I am current trying to do quantitation of LC/MS assay. It seems the data fit better with quadratic curve. However, at the high end (ULOQ), a small change in area ratio yields large change in concentration. Is this indicate that the calibration range should bring down (the curve almost reaches plateau)?

The second thing is that I noticed that quantitation could be down with 1/x or 1/x2 weighed. I am not sure what exactly does the "weighing" means. Anyone give me some lesson? Also, I will be appreciated if someone could give me some references. I remember I did not learn this in my chemistry "statistics" course.

Thank you.

Paul

Dear Paul
a very good discussion on the use (and the purpose) of weighted regressions is in the paper by N.V. Nagaraja et al., Journal of Pharmaceutical and Biomedical Analysis 20 (1999), 433-438. They talk, however, only about linear regressions, which are more commonly used than quadratic ones.

With MS and MS/MS, as well as with any other detector, there is a limit to the dynamic range (in addition, there might be also the problem of overloading the LC column!). From my experience, UV has the largest range, followed by fluorescence and MS.

What I normally do if I need to calibrate over a range that is too wide for the detector linearity, is to split the calibrators into 2 subsets, a low calibration range and a high calibration range with partially overlapping calibrators. The difference between the 2 ranges is very often just the volume used for reconstituting the extracted samples (larger for the high calibration), while keeping all the other chromatographic and extraction parameters constant. Of course, one has to demonstrate also that extraction recovery is constant over such a wide range of concentrations.

I hope this might help you

A. Guenzi