Quantitation by peak height

[Blazer]

I recently came across an HPLC method at my company that performs quantitation by peak height. This is the first method I've ever come across that performs calculations by height rather than area. Is this considered an acceptable method of quantitation? As far as I know there are no regulations requiring quantitation to be done by peak area. Any thoughts as to why a method would require the calculations to be done using height rather than area?

Thanks in advance for any thoughts you may share about this.

[tom jupille]

Peak height measurements have fallen into disuse because today's data systems do areas by default (and do a good job), but height may actually be more accurate than area when peaks are incompletely resolved, when there is severe tailing, or when the signal/noise ratio is low.

Regardless of area or height, the method will have to be validated to demonstrate that it works.

[Blazer]

Thanks Tom for the quick response.

[juddc]

I've used peak height occasionally. In a case where peak shape may be slightly variable, measuring height might be a bit more reliable. I have had an analyte that undergoes a keto-enol tautomerism and have found peak height to be more reliable for quantitative measurement than area because the two tautomers are rather broadly separated, yet "baseline" separation is not possible. Validation of the method was done with no difficulty.

[Blazer]

Thanks for the response juddc.

So using peak height to quantitate when peaks aren't completely resolved or if the peak shape is questionable seem to be the primary reasons, based on what I'm reading here, as well as some other online research. It just was a bit shocking to me since the method in question - an HPLC assay for biotin - seems to show good peak shape and doesn't show any resolution issues (at least none that I've seen thusfar).

Thanks again for your thoughts and input everyone.

[HW Mueller]

Maybe you saw only chroms of standard or nearly ideal samples. When the matrix can not be completely eliminated one might have a lot of overlap with it. I used the peak hight method in HPLC of cortisol in blood even though the internal standard was separated by about 10 min, but the matrix caused ovrlapping that didn´t allow the area method, especially at low cortisol conc.

[lmh]

I would worry that if the peak shape is questionable for reasons related to the matrix, then you need to be certain that the peak shape remains constant. Otherwise, if the peak gets wider, it's also likely to lose height.

Of course this goes back to Tom's point that whatever you use as signal, the assay needs to be validated, and presumably (though Tom didn't go so far directly), if it passes all sensible validation, then the method is fine whatever aspect of the peak you quantified.

[tom jupille]

if it passes all sensible validation, then the method is fine whatever aspect of the peak you quantified.

Yep, that was the point.

Stripped of all the regulatory baggage, validation is simply demonstrating that a method does what it purports to do.

In the specific case of area vs height measurements, we can generalize about which is better when. If it's a real question, try crunching the data both ways and see. I suspect that in most cases either way would work acceptably.

[juddc]

Actually my peak shape was quite nice when examined at full scale, but what happened was that the keto-enol bit caused the peak to be somewhat fronted and being that the two tautomers were separated by 3 minutes or so, the software (as it was set) had a hard time accurately discering the peak liftoff for the tautomer that I wanted to measure. Sometimes it would integrate a good chunk of what is essentially baseline, other times it would grab the front of the peak accurately. I didn't really want to mess with integration parameters too much because other peaks nearby (following) were being accurately integrated on a consisitent basis. Measuing height gave significantly better injection to injection reproducibility than measuing area, thus the reason for doing it.

[Blazer]

Maybe you saw only chroms of standard or nearly ideal samples.

It turns out this is the case with this client test method. The offical document shows example chromatograms of standards and only two sample types, but this method is designed to be used for a variety of products that contain biotin. When I looked back at the validation report which had example chromatograms for all the different products, resolution issues became apparent.

[gtma]

If a method was validated using peak height only with no data generated using peak area. Can the validation data support the same method using peak area? My feeling is accuracy, precision and linearity will need to be reassessed. Do robustness and solution stability studies need to be reassessed also? Thanks in advance.

[tom jupille]

My feeling is accuracy, precision and linearity will need to be reassessed. Do robustness and solution stability studies need to be reassessed also?

I agree about solution stability. You could make a case for robustness needing revalidation, because integration may be affected by things like peak shape and critical resolution.