The Peter Attia Drive - #240 ‒ The confusion around HDL and its link to cardiovascular disease ｜ Dan Rader, M.D.

00:00:00.000 Hey, everyone. Welcome to the drive podcast. I'm your host, Peter Atiyah. This podcast,

00:00:15.500 my website, and my weekly newsletter all focus on the goal of translating the science of longevity

00:00:19.840 into something accessible for everyone. Our goal is to provide the best content in health and

00:00:24.780 wellness, full stop. And we've assembled a great team of analysts to make this happen.

00:00:28.920 If you enjoy this podcast, we've created a membership program that brings you far more

00:00:33.320 in-depth content. If you want to take your knowledge of the space to the next level,

00:00:36.920 at the end of this episode, I'll explain what those benefits are. Or if you want to learn more now,

00:00:41.760 head over to peteratiyahmd.com forward slash subscribe. Now, without further delay, here's

00:00:48.100 today's episode. I guess this week is Dan Rader. Dan is a professor of molecular medicine at the

00:00:55.820 Perlman School of Medicine at the University of Pennsylvania, where he conducts translational

00:01:00.140 research on lipoprotein metabolism and atherosclerosis with a particular focus on the

00:01:05.740 metabolism and most importantly, function of high density lipoproteins, HDLs. Dan has received

00:01:13.060 numerous awards and has been elected to the American Society of Clinical Investigation,

00:01:17.560 the Association of American Physicians, and the National Academy of Medicine. He also currently

00:01:21.580 serves on the board of directors of the International Society of Atherosclerosis, the board of external

00:01:26.380 experts at the National Heart, Lung, and Blood Institute, and the advisory board for the clinical

00:01:31.320 research of the NIH. In this episode, we focus our entire conversation around high density

00:01:37.180 lipoprotein or HDLs. Now, as any listener of this podcast will know, we have no shortage of content

00:01:42.740 around lipids. And we focus a lot of that energy on the ApoB side of the family. That is,

00:01:48.620 to some extent, VLDLs, and of course, LP little a, which is a subset of LDL. However, this is the

00:01:53.840 first time we're doing a dedicated podcast on HDLs. Now, the reason for that, in other words,

00:02:00.520 the reason for that disparity is largely because HDL biology is so much more complex and we know so

00:02:06.820 much less. I mean, at the highest level, I think people generally think of HDL as quote-unquote

00:02:11.080 the good cholesterol, but you've no doubt heard me rail on the stupidity of such a designation.

00:02:15.800 But there is clearly something good about HDLs as in the lipoproteins, not the cholesterol. 0.62

00:02:21.740 And in this discussion, we really talk about everything from the biology of the HDL,

00:02:28.920 its genesis, its origin, its metabolism, its life cycle, and of course, its function. And we also

00:02:33.580 talk about why it has been so complicated to use pharmacologic interventions on the HDL side of the

00:02:39.920 equation to impact atherosclerosis. Conversely, of course, it's been the easiest thing, I would say,

00:02:45.620 in medicine, and perhaps the greatest success of modern medicine, especially as it comes to

00:02:50.280 cardiovascular disease, has been our ability to manipulate the APO-B side of the equation as

00:02:55.320 opposed to the APO-A side of the equation. That's one thing that's going to be important here when

00:02:58.960 we get into the terminology. When we talk about APO-A in the context of HDLs, we're talking about

00:03:04.280 APO-A as in a big A, which has no bearing whatsoever to APO little a, the thing that of course defines an

00:03:10.480 LP little a. But I digress. That is simply one of the many details we get into. So as I said,

00:03:16.240 we're going to talk here about the genesis of the HDL, the structure, its metabolism. We talk about

00:03:20.300 the differences between HDLs, LDLs, the difference between the HDL measurements, so what exactly is

00:03:25.640 HDL cholesterol versus APO-A concentration versus HDL particle number. We talk about the idea that

00:03:31.400 having a high HDL means you don't need to worry about cardiovascular disease and how that's obviously

00:03:36.000 going to be bunk. I'm just going to let the cat out of the bag on that one. We speak about CTEP

00:03:40.000 inhibitors, which are a class of drug that have been repeatedly used to try to increase HDL cholesterol

00:03:45.940 with the hopes that that would reduce cardiovascular disease. And finally, we end the conversation around

00:03:51.240 some of the new thinking around HDL and neurodegenerative disease, something again,

00:03:54.700 I learned a lot about. So without further delay, please enjoy my conversation with Dan Rader.

00:03:59.380 Well, Dan, thank you so much for making time to join us on the podcast today. Tom Dayspring really

00:04:10.620 recommended you highly. And anytime Tom Dayspring says to me, you should have so-and-so on the podcast

00:04:16.140 to talk about anything that has to do with lipids, I immediately pay attention.

00:04:20.520 Tom's an amazing guy.

00:04:21.780 Now, in particular, the subject matter I want to explore with you today is probably the area of

00:04:27.960 lipidology that I personally am the least familiar with. And that has to do with kind of one half of

00:04:36.280 the lipid family. You know, I explain this to patients as there are broadly speaking two families,

00:04:41.840 the APOB family and the APOA family. We spend obviously so much more time talking about the

00:04:46.400 APOB family based on, I think, two things. One is our clearer understanding of it and two,

00:04:52.080 the direct and causal relationship to pathology. But I often sort of waffle a bit when I'm talking

00:04:57.500 about the APOA side of the house. That's really why I'm excited to be sitting down with you today.

00:05:02.160 But perhaps for the person listening to this who doesn't even understand what a lipoprotein is yet,

00:05:07.440 never mind what the two families are that we're talking about, can you take it back to the beginning

00:05:12.840 for where do these things fit into the broader architecture of our existence?

00:05:17.880 Lipoproteins are these big complexes that really are evolved to transport lipids within the blood.

00:05:23.320 You know, lipids are like oil. They don't mix well with water. You know, oil droplets float to the

00:05:28.100 top of a puddle. We wouldn't be able to transport lipids in the blood if we hadn't evolved a mechanism

00:05:33.840 to do that. And lipoproteins are basically the mechanism to do that. As their name suggests,

00:05:38.920 they're lipoproteins. They have lipid in the core, and then they have proteins that dot the surface.

00:05:45.420 That allows them to be transported in very complex and sophisticated ways within the bloodstream

00:05:52.060 in terms of how they're metabolized, in terms of the receptors they bind to, and in terms of the

00:05:57.840 specific proteins and lipids that they carry. You've talked a lot about APOB lipoproteins. I've

00:06:03.180 listened to several of those podcasts. They're really fabulous. And of course, the APOB lipoproteins

00:06:08.420 really primarily transport or evolved to transport triglycerides as a source of energy, you know,

00:06:15.380 both from the gut to adipose and muscle and heart, as well as from the liver during times when we're

00:06:22.660 fasting. You've talked a lot about that. And as you pointed out, they are very important in terms of

00:06:28.280 causal relationship to atherosclerotic cardiovascular disease. HDL, which we're going to be talking about

00:06:33.000 today, is also a lipoprotein. It's a very complex lipoprotein. It doesn't have this key protein,

00:06:39.540 APOB. That's what differentiates them. That's why we often refer to the APOB-containing lipoproteins,

00:06:44.180 and then HDL is the other half, as you said. And HDL really is characterized by a different protein

00:06:49.520 called APOA1. HDL also transports lipids, especially cholesterol and some other complex lipids,

00:06:56.600 and we'll be getting into that more. But essentially, lipoproteins are lipid transport vehicles

00:07:01.740 within the blood and really evolved to do that function. A couple of things I'll just throw in

00:07:06.640 for folks, again, maybe new to some of this, which is all of this stuff we're talking about is so

00:07:11.960 important in terms of these lipoproteins, because unlike glucose, electrolytes, things that we kind

00:07:19.000 of take for granted that are water-soluble, we use our circulatory system as the freeway to move these

00:07:25.400 things around. And unfortunately, as you said, because of the insolubility of both cholesterol,

00:07:31.900 triglyceride lipids, et cetera, we have to come up with this more complicated system.

00:07:36.160 I think the second thing I'll throw in just for folks, because it could get confusing with the

00:07:41.220 nomenclature, when you talk about APOA, I want to make sure people understand we're not talking about

00:07:45.800 LP little a, which has another APOA little a. So maybe just clarify for folks the difference between

00:07:51.580 those two, because we'll do our best to not talk about APO little a today at all.

00:07:56.880 Yeah, APO little a is, or LP little a is really important, but not the topic of today's discussion.

00:08:03.080 I think as Peter already said, the main thing to remember is that that A is lowercase,

00:08:09.480 whereas the A we're talking about today is uppercase. I know it sounds crazy. It's just the

00:08:14.360 way things evolved historically, but we're talking about a APOA, uppercase A, and the most important

00:08:21.200 one has a one after it. So APOA1. So you're going to hear us refer to APOA1 a lot, totally different

00:08:28.260 than the LP little a, which is also an important topic, but not for today.

00:08:33.560 Yeah. It's a shame that the lipid community back in the 60s, 70s, and 80s didn't hire 1.00

00:08:38.880 kind of a PR firm to sort of help with the naming of this stuff, because it is hands down some of the

00:08:46.160 most complicated nomenclature. And if they realized back then, I'm half joking, that it

00:08:51.840 wouldn't be a bad idea for patients to actually understand this stuff, I think we could have done

00:08:55.420 better. Okay. So you've already alluded to kind of different APOAs and APOA1. Again, that's APO

00:09:02.120 capital A dash Roman numeral one is the most important, but let's go back even further than

00:09:06.780 that. So again, not to draw the APOBs into this again, but we kind of technically now have these two

00:09:12.420 lineages of APOBs, right? We have the APOB 100 and virtually all of the time that we say APOB,

00:09:18.100 we really mean APOB 100. And that's to contrast it with APOB 48, the one that you kind of alluded

00:09:23.720 to very briefly as transporting chylomicrons from the gut for the use of energy. Can you talk a little

00:09:29.720 bit about the, or not a little bit, maybe a lot of it, about the genesis of the high density lipoprotein?

00:09:36.120 What is its parent? Where is it formed? And how does it move through its evolution such that

00:09:42.400 when we're measuring HDL cholesterol, something we'll talk about in a minute,

00:09:46.760 we have a sense of where it came from in the same way that we understand VLDLs, IDLs, and LDLs.

00:09:51.800 With all due respect to the APOB community, and I'm part of that community too,

00:09:55.940 I would say the metabolism of HDL is perhaps an order of magnitude even more complex. And that's

00:10:01.580 obviously what we're going to be working our way through today. So I'll start relatively

00:10:05.420 simplistically. And then Peter, I know you'll lead me to increasingly more detail about the process.

00:10:10.920 And by the way, I don't think anyone's going to disagree with that, Dan.

00:10:14.840 Meaning, I don't think anybody who spent even a modicum of time looking at this literature,

00:10:20.120 it is hands down the most confusing stuff in the lipid space.

00:10:23.940 Again, I'll start with APOA1 as the key protein with HDL. The analogy is to APOB, as Peter alluded to.

00:10:31.440 Although one big difference is APOB, as your listeners know, APOB stays with that particle

00:10:36.720 throughout its lifetime. So it gets made by the intestine as B48 or the liver as B100.

00:10:42.840 It stays with that particle. There's one molecule of this huge APOB protein and that it basically

00:10:48.320 then ultimately gets taken up mostly by the liver after its time in the blood. APOA1 is also made by

00:10:55.220 the intestine and the liver. There's some parallelism there. It also is a core protein,

00:11:01.600 essentially, of HDL. But unlike APOB, there are several molecules of APOA1 on any given HDL

00:11:10.280 particle. We'll probably talk more about that, but anywhere from one to four, maybe in some cases a

00:11:15.240 little bit more. And also APOA1 doesn't stay with the particle. It can exchange onto other types of

00:11:22.540 mostly HDL particles, but even sometimes onto APOB-containing lipoproteins. What we've really

00:11:28.260 learned in the last decade or so is that APOA1 is put into the blood by either the intestinal

00:11:34.780 enterocyte or the hepatocyte in the liver as more or less a free protein. So it's secreted as a protein.

00:11:42.360 One of the first steps that has to happen for HDL to be formed is that HDL, once it comes out of the

00:11:48.400 cell, engages with a key transport protein called ABCA1. And Peter, stop me if I'm already getting

00:11:55.060 too detailed. ABCA1, we'll come back to that, which its role, ABCA1's role is to take lipid,

00:12:02.460 cholesterol, and other types of lipids called phospholipids from the cell and export them to

00:12:08.800 the newly secreted APOA1. So APOA1 has to acquire lipid, particularly phospholipid and cholesterol,

00:12:16.000 soon after it's been secreted in order for the so-called nascent or early HDL to start forming.

00:12:23.920 We'll come back to this when we talk about genetics, but we know this because humans that

00:12:28.340 lack ABCA1 have virtually undetectable HDL. And that is because they cannot, they make plenty of

00:12:36.000 APOA1, but they cannot protect that APOA1 with lipid once it's secreted and it goes out like a

00:12:42.360 rocket from the blood. Basically can't, almost can't measure it in the blood. So that's the first

00:12:46.980 step in terms of HDL biogenesis, if you will. What other phenotype do those patients have? So

00:12:52.780 in the absence of cholesterol accumulating in what would be an HDL, does it end up where? More of it

00:13:01.100 ends up in the LDL or back in the liver? Where does it go? This is the value of studying these very rare

00:13:07.540 human genetic disorders. And it just tells us a lot about what specific proteins or genes are doing.

00:13:13.420 This disorder, which we can talk about the history. It's a fascinating history. It's called Tangier

00:13:18.280 disease. It's named for a small, flat, bizarre island in the middle of the Chesapeake Bay where

00:13:23.840 the first patient was discovered. This disorder is primarily associated with accumulation of cholesterol

00:13:30.660 in macrophages throughout the body. So tonsils, spleen, places where there are a lot of macrophages,

00:13:37.860 the cholesterol really builds up. And they have huge sort of orange colored tonsils. They have big

00:13:42.960 spleens that sometimes rupture. They have big livers because there are a lot of macrophages and

00:13:47.000 macrophage-like cells in the liver. And they also have some neurologic type issues, especially neuropathy,

00:13:54.260 which maybe we'll leave to later when we start talking about HDL and its relationship to the

00:13:58.460 nervous system. They may or may not have some increased risk of atherosclerotic cardiovascular

00:14:03.160 disease. But as we'll talk about, that relationship between HDL and atherosclerosis is quite complex.

