Let’s talk Lipoprotein(a) and cardiovascular disease risk (hereafter, Lp(a)=Lipoprotein(a) and CVD=cardiovascular disease).
Lp(a) blood level show a dose-response relationship to cardiovascular disease risk in multiple studies, and Lp(a) is now widely considered a reliable marker for CVD risk.
Here is a graph showing ~4-fold increased risk at >95th-percentile Lp(a) level.
[Above is adjusted for age (left) and multifactorially (right). Latter: age, total cholesterol, LDL cholesterol, HDL cholesterol, triglycerides, apoB, BMI, hypertension, DM, smoking, lipid-lowering therapy, and, for women, menopause and HRT.]
Unadjusted, >95th percentile Lp(a) associates with >2x lifetime incidence of myocardial infarct (heart attack). At 80yrs, 15% at <22nd but 30% at >95th will have had MI. 30% @ <22nd but 40% at >95th have CVD. Lp(a) = CVD more frequent, more severe.
Why did investigators adjust for those variables in the first graph? Because risk factors go together: smoking goes with bad diet, lack of exercise, low income, etc. So if you do a study and conclude smoking causes X, it could actually be all those other things causing X.
So what investigators do is use statistical methods to estimate what the data would look like with all variables of non-interest equal, and just the variable of interest changing.
This multifactorial adjustment usually causes the risk attributed to the variable of interest to decrease relative to the unadjusted risk. Yet in this study, the risk doubled after adjustment. This is unusual, and investigators noted this but couldn’t explain it.
Could it be that somehow Lp(a) protected participants from diabetes, stopped them from smoking, reduced cholesterol, caused lower blood pressure? We may look at this more later in the thread.
Here is one more study (2016) showing the same epidemiological trend, using two different assays for Lp(a), analyzing data from the UK, showing the same results. 2- to 4-fold increase in risk in CVD.
Most people have ~1/3 chance of dying from CVD. CVD probably dramatically increases Alzheimer’s risk as well. This means that at the upper quintile for Lp(a) blood levels, we expect a large increase in absolute risk of both CVD and other diseases. It’s a really big deal.
What is Lp(a)?
Lp(a) is a lipoprotein.
A lipoprotein has three parts: a core containing triglycerides and cholesterol, a shell containing phospholipids, and a skeleton holding the shell together called apolipoproteins.
Attached is an image of a generic lipoprotein.
Lp(a) is virtually identical to LDL.
So we need to know what LDL is, to know what Lp(a) is.
LDL is a lipoprotein, the same as the above. However, the apolipoprotein that holds LDL together is called apoB. The apoB is what makes LDL LDL.
(Other lipoproteins, like HDL, have other apolipoprotein skeletons that make them what they are. In fact, the type of apolipoprotein skeleton that holds together the shell is what determines what kind of lipoprotein the lipoprotein is.)
From a recent review in Cell from the discoverers of LDL, here is a representation of LDL. As you can see, it’s the same as the image a few tweets back, except the apolipoprotein here is specifically apoB. ApoB is what makes LDL LDL and not HDL, etc.
To continue, Lp(a) is virtually identical to LDL, as explained here. So why is it not just called LDL? Because there is one characteristic that distinguishes it from LDL: another protein, called apo(a), is added to the apoB. This makes Lp(a) Lp(a) and not LDL.
As you can see, Lp(a) is just LDL with another protein added: apo(a). So we can use the following formula:
Lp(a) = LDL + apo(a).
Thas it mayne.
We can think of LDL as a baseball, and apoB as the stitching.
Inch-long spikes. Added to the baseball stitching. That help the particle shear through blood vessel walls.
So just eat healthy, right?
Alas, if only it were so.
Lp(a) levels, and their substantial influence on CVD risk, are almost entirely genetically determined.
This means Lp(a) is largely unmodifiable by lifestyle choices, with some modest exceptions, which we discuss later. So can nothing be done about Lp(a) levels? Not exactly. Teaser: one study found reduction of Lp(a) levels reduced risk of CVD in one patient population by 86%.
Before we get there, though… What this information does mean is that diet and lifestyle aren’t going to help with Lp(a). To make matters worse, it has been shown in several studies that weight loss actually increases Lp(a) levels. For example:
Indeed, studies such as this and this one have also consistently shown an inverse relationship between blood triglycerides and Lp(a), i.e. as fat in the blood goes up, Lp(a) goes down, and vice versa.
Another similar and perhaps even more confusing (…and consistently reproduced) paradox holds for type 2 diabetes (as it does for cholesterol more generally). But for now, the question must be asked:
How does any of this make sense?
I.e. that healthy lifestyle choices result in a worsening of a major CVD risk factor? That this risk factor is widespread, and is largely genetically determined (when not being made worse by good lifestyle decisions)?
Why would evolution do this?
To start answering this question, we need to buckle down and talk genetics. Apo(a), the apolipoprotein (if you’ll recall) that hangs onto the skeleton keeping the Lp(a) together, is a real oddball of a gene and protein. (Briefly, let’s look at the Lp(a) particle again.)
Now let’s introduce apo(a) protein structure. As you can see, it has all of these IV things and a P and it’s attached to the apoB. What the hell is this?
