Wednesday, August 21, 2019

The Inversion Pattern: Cholesterol Code or Cholesterol Slowed?

Dave Feldman is a prominent voice in the low-carb community. He is a software engineer who decided to try a low carbohydrate, ketogenic diet (KD) as a nutritional strategy to regain control of his health when he was diagnosed as being borderline pre-diabetic in 2015 [1]. However, this led to an increase in his low-density lipoprotein cholesterol (LDL-C) that left him concerned about his heart disease risk. He then decided to have his LDL-C measured more regularly than most for an extended period of time, even once per day for a while. He was also tracking his fat intake and was able to compare this data to his LDL-C data. From this he discovered that for which he is arguably most famous⁠— the Inversion Pattern [2]. His LDL-C was inversely correlated with his fat intake. His LDL-C on average was higher than before, but it dynamically responded to the fat in his diet.

He infers that his LDL-C is inversely tracking his fat intake because it is a reflection of energy distribution. That is to say that the more fat we have coming in from our diet, the fewer very low-density lipoproteins (VLDL) we have being secreted by our liver. He remarks that he's seen this before as a software engineer. This pattern resembles a network to him—a distribution network of objects that dynamically respond to inputs. He suggests that being fat-adapted allows us to uniquely tap into this energy distribution network [3]. Perhaps if we're "powered by fat", the average bump in LDL-C experienced by some on low-carb diets might not be a sign of a pathological state after all. Additionally, perhaps eating fat isn't so bad if eating more of it reduces LDL-C. Not a totally unreasonable hypothesis, in my opinion. However, I'm not at all persuaded by this. Let's explore why.

Before we dig into the Inversion Pattern, we need a crash course in basic lipid metabolism. This means we need to understand how the body moves lipids from one compartment to another, and how they're eventually broken down to make energy. Since every lipid in your body got there as a result of something you shoved in your mouth, let's start at ground zero: your diet.

You just ate a glob of fat. What happens now? Aside from the heart palpitations and anxiety you might experience if you're a fan of typical Western dietary guidelines, of course. Well, the fat hits your stomach and signals to the gallbladder to release bile acids (BA), phospholipids (PL) (also known as lecithins), and free cholesterol (FC). These are all components necessary to capture the dietary fat in a self-assembling structures called mixed micelles (MM). The MM is composed of, you guessed it; BAs, PL, FC, and dietary fat.

There is an uncountable number of these structures when you consume fat, and the sum of all of them forms what is known as an emulsion. An emulsion is a sticky substance consisting of both hydrophobic and hydrophilic components. Have you ever made mayonnaise at home? It requires a liquid fat and an egg. That's it. The egg contains PLs and CE, FC, and the fat contains, well, fat. You blend them together, and it produces an abundance of these self-assembling lipid structures. The sum of which is an emulsion. The same thing happens in your small intestine when you consume dietary fat. Your gut basically turns into a mayonnaise factory.

The MM migrates to your enterocytes. These are the cells that comprise the surface of the small intestine. The MM encounters a layer of higher acidity between the intestinal lumen and the enterocytes. The MM is readily disassembled by a series of proteins and its components are absorbed. The dietary fat and PLs are broken down by lipases and is taken up by transporter proteins. FC and some BAs are absorbed through the Niemann-Pick C-1 like 1 transporter protein (NPC1L1). Excess FC (and BAs that were accidentally taken up) are ejected back out into the lumen by ATP-binding cassette sterol transporters, G5 and G8 (ABCG5/G8). However, up to 95% of BAs are taken up by another portion of the intestine called the ileum and are taken to the liver via the portal vein. The net effect is that most of the dietary fat and PLs are absorbed, a regulated amount of FC is absorbed, and almost all BAs are eventually absorbed.

At this point, all of these lipids are reconstituted through a series of metabolic pathways that aren't important to understand. But, essentially the enterocyte will take in the materials from the MM on the lumen side of the cell and will use those materials to rebuild a similar structure on the basal side of the cell called a chylomicron (CM). A CM is similar to a MM, but it is in many ways very different. A CM is a lipoprotein (Lp). Lps are different sorts of structures. They contain special proteins called apolipoproteins (Apo) that play regulatory roles and help facilitate Lp metabolism in different ways. There are about a dozen or so Apo proteins that have been identified and characterized. Only a handful of them are relevant to this discussion, but we'll come back to that.