00:14:09.180 This disorder has helped to inform that a little bit. Another sort of structural question, Dan,

00:14:14.200 if you were to take a quasi-mature garden variety HDL particle, and again, it's more complicated there

00:14:20.960 because of what we're talking about and put it next to a comparable LDL particle. What's the

00:14:27.780 difference in size? And can you explain a little bit why they come up with this different name of

00:14:32.180 high versus low density? What specifically is that referring to?

00:14:35.540 That's a great question. Again, this is a historical artifact of how lipoproteins were discovered and

00:14:39.820 named. First of all, HDLs are much smaller, I would say, you know, in the neighborhood of one-fifth to

00:14:46.280 one-tenth the size of an LDL. And of course, much yet smaller than triglyceride-rich lipoproteins.

00:14:51.700 You know, lipoproteins have lipid. As I mentioned earlier, lipid floats. So lipids create buoyancy

00:14:57.780 to particles. And the way lipoproteins were discovered is that when you subject plasma to

00:15:04.000 spinning, the lipoproteins spin to the top. And they spin to the top at different rates under

00:15:10.020 different forces. The low-density lipoproteins have more lipid or bigger, more lipid. They

00:15:16.120 spin to the top more easily. The high-density lipoproteins still have lipid. They do spin,

00:15:22.180 but they don't spin quite as quickly or easily. So they're higher density than the low.

00:15:26.760 And then in another artifact of sort of the history, Peter, after low density was discovered,

00:15:32.660 something even lighter that spun up even more easily was discovered. And of course,

00:15:37.380 low was already taken. So that had to be called very low. And so there's the very low-density

00:15:43.840 lipoproteins that you've also talked about at length. And then finally, the chylomicrons,

00:15:47.680 the huge ones that come from the intestine, they're just so light and so buoyant that instead

00:15:52.300 of calling them very, very, very low, they were just given the name chylomicrons.

00:15:57.120 How long does an HDL particle last? And does it mean the same thing given that it's transferring

00:16:03.480 its APOA in a way that makes it a little different from, as you said,

00:16:08.480 the way the VLDL to IDL to LDL always stay with one lipoprotein, APOB, that allows us to kind of

00:16:15.720 track its life? Before we get into kind of how HDL is removed from the blood, could I take a little

00:16:22.020 bit more time to explain what happens after that? Absolutely.

00:16:25.540 APOA1, quote-unquote, nascent HDL particle gets formed because there's actually a bunch of stuff that

00:16:31.020 has to happen first before we start talking about how the HDL then gets removed from the blood.

00:16:35.680 But so the APOA1 or what we call the nascent HDL has, you know, APOA1 has phospholipids and it has

00:16:42.680 what I'm going to call free cholesterol. And this is a concept that you may or may not have addressed

00:16:47.500 before, I can't remember, which is cholesterol itself as a molecule can have a fatty acid attached

00:16:53.700 to it. When you attach a fatty acid to a hydroxyl moiety on the cholesterol molecule,

00:16:58.980 you create what we call a cholesterol ester. And the cholesterol ester is even more hydrophobic or

00:17:07.040 oily than the free cholesterol because it has this additional fatty acid sticking off it,

00:17:12.740 which is obviously oil. It's absolutely critical for HDL, for the HDL maturation process to have this

00:17:20.820 fatty acid attached to the cholesterol after the nascent HDL has been formed. And there's another

00:17:26.620 key enzyme that is responsible for that so-called lecithin cholesterol acyltransferase. It's the last

00:17:33.260 time I'm going to say that. I'm just going to call it LCAT. That's its nickname, LCAT, L-C-A-T.

00:17:37.840 So LCAT rides on the HDL particle, on that nascent particle. It takes a fatty acid from one of the

00:17:44.560 phospholipids on the particle and it transfers that fatty acid to the cholesterol and creates the

00:17:50.720 cholesterol ester. Remember, that cholesterol ester is more oily. It actually then moves to sort of the

00:17:56.980 center of the particle because it wants to be by itself and not interacting with the water.

00:18:01.740 And it starts to form the core of the HDL particle. So the core of the mature, what I'll call the mature

00:18:07.060 HDL particle, is basically cholesterol esters that are entirely dependent on LCAT for their formation.

00:18:13.920 Another rare condition are people who lack LCAT genetically. And as you might maybe expect from

00:18:20.760 what I just said, people who lack LCAT, who can't do that process of esterifying the cholesterol,

00:18:26.980 have extremely low levels of HDL, not quite as low as Tangier disease, but very low, and simply can't

00:18:34.900 form the mature HDL particles that we see in most people. So again, we learn very important things

00:18:39.920 about the key role of that enzyme. That ultimately leads to the mature particle that then is really

00:18:46.380 what we're measuring as HDL cholesterol in people is this mature HDL particle where most of the

00:18:52.500 cholesterol we're measuring is the cholesterol ester in the core. Just to make sure we're clear on

00:18:57.060 semantics, when you say that those folks have very low HDL, you mean low HDL cholesterol or a low

00:19:04.560 concentration of ApoA1 or any of the ApoAs? They have low HDL cholesterol. So again,

00:19:11.380 Tangier disease, HDL cholesterol is one, virtually undetectable. LCAT deficiency, HDL cholesterol is

00:19:18.420 10. Also, as we'll talk about, quite low compared to normal. They also have low ApoA1 levels. And part

00:19:25.400 of that is because part of the metabolism of ApoA1, which is where you were going, is that it's affected

00:19:32.200 by the mature HDL and the cholesterol in the core. So when you don't make that cholesterol ester,

00:19:38.120 that ApoA1 is much more rapidly removed from the blood. It's not the normal metabolism. And it

00:19:43.920 basically, the kidney sees that ApoA1, which doesn't have much cholesterol on it, and actually

00:19:50.020 filters it and degrades it. Not surprisingly, the biggest issue with LCAT deficient patients

00:19:55.600 is they get renal disease. There's a serious progressive chronic kidney disease in these

00:20:00.860 patients that leads usually to kidney transplant related to something around the lipid metabolism

00:20:05.960 in the kidney as a result of the lack of LCAT activity and esterification of cholesterol.

00:20:11.960 Before we go deeper on this, I just want to ask if there are any insights we have or how parallel is

00:20:17.500 this system in other mammals, whether it be primates, dogs, mice, et cetera. Because obviously a lot of

00:20:24.920 these animals have a very different ApoB side of the equation. I'm curious as to how similar they are

00:20:30.560 on this front. No, it's a great question. ApoA1 is pretty highly conserved among, certainly among

00:20:35.760 mammals. In many lower level mammals, the primary lipoprotein is HDL. That clearly includes ApoA1.

00:20:43.900 I'll just take mice, obviously a major model system for biomedical research. Mice, I'd say 90%

00:20:52.100 of cholesterol in the blood of mice is in HDL, even more than 90% in a normal mouse. That is the

00:20:58.300 primary lipoprotein in mice. It's true for dogs. It's true for really all mammalian species. I mean,

00:21:04.240 once you start getting below mammals, things change a little bit in terms of the lipoprotein metabolism.

00:21:10.280 Probably we won't get into that today. But there's even HDL-type particles in much lower

00:21:16.160 species, leading to one of the kind of thoughts that HDL is kind of like the primordial lipoprotein

00:21:22.840 and that the ApoB system kind of evolved subsequently, but that HDL probably very early

00:21:29.540 evolved as a transport mechanism for lipids within circulatory systems in lower level species.

00:21:35.240 You know, this to me is one of the most interesting kind of teleologic slash

00:21:40.620 evolutionary questions, which is, was God just pissed off at us? And is that why we as humans

00:21:46.360 have ApoB? Because I can't, I've never really heard a convincing explanation for why humans have

00:21:53.000 ApoB when it seems that every other species gets away without it, right? As you said, I mean,

00:21:59.380 if a mouse or a primate or a dog who clearly have similar energetic requirements can use

00:22:06.540 their ApoAs, i.e. HDLs for lipid transport and energy mobilization, why did we need to evolve

00:22:15.820 with this particle that in the short term really doesn't cause us any trouble, of course, but

00:22:20.100 long enough it doesn't. And so, you know, one answer might simply be that evolution had no

00:22:24.720 concern for the longevity of our species, which is pretty true, but there is some other short-term

00:22:31.740 benefit from ApoB that maybe outweighs the long-term death that it brings our species via ASCVD.

00:22:38.580 I do want to be clear. Mice, dogs, pigs, rabbits, they all make ApoB in both the gut and the liver.

00:22:44.140 So it's not that these species don't have ApoB. The differences in terms of looking at the steady

00:22:49.960 state plasma levels is that their ApoB system is just really markedly revved up. They take care

00:22:56.240 of their business.

00:22:56.980 They just must clear it out of circulation so much more.

00:22:59.360 And it serves a similar role, you know, in terms of transporting triglycerides as a source of energy

00:23:04.920 to tissues in the body that need energy and to adipose tissue for storing it. It's just that

00:23:10.820 the rate at which they do that is much more efficient. Whereas in humans, for reasons we could

00:23:15.800 talk about, but aren't totally clear, but certainly have to do partly with our lifestyle, humans are

00:23:20.700 not nearly as efficient at clearing the ApoB-containing lipoproteins. And they hang

00:23:26.000 around, especially as LDL, as a by-product of this metabolism in a way that then ultimately leads to,

00:23:32.660 of course, atherosclerotic vascular disease. We do need ApoB, honestly. Mammals evolved ApoB because

00:23:38.580 one, we have to be able to efficiently capture fats in the diet for energy purposes, very efficiently.

00:23:46.920 That's what the chylomicrons allow us to do. And during fasting, which of course in evolution

00:23:51.520 sometimes lasted a long time, we have to be very parsimonious about parceling out that available

00:23:57.660 fat we have in our adipose stores back out to heart and muscle to be able to have as a source of 0.84

00:24:03.560 energy. That's what ApoB 100 in the liver does. We need these, especially in evolution,

00:24:08.140 the latter. We need it very substantially. But now, unfortunately, in the modern times,

00:24:13.240 the ApoB system is mostly a problem, not something we need.

00:24:17.280 And I think the way you stated it earlier is obviously more accurate. And perhaps I want to

00:24:21.340 just make sure I'm restating it without being quite as tongue in cheek. It's really not that

00:24:25.420 we don't need ApoB. It's that we could be clearing it at a much greater and more efficient rate than we

00:24:32.400 are. And we could more mimic primates and other mammals and walk around with an ApoB concentration

00:24:39.100 of, call it 20 milligrams per deciliter instead of 100 milligrams per deciliter, a five-fold

00:24:45.260 difference, which is about, that's the difference between getting atherosclerosis in our lifetime and

00:24:51.280 never getting atherosclerosis in our lifetime.

00:24:53.540 I agree with that.

00:24:54.120 And that's the interesting thing because even absent kind of the quote-unquote lifestyle things

00:24:58.880 that will raise ApoB, it just occurs with aging. You know, Peter Libby has written about this very

00:25:04.300 eloquently. And children neonates have ApoB concentrations that are just like those of

00:25:09.360 these other animals. And all things equal, as we age, ApoB just goes up. That concentration goes up.

00:25:16.160 As you say, we're less efficient in the way we clear ApoB from circulation.

00:25:20.460 Anyway, it's a fascinating topic. And it's, I know, not what we're going to talk about today

00:25:24.000 because it's one of speculation. And we could speculate, but there are actually lots of things

00:25:28.220 today that we don't have to speculate about as it comes to ApoA and HDL. I want to go back to kind

00:25:32.540 of the nomenclature a little bit because I know I'm still kind of confused on some of this. And I

00:25:37.560 think it's going to be a little challenging because we're sometimes talking about Roman numerals and

00:25:41.000 we're sometimes talking about numbers. But is there just an ApoA1 and 2 in the Roman numeral

00:25:47.100 lipoprotein, Apo lipoprotein cover? No.

00:25:50.520 So it goes to 2, 3. The two other ApoAs that at least we know about or we talk about are ApoA4

00:25:57.340 and ApoA5. Don't ask me what happened to ApoA3. That's a mystery I've never been able to solve.

00:26:05.100 Okay.

00:26:05.980 But anyway, it's another one of those. ApoA4, we don't really know what it does. ApoA5,

00:26:12.020 I can't remember if you've talked about, but ApoA5 has a very important role in stimulating

00:26:17.360 lipoprotein lipase and in the metabolism of triglyceride-rich lipoproteins. It rides on

00:26:23.620 HDL, but it also rides on triglyceride-rich lipoproteins. And people who lack ApoA5 have

00:26:30.000 really high triglycerides and are also at major risk for atherosclerotic vascular disease. So anyway,

00:26:35.080 again, maybe not the topic for today, but it does illustrate one point, Peter, maybe that I'll make,

00:26:39.900 which is that HDL as a molecule, as a lipoprotein, I think of as a platform. It's a platform

00:26:46.600 for all sorts of proteins and lipids that get transported by HDL in the blood and then transfer

00:26:55.260 off HDL to other things, to ApoB-containing lipoproteins. So my example of ApoA5, it's named

00:27:02.960 ApoAA because it was discovered on HDL, but its primary role in metabolism is triglyceride-rich

00:27:09.440 lipoprotein metabolism. ApoA5 is basically kept within the blood and stored, if you will,

00:27:15.820 on HDL. But then when we eat a fatty meal, it transfers off to the triglyceride-rich lipoproteins

00:27:22.460 and it serves a role of promoting the hydrolysis of the triglycerides. So I think that's a general

00:27:27.560 concept for HDL that maybe we'll keep coming back to is this platform for all sorts of things

00:27:33.780 that are doing different things that involve not only lipid metabolism, but host defense and other

00:27:40.260 things that it evolved to do as a platform for transporting things within the blood.

00:27:46.580 Which I think explains, at least in part, the complexity of this system is it's much more

00:27:52.100 dynamic than what we see on the ApoB side. The ApoB side, as you said, it's a very monogamous

00:27:57.640 relationship, right? I mean, you sort of get your ApoB in the liver and it's with you for life and

00:28:02.240 you're not swapping, you're not trading, you're not increasing or decreasing, you're just marching

00:28:06.780 through life. And the HDL is, maybe the monogamy wasn't the best example because now I don't know

00:28:11.640 what to come up with here because it's not just polygamy, it's even more complicated than that.