Hold on. It’s easier than it looks.
Turns out, apo(a) came from gene duplication of plasminogen, which is responsible for blood clotting.
What’s gene duplication? As it sounds. During evolution plasminogen was copied twice in DNA by mistake.
And from that copying: Voila. Lp(a), plasminogen’s hellspawn offspring, was born.
This duplication event happened around 40 million years ago, making Lp(a) unique among apes, old world monkeys, and humans. Baboons have it. Chimps have it. And now you have it.
And European hedgehogs, which evolved Lp(a) independently.
So how did we get Lp(a) from plasminogen? Plasminogen consists of 6 protein domains: KI, KII, KIII, KIV, KV, and P. Think of these domains as substructures of the larger protein structure. (link)
Well, as you can see, after duplication, 3 out of 6 of these domains were lost. So that we’re left with KIV, KV, and P. Then, the KIV domain within the plasminogen is duplicated ten times, to produce KIV1-10. That’s the standard template.
What happened next? Next, KIV2 is itself duplicated, anywhere from two to >40 times. So, we either end up with a really huge apo(a) molecule, or a much smaller one. Or anything in between. And that’s what the apo(a) variation looks like among humans.
By the way, where does the ‘K’ come from in the domain name? Turns out it’s named after a Danish pastry, which is called a Kringle, and which it looks like.
Again, plasminogen has domains KI-KV, and P. These were duplicated to make the forerunner gene to apo(a). KI, KII, KIII were then deleted, leaving KIV, KV, and P. Then KIV was itself duplicated 10 times to create KIV1-10, each duplication itself then independently evolving.
Finally, KIV2 was itself duplicated anywhere between 2 and >40 times, with everything else remaining very similar. Thus, humans have a range of sizes of apo(a) in their Lp(a) molecules.
If you find any of this confusing, don’t hesitate to re-read or ask me on twitter.
But with this background, we can start to sketch out why different humans have different levels of genetically determined Lp(a). And what Lp(a) does.
Now we’re ready to understand why Lp(a) varies so much between people.
But first, a brief review with some a couple of new figures. As you can see below, the number of KIV2 copies dramatically affects the size of apo(a).
One last figure. Distribution of Lp(a) levels in the population. Where the peak is, the majority of the population is. As you move to the right, the risk goes up. The cutoff is arbitrary. Any double digit represents an incrementally increasing risk.
Figure (a) is the same as above, except using a less sensitive assay (thus the lower values).
Figure (b) is the meat. As apo(a) gets bigger, Lp(a) blood levels go down. And vice versa.
This phenomenon (apo(a) size) is thought to account for 15-70% of the variance in Lp(a) levels, depending on the population. White and asian populations are thought by investigators to be closer to the higher end of that range.
What’s the mechanism here? Why should apo(a) size be so strongly determinant? First of all, Lp(a) levels are thought to be due to differences in production, not breakdown.
Let’s look at radiolabeled apo(a) tracked over time in three subjects. The apo(a) in each subject was highly heterozygous for size, meaning that each subject had two apo(a) proteins of two very different sizes. Yet the breakdown rate is identical.
Let’s now compare production rates with breakdown rates in human subjects. FCR means fractional catabolic rate, i.e. fraction of protein broken down/day. Can see FCR is very similar between genotypes (apo(a) sizes), but production rate varies a lot.
Why does production rate vary due to apo(a) size? Investigators looked at liver cells in a petri dish, observing what happened when cells were made to express apo(a) of different sizes. Here’s a blot with apo(a) of different sizes.
Expression was highest for the lowest kringle repeat isoform, and practically nil at 34. (Since Western blot is not [typically] truly quantitative, this does not mean 34 was not expressed. But relative expression #’s are clear.
Through one experiment in particular (but also others), investigators showed that larger isoforms were retained longer in the endoplasmic reticulum of the cell, possibly because processing of the larger protein (glycosylation) took longer.
Left side (K10) represents 10 kringle isomer; right (K22), 22 kringles. apo(a)=mature, pr-apo(a)=pre-apo(a). K10, lower on gel because migrates faster because smaller. PNS=everything except nucleus.
So, more K22 pr-apo(a) is found in PNS and ER, and less of the mature form is secreted. This is in line with above tweets: larger isoform requires more and longer processing, and that this lower secretion accounts for lower Lp(a).
Another paper suggests that apo(a) is degraded intracellularly, and that the rate of this degradation can be modulated by environmental stimuli (in the paper’s experiments, oleic acid). So conceivably, the larger isoform may be degraded as well.
This discussion raises a few questions: why is KIV2 repeated at all? Is it an evolutionary mechanism to reduce protein expression? If so, why not repeat other KIV domains? Or make the promoter less efficacious, etc.? Furthermore, if evolution “wanted” to reduce expression, then why didn’t the high kringle isoform see a selective sweep? What is the function of KIV2 anyway? To address these questions, we will turn to the function of KIV2, and the major protein that it interacts with–a protein that is found in high concentrations in the blood.
And before turning back to the genetic control of human apo(a), we will also take a look at the apo(a) found in other species–and how it has independently evolved, and what this independent evolution might tell us about human apo(a).
