Interferences in LC-MS-MS

[Maristela]

Dear friends

I am working with bioequivalence study in plasma samples.
I verified that after one year, instead of degradation of the samples and decreased of the anayte in plasma, the oposite happens and besides, I had a huge variation (25-40%) between replicates in the same sample.

The drug was metoprolol and was administered one year ago and stayed in at -20°C until the analysis.
The method was completed validated and the results were in accordance with the accepted criteria.

Does anybody has any references on this or perhaps an explanation about it ?
Thanking you in advance and hoping to hear from you soon.
Maristela

[Uwe Neue]

Maybe matrix interference in your old analysis, and less with the new analysis. How do you clean up the sample?

I can send you a review article with references on the subject, if you contact me at the e-mail below.

[Maristela]

Dear Uwe

I read the article, and it is not clear how could I have an increase of the signal instead of decrease.

Related to the sample processing, I extracted 200ul of plasma at pH 11 (20%NH4OH) with dichloromethane-diethylether (8:2). The mobile phase was methanol-formic acid 0,05% (1:1). I did all the validation and the method was OK.
As I am analyzing old samples, before I started with the volunteers samples, I exctracted an stability test that was prepared almost at the same time as the samples was collected and they were OK (accurcay around 100%) compared with recently prepared quality controls.
I finished analyzing all the samples after 1 month from the validation and when I started re-analyzing some samples that had problems, the variation between them was big. I did replicates and sometimes the CV between them was 25-45% !!! And this was not due to any wrong procedure. I re-test that stability test analyzed before and the accuracy was around 137%!! So, it was happening the same as I had with my old samples.
But I still have no other explanation besides the proteases acitvity.
Do you have anyother idea ??????
Thank you a lot.
Maristela

[Uwe Neue]

I am not sure if I understand what you have described. Therefore let me rephrase it.

You recently reran an old stability test standard. You did a new extraction with this old standard (which does not contain plasma, right?), and you got 100% accuracy.

You reextracted the old samples again, and the results were significantly higher now than what they were 1 year ago. Also the variability between new repeat extractions of the same samples was very large (please verify, because this is important for the interpretation!).

You also retested the old stability test sample again (the one that still worked a month ago), and it now gave you 130%.

Please clarify, if I understood you correctly or what I missed!

[Maristela]

Dear Uwe

I will try to explain better:

1) at the time when the volunteers samples were collected, it was prepared an spiked sample (SS) that were frozen at the same time (-20°C).

2) I recieved all those samples almost 1 year after to analyze them

3) I validated the method (specificity, recovery, precision, accuracy, stability)

4) Before start with the volunteers samples we check the stability analyzing the SS and compared the results with samples recently prepared. The results were OK (around 100% of accuracy).

5) After finishing all the volunteers (1 month and ± 2500 samples) we made some repetitions with samples that had a strange profile (too high or too low or some that had been lost during the extraction). Now I started noticing that something was wrong.

6) while my controls were OK, some repetitions had CV around 25-50% between the replicates and even though we re-do it some times it was OK and others were even worse.

7) I analyze the stability samples (SS) again and they were also higher than before (around 130% compared with freshly prepared samples). I am sure that ther is no analytical problem.

If my samples are not stable anymore (what could happen even if the samples stayed frozen during all this period), I should have a degradation and the area should decrease.

9) why do I have an increase of the area ? Is it something related to the protease acitvity as metoprolol is 98% bounded to protein ?

Thank you a lot for the interest and please let me know if I was misundesrtood.
Maristela

[Uwe Neue]

Thank you for clarifying what you have done! It is now much easier to understand, and we can potentially find a solution.

Have you used the same column during all recent results? Has the column with the increased area for the SS been used extensively? What are the details of your chromatographic method (gradient/isocratic, column wash?)?

The reason for these questions is that you can also get INCREASED signal (signal enhancement instead of suppression) from coeluting interferences that may have accumulated on the column.

