Why LogD Matters / ΔLogD Cheat Sheet
This article explains what LogD is, why LogD (or LogP) is important in drug discovery, rookie mistakes in drug discovery that come from overlooking LogD or LogP, and contains a LogD reference poster that shows the estimated amount of potency attributable to lipophilicity considerations alone.
Lipophilicity is a drug discovery concept that I didn't take seriously enough when I first started. It made me focus on the wrong molecules and caused me to design and synthesize buckets of compounds that in retrospect were clearly headed nowhere. This post is here to save you that time.
If you're an experienced drug hunter, you can share this with your junior colleagues so you can stop repeating yourself -- skip down to the LogD cheat sheet section below highlighting an excellent resource by my Genentech colleagues, Matt Landry and James Crawford. (Landry, M. and Crawford, J. "LogD contributions of substituents commonly used in medicinal chemistry." ACS Med. Chem. Lett., 2019, doi: 10.1021/acsmedchemlett.9b00489) If you're newer to drug discovery, read on to claim your time back!
What is LogD and Why is LogD Important?
As a measure of molecular lipophilicity, LogD is among the most important measurements in the drug hunter's repertoire because it's our control experiment (Figure 1). Like actin bands on Western blots, placebo controls in clinical trials, or inflation for currency, changes in LogD tell medicinal chemists whether the affinity changes we observe as we modify our compounds are significant or “real.” ("Real" as opposed to "nominal" by analogy to inflation. The increased affinity is real in the literal sense, but whether it's a "real" gain that's worth the cost of added lipophilicity is what I mean here.)
If you want to startle a chemist, the next time you’re in a meeting with one and they say “we made change X to the molecule and this increased potency by Y,” ask: “but how did change X affect LogD?” They’ll either be (a) impressed that you’re fluent in medicinal chemistry (b) nervous that you’re on to them or (c) oblivious to LogD in which case you should find another chemist.
Figure 1. The partition coefficient D is the ratio of organic and aqueous concentrations of a molecule that is allowed to equilibrate between a bi-phasic pH 7.4 water layer and a non-polar 1-octanol layer. Hence, a molecular change that increases LogD by 1 unit (e.g. -n-propyl) results in 10x greater ratio of compound found in the lipophilic compartment vs. in bulk water. LogD measurements are used as gauges of compounds’ intrinsic lipophilicity.
If a molecular change results in a compound with 10x greater affinity to your target of interest, it might be because the change did something productive like introduce a hydrogen-bond or enforce a productive molecular conformation. However, if the change also made the compound 10x more lipophilic relative to the starting molecule (ΔLogD = +1), it’s much more likely that the apparent increase in target affinity is primarily due to the increased desire of the new compound to stick to any lipophilic region, which happens to include your binding site.
So while the new greasier compound now looks 10x better by an early biochemical assay (Kd, Ki, IC50…) where there’s nothing but buffer, target, substrate, and drug, the new compound is likely no closer to a drug candidate than your starting compound (Figure 2).
The increase in lipophilicity of the new compound also results in increased affinity to all other lipophilic pockets where the compound can fit, including indiscriminate binding sites in plasma proteins like albumin (reducing concentration of free drug available to engage the target in whole blood assays) and binding sites in undesired off-targets (e.g. hERG, BSEP, kinases, …). (Strictly speaking it’s water’s stronger preference to be in water that is the main “driving force” for this “hydrophobic” effect.)
Figure 2. A change which increases molecular lipophilicity corresponding to a change in LogD = +1 has multiple expected consequences. (Top left) Potency in biochemical assays is expected to increase if the molecular change is made in a region that binds to a lipophilic pocket. (Top right) Potency is also expected to increase against off-targets which have lipophilic cavities. (Bottom left) Affinity for serum proteins with lipophilic binding sites is expected to increase, lowering the ratio of free drug (fraction unbound, fub) to plasma-protein bound drug (fraction bound, fb). (Bottom right) Drugs which have greater affinity to their targets due to increased lipophilicity are not expected to be safer than their parent (no change in selectivity vs. off-targets) and are not expected to be more active in real biological settings where non-target compartments like plasma protein compete for drug.