As interesting as these questions are, another is more pressing: why is Lp(a) atherogenic? We will answer this, then work backward. It is necessary to answer this question in a way that accounts for the evolutionary context and normal physiology of Lp(a).
We can discuss cytokines, inflammatory markers, lipoprotein fractions and particle sizes at great length, but what does any of this mean? Why are these molecules behaving in this way?
What is the process that these molecules normally facilitate? And why does the body “misreact” to its own self, produce a spiraling immunological response against the vasculature, and eventually cause clotting that cuts off the blood supply to vital organs?
And, from an evolutionary point of view, why should the body fail to function properly at an age associated in our species with the rearing of our children, putting them in jeopardy at a biologically predetermined precarious age?
Let’s start with the evolutionary question first. Because it’s necessary to understand that from an evolutionary point of view, cardiovascular disease at a young adult age (<60) is abnormal. Let’s take a look at this table by Gurven and Kaplan (2007):
That table is the result of a review of all of the demographic data on mortality among hunter-gatherers, comparing it to the data among Americans and pre-modern Swedes (defining modernity as starting ~1800).
And what’s clear in that data is that no matter what population is studied, the modal population average life expectancy among humans is 65-85. For reference, here is a graphical representation of the same data.
Likewise an entirely different source seems to be making the same point, though here reported as an average (causing e.g. Roman politicians, in a system where the rules consisted of killing your political opponents, to fall short). But see: ~65. (link)
So the human body seems “hardcoded” to “strive” for at least 65, even in the worst of conditions. To illustrate this still further, let’s look at the specific breakdown of causes of mortality among the groups studied in the Gurven, Kaplan paper.
Here we see that degenerative disease before the age of 60 is extremely rare. Even after 60, only about 30% of the deaths are attributable to it. Instead, the human organism is even after 60 still dealing largely with violence, wounds, infection.
Degenerative disease is simply not biologically normal before the age of 60. And yet it is among Americans.
In the United States according to the most recent CDC data, the #1 cause of death for both males and females at ages 45-54, 55-64, 65-74, and 75+. Cardiovascular disease is #1 at all four age ranges. #4 at ages 35-44. (source)
And if the data on hunter-gatherers was not enough, we have still more data from multiple careful investigations that this high rate of cardiovascular disease is distinctly modern, and not due to living longer. From Staffan Lindeberg‘s book, autopsy study from mid-20th century.
Here’s another one, from related research also on Ugandans in 1958. From autopsies of white and black Americans, and black Ugandans, 245 in each of 3 groups, matched age.
Here’s yet another, age-adjusted, comparing disease rates of Americans to Japanese to Okinawans, from Willcox et al.
And everyone knows about, or should know about, the NIHONSAN Study, the one that showed that Japanese who migrated to the United States showed progressively higher rates of cardiovascular disease, eventually matching that of white Americans.
And that’s DESPITE the fact that Japanese were already thoroughly modernized (though not at the level of Americans). Similar differences between Japanese and American Japanese remain, pointing to a persistent distinctiveness in the type of modernity that Japan has chosen.
But what I hope I’ve established is: the incidence of cardiovascular disease in the United States is not normal human biology, and that it’s due to something in the environment, which is foreign to the homeostatic system that has developed over the course of human evolution.
Now let’s review the pathophysiology of cardiovascular disease. The rock solid autopsy findings and some of the associations and intervention data that have formed our present understanding. We will cover both the lipid and the inflammation theories.
After, we can hone in on apoB and unfold its physiology and discuss the growing data showing its determinative role in LDL composition and CVD pathophysiology.
ApoB also happens to be apo(a)’s binding partner.
Now, the review of CVD.
We are going to start very basic, textbook lipidology. What is LDL, VLDL, chylomicrons, etc. Instead of a strictly technical discussion, focus will be on mnemonics and trying to making sense of things. If you try to just memorize the particles, things are nearly impossible.
But if you understand and have handy mnemonics, this makes things easier. Let’s start with a lipoprotein particle. We have already seen one of these. Here is a lipoprotein particle.
Lipoprotein particles are used by the body to transport triglycerides and cholesterol to and from the tissues. Triglycerides are energy molecules: basically, fats. Cholesterol is a building block that the body’s cells use to make important biological molecules.
So we can think of lipoproteins as transport molecules, like big haulers on a freeway. Lipoproteins are our 18-wheelers. The blood vessels are our perfectly designed road system, with major highways, side roads, alleys, and tiny causeways to just barely squeeze through.
And the body’s cells are the individual houses that require delivery or shipping services. The lipoproteins stop at these to deliver and pick up.
But the lipoproteins aren’t simply cargo trucks. No. Because there is the constant threat of invasion by bandits.
We call these bandits microbes, e.g. viruses, bacteria, parasites. And although usually these bandits usually try to invade along small thoroughfares and alleyways, sometimes they make their way onto the highways (arteries).
And if they take over the highways, they can quickly, like a Batman villain, take over the city and try to destroy it. So in the blood we have patrolling–alongside the cargo trucks, 18-wheelers, and so on–police officers and hard hitting military vehicles (immune cells).
ut this ain’t no peace loving town. No.
The threat of bandits is constant. So even the cargo trucks are armed. Like this, the tanker from Mad Max 2. With all the same moral ambiguity as in the film.
