Chromatography Forum

Hello everyone- I have a perplexing problem and hope someone can offer some help!

First- the details:

extracting rapamycin from whole blood with a simple extraction procedure of adding 600ul MeOH to 300ul blood (with 50ul Triamcinolone @ 1ug/ml). vortex, then spun at 14,000 rpm RT. The supernatant is then injected into an LC/MS system.

I am spiking with Triamcinolone as the IS. We chose this compound because it was readily available in our lab.

LC/MS: positive mode, API2000 with APCI source, Agilent 1100 Autosampler injecting 40ul. Column: Supelco LC18 3.0 x 5cm(length)
mobile phase: 10mM ammonium acetate for 2 minutes, followed by 90% MeOH 10% 10mM Ammoniun Acetate for 6 minutes. rapa elutes at 5 minutes. IS elutes at 2.5 minutes.

I made a set of calibrates with a range of 0.3ng/ml - 50ng/ml. I extracted this set 9 times (a total of 90 samples or vials)

During the sample run, I am seeing that the Internal Standard is increasing in peak area/height by 20% from the beginning of the run to the end. However, I see that the drug of interest is not increasing at the same concentrations from beginning to end. (if that makes sense)

Any ideas what can cause this? (I am not sure if this is Mass/Spec related or just Chemistry related- so I posted this here.

Thanks for any and all input!

-Larry

On first glance, I do not see a problem on the LC side. Since you are seeing the problem only with the internal standard, it is possible that you get signal enhancement in the MS due to increased level of interferences. I think even more so that this is the case, since your sample preparation is rather simple.

I suggest two paths: inject a few neat standards and see if they exhibibit the same problem, and ask the MS expert about signal enhancement, especially from hydrophobic interferences such as lipids and phospholipids. If the latter is the case, a wash of your column with a suitable hydrophobic step gradient should help.

Thanks for replying!

I would like to also add that the supernatant for injection is colorless, but not clear. after an overnight run, I can see it settled out with time into an easily disturbed and very small pellet. Also- I forgot to mention that I run a column wash before each injection (10mM Ammonium acetate) to flush out salts from the injection. Could the problem be directly related to the amount of time the samples are sitting in the autosampler rack?

Hello

I've done a few drug metabolite studies on serum using GCMS, and I've come across variable peak heights too. Thought I'd add a few suggestions from my side. Just suggestions !

Seems to me Triamcinolone is getting protein bound to some extent. MeOH in my experience isn't very efficient in disrupting protein-binding and when you let the supernatent settle over a period of time, Triam gets slowly released into the supernatent, resulting in higher peaks.

You can prove / disprove this by injecting a few samples fresh and repeat the injections after letting the sample settle overnight. If you get enhanced but steady peaks heights for the IS, perhaps I am right.

I'd also suggest you try spiking whole blood with the IS but without the drug of interest and check again. This would be in conjunction with the pure standard injections that Uwe rightly suggested.

In case I happen to be right, then all you need to do is to wait overnight and then inject - assuming rapa is stable in solution. If not, I'd suggest you use either TCA or ACN for denaturation.

Hope this helps.

I have had some problems with analyzing plasma samples. I normally take 100 ul of plasma, ppt protein with 250 ul acetonitrile. Then spin down, and inject from supernatant by offsetting the syringe to keep from fouling the needle with the protein in bottom of 96 well plate.

Use a Quattro Micro with UPLC system.

My system often increases the response of components with time. Unfortunately the rate of increase of internal standard is not the same as the analyte causing some problems with the analysis. Often seems to level out after and hour or two.

You could try just doing the analysis out of water. For example, subsitute water for the plasma and do your standard workup. See if the same problem is ocurring, if so probably instrument.

I have done a lot of work on understanding the effect of lipids in plasma. If the lipids are not cleansed from the column, if can increase or decrease the response of the analyte/IS with time. This is because with time the lipids begin to come off the column. I monitor m/z 184 vs time in an MRM channel fragmenting the lipids in the source and passing through the collision cell without further fragmentation.

When the lipids become steady state, the respone for analytes/IS will normally increase, but not both at the same amount. So important to be steady state if not flushing the lipids from the column.

I have a paper in press in J. Chro. B, but will be a while before published. Some preliminary work at

http://users.chartertn.net/slittle/

In paper, found that reasonable to not elute lipids from column if using gradient, isocratic too varialbe, or if using labeled internal standard. If not, safer to elute before injecting again.

Good luck.. I have had many of the same problems you noted.

a few more things, very briefly stated below, much more detail in article.

My instrument usually shows increase of response over the first few hours, then then reponse levels out. Can be as high as 20%. Again, problem is that doesn's change the same for IS vs analyte, but drift in response somewhat cancelled by presence of internal standard vs not. After that time, the precision with or without internal standard is the same. Don't understand why, tried with letting flow go into column for several hours before analyzing sample with instrument interface heated, etc..

