Since Monday's posting was about the terrible consequences of changing one amino acid, I thought I'd stick with that theme. Today's outcome isn't life-and-death, fortunately for everyone involved.

But it does involve the patent rights to a $500 million/year antibody, so it's not without interest. Particularly if you think that another company has blatantly ripped off your patent, after doing a licensing deal with you that was supposed to preclude that sort of thing. The patent holder is Celltech, a UK company with a technique for generating humanized monoclonal antibodies in mice.

Antibodies are very useful, of course, but getting good ones isn't easy. There's no way to make them other than from a living organism (cell cultures, usually.) Ideally, you'd want to generate them in the same species that they're to be used in. Otherwise, you run a high risk of setting off an immune response in the treatment animal - instead of the antibody doing its thing and alerting the host's immune system to something else, it gets recognized as a foreign object all by itself. Whatever immune signal it's trying to send gets lost in the ensuing confusion, which can range from local irritation all the way up to anaphylactic shock. A humanized antibody has a generally human immunogluobulin structure with some specifically mouse-generated immunoreactive sections in it. These can pass, for the most part. It's a very useful technology, but rather tricky and sometimes hard to generalize.

Medimmune developed Synagis (palivizumab) as a treatment for respiratory syncytial virus (RSV,) which can be severe in infants. It's pretty much the only thing out there for it. No vaccine exists, for example: the first attempts to develop one in the 1960s backfired tragically when it turned out that treated children actually had a worse outcome if they were exposed to the real virus later on. (Here is a thorough report on Synagis and RSV, courtesy of Biopharma.com.)

Celltech claims that Medimmune's antibody, though, is equivalent to something they've already patented. It's a point worth arguing: of the over 1300 amino acids in the antibody's chain, there's one that differs between Celltech's patent and Medimmune's drug. (Specifically, it's a threonine for serine, which is a mighty small change indeed: just one methyl in the side chain. That makes the Dutch mutation I spoke of yesterday seem like a total makeover.)

I can hear the groans now from the cognoscenti who can tell what's coming next: yes, this is going to turn on the Doctrine of Equivalents and the Festo decision. I wrote about this at length back on June 5, but I'd caution you if you go back to read that post.

Gary Pulsinelli of Tennessee's law school (yep, just down the hall from Glenn Reynolds, I gather) wrote to point out that I'd thoroughly smeared together the issues of infringement and patentability. That I did! My own background biased me into thinking in terms of battling patents, and that led me astray. It's important to keep in mind, for example, that something can be found not to infringe someone else's patent, but at the same time not be patentable on its own.

The Festo case was all about infringement. Here's a distilled version:
1. Festo (of Long Island) starts selling a nifty mechanical cylinder device.
2. SMC (of Japan) starts selling a nifty (and hauntingly familiar) mechanical cylinder device.
3. Festo sues SMC for patent infringement. They claim that the doctrine of equivalents prevents SMC from making a minor modification and calling it theirs.
4. SMC claims that Festo doesn't get to invoke the doctrine, because they made changes to the relevant claims while the patent was going through the approval process. That's "prosecution history estoppel," which has been the counterweight to overly broad doctrine of equivalents interpretations. If a claim gets narrowed during the pre-approval prosecution of a patent application, you can't go back (for example) and try to get D of E protection based on the original broader claim. After all, the Patent Office presumably narrowed the claim because they thought the broader one wasn't patentable in the first place.
5. Festo wins in the District Court, and the decision is kicked upstairs on appeal, to no one's surprise. It bounces around the Appeals Court for a while, with another big patent case (Warner-Jenkinson) complicating the issues in the meantime.
6. To everyone's surprise, the Federal Circuit eventually reverses the decision and takes the hardest line yet on using the doctrine of equivalents. Saying that the case-by-case approach had led to a mess, the court decides to draw the line so everyone can see it: Just about any changes made to a claim during the prosecution of the patent, the court now says, trigger prosecution history estoppel. Their take-home is something like "write your claims better from the start, and we wouldn't have these problems." The decision applies to a huge number of patents that were written and prosecuted while the parties still had a stronger doctrine of equivalents in mind. Confusion reigns.
7. Festo appeals to the Supreme Court. They don't hear many patent cases, but this is the biggest one since Warner-Jenkinson. Heavy amicus curiae briefs are filed in support of both upholding and reversing the Circuit Court's decision.
8. The Supreme Court largely reverses the lower court. The doctrine of equivalents is given new life, and the court attempts to clarify the no-man's-land between it and prosecution history estoppel by setting some new criteria. All interested parties in the intellectual property world settle down to figuring out the new landscape, and waiting for the first major court cases that'll hammer out the details.

That leaves out a lot of tricky stuff (such as the details of those new criteria ,) but I think it gets the main points across. Celltech, as you might guess, filed a brief urging that the lower-court decision be junked. If you weaken the doctrine of equivalents, they said in essence, you're going to open up everyone in biotech to what MedImmune is doing to us. Meanwhile, MedImmune claimed that this sort of time-wasting litigation was just what the lower court was trying to prevent by drawing its clear, bright line. If Celltech wanted to claim this protein, then they should have claimed it, was their point of view - it's time to stop hiding in the bushes, waiting to whack people in the head with the doctrine of equivalents if they think they can lay claim to something valuable.

The rest of the biotech industry was split along roughly these lines, with the likes of Chiron siding with the original doctrine of equivalents, and Genentech (among others) wanting the limitations imposed by the Circuit Court to stand. I like the first approach better. While I take the point that there's a lot of deliberately broad or vague claim language out there, I think that's the lesser evil compared to what this case is showing us.

If single amino acid changes are enough to break a patent, then what isn't? Instead of wasting time on post-patent litigation, we'd all be wasting our time writing incredibly precise and intricate claims (and doing a lot more running in circles trying to come up with data to enable them.) Celltech should win this one, and I hope they do.