You see, the lipoproteins are not merely energy (triglycerides) and material (cholesterol) transporters. They are crucial agents in the regulation of the motorways, as well. And that includes law enforcement. The lipoprotein system is also a part of the innate immune system.
We will discuss the evidence for this later, but it is best to introduce the concept now. The current thinking is that to make sense of the data on CVD and especially Lp(a), we must consider the lipoprotein system an integral part of host defense.
To see how this might be, let’s circle back and look at the structure of a lipoprotein again. Let us finish this in detail. And then circle back to illustrate how a lipoprotein could play a role in host defense. Let’s start with the phospholipid shell of the lipoprotein.
The phospholipid shell is what allows the lipoprotein particle to exist in the blood. That’s why the lipoprotein particle is necessary: because it has a phospholipid shell. Take a look at the structure of the major phospholipids in the LDL particle.
We can see one end that is charged, and the other end is nonpolar.
The charged end interacts with water, which is polar.
Charged parts of molecules interact with polar molecules.
The nonpolar end, the fatty acid end, interacts with the fatty cargo inside.
Without this structure of phospholipids, the fatty cargo would exist in the blood, but it wouldn’t dissolve. So it would be like oil on water and would stick to the blood vessel walls and wouldn’t be transported. That’s why we need lipoproteins with their phospholipid shell.
What about the cholesterol in the shell of the lipoprotein? It exists to modulate the stability of the phospholipid shell, altering its rigidity and shape for purposes that will not be discussed here. Different amounts of cholesterol create shells with different properties.
OK, before discussing the apolipoprotein, let’s look at the cargo. We see esterified cholesterol. That’s cholesterol that has had something added to it (an ester) to make it even more polar, and thus better able to fit inside the lipoprotein.
Cholesterol, as we said, is a building block for cells to use. And triglyceride, as we mentioned, is the energy dense cargo that the lipoprotein can deliver. As we shall see, different lipoproteins have different relative amounts of cholesterol and triglyceride.
Let’s now discuss the apolipoprotein. The “apo-” part is from the Greek for “away” or “from”. When the apolipoprotein is attached to the rest of the lipoprotein molecule, well, it’s a lipoprotein.
The apolipoprotein “organizes” the shell and lipoprotein. Now let’s get to the meat. The apolipoprotein is a very complex particle. It interacts with many different proteins in the blood. Sometimes multiple simultaneously. It can catalyze reactions. It can change shape.
Perhaps most importantly: this figure is a gross simplification that really mystifies the immunological function of lipoproteins. In fact, the shells of many lipoproteins are studded with still more proteins with widely divergent functions.
This paper, published by researchers at Harvard in 2017, notes there are as many as 95 accessory proteins associated with HDL particles. Again, these additional proteins exist alongside the apolipoproteins inside or attached to the phospholipid shell.
And indeed in the same paper, the authors note that the functions associated with these accessory proteins range from inflammatory response, hemostasis, immune response, metal binding, and protease inhibition.
The first three of these are host defense functions. And this is just HDL. We will see a lot more of this in a moment with LDL and especially Lp(a). For now, we must only briefly mention it, as there is much more to cover.
Now let’s cover the lipoprotein types. Chylomicrons, very low density lipoproteins (VLDL), intermediate density lipoproteins (IDL), low density lipoproteins (LDL), and high density lipoproteins (HDL). It is worth noting the relative sizes.
Think Chylomicron = Cuisine. Chylomicrons are made in the intestinal cells from food (hence, cuisine). They traffic triglycerides, cholesterol, and cholesterol esters from the intestines to the muscles, heart, and adipose tissue. (link)
When chylomicrons become “depleted”, they are called chylomicron remnants and traffic the remainder of their cargo to the liver. And that’s the introduction to chylomicrons.
For VLDL, think Velocity. By providing energy, you move.
As we see, they are 50% triglycerides. If we look only at the core, they are 75% trigs, 25% cholesterol. They are produced in the liver and pass their cargo in a manner similar to chylomicrons. (link)
LDL are what happens to VLDL when it becomes depleted of triglycerides. VLDL’s proportion of cholesterol goes up, and it becomes LDL. For LDL, think Less (than VLDL).
As for intermediate density lipoprotein (IDL), IDL are in between VLDL and LDL. IDL is formed from VLDL as it becomes depleted of trigs. As IDL becomes further depleted, it becomes LDL. Not all IDL become LDL. In fact, most do not, and these are recycled to the liver.
It is thought to play a cardioprotective role, though this role has been disputed in recent years. There is much to discuss with HDL, and we will discuss it as necessary.
Now for Lp(a), a protein similar to LDL but synthesized through its own, unique pathway.
To understand Lp(a)’s uniqueness, it is necessary first to go slightly more in depth on what determines the class of each particular lipoprotein: the apolipoprotein, obviously. Here is a figure.
As can be seen, VLDL has the B100, C, and E apolipoproteins. But LDL has only B100. This is because as VLDL is depleted, it eventually has its C and E apolipoproteins stripped. LDL is distinguished from VLDL not just by depleted trigs, but also this loss of apoE and apoC.
Now here’s the interesting part. Lp(a), as we saw earlier in this thread, is characterized by apo(a) attached covalently to apoB.
