People tend to go into drug discovery and development from some kind of scientific background. Since I’m a chemist myself, I’d be the last one to say a word against that sort of training, but. . . there is one thing that we perhaps need to UNlearn. (And before anyone from a business background starts to feel relieved, I should tell you that you probably suffer from the same problem.)

When you study the fundamentals in any of these fields, you’re exposed to a lot of linear relationships. Put twice the amount of starting materials into your reaction, and you get twice the amount of product, right? I’m not sure if it’s exposure to those examples or just some hard-wired tendency of the human mind, but for many people, linear output is the default mode for how they expect things to work. But the problem is, I think it’s the LAST thing we should count on. As far as I can see, in the real world there are no linear relationships at all. Everything obeys more complex equations, with all sorts of odd exponents in them.

I’ll qualify that, because it’s true that for some systems, over some of their range, things do come out 1:1. But even these can be pushed at the high or low ends to go off the straight line, and it can be even more troublesome when they do. (After all, it’s the last thing you might expect from a well-behaved system that’s never troubled you before.) Most of the time, though, for most things, the functions we’re dealing with are a lot hairier, and we really should be ready for that from the start.

Look at some examples from pharmaceutical work: how many truly dose-proportional blood levels have you seen? Sure, over a narrow range things can come close, and if you’re lucky, that’ll correspond to the compound’s expected dosing. But every single pharmacokinetic plot will show some funny business at the bottom and top ends if you look closely. And as for efficacy, try to estimate how many U-shaped dose-response curves you’ve seen in your career. (If you’ve done CNS work, you may never have seen anything else.) Having a plot twang around and curve back on you like that is about as non-linear as it gets. And linear responses in toxicology? Since when?

Well, you might say, those are living systems, and you can’t expect them to behave in a simple manner. I’d agree with you there, only stopping to point out that those living systems are what make or break us in the end. Stepping back a bit to in vitro work doesn’t improve things much. Look at organic chemistry, my own first area of expertise. Many chemists have made the mistake (once) of thinking that they can run a reaction on five times the scale, to get five times as much product, if they’ll only put everything in a flask five times as large.

Of course, you run the risk of spending five times as much effort in scraping things off the ceiling if there’s any kind of exothermic reaction involved. As this experience teaches a person, the volume of solvent goes up as a cube, while the surface area of the flask, which is what you’re counting on for cooling, only goes up as a square. People who spend their time dinking things around with five-mL flasks never think about these things. But scale-up chemists have to internalize this bit of knowledge or look for another line of work.

OK, I promised some examples from the business end. How about doubling the advertising budget — does that ever sell exactly twice as much product? How about increasing head count? Would doubling someone’s salary improve their productivity by the same factor? (If anyone is minded to try that last experiment under controlled conditions, please contact me). Large-scale business issues are full of things that work by exponents instead of linearly. Such power-law distributions are responsible for the unequal splits that you can find everywhere — the way that a minority of your employees seem to be responsible for a majority of the work that gets done, or how 20% of your customers cause 80% of your problems.

There’s a tendency for people to assume that these phenomena are a sign that something has gone wrong. But while these things may (or may not) be troublesome, trying to fix them permanently (rather than hedge against their inevitability) may be a waste of effort. How did the company ever get itself into a position where only a handful of its products generated most of its revenue? Time to call in the consultants, surely. And look at the HR situation — think how great things would be if we could get everyone to perform like those stars! Training course time! Break out the three-ring binders! But think a bit before you start booking all the conference rooms. These major/minor splits aren’t perversities that happen only to you; they’re a reflection of how the world is put together. Perfectly equal distributions are the real oddities.

I’ve saved the most telling business example for last: what ever gives anyone the idea that doubling the size of the whole company is going to improve things by a factor of two? It’s true that some of the functions of a large company scale fairly well — sales, for example, over the range that most companies find themselves working in. Bulk production runs fairly close to the line, too, since two plants will produce roughly twice as much as a single one (as long as you’re not creating new distribution problems along the way).

But research? Research productivity follows perverse functions that no one has ever been able to graph. Why shouldn’t it? Research projects are the sum of all those nonlinear relationships I was talking about at the outset, and they’re being carried out by teams of people in which the best performers have influence out of all proportion to their numbers. The one result you can rule out is that the whole will equal the sum of the parts.

Unfortunately, when the case for a merger is being made, the argument is sometimes that things will indeed be nonlinear, but only in a good way. Yes, thanks to the magic of synergy, two plus two will surely equal five, if not some even more profitable sum. No one seems to plan for two and two to equal three-point-five-something, but I think that’s probably just as likely (and probably more). And that can be a real problem if everything else has enlarged, but the pipeline of new drugs doesn’t keep pace.

Everyone can think of some examples from recent history here, I’m sure. Even the better drug mergers I can think of fall more into the first-do-no-harm category, as far as I can see. Can anyone come up with some examples of those roaring bonfires of synergy that have supposedly been ignited? Here it is at last, the one part of the industry where nonlinearity is anticipated, even celebrated, and it’s fit to do us all in . . .

Derek B. Lowe has been employed since 1989 in pharmaceutical drug discovery in several therapeutic areas. His weblog, In the Pipeline, is located at www.corante.com/pipeline and is
an invaluable aid to Contract Pharma. He can be reached at [email protected].