GCMS precision and accurary problem

[Edde]

I am developing a quantitative method for ethylene glycol, diethylene glycol and triethylene glycol in serum. The sample preparation step is by adding methanol to precipitate serum protein. The supernatant is derivatized with BSTFA with TMS. Standard curve uses same serum as matrix. However, the precision and accuracy data of spiked QC (low, medium, high levels) were poor. Accuracy can be up to 150% in 1 replicate sample. This happens from time to time.
The tuning report is ok.
1. The temp. gradient is 40 deg C per min in part of the experiment. Can this high rate of change cause the imprecision and inaccuracy at some occasion?
2. The column changed to purple color. Will a deteriorated column affect the precision and accuracy?
3. The liner gets dirty quickly. Will a dirty liner causes imprecision and inaccuracy?

I am very puzzled. I appreciate if anyone can give me a clue. Thanks.

[Don_Hilton]

It sounds like you have a lot of extra stuff in your sample after you precipitate the protein. Dirty liners will make a difference.

You have a lot of degredation going on in the system. A bit more sample cleanup may be required.

It seems to me that a method for diethylene glycol has been developed in Dana Barr's group at Centers for Disease Control and Prevention. If so, it most likely has been published.

[Edde]

Thanks for your suggestion. You reminded me that dirty matrix may be the culprit. I will try using less serum.
Since the internal standard precision is bad too - RSD up to 33%. I will add the IS to MeOH pool instead of adding 50 uL individually to each sample or standard to minimize pipetting error.

[Don_Hilton]

Are you pipetting methanol with a pipetter that draw the sample up with an air column? If so, this can give precision problems. Adding 50 uL internal standard to each sample with a failry non-volatile solvent works quite well. The lab I am in does this all the time - but with a higher boiling solvent. (I believe one group is using toluene, the other nonane)

[Edde]

Thanks again. I already used a positive displacement pipettor, not the type with air column.
Do you think the culprit is dirty matrix? I intend to decrease serum volume to 30 uL. The glass wool on the liner turned brown after about 20 injections previously. Not sure how this affect the column, which already turned purple in color.
The paper you mentioned is a good one. But I am searching for the fulltext now.
Appreciate your help!

[JTM]

I am developing a quantitative method for ethylene glycol, diethylene glycol and triethylene glycol in serum. The sample preparation step is by adding methanol to precipitate serum protein. The supernatant is derivatized with BSTFA with TMS. Standard curve uses same serum as matrix. However, the precision and accuracy data of spiked QC (low, medium, high levels) were poor. Accuracy can be up to 150% in 1 replicate sample. This happens from time to time.
The tuning report is ok.
1. The temp. gradient is 40 deg C per min in part of the experiment. Can this high rate of change cause the imprecision and inaccuracy at some occasion?
2. The column changed to purple color. Will a deteriorated column affect the precision and accuracy?
3. The liner gets dirty quickly. Will a dirty liner causes imprecision and inaccuracy?

I am very puzzled. I appreciate if anyone can give me a clue. Thanks.

1. Don't exceed your GC's maximum temp ramp speed, otherwise it'll do what it can and be uneven, like you said. Not sure if that'll cause much difference in the signal though.
2. The ENTIRE column turned purple? Huh? If it's just the front 2-3 inches, then trim it, yes.
3. You need to change that liner if you can ever see discoloration or particulate in it...

[Edde]

The GC model is 6890 and the engineer said that 40 deg/min is tolerable.
In fact, the entire column turned purpish on top of the bronze color background.

[Don_Hilton]

From your description, additional cleanup of your samples might be a good idea. I can think of only two reasons for a column to be discolored from one end to the other. 1) it has been heated well above the upper temperture limit - well above. or 2) there are non volatiles contaminiating the column from one end to the other - typically darker at the inlet end. If the column is darkened from end to end - it is toast. You do not want to inject nonvolatiles - while the glass wool in the inlet will catch some, it will not catch all of them.

[Edde]

Thanks for your good suggestion. I will half the serum sample volume to reduce matrix. See if there will be any improvement.

