O--> ACN gradient is applied, and elution occurs at about 40% ACN. Compared to other experiments on separation, this acieves pretty good separation from known interfering substances.

To JGK:
...are you using "External STDs" rather than "Extracted" Stds?
--> Standards and samples are treated exactly the same. Standards are prepared by spiking a charcoal-stripped matrix, proven to be free of the substance of intererest.

Have you assessed the level of matrix "enhancement"?
--> I have spiked H2O and a real sample matrix with the same amounts of the substance and IS (single measurements only). Data from stripped std.matrix is not directly comparable, but I should probably do this to. Anyway, this is what happens to AUC:

Matrix substance IS
H2O 100% 100%
real matrix 88% 70%
stripped std.matrix ??? 100%


ok...this test should be done in replicates to be sure I can trust the results. Given that this trend holds true this would explain, at least partially, the "overestimated recovery". Somewhere IS is lost. Either during sample prep or due to ionsuppression, right?

How can I improve the situation? I rather not complicate the method more than necessary...or making it more laborsome.

Btw, I have made linear test in several real matrices by, and the slope seems quite constant.

[JGK]

Are you measuring an endogenous substance with this method?

It would appear that there is something present in the matrix which is suppressing the response of the analyte and IS which is not present in the "stripped matrix".

Have you tested the level of suppression across the range of the assay?

If it is constant ,then you may have to correct mathemetically if you cannot use the matrix for STD preparation.

[Anna01]

hmm...looks like I am in trouble here then!

Should the AUC of the same amount of the substance and the IS always be be equal (=no ion suppression)? Or is it allowed to have some ion suppression as long as both IS and analyte are generally equally effected? If so, are there any thumb rules for maximun ion suppression allowed?

I am trying to measure an endogenous substance, a precursor of an androgen steroid hormone. It's rather non-polar. Maybe there is some way to improve the sample cleanup? I utilize a Oasis HLB and wash the sample with (large amount) of 20% MeOH. If I increase the MeOH I start loosing the substance...

[Uwe Neue]

I don't think that significant levels of ion suppression or ion enhancement are acceptable. You do not know the cause of this, and it may vary from sample to sample. The only solution is a better sample cleanup. Been there, done that...

You need to develop a better sample cleanup technique. I can send you some publications how we went about this, either with the Oasis HLB or with other Oasis packings. My first suggestion is to use pH flips with the setup that you have to get a cleaner sample, but there are better tools around as well.

[Ken]

Yeah, I do agree with Uwe.

I forget to mentione that the pH of the samples are very critical; ion enhancements can also occured due to co-eluting peaks of interest and the endogeneous peaks.

How about column setup as well? I just encountered some ion enhancement problems due to co-eluting peaks and using a more suitable column helps !!!

Just a simple suggestion...

[JGK]

hmm...looks like I am in trouble here then!

Should the AUC of the same amount of the substance and the IS always be be equal (=no ion suppression)? Or is it allowed to have some ion suppression as long as both IS and analyte are generally equally effected? If so, are there any thumb rules for maximun ion suppression allowed?

I am trying to measure an endogenous substance, a precursor of an androgen steroid hormone. It's rather non-polar. Maybe there is some way to improve the sample cleanup? I utilize a Oasis HLB and wash the sample with (large amount) of 20% MeOH. If I increase the MeOH I start loosing the substance...

If as it appears from your data, that a fraction of your analyte is "bound" to part of your matrix (a part which is removed by the stripping process and therefore not seen in your STDs), you will have to modify your sample process in order to "unbind" that fraction for elution which may be tricky.

[yangz00g]

I assume you are using ESI, should APCI be a better option since it's a steroid?

[Anna01]

To yangz00g:
Yes, you are right. ESI. I have tried using APCI, but it lacks the sensitivity needed.


To Uwe Neue:
Today I tried to improve sample cleanup in several ways, including pH flips. I have added 2% formic acid to the washing solution, and increased it's MeOH content to over 50-60%. I start loosing my substance at about 30-40% MeOH, but compared to the sample prepared in H2O ion suppression is little effected (about 20-25%). It doesn't look very promising. Could the MCX column (or other) be a better choice?
One peculiar thing. When doing experiments on spiked H2O I notice that almost 50% of the AUC is lost by washing i 20% MeOH. Under the exact same conditions, this is not happening to spiked matrix. Uuuh?

To Ken:
I believe my separation is quite good, at least from other known interferences. This is not a short fast method. The analytical column is 150mm.
Regarding the pH of my samples - I add formic acid during the sample prep.

To JGK:
You might be right that a fraction of my analyte is "bound" to part of the matrix, although this comes very unexpected to me. Besides I use an isotopic anologue as IS, which should compensate for this. Earlier I have tried adding ACN to precipitate proteins and strong acids, neither did any difference.

Still (very) stuck

[Uwe Neue]

It is not clear what you did, what the nature of the analyte is, etc. Contact me to discuss this in detail in private, if you want.

If you have an acidic analyte, it will be more retained under acidic conditions, and you can wash with a higher concentration of organic before eluting it. You can then go to a basic pH to selectively elute it. You need to then adjust the pH in the online mode to get back to the pH that you need in the HPLC separation.

If you have a basic analyte, you can wash at alkaline conditions at a high concentration of MeOH, then go back to a lower concentration and acidic pH to elute the analyte onto the analytical column. After elution, you need to disconnect the precolumn to avoid that all the gunk from the precolumn ends up in your chromatogram (this is why I prefer off-line sample prep... gives you much more flexibility)

If you have a neutral analyte, you can wash the precolumn with acig and base just before elution of the analyte. Disconnect the precolumn after complete elution of the analyte and clean the precolumn in a separate step.

I have worked with on-line methods and off-line methods. I personally prefer off-line methods, since they give you a lot more flexibility (=less constraints).