Calibration Curve Problem

[Prometheus]

Dear experts,

I am currently doing stability studies of anti-ulcer drug in human plasma using HPLC-UV. We already validated the method in our lab but the problem now is we can't reproduced a good calibration curve, the low concentration level or LOQ in our calibration curve always have a % deviation from nominal greater than 20%. The linear range of our calibration curve is 0.02-2 ug/mL. Can you help me figure out the culprit here.

[bert]

Did you try to use a weighting factor?

regards Bert

[Prometheus]

Dear Bert,

Yes. I try but the % deviation from nominal of the LLOQ (0.02ug/mL) is still higher than 20%.

[bert]

Can you show us the data?

regards Bert

[bartjoosen]

Have you already investigated the possibility of carryover, instead of changing the way of calculation?

Have you made some stds with a range across the LOQ (like 70-85-100 (LOQ)-115-130%? All the same behavior?
Have you compared the amount of noise with the original validated the method and the amount noise currently?

Bart

[Mary Carson]

This may be kind of obvious, but have you checked your pipettors? Last time we had a "bad calibration" it was actually a pipettor that had quite suddenly gone off.

[Prometheus]

Hi All,

Yes,I already checked my pipettors,it is in good condition. Mary,we also experience that before so we are more careful now in handling our sample preparation, we usually check our pipettors .Thanks for the comments.

Thanks for your suggestion Bart, I guess I got some carryover and S/N ratio problem. I follow your suggestion, I compare the amount of S/N with the original validated method,there is a great difference. How would I deal with carryover and S/N problem? Please help.

[HW Mueller]

You don´t guess on carryover, you determine it by injecting sample solvent, H2O, or whatever. There has been lots of stuff on carryover here, generally you determine were it comes from, then clean up.

[bartjoosen]

You don´t guess on carryover, you determine it by injecting sample solvent, H2O, or whatever. There has been lots of stuff on carryover here, generally you determine were it comes from, then clean up.

As HW Mueller said, find the source of the problem and solve it.
Find why the S/N ratio is too high, and I think your problems will be solved.
Look at the lamp, has it more than 2000 hours? Check the flow cell, is it contaminated?, check your mobile phase, have you changed supplier for some reagent?, ....

Good luck

Bart
PS: Let us know what the final answer was, so we all learn from the topic

[leadazide]

An often overlooked reason for high S/N is electrical interference.. Are there any power cords runing close to your detector? Are your data cabels shielded? Are there any other instruments connected to the same power outlet?

One way of trying to find electrical "noise" is to wiggle to cords behind the instruments.. If you see a fluctuation in signal.. you have some electrical interference from the cords. A quick solution is to make sure all the cords are runing straight. So no rolling up powercords or datacables! And try to avoid your powercords running close to (unshielded) datacabels and your detector.. and also the A/D-converter if you have one of those lying around.

[Prometheus]

Hi All,

I apologize for not giving an update to this issue. During troubleshooting, I account the S/N problem with the flowcell, the D2 lamp that I am using operates 1010 hrs ( less than 2000hrs).The high amount of noise is accountable to a contaminated flow cell so after a clean up according to our detector manual,the noise was greatly lessen. High amount of noise greatly affect the quantitation in the low level of concentration but does not greatly affect those with higher level of concentration.This is the culprit , the reason why LOQ fail ( high % nominal deviation), greater than 20% in the calibration curve.

Regards with carryover, I determine that by injecting a sample spiked with the analyte highest concentration in my calibration curve ( 2ug/mL) followed by injecting a mobile phase. There is a small % of carryover.The needle washing solvent that I am using that time was MeOH:H2O (50:50).The analyte that I am analyzing is more soluble in methanol than water (slightly soluble). I try to replace the needle washing solvent with methanol, and repeat the procedure I mention earlier, the carryover was eliminated. But, I am still wondering why the carryover was not present in our validation stage but in our recent analytical run,there are some traces using MeOH:H2O (50:50) although it was eliminated using MeOH.

After the S/N ratio and carryover problem was addressed, I proceed for a new calibration curve. I am happy to inform you that it pass our criteria and comparable to the calibration curve in the original validation that we conducted 3 weeks ago.

Thanks for all the help!

[Alex Buske]

Hi Prometheus,

A reason for not finding the carryover problem in validation could be that often (at least with me) the lowest concentrations are injected first.

Alex

[Rollo]

If you have carry over in your system you really should check the kind of vials you use. Maybe in the validation state different vials were used than you did. Next thing is if you have a wash vial do not use a cap for that !!!
The washing vial must be open.
If this still does not help check when the last maintenance has done. An old needle/seat could cause carry over. What kind of HPLC you use ?

[Prometheus]

Hi All,

Rollo: Yes there could be a lot of carryover source, the vial,needle etc. We use HPLC by Shimadzu;autosampler- SIL-10ADvp model.

[Vingo]

Hi Prometheus,

As you are using the Shimadzu SIL-10ADvp autosampler, you are familiar with the possibility to clean the needle of the sampler before/after aspiration of the sample. Is this setting enabled? Also set the rinse dip of at least 5 seconds. And is the rinse solution adapted to the compounds that you are analyzing? The rinse solution should dissolve the compounds of interest very easily.


Best regards,
Vincent Goudriaan
Sales support engineer LC-LCMS
Shimadzu Benelux