Thursday, March 30, 2017

Amgen share price and PCSK9 inhibition with Repatha (2)

I had an email from a PR company representing Amgen, re Repatha and all cause mortality. Here's the bit of interest:


I respect your opinion, but did want to share some additional information with you regarding the 2-year length of the study.

FOURIER was an event-driven study and was to conclude when least 1,630 hard major adverse cardiovascular event (MACE) events were accumulated. Amgen expected the study to run for 43 months with a 2 percent annual event rate in the placebo arm. However, the annual event rate in the placebo arm exceeded 3 percent and led to a faster accumulation of hard MACE events. Since the relative risk reduction in the hard MACE composite endpoint grew from 16 percent in the first year to 25 percent beyond 12 months, Amgen anticipates that a longer duration trial would have led to further relative risk reduction.

Would you please consider correcting this sentence of your post?

“The study was stopped early, presumably to stop the hard end points of dead patients from becoming too obvious.”


You can see how Amgen made their decision. Am I incorrect in my presumption about why the study was terminated early?

Well, technically yes. The protocol is laid out. That's unarguable. So they have a point and have designed the study well, from their point of view.

The fact that 444 people died in the treatment arm vs 426 in the placebo arm was not statistically significant, despite representing a 4.2% increased relative risk of death over the study duration.

What seems to concern Amgen is the implication that all cause mortality had any influence on the decision to terminate the study early. Obviously I cannot know whether this is the case and Amgen are certain that my presumption is incorrect. So maybe some compromise:

If we go with this I can reword the sentence to:

“The study was stopped early due to an unexpected excess of combined cardiac adverse end points in the placebo arm. At this time point the 4.2% increase in relative risk of all cause mortality in the treatment arm was not statistically significant”.

I don't think these facts are arguable with.

Well, that's been interesting. I feel somewhat honoured to have been contacted by a company representing Amgen to correct my presumptions!



bill said...

So, in the same time period, they expected 283
MACEs in the placebo arm (426 x two thirds). So
their treatment arm would have been relatively
much worse, with a presumably significant
increase of 444 - 283 = 161 or 36% higher
MACE rate in the treatment arm. It's all
number games anyway. There's no good reason
to lower cholesterol artificially. Eat low
carb and lots of fat.

I hope they're reading this.

Antonio said...

The control group, named "placebo", being under statins...


Tucker Goodrich said...

Good for Amgen, for explaining the problem, and asking nicely.

raphi said...

the lady doth protest too much...

Boundless said...

This PR initiative suggests that Amgen isn't ready to give up on Repatha, as Pfizer did with Bococizumab. That might be welcome news, if Repatha has truly beneficial cases, such as those with high Lp(a) who are avoiding destructive standard diets. The question is whether we will ever learn of those applications.

On the other hand, I suspect that the market response to the trial results may suggest how the insurance industry is going to respond. Amgen needs to be a bit more concerned with that audience than with individual lipid bloggers.

Serdna said...

From Section Statistical considerations of Rationale and design of the Further cardiovascular OUtcomes Research with PCSK9 Inhibition in subjects with Elevated Risk trial (my bolds):

"Assuming a 2% per year event rate in the placebo arm, 27,500 patients followed up for a median of approximately 43 months should provide 1,630 key secondary end points. Assuming an annualized event rate of 4.5% for the primary end point, there should be 3,550 events at the end of the study, yielding >99% power to detect the hypothesized effect. The executive committee and sponsor will monitor the blinded aggregate event rate data as well as other trial metrics, and study duration may be adjusted to ensure assumptions are met."

How many primary end point events do they got? From Table 2 of Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease: 2907. Yes, it is lower than 3,550. Nevertheless they had untied their hands at next Section (Data monitoring and lipid-monitoring committees) of their protocol paper (my bolds):

"Reviews of safety are being performed by an independent data monitoring committee (DMC) (online Appendix) approximately every 3 months. All DMC recommendations will be made based on a review of the totality of evidence from the unblinded data from the trial. There are no formal statistical monitoring rules for stopping the trial based on safety or efficacy. The DMC will consider a Haybittle-Peto monitoring guideline as one component of the totality of evidence."

So much precision boggles the mind. Haybittle-Peto boundary of 0.001 seems to have been applied to either any of the two composites or myocardial infaction alone (yes, its p-value is lower than 0.001), so it doesn't matter to them the statistic-unsignificant higher mortality (and statistic-unsignificantly rising). What a joke! That criterion either shouldn't be applied (my opinion) to anything less than overall mortality or should be applied only when the overall mortality doesn't get worse with the intervention. Silly me.