Common Drug Discovery Mistakes from Overlooking Lipophilicity
The increased lipophilicity also results in increased binding to the non-polar active sites of metabolic CYP enzymes, reducing metabolic stability and increasing unbound clearance in vivo. Furthermore, more lipophilic drugs are typically less soluble in water, limiting bioavailability and maximum absorbable dose. Hence the hallmark of the novice medicinal chemist is to chase potency by continuously adding lipophilic groups, resulting in promiscuous, unstable compounds that look more potent in early assays but go nowhere in vivo. I was guilty of this when I first started -- nothing made a newbie like me happier than a nanomolar lipophilic amine!
Critically, lipophilicity's nearly universal influence on drug properties can cause even very experienced project team members to see "ghost trends" and draw incorrect conclusions. A few common examples are shown in the table below.
Table 1. Examples of common observations leading to erroneous conclusions due to LogD as a common "ghost variable."
It's important to clarify that I'm not saying all lipophilicity and LogD gains are bad - the ideal drug typically has a LogD in the sweet spot of +2-3, giving it the passive permeability properties necessary for cell activity, bioavailability, and to avoid the effects of many transporters. As an old boss of mine used to say, sometimes you need a little lipophilic "spice" to put a drug right where it needs to be potency and property-wise. Sometimes you get lucky, and your drug can tolerate a little more LogD without increasing all the other unfavorable properties. In general, I think it's helpful to imagine that you start each drug discovery campaign with only 3 LogD units to spend. Once you've spent them, if you want to add any more lipophilicity to your molecule, you've got to take it back from somewhere else in the molecule.
ΔLogD Cheat Sheet
Having a sense of how different molecular changes are expected to influence LogD is therefore critical to interpreting assay data and designing new rounds of molecules. Many large companies have automated workflows which experimentally measure LogD on virtually every new compound that is registered, but getting comprehensive LogD’s can be an expensive exercise for smaller companies that have to outsource LogD’s at rates of ~$200-$500 a compound.
Recently my Genentech colleagues Matt Landry and James Crawford published an excellent summary of expected changes in LogD from common substituents used in medicinal chemistry based on data mining a massive set of experimental measurements on matched molecular pairs from Genentech's database.
Here is a LogD cheat sheet created from their work, summarizing how much affinity is expected from a change based on lipophilicity considerations alone. For a methyl group for example, the median LogD change is about 0.3, indicating that you'd expect about a 2-fold increase in lipophilic potency due to the methyl group's contribution to LogD. So chances are, if a methyl group increases activity in the ballpark of 2x, it's not magic, it's grease -- BUT, if the addition of a methyl group increases potency by 10x or more, it probably really is "magic," generating improved binding affinity through effects like ligand conformational changes, solvent displacement, or protein conformational changes. Of course, these prospective estimates can’t replace actually measuring ΔLogD on your particular molecules, so when in doubt, get your compounds tested.
ΔLogD Cheat Sheet
Now, thanks to this table, instead of just asking your chemist how change X influenced LogD, you can really be a wise-guy/gal and declare, “but you’d expect change X to result in potency-increase of around Y-fold due to lipophilicity considerations alone, wouldn’t you?”
That’ll really raise eyebrows.
Purchase high resolution posters in our gift shop.
Further Reading
For more reading on the importance of lipophilicity metrics in drug discovery, I recommend the following recent articles. Let me know if you have any other favorite articles on this topic and I’ll put in the effort of linking them here.
This recent article by Pfizer scientists nicely visually explains the concept of “lipophilic efficiency” (LE, or LLE for AstraZeneca folks), which is a way of “normalizing” your activity measurements against lipophilicity to see if you’re making real progress in drug design: Johnson, T. W.; Gallego, R. A.; Edwards, M.P. “Lipophilic Efficiency as an Important Metric in Drug Design.” J. Med. Chem. 2018, 61, 6401-6420. doi: 10.1021/acs.jmedchem.8b00077
This short editorial by Pfizer scientists further emphasizes the importance of LE and cites several useful references linking lipophilicity to undesirable drug properties (in vivo tox, poor solubility, etc.): Freeman-Cook, K. D.; Hoffman, R. L.; Johnson, T. W. “Lipophilic efficiency: the most important efficiency metric in medicinal chemistry.” Future Med. Chem. 2013, 5, 113-115. doi: 10.4155/fmc.12.208