The CM is essentially a spherical layer of PLs, FC, and Apo proteins shrouding a core of triglycerides (TG) and cholesteryl esters (CE). The CM is secreted from the enterocyte into the lymphatic system. From there, it travels upward and is released from the thoracic duct into general circulation. This is where the CM's journey really begins. As it travels through the blood, it interacts with various tissues and delivers its main cargo, TGs. The CM will leave a little bit of TGs in the fat tissue, in the muscle tissue, the cardiac tissue, organ tissue, etc. Any tissues requiring energy at that moment will have a fair crack at the TG content of the CM. The way these tissues take up TGs is through interacting with one of the Apo proteins on the CM, apolipoprotein C-II (ApoC-II). ApoC-II interacts with an enzyme found on tissues all over the body called lipoprotein lipase (LPL), this enzyme acts like a drinking straw, using ApoC-II to pull TGs out of the CM.

After the CM has made the rounds and delivered the bulk of its TGs to peripheral tissues (tissues other than the liver), the change in its TG content causes the CM to shed its ApoC-II. This reduces the CM to a chylomicron remnant (CMR). It is now unable to further donate TGs directly to peripheral tissues. However the CMR is still able to trade TGs from high-density lipoproteins (HDL) in exchange for CE with the use of cholesteryl ester transfer protein (CETP). It is also hydrolyzed by hepatic lipase (HL). However, these are not major fates of CMR-TGs. This is because the loss of ApoC-II facilitates the binding of apolipoprotein B-48 (ApoB-48), a major structural protein of CMs and CMRs, with apolipoprotein E (ApoE), another structural protein that is present on nearly all Lps. This then makes ApoB-48 and ApoE a single ligand for the low-density lipoprotein receptor (LDLr) on the liver. LDLrs are responsible for pulling CMRs, and a few other Lps out of circulation. The CMR only has the opportunity to donate any of its remaining TGs to HDL or HL if hepatic LDLr expression is diminished. But, even if those were major pathways for CM-TG, those pathways are carrying TGs to the liver anyway. Nevertheless, pretty much all CMRs are TG-rich upon reaching the liver.

Once the CMR is inside the liver it is catabolized and all of its contents are released. The TGs now have one of two fates. Either they're burned for energy or exported from the liver back into circulation. Just to keep the story moving, let's assume the TGs are being released back into circulation. This occurs similarly to how TGs are exported from the enterocyte. The TGs are packaged up into a Lp called a VLDL. This Lp is a little more complicated than a CM. It has a long and complex life cycle. It has its own specialized Apo proteins and interactions with tissues in a very different manner. However, the means by which the VLDL sheds its TGs is identical to that of the CM.

The VLDL traverses the blood, and the ApoC-II interacts with LPL all over the body to deliver TGs. Once the VLDL sheds a certain amount of TGs, it sheds its ApoC-II and becomes an intermediate-density lipoprotein (IDL). An IDL is very similar to a CMR. It is TG-rich, lacks ApoC-II, and it can also be taken up by hepatic LDLr the same way as the CMR. The IDL has three possible fates. Either it will lose its TGs and become a low-density lipoprotein particle (LDLp), or it will be catabolized by the liver. At least half of IDLs are pulled out of circulation by the liver. The other half will be hydrolyzed by HL or CETP, lose their TGs and thus become LDLp.

Now we have ourselves a bunch of LDLp. LDLp are primarily responsible for carrying out reverse cholesterol transport (RCT). Contrary to Dave Feldman's claim that the primary job of LDLp is to deliver fat-based energy, LDLp are actually TG-depleted. They collect cholesterol (CL) from HDL through CETP and help HDL carry CL back to the liver. LDLp have many potential functions, but generally LDLp have two possible fates. They can be taken up by the liver via LDLr, or they can be taken up by peripheral tissues. Only about one fifth of LDLp are taken up by peripheral tissues, and the remainder are taken up by the liver. Once in the liver, the LDLp are catabolized and their contents are released. Most of the contents are FC and CEs.

