Accuracy results

[smartin]

Hi!!

I'm in the first steps of the development of a new LC/MS-MS method.
I've just check the lineality, precision and accuracy, apart from recovery and LD and LQ estimated.

Everything is correct except accuracy results.
Could anybody suggest me which could be the reason of it? Maybe the points of the calibration point will be differents? I've realised that maybe there are a lot in the low concentrations and not so much in the upper limit.

Please, tell me!!!

Thank you!!

[Don_Hilton]

With the level of detail you've given, it is a bit difficult to pin the problem down to any one area. Give some hints about what you are doing: Analyte ranges, method parameters, compunds of interest, bits of relevant data, etc.

[smartin]

Ok,

I'm working on Lopinavir in blood by LC-MS/MS, using a C18 column.
I used an isocratic gradient, as I only want to determine this analyte, using formic acid in water & acetonitril.
I've tried to use a calibration curve of 7 points: 0.015 µg/mL, 0.1µg/mL, 0.2µg/mL, 0.8µg/mL, 3.2µg/mL, 8µg/mL and 20µg/mL.

I think that maybe the calibration points are not homogeneously distributed among the calibration curbe...
What do you think? It could be the cause of the bad accuracy?

Do you need some more information?

Thanks!!!

[Don_Hilton]

What about your data tells you that there is an accuracy problem? And have you made dilutions from a single stock to make the all concentrations of analytes or have you performed serial dilutions? Are you using an internal standard and what is your injection method - autosampler, manual?

[smartin]

Hi,

This is only the first assay to validate the method. I did the calibration curve (4 samples per level). I calculated the equation of it and after that I calculated the "Estimated concentrations" for each sample using the resulted equation.

From this "Estimated concentrations", I calculated the "Accuracy" as:

((Conc.estimated - Theorical conc.)/Theorical conc.)*100

I use a value of 15% for all the samples except of those around LLOQ, that should be xithin 20%.

Regarding dilutions:
I did a working stock solutions (that is the highest calibration point, CAL7). From this, I did the CAL6 & CAL5; from CAL6 I did CAL4 & CAL3; from CAL4 I did CAL2 and from CAL2 I did CAL1.

Regarding IS:
Yes, I am using an internal standard, and it is correct. I've obtained similar areas in all the samples.

Finally, regarding injection method, this is an autosampler one.

And so????
Where do you think the problem could be?

[Don_Hilton]

I still don't know how the individual values missed the target. Was the mean value high at one level and low at another? Back to bits of data: Perhaps you can list for each injection nominal concentration calculated, and areas of analytes and internal standard. While you have calculated concentration, the magnitude of peak areas may catch the attention of someone familiar with the make and model of instrument you are using. And a look at how much internal standard varies may give a clue as to variability of results.

If I were to walk into your lab and offer to help, you would show me data, and perhaps a chromatogram or two for troublesome samples and for samples that were not so troublesome. You would point out instrument settings, showing what you had considered in setting up the instrument and in looking to see what you could do to solve the problem before looking for help. The more you post in the way of details, get better a chance there is that someone reading the posts will see something that will lead to an answer.

You made up solutions of varying concentrations of analyte. In volumetric glassware of by computing combined volumes of solution added by a pipetter? How did you add internal standard? An alequot of sample into a vial and an alequot of internal standard solution or by some other set of steps?

Injection volume? To help out the LC folks who might have experience with the compound, who I hope are still looking at the posts on this thread, give details about the solution(s) used for the isocratic gradient. And what is the sample solvent? The solvents are also important in consideration of ionization in the MS.

[smartin]

A lot of things to think about....

I can't either understand why the nominal values missed the target. But the areas are more or less homogeneous, either for the analyt and for the IS.... That's the reason why I thought the cause could be a "bad" adjusment of the calibration curve....

The IS is added by an aliquote, sample by sample.... As I did for other methods...

What about a full-loop injection vs partial-loop injection? I'm not sure about the difference...

[Don_Hilton]

Are you willing to share the numbers? Did the mean result at each level fail to fall within the expected range?