Linearity issues using UPLC/MS

genpennick

<p>Hi,</p><p>I have the WATERS Acquity with single quad mass spec. I have developed a method to quantitate six drugs with internal standard; however, after extraction my standard curves no longer are linear even though there is great linearity (r² >0.999) for each drug’s methanolic (unextracted) curve. My analytical measurement range has been determined to be 0.25 - 3.0 µg/ml to achieve linearity. I have several concerns and hope that someone can provide answers to my questions. We have experience with HPLC however introducing the MS into the equation has presented some issues that I do not have the answer to. First, my drugs are hydrophobic and during some testing it was determined that the sensitivity appear to be greater when injected into the mass spec if methanol is used rather than water to reconstitute the dried residue after elution. Because my methanol standards for all drugs are linear, something must be happening during extraction. I have checked all phases of the extraction for breakthroughs but have not discovered any obvious problems. The plasma matrix is acidified with 4% phosphoric acid loaded unto WCX SPE columns, drained by gravity, washed (1) with 5% NH₄OH, washed (2) with 5% methanol; eluted with two elutions of 0.5ml 100% MEOH, and 0.5 ml 2% formic acid in MEOH. The combined eluent is dried under a stream of nitrogen, reconstituted with 100% MEOH and injected under the following conditions: gradient run ranging from 95% water / 5% acetonitrile changing over a 6 minute run. The column used is the Acquity UPLC BEH130 C18 1.7 µm 2.1 x 100mm column, electrospray ionization and SIR mode. The peaks are sharp and all drugs are present, however my r² ranged from 0.95 – 0.98. This is a clinical laboratory and probably unacceptable. I need to achieve at least r²= 0.99.</p><p>Thanks,</p><p>GP</p>

lizh

Hello:

This is interesting, can you supply the curve, it is easier to visualize where the issues are with it. Can you supply the full instrument method, gradient table flow rate, injection volumes, wash solvents, needle surface and loop size etc etc.

It could be a sample and instruemnt interaction, but I will check tomorrow with better MS users than I (which is most people here) and see what they say.

All was well until you ended up with samples in 100% methanol, if I understand what you are stating correctly? Therefore, it may be that your hydrophobic samples are sitting happily in 100% methanol and then they hit 5% aqueous in the gradient and decide they would rather not go there. So a simple test might be to see if changing injection mode helps. Same method but try Full loop and see what happens. Another test, would be again, same conditions, but start the gradient from >5% say 30% and see what happens.

What do you think?

Liz

genpennick

Hi Liz,

I am so excited to receive a response to my question. I am a medical technologist with experience with developing HPLC assays however my knowledge in this arena is lacking. I have attached a PDF file (17 pages) showing the complete instrument method and curves (both neat & extracted) for six drugs + internal standard. It is so obvious that something is happening either during the extraction or injection. As mentioned previously, my peaks are sharp on all channels. Also I have developed another issue which is probably all related. The SQD, we think, has a plug in the POD valve. It has been suggested that this may have been caused by precipitation during the injection which again may point back to my injection method. The WATERS engineer is coming to fix that problem. So hopefully we can iron out some of these issues. Again thanks and I look forward to receiving your input after review of the methods and curves sent.

GEN

genpennick

Hi Liz,

I think your suggestion to increase the gradient from >0.5% solved the problem. I started off with 30% acetonitrile however one of my peaks split. However the best peak resolutions were with 90:10 (0.1% FA, water/acetonitrile). Along with a few other minor changes with the extraction method, the linearity issues seems to have been solved.

Thanks...

lizh

That is great news!