It’s been six years since I titled a post “Remember Apo-A1 Milano?” If you go back even further, I wrote about the topic on this blog back in 2003 (!); scroll down to the November 11 post on that page. You can see from that one that the saga goes well back into the 1990s, but it’s really even older than that. New developments this week make the whole story wel worth examining, because it tells us a lot about both the good and bad sides of using human genetic markers to advance new therapeutics.
Here’s a brief summary. The name refers to a mutated form of the Apo-A1 lipoprotein, first identified by researchers at the University of Milan, when a patient presented from a village in northern Italy (Limone sul Garda). This person had blood chemistry that could only be described as alarming: very low HDL and elevated triglycerides. One would normally think that someone with that profile was at great cardiovascular risk, but the patient showed no signs of atherosclerosis or any other problem of the sort. It turned out that the patient had a mutated form of lipoprotein A-1, a major constituent of HDL particles. It’s a single amino acid replacement – arginine for cysteine at position 173 of the protein. Careful work with blood testing of all the residents and genealogy established that all the people with this mutation (just over 3% of the village, so maybe three dozen people in all) descend from a single person, Giovanni Pomarelli, who was born with it in the 1700s. Further study of the population in Limone showed that the mutation did seem to have an unexpected cardioprotective effect – how else to explain the unfavorable blood profiles with the lack of the usual problems?
Well, good. What we have here is one of nature’s random shots that tells us something very interesting about cardiology and human lipid handling. A great deal of work has gone into trying to figure out what it is about the mutation that makes it different – as far as I know, current thought is that it’s both the loss of the arginine and the addition of the cysteine contribute. But the next step was to try to figure out how to turn these insights into a therapy for the large number of people in this world who are not descendants of Giovanni Pomarelli. As is so often the case, that was where things started to get really tricky. Gain-of-function mutations like this are rare, and they present a real problem for drug development. Put simply, we’re a lot better at messing things up. We can make inhibitors and antagonists, ligands that block interactions and so on, so if you have a beneficial loss-of-function mutation, we can often come up with some way to mimic that. Might be a small molecule, might be a big ol’ antibody, but one way or another, we can often take some good cracks at blocking whatever needs to be blocked.
But a mutant protein that’s actually working better poses a problem. How do you recapitulate that with a small molecule? In this case, it’s hard to see a way to do it – the only thing that seems workable is to find a way to get that new protein into a patient. Out-and-out gene therapy in mice, using a viral vector, looks impressive: forcing the mice to make the new protein that way, along with a low-cholesterol diet, led to “rapid and significant regression” of atherosclerosis, but gene therapy in humans is still in its early days, and no one is going to line up to be the first for this one. To go even further, eventually we’re going to have the nerve (and the knowledge that lets us get that nerve!) to mutate the human germ line itself, and for all I know, everyone a hundred years from now will be born with something like this mutant lipoprotein. But we’re definitely not there yet. So the brute-force approach, which is available right now, is to just give the mutant protein itself as an injection, let it circulate around in the blood and do its thing as if the body were producing it. That gives you a mixed situation that’s not found in nature – a handful of people have only Apo-A1-Milano, the rest of us have none – but it seems plausible.
And it certainly seemed plausible in animal models, that’s for sure. Infusion in both mice and rabbits seemed to reduce atheroma lesions in their blood vessels. Safety trials were done in humans with recombinant protein, and in 2003, a study funded by Esperion Therapeutics was published. Patients got five weekly treatments, and the results were positive – but were they positive enough? Esperion couldn’t produce enough of the protein for a longer trial, as I understand it, and what they saw was an increase in atheroma volume in the control group (+0.14%) and a decrease in the treatment group (-1.06%). That certainly seemed real, but you would have to have more to declare victory: a longer trial in more patients, long enough to determine real cardiovascular outcomes. Esperion could not deliver that, but they were bought up by Pfizer (for an impressive $1.25 billion dollars) around the time that work was published, and Pfizer can deliver that sort of big, expensive effort.
That did not work out. Not even a little bit. Five years later, the whole thing was a wipeout – Pfizer spun out what was left of Esperion and sold Apo-A1 Milano to The Medicines Company for $10 million, a whacking loss of greater than 99% on that deal. The Medicines Company are turnaround specialists – their thing is to buy what they feel are undervalued drug programs from larger companies and try to get them to market in some form. They’re smaller, so they don’t need multibillion dollar payouts to be successful (although I’m sure that they wouldn’t be averse).
Cerenis Therapeutics also took up the challenge. They’re a French company founded by a former research head at Esperion, and they went into the clinic with their own variation on Apo-A1 Milano HDL, but reported disappointing results. “Disappointing” is hardly even the right word: no change at all in atheroma volume between treatment groups and controls. Instead of a dose/response, they got what looks like random noise, a complete bomb. That first link has some interesting comments from cardiologists about these results – they were reluctant to think that this whole approach was in trouble, but kept holding out for better trials.
Meanwhile, the Medicines Company trial got underway, but back in August came some disquieting news. An independent committee was examining the data, and the optimistic case said that they might find a strong enough response to stop the trial early. But they recommended that it continue as planned. And now a later look at the numbers have led the company to completely stop development. That leaves the whole Apo-A1 Milano story in disarray; to the best of my knowledge, no one is is working on it, and after these results, you’d have to have a pretty compelling reason to get back in.
So what’s going on? You can come up with several explanations, not all of which exclude each other:
- The original cardioprotective data from Italy weren’t correct. As far as I know, though, these still stand up, so the human mutation probably is still considered to be beneficial. But there are some data from both these human patients and the animal studies that don’t quite fit the story (see here).
- Having the mutation protects you, for reasons yet to be figured out, but infusion of the protein itself doesn’t. That would seem to fit the facts, but it’s hard to see how that can be true, given the animal data. Either there’s something wrong with those numbers, or the split between humans and the animal models is very wide, or. . .?
- The human trials just weren’t done the right way. That was (as mentioned above) one explanation that was offered after the Cerenis failure, but it’s getting pretty beaten-up-looking by now. You’d have thought that there would be at least a weak signal from one of these trials, but nothing.
Back when Pfizer bought up Esperion, there were mutterings that it was all done just to take them off the market so that Lipitor could have the field all to itself. But it doesn’t look as if Apo-A1-Milano, as a drug, is able to do anyone any good at all. No one could have believed this outcome back in the early 2000s, but the numbers are the numbers, and (once again) we don’t know as much about human lipid handling and the cardiovascular system as we think we do.
Update: John Carroll of Endpts.com notes that CSL Behring just announced results of a trial giving extra regular plasma-derived Apo-A1 protein to patients at cardiovascular risk. After four infusions, they seem to be OK on toxicology/safety, but seem to have found no difference in efficacy in the treatment group. If they’re going to see anything, it’s going to have to be in a big, long Phase III, apparently. . .