00:28:15.160 As you said, it's swapping, it's moving things around, it's carrying things that it doesn't

00:28:20.160 really use, but loaning them out, it is indeed a complicated system. The next thing that seems to

00:28:25.280 add a lot of confusion, at least to me, is some of the nomenclature around the numbering of the

00:28:31.660 HDL particles. And I bring this up in part because various commercial labs, you know,

00:28:37.700 someone listening to this goes out and gets a really fancy, fancy blood test. We use Boston

00:28:42.160 Heart Labs with our patients. I know that this is, you know, something you see with a number of other

00:28:47.140 labs. It might say things like HDL1, HDL2, HDL3. What's the relationship between that nomenclature

00:28:54.440 and the APOA1, APOA2, APOA4, et cetera? This is ridiculously complex. The short answer to your

00:29:02.680 question is actually, believe it or not, there's no relationship. So the numbering of the APO

00:29:06.520 lipoproteins or the APOA lipoproteins, 1, 2, 4, 5, is totally independent of the numbering of the HDL

00:29:14.140 particles, like especially HDL2 and HDL3, which are the classic two main, quote, subclasses of HDL.

00:29:21.100 These are different sizes and different densities of HDL that, again, were isolated by centrifugation

00:29:27.400 and by their flotation properties. HDL2 is bigger. HDL3 is a little smaller. HDL2 has mostly four

00:29:36.080 molecules of APOA1. HDL3 has mostly three molecules of APOA1, but those threes don't

00:29:43.020 have anything to do with each other. And then to make things more complicated, there are other ways

00:29:47.280 of sub-fractionating lipoproteins that I know you've discussed, like the NMR type methodology

00:29:52.520 that don't use HDL2 and HDL3, which frankly is a little outmoded now, but use things like

00:29:58.260 very large HDL, large HDL, medium HDL, small HDL, very small HDL. HDL certainly comes in a whole

00:30:06.060 series of sizes and densities that can be isolated by these different methodologies.

00:30:12.040 There's been a cottage industry, as you know, Peter, of trying to relate these different

00:30:17.160 fractions of HDL to cardiovascular risk in a way that might give us some advantages in

00:30:24.320 terms of trying to predict risk. We can come back to that, but I would say in general, I'm

00:30:29.380 not very compelled that fractionating HDLs gives much clinically valuable information that

00:30:36.620 allows us to predict risk. In contrast, I just want to say, in contrast to the APOB-containing 0.71

00:30:42.240 lipoproteins, where, you know, whether it's APOB or whether, but I think the smaller, denser LDL

00:30:48.400 particles do have a relationship to increased risk, as you've discussed in the past. But I think the

00:30:53.820 HDL fractionation is fascinating for understanding HDL biology and metabolism, but is relatively

00:31:00.400 unimportant from a clinical relevance standpoint.

00:31:02.500 I'm actually glad to hear you say that because I was wondering if I need to be changing our

00:31:07.460 clinical practice. In about 2016, we basically stopped paying attention to any of the HDL

00:31:16.300 fractionation metrics. And really it, you know, had to go out of our way to make sure that the lab

00:31:22.580 was not running those because, you know, in some areas we're running complicated labs and we were

00:31:27.800 sort of saying, look, we don't want any of this stuff. A, we don't want the additional cost of

00:31:32.440 it. But more importantly, we don't think it's clinically actionable. We still get labs from

00:31:38.040 other doctors who might've seen our patients in the past and they're chock full of these things

00:31:42.540 to which I don't know what to do with them.

00:31:45.200 But while we're on this topic of HDL fractionation, unless you want to ask me another

00:31:48.420 question, I wonder if I could use this as a springboard to continuing this complex saga of

00:31:54.880 the metabolism of HDLs once they're formed and once they have cholesterol ester. And the reason

00:32:01.380 that we have this whole sort of set of different size and density HDLs is because of the complex

00:32:08.800 metabolism that occurs on the HDLs once they're in the blood and once they've kind of been made

00:32:15.340 mature HDLs. I want to give you maybe a couple examples of that. One is that HDLs are acted upon by

00:32:23.140 so-called lipases. And you've talked a lot about lipoprotein lipase. It's critically important for

00:32:29.540 triglyceride metabolism and energy metabolism. Lipoprotein lipase has two other very close cousins

00:32:35.480 that I don't think you've talked nearly as much about appropriately because they're more HDL related.

00:32:40.100 One is called, quote, hepatic lipase and what is called, quote, endothelial lipase. We actually first

00:32:46.560 reported a long time ago endothelial lipase as a member of this family. So both of these lipases,

00:32:52.200 they're made in different places as their name suggests, but both of them fundamentally

00:32:56.540 chew on HDL, specifically the phospholipids on HDL and result in modification of the phospholipid

00:33:04.460 composition of the HDL particle in a way that does different things. It certainly changes the sizes of

00:33:10.440 the HDLs and they contribute to this distribution of different sizes. It changes the protein composition

00:33:16.300 of the HDLs in ways that we don't fully understand. And almost certainly has other important

00:33:21.780 biological effects that we don't fully understand. But of course, these evolve for some reason,

00:33:27.240 one of which I'm going to get back to later when we talk about the central nervous system.

00:33:31.360 So that's one. Lipases are really critical for HDL metabolism. There's also this protein that I know

00:33:38.000 you've had some discussion about, cholesterol ester transfer protein. Happy to go into more detail,

00:33:43.240 and I'm sure we will because it's very relevant to HDL. But just for the metabolism part of it,

00:33:47.880 this so-called CETP, cholesterol ester transfer protein, essentially transfers cholesterol esters

00:33:54.960 between ApoB-containing lipoproteins and HDL. And it's a major modifier of the HDL particle in terms

00:34:02.740 of its size and composition. We know this because people who lack CETP have hugely elevated HDL cholesterol

00:34:10.960 levels, big time, like way over a hundred. That really told us that cholesterol ester transfer

00:34:16.960 protein siphons cholesterol out of HDL. And when you don't have it, you have a lot more cholesterol

00:34:23.580 in HDL. And just point back to our previous discussion of mice, mice don't have CETP. So

00:34:30.160 among the differences, one of the things is that mice have a lot more HDL cholesterol relative to ApoB

00:34:35.160 because they lack this CETP protein. I mean, I know we're going to come back to this,

00:34:40.300 but I just, I worry that for the listener, this would be an opportunity missed if we don't dive

00:34:45.780 into CETEP a little bit because of the following. If we take a step back, a listener is going to be

00:34:52.580 saying, Peter, Dan, what the hell are you guys talking about? Like, I am so lost. The only thing I

00:34:58.000 know is when I go to the doctor, the doctor says my good cholesterol is high. I'm in good shape.

00:35:03.580 So let's do this. Let's hit pause for one sec and acknowledge that a standard lipid panel

00:35:10.380 spits out a bunch of numbers, total cholesterol, LDL cholesterol, HDL cholesterol. And if you're

00:35:17.000 lucky, VLDL cholesterol, non-HDL cholesterol, and triglycerides. That's basically the standard metric.

00:35:23.340 If the lab is competent and they're using direct measurements, your HDL cholesterol, LDL cholesterol,

00:35:30.440 and VLDL cholesterol should sum to your total cholesterol. And of course, your non-HDL

00:35:37.040 cholesterol should be the same number as your total cholesterol, less your HDL cholesterol.

00:35:42.880 I want to pause and also insert that with nomenclature, HDL is not a laboratory metric.

00:35:50.240 It is a lipoprotein. The laboratory metric is HDL cholesterol, HDLC, or if using NMR,

00:35:56.720 HDLP, HDL particle number. Or APOA1, an analogy to measuring APOB.

00:36:02.240 APOB.

00:36:02.620 Exactly.

00:36:03.400 The units that you've been throwing around, of course, are the HDL cholesterol. So a moment ago,

00:36:07.100 you said, hey, people who are deficient in CTEP could have HDLs over 100. Of course,

00:36:12.120 that means HDL cholesterol over 100 milligrams per deciliter. Now, there is an observation that goes

00:36:18.360 back probably to the late 70s, right? I mean, it probably goes back to late 70s, early 80s,

00:36:24.040 which is in some of the earliest Framingham cohorts, which observed the risk of ASCVD in five

00:36:33.000 cities. We don't need to go into what Framingham did in the first cohort. But what came out of that

00:36:37.700 was higher HDL cholesterol was better than lower HDL cholesterol. In fact, that was four times greater,

00:36:47.080 if my memory serves correctly, as a predictor of ASCVD than high LDL cholesterol was a negative

00:36:54.700 predictor. Am I remembering that correctly?

00:36:56.500 Yeah, that's about right. Yep.

00:36:58.120 So by now, we're talking about 82, 83, 84. This has been a long time since I've looked at this

00:37:04.440 stuff. But it emerges that, hey, high HDL cholesterol seems to be positively associated

00:37:10.920 with outcomes. And obviously, that's a big part of why HDL cholesterol became known as good cholesterol

00:37:15.760 cholesterol. And LDL cholesterol became known as bad cholesterol. We'll obviously talk about why

00:37:19.940 those terms are inaccurate. But it wasn't long until companies, drug companies, were saying,

00:37:27.440 hey, wait a minute. We know that if CTEP is inhibited, this enzyme is inhibited, HDL cholesterol

00:37:34.880 goes up. That must be a good thing, right? That led directly from this observation I mentioned,

00:37:40.420 that people who lack CTP genetically have these hugely elevated levels of HDL. One directly from

00:37:48.220 the other. Well, gosh, that must mean that if we could pharmacologically inhibit CTP, it would be a

00:37:55.160 way to raise HDL. And of course, as you know, that absolutely turned out to be the case. But the story

00:38:00.580 goes beyond that.

00:38:02.040 So the first company to do this was Pfizer, right? Wasn't the first CTEP inhibitor?

00:38:07.000 The first CTB inhibitor, torsatripib, was from Pfizer, correct?

00:38:10.720 And that was, you could take a skeptical view of the trial, which was Lipitor was about to go off

00:38:16.320 patent. And so they came up with a trial that was Lipitor by itself versus Lipitor combined with the

00:38:24.860 CTEP inhibitor. And I remember actually, this was probably early 2000s, right? Thinking, oh, this is

00:38:31.080 super, I mean, this was, you know, back when I was in my surgical residency. So I was only tangentially

00:38:35.520 interested in this, right? It wasn't my field. But interesting in that I knew my family history

00:38:39.160 for ASCVD was high, so I was paying attention. And I remember thinking, well, God, this has got to be

00:38:43.640 great, right? You got one drug that's going to lower LDL-C, one that's going to raise HDL-C. This

00:38:48.420 is a can't miss drug. And then in September of 2006, lo and behold, not only did it not get better,

00:38:55.340 it was slightly worse, right?

00:38:56.300 Yeah. Bombshell. So first I want to emphasize, the genetics did predict what happened. That is,

00:39:02.300 pharmacologic inhibition of CTP is extraordinarily effective at raising HDL a lot. In fact, putting

00:39:07.860 people kind of over 100 HDL cholesterol, like the folks who lack CTP. But you're absolutely right.

00:39:13.660 There's a lot of excitement about that. A lot of excitement about HDL as the good cholesterol.

00:39:18.160 And that trial, the first of several with different CTP inhibitors, not only didn't show benefit,

00:39:24.860 but showed an actual adverse effect. Now, it has to be said that in retrospect,

00:39:31.040 we're pretty sure that adverse effect, more people, literally more people died. That adverse effect was

00:39:36.900 due to off-target effects of the drug, not due to CTP inhibition itself. So that drug, of course,

00:39:44.100 was discarded. But it didn't kill the field because the idea was, well, that was just a bad drug.

00:39:49.880 Let's get a clean CTP inhibitor and see what that really does. I don't know if you want to ask me

00:39:54.220 questions or I can continue the story. Yeah, let's go down that path because it's still a little,

00:39:59.320 it got progressively less murky as time went on. But let's go down the path of the evolution of

00:40:06.440 CTP inhibitors, which let's be honest, it's for what, we're almost 15, well, actually 16 years post

00:40:13.560 the halting of that trial and the discarding of that drug. And we still don't have a CTP inhibitor on

00:40:20.220 the market, right? We don't. There is still a CTP inhibitor that's in development, which I'll get

00:40:25.240 to, but there is certainly not one on the market. As you know, Peter, and certainly probably many of

00:40:30.140 our listeners do, three additional CTP inhibitors were then taken into late stage clinical development,

00:40:37.780 including large cardiovascular outcome trials. To summarize the results of that, one of them was

00:40:43.720 just flat, didn't do anything, didn't hurt people, but didn't help. A second one was stopped early

00:40:49.320 because it really didn't look like it was doing anything, helping. And the third one was followed

00:40:55.300 through and did show about a disappointing 9% reduction in cardiovascular events. But I also

00:41:02.980 want to point out that it lowered LDL and ApoB pretty well also. 9%, given that you also lowered LDL and

00:41:10.180 ApoB, was not exactly exciting. And that drug was not taken further in terms of approval. I have to say

00:41:17.060 one other thing that's just fascinating for sort of our field and maybe in general. One of these drugs,

00:41:22.900 dalsetrapib, that failed in its clinical trial, a detailed genetic study was done post hoc. And there

00:41:29.740 were individuals who were found to have a particular genetic variant that looked like on post hoc analysis,

00:41:37.420 that group actually benefited from the drug. And that led to a subsequent attempt to do another trial

00:41:45.200 focused specifically on people of that genotype with the CTP inhibitor, which was just reported

00:41:51.860 out really within the last year as a negative trial. That CTP inhibitor actually had two different

00:41:58.200 trials, one in kind of a very focused precision medicine way, but it still didn't produce any benefit.

