The Washington Post made quite a splash with this story about Pfizer, Enbrel (etanercept), and Alzheimer’s disease. There’s already been a lot of comment about it yesterday on Twitter and in some other venues, but I thought it might be useful to try to sum things up in an easily accessible place. Here we go:
What’s the accusation in the Post’s story?
Basically, that Pfizer had evidence that Enbrel could be useful in Alzheimer’s disease, and didn’t do anything with it. This came from an analysis of insurance claim data: a set of about 127,000 patients with an Alzheimer’s diagnosis and a set of 127,000 without. It turns out that more people in the second group had been treated with Enbrel (302 patients) versus the first (110 patients). The Post obtained internal Pfizer documents discussing this and whether it was worth further investigation, and the company had concluded it wasn’t.
Why wouldn’t they?
Several reasons. The biggest, though, is that no one undertakes an Alzheimer’s trial lightly. The clinical success rate for Alzheimer’s trials is arguably zero per cent. There are a couple of therapies approved, but they don’t affect the disease progression at all – they help a few people, a bit, for a little while, and that’s it. That argues for a huge unmet need in the area (which no one disputes, God knows), but it also argues for caution. Many, many approaches have been tried already, from many angles, and every single one of them has gone down in flames.
The story quotes the Pfizer documents as saying that three to four thousand patients would be involved in a clinical trial if the company were to run one (which sounds about right), and also says that the trial would cost $80 million. Which sounds ridiculous. You’d have trouble running a 4000-patient study for the flu with $80 million in hand, much less an Alzheimer’s trial. It’s a slow-moving heterogeneous disease; Alzheimer’s trials go on for years because of those factors, and even just choosing your patients at the start is not an easy process, either.
It has to be emphasized as well that the data in the Pfizer presentation are not amazing. I’ve seen people on Twitter and the like going on about how Pfizer had a drug that was 64% likely to cure Alzheimer’s or some other crazy statement (two out of three, y’know), and that betrays an extreme lack of knowledge about clinical data and drug development. Not that that’s uncommon. But no, this is a noticeable-but-small signal, and by itself (I cannot state this strongly enough), it would not be enough for anyone to launch an Alzheimer’s trial. See below for more reasons why this is true, but that has to be stated up front. Note, for example, that Pfizer only sells Enbrel in Europe: in the US, it’s sold by Amgen, who acquired Immunex years ago largely just to get the drug (Pfizer ended up with the other marketing rights when they bought Wyeth). The Post article mentions, in passing, that Amgen had also looked at these numbers and passed. You would not be able to get venture capital money to run such a trial based on these data, nor a grant from any government agency. If you still doubt those assertions, perhaps the case for them will become clearer as you read on.
Any other reasons?
The article does note that Pfizer was getting out of Alzheimer’s in general at the time (2015), but it also explicitly makes clear that Enbrel was nearing the end of its patent lifetime and brings up the idea that Pfizer deliberately took a pass because they weren’t going to reap as much profit.
Well, you’ll have to trust me on this, it’s a little out there, but drug companies don’t generally walk away from big profits if they can help it. I’ve had my problems with Pfizer over the years, but I have never called into question their ability to make money. If Pfizer really thought that this was a promising lead into an Alzheimer’s therapy, they would have found a way to turn a profit off of it. One of the biggest objections to the idea inside the company was that this finding was unlikely to be real because Enbrel doesn’t really penetrate into the brain: a believable signal for Alzheimer’s, unlikely though it would be, would have been the occasion to search through similar antibodies or modified forms of Enbrel looking for ones with better penetration. Those you could patent fresh, it hardly needs to be said – and here are some people who have worked on just this idea. Antibody therapies (directed against the beta-amyloid protein) have already been tried several times against Alzheimer’s, so that’s not a crazy idea per se. Now, it’s worth remembering that all of those amyloid-antibody trials have ended in grievous, expensive failures – every single one, apparently because targeting beta-amyloid is not exactly the road to success that people had hoped – but these antibodies have gotten into the brain.
