Search This Blog

Tuesday, 29 April 2014

Saturated fat IS good for you, but how much polyunsaturated fat do you need?

Working on the "Real Food" AUT dietary guidelines submission it became clear that neither the epidemiology of the lipid hypothesis nor the RCT experiments testing it were designed to separate the supposed artery-clogging effect of saturated fat from the purported heart-healthy effect of polyunsaturated fat.
What we had was a series of comparisons of higher SFA and lower PUFA vs. lower SFA and higher PUFA. There were no investigations of either higher SFA in a diet with optimal PUFA, or of deficient PUFA in a diet also low in SFA.
The former of these, higher SFA in a diet with optimal PUFA, is what the low-carbohydrate, high-fat (LCHF) diet supplies. The latter is a danger on a low-fat diet, so "let them eat margarine".
1946 Dietary Guidelines - not bad at all.

Richard Lehman described the mindset in a recent BMJ blog
If I despair of meta-analyses, what can I say about dietary surveys? Actually, they’re quite fascinating, though they rarely bear any definable relation to human health. Diet and religion are often closely aligned. The authors of this paper have got religion bad: “Diet is one of the fundamental risk factors for health, disease, and disability in the world. Indeed, given that trends in metabolic risk factors such as blood pressure, cholesterol, glucose, and body mass index are being largely driven by nutrition, suboptimal diet is the single leading modifiable cause of poor health in the world, exceeding the burdens due to tobacco and excess alcohol consumption combined.” They then go on to look at fat. Why, I know not. But it is quite interesting to see which countries eat what types of fat. Where people eat more saturated fat, they often eat more unsaturated fat. For all I know this may help to explain why nearly everyone everywhere is enjoying their food more and living longer.
In the paper Richard Lehman is discussing, we find the alarming statement that the Centre for Disease Control (which really should be worrying more about emergent viruses and antibiotic-resistant bacteria and less about fat) thinks that getting 13% of our energy from linoleic acid (omega 6 PUFA) will be "optimal". If that's optimal, what do they consider an excess? This crazy figure is certain to be based on the amount of linoleic acid required to reduce serum LDL to a particular "healthy" target across a Western population, with no thought given to metabolic health, and bearing no relation to any intake humans might have been exposed to for any period in their evolutionary history.

As Peter D. said in a recent comments thread, "
You have a basic assumption that blood lipids cause CVD. You will make many wrong choices based on this hypothesis." Now, I don't know what causes CVD. It could be blood lipids - there are intelligent people who think that small, dense LDL particles typical of refined-carbohydrate diets rip up the arterial walls like stilettos on a ballroom floor. But I do know that the assumption that blood lipids cause CVD has led to many wrong choices. Wrong choices in the supermarket. Wrong choices in the design of experiments. Wrong choices in the interpretation of epidemiological data. Imagine what useful drugs statins might be if they were not over-prescribed on the basis of LDL counts, with results like this.

Many analyses of the "lipid hypothesis" epidemiology and RCTs failed to distinguish between omega 6 and omega 3 PUFA, or between vegetable PUFA (linoleic acid and alpha linolenic acid) and long chain polyunsaturated fats found in animal foods like EPA, DPA, DHA and arachadonic acid. This is understandable because, when Ancel Keys cooked up the lipid hypothesis, no-one knew that omega 3 fatty acids existed, let alone that they were essential nutrients. I'm not even sure if anyone realised that omega 6 was essential back then.
We are stuck with an article of faith from the Dark Ages. A diet high in saturated fats that happens to be deficient in polyunsaturated fats - say with fats supplied exclusively from tallow and hydrogenated shortening, with over-cooked boiled meat or deep-fried fish, milk, boiled greens, sugar, and highly refined grains - which was the mid-20th century diet for many in the Westernised world - can be deficient in PUFA. Smoking and drinking and the general lack of antioxidants in the diet, as well as the effect of insulin from the refined carbohydrate, will tend to squander what PUFA there is. Not surprising if this population has more CVD. What will be surprising is, if the saturated fat has anything to do with it. That diet will also be high in monounsaturated fat (beef olein is the MUFA from tallow, used in old-school chip shop fryers), but somehow MUFA gets a pass. It's all very unsatisfactory.

Is there any research that can help us sort out the difference between a lack of PUFA and an excess of SFA? Well, maybe. We can at least see the difference between an excess of PUFA, and a lack of SFA.
This study by Amin A. Nanji and Samuel French is an ecological study comparable to Ancel Key's "7 Countries" study, except that it includes 17 countries.
Mortality from cirrhosis in many countries deviates markedly from that expected for a given per capita alcohol intake. We investigated the possibility that dietary factors might explain the deviation expected and actual mortality rates in different countries. Deviations from expected cirrhosis mortality was calculated as a percentage for 17 different countries, all of whom had carrier rates for hepatitis B virus of less than 2%. The percentage of deviation was correlated with dietary intake of saturated fat, polyunsaturated fat, cholesterol, and also with mortality from ischemic heart disease. The percentage of deviation correlated inversely with dietary cholesterol (
r= -0.86, p 0.001) and saturated fat (r= -0.80, p 0.001) and positively with polyunsaturated fats (r= -0.55 p 0.05). This suggests that both saturated fat and cholesterol protect against alcoholic cirrhosis while polyunsaturated fats promote cirrhosis. The correlation between percentage of deviation and ischemic heart disease (r= -0.78, p 0.002) suggests that those factors that promote ischemic heart disease protect against alcoholic cirrhosis.