LDL is defined by apoB by itself.
So is Lp(a) simply a modification of LDL as this image by Wolfson implies?
Whereas LDL is derived from VLDL by loss of apoE and apoC and depletion of trigs, Lp(a) is synthesized DE NOVO in the liver, already with trigs depleted and C and E gone.
Lp(a) not some additional lipoprotein evolutionarily grafted onto existing lipoprotein metabolism, as an afterthought. It’s its own special beast.
Lp(a) is not LDL with apo(a) attached. It follows its own rules.
And in more ways than this.
Before we continue, let’s point out one more feature of Lp(a). Remember the convergently evolved European hedgehog Lp(a)? It is also characterized by an apo(a) molecule bound to apoB.
Here is an image of Lp(a) that could be hedgehog or human.
Is there something about this relationship to apoB that makes Lp(a) what it is? Well, let’s ask: what might be special about apoB?
apoB is the apolipoprotein that is the best surrogate for LDL particle number.
Now not all apoB is bound to LDL. Some of it is on chylomicrons, VLDL, IDL, as we have seen. But the vast majority of apoB is associated with LDL.
So much better, indeed, that the American Association for Clinical Chemistry in 2009 recommended a shift away from LDL-C to apoB as the preferred predictive biomarker and intervention target.
Still, in 2015, while acknowledging apoB was marginally better than LDL-C for CVD risk, a group of researchers concluded that the difference wasn’t significant enough to warrant transition from current guidelines to measure LDL concentration.
Why should apoB be special? First, it was the first lipoprotein thought to be concentrated in atherosclerotic lesions, and this remains its claim to fame within the dominant paradigm, which gives apoB a special place. (link)
Unfortunately the studies commonly cited as “proof” of apoB’s causative role seem to only assay for apoB–and don’t report assaying for any other apolipoproteins! This means, while they handily show apoB’s concentration in atherosclerotic lesions…
Nonetheless, they don’t seem to be able to rule out the possibility that other lipoproteins could be of greater importance causally in the process in humans. It is very frustrating to be unable to find these studies.
To document this, I will provide links and screenshots to many of the relevant studies. I have also looked through many others, but been unable to find any study that profiles and quantifies ALL of the lipoproteins in actual human atherosclerotic lesions.
Commonly cited studies, one.
Commonly cited studies, two.
Commonly cited studies, three.
Commonly cited studies, four.
Commonly cited studies, five.
Commonly cited studies, six.
We’re supposed to believe that this establishes apoB as a primary causative agent of CVD on the basis of this data? No thanks.
Let’s look at some data and try to make sense of these, from the ground up. Once done, we can run the major CVD paradigm and show Lp(a)’s role.
Here is a panel from 2016. ApoA1, apoB, and apoE in progressively more damaged brain aneurysms. (link)
The top panel is negative control, without antibody. Second is minor damage. Third is moderate. Fourth is severe. Bottom is intracellular staining. Let’s look at some notable features.
First, in even mildly damaged walls, there was some aggregation of apoB (LDL, VLDL) and apoA1 and apoE. The text says apoA1 is associated with HDL. (Also associated with VLDL.) Also that apoA1 aggregates because apoE interacts with endothelium.
This seems to belie claim by dominant paradigm that LDL is initiator of atherosclerotic process, at least in brain aneurysms. apoA1, an HDL marker, and apoE, a VLDL marker, are there from the beginning. So how can we know apoB starts everything?
Most lipoprotein classes seem to be in the vessel walls, in brain aneurysms, from the very start of the process. This is not what I thought I would be reporting in this part of the thread, but it is what I have found. Now, what do the other papers show?
Here is another paper, showing co-localization of apoE with an important proteoglycan biglycan. Proteoglycans are molecules that are also thought to “trap” apoB-containing particles, as we will discuss in a moment. For now…
What does this paper say about this finding? It says it doesn’t understand–maybe this apoE is macrophages?–so it will do some more experiments. Let us take a look at these.
What do these experiments show? This:
apoE overlaps with apoA1, a marker of HDL, but not with macrophages. This suggests that the apoE is coming from HDL aggregation in the vessel walls. Ouch.
With still more images, the paper shows the same thing again in some very severe lesions. Overlap between apoE, apoA1, apoB, and biglycan staining, but not macrophages. Again, this implies that HDL may be atherogenic.
The researchers then decided to try to quantify their findings. To be clear, this paper examined 68 arterial segments from 14 patients. And 20 more from an additional 8 patients.
So for the quantification, they took images from…
These 68 segments and divided the images into quadrants for a total of 256 quadrants. (16 were not suitable for analysis.)
And 80 images from the 20 segments from the additional 8 patients.
What did they find? (A-first set of pts; B- second set.)
As we can see: biglycan, a proteoglycan that attracts apoB and apoE is laid down in 100% of lesions. apoE is present in almost 100% of lesions; apoA1 in 100%. And apoB, the initiator according to the dominant paradigm, is only present in 90%. Shit.
What’s more, apoE and apoA1 are present in nonatherosclerotic lesions, whereas apoB isn’t. This finding is a bit more ambiguous, since it either implies that apoE and apoA1 are “normal.” Or apoE and apoA1 are more important than apoB in initiation.
