Replies to post #157712 on Biotech Values

[iwfal]

RVX -

a1-HDL is not he majority of HDL in the measurement of HDL-C. In fact, a1-HDL is around 15% of HDL-C as noted by the measures in the 2004 Asztalos paper. You had stated before that a1-HDL made up the vast majority of HDL-C, so I just wanted to correct that.

FWIW I agree completely that you are correct in the context of our on-going discussion regarding the papers with detailed assays of the subspecies. BUT in the context of the way in which I used the factoid I'll quote from Lipid On-line:

There is a strong correlation with total HDL-C and a1 subpopulation as it is the largest HDL particle.

and

HDL-C primarily reflects cholesterol levels within large, cholesterol-rich particles [graphic pointing to alpha-1 and 2] and lacks sensitivity to detect small cholesterol-poor particles [e.g.alpha-3/4 and betas]

http://www.lipidcenter.com/pdf/HDL_Structure_and_Classification_with_slides.pdf

I.e. my logic was correct (alpha-1 couldn't have such a large effect on CHD because HDL-C doesn't)

Finally note that it is very clear that the HDL subtypes do not add up to HDL-C because the assay effectively overweights the bigger particles - but exactly what this overweighting process looks like is not something I have figured out. I've just taken the word of Lipid On-line and another paper that the correlative math of the detailed assays supports it. Alpha-1 cannot be much more potent than HDL-c because the two are highly correlated.

If you think the second Asztalos paper erases the conclusions of the first paper, I think you need to read closer.

Suggest you re-read what I said of the conflict between the two Asztalos papers: "The authors who concluded that alpha1 was the font of cardiac beneficence subsequently un-concluded it within 1 year." the bolded part is what is important and consistent with what I have stated again and again in this thread I.e. the first A paper said every mg/dl of alpha-1 decreased risk by 25%, the second calculated it only decreased risk by 3.3% in the coherent population (see Table 4). A factor of 7 difference. And thus RVX-208's imputed benefit to CHD via alpha-1 is a 10% reduction in risk - which is not remotely competitive.

BTW - I find A's fig leaf of 'still finding alpha-1 as outsized important' amusing. They do quite a bit of that - e.g. trying to combine two trial populations which is a generally a big no no (see, for instance, Simpson's Paradox). Going to a subgroup. ... . That said, I agree that alpha-1 is likely to be a marker - just no more important than HDL-C, which should be no surprise given the strong correlation between alpha-1 and HDL-C (one paper gives it as r=0.87, which is exceedingly high).

And finally - I would suggest we agree to disagree about the math/science just because it is probable that neither of us have the time (this is NOT meant to be confrontational - just shear truth). But for the record I predict:

a) If RVX hits stat sig vs baseline and vs placebo. Glorious! I am sure they will release most of the data. (And, note, I do think this is possible. However it is a complete wildcard dependent on them having correctly identified the correct portion of RLT to change)

b) If RVX hits stat sig vs baseline - but not versus placebo they will: a) tout the "primary endpoint was stat sig", b) say the test vs placebo not sized big enough (true), c) not release the placebo data unless the difference is (WAG) 0.6% or more. So ultimately in many drug-is-a-failure scenarios you will not be able to tell how good their drug is - but they will make it sound great.

c) If RVX-208 misses stat sig vs baseline (and not versus placebo either) they will: a) tout the trend, b) talk of a subgroup, c) release minimal data other than a subgroup.

d) If RVX-208 actually increases the rate of atherosclerosis they will, of course, kill the drug.