The FC and CEs can then be metabolized to FC as well as BAs⁠—ready to emulsify more dietary fat. Now we've come full circle. We've gone from dietary fat metabolism to Lp metabolism and back again, and observed a bird's eye, macro-level view of the entire pathway. Unfortunately, we're not quite ready to reconcile this information with the Inversion Pattern just yet. I mean, we are. At least I feel as though we are. But I know I won't be able to escape certain baseless criticisms unless I explore nutritional ketosis. So, let's take a moment to review the basics of the fat-burning metabolism and ketogenesis. Don't worry, this will be brief.

When we consume glucose, there are generally three different things we can do with it. First, to whatever extent we can we replete our glycogen⁠— a storage form of glucose similar to starch. This glucose is stored primarily in our liver and muscle tissue. Secondly, if glycogen is replete we burn the carbohydrates (CHO) as fuel. Lastly, if we are in a calorie surplus and can't burn the glucose, we can convert it to TGs through a process called de novo lipogenesis (DNL) [4].

Let's focus on liver glycogen, since this is the crux of ketogenesis. Glycogen is used by the liver to maintain euglycemia between meals and during sleep. When we sleep, we are fasting. There is no glucose coming into the body from outside, so the body must rely on the liver to stabilize blood glucose during this time. But let's say we decide to replace nearly all of the of CHO in our diet with dietary fat. This type of glucose-deprivation causes hypoglycemia, and the liver is called upon to defend our blood glucose until we run out of liver glycogen. Falling blood glucose signals to the pancreas to release glucagon, which acts on the liver to hydrolyze its stored glycogen and release glucose into general circulation.

However, this counter-regulatory hormone can only do so much. As glucose-deprivation continues, soon the liver's glycogen gets so low that glucagon is no longer able to maintain blood glucose by itself. When liver glycogen is sufficiently depleted, the adrenal cortex is called upon to secrete cortisol. Cortisol acts on skeletal muscle to liberate particular amino acids called gluconeogenic amino acids. These amino acids travel to the liver and make unique contributions to a process called gluconeogenesis (GNG). GNG uses non-glucose substrates to generate glucose. These amino acids are used to synthesize pyruvate and a number of intermediates necessary to the TCA cycle that would normally require glucose. In the initial phase of ketosis, most of the gluconeogenic substrates are amino acids from skeletal muscle and lactate shunted out of the cell by the deactivation of the pyruvate dehydrogenase complex (PDH). As time goes on the body can learn to rely on other substrates, like, glycerol from the adipose or the liver, and and even small aldehydes generated by normal metabolism.

The same loss of liver glycogen that leads to GNG also leads to ketogenesis. As blood glucose falls to a steady state, insulin also falls. As insulin falls, adipocytes release more free fatty acids (FFA). These FFAs traverse the blood stream, bound to a serum protein called albumin. They're taken up by tissues to provide energy when both glucose and insulin are low. But they're also taken to the liver along with dietary fat from CMRs. Once in the liver, these fatty acids (FA) and CMR-TGs are fed into the TCA cycle to generate acetyl-CoA, and then that acetyl-CoA is carried down through a pathway to generate acetoacetate and beta-hydroxybuterate. These are the water-soluble "ketone bodies" that enter the circulation and serve as an alternative fuel when glucose availability is low.

Well, maybe that wasn't very brief after all. Oh well. Let's continue. Just kidding. Now we have to discuss why ketosis is irrelevant to the Inversion Pattern. Believe me, I don't want to type this. But I will.

To understand what is really going on with ketosis and the Inversion Pattern, we have to understand three variables relevant to the liver: fat uptake, FA utilization, and TG export— fat in, fat used, and fat out. Ketosis will generally only independently affect one of these variables, FA utilization. Ketosis will conditionally affect TG export. Dietary fat will generally only independently affect one of these variables, TG uptake. Dietary fat will conditionally affect TG export as well. But, your overall energy status generally affects all three variables together. This is why I'm not convinced that ketosis is relevant to the Inversion Pattern. Especially as it relates to fat overfeeding, which we will discuss in a moment. Not only are you suppressing the amount of ketosis that is necessary to support energy metabolism, but the amount of FAs that contribute to ketosis is capped regardless of dietary fat intake.