If the factors that promote IHD protect against alcoholic liver disease (ALD), what are they? 
We can perhaps discard the red herring of cholesterol, which is a marker for animal fat, and which indicates consumption of foods (such as eggs) which are also rich in the essential hepatoprotective nutrient choline. Cholesterol does protect against ALD tested separately, but coconut MCTs (no cholesterol, no PUFA) gives the same protection as tallow (cholesterol, 1% PUFA) when the two are compared.
There has been a large body of animal experimentation designed to elucidate the link between alcoholic liver disease and dietary fats, with occasional testing of acetaminophen (paracetamol) toxicity in the same model. These papers have been analysed expertly by Victoria Prince on her blog and I will not duplicate her efforts.
It can quite clearly be seen in Victoria Prince's third liver-and-lipids post below that there are mechanisms whereby saturated fats can have health benefits, distinct from the restriction of PUFA, because saturated fats from coconut and cocoa protect the liver in different ways.
Thus it seems that dietary MCTs work in a way that maintains the expression of gut tight junction proteins, preventing endotoxin from making it into the circulation, while long chain saturated fats work in a way that increases endotoxin-binding proteins in the liver.  Both prevent endotoxin-induced damage in the liver, but in very different and distinct ways." This is not just relevant to alcoholics and people who consume paracetamol; the fact that long-chain SFAs make the liver less sensitive to LPS is relevant to the studies showing serum markers of dairy fat intake are associated with diabetes protection, because liver inflammation is a precursor to diabetes. And the fact that MCT, made from fats you can only get in significant amounts from coconut or dairy fat, maintains the integrity of the gut, is relevant to the epidemiological studies showing that children who eat margarine and drink low-fat milk have a higher rate of allergies and asthma compared to children who eat butter and drink full-fat milk.
Saturated fats are good for you. Get over it.

The high fat model of alcoholic liver disease was developed because rats fed normal chow plus alcohol would not develop ALD. The researchers concluded that polyunsaturated fats are "essential for the development of alcoholic liver disease", and that saturated fats are protective.

Diets enriched with saturated fatty acids protect against alcohol-induced liver injury, whereas diets containing polyunsaturated fatty acids promote liver injury (Nanji and French, 1989Nanji et al., 19891994a). Saturated fatty acids have also been reported to reverse established alcoholic liver injury (Nanji et al., 19951996,1997b).

Now, the fats that promote liver injury are vegetable oils, typically in these tests corn oil and soy oil*. Thanks to the campaign against saturated fats, these are the kind of oils that the fast food eaten by drunken New Zealanders is prepared with today. This seems to be a failure in public health planning.

"You have a basic assumption that blood lipids cause CVD. You will make many wrong choices based on this hypothesis."
(*interestingly fish oil, which has a stronger association with CVD prevention than omega 6, also has a greater promoting effect on alcoholic liver disease, albeit in quantities many times greater than anyone would ever include in their diet. And deficiencies of PUFA, particularly AA, EPA, and DHA, also play a part in ALD, so Samuel French suggsts that alcoholics be advised to replenish these fats by eating "Mediterranean diet" type foods whenever they're in recovery).

The rat model of alcoholic liver disease demonstrates that in this case the effects of high PUFA, while they can be confused with those of low SFA, are not the same. It also demonstrates that the SFA/MUFA ratio of the non-PUFA fat makes little difference (olive oil is almost as protective as beef tallow against acetaminophen liver damage; it seems an oversight that olive oil has not been included in the alcohol experiments, considering how widely it is consumed).
The link between SFA and CVD disappears when SFA is considered separately ("monotonically"), rather than as a marker for PUFA deficiency. The protective association between omega 3 PUFA (the PUFA less likely to be optimal) and CVD remains about the same when these nutrients are considered monotonically, as in the 2013 Singapore Chinese Health study by Koh et al. People who ate more DHA, EPA, and ALA had significantly lower rates of CVD mortality - and they also ate more saturated fat, monounsaturated fat, and omega 6 PUFA, all increasing stepwise with omega 3 across the quartiles as CVD mortality decreased.
"Where people eat more saturated fat, they often eat more unsaturated fat. For all I know this may help to explain why nearly everyone everywhere is enjoying their food more and living longer." 
The inverse association between omega 3 and CVD (0.83 OR) is seen with total omega 6 intakes of, on average, 9.9g per day, or 4-5% of energy - almost exactly the same as healthy New Zealanders ate in this paper, and much less than the crazy CDC projection of 13%. 
All things being equal, the people who eat the most fat, assuming they are not eating solely hydrogenated industrial fats or exclusively low-PUFA fats such as tallow, but are instead getting some fats from wholefoods, including fish, are surely eating enough PUFA to have an optimal intake for CVD prevention, regardless of any effect on blood lipids.

