Thread by @realDaveFeldman

1/ ☠️ Thread ☠️
Looking at Heart Disease vs All Cause Mortality

Let's start with a poll. Would you prefer:
1. Being less likely to die of heart disease
2. Being more likely to die of something that isn’t heart disease

2/ The “how” you die is important, but probably not as important to you as “when” you die.

But could we just look at the "how" and "when" of one kind of mortality and know whether it will truly extend lifespan?

3/ If I hold an experiment and my intervention group died 50% less by heart disease than my control.

Can I conclude this intervention group must necessarily live longer than the control group on average – particularly given heart disease is our number one killer?

4/ It cannot. We don't have enough information when looking at only one kind of mortality. What we *want* to infer is that heart disease went down and all other risk factors remained the same.

That's very intuitive. That's how we think of it like the poll answer in (1) above.

5/ What if something makes you more vulnerable to another mortal illness that isn't heart disease?

Technically speaking, that will reduce your risk of dying of heart disease.

For a hyperbolic example, consuming cyanide would reduce your risk of dying by heart disease 100%

6/ Which brings us to genetically high and low LDL cholesterol (LDL-C). There is a known association between LDL-C and heart disease, and this includes genetics.

So an obvious question comes to mind, do those with naturally lower LDL-C live longer?

7/ That was actually the very first question I wanted to find out at the very beginning of my research in Nov' 2015.

It seemed relatively simple -- if people with genetically low LDL-C were blessed with longer lifespans, then the case was made, full stop.

8/ There are actually many different genetic ways one gets to low/no LDL-C/-P with technical names like abetalipoproteinemia, familial hypobetalipoproteinemia, and PCSK9 LOF.

Do they live longer?

They don't.

Why?

9/ This brings us back to the beginning. Three scenarios:

1. Less heart disease risk, all else the same
2. Less heart disease risk, one or more risk factors worsen (trade off)
3. Heart disease same, one or more risk factors worsen (net loss)

10/ In all three of the above scenarios we get less death by heart disease and more death by something else.

There's just one way to help understand the trade off -- you have to track mortality by non-heart disease as well.

You need All Cause Mortality (ACM)

11/ A couple weeks ago @JakeKushnerMD brought this up regarding the podcast @ethanjweiss had with @skathire. (I only just now listened to the podcast today) Here's Jake's tweet:

@JakeKushnerMD @ethanjweiss @skathire 12/ There's an interesting exchange that follows where @ethanjweiss links @mvholmes graph from @mrbase2 where he compares two metrics very close to what I wanted to see:

Genetically higher LDL-C
vs
"Father's Age at Death"

@JakeKushnerMD @ethanjweiss @skathire @mvholmes @mrbase2 13/ To be sure, I was hoping for LDL-C vs simply "Age at Death". I found later that this endpoint actually does exist and ran the data for myself. Finally seeing

Genetically higher LDL-C
vs
"Age at Death"

...

@JakeKushnerMD @ethanjweiss @skathire @mvholmes @mrbase2 14/ But I'm not going to link the graphic yet. Yes, I'll hint that it doesn't look good for pro-lipid lowerers, but I'm going to take more time to better understand from their position why they'd say it is.

That's why I wanted to understand @mvholmes position and a few others...

@JakeKushnerMD @ethanjweiss @skathire @mvholmes @mrbase2 15/ Which brings me back to the original topic.

I don't see much value in looking at outcomes with heart disease alone. That simply tells you the how.

If lowering LDL-C had no effect on any other disease outcome, then we should see a net longevity, particularly with genetics...

@JakeKushnerMD @ethanjweiss @skathire @mvholmes @mrbase2 16/ But we already have many examples in the literature that suggest benefit from LDL-C/-P, particularly with the immune response. Naturally I have point to @siobhan_huggins's recent article here: cholesterolcode.com/lipoprotein-po…

@JakeKushnerMD @ethanjweiss @skathire @mvholmes @mrbase2 @siobhan_huggins 17/ So do I think there's a trade off?

I think it's quite likely.

To @JakeKushnerMD's point, we can only know when good scientists like @skathire can take these massive population analyses beyond heart disease and apply them to all cause mortality as well.