Ketogenesis only requires somewhere between 4-44g of fat per day for the average person [5]. More if starved; less if fed. On a 2000 kcal ketogenic diet that is 75% fat, only 26% of that fat is committed to ketogenesis. So, even if we are very generous and assume that 100% of the FAs being used for ketogenesis are coming directly from CMR-TG (coming from the diet directly), that's still only 26% of the amount of fat one would typically eat on a ketogenic diet. This doesn't explain or contribute anything revelatory to understanding the Inversion Pattern.

For example, if you consume no fat, ketogenesis will be supported by FFAs liberated from your adipose tissue. If you consume a large amount of dietary fat, TG uptake from CMR will increase, and TG export through VLDL will increase, but you need not commit any of those TGs to supporting ketogenesis while those TGs are in the liver. That is determined by the liver's perception of whole body energy status at that moment. At best there will be a slight decrease in TG export if your fat intake is perfectly eucaloric. But a marginal decrease in TG export under eucaloric conditions isn't what is being explored by the Inversion Pattern. The Inversion Pattern demonstrates a decrease in LDL-C that is inversely proportional to dietary fat intake, with a correlation of approximately 0.9.

Because the conversion of CMs to CMRs is based on the absolute TG content, there will be no difference in the number of TGs reaching the liver on a KD when compared to a mixed diet with the same amount of fat. The only thing that will change is the time-scale involved. CMs will be converted to CMRs and then taken up by the liver faster on the KD under eucaloric conditions with fat clamped. The CMs will become CMRs and will be taken up by the liver slower on the mixed diet. However, total hepatic TG uptake from CMRs will be equal in both scenarios and have an equal area under the curve (AUC). As far as the liver is concerned the fat-in side of the equation is unaffected by ketosis. This means the burden of fat on your liver will be proportional to your fat intake. Ketogenesis is also not necessarily limiting for TG export, because FA utilization need not change as a function of TG uptake. But your liver will take up those TGs no matter what, and it will export some proportion of them no matter what. Ketosis doesn't matter here. Sorry, I just can't emphasize it enough.

We're powered by VLDL-TG regardless of ketosis or our CHO intake. Our overall energy status dictates just how much of our dietary fat is going to be retained by the liver to support ketogenesis. In a high energy state, the liver will commit fewer-to-none of those TGs toward supporting ketogenesis, and may even re-esterify incoming FFAs from the blood to form new TGs. In a low energy state, the liver will commit more of those TGs as well as incoming FFAs toward supporting ketogenesis. Sure, if ketogenesis is maximally suppressed with dietary CHO, you will have fewer TGs that entered the liver being committed to supporting energy metabolism in that way. Under eucaloric, fat-clamped conditions, you may have a slightly greater amount of TGs leaving the liver with the CHO-inclusive diet. This isn't relevant to the Inversion Pattern, which is why I didn't even want to bring ketosis into this discussion. It's just straight up irrelevant.

The state of ketosis will not effect the flux of dietary fat through the liver more than marginally in eucaloric conditions, and probably more if fewer fat calories are consumed. But ultimately you're looking at a relatively small change in either case. However, Dave Feldman's Cholesterol-Drop Protocol (CDP) (a short-term dietary protocol designed to maximally leverage the Inversion Pattern) encourages you to eat approximately 1.38kg of dietary fat, equating 82% of calories, in three days [6]. I think everyone can agree that this generates a hypercaloric state in most people. Further emphasizing just how irrelevant ketosis is to this discussion. Under these conditions, potentially less than 1% of dietary fat would be contributing to ketogenesis. Being "powered by fat" does not fundamentally alter the physiology of hepatic TG uptake or export, and claiming that it does is tantamount to admitting that you believe in magic.

Alright. Now we're ready to reconcile what we've discussed with the Inversion Pattern. As disappointing as it may be, this is likely to be the shortest part of this blog post. Here we go.

