I got into a discussion not long about about the various meanings of what should be a simple question: “What does this compound do?” The answers are many, and to look them over is to take a tour through the thoughts of many people in every part of a drug company.

For the chemist who just made the thing, the answer might be, “This compound gets my immediate supervisor off my back, because he’s been on me to get this analog into testing for the last month.” Or perhaps it’s, “This compound will finally tell the modelers that no, that extra basic nitrogen really doesn’t pick up that polar interaction, and it’s time to think of something else.” Most of the time, though, I think the answer would be something like, “This compound is the latest in the lead series, and might just be the one that has good enough PK/high enough selectivity/strong enough activity to be the clinical candidate. And I sure could use one of those.”

Now, none of those answers may sound very satisfactory, but that’s because these are early days. Most people, when they think of what a compound does, are thinking of some sort of biological effect, and only the chemists get to see these things before they’ve had any such assays run on them. So it’s not surprising that their perspective might be a bit warped, at first. The honest answer to “What does this compound do?” at this point is, “No one knows yet”.

Once someone has run the primary assay, asking the question will almost certainly produce a number in return. It might be a full-fledged Ki, or an IC50, or just a per cent activity at some given concentration, but there’s always a ready number to tag each compound with and rate its immediate importance to the project. And while that’s useful, and while human nature probably makes that process inevitable, reducing everything down to one number is actually somewhat dangerous. There’s a lot more to know about a compound other than how potent it is in the primary assay.

But unless that number reaches a certain level, no one will pay attention to the others. (The flipside of this effect is that if the compound’s number is wildly impressive, no one will pay enough attention to the other numbers, either.) But if the compound does make it to the cell assays, or especially into animals, “What does it do?” then gets answered by a phenotype. The compound causes A, B, and C to happen. Note that these activities probably have something to do with the numbers in the primary assay, but it’s by no means certain that all of them do. This takes us right into the difference between target-driven molecular biology and old-style pharmacology, and right into two very different worldviews. If we had the biological pathways mapped out well enough — or if we thought we did — then the potency at the target would be enough to tell us what the phenotype in a real animal is likely to be. But when do we have that level of confidence?

So we run the tests, for both good and bad effects. A toxicologist will give still another answer to the question, because they’re focused (quite rightly) on the side effects. You’d prefer the tox people to be at a loss for words when you ask them what a particular compound does, but they rarely are.

As Willie Stark, the corrupt governor in All the King’s Men put it, “There is always something.” By this point, the effects are probably completely divorced from all those earlier assay numbers (unless you target has something intrinsically nasty about it). Whole-animal toxicology is a phenotypic screen if ever there was one, and a close study of it will bring home just how little we understand what we’re really doing. If you’re lucky, you’ll be able to say something about what the compound does, but how often will you be able to say why?

We’re up to a pretty advanced state of affairs by now, and “What does this compound do?” can be answered in several new ways. “It generates a nasty waste stream in the pilot plant,” is one you don’t want to hear very often, as is, “It calls for a starting material that no supplier seems to be able to deliver reliably,” or, “It forms 14 polymorphs.” To a large extent, the original purpose of the compound — and the actual one, according to many other people — is beside the point to many people at this stage. It could be used for anything, but first you have to get enough of it, and in reliable fashion. And by now, the big question is also starting to be answered by clinicians. We can start, finally, to talk about the answers that everyone was thinking about back when the first compounds were being made. Is Compound X really going to help people with Disease Y? What does it do for them?

And that brings us to a very important answer indeed. Because, “What does this compound do?” is exactly the sort of question that the FDA will be asking eventually, and they’ll be going into great detail about all the good and bad answers that have been generated so far. Once they’re all put together, “This compound provides substantial benefit relative to its risks,” is the magic answer that no one ever gets to hear very often. With any luck, the question of what the compound does will be the one that patients will soon be asking their doctors.

Past this point, the question will start to be answered by accountants and by rarified parts of upper management during press conferences. The “do” part of “What does this compound do?” gets applied to the market share it carves out, the peak sales it will (or might) reach, and the profit that it had better bring in. “This compound fills a slot in our oncology portfolio,” is one possible answer, as is, “This compound will become a first-line agent for Disease X,” or, “This compound will become our biggest seller during the peak of its product cycle.”

That last one moves us into investor language, and investors come in several species. The sorts who buy early into smaller, speculative companies tend to be very much into the science behind the coming compounds, even if they turn out, at times, not to understand it very well. But the ones who are buying the bigger, more established outfits often want to hear about things like inventory levels, quarter-to-quarter prescription trends, and the legal defenses against the case for an early generic switch. What does this compound do? It keeps the institutional investors happy and keeps the company’s bond rating high. Sometimes, it keeps the lights on and the doors unchained.

We’re a long way from the days of taking the first NMR spectrum of a new reaction, or getting the first readout in the primary assay. The people who did those experiments have long since moved on to other projects and are asking what some new molecule does, even as the original one starts to pay their salaries.