Half-life & bioavailability

A compound's half-life is how long it takes for its concentration in the blood to fall by half. A short half-life means it comes and goes quickly and may need redosing; a long one means it lingers and builds up over days until it reaches a steady level ^[1].

That plain version hides two things worth correcting, because both are misread constantly. First, terminal half-life is the time to halve the concentration after pseudo-equilibrium is reached, not the time to eliminate half the administered dose ^[3]. Those are different quantities and they diverge sharply early on, which is why the first hour after a dose rarely behaves the way a half-life figure suggests. Second, half-life is not a fundamental property at all; it is a hybrid of two things that are: how fast the body clears the compound, and how widely the compound spreads into tissue. A drug that hides in fat has a long half-life not because clearance is slow but because most of it is not in the blood where clearance happens.

The practical consequences all follow from one number. Steady state takes about four to five half-lives, whatever the dose, which is why a compound with a 24 hour half-life takes most of a week to reach its real level and why judging it on day two is judging something that has not happened yet. The same arithmetic runs backwards on stopping: clearing a long-half-life drug takes days, and stopping it produces a slow taper rather than a cliff. And half-life controls how much the level swings between doses, which is why the same daily amount split into two doses feels different from one dose ^[3].

One trap deserves naming because it inverts the whole picture. When absorption is slower than elimination, the terminal half-life stops reflecting elimination at all and starts reflecting absorption; pharmacologists call this flip-flop kinetics ^[3]. Depot injections and slow-release formulations work this way on purpose. It means a long duration of action is not always evidence of a slowly cleared molecule; sometimes it is evidence of a slowly released one.

A last caveat that matters more than any of the above for reading this wiki: the duration of an effect is not the half-life. The two can come apart in both directions. An effect can outlast plasma levels because what the drug did takes time to undo, and it can fade while plasma levels are still high because the receptor has adapted. Any dosing advice that maps the two onto each other one for one is guessing.

Bioavailability is the fraction of a dose that actually reaches the bloodstream intact. By definition an intravenous injection is 100%, because nothing sits between the syringe and the circulation. Swallowing something is often far less, and the loss happens in two places.

The first is simple absorption. A molecule has to dissolve in gut fluid and then cross the gut wall, and those two requirements pull in opposite directions: water solubility helps the first and hurts the second. The rules of thumb that came out of drug discovery capture this well enough to be useful. A compound is more likely to be poorly absorbed when it has more than five hydrogen bond donors, more than ten acceptors, a molecular weight above 500, or a calculated log P above 5 ^[5]. Whether a drug is then cleared by the liver or by the kidney turns out to be predictable from the same two properties, solubility and permeability, which is the basis of the classification systems used across drug development ^[1].

The second is the first-pass effect, and it is the one that surprises people. Everything absorbed from the gut travels to the liver before it reaches the rest of the body, and the gut wall itself is full of metabolising enzymes. A compound can therefore be completely absorbed and still have low bioavailability, because most of it was destroyed on the way through ^[4]. Absorption and bioavailability are different questions, and a product that claims high absorption has not answered the second one.

This is the whole reason most peptides are injected: the gut treats them as food and digests them. It is why some nutrients are sold as more absorbable prodrug forms; plain uridine absorbs poorly, so uridine triacetate is used instead. And it is why the same compound can feel completely different taken orally, sublingually, intranasally or by injection. The numbers in the table below are measured values from human studies, not estimates, which is the standard any claim about a route should be held to.

Reaching the bloodstream is not the same as reaching the target, and for anything meant to act on the brain the gap between the two is the whole problem.

The blood-brain barrier is a genuine bottleneck, not a filter with a few gaps in it. It is the reason a compound with excellent oral absorption can be pharmacologically useless in the brain, and the reason brain drug development lags so far behind development for other organs ^[6]. What crosses it easily is small, uncharged and moderately fat-soluble. What does not cross includes essentially all peptides, almost anything permanently charged, and most large molecules. GABA itself is the instructive example: it is the brain's main inhibitory transmitter and swallowing it does very little, because at body pH it carries a positive and a negative charge at once and doubly charged molecules do not cross.

Volume of distribution describes how far a compound spreads out of the blood into tissue, and it is half of what determines half-life. A large one means most of the drug is sitting in fat, muscle or brain rather than circulating, which is why fat-soluble compounds have long tails: the tissue gives back what it absorbed long after the last dose.

Protein binding is routinely oversold. Many compounds travel bound to albumin, and only the unbound fraction is free to act or be cleared, so it is common to see interaction warnings built on one drug displacing another. For the great majority of drugs that reasoning is wrong: displacement raises the free fraction transiently, clearance of the free drug rises to match, and the free concentration returns to where it was. Changes in plasma protein binding have little clinical relevance in almost all cases ^[8].

