Failing Check Solution

[Jade.Barker]

Can anyone point me in the most common reason for a failing check solution? (Other than the obvious... analyst error )...

The check solution is made daily, at the same time as the calibration standards. Then I vial up enough for the whole day. Only some of control vials are failing, but it seems random: Not all the check solutions in a day fail, not all the check solutions in a sample set fail. The two analytes have failed both high and low, independant of each other. Early and late runs have been effected equally. The failing vial is not always the final vial.

My calibration curves are good. My system suitability is good. The analyte amounts are in line with expectations. I'm scratching my head on this... any hints? What other information do you need?

[Peter Apps]

Hi Jade

What are you analysing, and how ?

Peter

[gcguy]

Have you tried running your calibration solution multiple times? Check that the areas for the peaks fall within the specified precision criteria. If they don't then it is probably the instrument not the analyst.
I have seen this problem on HP1100. Belive it or not we fixed it by using Agilent autosample vials. Some third party vials can form a slight vacuum when the instrument is drawing up the solution.

[Ron]

In general GC autosampler needle points are sharp and LC autosampler needle points are blunt. They are made that way because GC needles are designed to pierce the septum cleanly, and LC needles tend to tear the septum. For GC you are typically pulling a 1 or 2 microliter sample, while in LC you are pulling 10 to 25 microliters or more. It is rare to affect sampling by creating a partial vacuum whlie drawing up the sample in GC work, but not that uncommon in LC work, particularly when using vial insert or conical vials.

I have heard, but have never seen this, that if you use conical vials and the needle penetration depth is too low that it is possible to have the needle contact the taper in the vial, form a partial seal, and affect sample volume that way.

I think the previous post could be right on target, try a different vial septum optimized for LC use. I don't think it is another part of the system since the problem is so random.

[Jade.Barker]

Thanks everyone for the quick feedback - This board is a lifesaver!

The analytes are two forms of a steroid drug. I'm working on validating a new elution method using 65% acetone (35% water), the previous method was 90:10 MeOH: Water. We can't stay with the previous method because of a polymophy issue with the drug behavior in methanol. Acetone seems to clear up the polymorph issue beautifully, but now I'm having this odd problem with the check standards. Rumor is that the other analyst (currently on leave) has also had problems with the check solution... presumably with the Methanol method... but this is just hearsay.

Also I didn't mention 3 sample sets were run on a different HPLC, with reagents made, and samples vials by a second analyst. One of the three sample sets had failing check standards. I'm not sure if that makes me feel better about this - or worse

...If it was a vial issue wouldn't the system suit also have a problem?

You guys are my heros - keep the ideas comming! Thanks again!

[Peter Apps]

If one analyte is above spec and the other one below spec in one run we can eliminate anything that affects only injection volume (such as vacuum in the vial, or needles sealed to vial inserts - which I have actually seen when doing manual syringe operations).

By how much are the check standards failing ? in percentages of what they should be.

Are the calibration standards, samples and checks all dissolved in the same stuff for injection ?, what is it and what is the mobile phase ?

What detector do you use ?, if a UV is the wavelength on top of a spike of absorption, on on a gentle slope ?

How are the peak shapes, and are the peak areas about the same for the checks as for the samples and calibration stds ?.

Do you just dilute and shoot the checks, or do you run them through the sample prep ?

Peter

[Jade.Barker]

Peter,
Thank you for your interest. I'm gathering some of that info now. I'll post back later with that data.
Your the best!

[Jade.Barker]

- If one analyte is above spec and the other one below spec in one run we can eliminate anything that affects only injection volume...
- how much are the check standards failing ?
- all dissolved in the same stuff for injection ?
- what is it and what is the mobile phase ?
- What detector do you use ?,
- UV is the wavelength on top of a spike of absorption, on on a gentle slope ?
- How are the peak shapes?
- and are the peak areas about the same for the checks as for the samples and calibration stds?
- Do you just dilute and shoot the checks? vs sample prep ?

1. In a single sample set I have both high and low failures for the analytes. So I think your right it can't be injection volume related.

2. The checks are failing with deviations of 5 - 11%

3. The stock standards are made with ACN. The mobile phase is ACN:Water gradient. The stocks "B" (1mg/ml) are diluted into a substock and this is diluted into the calibration standards. A different batch of stock "A" (1mg/ml) is diluted just one time for the check solution. The diluent is the 65% acetone elution medium.

4. It's a UV detector, but we should be well within the range... I think. Does 242 nm sound like anything important? I will have to double check the instrument and get back on this one.