And the idea that Enbrel’s patent is about to expire is also up for debate – in fact, it’s being debated in court right now. There are biosimilars on sale in Europe, but the in US Amgen is arguing that they have legal grounds for ten more years of patent protection, and that case hasn’t been decided yet. One would also be able to argue that these biosimilars would need to prove their own efficacy against Alzheimer’s, rather than (as is the case now) against the other main targets for Enbrel, which are rheumatoid arthritis and psoriasis.
In short: the Evil And Nasty interpretation, which I’ve seen proposed and which can be summed up as “Drug Company Finds Cure For Alzheimer’s And Sits On It“, is ridiculous.
What is Enbrel and why might it do something against Alzheimer’s?
Enbrel is a fusion protein (of the TNF receptor and an antibody domain) targeted against the TNF-alpha protein. (Edit: clarified this, as opposed to a straight antibody). That’s a major player in inflammation, which is why you see uses like the ones just mentioned. It’s been noted for years (decades) that the brains of Alzheimer’s patients show signs of a persistent inflammation response, leading some to wonder whether Alzheimer’s leads to inflammation or inflammation leads to Alzheimer’s. With that in mind, basically every single anti-inflammatory therapy has had a look at it for potential Alzheimer’s use, starting with aspirin and ibuprofen and going up to the anti-TNF antibodies like Enbrel.
Are you saying that others had thought about TNF for Alzheimer’s?
Yep, absolutely. If you go through the literature, you can find any number of papers proposing such a connection – here’s a review of the field. Here’s a study (in mice, if you believe Alzheimer’s mouse studies, and some do) with one of Enbrel’s main competitors, Humira (adalimumab), and there are a lot more animal studies out there as well. Note that it’s a complex field; there are several forms of TNF and they have different activities. But it’s not just TNF in general that’s been proposed for Alzheimer’s before, it’s Enbrel in particular. Here’s a review on its effects when injected directly into the cerebrospinal fluid of Alzheimer’s patients, for example. Here is an actual clinical trial of Enbrel in human Alzheimer’s patients – it’s mentioned quickly in the Washington Post article, but it’s easy to miss that. (It was a small trial, and results were not compelling).
But didn’t Pfizer bury its data showing that Enbrel might have some use for Alzheimer’s?
If you are curious about what insurance claim data show about rheumatoid arthritis, Alzheimer’s, and treatment with Enbrel, let me refer you to this 2016 paper. The authors ran just such an analysis across 8.5 million patient records, and for all I know this is the same data set that Pfizer analyzed (unlike the Post, I haven’t seen their slide set, so I don’t know). The paper concludes that (1) Alzheimer’s disease is significantly more prevalent in rheumatoid arthritis patients than in those who have not been diagnosed with the disease, (2) the presence of other chronic conditions such as diabetes or cardiovascular disease significantly increased the risk of AD, (3) that treatment with anti-TNF antibodies as a class was associated with lower risk of AD, and finally (4) that when analyzed on a drug-by-drug basis that only Enbrel (of the various anti-TNF treatments) was associated with that decreased risk. And if you read the full paper, you find that it references a report from 2010 that showed improvement in cognitive function in a small set of Alzheimer’s patients treated with Enbrel for arthritis.
Why only Enbrel? The authors have several speculations based on the differences between the three approved anti-TNF antibodies, all of which are reasonable, but conclude that “Further studies directly comparing the efficacy of each agent in treating or preventing AD are required to determine the difference in the clinical benefits among these three agents“, and further note that since none of them get into the brain, that the mechanism for any benefit remains unknown. But the main point is that if you wanted to act on clues that Enbrel might have benefit in Alzheimer’s, they have been out in the open literature for years now. Not buried by an evil drug company. Out there for anyone to see and act on.
You can purchase Enbrel on the open market and run a trial of your own, if you so desire – but what you will find (unless you are fabulously wealthy and spending your own money) is that coming up with funding will be very difficult, because these data are still insufficient to go right into a large human trial in an area as fraught as Alzheimer’s. The use of anti-TNF antibodies for Alzheimers is not a crazy idea, but a lot of even better ideas have failed ruinously in this area. It’s worth investigating more, but anyone saying already that this is an overlooked cure has not worked in the field. Or, apparently, even spent fifteen minutes looking through PubMed. The Washington Post article would have benefited from doing just that. As it stands, it’s just worked up people for very little good reason. I like the newspaper, and I subscribe to it online, but this is not a good moment for it.