Let's talk about Dave Feldman's CDP. This protocol is designed to fully reveal the Inversion Pattern, and encourages one to eat an enormous amount of dietary fat. Dave Feldman suggests that dietary TGs downregulate the release of VLDL-TGs. He postulates this is because there is no need to mobilize stored fat as VLDL-TGs when you have an abundance of CM-TGs coming in from the diet. The CDP is meant to be the ultimate example of this principal in action— eat a massive amount of fat, hepatic TG export decreases, LDL-C naturally decreases in tandem. Or so the story goes.

Sounds compelling. However, this is not actually how the liver works, as we've discussed. With regards to dietary fat, the liver simply takes in TGs from CMRs, HDLs, IDLs, and LDLp to a lesser extent, and decides what the do with them. The liver exports some proportion of them (or all of them) back into circulation as TGs in VLDLs. That's actually it. You eat fat, your body takes what it wants, the liver gets the remainder, the liver takes what it wants, and sends the rest out into the blood. Simple.

But wait, this actually sounds like eating dietary fat should increase VLDL secretion. So, why does LDL-C tend to go down proportionate to the amount of fat consumed as seen in the Inversion Pattern? This is because the liver does not generally have the resources on hand to export very much dietary fat at a given moment. The more fat you eat, the worse it gets. In fact it can take many hours to mobilize all of the dietary TGs from the liver after a single high-fat meal [7][8]. I'm not saying KDs cause fatty liver. In fact, experiments have revealed reductions in hepatic steatosis among human subjects following KDs [9]. Not my point. Those findings can be explained by variables unrelated to nutritional ketosis itself. Adiposity and high energy status are overwhelmingly the largest predisposing risk factors for hepatic steatosis, and a mere change in protein intake can have huge effects on this. However, the KD is a useful weight loss tool in free-living humans, and animals, due to its effects on appetite. But what happens if you deliberately overeat fat on a KD?

Let's refer back to earlier in our discussion. When we eat fat, the gallbladder secretes an assortment of lipids that are essentially raw materials for building MMs and Lps. These raw materials include: PLs, FC, and BAs. The PLs are mostly reabsorbed, the FC is only partially reabsorbed, and the BAs are almost completely reabsorbed in the ileum of the small intestine. However, these are under normal conditions.

Eating over a pound of fat per day is hardly representative of normal conditions. In fact, eating more than 77-83g of fat per day would be considered abnormal by national standards [10]. Under normal conditions BAs are typically reabsorbed, sure. But, it has been well documented that merely feeding a high-fat diet (an extreme condition from my perspective) can significantly increase bile acid loss in the stool [11][12][13][14][15][16][17].

I surmise that this is actually the cause of the Inversion Pattern. The rapid BA excretion and loss of BA turnover leads to hepatic BA and CL depletion. BAs are synthesized from CL. In this state, the liver is forced to work overtime, doubling down on CE and FC synthesis to support VLDL production as well as BA production. Every time you eat dietary fat, you lose a little CL as BAs, as well as FC, in the stool. The effect is highly sensitive to the amount of fat in the diet. It is even sensitive to a particular high-fat meal, as the body can often need several cycles of enterohepatic BA circulation to handle just a single high-fat meal. 

But, this isn't a single high-fat meal under normal conditions. This is balls-the-wall-inhale-over-three-blocks-of-butter-in-three-days level fat overfeeding here. There is no reason to believe that the established effect of BA loss due to high-fat feeding wouldn't hold true at even higher levels of fat intake. Keto isn't magic. It is highly likely that you're going to lose more BAs in the stool when you do this protocol.

In essence, the Inversion Pattern is likely a reflection of hepatic BA and CL depletion, which means the CDP is actually a BA/CL-depletion protocol first and foremost. The Inversion Pattern is not correlating with energy distribution as Dave Feldman sees it. It's correlating with the liver's struggle to handle the fat burden of the diet within a given period of time. The liver is not recruiting fewer TGs from storage in response to incoming dietary fat as Dave Feldman suggests. The TGs from the diet are the first ones on the scene in the liver, and that amount of fat can back up hepatic lipid metabolism such that LDL-C drops in tandem. This is a very easy mechanism to understand, it is well-documented, and as such it requires very little speculation.

