External stds reference

[dth]

Does anyone have a reference for the use of external stds rather than the use of and internal std for HPLC.
Thanks

[tom jupille]

External standardization is the "default" technique in HPLC, so I'm not sure why you'd need a reference, but:

Introduction to Modern Liquid Chromatography, 2nd ed.; pp 549-552; L.R.Snyder & J.J. Kirkland; Wiley-Interscience (1979). ISBN: 0471038229

The book is still in print (even after 25 years!).

[dth]

Thanks Tom. I don't have the 'Introduction' book but I do have 'Practical HPLC Method Development' and I found a reference in that.

I think a lot of people would argue that using internal stds is the 'default' - I'm trying to justify using external stds which is the practice in our lab but difficult for some reviewers to accept. The only argument I have is that our sample prep is very simple, but I'm trying to find a better argument in the literature. If you have any further references I'd be grateful for them.

Diane
--------------

[Mark]

For a better reference, how about the USP? The vast majority of the LC and GC methods in the USP are external standard quantitation. If it is good enough for the USP it should be acceptable to most other reviewers (my humble opinion). The company O work for has submitted numerous validated methods all using externhal standards to the FDA without any comments about not using internal standards. I agree that in practice an internal standard method is generally better but not always practical in terms of time and effort to develop.

Regards,
Mark

[tom jupille]

The choice of internal vs external standardization depends to a large extent on the dominant source(s) of error in the analysis.

If the errors in all peaks are correlated (i.e., an error in one peak implies a proportional error in other peaks), then the use of internal standards will improve precision. In the limiting case, if errors are perfectly correlated, then they will exactly cancel.

On the other hand, if the errors are uncorrelated (i.e., the error in one peak is unrelated of the errors in the other peaks), then the use of internal standards will degrade precision. In the limiting case, the overall result of independent errors is the square root of the sum of the squares of the individual errors.

To oversimplify a bit, if the dominant source of error occurs before the sample gets onto the column (e.g.., sample workup or injection), then internal standardization will give better precision. If the dominant source error occurs after the peaks have begun to separate (e.g., integration problems due to tailing or baseline noise), then external standardization will give worse precision.

[Bill Tindall]

There are yet other considerations. Use of an internal standard involves measureing the area of an additional peak. This fact increases the chances of an interference affecting results, a real issue in complex samples. Second, because the results of two area measurements are required to calculate a result, the precision will be poorer, all else being equal.

I don't believe an internal standard should ever be used until it is demonstrated statistically that its use improves precision. But, you are right that some people use (incorrectly in my opinion) the internal standard as their default approach.

A serious misuse of internal standards can be to compensate for analyte degredation, or other losses. In these cases it is quite possible to get no, or even reverse, correlation between what happens to the analyte and what happens to the internal standard. Yet one seldom sees in the literature any rigorous investigation to conclude that the internal standard was suitably chosen.

A common "justification" for internal standards is to account for injector variability. But, if you need an internal standard for this factor it would be better to fix the injector.

[Consumer Products Guy]

We use almost exclusively external standard quantitation for both HPLC and GC, even for pharmaceutical assays; today's autosamplers are very good, and take away operator bias in the injection step. I've seen a lot of procedures which use internal standard that are usually old procedures from before autosampler days, just like I'm clued to an out-of-date procedure when detection wavelength of 254 nm is used, tells me that a single-wavelength detector was likely used.

[dth]

Thank you all for your input.
Question regarding USP, are there specific guidelines for methods development and validation? I'm only used to the FDA guidelines and I can't find equivalent on the USP website.
Thanks

[Consumer Products Guy]

These are included in the NF part (the rear part) of the full USP/NF. Not everything is always available for free on the internet.

[HW Mueller]

We have been through this before, but it seems it also should be pointed out again that nowbody prevents anybody to use the external and internal st. method simultaneously. This can remove the possibility that the internal st. deteriorates things, but it can show immediatly (in every injection) that something is going wrong with the analysis.

[sluggo]

If you were having problems getting full recovery of your analyte from SPE prior to analysis using HPLC, could it make sense to inject an internal standard prior to SPE?

thanks

[Mark Tracy]

Just to add fuel to the fire...

The term "internal standard" has been used to mean two different things. One is a calibration substance added to the prepared sample at the last step. This is used to overcome instrumental variability in the determinative step of the analysis: injector problems, detector calibration drift and the like. These are rarely a problem in modern HPLC/UV methods, but still can plague LC/MS. As previous authors pointed out, unnecessary use of IS in this context can degrade precision. And as some other authors pointed out, there are some types of errors that IS can't fix at all, such as non-linear detector response due to coeluting substances.

Then there is the addition of a standard to the raw sample at the beginning of the sample prep. The term "surrogate" has been introduced to replace IS in this situation. Here the intent is to monitor recovery of the entire analyis. The use of surrogate recovery to correct analytical results is controversial; expect to rigorously justify any such use, and be sure to ask your regulatory experts, as the policies vary. No one will argue with surrogate recovery as a QC tool to flag a suspicious result.

One notable exception is isotope-dilution MS, but that is way off-topic.

One other approach that I have used when the sample preparation is tricky is to carry the calibration standards through the same process as the samples, and use external calibration. I used to do selenium analysis (not by LC) that way because nothing else worked.

[Ron]

The use of internal standards is usually not needed with modern instruments. Internal standards were often used 15 to 20 years ago because instruments could show considerable drift in response over the course of a day. As noted above, autosamplers are in common use today and help minimize injection variation. I use external standard unless there is a significant improvement in the precision, which is rare, or a regulated method requires use of internal standards, which unfortunately is not as rare.

I agree with Mark that the addition of a known quantitiy of a standard prior to processing is a surrogate, not a true internal standard. When I was in school an internal standard was defined as a constant amount of a compound added prior to analysis to compensate for instrumental instability. A compound added prior to processing could be used to evaluate the sample processing, but was typically used to identify suspect sample preparation, not to correct for any problems in the processing. Unless the compound spiked into the samples is very similar to the target analytes the recovery can be different and precision can be adversely affected. It is not uncommon when using a compound added before sample preparation as an internal standard to see precision of analysis of some target compounds improved, and the precision of other target analytes to be worse, as Bill pointed out.

Mention was made of most USP methods being written as external standard methods, but in the general chromatography section the use of internal standards is allowed in a method written as an external standard method as long as the precision criteria of the method are met. I have used internal standards in some USP methods to compensate for matrix effects, and in a method with 5 analytes I had to have 2 internal standards because one analyte behaved differently from the other 4, and the precision of this target was degraded with the first internal standard used.

[tom jupille]

This thread continues under the topic "Internal Standards for LC-MS" on the "Hyphenated Techniques" board.