As it stands, it is not an event-driven trial: it is a whatever we see fit in order to stop the trial early at our convenience trial. And don't doubt it: its protocol was perfectly fiddled with to stop as early as convenient.

Bob said...

What a great company Amgen is! Their people read Hyperlipid!

cavenewt said...

"Amgen needs to be a bit more concerned with that audience than with individual lipid bloggers."

"What a great company Amgen is! Their people read Hyperlipid!"

While I'm tempted to agree with the latter, the truth is most big companies employ clippers to scour through publications searching for mentions of their products.

In the 90s my spouse-equivalent wrote a humorous popular culture column for the local small town newspaper. One week he made some parenthetical reference about kitty litter (kitty litter in a generic sense.) A few weeks later he received a letter on very heavy creamy stationary with gilded logo from the Golden Cat Corporation, manufacturer of the product Kitty Litter. In surprisingly friendly legalese he was informed that that is a trademarked name and could not be used for such purposes. To this day he treasures that letter.

With the explosive growth of the Internet in subsequent years, the army of people being employed to find references to trademarked products must have increased exponentially. Though I'm sure Google makes their job easier.

Boundless said...

re: What a great company Amgen is! Their people read Hyperlipid!

As cavenewt points out, it was likely their media minding minions doing the reading, but the detail in the resulting response suggests significant corporate involvement. It is, however, forthright engagement on the part of Amgen.

This sort of official reaction is surprisingly rare, and I suspect deliberately so, due to growing awareness of the Streisand Effect. What happened this time? My guess is that Peter's post got too widely linked to. Amgen may even have been asked about it through other channels.

Under another username, on another blog, I often answer questions of this sort:
"On this plan, can I eat Addictomate® brand Frosted High-PUFA Embalmed Durum Crisps with Calcium?"
to which I almost always answer:
"Umm, I wouldn't, and here's why."
The brands, despite being identified by name with specific product descriptions, for some reason, never ever sail in to defend their toxic food-like substances.

What I do see, on some other blogs, however, is evidence of company agents trying to pretend to be ordinary blog readers. Discussions of MDR for dietary glyphosate seem to draw out such astroturfers.

Bob said...

"re: What a great company Amgen is! Their people read Hyperlipid!"

As you might imagine, I was being ironic. I understand cavenewt's comment, but I'm inclined to agree with Boundless. The attention they lavished on Peter suggests they took the fact of a critical comment seriously. Which says much for Peter's impact "out there".

If they really were a great company, of course, there wouldn't be a Repatha or a FOURIER study. And if they really read Hyperlipid (REALLY read it), they might develop and test a drug to restore Complex I and the TCA and Leptin sensitivity in LCHSFA-eating post-obese people. And / or a really good cure for baldness (I could use it). And / or genuwine bottled water from the Fountain of Youth (I could use that, too). And I'm sure a bunch of other things as well.

Bob said...

P.S. I think Tucker was being ironic, too.

karl said...

There is another way to look at this -

1- With no statin skin in the game - this research had more deaths on statin than expected.

2- Lowering LDL once again does not seem to be the magic bullet.

3- All cause mortality in both groups was higher than expected even with LDL greatly lowered.

I don't think we evolved LDL to cause heart disease - and I don't think it does ( there could be an exception in the favorite study group of folks with sky high cholesterol).

I ended up taking Repatha based on what a cardiologist told me - that they were seeing regression. (My case is not typical - I can not take statins - so I'm only on Repatha ).

So what happens if we suspend the accepted narrative for a moment? What if the healing process that produces foam cells that crimp off arteries is only slightly slowed when they reducing LDL? That would explain what we see. What if reducing LDL is a costly dead end?

What if modern medical science is missing the point that they should really be focused on stopping the initial damage?

What if meldonium is compared with statins? ( Because it is not owned by US Pharma - we will never know ).


What if the proteins in LDL are like most other proteins we make - they have a many functions - not just one?

Peter said...

"What if reducing LDL is a costly dead end?"

What indeed!


George Henderson said...

There seem to be a few studies that say if you go to hospital with low LDL your chances of dying are higher, explaining how reduced events through this method can still = more deaths.
LDL is supplying fat-soluble vitamins and antioxidants, CoQ10, cholesterol, all of which are useful to struggling myocardial cells. It's also great for binding LPS and regulating the response to it, making it a key control on inflammatory response and sepsis. And it usually means you have enough fat energy in your diet that you're not limited to glycolysis at a time when that might not be optimal.