But, let's hammer this point home a little more. Let's take a look at Dave Feldman's TGs compared to his fat intake. 

Notice how his TGs have a poorer correlation with fat intake than does his LDL-C? 

Interesting, considering that his argument is that the Inversion Pattern represents the inverse relationship between dietary fat and fat recruited from "storage" being mobilized from the liver as VLDL-TG. Presumably this graph is a representation of fasting TGs held primarily in VLDLs, IDLs, LDLs, and HDLs. Pretty odd if you ask me. He even remarks that this is a disappointing finding in his article. Probably because it cuts against his hypothesis so sharply. Why he maintains this hypothesis in light of this is a mystery to me. If his claim is that the Inversion Pattern is tracking energy (not necessarily CL availability or synthesis), you'd expect the Inversion Pattern to track with TGs far more tightly than LDL-C. But it doesn't.

The truth is you can't manufacture VLDL in the liver without either CL or PLs, but PLs become limiting much sooner than CL. You can export enormous amounts of fat from the liver using very little CL, but you can't export ANY fat from the liver without PLs. Dave Feldman's protocol also involves eating 194g of protein per day. This is only 15% of calories, but it is a LOT of protein. In this blog post, I explain how protein is a source of methionine (and is often a source of choline), which the body uses to synthesize PLs. It is also likely that the diet itself is very high in PLs. As far as I can tell, Dave Feldman's protocol is designed to lower LDL-C by overburdening hepatic CL synthesis. Whether or not transient steatosis is also a contributor is likely anyone's guess at this point. Essentially, in high-fat overfeeding you may see TGs go up or down inconsistently. You may see HDL fluctuate. You may see LDLp go up or down. But, the actual blood CL itself maintains a tight inverse correlation with fat intake.

At this point someone might argue that Dave Feldman's protocol could actually be pretty high in dietary CL, so CL-depletion may not be a sufficient explanation for the finding. Maybe. But, CL-uptake in the gut is tightly controlled by your enterocytes. Not to mention that most dietary CL is esterified and is less bioavailable. I find it highly unlikely that exogenous CL contributes meaningfully to the hepatic CL pool in normal people. I mean, heck, isn't dietary CL's poor absorption rate a typical low-carb talking point anyway? However, there are subsets of people for whom I'd expect CL to be less limiting in this way. For example, people with familial hypercholesterolemia (FH) resulting from impaired ABCG5/G8 activity in the gut [18]. These people have accelerated and highly indiscriminate sterol uptake in the gut, and as a result maintain a state of hyperlipidemia. I would suspect that the Inversion Pattern would not hold true as tightly for these people.

Unfortunately, I could only find one study investigating changes in hepatic lipids in the context of high-fat overfeeding [19]. Doubly unfortunate⁠— it's a rodent study. The study was performed on a special kind of mouse called an ob/ob mouse. This is what is known as a "knock out" mouse. Basically, the researchers select genes they want to "knock out" and essentially deactivate them. This has different effects depending on the genes we want to knock out. In ob/ob mice, their leptin genes have been deactivated (another name for an ob/ob mouse is a "leptin-deficient" mouse). This causes them to overeat drastically. Since it is overeating we wish to investigate, they're a reasonable choice.

Essentially there were two groups of mice in the experiment. The control group were standard 6J mice fed an ad libitum (as much as desired) diet of either rodent chow or high-fat chow (60% of calories as fat). The experimental group was the ob/ob mice fed the same two diets. As predicted, the ob/ob mice overate both diets. They were killed at different time points and their livers were removed, frozen with liquid nitrogen, and their hepatic lipids were analyzed. Lipids were measured by molar weight, and were represented as percentages of hepatic lipid droplets.

