Thread by @udarnik

I think @JohnCendpts did a great job in laying out some of the hype behind the @washingtonpost story on Enbrel and Alzheimer's. If you don't subscribe to Endpoints and have any interest in the life sciences at all, do so.

endpts.com/the-washington…

John's expose of the bad Academic behavior of the guy behind the 'whistleblowing" is great. It's a shame the post ran with this story without checking out the full background. Make no mistake, the post story is just *bad* science journalism.

It's not like the Academic world doesn't have legitimate beefs along this line with Pfizer: ask any transplant surgeon about Xeljanz, and you are likely to get an earful.

But the Enbrel /Alzheimer's story is not, IMO, a case of Pharma abandoning a potentially beneficial therapy.

I find this story a bit ironic, because it's not like the medical community doesn't have legitimate beefs with Pfizer along these lines. Ask any transplant surgeon about Xeljanz, and you are likely to get an earful.

But let's dig in a little deeper to the evidence presented by the Post. This alleged benefit is based on an insurance claims database. After I left the big blue egg, I spent a few years in the world of big data, and let me tell you, EMR / claims data is *dirty*.

Not every doc enters data in a standard way, not every doc is competent at differential diagnosis. At *best* claims data is enough for hypothesis generation. At worst, it's useless. *No* Pharma or Academic institution is going to launch a major trial based on claims data.

So that leads to the question: were there any preliminary studies to test the hypothesis? (And if there were, that really puts lie to the Post's claim that Pfizer hid this from the world, doesn't it?)

Oh, look! Way back in 2008 some, shall we say, less than well established researchers in the field published a 12 - count 'em 12, (if you have any statistics background at all: TWELVE!) patient study purported to show a benefit:

bmcneurol.biomedcentral.com/articles/10.11…

That "study" is little more than an oversized case report, so did anyone else look at etanercept in Alzheimer's? Why yes, yes they did. Were now *really* putting holes in the "Pfizer kept this from the medical community" narrative, aren't we?

This study, from 2015, looked at 41 patients, and was properly blinded. This is little more than a Phase I type of study, charitably it might be called a Phase IIa, but the conclusion is interesting:

ncbi.nlm.nih.gov/pmc/articles/P…

"While there were some interesting trends that favored etanercept, there were no statistically significant changes in cognition, behavior, or global function."

Yeah.

Alzheimer's is a graveyard of druggable compounds, this is not exactly a ringing endorsement for moving forward.

One of the main questions for an approach like this is: *how* does this work?

You, as a layman might think this is a silly question, as long as it *does* work, but if you do think this is a silly question, you should go read the raison d'etre for the blog Science Based Medicine.

*Clinical* based or "Evidence" based medicine relies on clinical trials. But Science Based Medicine seeks to remove the noise in the evidence by requiring at least a plausible biological rationale for an approach.

The reason is that in the statistics of small numbers we must use in clinical research, noise abounds.

If you toss a coin 100 times, there will be chains of heads or tails, sometimes 8 or 10 long, in that sample. That does NOT mean the coin is biased.

Similarly, when doing clinical trials without a rationale, some approaches that are objectively bunk will show a spurious positive result. Asking for a rationale helps eliminate that noise.

Hence, *Science* Based Medicine.

sciencebasedmedicine.org

So, to be science-based, the next thing you should ask yourself is: what is Enbrel (etancercept), what does it do, and why would it work in Alzheimer's?

Well, etanercept is what we call a molecular sponge. It soaks up a chemical called Tumor Necrosis Factor Alpha from the blood.

TNFα is a signaling molecule in the immune system. To be *really* brief, it promotes inflammation in a lot of diseases where the body confuses self for non-self: autoimmune diseases.

Functionally, the normal receptor for TNFα that is on the surface of immune cells was cut off and attached to the bottom chain of an antibody (IgG1). The cell receptor binds to TNFα as if it were on a cell and takes it out of circulation so it can't promote inflammation.

The antibody tail makes it harder for the body to flush the molecule through its normal elimination systems, improving its circulatory half life.

The point is that this is a big honking molecule.

enbrel.com/hcp/about

Now, if you know anything about drugs & anatomy, you know that many drugs don't make it into the brain because of something called the blood brain barrier.

Evolution put this in place to keep random crap in the bloodstream from contaminating the brain.

en.wikipedia.org/wiki/Blood%E2%…

Random crap might plausibly include honking big molecules like Enbrel. But assume that Enbrel does make it past the BBB. Ascience based approach would also demand a plausible explanation for what it's doing there. In other words what role does inflammation play in Alzheimer's?

Well, the primary immune cell in the brain, microglia, are implicated in the pathogenesis of Alzheimer's via inflammation:

sciencedirect.com/science/articl…

However there is no evidence that the inflammatory signaling noted as being correlated with early Alzheimer's involves TNFα. In fact, there is evidence that TNFα is not involved:

jnnp.bmj.com/content/89/4/3…

OK, let's wave our hands and postulate that Enbrel does *something* anti-inflammatory in Alzheimer's patients' brains. Maybe it's a downstream effect!

Well, it would have to get into the cerebrospinal fluid to do that, right?

