ADME

Swallow a tablet and the drug inside it has done nothing yet. Before it can lower your blood pressure or kill a cancer cell, it has to survive four separate stages, any one of which can fail.

Absorption. Can the drug cross the gut wall and reach the bloodstream at all? Stomach acid destroys some drugs outright; others simply can't cross the gut lining, which is why insulin is injected rather than swallowed.

Distribution. Once in the blood, where does it actually go? A drug that never reaches the tumour, or never crosses into the brain, is doing nothing useful no matter how much of it you give — it might just sit in fat tissue instead.

Metabolism. Mostly in the liver, enzymes — chiefly the CYP450 family — attack the drug and convert it into other molecules. Sometimes that deactivates it; sometimes the resulting metabolite is itself active, or toxic.

Excretion. The body has to be able to get rid of what's left, mainly through the kidneys or bile. A drug that can't be excreted accumulates, and accumulation is how a safe dose becomes a toxic one.

ADME is the science of measuring and predicting each of these four stages for a given drug candidate, before it ever reaches a patient.

Where it sits in development#

ADME is a gatekeeper science, front-loaded into discovery and preclinical work, then carried into early human trials to confirm what the lab work predicted:

Discovery → ADME screens compounds out early
Preclinical → full ADME characterisation
Phase I → human ADME studies confirm or contradict the preclinical model

A compound with poor ADME properties — one that never gets absorbed, or gets destroyed by the liver before it can act — is usually dropped here, long before anyone spends money on a Phase II trial to find out.

What's actually settled, and what isn't#

The four-stage framework itself has been standard since the 1970s, and for a conventional oral small molecule, ADME is well-characterised biology, not a research frontier. What's still genuinely open:

• Predicting how a drug will behave in humans from animal or cell data alone — the inter-species gap is still poorly understood, which is why animal studies remain a required step rather than a formality.
• Blood-brain barrier penetration is notoriously hard to predict, which is a direct bottleneck on CNS drug development — Alzheimer's, Parkinson's, psychiatric conditions.
• ADME in populations that don't get studied much — neonates, the elderly, patients with liver or kidney disease — is thin.
• The gut microbiome's effect on drug metabolism is a real and largely unmapped source of the patient-to-patient variability doctors see and can't explain.
• ADME for biologics — proteins, mRNA, gene therapies — doesn't behave like small-molecule ADME at all, and the field is meaningfully behind on it.

None of those numbers above are mine to put a percentage on with any confidence; I'd treat any specific "X% of drug failures are ADME-related" claim you see elsewhere as a rough historical estimate, not a measured constant — it depends heavily on therapeutic area and on how early ADME screening happens in that company's pipeline.