SP-08 · The First Humans

Guillaume Molinet was not supposed to be in the room.

He was a stand-by — recruited, screened, kept in reserve in case one of the booked volunteers didn’t show. On the morning of Sunday, 10 January 2016, in a clinic in Rennes run by a contract research firm called Biotrial, someone dropped out, and the slot fell to him. He was forty-nine. He was a painter, a singer, a composer, a poet, a father of four. For five days in a clinic he would be paid €1,900. That morning he took the fifth of ten planned doses of an experimental molecule called BIA 10-2474. By the afternoon he had a headache and his vision had blurred. He never went home. Four of the five other men dosed alongside him went to the hospital with neurological damage that, for some, did not fully heal.

It is the worst thing that can happen at this stage of drug development, and it is genuinely rare. But hold the scene for a moment, because it tells you — more honestly than any textbook definition — what this stage actually is. A molecule that had only ever existed in cells, in mice, in a beaker and a binding assay, met a human nervous system for the first time. And the thing that turned a clean record into a catastrophe was not the drug’s existence. It was its amount.

That is the secret the name hides. We call it Phase 1, the first time a drug is given to people, and almost everyone — including a fair number of people who work in the industry — will tell you that Phase 1 is the part where you test whether a drug is safe. It is the single most common thing said about this stage, and it is wrong in a way that matters. Phase 1 does not ask whether a drug is safe. It asks how much of it a human body can take. Safety is not a box you check once and move past; it is a question that stays open for the entire life of the drug, through Phase 2, through Phase 3, through twenty years on the market and millions of patients. What Phase 1 does — its real, narrow, indispensable job — is find a dose.

The Staircase

Picture the problem from the inside. You have a molecule. Animal studies have given you a guess at a starting dose — deliberately, almost insultingly low, a fraction of what you think might do anything. You do not know what it does in a person. You cannot know. So you climb.

The first part is the single ascending dose, or SAD. A small group of healthy volunteers gets one low dose, and a team watches them — blood draws, heart tracings, the careful boredom of someone looking for the first sign of trouble. If nothing happens, the next small group gets a higher single dose. Then higher. Each dose level is a cohort; each cohort is a landing, and you do not climb to the next until you have looked hard at the one you are on.

The climb has its own safety rails, built from hard experience. The starting dose is set far below the level that did anything in animals, with a wide margin folded in deliberately. The first person in a cohort is often dosed alone, ahead of the others — a sentinel — so that if something goes wrong, it goes wrong for one, not six. Time is left between volunteers, and between cohorts, on the principle that a problem you cannot yet see is worth waiting for. None of this is bureaucratic delay. Each rail is a lesson somebody once paid for.

Then comes the part that killed Guillaume Molinet: the multiple ascending dose, or MAD. Now volunteers take the drug repeatedly — once a day for days or weeks — because that is how medicines are actually used, and because a molecule that the body clears cleanly after one dose can quietly accumulate when the doses keep coming. The BIA 10-2474 single-dose cohorts had been uneventful. It was on repeat dosing, in the fifth cohort, at fifty milligrams a day, that something in the drug’s chemistry crossed a line and stayed there. The staircase had looked safe right up until the stair that wasn’t.

What you are hunting for on this climb has a name: the maximum tolerated dose, the highest amount a person can take before the side effects become unacceptable. You are not hoping to reach it gently. You are, in a controlled and consented and heavily monitored way, trying to find the ceiling by approaching it — because the dose you will eventually give patients, the recommended Phase 2 dose, lives just below that ceiling, in the narrow band where the drug is present enough to work and not so present that it harms. Find that band and you have done the job. Miss it high and you poison people. Miss it low and the drug looks worthless in Phase 2 when really you just never gave enough.

So the better analogy is not a safety inspection. It is tuning an instrument you have never heard played — turning the peg, listening, turning it again, knowing that the note you want sits somewhere between slack and snapped, and that the only way to find it is to come at it carefully from the slack side.

What the First Doses Actually Measure

There is a second question riding alongside how much, and it is just as much the point of this stage: what does the body do to the drug?

When a molecule goes in — swallowed, injected, infused — it begins a journey the laboratory could only guess at. How much of it actually reaches the bloodstream? How quickly does it climb to a peak, and how high? How long before the liver and kidneys clear it away — hours, or days? Those numbers, plotted as the rise and fall of drug in the blood over time, are the molecule’s pharmacokinetics, and Phase 1 is where they are first measured in a living person (F-05). The blood draws that punctuate a volunteer’s day are not only safety monitoring. They are tracing a curve.

That curve decides things that matter enormously downstream. A drug cleared in two hours and a drug that lingers for two days are, in practice, different medicines even when the molecule is identical — one may need dosing four times a day, the other once a week. And it was exactly this property, accumulation across repeated doses, that turned BIA 10-2474 lethal: a compound the single-dose cohorts had tolerated built up, dose upon dose, into something the body could no longer absorb quietly. The pharmacokinetics were not a footnote to the safety story. They were the safety story.

So the stage is really two measurements braided together — how much a person can take, and what becomes of the drug once it is in. Neither can be read with confidence from a mouse, because absorption and clearance are among the things that differ most between a mouse and a person. The first humans are where the guessing stops and the measuring starts.