There's no strong evidence that it does, in fact groups are working on modifying TNFα antibodies specifically to pass the BBB and test hypotheses like this:

pubs.acs.org/doi/abs/10.102…

In fact, our best guess is that in order to get therapeutic levels of etanercept (I'm using the generic name and brand name interchangeably) into the brain, the administered dose would need to be 70 times as high as that given for autoimmine disease:

onlinelibrary.wiley.com/doi/full/10.11…

This brings me back around to the study Pfizer supposedly hid from the world. Immunomodulators such as Enbrel carry significant dangers, & one of those dangers is that they reduce the body's ability to fight off the Cunninham virus and cause a neurodegenerative disease called PML

A competitor to Enbrel, Rituxan, almost got pulled from the market because of three cases of PML in Lupus patients (an off label use), and the drug is still being monitored for this side effect:

ncbi.nlm.nih.gov/pmc/articles/P…

This is an indication that Rituxan may in fact slip through the BBB in some patients (the Lupus cases were explained away as already having compromised BBBs, IIRC).

Enbrel has very little of such history, but it is of some concern for all such drugs.

So when I looked at the insurance database study, one thing immediately sprang to mind - all the AZ patients had to have had a diagnosis of some *other* - autoimmune - disease in order for an insurance company to be paying its *very* expensive price tag in the first place.

In case you don't know, fatigue is a major, major symptom of auto-immune disease, and it exacerbates the symptoms of AZ.

So extrapolating from "we see a difference in a couple of hundred controls and about a hundred Enbrel AZ patients" to "this is a great idea to put into normal geriatrics" is a HUGE leap of faith.

Especially given the risks of these drugs. I mean, it's called TUMOR Necrosis Factor Alpha because it plays some role in the natural process ferreting out cancer cells before they grow (not as much as originally believed, though).

One of the major side effects - very common - is an increase in infections. You're crippling a pathway in the immune system. So there is a very real danger of using a drug of marginal utility in a geriatric population especially susceptible to infectious sequelae.

But the optimists might keep asking: what about that glowing 12 patient study?

Leaving out the larger study that found bupkis over the same time frame, the doc who ran it is ... problematic.

Very problematic.

sciencebasedmedicine.org/enbrel-for-str…

So problematic he sued the fine folks at SBM for running that piece ... and lost:

sciencebasedmedicine.org/tobinick-lawsu…

So what do we have at this point? We have an insurance database study where the patients have multiple serious co-morbidities, where the progression of AZ was not a pre-specified endpoint so the means of assessing the disease is all over the map.

We have a positive 12 patient study of Enbrel *infused into the spine* of AZ patients conducted by a community practice, non-Academic doc who sued SBM when they questioned his methods.

And we have a largely negative follow-up trial.

i'm being very generous to the insurance database study in assuming all the docs made the right diagnosis and the patients actually has AZ and not some other form of dementia. Based on my previous work with claims databases, that is one hell of an assumption.

If you're Pfizer management, what do you do? Now remember, Pfizer had Aricept. They know the market & they know how to make money in it. If this was good evidence, why *wouldn't* they go after it?

First, as John reported, AZ is tough, & no one has had much success since Aricept.

Second, sure as God made little green apples, if you stick Enbrel into a geriatric population with Az, all sorts of side effects will come up. Remember the Rituxan in Lupus example a few posts back in this thread? The AZ trials might get the drug withdrawn by the FDA.

Now, Pfizer's management is probably thinking about how the loss of a major revenue stream will hit them *straight* in the bonus pocketbook, but there is an ethical argument to be made:

If the evidence for the new indication is weak & risk of side effects high, you are doing a disservice to the autoimmune patients benefitting from Enbrel right now by running the trials.

So looking at the totality of the evidence, I think there is a huge chance the claims data dredge is garbage. I think Pfizer management felt that, too.

Then there are the questionable people floating around this, from the Harvard dude who has lost Biogen and Eisai so much money to the spinal infusion quack. This is starting to look radioactive from a Pharma decision making standpoint.

Best not to step into that minefield.

But there is more. One thing John alluded to, and one he may not be aware of.

John alluded to the fact that Enbrel is actually Amgen's drug, Pfizer just co-promotes it. Enbrel means a LOT more to Amgen than to Pfizer, and they would be even less likely to run with crappy data like this than Pfizer would.

The Post puts all the blame on Pfizer, showing massive ignorance of how these kinds of partnerships work.

But there is another reason Pfizer might hesitate to run with publishing this data.

Back when Pfizer acquired Neurontin in the Parke-Davis merger, PD was trying to pull a fast one by having its reps hand out reprints of three studies that showed worked in neuropathic pain (it was developed as an anti-epileptic).

But they never bothered getting FDA approval.

If you're familiar with FDA regulations, that is off-label promotion, and PD tried to skirt around this with a bogus First Amendment claim that cost them a lot of money. Pfizer, mostly rightly, claimed this was PD's management and not Pfizer, so somewhat got off the hook.

Unfortunately for Pfizer, a couple of rogue teams (Bextra and Geodon) pulled some *spectacular* bullshit soon after they reached an agreement with the DoJ about Neurontin.

This bullshit included not only off-label promotion, but market research to ascertain that the off-label promotion had indeed worked. It cost Pfizer $2.3 billion:

nytimes.com/2009/09/03/bus…

The DoJ was *pissed* when the off-label promotion came to their attention, so in addition to the fine, they told Pfizer in no uncertain terms that the next time something like this happened, the CEO was going to jail.

So naturally, management would have been thinking about *that*, too, when they decided that not publishing that little insurance claims study was the better part of valor.

I have no love lost for Pfizer management. But they made the right call, here.

For multiple reasons.

Thus endeth my presentation of evidence for why I think the @washingtonpost is guilty of both lousy science reporting AND lousy business reporting.

Do better, guys.