00:42:04.480 This has been a long saga of using CTP inhibition to raise HDL as a way to reduce risk. And as we're

00:42:13.760 going to be talking about more, it really is one of the key planks that has led to, I think, what is now

00:42:19.860 rock solid, which is HDL cholesterol itself. The HDL cholesterol itself is not directly and causally

00:42:28.120 protective against atherosclerotic cardiovascular disease. There's a lot of nuances there, which we'll

00:42:33.080 come back to, but I feel pretty confident making that statement. And I just think this is a great

00:42:37.700 opportunity to also talk about why it's so important to have hard outcome trials in the field of

00:42:45.740 cardiovascular medicine. I think it's true in all medicine, but let's go back and think about the

00:42:50.700 first lipid lowering drug introduced in the United States in, God, about 1950, right? Maybe 1959,

00:42:57.260 something like that, right? So trypanorol, which I can never remember the name of the enzyme,

00:43:03.020 but it's the enzyme that converts desmosterol to cholesterol in the cholesterol synthetic pathway.

00:43:08.880 Believe it or not, this is a sign of my aging, Dan. I used to know the names of these enzymes. It

00:43:13.960 was like, and it's not Delta 24 desaturase, but it's a cousin or derivative, isn't it, right? But

00:43:18.940 anyway, so you had this drug, triparanol, that inhibited that enzyme and lo and behold, it lowered

00:43:24.640 cholesterol. Now this was for the listener. This was back in the day before we knew anything about

00:43:28.520 the sub-fractions of cholesterol. So we just knew from some of the early work of Ancel Keys that if

00:43:34.200 you looked at very, very high levels of total cholesterol and compared that to people who had

00:43:39.020 very, very low levels of total cholesterol, there was a difference in outcome, cardiovascular outcomes.

00:43:44.740 This was an observed finding. There was no intervention to test that. So this drug came along

00:43:49.060 and it really lowered total cholesterol and it likely would have been lowering LDL cholesterol and

00:43:53.900 NAPO-B. It basically got approved on the basis of that without, of course, the hard outcome. So it

00:43:59.620 gets approved, it goes into circulation, and it's really only after it was approved, gosh, probably

00:44:05.460 what, eight to 10 years later, that they had enough post-surveillance data to say, you know what,

00:44:12.020 this drug is lowering cholesterol all right, but it's actually increasing mortality, cardiovascular

00:44:17.980 mortality. The drug was pulled off the market. I don't know what is known about the why,

00:44:23.520 but I know Tom Dayspring and I have speculated that it was probably the desmostrol spike that

00:44:29.440 was causing the problem. So you might've been trading one problem for a worse problem. Are

00:44:33.980 you familiar with that drug or that story? Only peripherally, but of course there's been

00:44:38.300 a ton of research in desmostrol really over the last several years. We now, based on that,

00:44:43.700 the plausibility of the fact that the adverse effects were related to increasing desmostrol I think

00:44:48.800 is quite real. That very well may be the explanation. By the way, this is a totally

00:44:53.640 tangent off-topic point, but it's become quite in vogue to use clomiphene or clomid for testosterone

00:45:02.140 replacement in men. And the reason for it is, A, it's quite effective. So if you give clomiphene,

00:45:09.460 you are telling the pituitary to make a lot of LH and FSH, which of course is telling the testes to

00:45:16.380 make a lot of testosterone, this would certainly in the short term have the advantage of preserving

00:45:22.320 testicular function, unlike giving exogenous testosterone, which suppresses testicular

00:45:27.180 function. And I think for short-term use, it probably isn't a bad thing. It's an off-label

00:45:32.760 use of course for clomiphene. And we used to do it for this purpose, right? So if a guy was still

00:45:38.680 considering reproduction or we were considering this a ridge treatment between testosterone,

00:45:44.540 we would use it. But because we always measure desmostrol, lithosterol, compesterol, cytosterol,

00:45:50.260 these sterols, when we measure our patients' cholesterol levels, we noticed how high the

00:45:55.300 desmostrol levels were getting in those patients. And we figured out it was pretty quickly, it was the

00:46:01.260 clomid that was doing this. And it would reverse, but it could take a year to return desmostrol levels

00:46:07.120 to normal after you stopped the clomid. So about four years ago, once we put two and two together,

00:46:12.500 we stopped using clomid. But it's interesting to me that the use of that hormone has gone

00:46:18.220 through the roof. There are now like clomid clinics opening up. And I don't understand why

00:46:24.500 someone, and we've done a lit search, I don't understand, and we just need to write this up,

00:46:27.840 I think at some point. Yeah, it's fascinating. Because I don't understand why someone hasn't

00:46:32.040 put the two and two together, especially long-term use. Again, I don't think using clomid for

00:46:36.160 a couple of years is going to be problematic. And I certainly don't think it's problematic for women 0.99

00:46:40.600 using it for IVF. When I think about a guy being on this for 10 years, and I'm talking desmostrol

00:46:46.320 levels going up by 20-fold, that would be concerning. And we went back and tried to see

00:46:52.500 if we could find out how high the desmostrol levels were in the triparinol trial, not trials,

00:47:00.680 but in the levels that if anybody had pulled serum or saved serum from those subjects, we couldn't find

00:47:05.320 it. So we don't really know how high it was, but the assumption has to be that that drug

00:47:10.200 and clomiphene both interfere with the same enzyme. Really fascinating. I would encourage you to write

00:47:15.140 that up. So what's the point of that whole long-winded story? The point of that long-winded story is

00:47:19.480 things can make a lot of sense until they don't, right? That's for sure. Before we leave CTP

00:47:25.600 inhibition, though, I do want to just remind your listeners that there is still one CTP inhibitor,

00:47:30.440 ovacetrapib, that is still in clinical development. What differentiates it, it's much more effective

00:47:36.140 at lowering LDL and ApoB. We now no longer think that raising HDL cholesterol with CTP inhibition

00:47:43.020 is going to help you. We don't think it hurts you, but we don't think it's going to help you.

00:47:48.120 But we do think that there's still merit. Of course, we know there's merit to raising LDL cholesterol

00:47:52.460 and ApoB. You mean lowering? I'm sorry, rate lowering. Perhaps this CTP inhibitor could be

00:47:57.600 part of the armamentarium to do that. So we'll see. Stay tuned.

00:48:01.240 Now, Dan, maybe this was naive of me, but after the third failure of a CTP inhibitor,

00:48:07.600 my very crude interpretation is not unique. I'm sure a lot of people have speculated this,

00:48:13.840 is that if at least part of the benefit of HDL involves stuff we're going to talk about, right?

00:48:22.980 Delipidation, reverse cholesterol, transport, all of those things, and you slap a CTP inhibitor on one

00:48:29.180 of those particles, thereby making it harder for the particle to efflux its cholesterol, to get rid

00:48:35.900 of its cholesterol, which is why you now measure much higher HDL cholesterol, which on the surface

00:48:39.960 looks good, that could actually be problematic. In other words, if you see a lot of people in a room,

00:48:45.900 it might be tempting to conclude that there's something awesome going on in that room.

00:48:50.300 But what you don't know, if you can't measure all the ins and outs, is it might be that those

00:48:55.240 people are all stuck in the room because the door is locked. I know that's a bit of a crude analogy,

00:48:59.480 but is there any merit to that sort of thinking around the different ways in which one might

00:49:05.120 boost HDL cholesterol and how some of those could actually be deleterious if they prevent function?

00:49:11.020 There absolutely is, but I think it's probably not relevant to CTP for reasons we can return to.

00:49:15.660 But it is relevant to another protein that I think now's a good time to introduce,

00:49:19.940 which is another key protein, a receptor that is basically the major HDL receptor,

00:49:25.300 essentially the equivalent of the LDL receptor, but for HDL. And it's known as SRB1,

00:49:30.740 another unfortunate nomenclature. But SRB1 is a receptor that basically is on a lot of cells,

00:49:38.940 but the liver is the most important with regard to HDL. And it essentially binds the HDL particle,

00:49:44.220 basically sucks the cholesterol ester out of the HDL particle, and then releases the

00:49:49.540 cholesterol depleted HDL back into the circulation to go back around and do whatever it's doing.

00:49:55.680 So the analogy I like to use, Peter, is that of garbage trucks. So a very simplistic view of HDL,

00:50:01.620 which we're going to return to because it's much more nuanced, is that HDL, to a certain extent,

00:50:06.180 functions like garbage trucks that are picking up things, trash in places where you don't want it,

00:50:12.260 and returning it to the liver, dumping it off via SRB1, and then going back now empty to do more of

00:50:21.540 its role. Again, it's more complicated than that, but that's an analogy. So this brings me back to

00:50:27.120 SRB1. So there are humans who lack SRB1. What do you think the problem is? They have very high HDLs.

00:50:35.100 They have very high HDLs because they can't unload the dump trucks, but they actually have increased

00:50:41.800 risk of heart disease because they are not efficiently unloading that HDL. So very similar

00:50:48.880 to your analogy of the locked room, you're not clearing the HDL and doing that normal process

00:50:54.380 of recycling the trucks back to the periphery to do what they need to do. And so that's, I think,

00:50:59.560 our best example of this concept of constipation of the system leading to high HDL, but paradoxically

00:51:06.900 to increased risk. And of course, the analogy would be that, not the analogy, but the interpretation of

00:51:12.700 that would be, you would never want to develop an SRB1 inhibitor. Yes, it would raise HDL quite a lot,

00:51:18.880 but it wouldn't protect against heart disease and probably would hurt people.

00:51:23.200 Yeah. I remember the first time Tom shared a case study with me of a patient with presumably

00:51:30.320 defective, not necessarily completely absent, but SRB1. And this was one of those things where

00:51:37.260 you could tell just reading the journal, it must've been 40 years old, but her LDL cholesterol was

00:51:41.520 whatever it was, 100 milligrams per deciliter. Her HDL cholesterol was like 150 milligrams per

00:51:45.780 deciliter. So she had quote unquote high cholesterol, but initially people assumed it was not of concern

00:51:52.600 because the fraction that was HDL cholesterol was what was contributing to it. And of course,

00:51:57.280 on further exam, I mean, she had very, very advanced atherosclerosis for a woman her age.

00:52:01.340 Yep, exactly.

00:52:02.600 We don't see this often. I can tell you that I have not yet seen a case of that. Now, admittedly,

00:52:07.540 my practice is not large, but we're certainly looking for it when we see people with high HDL

00:52:15.200 cholesterol, typically, you know, north of 90. I guess we should talk about this as well,

00:52:19.840 but we can do that in a moment after this question, which is why is there a sex difference

00:52:23.800 between these two? Because there is quite a significant sex difference between men and

00:52:27.940 women in terms of HDL cholesterol. But at what point, Dan, as a clinician, do you start to worry

00:52:33.080 about that? And if there are doctors or patients listening to this, when would you recommend that

00:52:37.120 somebody go and get checked out for a genetic deficiency there?

00:52:41.220 SRB1 deficiency is not common. I completely agree with you. A huge project we've had for a long time

00:52:46.620 now is essentially collecting people or consenting people with extreme high HDL with the goal of

00:52:51.820 trying to find what the underlying genetic cause is. And I will tell you that that's how we found

00:52:57.440 a few of these SRB1 folks, but most of them don't have any identifiable gene that we can point to and

00:53:03.640 say, here's the smoking gun. Here's why your HDL is high. HDL is very heritable, meaning there's a lot

00:53:08.800 of genetic determinants of HDL, but it's more about so-called polygenic inheritance where

00:53:13.400 multiple different genes, including SRB1, but not major genetic defects, just sort of more common

00:53:19.160 variants, but many others, like some of the things we've been talking about, including SCTP and ABCA1,

00:53:24.340 all contribute to the heritability of the HDL cholesterol. But to get back to the clinical

00:53:29.660 implications, my view at this point, we can talk more about the data. It's pretty clear that

00:53:36.220 high HDL is not uniformly associated with protection. And there are certain circumstances

00:53:43.620 where that's true, especially if it's extremely high. Perhaps we'll get back to this, but there's

00:53:48.060 a fair amount of evidence now, including a recent paper that individuals of African ancestry who tend

00:53:53.380 to have higher HDLs anyway, that the high HDLs are not as protective or maybe not even HDL is that well

00:53:58.960 associated with risk. So my clinical advice to anyone with high HDL is not go get yourself

00:54:05.520 sequenced to find out if you're deficient in SRB1. The odds of that are extremely low. But never use

00:54:12.320 a high HDL as a reason for not using a statin or some other LDL lowering or preventive therapy. That

00:54:20.960 is, you should never be dissuaded from doing what you would have otherwise done in that patient just

00:54:25.780 because their HDL is high. Does that make sense? It absolutely makes sense. And I have a very short

00:54:30.380 paragraph in my book that'll be coming out in a while where I cite two Mendelian randomizations

00:54:36.820 that really look at this in some detail. I think I've explained Mendelian randomization before on the

00:54:43.420 podcast, but it's always worth, I think, a short explanation. Again, in case you haven't heard it,

00:54:47.700 this is basically a technique where you look for genes that impact traits that you are interested

00:54:56.800 in. So in this case, because as you pointed out, HDL is very genetic. That means that there are sets

00:55:03.120 of genes that we would identify that predispose people to having very high versus very low HDL

00:55:08.080 cholesterol. And because those genes are randomly occurring, you can look at that as though it's a

00:55:15.660 natural experiment. And you can look at, based on the natural occurring spreading or scattering of

00:55:22.420 those genes, how are outcomes affected? And it turns out that low HDL cholesterol is not causally

00:55:32.240 linked to atherosclerosis and high HDL cholesterol, genetically high, is not causally linked to protection

00:55:40.780 from ASCVD. So I think those are very important findings. And I think it speaks to why, as you say,

00:55:47.340 you can't use high HDL cholesterol as a reason to not treat in the presence of other risk factors.

00:55:53.300 I completely agree.

00:55:54.640 It's interesting though, that that persists, isn't it? I do find this to be probably top three most

00:56:01.940 vexing discussions I have with other physicians, which is, I know his LDL cholesterol is 140 milligrams

00:56:08.940 per deciliter, but God, I mean, his HDL is 80. I mean, you know, his ratio, I mean, that's when they

00:56:14.440 say the ratio of total cholesterol to HDL or LDL to HDL is such and such, and therefore I don't need

00:56:20.460 to treat. I mean, I, it makes me wish I had hair to pull out.

