We purchase a human recombinant trypsin and need to assay the purity of the incoming material. I have developed an HPLC method for the assay and need to validate it. How do you demonstrate accuracy for a method when there is no standard available? Up to this point, I have been using a purified porcine trypsin from Sigma as a standard for quantitation.

Can't you get a analysis certificate for that Sigma product?

No purity is given on the C of A. The C of A for this product uses the result of an enzymatic assay. The value for the lot that I have is 16,000 BAEE units/mg of protein.

We have had a similar problem with an antibiotic that is a mixture of about a dozen analogs. The supplier quoted "98% purity by HPLC" on their C of A, ie the summed peak integration area for the antibiotic mixture was >98% of the peaks detected at that wavelength (which is what we found). However, on a weight/weight basis, we found that the antibiotic mixture was only 70% purity.
We have addressed this problem by purifying the antibiotic mixture by solvent extraction, prep HPLC, etc, etc, and monitoring the HPLC integrated area for the toxins. Quantitation is by UV(DAD), with ELSD in line to look for non-UV adsorbing components. LC-MS is also used to confirm the identities of each peak. When the integration area vs weight plateaus as we clean up the antibiotic, we have assumed that we are approaching 100% purity. This purified product is then used as our standard.
Of course, there are no guarantees that we are in fact close to 100% purity. Over the years, our "gold standard" has been replaced several times as we have improved our purification procedures. In the absence of certified standards it was the best we could come up with. If anyone has a better solution, I'd love to hear it!

I have never dealed with this kind of problem but I think that the indispensable detectors for this kind of job would be the MS, ELSD and chemiluminescence nitrogen detector (CLND). The MS for identification reason, the ELSD is a universal detector (for non volatile compounds) and "equimass" capabilities, the CLND can detect any nitrogen containing compound and has equimolar capabilities (that is why these last two are used a lot in combinatorial chemistry).

For example if your compound contains nitrogen in its molecule you could run it in HPLC and see already if you have any nitrogen impurities (other peaks). As this detector is equimolar you could then use another compound (caffeine for exemple) to make a calibration curve and check if the injected amount in mM corresponds in the the theoretical purity or you can even determine the purity assuming that you don't have anything that is coeluting with your compound of interest.

In the same HPLC method the ELSD will monitor any non-UV adsorbing compunds as described by Neil and of course any compunds that CLND might miss (non-nitrogen containing compounds) or the MS (compounds that have not been ionised).

You can find several bibliographic examples of the use of ELSD and UV for this kind of purposes.

In the case of CLND I have used its equimolar capability in order to calculate the amino acid stoichiometry of peptides (after acid hydrolysis) without the need of calibration curves. If you are interested you may see: K. Petritis et al. LC GC Europe 14 (2001) 389-395 (I can e-mail you a pdf file upon request).

You may also look at Taylor et al. Anal Chem. 70 (1998) 3339-3347 which is the article which introduced the CLND it the combinatorial chemistry.

Hope the above helps,

Kostas