Lipophilicity: logP, logD, and why they drive ADMET

If you could keep an eye on only one physicochemical property across a medchem campaign, it should be lipophilicity. Nothing else has its reach: the same number that governs how well a compound dissolves also governs how well it crosses membranes, how fast the liver clears it, how promiscuously it hits off-targets, and how likely it is to block hERG. Push it too high and every downstream property degrades at once. This is a practical guide to the two numbers people actually use — logP and logD — and to the rules of thumb that keep a series in safe territory.

logP vs logD: they are not the same number

Both describe how a molecule partitions between octanol (a stand-in for membrane lipid) and water, but they measure different things.

• logP — the partition coefficient of the neutral species only. It ignores ionization. For a compound with no ionizable groups, logP is the whole story.
• logD — the distribution coefficient at a specified pH (almost always logD_7.4, physiological pH). It sums the neutral and ionized forms, so it captures what the molecule actually does in blood and gut.

For a neutral molecule, logP and logD are identical. For an acid or base, they diverge sharply: a basic amine that is mostly protonated at pH 7.4 has a logD far below its logP, because the charged form stays in water. That gap is a lever. Adding a basic nitrogen to dial down logD is one of the most common ways medicinal chemists rescue solubility without touching the neutral-form lipophilicity that drives target binding. The flip side: the protonated base is exactly what gets you into trouble with hERG. logD is the number to track for ADMET; logP is the number that often correlates with the binding pocket.

Why it touches everything downstream

Lipophilicity is upstream of most of the ADMET properties that kill compounds, which is why optimizing it pays compounding dividends.

• Aqueous solubility — falls roughly log-linearly as lipophilicity rises. High-logD compounds crash out, limiting oral absorption and complicating formulation.
• Permeability — rises with lipophilicity up to a point, then membrane retention and efflux take over. Permeability and solubility pull in opposite directions on the same axis, which is why a mid-range logD is the sweet spot for oral drugs.
• Metabolic clearance — the CYP450 enzymes have lipophilic active sites, so more lipophilic compounds are generally better substrates and clear faster. Lowering logD is a standard tactic to improve metabolic stability.
• hERG and off-target promiscuity — lipophilic, basic compounds hit the hERG channel and a long tail of unrelated targets. Promiscuity scales with lipophilicity; cleaner selectivity tends to live at lower logP.
• Plasma protein binding — climbs with lipophilicity, lowering the free fraction available to engage the target.

The 3/75 rule and the lipophilicity ceiling

The most-cited single data point here comes from Hughes et al. (2008), a Pfizer analysis of in vivo tolerability across 245 preclinical compounds. Molecules with ClogP > 3 and TPSA < 75 Ų were markedly more likely to show toxic findings than molecules in the opposite corner. The practical heuristic that fell out — keep ClogP under ~3 and polar surface area above ~75 — became a fast triage filter for “greasy and non-polar means risky.” It is a correlation, not a mechanism, but it has held up well enough to earn a permanent place in property dashboards.

This sits alongside the broader lesson from Leeson & Springthorpe (2007): mean lipophilicity of marketed oral drugs has crept upward over decades, and that drift correlates with attrition. Lipinski’s Rule of Five already capped logP at 5, but the modern target is more conservative — most oral programs aim to land logD_7.4roughly between 1 and 3.

Lipophilic efficiency: spend your lipophilicity wisely

Raw potency is easy to buy by adding grease — a bigger, more lipophilic compound almost always binds tighter, but it does so by burying hydrophobic surface, not by making a better-quality contact. Lipophilic ligand efficiency (LLE, sometimes LipE) corrects for this:

• LLE = pIC₅₀ (or pK_i) − logP — potency normalized against the lipophilicity that bought it.

A series whose potency is climbing only because logP is climbing shows flat LLE — a warning that you are inflating affinity with liabilities attached. High-quality optimization raises potency and LLE together, which means the new binding is coming from specific polar contacts rather than from generic hydrophobic burial. Most teams target LLE in the 5–7 range. Tracking it alongside the docking score is one of the most useful habits in lead optimization.

Try the docking yourself

Open Studio and dock a candidate against your target — the ADMET panel reports predicted lipophilicity alongside the docking score, so you can read potency and property risk on the same screen. Pair the predicted logP with the binding score to estimate lipophilic efficiency on the fly: if a more potent analog also carries a higher logP, your LLE may not have moved at all, and that is exactly the trap this post is about. The panel also flags the hERG and solubility liabilities that ride along with high lipophilicity.

Liganx is molecular docking online: free, browser-based, and set up to show structure-based potency and property risk together. If you want to run molecular docking while keeping an eye on lipophilicity, that is the fastest path.

Primary sources

• Hughes JD, et al. Physiochemical drug properties associated with in vivo toxicological outcomes. Bioorg Med Chem Lett 18, 4872–4875 (2008). doi:10.1016/j.bmcl.2008.07.071
• Leeson PD, Springthorpe B. The influence of drug-like concepts on decision-making in medicinal chemistry. Nat Rev Drug Discov 6, 881–890 (2007). doi:10.1038/nrd2445
• Gleeson MP. Generation of a set of simple, interpretable ADMET rules of thumb. J Med Chem 51, 817–834 (2008). doi:10.1021/jm701122q