With a labelled IS, the increase was noted, but the amount of increase for labeled vs unlabeled was same.

In work in press, generally noted an increase in response of the analyte and internal standard as the phospholipids not eluted began to break through the column. Could meet FDA guidelines for tests if used a small gradient that didn't elute lipids initially with unlabeled standard, but became steady state after 5-7 injections. However couldn't meet guidelines when isocratic conditions used.

However, could meet guidelines with isocratic run with labeled internal standard and didn't need to necessarly wait for the phospholipid signal to become steady state.

Learned a lot, but still have a lot that don't understand about matrix effects. For a lot of the anlaytes if you do the post column addition of analyte while injecting sample and eluting lipids, the analyte is suppressed. That same analyte response is enhanced in counts (not necessarily S/N) when allowing the lipids to bleed off column by breaking through from previous analyses.

Some people have shown that lipids can enhance analyte response at low levels and suppress at high levels.

For now, I usually take the time to elute after each run since I can usually get that to occur with a 5 min cycle time including the autosampler sequence.

If that is too long, use a gradient and let lipid signal become steady state or use either gradient or isocratic with labeled internal standard.

Thanks for all the input/help!

I went on the hunch that there was some protein binding going on in the autosampler rack- so I added 50ul of 9% formic acid solution to each vial. This seemed to help tighten up the CV's dramatically! While it's still not perfect (5-10% variability), the Internal standard is trending upwards MUCH less than before. I also notice that the IS increases in peak area/height at the beginning, then it stabilizes.

I suspect there is still some ion suppresion issue due to phospholipids as james suggested. Thanks to ALL of you for you input and help!! You have offered ideas I would never have thought of!

-fixed typo's

Looked a little more carefully at your analysis and have a few suggestions. Looked at the structure of your internal standard. Could be forming the dimethyal acetal of the ketone with methanol? Would break up with formic, but might worry about decomposing some phase II metabolites to free metabolites.

I ususally try adding several internal standards initially, and choose the one that tracks the analyte of interest the best in relative response. Often leave several IS in the analysis which gives me some flexibility if something affects the internal standard response during the real study samples.

i was also looking at your chromatogrpaphy. Pretty sure you are probably not eluting the phospholipids after each analysis, but probaly are eluting the lyso phospholipids. Since you are running a gradient, probably after injecting 4-5 sample, the lysophospholipids and phospholipids reach steady state.

Since not using labeled internal standard, the response of the analyte will be changing during the first 4-5 injections. However, doesn't explain the increase in the IS since it is probably eluting in a region not affected by the lipids, but I think it could lead to more variablity in your results if not understood.

Should try monitoring the lipids using m/z 184 ion. Idea is to form 184 fragment up front in source and pass through collieion cell withou fragmenting. Can optimize the m/z 184 by infusing some of the supernatant from the protein precipitation. Dont know much about the API 2000 instrument, on the 4000 used the following..

The primary method for simultaneously detecting all GPCho's on the Applied Biosystems 4000 Q TRAP employed a declustering potential, entrance potential, collision energy, and collision cell exit potential of 165, 10, 7, and 5 volts, respectively. The mass transition of m/z 184.1184.1 was monitored in the positive ion electrospray mode. Similar parameters were employed for monitoring 2-lyso-GPCho's via the m/z 104.0104.0 transition. The collision gas (nitrogen) pressure was set to medium.

james- Again- thankyou for your continued help! Can you please explain a little more about what phase II metabolites vs free metabolites? Is this in reference to rapa? or the IS? It seems that while I thought that adding formic acid would disrupt protien binding and give a more constant peak area, my theory could be incorrect, but the result from adding the acid is what I was hoping for. My shortfalls as a biology major has been knowing some of the basic chemistry going on here. I am finding that working with our new mass spec forces me to review those basics!! If I can get some time, I will try monitoring the MRM pair you mentioned for the lipids.

-Larry

I had read somewhere that acid addition could turn a gluronide of the product (a phase II metabolites) back into the original drug. I'm not familiar with the metabolites of your particular drug, so don't know if it would be a problem.

Your internal standard might be forming some type of acetal with methanol and acetals are not stable in acid. Thus the addition of acid might reverse this process??

Might try using 2:1 ratio of acetonitrile to plasma to precipitate the protein in your plasma to avoid using methanol and see if internal standard problem goes away.

Also swithcing to an internal standard closer in character to your drug might be useful.

All just conjecture, I have doing these types of analyses for 2 years now and still find it very challenging to develop a robust analysis.

Good book to buy would be

H. Mei in: W. Korfmacher (Ed.), Using Mass Spectrometry for Drug Metabolism Studies, CRC Press, Boca Raton, FL, 2005, p. 103 (Chapter 4).

see http://www.sigmaaldrich.com/catalog/sea ... rand=SIGMA for table of contents.

Good luck.