ADMET@TheSpeedOfThought

Accurate estimation of ADMET effects still remains the Waterloo of drug development, with up-to 40% of promising candidates failing in clinical trials because of unfavorable pharmacological properties. Over the last few years many algorithms and descriptors have been developed and implemented into programs- many of them proprietary- to predict ADMET. However at some point these programs become far from intuitive, being of no real help to medicinal chemists engaged with actual lead design.

Paul Gleeson from GSK has come up with a set of rules of thumb for remembering the different properties of drugs that influence their ADMET behavior. He has used large databases of molecules, sound statistics and confidence intervals and GSK's internal studies to derive these rules. In my opinion he has done us all a service, because most pharmaceutical companies would want to and in fact do keep such data proprietary. He narrows down these properties to three important ones which medicinal chemists are comfortable and familiar with- molecular weight, clogP (lipophilicity) and ionization state (+ve, -ve or neutral). The major ADMET parameters he looks at are:

1. For Absorption:solubility, permeability and bioavailability
2. For Distribution: volume of distribution, blood-brain barrier penetration and plasma protein binding
3. For Metabolism: clearance, half life, P450 metabolism, hERG inhibition, P-glycoprotein efflux

While some of the rules are what they are supposed to be- intuitive- there are also some revelations. I will let the reader muse over the details of how the three aforementioned properties affect the ADME parameters. The author also interestingly considered the effect of changes in clogP- important for medicinal chemists- on the properties. For now I will list a few effects I personally found most interesting:

1. Basic compounds have a larger volume of distribution (which means a larger half-life for a given clearance rate) than acidic compounds. That's because basic compounds can interact with the negatively charged head groups of phospholipid membranes and get distributed easily throughout the body. Acidic compounds in contrast bind to positively charged residues on ubiquitous albumin, which limits their volume of distribution. Thus acidics show greater plasma protein binding.
2. Because of the same property basics also can pass more easily through the gut wall membrane (permeability). However, lesser plasma protein binding can also increase the clearance rate for basics. Thus there's a balance between volume of distribution and clearance, and since half life is given by

half-life = 0.693 * Volume of distribution/Clearance

one often has to face a compromise. In general having basic side-chains can be advantageous (also see effects on CNS penetration below), notwithstanding higher clearance rates. There is an instructive example that comes to my mind; the anti-TB antibiotic rifamycin has an unfavorable Vd which can be improved by adding a basic side chain, thus converting it to rifampicin with a radically better Vd and half-life.

3. As for CNS penetration, it's well-known that small, non-polar molecules easily make their way across the blood brain barrier (ask any druggie). However the study shows that again, basic molecules are on average more CNS permeable than neutrals, followed by acidics. This trend mirrors the permeability trend above. CNS penetration may also be complicated by active transport mechanisms that are hard to predict. For non-CNS drugs, preventing CNS penetration is what's most important. In general I would stay away from making predictions about CNS effects of drugs.

4. hERG inhibition: The human-ether-a-go-go-related-gene ion channel is notorious for flagging toxic molecules in studies. Excessive hERG inhibition leading to QT interval prolongation of heartbeat can assuredly be a death warrant for your molecule, not to mention for yourself. For some years now people have been trying to figure out pharmacophores for hERG inhibition that would enable them to find common features among hERG-unfriendly molecules. To my knowledge nobody has been spectacularly successful although there are have been some interesting results. It is generally accepted now that basic molecules tend to block hERG more than neutrals or acidics. Interestingly as this study shows, effects of clogP on hERG inhibition are also the greatest for basic molecules.

5. Finally, there is no meaningful relationship between molecular weight and many of these factors. Parameters like permeability will naturally be greatly influenced by Mol. Wt. I think it's more accurate to say that relationships between MW and these parameters will be masked by factors like clogP and ionization state.

Readers are urged to go through the article for more details. The sections on PgP and P450 inhibition are interesting and these properties are extremely important (Taxol for example is pumped out by PgP in resistant cells) but it's also difficult as of now to come up with predictive models for these. It's still very difficult to predict ADMET for budding clinical candidates, mainly simply because the body is still too complex a creature for us mortals to ponder. But medicinal chemists will greatly benefit from intuitive rules, which even if they break down in certain scenarios, do provide a rough and ready guide for checking off factors from your list of adverse ADMET effects. Such studies would help.

Gleeson, M.P. (2008). Generation of a Set of Simple, Interpretable ADMET Rules of Thumb. Journal of Medicinal Chemistry, 51(4), 817-834. DOI: 10.1021/jm701122q