5. Peak shape is good, and consistant. RT also. Peak Area for Check solution is about 120,000 - 150,000 peak area for analyte is similar. Explain more how this could be an issue though? I had not thought about this at all.

6. The system suit (cal 5) and Cal 1-5 are all just diluted. Every run includes a elution medium blank that is treated like the samples, ie. in the same shaker vials, on the heated shaker with the samples. Other sample set that do not have elution... like LOQ etc. the blank is plain elution medium right from my main bottle... Speaking of sample prep: We have been TEA coating the shaker table vials to prevent drug loss to the glass (cheaper vials) The vials are dry before elution medium is put in them, so there should be little chance the TEA is an issue right?

Thank you for your help/ interest. I'm meeting with my Principal/ Mgr tomorrow to try to needle this out... It's so nice to have the board as an extra resource. Thanks again!

[Peter Apps]

Hi Jade

Detailed answers, great !

1. Just to be sure; Do you ever get one compound high and the other one low in a single injection ? If both compounds are affected the same in each injection it could well be a volume issue, but it cannot be volume if the peak sizes are going in opposite directions.

3. 65% acetone is pretty strong stuff to be injecting into an HPLC column, but if it was the cause of the problem I would expect you to see some peak distortions. Are you using this for all the injections ?

5. If the check peaks are much smaller or bigger than the calibrations or samples you might get adsorption or detector saturation respectively on the checks but not on the others. Once again this would not send peak areas in different directions.

6. This might be it - do you TEA treat the vials for the checks ?, if not you might be getting adsorption to the vials, and this might be different for the two compounds, giving different behaviours.

Good luck Peter

[Jade.Barker]

Peter,
Thank you again for your thoughts -

1. The theoretical amounts should be right around 18.5 ug/ml for both Drug A and Drug B. In a single sample set run: the First check solution showed Drug A at 17.9 and Drug B at 20.0. The second check solutution of that same sample set showed Drug A at 19.7 and Drug B at 17.5. The check solutions are from 2 different vials aliquoted at the same time. In other words, first the "A" was low and the "B" was high. The very next check from a different vial showed the oppocite: the "A" was high and the "B" was low.

I think you got me going down the right path. So I injected 3 times from another two different check vials.
Check Vial One was: "A" 20.3, 20.5, and 20.3 ug/ml and "B" was 17.2, 17.3, and 17.4. ug/mL.
Check Vial Two was: "A" 22.6, 22.7, and 22.7 ug/ml and "B" was 14.9, 14.9, and 14.9...

3. The 65% acetone is fairly strong, but it was decided upon before I got here. My understanding is that this solvent concentration combined with the shaker table speed and heat produces elution results that meet the regulation requirement of less than 20% elution before the first timepoint and greater than 80% drug elution at the final time point... It's possible this could be treaked maybe a lower solvent concentration with more heat on the shaker table... Tough to say at this point. But you say the peaks would be badly shaped right? I would see shouldering or spliting maybe?

5. I think I understand - basically if the check solution peaks were huge compared to the sample peaks this could be a saturation issue? In any case, the peak areas are about the same in this situation AND the peaks are going kitty-whumpus (different dirrections).

6. Well you bring up an excellent point. I think I may have raised this question at one point - the thinking was that it had been a problem in the elution vials, but not the HPLC vials with the old solvent... - but maybe the same interaction that was causing the polymoroph issue in the Methanol was also slightly protecting the drug from sticking to the glass - the acetone elution drug samples might be "stickier"... Only one way to find out. Maybe I will try TEA coating some of the HPLC vials today and see how we do. Especially, *Especially*, since the two control vials with the 3 injections each were repeatably different. Oh my...

[Peter Apps]

Hi Jade

Adsorption in the HPLC vials starts to look like a likely culprit. If you have -OH groups on your analytes they will try to stick to silanols on the glass. I would expect 35% water would be enough to out-compete the analyte for the binding sites, and so you might have some other kind of interaction. Easy enough to test anyway.

Peter

[Jade.Barker]

...I would expect 35% water would be enough to out-compete the analyte for the binding sites, and so you might have some other kind of interaction.

The other suspicion being tossed around is solution equilibrium, that the substock and cal standards aren't getting enough "rest" time before aliquoting - so the effective concentrations in the HPLC vials might be different enough to cause a problem? What are your thoughts on this?

[Peter Apps]

HI Jade

It's possible - and easy to test.

Peter