variation slope of calibration curve

[Fluterd]

Hello,

I constructed a calibration curve for validated LC-MS/MS method in urine (at 6 concentration levels, n=6), using an internal standard (that works great, MR and RE are about 100%).
Let's say the equation for the compound is y= 20x-0.01 (weighted (1/x²) linear regression), R²>0.99.

Now, some months later, I am doing the same in blood. Again the internal standard works fine (ME and RE both around 100%). I use the same calibration concentration levels and the same weighted linear regression model. Everything works fine in blood. However, when I just compare the equations in blood and urine, the slope is totally different, (20 is blood versus 0.1 in urine).

Is this normal?

[Alp]

I'd like to know what are meant by MR and RE?
Some sort of recovery?

You might want to think about what the equation is saying. It depends on how you are plotting the data. Are you doing Ratio drug area/IS area vs conc? Y=drug/is, x=conc.

If you are using the same method, only difference is the matrix, same concentrations, same amount of IS, same extraction procedure, same mobilephase, same column... why is your drug/IS response a factor of 200 higher? (y=20x+b vs Y=0.1X+b) What could cause this to happen? This is the key to the solution.

Maybe doing a blood and urine at the same time and have the samples injected back to back (or better, randomly assorted in the same run) can help narrow down between system variables and method variables.
In this case I want to be able to look at either matrix as the standard curve.
I would also want to compare analyte peak areas between the two methods (does a blood standard give same response an equivilant urine standard?) and also the IS areas between the two methods.

Alp

[zokitano]

Alp has already given you a hint about the observed "anomaly" with your method.
I agree that probably the matrix effects due to the different sample matrices are giving you different calibration slopes for your analytes. You developed and validated the method initially for determination of your analytes in urine, right? And now you're transferring the method for determination of your analytes in different matrix, i.e. blood. You're aware that the different compositions of the matrix in both cases could have substantial influence on the ionization.
Having 20 times higher slope for the calibration in blood, could suggest ionization enhancement if the area of the internal standard is not changed due to the matrix (we assume only change of the analyte peak area), or this could be also a case of internal standard ionization suppression - having the analyte peak area consistent between the standards and samples.
But, also you have lower slopes even for the urine samples, for which you initially have developed the method. This also could be a issue of matrix effects (in this case probably some ionization suppression of the analyte, if the IS peak area is consistent between the Std and samples).
You should observe the change of the IS peak area between the std and samples in both cases (blood and urine) in order to diagnose your problem, as suggested by Alp.

Also, it is a practice in biochemical/toxicological LC-MS method development and validation to include at least 6 different sources of urine or blood in order to test the robustness of the method with different samples, in order not to face this kind of problems in the future. If you find a ionization suppression or enhancement with one/some of your samples in this stage you should tweak the chromatographic conditions or choose other ionization method (from ESI to APCI for instance) in order to avoid the matrix effects during the application of the method.