[Maristela]

Dear Uwe

Some other points to clarify:
1) I used the same column during all the study (Zorbax Eclipse XDB-C* 150 x 4.6 mm, 5um). The mobile phase was methanol-formic acid 0,05% (1:1), isochratic, flow 0,7 ml/min (split 1:1).

2) Before any set of samples (around 200 samples) I inject standards os anayte and IS in mobile phase to check the chromatography and the MS.

3) Each set of volunteers samples come along with 2 calibration curves (one in the beggining and another in the end of the set) and quality control samples that are put each 10-12 samples. All this controls and calibration are extracted together with the volunteers samples. The results of the accuracy of my controls allow me to accept or not the set of the samples anayzed.

4) At the end of all set, I wash the column with methanol-water (80-20) during 40 min at flow 0,5ml/min. After this, we re-equilibrate with mobile phase for approximately 40 min, inject mobile phase, the standards and if they are in the expected retention time and area, we run all the set. The same standard is injected in the end of the set to guarantee the quality of the chromatography.

Thank you for helpíng me to find out some explanation for this......
Thank Thank Thank Thank Thank......

[Uwe Neue]

Here is something to consider:

hydrophobic compounds from the plasma samples have survived the extraction process and accumulate on the column. The washing process does not remove them, and they continue to accumulate. Now they elute and interfere with your analyses, no matter what you inject. The interference does not create the commonly expected ion suppression, but ion enhancement.

Sounds complicated, but I believe that it fits to your observations.

Solution: wash the column with something stronger than 80/20 methanol water, e.g. with THF, and try again.

[Maristela]

How can I have problems with my volunteers samples and not with my controls that is injected every 10-12 samples ?
If I had something retained in the column, why is it only run with the volunteers samples and not with my control (of course that I have much more samples than controls, but it is a lot of coincidence that the " trash" would run only with the volunteers samples ).

Do you think that protease activity is something from my creative imagination ?????
Maristela

[Uwe Neue]

Under 7 above, you said that your stability samples also had a higher (=130%) value.

Something else is bothering me now: what do you do to prevent the protein binding of your analyte? I had implicitely assumed that you had taken steps to eliminate protein binding, and the treatment of the sample at pH 11 would take care of this. The question now is, if you are sure that you did not have protein binding in the past, and less now, as the protein in your sample has aged?

[Maristela]

Uwe

I did nothing to prevent protein binding. The drug is highly conected to protein (98%), but I did not have to precipitate (with acetonitrile or methanol) them to analyse because I had enough drug free in the plasma.
Maristela

[Uwe Neue]

So you do not know if you recovered 100% of your analyte in the early studies. Now, with the new studies, the protein might have degraded in the meanwhile and more drug could have been liberated.

Is this a possible conclusion?

[Maristela]

I think so. At least, it is reasonable.

I never thought about this, but I think I have to consider a deproteinization step every time my drug has protein binding, independently if it is detectable with deproteinization.
Maristela

[Maristela]

I think so. At least, it is reasonable.

I never thought about this, but I think I have to consider a deproteinization step every time my drug has protein binding, independently if it is detectable without deproteinization.
Maristela

[Uwe Neue]

Well, it may be possible to prove or disprove this idea. Please do not think about deproteinization (=removing of the protein) but think about means to destruct the structure of the protein. Removing the protein will remove the bound analyte with it. A destruction of the protein will prevent the binding.

This is usually accomplished by changing the pH (commonly to a very acidic pH, but alkaline pH is possible as well) or by the addition of organic solvent (acetonitrile precipitation). In your case, it could be question of exposure time to the alkaline pH, or maybe with an increase in temperature, before the extraction. You could compare a sample treated the old way to the same sample where you did things to elimate any possibility of binding, and compare the results of the extractions. You have to keep in mind though that the extraction conditions (pH etc.) can not be altered.

Just some ideas... You may be able to come up with better ideas.