I wrote some time ago (Ay! Four years ago - have I been doing this for that long?) about the Roche/Trimeris HIV drug Fuzeon (T-20, enfuvirtide), and its costly manufacturing process. Roche built a factory in Colorado just to make the drug, which is a 26-amino acid peptide. And instead of doing it recombinantly, they're producing it the good old chemical way, by peptide coupling. (Here's a not incredibly competent collection of whiz-bang photos of the place, which at least have no purple spotlights in them)
Back in 2002, I had some thought that Roche had perhaps lost its corporate mind. But as this article from Chemical and Engineering News points out (subscriber-only, I think), they've actually done everyone a favor, whether by losing their minds or not. Their decision to go fully synthetic, and the massive investment that followed, has lowered the cost of all sorts of peptide synthesis reagents, starting materials, and equipment, to the point that it's now become enough of an industry to attract a lot more production interest. (And one of the big players in the contract business is. . .Roche's Colorado facility!)
As the article points out, recombinant technology (producing the peptide in engineered cells) is a wonderful thing, but only when it's working perfectly. And getting it to that point can be a long, expensive task. There are a lot of potential cell lines to choose from, each with its own advantages and disadvantages, and uncountable ways to engineer them and culture them. Even then, the purification of the target protein can be a whole new nightmare - as one chemist interviewed by C&EN says, at least synthesis doesn't give you back ten times as many different things as you put into it.
Peptides still aren't anyone's first choice for development when there's a small-molecule alternative. But for the targets that no small molecule is going to hit, they're worth looking at. Recent years have seen improvements in metabolic stability and duration of action, as people come up with all sorts of nifty delivery systems and conjugate polymers. You could do a lot worse.
But perhaps Roche could have done better. There were all sorts of glowing forecasts about Fuzeon when it was first approved, and all sorts of grumbling from people who took the optimistic numbers and calculated that Roche would be making its money back in two or three years at the prices they'd set. Well, that hasn't happened yet, since the drug isn't selling nearly as well as had been hoped.
Another two or three years should do it, if nothing better comes along to cut into Fuzeon sales. And stipulating that (which is no sure bet) Roche might be selling it for a long time to come, since the barrier to generic manufacture is going to be rather high. So, even after that wild factory in Colorado, they're still probably going to go into the black on Fuzeon, but it does make you wonder how the return compares to some of the other drugs in Roche's portfolio.
But that's their problem. In the meantime, it looks like they've helped everyone else in the business by making industrial peptide synthesis more affordable. Adam Smith's invisible hand strikes again. . .