Is internal standard inevitable?

[Karvezide]

Hi, everybody!
I have to quantitate a contrast agent in serum using RP-HPLC.

To constitute the standard, I use the agent from the suplier.
I deproteinate the sera samples using 4 Vol. of 5% perchloric acid and then inject the supernatant for the analysis after centrifugation.

To examine the accuracy of the standard, I constituted separately 4 QC-samples and calulated their concentration by means of the satndard.

The linearity of my standardcurve is quite well (> 0,999) and the relative error (in %) between the back-calculated value and the theoretic value of my QC samples are < 1%. So I thought an added internal standard was not neccessary.

But a professor from other working group who has worked with HPLC for a long time has other opinion and has argued that an internal standard is inevitable, at least it should be added to the supernatant before the analysis.

What is your opinions? I really appreciate your help and advice.

Thanks a lot

[Don_Hilton]

Internal standard is a good idea. There will be variations over time - and the internal standard will correct for variations as well as indicate problems. If there is variation because of a slight variation in injection volumne or dilution of the sample during preparation (after the internal standard is added, the internal standard will vary with the analyte - and the ratio of analyte to internal standard will be constannt.

If monitor the internal standard area during your review of results, large variation of the internal standard area will indicate that there is a problem with the sample or the instrumentation, which needs to be resolved and the sample rerun.

[thohry]

Internal Std methods always give better results, especially the matrix is kind of serum.

[HW Mueller]

This has been discussed extensively here. Some have pointed out that an internal standard introduces a further variable, and thus a further uncertainty. If one fears that one can use the IS as a very nice check on each run, as Don_Hilton suggests, and leave the quantitation to the calibration of the external standard.

[tom jupille]

To elaborate a bit on the previous comments, when you make a ratio of two quantities (i.e., the analyte area to the IS area), any *correlated* errors will tend to cancel, so that the ratio will contain less error than either of the two original quantities. However, *uncorrelated* (independent) errors will accumulate, so that the ratio will contain more error than either of the two quantities.

Whether an IS helps or hurts depends on the dominant sources of error. If the dominant sources of error occur when the IS and analyte molecules are together (e.g., dilution, or injection volume), then the effects will tend to be correlated and the use of an IS will help. However, if the dominant sources of error occur after the IS and analyte molecules have separated (e.g., integration errors or signal/noise issues), then the effects will tend to be independent and the use of an IS will hurt.

As HW pointed out, if you include an IS, you always have the option to do the quantitation without it.