00:56:24.940 You're absolutely right. And I'll just reiterate, high HDL is never a reason not to treat someone who

00:56:28.980 would have otherwise merited treatment. I do want to make clear though, there are lots of people

00:56:33.380 were on the fence about whether to start a statin who have LDLs that are kind of borderline or even

00:56:38.300 not that terribly high, but, and of course we look at the whole patient, we look at the risk,

00:56:42.920 we look at their calculated risk, other risk factors, but I'll just say that a low HDL in the

00:56:49.740 setting of someone who you're generally on the fence about treating perhaps could contribute to

00:56:55.400 tilting toward, yes, I'm going to treat this person. In other words, a low HDL, at least in most

00:57:01.000 populations. Again, I think individuals of African ancestry, perhaps we have to be a little bit more 0.90

00:57:05.600 careful about using HDL as a predictor. Well, maybe we'll come back to that. Otherwise, I think a low

00:57:10.960 HDL can be used not absolutely, but relatively to tilt toward being more aggressive, but only in the

00:57:18.340 context of the overall risk profile. And obviously part of that, Dan, has to do with the relationship

00:57:24.580 between low HDL cholesterol, if my memory serves me correctly, it's the HDL2 fraction, and the

00:57:31.280 association with that and insulin resistance. So there's no question that a phenotype, and as you

00:57:37.380 point out, and we should come back to this, we don't even do this calculation for African-American

00:57:41.640 patients because we've long observed it's not helpful, but in non-African-American patients,

00:57:47.300 the ratio of triglyceride to HDL cholesterol, when both are in milligrams per deciliter,

00:57:52.680 you know, is reasonably associated with insulin resistance. And the higher that ratio, the more

00:57:57.300 insulin resistant they are. And obviously that ratio is driven up by an increase in triglycerides

00:58:02.040 and a reduction in HDL cholesterol. Why is it that, I mean, we've spent hours on this podcast talking

00:58:07.540 about why insulin resistance would lead to or be associated with high triglycerides. We haven't

00:58:12.880 done the opposite or the reverse of that. What is it about IR that drives down HDL2?

00:58:18.660 I'll just say first though, that one of the other analogies I like to make is that HDL cholesterol,

00:58:25.400 while not causally related to disease, is sort of like an HbA1c for cardiovascular risk factors.

00:58:34.640 It's an integrator of information related to insulin resistance, related to triglycerides,

00:58:40.680 related to inflammation, that in one number, in most people, when it's low, it's telling you

00:58:46.620 something about cardiovascular risk, even though itself, it isn't directly impacting on risk.

00:58:51.840 So in answer to your question, I think one of the big issues with HDL cholesterol is that it's an

00:58:57.100 inverse barometer of triglyceride efficiency of triglyceride metabolism. And again, we talked

00:59:03.820 about efficiency earlier. You can only learn so much from measuring a fasting triglyceride after a 12

00:59:09.800 hour overnight fast. It's useful. It's the way we do it in terms of lipid panels, but a lot happens

00:59:16.420 after a fatty meal. And a lot of that action is basically over in most people by 12 hours.

00:59:23.500 Some people are extraordinarily effective at clearing their dietary fat and even their liver-derived fat,

00:59:29.680 like mice. Others are not so effective, but their fasting triglycerides may not necessarily

00:59:34.780 even reflect that. The HDL cholesterol does reflect that. There's a complex, as we discussed earlier,

00:59:41.320 complex frequent interaction and exchange between triglyceride-rich lipoproteins and HDL,

00:59:47.200 with the net effect being the higher the triglycerides at any given time, the lower the HDL is as a result

00:59:54.400 of the complex interactions. So if you can picture, we do a lot of these experiments where we bring in

00:59:59.700 people and we give them a high-fat milkshake. We draw blood multiple time points after that milkshake,

01:00:04.640 and we measure triglycerides and all sorts of other things. People differ a lot in their response to

01:00:09.500 that high-fat milkshake challenge. The higher that triglyceride goes, the area under the curve,

01:00:16.020 the lower the HDLs are. That relationship is extraordinarily strong. So just like HbA1c,

01:00:22.400 that HDL cholesterol is sort of reading out the 24-hour triglyceride metabolism much better than the

01:00:29.980 overnight fasting triglyceride measurement is. Sort of why HbA1c is better than fasting glucose.

01:00:36.640 You know, that's the analogy. First of all, I've never heard that before, and that might be, if I learn

01:00:41.440 nothing else on this podcast, Dan, that is hands down the most amazing thing I have learned, not just

01:00:46.460 today. But I'm going to go out on a limb and say this week, potentially this month. That is super fascinating.

01:00:51.840 First of all, people on this podcast have probably heard me gripe over the lack of integral functions

01:00:58.140 in biology. You know, my background, of course, in math and engineering, we love integrals, right?

01:01:03.080 As imperfect as the A1c is, as you said, it is an integrator, and it's so hard to find integrators.

01:01:11.260 You know, ferritin sort of does it a little bit for iron, but not nearly as well. And, you know,

01:01:16.140 the holy grail, of course, would be to find an integral of something like mTOR activity or

01:01:20.180 something like that. But I've never before been presented with or confronted with this. So I want

01:01:25.660 to make sure we all understand this a little bit better. So there's a couple of things that you've

01:01:29.720 said there. One is, if you stick an IV in a person's arm, let's just do that to go easy on

01:01:34.440 them, right? So we're going to just be able to draw blood continuously or call it every five minutes

01:01:38.580 without poking them. You bring them into a lab fasted. You measure their HDL cholesterol. You measure

01:01:44.040 their trig. So let's just say they show up in the morning fasted. Their trigs are 100 milligrams per

01:01:48.340 deciliter. HDL is 50 milligrams per deciliter. So most people would look at that and say,

01:01:52.160 oh, that's great. Person looks super healthy. You give them a high fat shake and 30 minutes in,

01:01:58.400 you just start measuring every couple of minutes the concentration of those two things.

01:02:02.980 I think most people somewhat familiar with metabolism would not be surprised to learn that

01:02:08.620 the triglycerides will go through the roof. As an aside, anybody who has accidentally done a blood

01:02:13.340 draw with their doctor after eating a meal when it was supposed to be fasting will be

01:02:18.200 familiar with this, right? We see this from time to time when patients make mistakes. They have a big

01:02:22.960 fat breakfast before the blood draw and you get their trigs back and they're 400 or 500 milligrams

01:02:26.780 per deciliter. But highly variable from person to person, just to be clear.

01:02:30.780 And highly meal dependent as well. Exactly.

01:02:33.120 Very much depends on what the meal is. So in this experiment though, you're specifically giving

01:02:37.880 them something to elicit the biggest triglyceride response. In the example I gave, Dan, let's say

01:02:43.060 triglycerides go 100, 125, 150, all the way up to 400 before they start to come down. That generates

01:02:48.560 an area. You could integrate that on the curve. Can you give realistic ideas or values for what

01:02:54.740 the HDL cholesterol would do during that period of time? The HDL cholesterol definitely dips during

01:03:01.020 that time in a way that's roughly proportional to the triglyceride levels, but not anywhere near the

01:03:07.380 same degree of magnitude. Right. Simply because it's starting lower and you have a constraint bottom

01:03:12.100 versus no constraint top. Starting lower, but also the dynamics of its turnover are very different.

01:03:18.280 I think the key point here is HDL is not just an integrator sort of acutely over that meal, but

01:03:25.320 chronically. In your example, the HDL cholesterol might go from 40 to 38 or 37, which doesn't sound like a

01:03:34.100 lot after that one meal. But the repeated meals that that individual is eating that are high fat

01:03:40.320 and that repeated triglyceride excursion has a more chronic effect, a little like glucose and HbA1c

01:03:47.100 that keeps taking the HDL down over time until you get to kind of a new steady state. So the acute effect

01:03:54.020 on the HDL is real. It's modest. The chronic effect on the HDL of that abnormal postprandial

01:04:00.920 triglyceride metabolism is quite substantial. And that's why HDL is a good, as you say, integrator of

01:04:07.060 this effect. And actually what you just said, Dan, is kind of what I was hoping to go to next, which is

01:04:13.400 if everything I said were true, that might not be enough to explain the full integral function because

01:04:20.380 it probably captures some of what's happening outside of the meal, just as hemoglobin A1c doesn't

01:04:27.040 only reflect what would be captured in an oral glucose tolerance test. It captures what's

01:04:31.860 happening over 90 straight days when you're eating when you're not eating. Exactly. And that's also why

01:04:37.440 there's a very strong statistical relationship between fasting triglycerides and the HDL. So the

01:04:44.480 fasting triglycerides themselves are still affecting HDL metabolism. It's just that the postprandial part is

01:04:50.320 also a key component. So let's then, again, just one more time, sort of reiterate the lagging nature

01:04:58.320 of HDL cholesterol through HDL biology to what's happening with the efficiency of, maybe for lack

01:05:07.300 of a better word, lipid partitioning. The lagging nature, meaning the integrating over time?

01:05:13.960 Yeah, exactly. Let's use an example. So you've got a person who is exercising. And so you have two

01:05:21.300 people who are similar, except that one is very insulin resistant and one is quite insulin sensitive.

01:05:27.480 The insulin resistant person is going to have a higher level of insulin. All things equal,

01:05:31.960 they're going to have a more difficult time oxidizing free fatty acid. So as energy demand

01:05:37.500 goes up from the muscle, they're going to be more likely to utilize glycogen as opposed to utilizing 0.75

01:05:42.380 triglyceride. So at any point in time, you might measure differences in glucose, differences in

01:05:47.740 lactate. You might, well, you would see lactate as well, but differences in triglyceride level.

01:05:51.440 You might also notice more fat in the liver or some fat in the liver of the insulin resistant person,

01:05:56.740 none in the insulin sensitive person. Into that situation, even when they're not eating,

01:06:01.560 they're quite different physiologically. What is the HDL discerning or doing in those two scenarios

01:06:07.760 that's being reflected in the ongoing integral function of it?

01:06:12.380 The simplest, most direct model is that the HDL is reflecting the 24-hour excursions in

01:06:17.960 triglycerides, like we were just talking about. I think what you may be getting at is there are

01:06:24.380 almost certainly other components of metabolism that the HDL is integrating and reflecting,

01:06:31.220 particularly in the complex insulin resistance world. We know insulin resistance has a lot of effects

01:06:36.580 other than affecting triglycerides. And insulin resistance, I'm sure, has effects that are

01:06:44.200 affecting and are reflected by a lowering of HDL. What I can't tell you is exactly what those are.

01:06:51.460 That's an area that is still a topic of investigation. I will give you one hypothesis

01:06:56.460 though, and one that maybe I think has some merit. As you know, adiponectin is an adipokine secreted by

01:07:03.020 fat that has an inverse relationship to insulin resistance. So basically, people who are insulin

01:07:08.640 resistant have lower levels and people who are more insulin sensitive. And adiponectin itself

01:07:14.240 appears to have some direct effect on HDL metabolism in the right direction. We could talk about it

01:07:20.480 possibly by tweaking the liver in ways that then the liver affects HDL. So where I'm really going with

01:07:27.480 this, although I have to say the data still are not completely solid, is that another way that

01:07:33.200 insulin resistance is impacting on HDL is through adiponectin secretion affecting HDL metabolism in a way

01:07:42.540 that's completely different than the triglyceride hypothesis that I just put forth. You follow me?

01:07:47.380 Yeah. And it's funny. We used to measure adiponectin and leptin levels. Again, one of those things I sort of

01:07:52.780 stopped measuring, do you think that that's a helpful biomarker independently? And should that be

01:07:57.240 something that's back on our plate? Absolutely fascinating to try to put together these metabolic

01:08:01.740 pathways. In terms of its utility as a clinical marker, I guess I'd have to be a little bit

01:08:07.680 skeptical. I'm not quite sure how I would use it in terms of guiding clinical care.

01:08:13.120 I mean, I think that was sort of where we ended up, which was, look, we get more actionable insight

01:08:17.420 out of other metrics. Okay. That was really interesting. And again, before we leave the

01:08:24.400 pharmacoside of this, can we talk for a moment about niacin? Oh, sure. It's been a while since

01:08:28.920 niacin came up on a podcast, but this is an interesting drug because it clearly raises HDL

01:08:34.220 cholesterol and it lowers ApoB, doesn't it? It does. It lowers triglycerides. It lowers ApoB.

01:08:39.820 It lowers LDL, all modestly, but really. Peter, in niacin, I just chuckled because niacin is one of

01:08:45.840 these areas where we as lipidologists and me personally have to really eat a bit of crow.

01:08:50.620 There was a time when I prescribed niacin to a lot of patients with primarily the idea that it was the

01:08:58.400 only thing we had to raise HDL. This was in an era where HDL is the good cholesterol and raising it

01:09:04.260 must be good, right? I also told myself, well, niacin does lower triglycerides. It lowers ApoB.

01:09:09.720 It lowers LDL. It also, as I'm sure you know, modestly lowers LP little a as well.

01:09:15.220 About 15% lowering of it?

01:09:17.160 Yeah, 15, 20%. I basically was like, this is a nice broad spectrum lipid lowering drug that,

01:09:22.560 you know, not in place of, but on top of a statin for certain people who had certain lipid profiles,

01:09:27.620 mostly high triglycerides, low HDL, has to be providing some benefit, even in the absence of,

01:09:32.760 as you pointed out, the clinical trials that really are a cornerstone of cardiovascular medicine.

01:09:38.420 Well, the clinical trials, you know, got done. The AIM-HI trial, which was a very well-done trial,

01:09:44.340 which frankly just didn't show much benefit of niacin, a blow to us, but the field comes up

01:09:50.140 with reasons why that trial may not have been right. Bottom line is, one more trial run by

01:09:54.840 Merck with a drug that also helped to potentially address some of the issues with niacin was done,

01:10:01.120 a very large trial, very well-powered trial, and that also had really minimal or very disappointing

01:10:07.520 effects on reducing cardiovascular events. Essentially, those two trials killed niacin.

01:10:13.380 So, over the next, you know, year as patients would come back, I would have the discussions that

01:10:18.000 maybe you've had discussions with patients too like this. You know that niacin I put you on,

01:10:22.440 you know, eight years ago that you've been taking religiously despite the question that it causes

01:10:27.120 when you take it? I really think I have to tell you that I'm not sure you need to take it anymore.

01:10:31.380 And it was a humbling experience to basically have a drug that I had prescribed quite a lot,

01:10:37.640 basically to tell patients that I don't think that in retrospect, this is helping you much.