This is not something I wanted to find, since it throws my theory about Lp(a) out the window, unless I can figure out a way to save it.
Nonetheless, let’s follow these findings. Perhaps they will turn around.
Next, investigators wanted to test whether it was… apoE or apoA1 that bound biglycan, specifically in the HDL3 subfraction of HDL. So they mixed together increasing concentrations of HDL3 with radiolabeled biglycan. In another tube, they mixed HDL3 (with apoE removed), with radiolabeled biglycan.
The found that without apoE, HDL3 could not bind biglycan. With apoE, the HDL bound more and more biglycan. This showed that it was the apoE in HDL that was binding to the biglycan.
Next, they did the same experiment with just LDL. Increasing concentrations of LDL mixed in different tubes, run on a gel. Yep, the LDL binds to the biglycan.
In other words, could the same basic process that makes LDL atherogenic also be making HDL atherogenic?
This also opens up the possibility that through apoE, virtually all other classes of lipoproteins could be retained in the subendothelial space in precisely the same way that apoB is thought to be. This includes not just HDL and LDL, but VLDL too.
In line with this: another study with immunoblots for these same proteins fractionated from human atherosclerotic lesions via sepharose 6 gel filtration column (i.e. proteins in the lesions were separated by density). What was found? All present.
Nonetheless, the largest observational study evaluating the role of apoE in CVD risk found no association between plasma apoE and CVD risk.
Is apoE in the atherosclerotic plaque a mere epiphenomenon, i.e. not a cause but a consequence of CVD? (link)
And are we to make that judgment on the mere basis of the observational data?
These are questions to keep in mind moving forward.
In any case, it’s time to jump into response-to-retention model, the mainstream view of CVD. As has been alluded to, according to RTR, apoB-containing lipoproteins being retained in the subendothelial space is the initiating event in CVD. (link)
What’s the subendothelial space? It’s the space right under the first cell layer of the blood vessels. When atherosclerosis occurs, it’s not by “clogging”. It’s by pushing the first cell layer upward into the formerly empty blood vessel space, as the image below shows.
The key idea behind RTR is that to reduce CVD risk, people should reduce blood apoB. The fewer apoB-carrying lipoproteins, the lower the risk they will become “lodged” in the subendothelial space and spark runaway inflammation and atherosclerosis.
Again, this is supported by studies showing apoB has an advantage over LDL concentration in predicting CVD risk. While LDL-C is a measure of total cholesterol held by LDL in blood, apoB is a direct number of apoB-containing particles.
(Thanks Fib Co)
ApoB are thought to be particularly atherogenic because, as we mentioned earlier, apoB are prone to being “retained” in the subendothelial space. (link)
(See top right of this figure.)
Proteoglycans are basically tissue glue. They keep tissues together. They can also attract proteins for binding. apoB has multiple binding sites–as many as three, and according to some sources, four–for binding subendothelial proteoglycans.
(Whether the subendothelial proteoglycans are designed for attracting apoB, or if this is merely a coincidence, is an important issue that we shall discuss later.)
Now, why are apoB retained? Researchers believe, as mentioned, that concentration of apoB is important–simply due to increased chance to be retained, over time–but also that some people’s endothelium is more prone to retention than others’.
As a brief aside, 97% of diabetics are dyslipidemic. 56% have LDL >130 mg/dL. This is because the metabolic dysregulation of diabetes affects not just blood glucose but blood lipids. (link)
As mentioned, increased apoB increases the risk of CVD. Diabetics have substantially increased apoB, independent of LDL-C. (link)
Diabetics also usually have high blood pressure, which causes chronic stress on the endothelial cells of the blood vessels and damage, and thus predisposes the blood vessels to retention of lipoproteins (especially, according to RTR, apoB).
Finally, as discussed, diabetes causes endothelial cell dysfunction, and predisposes the endothelium to damage and retention of apoB. The molecular mechanisms are complex and beyond the scope of this thread, but here’s a famous article on the subject.
Any chronic or unrepaired damage to the endothelium probably increases the retentive properties of the blood vessel walls and thus predisposes to CVD. This includes smoking and probably many types of environmental toxin exposures.
A controversial issue in the field is whether one is still at elevated risk for CVD after reversing endothelial dysfunction by normalizing metabolic function, yet maintaining a high LDL-C and apoB, such as can occur in some people on low carbohydrate diets.
Here are some images of classic atherosclerotic progression from a 2007 autopsy study. As can be seen from the middle and right-side vertical panels, lipid accumulation kicks off the process, and then macrophages infiltrate in response. (link)
The paper grades atherosclerosis 0 to 3, where 0 shows diffuse intimal thickening (thought normal), 1/2 show fatty streaks with extracellular lipids colocalizing with biglycan/decorin (proteoglycans), and 3 shows foam cells (lipid-laden macrophages).
Again, it is thought that apoB-associated lipoproteins are retained in the subendothelium and accumulate into the extracellular lipids in the images, and then macrophages and other immune cells enter to try to clean up the mess.
