Robert Plenge has an excellent overview of the PCSK9 story up on his site; I recommend it. His take may sound different from mine at first, but I think we’re actually in agreement on a lot of important points. I said that “PCSK9 is about as compelling a story as we’re likely to see in this space, and if it has indeed come up a bit short, that’s food for thought,” but Plenge’s focusing more on the first half of that sentence. And he’s right about that – this target really is the best story we have so far in what I think we may start calling the new genomics era of drug discovery.
That era has been a while getting off the ground. Longtime drug researchers will remember that it was all supposed to take off like a rocket back around 1999 or so, when the human genome was sequenced. That was indeed a weird period, because a lot of people figured that if you hadn’t boarded said rocket in time, you’d find yourself stranded on a desolate planet bereft of all drug targets and all hope. Some rather silly decisions (in retrospect) were made on that basis, but the fear was very real, as were the hopes on the other side of the issue.
But as that PlengeGen post details, PCSK9 is indeed the real thing, and just what people were expecting to find back then. He’s right that the gene was basically unknown before 2003; no one had studied it and no one cared, because no one knew what it did or that its function could be important. From that flat standing start, a drug was on the market in 11 years, which by real world drug industry standards – particularly in cardiovascular disease – is about as fast as we can possibly go. People outside the business might well look at that and be aghast (in fact, that seems to be the reaction driving a lot of the “Burn down the FDA” sentiment). But it wasn’t the FDA that made this take 11 years; it was (once again) the flippin’ science.
If you take someone through the whole process, showing them what had to be done at each step and what the issues and pitfalls were, the timeline becomes a lot more believable. The first report of PCSK9 mutations in humans were gain-of-function, which (as mentioned before) is unusual. The people with these mutations have far worse LDL profile than normal, and that’s an interesting result, but it by no means guarantees that loss-of-function mutations would turn around the other way and have a beneficial effect. The report a couple of years later of just those sorts of mutations is what really jump-starting things – that was proof that something good could happen by knocking down PCSK9 somehow.
Then you get to the “somehow”. Do you do it with a small molecule? The PCSK9 protein has a serine protease functionality in it, so people thought that it might be approachable that way, but for many reasons that just didn’t work. Time spent, and that took care of the plausible small molecule approaches, so it’s on to antibodies. Then you have to take those antibodies, once you’ve generated a good one (which doesn’t happen overnight) and validate the whole approach in one of the animal models for lipoproteins. Takes time, and you can’t skip it or speed it up much, not and get actionable data that’s worth starting a human clinical program around. A big cardiovascular program is going to chew up many hundreds of millions of dollars in the clinic at the barest minimum, so you’d better be sure that you’re making the right call. The first humans didn’t get dosed with such an antibody until 2010, and that’s not because people were dragging their feet, and the first Phase III study results didn’t get generated until 2013, which for a long, slow process like cardiovascular disease is just about as fast as you can go. And even then, those numbers didn’t address real-world outcomes (mortality and morbidity); we only got those this year. If you can find a way to speed that up and still generate something meaningful, go collect your billion dollars. Frankly, a lot of the calls for FDA reform start to sound to me like “But we don’t care about meaningful”.
PCSK9 therapy was approved as quickly as it was (as the Plenge article details) on a surrogate endpoint: LDL levels. We’re fairly sure that those are connected to cardiovascular disease, but lowering them through a different mechanism (like this one) might or might not partake of the benefit that we’re seeing. You don’t know. That’s what the outcomes data this week addressed, and it’s true that the big picture is that yes, people were right – targeting PCSK9 really does work the way we thought it did. (The flip side, which I emphasized in my own post, was that it may not be working quite enough, compared to expectations). Surrogate endpoints are going to be crucial for slow-running diseases in the clinic, and this is one of the best example we have so far. Try finding one for Alzheimer’s, though, if you have few spare billions and a brave heart.
So there’s a case to be made that this target really is the dawn of the era that we all thought was dawning years ago. I can appreciate the celebratory tone of Plenge’s post, but at the same time, if you’d told people back in 1999 how things had worked out, they would likely have been (at some level) horrified that (a) it took until the mid-2010s for something like this to happen and (b) that this is the main example that we can point to, and that the landscape is not littered with similar stories. That brave new world is not the one we live in, though, and for the one we’re in, this is a good result and we should make the most of it.