Why the People in the Room Are Different

There is a fork in this stage that surprises people, and it is worth slowing down for, because it is where the ethics of the whole enterprise live.

Most Phase 1 trials enroll healthy volunteers — people with nothing wrong with them, who cannot benefit medically from the drug and accept some risk for payment and, often, for the genuine sense of contributing. That arrangement carries a quiet moral weight. The volunteer is not a patient. There is no therapeutic intent. The €1,900 is not a treatment; it is compensation for exposure. The entire justification rests on the risk being small and the oversight being real — which is exactly why the Rennes failure became a national reckoning in France, and why a manslaughter investigation followed. When a healthy person is harmed, the bargain that makes healthy-volunteer research permissible has been broken.

And then there is oncology, which plays by other rules. You cannot give a promising but brutally toxic cancer drug to a healthy person — the risk is far too high to justify with money. So cancer Phase 1 trials enroll patients: people who usually have advanced disease and have exhausted their standard options. Now the calculus inverts. There is therapeutic intent — a real, if often slim, chance the experimental drug helps. The volunteers are not trading safety for cash; they are trading it for hope, which is a different and heavier currency. The dose-escalation logic changes too, into careful schemes — the classic “3+3” design and its modern successors — built to climb toward the tolerated dose while exposing as few people as possible to a dose that is too high or, just as cruelly in this context, too low to do any good. For someone with advanced cancer, such a trial is sometimes the only door still open, which makes the consent conversation one of the most delicate in all of medicine. The honest version of it admits something uncomfortable: that the dose this patient receives is chosen partly to protect the patients who will come after — that they are, even at the bedside, helping to build the staircase as they climb it.

Same stage, same word on the protocol — Phase 1 — and two almost entirely different human situations. A primer on how trials are built can take you deeper into the machinery (F-08), and the statistics of small numbers — how you draw any conclusion at all from a dozen people — are their own discipline (F-07). How the dose-finding actually changes shape across drug types is a story in itself: a first-in-human cell therapy (M-AUTO-08) is nothing like the first dose of an antibody-drug conjugate (M-ADC-08), which is nothing like a vaccine given to thousands of well people (M-VAX-08). And to see how strange and wonderful Phase 1 can get when the drug simply works, there is no better story than the 1998 trial of a leukemia drug where the response was so total it broke the conventional logic of these stages entirely (CS-GLEEVEC-03).

The Bet

Step back and look at the numbers, because they reframe everything.

Of all the drugs that enter Phase 1 — that have already survived years of chemistry, the brutal filter of animal testing, the regulatory clearance to touch a human at all — only about 13.8% are ever approved. Roughly seven in eight do not make it. And that is the figure across all of medicine; in oncology, the hardest arena, the number falls to about 3.4%. Fewer than one in twenty-five. (Vaccines, by contrast, clear the gauntlet far more often — about a third, 33.4% in the same dataset — a gap worth sitting with.)

These are not the odds of a careless field. They are the odds of a field operating at the edge of what is knowable. Animal studies said yes; the human, who is not a large mouse, gets the final word. Phase 1 is where that conversation opens. It is the first place the drug can speak for itself, and most of what it says is no.

And reaching this room is itself an achievement most molecules never manage. For every compound that earns its first human dose, thousands were made and thrown away in the lab; the survivors carry years of work and tens of millions of dollars behind them before a single person rolls up a sleeve. To then watch seven of every eight fail anyway is not waste. It is the cost of finding the eighth — and the failures are how the field maps where the cliffs are, so the next molecule can be walked past them.

Which is why the framing we started with matters so much. If you believe Phase 1 is a safety test, then a death in a trial looks like the system failing at its one job. But Phase 1 is not promising safety; it is searching for a dose, in the dark, with human beings holding the candle. The promise it makes is smaller and more honest: that the search will be slow, that the steps will be small, that someone will be watching at every landing, and that the people who take the risk will have been told what is known and chosen freely on that basis. The Rennes trial is remembered not because Phase 1 is reckless but because, that week, those promises were not kept.

Guillaume Molinet went to Rennes as a stand-by, expecting to sit out the day. He was dosed because a chair came open. There is no metaphor that improves that fact, and it would be obscene to try. But there is a reason to carry it forward into everything that follows in this curriculum. Every approved drug you have ever taken passed through a room like that one — through a first human, a first dose, a careful climb up an unfamiliar staircase. Most candidates never reach the top. The ones that do are not the ones that were merely safe. They are the ones that, at some tolerable dose, turned out to be worth it.

That is the bet of the first humans. The rest of the pipeline is just finding out whether it pays.

This is the free spine of The Lead Compound.

You’ve just read one stage of how a medicine actually gets made. The spine is the free through-line — the whole pipeline, start to finish. The full course goes deeper: every drug class (antibodies, mRNA, cell and gene therapy, peptides, and more) and the real, documented stories behind the medicines that defined them — Ozempic, Keytruda, Gleevec, Comirnaty, and dozens more.

← Previous (SP-07) · How to read this course · Next: (SP-09) →