Transporters do as much work as the enzymes. Proteins such as P-glycoprotein actively pump drugs back out of the gut wall, out of the brain and into bile, and there is now a formal framework for deciding which of them a new drug has to be tested against before approval ^[7]. Two compounds can interact through a shared transporter with no shared enzyme at all, which is why any interaction tool that only knows about CYP enzymes is looking at half the picture.

Most compounds are chemically altered before they leave, and the family responsible for most of that work is the cytochrome P450 enzymes, written CYP. The human genome contains 57 of them, but only about a dozen, from the CYP1, CYP2 and CYP3 families, handle the biotransformation of most foreign substances including the large majority of drugs in clinical use ^[9].

Here is the catch that generates most drug interactions. One compound can block the enzyme that clears another, or speed it up. Block it and the second compound accumulates: a stronger, longer, sometimes dangerous effect from an unchanged dose. Speed it up and the second compound is cleared before it can work, which is the quieter and more dangerous failure because nothing appears to be wrong ^[2].

Induction has a second-order consequence people miss. If a drug is a prodrug that the enzyme activates rather than destroys, inhibition and induction swap their usual meanings: blocking the enzyme makes it useless, and inducing it makes it stronger. Any interaction rule stated as inhibitors make drugs stronger is wrong roughly half the time.

Timing is also asymmetric. Inhibition is usually fast, appearing within a dose or two, because it is a matter of one molecule occupying an enzyme. Induction is slow, taking days to weeks, because it works by increasing how much enzyme the liver makes. A newly added inhibitor causes trouble almost immediately; a newly added inducer causes trouble a week later, when nobody is still thinking about the change.

That asymmetry is exactly what this site's interactions and stacks tool watches for on the serious end. A strong enzyme inhibitor meeting a drug that depends on that enzyme is among the most common hidden interactions in medicine, and it is invisible from the label of either product.

Pharmacokinetics is the main reason two people taking the same amount of the same thing have different experiences, and most of that variation is in metabolism rather than in receptors.

Genetics is the largest single source. Several of the enzymes above are multiallelic and strongly dependent on ancestry, which produces distinct metabolic phenotypes rather than a smooth spread: poor, intermediate, extensive and ultrarapid metabolisers of the same drug ^[9]. For CYP2D6, CYP2C19 and a handful of others, that difference is large enough to be clinically actionable, and the evidence for using genotype to pick a dose is strongest exactly where the enzyme is polymorphic and the drug has a narrow margin ^[11].

Everything else stacks on top of that. Liver disease reduces metabolic capacity; kidney disease matters more for compounds excreted unchanged, which is the group most nootropics belong to. Age moves both, in opposite directions at the two ends of life. Smoking induces CYP1A2 and quitting reverses it. Diet does the same in miniature, which is why grapefruit is a genuine warning and cruciferous vegetables are a real if minor one ^[10][12]. Pregnancy changes the activity of several enzymes substantially.

The habit worth building from this is to treat a published half-life or bioavailability as a population average with a wide distribution behind it, and to treat your own response as the more informative number. That is also why starting low and changing one thing at a time is not caution for its own sake; it is the only way to find where in the distribution you sit.

Four questions turn a number on a page into something usable.

Was it measured in humans, or calculated? Log P values on compound pages are usually computed estimates useful for ranking and nothing more. Bioavailability figures are sometimes measured against an intravenous reference, which is the real thing, and sometimes inferred from urine recovery, which is not. Was it the parent or a metabolite? A short-lived parent with a long-lived active metabolite behaves nothing like its own half-life suggests; aniracetam mostly reaches the blood as a metabolite rather than as itself. In whom? Healthy young volunteers, fasted, is the usual population, and it is the least representative one. By what route, and with food or without?

Then the gaps, which for this subject matter are large enough that naming them is the honest thing to do.

Most compounds on this site have no published human pharmacokinetics at all. Dose ranges for research chemicals and most of the racetam family beyond piracetam and aniracetam come from clinical protocols where a trial happened to exist, from analytical work on marketed products, or from convention. They are not the output of dose-finding studies in volunteers, because those studies have not been done. See research chemicals and harm reduction.

Brain concentration is almost never measured. Plasma is what gets sampled, and the relationship between a plasma level and a brain level is compound-specific and usually unknown in humans ^[6]. A claim that something reaches useful brain concentrations is almost always an inference from animal work or from a calculation.

Interaction data is generated for approved drugs and nobody else. The enzyme and transporter testing that produces reliable interaction warnings is a regulatory requirement for a new medicine ^[7]. For a supplement or a research chemical the equivalent data mostly does not exist, so the absence of an interaction warning carries no information at all.

None of this makes the numbers useless. Half-life still tells you when to dose, bioavailability still tells you why a route was chosen, and the enzyme table above still predicts most of the interactions that matter. It means the numbers are a starting estimate for a person rather than a specification, and that confident precision on a product label is usually borrowed from somewhere it does not apply.