Would you agree that smoking is causally related to lung cancer? Yes. So just to be clear, Andrew, you do not think that it's just an association that smokers get more lung cancer? No, I do not. You, in other words, you believe that smoking causes lung cancer then? Yes. Okay.

I mean, there are a number of mechanistic steps in between. I mean, if somebody was really wanting to drill into the logic, they could say, okay, it's not actually the smoking, it's a, you know, some disruption of the endothelial cell lining that, you know... But smoking triggers that, that triggers that.

I assume so. And I agree with you, by the way. I think the data are very clear. I'm very relieved to hear that. Yeah, yeah. But I'm going someplace very important here, because if there's one topic that doesn't get enough attention in medicine, it's causality. And causality is an obsession of mine.

Like most of the day on some level, I sit around thinking about causality. And I think the hardest part about studying medicine with respect to human beings is how difficult it is to infer causality for most things that we do. So if you believe that smoking is causally related to lung cancer, then smoking cessation reduces the probability of lung cancer.

That is a logical equivalency. There can be no debate about that. What if I said to you, Andrew, this is going to be our new philosophy around smoking cessation. I'm going to anoint you the czar of smoking cessation. So if people pick up smoking, no problem. We're going to let them smoke.

But we're going to assess their risk for lung cancer using a model that predicts when their 10-year risk of lung cancer gets above a certain level, we're going to recommend that they stop smoking. So we're going to look at their age, their sex, their family history, some biomarkers that might help us.

We're going to even do scans of their lungs. And once we think they cross a threshold where their risk of lung cancer is high enough, let's just say it's 25%, boom. You make them stop. You tell them it's time to stop. Is that a logical approach to treating smoking and lung cancer?

Or would it be better to say, given that we know cigarettes are causally related to this, how about you never start smoking, and the minute you do, we pull the cigarette out of your mouth and explain to you that you're doing something that is causally related? Of course, it would be the latter, not the former.

It would be idiotic to suggest that we endorse smoking until you cross a certain threshold. Well, this now becomes the germane question. There is no ambiguity that ApoB is causally related to atherosclerosis. You know, how can I tell you that? I can tell you that looking at all of the clinical trial literature, all of the epidemiologic literature, and perhaps even most importantly, the Mendelian randomizations.

All of these things tell us... Because by the way... Mendelian randomizations meaning genetic mutants, humans out there that make very little ApoB or excessive ApoB. And very much, exactly. So we have the whole gradient. So you can say if you make very little, you aren't going to die as quickly in your life as if you make too much.

That's right. So the Mendelian randomization is such an elegant tool where you basically let genes do the randomization. And as you said, there is a gradation of LDL concentration or ApoB concentration that occurs from insanely low to insanely high. And this is a wildly polygenic, polymorphic set of conditions.

And we can look at the outcomes of those people based on the random sorting of those genes and there's no ambiguity. LDL is causally related, LDL cholesterol or ApoB, causally related to atherosclerosis. Well, if that's true, and I haven't seen a credible argument that it's not... There are people who argue that it's not, by the way, but they just don't have credibility in their arguments.

Then you have to say that what we're doing in medicine today is very backwards. Because what we're doing in medicine today is the following. We're saying, I'm coming at this in a long way, but your question is so important that I want to answer it this way. We're answering your question today as follows.

We're saying, "Andrew, let's do a 10-year risk calculation of your risk of MACE." MACE stands for Major Adverse Cardiac Event. It is the metric we use in medicine. So a Major Adverse Cardiac Event is a heart attack, stroke, or death basically resulting from these things. And we have calculators that are pretty good at predicting your 10-year event risk.

They'll look at your cholesterol levels, your blood pressure, they'll ask if you smoke, they'll ask some family history questions, and they'll spit out a number. Now we should do yours after the fact. And I don't know, if we did it for a person who's as, you know, you're in your mid-40s, like it would probably spit out less than 5% risk for a Major Adverse Cardiac Event in the next 10 years.

In fact, the models don't even work if age is below 40. So the first time I went to do one of these tests when I was in my mid-30s, I couldn't do it. The algorithm breaks. That's sort of like, you know, just doesn't work. So the implication there is if your MACE risk is less than 5%, the thinking is you do not need to treat LDL or ApoB.

I argue that that makes absolutely no sense. It's just as idiotic as the analogy I used around smoking. If a risk is causal and it is modifiable, it should be modified regardless of the risk tail in duration. So then the question becomes, to what level? And again, the earlier you start, the less aggressive you need to be, the less damage that's there already.

So for example, we do CT angiograms on our patients. If the CT angiogram shows no evidence of calcification, no evidence of soft plaque, that means grossly their coronary arteries are still normal. Histologically, they're probably not because nobody probably makes it to our age with histologically perfect coronary arteries. You know, we might be satisfied with a person's ApoB being at the fifth percentile of the population, which would be about 60 milligrams per deciliter.

But if we have any other factors, meaning we're starting later in life, you know, or a person already has gross evidence of disease, calcification, soft plaque, family history is significant, any other risk factors are present, I mean, we'll treat ApoB to 30 to 40 milligrams per deciliter, which is, you know, probably the first percentile.

And if somebody's sitting up in the, say, low 130s, where does that, what kind of flag does that raise for you? And I realize it's highly contextual, age, et cetera. No, no. It's a huge red flag. Again, just because something is causal doesn't mean you're guaranteed to get it.

There are smokers who don't get lung cancer. So, you know, there's going to be somebody listening to this who says, "My grandmother is 95 years old, she's, her cholesterol is sky high and she's alive and well." And I will say, "Absolutely. There are a lot of people walking around that way." Just as there are a lot of smokers walking around who don't get lung cancer.

You can't, you can't impute these things on an individual basis. You basically have to ask the question, "How do I make the best judgment about an individual from heterogeneous population data and based on what are causal and non-causal inferences around risk?"