LogP by HPLC

[Broesen1]

Hi.

I have setup a basic HPLC method with water and Methanol on a C18 that measure LogP.
To validate the method I have selected reference compounds from the chemistry bible.
The correlation was unbelievable. It also gave very good correlation to already known drugs.
Now suddenly I see that some new compound classes like sulfonamides or give bad correlation to shake or titration.
Now people don't trust the method any more. Someones thinks that if an analysis method is fast than it can't be accurate.
I can see that other people also find good correlation with or without adding octanol. Someone claim that you have to calibrate for each compounds classes. The last thing isn't an option if you are dealing with thousands of compounds. I personal find it a little bit odd that you can have method that only fit the reference compounds.
So what are you experience with Shake Flask, titration Sirius or HPLC?
It also looks that intramolecular hydrogen bonds place a larger role in the HPLC assay. Have you also seen that?

BP

[M Farooq]

Hi.
I have setup a basic HPLC method on a C18 that measure LogP.
To validate the method I have selected reference compounds from the chemistry bible.
The correlation was unbelievable. It also gave very good correlation to already known drugs.
Now suddenly I see that some new compound classes like sulfonamides or give bad correlation to shake or titration.

Unfortunately, there is no fundamental theory for retention on C18 bonded silica or any other phase for that matter. The logP behavior of your solutes cannot be easily predicted even in terms of classes of compounds. One cannot simply ignore the underlying silica because it is well known that the hydrophobic interaction, which your octanol water system measures, is also coupled with other phenomena such as cation exchange, hydrogen bond donor, hydrogen bond acceptor relationships from underlying silica.

You may wish to read the Hydrophobic Subtraction Model by Snyder et al, to see what other interactions are possible on the so-called reversed phases.

With that there is a sneaky side of the commercial nature of stationary phases in the sense that no nobody would be willing to share what is on their stationary phase. This undermines any fundamental attempt to understand the chromatographic behaviour. A C18 may or not have other functional groups as end capping agents etc or embedded polar groups to name a few.

[bunnahabhain]

Hi.

I have setup a basic HPLC method with water and Methanol on a C18 that measure LogP.
To validate the method I have selected reference compounds from the chemistry bible.
The correlation was unbelievable. It also gave very good correlation to already known drugs.
Now suddenly I see that some new compound classes like sulfonamides or give bad correlation to shake or titration.
Now people don't trust the method any more. Someones thinks that if an analysis method is fast than it can't be accurate.
I can see that other people also find good correlation with or without adding octanol. Someone claim that you have to calibrate for each compounds classes. The last thing isn't an option if you are dealing with thousands of compounds. I personal find it a little bit odd that you can have method that only fit the reference compounds.
So what are you experience with Shake Flask, titration Sirius or HPLC?
It also looks that intramolecular hydrogen bonds place a larger role in the HPLC assay. Have you also seen that?

BP

The HPLC method works better the more hydrophobic your substance is. Interactions with the silica material play a smaller role then. However, if you have a certain substance class you are working with, it is worth the effort to create your calibration from this class (in fact, this is what the respective OECD 117 guideline requires for a valid test). Check out the Log Kow data bank for members of your target substance class and use the recommended values to build up your reference library. Be aware that logP < 0 should not be measured using the HPLC method.
I never tried a polymer based column for Pow determination, maybe that would also help? See if my boss is willing to spend some money for such a column... BTW, are there C18 polymer columns except apHera?

Jörg