What they discovered was that the ob/ob mice consuming the high-fat diet had significantly lower hepatic FC, CE, impaired VLDL secretion, steatosis, a minor decrease to hepatic PLs, and a higher loss of BAs in the stool. Not only this, but the ob/ob mice eating the high-fat diet had diminished HMG-CoA reductase (HMGCR) activity as compared to the ob/ob mice eating the control diet. This means the ob/ob mice eating the high-fat diet had impaired CL synthesis compared to the same mice eating the control diet. Lastly, the ob/ob mice eating the high-fat diet also had lower CYP7A1 activity as compared to the control mice eating the same diet. Which means that their BA synthesis was lower as a function of them having overeaten fat. Which is an illustration of the BA/CL-depletion I speculated about earlier. Essentially, what they're investigating here is the lipid milieu of the liver in the context of high-fat, diet-induced BA and CL depletion.

This is the only data I could find that investigated anything even remotely close to Dave Feldman's CDP. It coheres with my original interpretation of what is likely occurring in the liver during both the CDP protocol and the Inversion Pattern. High-fat diets may not necessarily lead to steatosis in human livers, but it is highly likely that both the BA pool and CL synthesis takes a hit from the sheer volume of fat overall. Keeping in mind that steady-state blood CL levels will be on average higher while on virtually any high-fat diet. But, in essence Dave Feldman's Inversion Pattern, as well as his CDP, merely reflect the liver's struggle to handle the burden of dietary fat over time. They are not a reflection of a dynamic energy distribution system that turns VLDL-TG export up or down like a faucet in response to dietary fat. Dave Feldman's model is reasonable if one lacks certain critical knowledge about lipid metabolism. But, ultimately the model as it is currently described makes very little sense, and it is completely at odds with the basic physiology of lipid metabolism.

PS. If you like what you've read and want me to continue writing, consider supporting me on Patreon. Every little bit helps! Thank you for reading!


[1] Cholesterol Code. About Dave Feldman.

[2] Cholesterol Code. Cholesterol Code – Part I : More Fat, Less LDL-C.

[3] Low Carb Down Under. Dave Feldman - It's About Energy, Not Cholesterol.

[4] Acheson KJ1, Schutz Y, Bessard T, Anantharaman K, Flatt JP, and Jéquier E. Glycogen storage capacity and de novo lipogenesis during massive carbohydrate overfeeding in man. Am J Clin Nutr. Aug 1988.

Patrycja Puchalska and Peter A. Crawford. Multi-dimensional roles of ketone bodies in fuel metabolism, signaling, and therapeutics. Cell Metab. February 2017.

[6] Cholesterol Code. Cholesterol Drop Protocol (“Feldman Protocol”).

[7] B. O. Schneeman, L. Kotite, K. M. Todd, and R. J. Havel. Relationships between the responses of triglyceride-rich lipoproteins in blood plasma containing apolipoproteins B-48 and B-100 to a fat-containing meal in normolipidemic humans. Proc Natl Acad Sci U S A. March 1993.

[8] Cohn JS, McNamara JR, Krasinski SD, Russell RM, and Schaefer EJ. Role of triglyceride-rich lipoproteins from the liver and intestine in the etiology of postprandial peaks in plasma triglyceride concentration. Metabolism. May 1989.

[9] Jeffrey D. Browning, Jonathan A. Baker, Thomas Rogers, Jeannie Davis, Santhosh Satapati, and Shawn C. Burgess. Short-term weight loss and hepatic triglyceride reduction: evidence of a metabolic advantage with dietary carbohydrate restriction. Am J Clin Nutr. May 2011.

[10] Susan K. Raatz, Zach Conrad, LuAnn K. Johnson, Matthew J. Picklo, and Lisa Jahns. Relationship of the Reported Intakes of Fat and Fatty Acids to Body Weight in US Adults. Nutrients. May 2007.

[11] Kasbi Chadli F., Nazih H., Krempf M., Nguyen P., and Ouguerram K. Omega 3 fatty acids promote macrophage reverse cholesterol transport in hamster fed high fat diet. PLoS One. April 2013.

[12] Müller V.M., Zietek T., Rohm F., Fiamoncini J., Lagkouvardos I., Haller D., Clavel T., and Daniel H. Gut barrier impairment by high-fat diet in mice depends on housing conditions. Mol Nutr Food Res. April 2016.

[13] Yoshitsugu R., Kikuchi K., Iwaya H., Fujii N., Hori S., Lee D.G., and Ishizuka S. Alteration of Bile Acid Metabolism by a High-Fat Diet Is Associated with Plasma Transaminase Activities and Glucose Intolerance in Rats. J Nutr Sci Vitaminol (Tokyo). 2019.

