A potent compound is just the starting point. To become a drug, it has to be absorbed, reach the target tissue, persist long enough to act, be eliminated safely, and avoid toxic effects. ADMET — Absorption, Distribution, Metabolism, Excretion, Toxicity — describes these properties. ADMET failures account for a major share of late-stage drug attrition.
A: Absorption
Most drugs are taken orally and must cross the intestinal wall to reach the bloodstream. Key factors:
- Solubility: Drug must dissolve in GI fluids. Poorly soluble compounds have erratic absorption
- Permeability: Crossing the intestinal epithelium. Lipinski’s rule of five (MW < 500, logP < 5, HBD ≤ 5, HBA ≤ 10) historically guides oral drug-likeness
- P-glycoprotein efflux: P-gp pumps drugs back out of cells. P-gp substrates often have low oral bioavailability
- First-pass metabolism: Liver metabolism on first passage from the gut
Common assays: Caco-2 monolayer permeability, PAMPA (parallel artificial membrane permeability assay), kinetic solubility, animal PK with oral and IV dosing.
D: Distribution
Once in the bloodstream, drug distributes between plasma and tissues. Important parameters:
- Volume of distribution (Vd): Apparent volume the drug occupies. High Vd suggests tissue binding
- Plasma protein binding: Most drugs bind albumin or α1-acid glycoprotein. Only the unbound fraction is pharmacologically active
- Tissue penetration: Drug reaching the site of action — particularly critical for CNS drugs (blood-brain barrier) and antitumor agents
M: Metabolism
Most drugs are metabolized in the liver, primarily by cytochrome P450 enzymes. Considerations:
- Metabolic stability: How fast the drug is metabolized. Determines half-life and dosing frequency
- Major metabolic enzymes: CYP3A4 (most drugs), CYP2D6, CYP2C9, CYP2C19, CYP1A2
- Drug-drug interactions: Inhibition or induction of CYP enzymes affects co-administered drugs
- Active metabolites: Some metabolites contribute to or replace the parent drug’s activity
Common assays: Liver microsome stability, hepatocyte metabolism, CYP inhibition panel, time-dependent inhibition assays.
E: Excretion
How drug or metabolites leave the body:
- Renal excretion: Filtration, secretion, and reabsorption in kidneys
- Biliary excretion: Conjugated drug or metabolites secreted into bile
- Clearance (CL): Volume of plasma cleared of drug per unit time. Combined with Vd, determines half-life
Renal impairment changes excretion of many drugs and often requires dose adjustment.
T: Toxicity
The most heterogeneous category. Common ADMET-related toxicities:
- Hepatotoxicity: Drug-induced liver injury (DILI) — a major cause of withdrawal
- Cardiotoxicity: hERG channel inhibition leads to QT prolongation and torsades de pointes. hERG is screened early in drug discovery
- Genotoxicity: Mutagenic potential — Ames test, micronucleus assay
- Reproductive toxicity: Pregnancy and developmental effects
- Immunogenicity: Especially for biologics
- Off-target pharmacology: Activity at unintended targets
In silico ADMET prediction
Computational tools predict ADMET properties from structure:
- SwissADME: Free, web-based property prediction
- QikProp (Schrödinger): Commercial tool widely used in industry
- ADMET-AI, Chemprop: Deep learning predictors
- PubMed/ChEMBL: Pull data on related compounds
Predictions are most accurate for chemical scaffolds similar to training data. Always validate experimentally before committing.
Practical strategy
- Build ADMET filters into early hit identification — don’t pursue compounds with obvious liabilities
- Run rapid in vitro assays (microsomes, hERG, kinetic solubility) on every series
- Generate animal PK data to validate in vitro predictions before committing to lead optimization
- Track structure-property relationships alongside structure-activity relationships
ADMET is where many promising compounds die. Building ADMET awareness into early drug discovery — not waiting until lead optimization — saves time, money, and gives you the best chance of nominating a development candidate that will survive the clinic.