01:10:43.220 I do have a subset of patients who are so wedded to their niacin that despite that,

01:10:47.840 they haven't wanted to stop. But the vast majority of my patients have stopped their niacin.

01:10:52.280 And what's the mechanism by which niacin raised HDL cholesterol?

01:10:56.060 I think one is certainly triglyceride lowering, like we were talking about earlier. But probably

01:11:03.020 not just that because the increases in HDL were somewhat disproportionate to what you'd predict

01:11:08.320 from the triglyceride lowering. Although, again, with the complexity, it's a little hard to make

01:11:12.120 that calculation. So, there probably is some other mechanism. And to this day, I don't think we

01:11:18.080 really understand it. We tried for a while to try to figure that out. Niacin is a very complex

01:11:23.260 drug. Of course, we're talking about niacin here in pharmacologic dosing, not in the kind of vitamin

01:11:29.380 dosing. We don't really know, I think it's fair to say, how niacin raises HDL beyond its triglyceride

01:11:35.960 lowering effect. It's not a drug we used much, if at all. But the few times we did, I was actually

01:11:43.620 really amazed at how much it raised HDL cholesterol. I mean, it wasn't uncommon to go from 50 milligrams

01:11:49.300 per deciliter to 90 milligrams per deciliter on a strong dose. Let's shift gears for a second and

01:11:55.280 talk about something that we briefly touched on, but I think we now want to go into a little bit more

01:11:59.680 detail. If everything we've talked about so far is somewhat complicated, I actually think for me,

01:12:05.800 at least, this next part, especially the RCT stuff, gets complicated. So, let's talk about what HDL

01:12:13.000 lipidation, delipidation, and reverse cholesterol transport are, because this is really where we

01:12:19.380 start to get into some of the sophistication of the HDL and even the interaction with other cells

01:12:24.920 like macrophages and things like that. I'll start and then you can kind of lead me along because this

01:12:29.340 can get quite complicated. And maybe just to orient you, Dan, why don't we start with,

01:12:35.500 unless you have a better way to do it, I'm totally open, but if you're looking for a goalpost to start

01:12:38.900 in, do you want to start with a foam cell stuck in an artery wall or is there a different place

01:12:44.980 you'd want to start the discussion? That's a good idea. So, many of your listeners know that

01:12:49.540 a core concept in atherosclerotic vascular disease is the macrophage that's taken up lipids and is now

01:12:57.400 a so-called foam cell, meaning it looks foamy under a microscope because it has all this lipid and when

01:13:02.400 you stain tissues, the lipids become like bubbles within the cells and they look foamy.

01:13:07.800 The foam cell, the lipid-loaded macrophage is a core pathologic feature of atherosclerotic

01:13:15.620 vascular disease. It's also the first thing you see. So, careful studies that have really looked at

01:13:21.120 vascular tissues in children and teenagers and young adults on the way up, you basically see

01:13:28.240 lipid-loaded macrophages accumulating in the sub-intimal space and the intimal space in the large

01:13:33.240 vessels before you start seeing some of the more complex features of infiltration of other leukocytes

01:13:39.980 and extracellular matrix and all the stuff that ultimately becomes the complex atherosclerotic plaque.

01:13:44.780 There's been a strong belief for a long time that this is one of the core initiators of the process

01:13:49.020 and I think that's probably true. An analogy would be like with Alzheimer's, the A-beta being kind of

01:13:54.500 the core initiator of the process leading to much more complex pathology. So, macrophages have very

01:14:00.560 well-established mechanisms for ridding themselves of cholesterol. You know, keep in mind, no cell

01:14:07.800 except the liver cell has the ability to metabolize cholesterol to other sterile species. Only the liver

01:14:15.360 can do that. Cells can make plenty of cholesterol, but the only way cells can deal with their cholesterol 0.66

01:14:20.960 is to, quote, efflux the cholesterol, to push the cholesterol out of the cell and get rid of it.

01:14:26.720 So, if you think at the whole body level, all of the body is making cholesterol, but that cholesterol

01:14:32.960 ultimately has to come out of those cells into something. This is where HDL comes in and ultimately

01:14:38.900 get back to the liver where the liver then can metabolize it or directly excrete it into the bile.

01:14:44.700 Then it goes out the intestine and the feces. So, all cells have to be able to do this and all cells,

01:14:49.820 in fact, have the ability to push cholesterol out of their cells, but macrophages really have to do it.

01:14:54.860 That is because macrophages are like kind of the dump truck of the body. They're picking up not only

01:15:01.160 LDLs and lipoproteins with cholesterol, they're scavenging cells, dead scales, apopsotic cells,

01:15:07.440 and of course, all those cells have lots of cholesterol. So, macrophages need very, very

01:15:11.780 effective ways to rid themselves of cholesterol, and they do have effective ways of doing that.

01:15:18.040 When those pathways get overcome or less efficient, the macrophage then builds up cholesterol and

01:15:24.040 becomes the kind of foam cell that we're talking about. Macrophages have transporters. They have

01:15:28.820 a very abundant amount of this ABCA1 that we talked about earlier. Earlier, we talked about it

01:15:34.620 in the gut and the liver. Now, I'm saying that macrophages have ABCA1. ABCA1 is one of the,

01:15:40.580 not the only, way that macrophages rid themselves of cholesterol. If you recall, the main acceptor of

01:15:46.800 cholesterol via ABCA1 transporting out of a cell is APOA1. This led now quite a long time ago to the

01:15:54.940 general paradigm that macrophage foam cells are building up cholesterol because they're not getting

01:16:01.200 rid of it effectively, and that one of the best ways to get rid of it would be to promote these

01:16:06.800 efflux pathways via ABCA1 and other transporters that are being driven by APOA1 to HDL, which then,

01:16:15.560 by my garbage truck analogy, the garbage truck, i.e. the HDL, is picking it up in the periphery like

01:16:21.940 the blood vessel and returning it to the liver, dumping it off in the liver, and then going back

01:16:27.120 and doing its job again. This process of, quote, efflux, cholesterol efflux from macrophages and,

01:16:34.400 frankly, other cells, particularly in the blood vessel wall, has been for a long time now thought of

01:16:40.760 as a key process that would help to protect against the early initiation and progression

01:16:47.440 of the atherosclerotic plaque. Peter, that was long, but that's a start.

01:16:51.960 Well, and I think that's, look, let's just make sure everybody understood what we're talking about

01:16:55.500 here, which is when we really talk about the positive valence of HDL as a particle, it's because

01:17:02.980 of this, right? In large part, this is a big piece of the positive association, presumably,

01:17:09.560 or negative association, depending on the direction of HDL, we might see. As we've been

01:17:14.480 talking about it, it really now speaks to function. Part of the problem, it would appear, is that very

01:17:21.060 crude metrics, like the amount of cholesterol in an HDL or even the number of HDL particles or the

01:17:30.060 size of an HDL particle, those are basically the only things we can measure, clinically at least.

01:17:34.900 Those are so crude and so far removed from providing any quantification of the process you just

01:17:43.140 described that I suspect in part that's why we are stuck in a way that we are not on the ApoB side

01:17:50.760 of the ledger. Because so much of the damage caused by ApoB is simply captured in the number of them,

01:17:58.280 given the stochastic nature with which they enter artery walls and get retained. Would you agree with

01:18:02.540 that assessment? I would absolutely agree with that. So this process of cholesterol efflux,

01:18:06.780 as these sort of ideas evolved, is, if you will, the first step in this broader physiologic process

01:18:13.160 that Peter briefly referred to that we call reverse cholesterol transport, or RCT, different than a

01:18:18.740 randomized controlled trial. If forward cholesterol transport is basically the cholesterol coming out

01:18:24.120 of the liver into VLDL and LDL and then depositing in tissues, like the artery, the reverse

01:18:30.360 transport is the picking up of the cholesterol putatively via ApoA1 and HDL and returning it

01:18:36.840 back to the liver. This process of reverse cholesterol transport plausibly, and I think some data, at least

01:18:44.200 in animal models, is related to protection against atherosclerosis. That is, the more effective you

01:18:50.180 are at the integrated process of not only picking up the cholesterol via efflux, but effectively

01:18:56.120 returning the cholesterol to the liver for excretion, the more you would protect against atherosclerosis,

01:19:02.760 as the theory goes. And I'll just remind you about our example with SRB1. That's the terminal,

01:19:08.680 sort of one of the terminal steps where the HDL is dumping off the cholesterol. If you interrupt that

01:19:13.660 process, your HDL goes up, but the process of reverse cholesterol transport is being constipated,

01:19:20.300 and therefore there's increased risk of atherosclerotic cardiovascular disease. One of the great goals

01:19:26.460 of the field has been, can we promote that first step of the process? Can we figure out a way to

01:19:32.860 promote the driving of the efflux process from the macrophages and maybe other cells in the

01:19:39.200 atherosclerotic plaque to acceptors like HDL in order to protect or maybe even regress atherosclerotic

01:19:48.020 plaque, shrink it as a way of trying to reduce risk? And I'll just say that as we move into this

01:19:54.380 next phase of complexity, I think what's pretty clear is it's not the mature HDL particle. You

01:20:00.340 know, when we measure HDL cholesterol, that's what we're measuring is the cholesterol in the mature

01:20:04.760 particle. That's almost certainly not the particle that is driving this first step of the efflux of the

01:20:11.420 cholesterol from the foam cell and the macrophage. And that's maybe why simplistically

01:20:16.760 raising HDL cholesterol, like with CTP inhibition, doesn't actually reduce atherosclerotic cardiovascular

01:20:24.880 disease events. But are there other ways more creatively that we might be able to drive that

01:20:30.880 process? And also to Peter's earlier functional point, might it be that different people, even with

01:20:37.160 identical HDL cholesterol levels, have different function of their HDL? One person with HDL cholesterol

01:20:43.980 of 50 might have something about their HDL pathway that is like super functional at driving efflux.

01:20:51.120 Whereas another person with the same HDL cholesterol doesn't do nearly as well. And if we could measure

01:20:56.960 function efficiently, might we have a better way of assessing risk and maybe even targeting interesting

01:21:03.840 therapies more so than just measuring this fairly not so dynamic measure of HDL cholesterol itself.

01:21:10.700 There's more to say there, Peter, but I'll turn it back over to you.

01:21:13.120 Well, no, I mean, I think you illustrate the very important distinction between static biomarkers

01:21:17.380 and dynamic biomarkers. And we have very few dynamic biomarkers. An OGTT, an oral glucose tolerance,

01:21:23.260 that's a dynamic biomarker in a sense. But most things are very static and static things allow us to miss

01:21:30.980 flux. So that example you gave of two people whose HDL cholesterol is both 50 milligrams per

01:21:37.540 deciliter. We have no clue what the velocity through the HDL is. In one of those cases,

01:21:44.700 it could be a bump on a log and the HDL is not really doing a heck of a lot. And in the other,

01:21:50.720 that could be the busiest beaver on the face of the earth, just transporting lipid out of foam

01:21:55.700 cells, right back to the liver, back and forth, back and forth, back and forth. And that could be the

01:21:58.680 most industrious little guy on the block. And you could be in a totally different risk situation as a

01:22:03.900 result of that. So going back to the RCT, how often is it happening that an ApoB,

01:22:10.060 bite-bearing lipoprotein, let's just call it an LDL, is floating around. So not yet in artery wall,

01:22:17.300 but on his way there, presumably. And an HDL come along and they collide in the artery itself

01:22:23.760 and delipidation takes place. So the HDL says, hey, I'm going to take some of your cholesterol away

01:22:28.820 and take it back to the liver now, which by the way, LDLs do as well. I mean, LDLs are obviously

01:22:33.480 carrying cholesterol back to the liver. But not as a reverse cholesterol transport process,

01:22:37.240 but yeah. The scenario I described, does that occur often? No, I don't think so. In fact,

01:22:42.600 as we discussed earlier with CTP, the directional flux of cholesterol in the blood is more from HDL

01:22:50.440 to ApoB-containing lipoproteins rather than the other way around. So I'm pretty comfortable in saying,

01:22:55.500 I don't think what you just outlined is a way that HDL, HDL doesn't directly delipidate LDL. No,

01:23:01.240 no. I think it's more about if it's happening at all. I want to say all this is couched in,

01:23:05.800 this is the paradigm that we're dealing with, but there's still a lot of uncertainty about it.

01:23:10.200 But if it's happening at all, it's really more that something about HDL or something,

01:23:14.000 ApoA1 is more interacting with cells to promote efflux of cholesterol rather than with other lipoproteins

01:23:20.380 like ApoB-containing lipoproteins. What do you think the future could look like here in terms

01:23:26.180 of commercial assays to measure HDL function? I know that we're a long way away from that,

01:23:30.100 so maybe I should start with something even more basic, which is what would it take in the lab

01:23:36.740 to measure HDL functionality if resources were unconstrained? No, that's where I wanted to go.

01:23:42.260 So first I want to say that several years ago with this concept of reverse cholesterol transport as a

01:23:47.320 dynamic process. We developed an assay in mice that simply speaking involved taking cholesterol

01:23:54.480 loaded macrophages with labeled cholesterol, injecting them into the mice, and then following

01:24:00.020 that label all the way through the HDL to the liver, to the gut, to the feces. And we called that

01:24:06.800 integrated reverse cholesterol transport. And we showed in a variety of different genetic and

01:24:11.560 pharmacologic approaches that when you tweak that process either up or down, it mirrored the effect

01:24:18.160 of that process on atherosclerosis. What I'm really trying to say is something that promoted that

01:24:22.380 process and made it more efficient also reduced atherosclerosis. Something that constipated that

01:24:27.780 process increased atherosclerosis. It gave us a lot of confidence. Frankly, it gave the field a lot of

01:24:32.640 confidence that integrated measure of reverse cholesterol transport is actually relevant to atherosclerotic

01:24:39.260 cardiovascular disease, at least in mice. So that brings me to humans. We can't do the type of

01:24:44.820 experiment I just described in humans. It's not feasible as a research or as a clinical assay.

