You choose the science. The disease chooses the bill.

What your first indication costs you, and why most founders pick it by accident.

When you build a biotech company, you spend equity to buy a piece of evidence, and if the evidence is enough, you keep going. If it is not, the company fails whether or not the science was right. Investors and pharma partners want to see that the programme has moved far enough to be worth more than it was last time. The number that governs survival is how much money it takes to buy evidence outsiders will believe, and that price is set by the disease. It decides whether regulators already accept a measurement as a stand-in for benefit and whether the patients who qualify can be found. It also decides whether the effect shows in three months or in three years. Decades of other people’s trials fixed these things long before your company existed, so two teams with identical science, working in different diseases, pay different prices for the same step.

Thomas Moore and colleagues, in JAMA Internal Medicine, costed all 138 pivotal trials behind the FDA’s 2015 and 2016 approvals. The 73 trials built on an accepted surrogate endpoint averaged $24.0 million. The 27 that had to measure a clinical outcome averaged $64.7 million. A third group scored on clinical scales came in at $20.5 million, close to the surrogates. Where a surrogate already exists, somebody else paid to validate it, and you get it for nothing.

The gap opens before the pivotal trial. Seed money buys lab work that costs about the same in any disease. The Series A has to buy the first result an investor will pay for, and the disease decides how far away that result is. In an expensive disease the A has to be bigger, or it runs out before it gets there. Fewer firms lead rounds that size, and the ones that do take more of the company.

That is why the first indication, meaning the patients, the line of therapy and the endpoint you aim at before anything else, is the most expensive decision you make as a founder. It sets how much cash you spend to get a result, how long you wait for it, how much of the company you sell to get there, and whether anyone will pay you for a partial answer along the way.

I ended the trials piece a few weeks ago asking what one answer costs, in money and in months. I meant it as a question about candidates. It is also the question that decides which companies can be built at all.

The best firms choose the disease on purpose

Few founders choose their first indication. A lab builds on the tissue it can get, and the disease comes with it, years before the company exists. When we ask spinout founders at ZAKA why they picked their disease, most give two reasons: the lab saw early signs there that the technology works, and the market is big. Both are usually true. Neither tells them what it costs, in that disease, to produce the first result an investors will pay for. By the time they price that trial, the company is built around the disease, and moving the science means starting over.

A handful of firms create biotech companies rather than back them. STAT ranks biotech venture funds every year, and Third Rock, Flagship and ARCH have all ranked near the top. Third Rock’s first three funds returned a median 3.58 times invested capital, against roughly two times for the median fund under $500 million and 1.77 times for the median fund over a billion. That does not show the returns came from choosing the disease well, but somebody at these firms picks the disease on purpose.

Flagship is recruiting a scientist to take “the implicit frameworks used by senior strategists (how unmet need is assessed, how patient populations are sized and stratified, how clinical and commercial landscapes are weighed)” and turn them into software. The listing wants someone with “a track record of personally making or directly influencing portfolio and indication-selection decisions.”

The method exists, but it lives in a few hundred people’s heads and sells through consultants at rates a seed company cannot pay. Flagship is writing software to capture it, because nobody has written it down as a system.

Atlas Venture is one of the few firms that writes this reasoning down in public. Julia Pian sets out when a small market still works: “smaller potential market sizes may still be viable if the capital required to get a drug approved and to patients or to a key value inflection point is relatively small and well-understood biology or precedented mechanisms increase the probability of success.” Two conditions, and the size of the market is not one of them. Her colleague Aimee Raleigh, writing about the crowded obesity market, puts the commercial question before the bench work: what has to be settled before discovery starts is whether the profile will be differentiated enough to earn adoption by physicians and reimbursement by payers. 

The pieces are public. Knowing how to weigh them against each other is not, and the people who can are not the ones starting companies out of university labs.

Choosing the indication well does not rescue bad science

If the target is wrong, no capital structure saves you. Nor is every failure foreseeable. The FDA published twenty-two cases in which Phase 3 contradicted Phase 2, and noted that this happened “even when the phase 2 study was relatively large and even when the phase 2 trials assessed clinical outcomes.” In two of them the drug increased the frequency of the problem it was designed to prevent.

Choosing the indication for the price of its evidence does not make weak science strong. It stops companies dying for reasons that had nothing to do with the biology.

AstraZeneca published its numbers before and after. From 2011 it put five questions at the centre of every discovery decision: right target, right tissue, right safety, right patient, right commercial potential. Its rate of getting from candidate nomination to a completed Phase III went from 4% across 142 molecules in 2005 to 2010, to 19% across 102 molecules in 2012 to 2016. Two of those five questions are about the indication, so read it as evidence that answering all five more rigorously changes what comes out, rather than as proof that the indication did it.

Start from the answer you can afford to buy

Ask one question: who funds the next step, and what is the cheapest result that would persuade them? It has to be a result that moves an outsider with no reason to believe you, not one that impresses your scientific advisory board. If the only convincing evidence in your indication is an outcome that takes four years and three thousand patients, you do not have a company, but a research programme looking for a sponsor. You should either find a different first indication for the same mechanism or accept that what you are building has to be financed some other way. Picking for the price of proof is not the same as picking the easy disease. You buy the cheapest credible result first and use it to fund the trial you could not have afforded at the start.

Then run the question forward. Each round buys one answer, and that answer has to be worth enough to raise the next one. At the end of the chain there has to be a result a pharma buyer (or a public investor) pays for. Lay the whole sequence out before you start and you can usually see which link breaks. Most founders find out which one it is when they get there.

Check these four first. They cost nothing to look up:

• Whether your indication has an endpoint the FDA already accepts, because that is the difference between a $24 million trial and a $65 million one.
• Whether a test can tell you in advance who will respond, because enrolling only those patients makes the trial smaller and the effect easier to see.
• How many clinical-stage assets are already aimed at your target, because Fougner and colleagues call more than five on one target herding, and past that your buyer can shop.
• What the standard of care costs, because the cheaper it is, the more you have to prove to beat it.

The ordinary sequence has three steps: get a data asset, build a platform on it, raise. The indication is not one of them. It came with the tissue. Put it first instead. Start from the answer you can afford to buy and that an investor or a partner will believe. Pick the indication where that answer counts for something. Then pick the science that can produce it.

Your first programme will look smaller than the company you set out to build. It is the one that can still raise when the result comes back, and it is how you pay for the rest.