[14] Reddy B.S., Hanson D., Mangat S., Mathews L., Sbaschnig M., Sharma C., and Simi B. Effect of high-fat, high-beef diet and of mode of cooking of beef in the diet on fecal bacterial enzymes and fecal bile acids and neutral sterols. J Nutr. September 1980.

[15] Sakaguchi M., Minoura T., Hiramatsu Y., Takada H., Yamamura M., Hioki K., and Yamamoto M. Effects of dietary saturated and unsaturated fatty acids on fecal bile acids and colon carcinogenesis induced by azoxymethane in rats. Cancer Res. January 1986.

[16] Stephen J.D. O'Keefe, et al. Fat, Fiber and Cancer Risk in African Americans and Rural Africans. Nat Commun. April 2015.

[17] Sato Y., Furihata C., and Matsushima T. Effects of high fat diet on fecal contents of bile acids in rats. Jpn J Cancer Res. November 1987.

[18] Miwa K, Inazu A, Kobayashi J, Higashikata T, Nohara A, Kawashiri M, Katsuda S, Takata M, Koizumi J, and Mabuchi H. ATP-binding cassette transporter G8 M429V polymorphism as a novel genetic marker of higher cholesterol absorption in hypercholesterolaemic Japanese subjects. Clin Sci (Lond). August 2005.

[19] Arisqueta L., Navarro-Imaz H., Labiano I., Rueda Y., Fresnedo O. High-fat diet overfeeding promotes nondetrimental liver steatosis in female mice. Am J Physiol Gastrointest Liver Physiol. November 2018.


  1. Hi!
    What about the fact that fasting raises ApoB lipoproteins and LDL cholesterol? It would seem that this supports the energy distribution system hypothesis. Or do you also explain this phenomenon by saying that since the liver is not overflowed with dietary fat, it has the capacity to make more cholesterol and LDL particles?

    1. ApoB particles go up commensurate with a drop in blood glucose. You need energy. FFAs arrive at the liver to be burned or converted to ketones. The remainder leaves the liver in VLDL and TGs are distributed to other tissues. A big problem with Dave's model is that it is extremely vague and poorly articulated by him. I don't even think he understands it himself to be perfectly honest.

    2. That is a fair criticism. My profession is totally unrelated to lipidology or biochemistry of any sort. What I found valueable in Dave's model is that at its core it says that for a body that gets its energy needs primarily from fat (ffas), the increase of ApoB and LDL cholesterol is not a physiological dysfunction caused by the system being overwhelmed with fat above its capacity, but a logical adaptation. Of course that is no proof that this altered energy metabolism is not dangerous long term, but it is a welcomed respite from the dogmatic 'LDL is bad nomatterwhat' message by media and the medical profession.

    3. Well, LDL causes CVD, and we haven't really identified a lower threshold to benefit when it comes to LDL lowering. It really does appear as though the lower the LDL, the better.

  2. This is really great stuff. A lot of it went over my head, but it is really fun to hear detailed alternate explanations of high LDL on a HFLC diet. I do have one question though. Can you explain why in all the groups I've found of people on a "vegan keto" diet, barely anyone is talking about elevated LDL? The extremely few people that mention it are quickly bombarded by comments to remove coconut oil from their diet. I feel like there is something completely obvious that I am not thinking about...but why would there not be more people with high LDL on a ketogenic diet that was high in MUFA, but relatively low in SFA? It's possible that vegan keto people are just not talking about their high LDL, but I doubt that. What am I missing? Anyway, thanks for writing this blog!

    1. I'm working my way backwards through your blog. Is it the lack of choline in the vegan keto diet that is preventing high LDL?

    2. without saturated fat, in a diet high in MUFA/PUFA the LDL just gets cleared out of the blood faster? So a vegan keto person eating a ton of olive oil with every meal would have high LDL after every high-fat meal, but this would get cleared out of the blood during the fasting period before the blood test, giving a low LDL result? Am I on the right track? Or am I still completely off? Thanks! :)