Worth noting: even Staffan Lindeberg pointed out that the diet known to best reverse CVD was Ornish’s. Suggesting, a sufficiently low fat (and apoB) can allow the immune system to do what it seems to be “trying” to do on histological examination. (link)
Interestingly, the other trial Staffan discussed that achieved some success also attempted a rather severe reduction in fat (<20% energy), though not at much as Ornish.
Do we know Ornish’s patients’ plasma apoB levels? We do.
Reduced 25% at one year, similar to total cholesterol.
LDL’s reduction was even more dramatic, to nearly half of that of the control group. (link)
Interestingly, at 5 years Ornish’s patients had returned to baseline apoB, yet continued to see apparent atherosclerotic lesion regression. This suggests that Ornish’s patients were probably less adherent at this time-point.
Yet the patients still continued to see improvement despite this. It is not clear why.
The only biomarker that remained depressed at 5-years was apoA1. This is somewhat surprising, since apoA1 is a marker of HDL, and HDL is supposedly cardioprotective (now thought not so).
Why should HDL go down? An earlier study addressed this, noting that while in standard populations, HDL was associated with protection from CVD, in traditional populations, HDL is remarkably low.
This makes great sense, since it implies energy intake is simply lower in such populations (thus, less apoA1 in chylomicrons [cuisine] are necessary). This is consistent with the “more adherent” period of Ornish’s study, which saw >20lb weight loss.
But how can we explain Ornish’s findings of continued improvement in stenosis despite a loss of adherence? We might appeal to a core idea in the response-to-retention model: decreasing reversibility with progression of atherosclerosis. (link)
The orange arrow indicates reversability. As we see, it gets smaller over time. This suggests that the more normal the vasculature, the less drastic the intervention necessary to see improvement in cardiovascular health.
Well, when does atherosclerosis start?
1951 autopsy study of American and Korean soldiers killed in war: >30% of American soldiers (mean age 22) had significant atherosclerosis, and >70% had fatty streaks or fibrosis. Koreans rarely had atherosclerosis, only (normal) arterial thickening. http://epi.umn.edu/cvdepi/study-synopsis/korean-soldiers-study/
1997 autopsy study of spontaneously aborted or premature newborns that died within 12 hrs of birth: majority had fatty streaks. Signif more in children born to moms with high chol. White bars: normal chol. Lined: temporary high chol. Gray: high chol.
Huge difference in total area with fatty streaks in fetuses between normal cholesterol mothers and hypercholesterolemic mothers.
How do retained apoB and other lipoproteins cause these problems? In a word: oxidation. In fact, oxidation of LDL retained in the subendothelial space occurs readily and rapidly. It seems to be a normal physiological process. (link)
Most oxidized apoB-containing lipoproteins are rapidly removed from the plasma. Retention seems to precede oxidation.
With one exception.
Some researchers believe many of those apoB lipoproteins in the artery wall are not LDL but Lp(a).
Here is an image of apoB colocalizing with apo(a). Panel a is apoB. Panel c is apo(a). (link)
Indeed, Lp(a) seems to be the preferential carrier of oxidized phospholipids. Remember that phospholipid shell on that lipoprotein particle? That outer covering? Well, when those phospholipids are damaged, guess who picks em up? Lp(a).
Just after percutaneous coronary intervention (PCI), where narrowed blood vessels are re-widened by a physician, Lp(a) and oxidized phospholipids spike in the blood. Only 50% of that associated with apoB is associated with Lp(a) at 1hr…
The other 50% is associated with the other apoB proteins. At 6hrs, all oxidized phospholipids are now on the Lp(a). That’s because Lp(a) has the ability to soak up oxidized phospholipids in the circulation. That seems to be its job.
That, and a lot of other things, as we’ll see. But Lp(a) takes up so much of the phospholipids from oxidized LDL in the blood, that measurements of oxidized LDL are nearly synonymous with Lp(a). (link)
We can see the close correlation between oxidized phosopholipids per apoB (oxPL/apoB) and Lp(a) levels. (link)
Here is a figure that demonstrates the close correlation between oxPL/apoB and Lp(a) with crystal clarity. (link)
And it has been found that those in the top tertile of apoPL/apoB have a 2.4x risk of CVD, tracking Lp(a) closely.
Here is a biochemistry-oriented study showing much the same thing, using density gradient ultracentrifugation. You can see that the apoB, apo(a), and oxPL are all sitting in the same fraction. (link)
Here are some with evidence for protein-protein interactions, using Western blotting. Note only the lanes with apo(a) show staining for OxPL (left).
We have established that oxPL on apoB (oxPL/apoB) correlates with Lp(a). What about oxPL/apo(a)? If Lp(a) concentration is driving oxPL number, we would expect to see a close relationship between Lp(a) concentration and oxPL attached to Lp(a). We do.
This suggests that the concentration in the blood of oxPL associated with LDL or LDL-like particles is entirely determined by Lp(a).
This in turn implies that the claim by bloggers (whose names I forget) that polyunsaturated fat intake results in oxPL oxidation–IS ENTIRELY THEORETICAL. oxPL associated with LDL or LDL-like particles is almost entirely determined by Lp(a), i.e. by genetics.
This is why one should base conclusions on evidence, not speculations. Speculations are notoriously prone to bias. Speculations are great for generating hypotheses, theory-crafting, or arguing with a buddy, but foolish to base actual health recommendations on.