01:24:49.820 So we thought long and hard about how can we start to think about doing this in people

01:24:54.500 and developed what I'll call an ex vivo cholesterol efflux assay, where the concept is we took someone's

01:25:02.440 blood or plasma, we isolated the HDL specifically, we got rid of the APOB-containing lipoproteins,

01:25:08.340 and then put that on cells that were labeled with cholesterol. And we measured the effectiveness by

01:25:14.880 which that HDL removed cholesterol from cells.

01:25:19.020 What kind of cells, Dan?

01:25:19.480 And what we found, as I think you know, is we were really struck with how different

01:25:24.500 individuals' HDLs were at their ability to extract cholesterol from cells, even with the same HDL

01:25:31.080 cholesterol level. And it sort of affirmed this concept that HDL cholesterol is not really telling

01:25:37.760 us or informing us on the efficiency of the function of HDL, at least in this case, the function

01:25:42.980 defined as the ability to extract cholesterol from cells. And so we went on to use that in larger numbers

01:25:49.260 of individuals and showed that actually that was much more predictive of risk of coronary heart disease

01:25:55.840 than just measuring HDL cholesterol, even when you controlled for HDL cholesterol statistically.

01:26:01.300 And many others have shown the same thing. So I think it's pretty well established at this point

01:26:06.560 that this so-called cholesterol efflux capacity measurement of HDL is a better marker or predictor

01:26:13.500 of risk than just measuring HDL cholesterol, consistent with the idea that function is important

01:26:19.740 and that perhaps if we could increase function, we could maybe have a mechanism for reducing risk.

01:26:26.460 Now, Peter, you asked about the clinical applicability of that assay. You know, I get asked all the time

01:26:30.820 by patients and referring doctors, can you measure my patient's HDL function? There's a lot of effort

01:26:37.680 in this regard. Full disclosure, I was part of starting a little company now several years ago called

01:26:42.600 Vassar Strategies. And Vassar Strategies does do this measurement in a very reproducible sort of way.

01:26:49.740 Mostly related to biomedical, biopharma research, but there is a lot of interest in trying to bring

01:26:55.300 this clinically. And there are other assays that are being developed. You know, when you're using

01:26:59.120 radioactivity, it's a little, you know, it's cumbersome for a high-throughput clinical assay.

01:27:03.740 And others are developing assays that are cleaner and simpler and faster and cheaper. I think there's

01:27:10.760 a chance we could have a more widely available clinical assay to us for this purpose within the next,

01:27:16.860 say, two to three years. Right now, it's still under development.

01:27:20.660 Two follow-up questions there, Dan. One is just a technical question on the assay.

01:27:24.000 What type of cell are you using to measure the efflux capacity?

01:27:29.440 Well, when we first sort of pioneered this work, we used a mouse macrophage cell line called the

01:27:35.240 J774 cell. We have done it with human macrophages as well, and so have others. We like to think that the

01:27:42.300 macrophage is the most relevant cell type for reasons we discussed. Yeah, that's the answer.

01:27:46.740 And then I guess the second question is, using the hypothetical but probably somewhat real example

01:27:53.340 of we'll take 100 people that have an HDL cholesterol that's about the same. We do this

01:27:59.720 assay and we rank order them in effectiveness. You're saying that that rank order of effectiveness

01:28:05.440 correlates directly to risk. Are you saying directly related to risk by proxy, i.e. other

01:28:13.180 measurements like insulin resistance, ApoB, or are you saying actual outcomes? The first study we did

01:28:18.480 and published was cross-sectional with clinical coronary disease. So basically it was-

01:28:23.260 Calcium scores or things like that. Association cross-sectionally with prevalent disease.

01:28:27.200 Then we basically said, we need to know if this is predictive of incident disease,

01:28:31.940 events that occur. So we went to colleagues who did the Eric Norfolk study, which you're probably

01:28:37.180 familiar with, a very large prospective study in the UK, in Europe. They provided samples and we

01:28:43.460 measured this in a very large number, many thousands of samples in a baseline in people who had been

01:28:48.980 followed for 10 plus years. And we showed even in that setting that the efflux capacity was predictive

01:28:55.060 of incident cardiovascular events. So yes, these are hard events, not just proxies like measurement of

01:29:00.720 coronary calcium, for example. Which is, again, fascinating. How well does that prediction hold

01:29:07.980 up if you corrected for other things that could be measured in the blood, such as insulin resistance,

01:29:14.660 triglyceride, or ApoB? We did do these statistical analyses correcting for clinical risk factors,

01:29:20.460 including, of course, HDL cholesterol itself. We corrected for things like BMI and presence of

01:29:26.500 diabetes, dichotomous presence of diabetes. If memory serves, we didn't attempt to correct for

01:29:31.500 like a sophisticated marker of insulin resistance, like HOMA or anything, because I'm not sure we

01:29:36.000 even had that data. So your point is well taken. At some point, is this causal or is it still

01:29:41.700 associative, but just even better associative because it correlates with things even better

01:29:45.820 than HDL cholesterol? And I think that's why we need, of course, interventional studies that

01:29:50.020 actually test the hypothesis. Well, or even could we use it? There could be a hybrid there, right? Which

01:29:56.240 is, it's absolutely causal, but because it's so difficult and complex to initiate at scale,

01:30:03.860 what if we use AI to figure out that it's equivalent to a new metric that is a composite metric of things

01:30:15.920 that we can measure as biomarkers, if that makes sense? I hear what you're saying, and that would

01:30:20.800 be a great thing to think about. In our paper in Epic Norfolk, we did correlations of the efflux

01:30:25.820 capacity with lots of different things. I frankly have to go back and look at that and remind myself

01:30:29.880 if we had like fasting insulin or anything, because I think that starts to get more at the insulin

01:30:34.260 resistance component. But you're absolutely right. We may be able to find markers that in composite

01:30:40.340 would predict efflux capacity rather than having to measure the efflux capacity itself.

01:30:45.400 I'm not convinced we'll be able to do that, but I think that it's a good thought.

01:30:49.460 Because the analogy that comes to mind is, you know, certain labs use a series of NMR metrics

01:30:56.760 to predict insulin resistance, which I've always found not that interesting, frankly,

01:31:01.060 because we can measure insulin resistance so easily that using NMR to add another insulin

01:31:05.600 resistance metric, it seems a bit backwards. But the idea was they were able to basically look

01:31:11.600 at the NMR spectra of all of the lipoproteins and impute a composite metric that came back. So

01:31:18.300 if you could do that in the other direction, which would actually be the valuable direction,

01:31:23.240 it could be amazing. In other words, if there was an HDL function score built out of X, Y, and Z.

01:31:29.240 Anything else before we leave the topic of kind of measurement? We haven't really talked about

01:31:35.040 the NMR side of the equation. That's probably a bias because we don't use NMR, but clearly there

01:31:40.760 are labs out there that will count the number of HDL particles and spit out an HDLP. Anything you

01:31:46.340 want to say about HDLP or anything like that? Yeah, the overall number of particles is obviously

01:31:52.880 correlated with HDL cholesterol, but is a different measure because HDL cholesterol is just that amount

01:31:58.200 of cholesterol carried in the HDL, whereas the particles are more analogous to sort of measuring

01:32:02.700 APOB, the meaning total particle number. I think the data on balance suggests that HDL particle

01:32:08.780 number is a little bit better than HDL cholesterol at predicting risk. We haven't formally compared HDL

01:32:16.900 particle to cholesterol efflux capacity, the functional measurement. Fundamentally, it's another

01:32:22.520 static measure though. And so while it has maybe some limited predictive value, I don't think it takes

01:32:28.560 us that much further in terms of trying to get at where we're really trying to go is better

01:32:34.360 predictions of risk, particularly in a setting where we might want to intervene to actually try

01:32:39.440 to reduce risk. Let's talk about something you mentioned kind of in passing a couple of times.

01:32:44.640 Peter, before we leave this topic though, I do think it's important to remind your listeners that

01:32:48.600 this concept of promoting the efflux and reverse cholesterol transport pathway

01:32:54.200 is being tested with another intervention called CSL 112. It's a form of APOA1 that's been complexed

01:33:03.780 with lipids in a so-called recombinant HDL particle is being tested for impact on cardiovascular outcomes

01:33:10.660 in the setting of acute coronary syndromes. So people have come in, been randomized to four weekly

01:33:17.420 injections of this APOA1-containing recombinant particle, which is very effective at promoting

01:33:23.900 cholesterol efflux based on all the work that's been done with the idea being that this may

01:33:29.500 directly impact the plaque and impact on cardiovascular events. So this is, I would say,

01:33:36.260 the closest thing we have to a formal test of the cholesterol efflux hypothesis as an intervention

01:33:43.840 in terms of whether it will reduce risk. Those of us in the field are on the edges of our seats,

01:33:50.000 it'll be a little while to really see what this trial shows. I do think it's going to be interesting

01:33:55.620 either way. So earlier, Dan, you mentioned kind of briefly the association of plasma APOA1 and HDL

01:34:03.420 cholesterol with neurodegenerative diseases such as dementia or Alzheimer's disease. I don't know if it

01:34:07.760 extends to Parkinson's or Lewy body. What do we know about that? Well, this is, I think, a fascinating

01:34:13.560 new area of HDL biology that has a number of components and that that's still being investigated.

01:34:19.640 I want to maybe start by bringing in APOE. We haven't talked about APOE. APOE clearly has a role

01:34:26.240 in APOB-containing lipoprotein metabolism and mediating the uptake of those remnant particles

01:34:32.300 into the liver. Can you actually explain this a bit more? Because most people hearing this are going

01:34:37.540 to think of APOE, the gene. And of course, the gene for APOE, people are going to be familiar

01:34:43.420 with. It exists in three isoforms, two, three, and four. But of course, these things combine. So you

01:34:49.400 have six combinations of an APOE genotype. And of course, the gene codes for a protein. So I mean,

01:34:56.120 I assume everybody has the same APOE protein. You're going to have different amounts of it and

01:35:00.500 different potentially functionality depending on the combination of which gene produced it.

01:35:04.680 Exactly. APOE is possibly one of the most fascinating genes and proteins in human biology

01:35:11.380 in terms of its roles and the isoform issue and its relationship to disease. So as you point out,

01:35:17.820 there are three common isoforms of APOE, APOE3 being the so-called wild type or the most common.

01:35:23.660 You know, in lipidology, we've focused a lot on APOE2 because if you inherited two copies of APOE2,

01:35:30.740 that form of APOE is defective in binding to the LDL receptor and the other receptors that mediate

01:35:38.460 uptake of remnant lipoproteins, both chylomicron remnants and BLDL remnants. And therefore,

01:35:44.180 if you homozygous for APOE2, you're at risk for so-called, again, the terminology is bad,

01:35:50.120 type 3 hyperlipidemia, which is basically a remnant clearance disorder. These individuals

01:35:56.880 can't clear their remnants appropriately. They get high triglycerides and cholesterol. They also are

01:36:01.740 at increased risk of atheroscardial cardiovascular disease. It's a classic lipidology thing.

01:36:06.160 And these patients require phenofibrates to bring down the VLDL and triglyceride?

01:36:10.960 They respond to a lot of the standard LDL-lowering therapies like statin and azetamide. They even

01:36:15.660 respond to PCSK9 inhibition. But sometimes we have to add on fibrates as well to maximally control

01:36:22.940 them. Depends on how severe they are. And yet those people, and by the way,

01:36:26.660 that's the only genotype I've never seen. It's much more rare than the 4-4 homozygotes. But the

01:36:33.700 2-2 would come with about a 20% risk reduction, relative risk reduction in Alzheimer's disease,

01:36:38.800 but paradoxically comes at this higher risk of ASCVD.

01:36:42.440 Exactly. So comes at higher risk of lipid disorder in ASCVD. But that's exactly why I brought up APOE

01:36:48.640 in response to your comment about neurodegenerative disease. So as probably a lot of people know who

01:36:54.480 are listening, the APOE4, as you just pointed out, the APOE4 form, which is present in about 25%

01:37:01.460 of people, so it's quite common. And even in the homozygous form, certainly plenty of people out

01:37:06.720 there who have both two copies, is in a dose-dependent way the major genetic risk factor for Alzheimer's

01:37:14.100 disease. No question about it. And as you said, Peter, APOE2, which is bad for your lipids in the

01:37:20.560 blood, is actually protective against Alzheimer's. It's absolutely fascinating biologically. There has

01:37:27.180 been so much work to try to understand how APOE is interfacing with Alzheimer's disease.

01:37:34.240 So APOE is made in the brain. It's made by cell types like microglia in the brain. And it's a lipid

01:37:41.060 binding protein that, you know, transports lipids. And now we know from genetics of Alzheimer's that

01:37:45.860 there are lots of other lipid genes that are related to Alzheimer's. So the story starts with

01:37:50.460 the fact that APOE made in the brain is somehow impacting on Alzheimer's risk. And if you have

01:37:56.260 this form of APOE4, and especially if you have two forms of APOE4, you are at very substantially

01:38:02.200 increased risk of Alzheimer's for reasons that we still don't fully understand. But as a lipidologist,

01:38:09.220 I think it has something to do with lipid transport in the brain. So that brings me to HDL. The APOE in

01:38:14.640 the CSF and in the brain is mostly made there. The newer developments that are relevant to APOE1 and HDL

01:38:22.320 are that there is clearly APOE1 in the brain and in the CSF. APOE1 is not made in the brain.

01:38:30.460 So APOE1 gets there somehow through the blood, which I'll come back to. But there's now quite a lot

01:38:38.240 of observational data that strongly suggests that APOE1 is protective against neurodegenerative

01:38:45.980 disease. Now, this is associative data. It doesn't prove causality. But the reason I think it's

01:38:52.800 plausible is that two of the major genetic risk factors for Alzheimer's that are expressed in the

01:38:58.880 brain are ABCA1. Remember that lipid transporter that rids cells of cholesterol and that APOE1 interacts

01:39:07.920 with and promotes. And ABCA7, a very close relative of ABCA1 that structurally looks very similar and that

01:39:18.060 we don't know exactly what it does. But I think it's a safe bet that it's transporting some sort of

01:39:23.740 lipid to something in the extracellular space, whether that's APOE1 or APOE or something else.

01:39:29.200 So the plausibility of APOE1 being protective against neurodegenerative disease, particularly

01:39:37.140 Alzheimer's, is quite high. And frankly, there's a lot of work to be done to try to figure it out.