[AICMM]

Edde,

Is your 6890 120VAC or 220VAC? 40C in the early part of the run is not as challenging as 40C in the latter part of the run (above about 150 it will be harder to do consistently.) Why do you need a ramp rate of 40 C....?

Have you tried the same experiment without any serum? What did your precision and recoveries look like then? You are using an alcohol and derivitizing diols so I would be interested in knowing what the recoveries are just from the alcohol. Also, methanol is an awful solvent to have to work with in GC if you don't have to.

Best regards.

[Edde]

Thank you very much for your advice.

My GC-MS is 6890 220V AC. 40 deg C was used in the latter part of the experiment in order to shorten the analytical run time.

You are right. The GC oven specification states that the typical 120V ramp rate at temp. range 115-175 is 40 deg C/min. No specification at 220V.

I have not checked precision and accuracy without serum.

I use methanol to precipitate protein. I tried acetonitrile and acetone but visually there is no difference in the precipitating capacity. I chose methanol for consistency sake because it is the solvent I used to prepare glycol standards.

I suppose methanol has evaporated entirely due to its low MW and there is no sign of methanol left over to use up the derivatizing agent as evidenced by the high BSTFA peak in solvent front.

Why is methanol not suitable for GCMS?

In your expert opinion, what should I do to improve the precision and accuracy?

[AICMM]

Edde,

Methanol has a big expansion volume, not as bad as water but one of the more extreme ones. You might want to switch to something like iso-propanol or acetonitrile.

I assume your derivatized peaks, such as they are, are coming out very early in the chromatogram. If so, the inability to maintain consistent oven performance late in the run should not be affecting your early components in my opinion. With 220VAC you can certainly take the oven up to higher temperatures at fast ramp rates but you probably don't need to go much over 220C from your application description. You don't describe your column phase anywhere so it is difficult to set an upper temperature limit. This may be why your column has turned purple in which case that is the place to start, replace your column.

You should also consider that once you derivatize the glycols they are going to be much more volatile and you can loose them much more easily in sample prep. You might add an un-reactive keeper solvent to hold on to your compounds as they derivatize.

Best regards.

[Edde]

Since there is a step to blow dry the supernatant of protein precipitation (containing glycols + methanol), during which the low MW methanol should have evaporated, I assume there is no methanol left when I add the derviatizing agent. Should there not be any expansion problem on injection?

You are right that I don't need to go beyond 220C. I raised the temp. in order to "clean" the column. The HP-5MSi column used should stand the maximum temp. I set the temp. about 30C below the column max temp.

Since ethylene glycol have low MW, I thought of using some un-reactive keeper solvent to prevent it from being blown away - in the sample prep stage, not during derivatization. But I have no idea what to use.

Could you kindly enlighten me please?

Thanks.

[Don_Hilton]

You might use something like DMF. Your analytes are soluable in it and it is compatiblewith your derivatization. but be aware that while you are precipitating the protein, you are carying along lipids and other somewhat polar molecules. And this may be partial. Some lipid types of material may not transfer well in the methanol, and may partion to the protien. If so your analytes may partiiton between precipitate and methanol. So, resuspension of the precipitate and collecting supernatant several times may help recovery and reproducability. And with several washings, you will drag along some of the lipid material. A GPC step would get rid of most of the lipid material. You will have othe small molecules that are soluable in methanol.

[Edde]

AICMM and Don_Hilton,

Thanks for pointing out the reason for the purple column. I guess although the max. chromatographic temp is 30deg C below the column max. temp., repeated injections during validation cause the column to deteriorate.

I have replaced the column with a new one today.

AICMM, you are right that my target compounds come out before 200 deg C. I have adjusted the max. chromatographic temp. to 280, hold for 5 min to clean the column.

Do you think choosing methanol as precipitating agent is not appropriate?

Is there any benefit to change to acetonitrile? I use methanol to dilute the glycols stocks. Is this an appropriate solvent for glycols?

Any recovery or partition problem when mixing the diluting solvent, methanol with precipitating agent, ACN during standard spiking?

BTW, what is GPC step?

Grateful for your valuable comments.