strange increase of sensibility during the analysis

[Maristela]

I am working with bioequivalence of loratadine in plasma samples using carbamazepine as internal standard.

The mobile phase is acetonitrile (90%) and water 20mM amonuum acetate 0,1% formic acid (10%), flow 0,6 ml/min, API 2000, Gemini column from Phenomenex 150mm, 5 min run.

I injected one calibration curve and after the injecton of 80 samples I run the same curve again.
I have a increase of 3-4 times in loratadine area, while the carbamazepine area did not change (around 3%). That means that both curve are linear but the slope is diferent.
I though about some hardware problem during the injections that could cause this kind of variation, but I think that would not be linear.

Did anybody could give me a help on this ????

[Uwe Neue]

under these mobile phase conditions, I would bet that your sample is not retained a lot. If that is the case, you may be seeing ion-enhancement from plasma components. If this is the case, you need to improve either the sample preparation method or the LC method or both.

[Dirk Hansen]

Hello Maristela,

could you give us also some details on the sample preparation ?
Is your LC/MS method really an isocratic method ? Have you already tried a gradient method ?

Dirk

[Maristela]

Thank you for the replay.
It is an isocratic method: 10% water with 20mM amonium acetate and 0,1% formic acid and 90% of acetonitrile.
I had already validated this method. The problem is when I analyzed the volunteers samples. I am testing now some matriz interference from the volunteers. I don´t thnk I have a mobile phase problem....

[Dirk Hansen]

Hello Maristela,

the mobile phase is not the problem as long as you are sure that you have enough separation of your analyte peak from the matrix peak. According to that what you wrote in your two posts there is a high probability a matrix effect. If we keep the LC method as it is, we should look closer to the sample preparation. Are you performing any kind of liquid-liquid or solid phase extraction ? If not, this may help you to get rid of interfering substances from the matrix.

What is the ionisation mode you are running your API2000 ?

Dirk

[Uwe Neue]

This clearly points ot plasma interferences. SPE is the best solution. I it also not clear, how much retention you are getting. If you are using a 4.6 mm i.d. column, your retention factor is about 1, and you have little chance to improve your method under the conditions described. If you are using a 2 mm column, your chances are better, but you will need some solid sample preparation techniques.

If you are using a 2 mm column, another thing worth trying is to include a step gradient to 100% to wash off residual compounds from the plasma that are accumulating on the column. To be safe, I would do this step after each injection, but you may get away with several injections.

[Maristela]

Dear Collegues

I had tried a gradient that comes from 20 to 95% of organic. It worked very well.
I could validate the method, but when I run the volunteers samples, the analyte area decreased during the run (after 20-30 samples). The internal standard did not decrease. I know that the IS is not the most adequate for loratadine, but is what I have in the moment.
It seems to me a matrix problem, but I can not use solid phase extraction (due to the cost of the cartridge). I am using hexane: diethyleter (7:3) as a solvent extractor.
Both qualiy control samples and calibration curves respond the same way, I mean the curve in the end of the run has a loratadine area that is 50% of the area in the beggining, and the internal standard keeps the same.
I do not know what else should I do.
Now I am just praying....

[james little]

might take a look at my web page..

ASMS Poster: "Simple Method to Monitor Lysophospholipids and Phospholipids During LC-MS Method Development via In-Source CID"

see

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

Top subject.
_________________

[Kostas Petritis]

If sensitivity is not a problem, you could always try to inject a smaller amount. This could help with your problem, although it won't eliminate it...

[james little]

I think in the following article they talk about the sensitivity increasing versus time.

P. Rudewicz et al, plasma analysis method, J. Chrom B, 799 (2004) 271-280.

Their explanation was binding of the drug of interest to the protein present in the sample after liquid-liquid extraction. They noted the problem going away when employing protein precipitation instead of liquid-liquid extraction.

I talked to someone at ASMS in San Antonio and they had noted the sensitivity of analytes increasing after sitting in solution in the refrigerator overnight. Again their explanation was decreasing of protein binding of the drug of interest.

Others indicated they sometime acidify plasma with formic acid before protein precipitation to speed the elimination of protein binding and analyte recovery. However, heard others say they don't like to change pH since they had seen the conversion of phase II metabolite to a phase I metabolite.

[Uwe Neue]

Usually, liquid-liquid extraction is a better sample preparation method than simple SPE, but there are sophisticated SPE methods that will beat even LLE.

Please keep SPE in mind - I would think that the cost of about one dollar per analysis is much less than what a volunteer is paid for his blood sample.

I am not able to point to the source of your interference. I am fairly sure that proteins won't pass through your LLE technique. I also doubt that phospholipids do, but James has potentially more experience with this than I do.

[MP3]

Did you condition you system using extracted plasma sample before you submited you run? We usually inject the same pooled plasma sample at least 10 times, and see what's the RSD. If we feel comfortable, then we submit the sequence.

[Maristela]

Did you condition you system using extracted plasma sample before you submited you run? We usually inject the same pooled plasma sample at least 10 times, and see what's the RSD. If we feel comfortable, then we submit the sequence.

We had tried this when using isocratic mobile phase. It worked well for a period. When we changed to gradient, we stoped.
We observed a great variation of the loratadine concentration (most of the time increse of concentration), even if we inject 100 samples.