In fact, things are worse than this. I decided to actually see if anyone had done a study comparing different meals with different types of fatty acids, and then measuring Lp(a).
l- linoleic (polyunsaturated)
o- oleic (monounsaturated)
p- palmitic (sat)
s- stearic (sat)
And we see the biggest increase in Lp(a) with… wait for it… saturated fat! Why should this be?
By the way, the authors claim that the difference between meals is statistically significant.
So why the difference? Remember that Lp(a) preferentially binds up all the oxidized phospholipids in the body. When do oxidized phospholipids get generated?
In response to oxidative stress, i.e. damage.
Now it turns out that Lp(a) may be an acute phase reactant. What’s that? It’s a blood molecule that rises in response to injury, stress, or illness. (link)
We can see that Lp(a) behaved exactly like an acute phase reactant in this study (looking at blood markers just after myocardial infarction), where we see an acute rise in acute phase reactants… and Lp(a), apparently also an acute phase reactant.
Now I won’t claim that this means that saturated fat causes stress or damage to the body, though there are plenty of short-term studies that purport to show exactly that: an increase in inflammatory, acute phase blood proteins in response to sat fat intake.
But post-prandial inflammation is a well known phenomenon, anyway. Eating is a stress, and it seems to show that saturated fat is a stress, too. (It also shows the lowest response from trans fat, so who knows? Trans fat also uniquely does not induce autophagy.)
But the point here is two-fold:
- Lp(a) and by extension oxPL/apoB has zero to do with eating polyunsaturates (possibly the reverse);
- Lp(a) is an acute phase reactant, i.e. a response to bodily injury.
Now: Lp(a) sequesters oxidized phospholipids AND responds to bodily stress.
Are we starting to see a pattern? Let’s take a quick look at how Lp(a) gene expression is regulated. One thing we know is that IL-6 increases it. (link)
Moreover, aspirin reduces it. Now it’s not so simple–TNF-alpha seems to reduce it. But Lp(a) is certainly involved in inflammation.
The question is why? Before answering that, let’s look at some other interesting facts.
VLDL production in rats is dramatically upregulated in response to TNF-alpha. And relative to control, the LDL/HDL ratio seems to have spiked later on.
If we WAAAY overextrapolated from this data, the association with a low LDL/HDL ratio with low CVD risk, yet without being able to directly modify HDL to improve that risk, would make total sense. (HDL would be a marker of inflammation, not a drive of pathology.)
But we’re not ready to overextrapolate from the data like that yet. But let’s look at some more relationships between inflammation and lipoprotein changes…
By the way, TNF alpha is an inflammatory signaling molecule, if this was missed. In any case…
To start, inflammation causes drastic increases in VLDL and triglyceride levels, as we just saw. The above study has been repeated. (link)
Similar situation with cholesterol. (link)
And with HDL-associated particles. (link)
In all, if I had to summarize the contents of this review, it would go something like this.
But seriously. Lipoproteins can smother invading microbes. They can suck up microbial toxins and inactivate them. (link)
In fact, lipoproteins express and can upregulate a suite of proteins to decrease the toxicity of microbial toxins.
Lipoproteins can engulf viruses and inactivate them.
There is even quite a substantial literature on the role of lipoproteins in CUTTING APART (lysing) invading parasites.
And mice with lots of LDL cholesterol? Great at fending off malaria infection.
Infection is associated with injured cell membranes; and repairing these membranes requires cholesterol.
Inflammation turns LDL from a friendly cargo delivery truck to a KILLING MACHINE that oxidizes itself–and everything around it–to death.
From the above, let’s extract some lessons with regard to some more specific changes in lipoprotein composition in the blood.
Many studies have found an increase in small, dense LDL in response to inflammation and infection (“bad”). These more easily get trapped inside the subendothelial space and are more susceptible to oxidation and being taken up by macrophages.
As for oxidation, the more oxidation, the better the uptake of the lipoproteins by macrophages. So all that gunk being trapped by those lipoproteins? It’s going to be easier to suck up.
But why should lipoproteins become easier to oxidize? Because when they’re in subendothelial space (retained there for whatever reason), they’re in a particularly oxidizing environment. Anything that gets in there should die. It should be a killing zone.
And so it is. And so a lipoprotein that gets into a killing zone probably belongs there, and it should be oxidized, and then sucked up by a macrophage after it’s done its job. That’s my teleological argument, anyway.
Update 2020/06/20: Today I cannot find any good evidence that Lp(a) protects against infections. The most up to date meta-analyses suggest that it does not. This may be because the nature of present-day infections is different from the nature of infections in the ancestral past in which Lp(a) evolved: it may be that the pathogens for which Lp(a) is the response. If therefore Lp(a) does serve an immunologic role, this role may be so small in infectious-disease-environmental conditions that the effect is not detectable. If this is the case, it does not mean that Lp(a) never had such a role. However, it does suggest that this role is no longer relevant, on average, to modern health outcomes. This suggests that the consensus view–that lower Lp(a) is better than higher, all else equal–is correct, until more data emerges that can question this view. Most recent literature supports this view, with a new paper showing that genetically lower Lp(a) levels associates with lower mortality from all causes.