01:39:43.740 There's one other point I'll make, which is the blood levels of APOE1 and the CSF levels of APOE1

01:39:50.020 and the relatively small number of studies that have measured both in the same people

01:39:53.700 are not that well correlated. So it's not simply a matter of if you have a high level of APOE1 in

01:40:00.340 the blood, that's going to generate a high level of APOE1 in the CSF. So what that probably means is

01:40:06.400 the processes that are happening that get the APOE1 across the blood-brain barrier

01:40:12.420 are highly regulated processes that, a little like the cholesterol efflux story, probably differ

01:40:18.340 from person to person and aren't directly being driven by the level of APOE1 in the blood.

01:40:22.560 But where I'm really going with this is this concept that if we could figure out how that

01:40:28.060 happened, if we could somehow promote more APOE1 going into the brain, into the CSF,

01:40:33.660 the data suggests that that might be a very interesting opportunity to blunt or reduce

01:40:39.860 risk of Alzheimer's. I find it fascinating, as you can tell.

01:40:43.880 Dan, do we know if this relationship is stronger or weaker as a function of APOE genotype?

01:40:52.560 Is the effect more pronounced? Do we have enough data to parse out that type of a relationship?

01:40:57.440 So I'll give you an example.

01:40:58.160 Yeah, that's a really super question. And I don't think we have big enough data sets yet

01:41:04.100 to be able to parse it out in terms of by APOE genotype to figure that out. But I have to go

01:41:09.020 back and look for that. But I don't think we know that yet.

01:41:11.220 Do you have any idea why labs have not developed what would be very easy to do,

01:41:17.580 which is just a commercial assay for APOE concentration? There's certainly some data

01:41:22.080 to suggest that APOE concentration might be more relevant than APOE genotype, although of course,

01:41:29.120 it's highly influenced by APOE genotype. But you could almost think of it as a gene expression

01:41:33.060 measurement, a crude measurement of gene expression for APOE. Is that something that you've seen or used,

01:41:40.600 even in the lab?

01:41:41.980 You're talking about APOE in the blood, right?

01:41:43.620 Yes.

01:41:43.780 Not in the CSF, just to be clear. There are automated assays for APOE. We run one routinely

01:41:48.840 in our lab. I think it's not a clinically used assay.

01:41:52.300 No, I've never even seen a CLIA-based assay for it.

01:41:54.960 Right. I think mostly because it isn't that helpful in the limited studies that have been

01:42:00.120 done, which admittedly maybe aren't as voluminous as they should be. It isn't that really helpful

01:42:04.760 in terms of predicting cardiovascular risk. Total plasma APOE...

01:42:08.480 Oh, I was thinking more for neurodegenerative disease. Has that been looked at?

01:42:11.660 Oh. Well, I didn't say it, but the levels of APOE in the CSF and the levels of APOE in the blood

01:42:19.020 for the few studies that have measured both in the same people also do not correlate much at all.

01:42:24.360 You would need CSF APOE if you wanted to do anything with this.

01:42:27.680 Exactly. That's what I'm getting at. You would need to be able to do it.

01:42:30.220 I would not be surprised if someone develops an assay for APOE concentration in the CSF as a

01:42:36.120 clinical tool, because there I think that might be highly relevant.

01:42:40.360 Interesting. So here we have basically HDL potentially being protective in the brain

01:42:48.000 and presumably doing so via APOA1, potentially offsetting some of the APOE problem, APOE being,

01:42:58.360 as I think one of our previous guests referred to as kind of the general contractor of cholesterol in

01:43:04.200 the brain. Does HDL do anything with nitric oxide? I think I remember reading something about

01:43:09.900 promoting sort of endothelial NOS activity, nitric oxide synthase activity. Is that a major issue?

01:43:16.180 There's been a variety of other, quote, functions that HDL has been reported to do,

01:43:21.260 which I think are absolutely real. Their relevance to human disease and pathophysiology,

01:43:25.600 I think are less clear. But one example is what you just said. So a couple of investigators,

01:43:30.720 most notably Phil Shaw in Dallas, has shown very clearly and beautifully in cells in mice that

01:43:37.040 HDL can promote nitric oxide production in a way that would be expected to be beneficial,

01:43:43.600 both in terms of blood pressure lowering and maybe protection against atherosclerosis.

01:43:47.920 And that SRB1, the receptor we've been talking about, is also present on endothelial cells and

01:43:53.220 mediates at least part of that effect of HDL. It's a fascinating observation that is absolutely,

01:43:59.200 I think, solid. Translating that observation to relevance in humans, I think is challenging,

01:44:05.140 but I think plausible. You know, another, we've been talking about insulin resistance. I'm sure

01:44:09.000 you're aware of the data that suggests that HDL can interact directly with skeletal muscle

01:44:15.120 in a way that promotes insulin sensitivity.

01:44:17.920 No, I'm not.

01:44:19.880 Fascinating work. Again, I'd put it in the same category of solid in terms of what it's been able

01:44:25.480 to show, mechanistically a little bit unclear, and relevance to human disease and physiology,

01:44:31.480 I think, still a little bit unclear. These are just two of the other types of things that HDL

01:44:37.060 has been shown to do that would be putatively beneficial, but are of uncertain relevance to

01:44:42.260 real human disease. You know, it's even more sort of removed than that because even if we knew

01:44:48.900 mechanistically that this were sound, we're still back in the same area we are with ASCVD.

01:44:55.080 In ASCVD, there's really very little ambiguity about the utility of HDL, the beneficial utility of

01:45:01.200 HDL, and yet we're standing here with our hands in front of us saying, well, what can we do about it

01:45:08.680 clinically? As a physician, as a patient, what do I get to do about this knowledge that HDLs are

01:45:16.660 helpful particles when I can't measure the manner in which they do their job? And the only things that

01:45:23.600 I can measure are as useless to me as my eye color, basically. And then we're expanding into,

01:45:29.860 and look what they can do in muscles, and look what they can do in insulin sensitivity, and look

01:45:33.400 what they can potentially do in the brain. It really comes back full circle to how we open the

01:45:37.440 discussion, Dan, which is this is such a complicated area of biology that I would guess it has to be

01:45:45.580 considered the next frontier of the lipid space. I mean, when we take a step back and put ASCVD in

01:45:52.000 the context of cancer and neurodegenerative diseases, right? So these are the big three

01:45:55.600 killers in the modern world. We know so much more about ASCVD, and we have so many more tools to

01:46:02.200 effectively treat it. We know a bit about these other diseases, but in the case of Alzheimer's,

01:46:06.300 we don't have a single tool to do anything about it. In the case of cancer, most of our tools do

01:46:10.940 nothing. Ninety percent of the tools do nothing, i.e. they barely extend median survival but don't

01:46:16.060 cure people. But in ASCVD, we can really move the needle, and yet you could argue half of the field

01:46:22.220 we still know nothing about. And is there going to be a renaissance, you think, or are we up against

01:46:27.940 some technical limitation on this inability to measure function? And I say that in a practical way.

01:46:33.980 I mean, yes, I think you're already measuring function in the lab, but is this a bridge too far

01:46:38.800 for the clinician? Well, I'm going to separate that into using a measurement of function for

01:46:44.540 risk prediction and then the implications for intervening therapeutically. I think that HDL

01:46:51.100 function, say cholesterol efflux, has the ability to allow us to more specifically assign risk

01:46:58.540 better than just measuring HDL cholesterol. I think realistically, we need a good reproducible,

01:47:05.600 easy to run, automatable assay that then is tested in large numbers of people and shown to predict risk

01:47:14.020 better than HDL cholesterol itself. I feel like knowing what I know is going on in terms of developing

01:47:19.840 those assays, that there's a decent chance that we're going to get there. But it's not just the

01:47:25.220 development of the assay, which actually there are several assays now that are, but it's the proving

01:47:29.240 that the measurement actually enhances risk prediction enough, not just incrementally, but

01:47:36.020 enough to make it actually worth doing in clinical practice. I think there's a, I'm not going to say

01:47:40.780 a hundred percent, but I think there's a more than 50% chance that we're going to see at least some

01:47:45.740 assay come out that will allow us to do that. And then it will be, you know, frankly, physicians like

01:47:51.180 yourself and like Tom Dayspring, who pick up that assay and start using it in their patients and

01:47:57.120 start trying to get a sense for whether it's useful in the context of a sophisticated preventive

01:48:02.060 clinical practice. I think with regard to intervention, the study I talked about is huge

01:48:08.200 in terms of this. Let's just assume that that study is positive. Four infusions, weekly infusions,

01:48:14.980 significantly reduces cardiovascular events in patients with ACS after 90 days.

01:48:19.820 If that study is positive, if you think about that, the implications for that product, but more

01:48:26.500 importantly for the field that showing that actually triggering cholesterol efflux via infusion of that

01:48:34.920 particle, that type of particle, that will rejuvenate the field of cholesterol efflux and reverse cholesterol

01:48:40.980 transport as a therapeutic target for intervention. I think if that trial is not positive, I think the

01:48:47.600 concept of intervening around HDL and reverse cholesterol transport for purposes of ASCVD is

01:48:53.960 probably past recovery. That would be my view. And even that CTP inhibitor that I mentioned that's

01:49:01.160 still in development, it's really more about reducing ApoB than it is about raising HDL.

01:49:05.740 When does that trial read out, the function trial?

01:49:08.280 I should have refreshed my memory on that. I just can't remember.

01:49:12.120 Where is it being done? How many centers, roughly?

01:49:14.380 It's a global study. There's centers in US and Australia and Europe. It's a classic global large

01:49:21.180 cardiovascular outcome trial. I think the leaders of the trial, I think, are in Boston.

01:49:24.980 Remind me, does it have a fancy acronym yet?

01:49:26.960 Yes. And I'm embarrassed to say that I'm not quite coming up with that either.

01:49:31.960 We will figure that out and it'll be in the show notes.

01:49:34.200 If I could just follow up on your next frontier comment. I don't think HDL per se is the next frontier,

01:49:39.140 but I do think lipid metabolism in the brain is one of the next frontiers. I really do.

01:49:44.740 With regard to neurodegenerative disease, but also other brain function, I say that as a long-term

01:49:51.360 lipidologist who has focused primarily on the blood. So I admit my biases and looking for things that we

01:49:57.920 can apply as lipidologists, our expertise to. But I do think understanding lipid metabolism in the

01:50:04.640 brain and its relationship to disease. The brain is, of course, the most lipid-rich organ of any

01:50:09.920 organ in the body has been an under-investigated area that we have a lot to do, but where I think

01:50:15.900 we're going to be uncovering some very interesting things that, with luck, have implications for

01:50:21.320 therapeutic intervention to prevent neurodegenerative disease.

01:50:25.140 Dan, it's hard to go anywhere from here. That is, I mean, both exciting and important in ways that

01:50:30.780 are rarely captured, even in this subject matter. So I want to thank you for taking a subject matter

01:50:36.480 that is so complicated that I've largely shied away from really doing anything on it in the AMA

01:50:42.540 series that we do, and so eloquently describing it and really using, if someone's listening to this

01:50:48.140 and this is the first podcast they've ever listened to of ours, they might think, oh my God,

01:50:51.180 that podcast is too technical. But I think for regular listeners, they will appreciate that

01:50:55.440 you really did a great job simplifying a very complicated subject matter and certainly did a

01:51:01.220 better job than I would have ever been able to do, not simply because I don't know as much,

01:51:05.260 but because the ease with which you were able to speak about this is impressive. So thank you for

01:51:10.300 taking the time. Thank you for doing that. And most importantly, of course, thank you for your work

01:51:13.400 in this field. We, I think, collectively look forward to seeing how things shake out over the next

01:51:17.720 decade. Thanks very much, Peter. I really enjoyed it. Thank you for listening to this week's episode of The

01:51:22.600 Drive. If you're interested in diving deeper into any topics we discuss, we've created a

01:51:26.980 membership program that allows us to bring you more in-depth exclusive content without relying

01:51:31.740 on paid ads. It's our goal to ensure members get back much more than the price of the subscription.

01:51:37.380 Now to that end, membership benefits include a bunch of things. One, totally kick-ass comprehensive

01:51:42.660 podcast show notes that detail every topic, paper, person, thing we discuss on each episode. The word on

01:51:48.740 the street is nobody's show notes rival these monthly AMA episodes or ask me anything episodes,

01:51:54.840 hearing these episodes completely access to our private podcast feed that allows you to hear

01:52:00.180 everything without having to listen to spiels like this. The qualies, which are a super short podcast

01:52:05.980 that we release every Tuesday through Friday, highlighting the best questions, topics, and tactics

01:52:10.680 discussed on previous episodes of The Drive. This is a great way to catch up on previous episodes

01:52:16.180 without having to go back and necessarily listen to everyone. Steep discounts on products that I

01:52:21.760 believe in, but for which I'm not getting paid to endorse, and a whole bunch of other benefits that

01:52:26.320 we continue to trickle in as time goes on. If you want to learn more and access these member-only

01:52:30.940 benefits, you can head over to peteratiamd.com forward slash subscribe. You can find me on Twitter,

01:52:37.700 Instagram, and Facebook, all with the ID peteratiamd. You can also leave us a review on

01:52:43.440 Apple Podcasts or whatever podcast player you listen on. This podcast is for general informational

01:52:49.380 purposes only and does not constitute the practice of medicine, nursing, or other professional

01:52:53.840 healthcare services, including the giving of medical advice. No doctor-patient relationship is

01:52:59.960 formed. The use of this information and the materials linked to this podcast is at the user's

01:53:05.280 own risk. The content on this podcast is not intended to be a substitute for professional

01:53:10.360 medical advice, diagnosis, or treatment. Users should not disregard or delay in obtaining medical

01:53:17.280 advice from any medical condition they have, and they should seek the assistance of their healthcare

01:53:22.320 professionals for any such conditions. Finally, I take conflicts of interest very seriously. For all

01:53:28.900 of my disclosures and the companies I invest in or advise, please visit peteratiamd.com forward slash

01:53:36.300 about where I keep an up-to-date and active list of such companies.

01:54:06.300 Thank you.

The Peter Attia Drive - January 30, 2023

#240 ‒ The confusion around HDL and its link to cardiovascular disease ｜ Dan Rader, M.D.

Episode Stats

Summary

Transcript