Polyunsaturated fats are essential nutrients, but also highly reactive and bioactive molecules. This means that the association between intake and mortality will follow a fairly steep U-curve, with a "sweet spot" in the middle. Saturated fats are inessential nutrients with low reactivity. The association curve between intake and mortality is likely so flat my dog could skate it.

(not my actual dog)

PUFA, MUFA and SFA occur together in whole foods and in traditional fats and oils. They are not easily separated, even in the laboratory. You can buy pure sugar for a few dollars in the supermarket, it's easily separated from the foods it occurs in, usually to be ladled back into some other food in excessive quantities. You can probably buy purified long-chain saturated fats like palmitic acid from Sigma Aldrich if you have the money and inclination, and MCT is a processed form of medium-chain saturated fat that is sometimes supplemented by ketogenic dieters; I've never seen it, but I've heard it's expensive.
Most people eating a LCHF diet, even those who think "saturated fat is good for you", are not going to go that far. Few are even going to use butter or coconut oil as their only fats, and anyone eating a variety of wholefoods that have been made palatable but not cooked to death is likely to have an optimal PUFA intake, whatever that is.



Michael Frederik said...

I think this is your best piece yet; a great piece of writing. You are really coming into your own here.

Bill said...


Bill said...

"I'm not even sure if anyone realised that omega 6 was essential back then."

For what it's worth, I'm not entirely convinced that linoleate is even truly "essential." Rodent studies by Le and Ling have shown that EPA & DHA supplementation, with or without arachidonate, can mask any signs of linoleate deficiency.

George Henderson said...

Yes, that's interesting research.
Cardiolipin is a mitochondrial membrane lipid which incorporates LA, and cardiolipin oxidative status determines mitochondrial lifespan, so I am reluctant to call inessentiality on LA. But IRL any food that supplies AA, or EPA and DHA, will supply at least a little LA, and all except marine foods supply quite a lot.
(which makes me wonder why the Dietary Guidelines specify "lean poultry"; if they think omega 6 PUFA is the bees knees, there went a great source of it).
Fish oil is around 1-3% LA (18:2) according to this

George Henderson said...

If you're truly PUFA deficient the body makes Mead acid, a omega-9 PUFA, to compensate, but nowhere can I find that Mead acid is a marker for CHD. So it seems unlikely that CHD is mainly a PUFA deficiency disease. It's not listed in the symptoms of deficiency.

Ash Simmonds said...

Sweet, will have to add this to my Fats vs Booze vs Liver thread.

Ash Simmonds said...

@Bill "For what it's worth, I'm not entirely convinced that linoleate is even truly "essential.""--

Another case of applying precursors to outcomes.

I've got a feeling that a bunch of things we consider we require in certain amounts will change with further research into people dodging "modern" foods in the longer-term.

Most everything we base our food belief system on are on a correlative basis - founded upon a population of people who eat a "healthy balanced diet" - almost none of it is demonstrably causative/protective in all cases.

But it's delicious.

AlmondD'oh said...

Thank you for taking the time of updating such a well- kelt blog. I am rereading some of the old entries and there is a wealth of information on here.

In regards to the "old" dietary guidelines, I couldn't but smile when I see the line "peanuts occasionally". Nowadays legumes have taken precedence over animal proteins. What a change.

AlmondD'oh said...


George Henderson said...

I feel I should explain why, if PUFA is a hazard for alcoholics and should be avoided, PUFA deficiency needs to be addressed too:
linoeic acid and alpha linolenic acid are both peroxidised and converted to toxic metabolites in the presence of alcohol, causing an eventual deficiency of AA and DHA (which would also be toxic at high intakes). AA and (especially) DHA are hepatic fat-burning signalling molecules that prevent steatosis, and allow normal immune function etc, and phospholipids (which are highly unsaturated) are also protective in the absence of alcohol. So the deficiency should be made up when not drinking, from fish, nuts and other wholefoods and olive oil. But when drinking alcohol, beef and dripping, chocolate and coconut are the best fats. Olive oil not terrible, but any high-PUFA oil including CLO should be avoided.

Tucker Goodrich said...

That linoleic acid causes liver disease in humans is a well-established fact.

"Chronic exposure to total parenteral nutrition (TPN) can cause parenteral nutrition associated liver disease (PNALD), a progressive condition that may severely affect the liver and lead to end-stage liver disease."

"Fish Oil for Patients With Liver Disease Due to Parenteral Nutrition"

The traditional fat used for TPN is soybean oil, rich in linoleic acid.

They're doing this phase III study because the clinical evidence is so overwhelming.

Nevertheless, currently the FDA will only allow you to use non-linoleic-acid-based TPN once you already have liver damage.

dr j said...

Taking up the diet as religion angle, the Sri Lanka Cancer incidence data is interesting when one looks at it from a religion based perspective.

While Muslims make up only 9.7% of the population, i think I better make a trip there and see how the Muslim women diet works and maybe follow their approach.

George Henderson said...

in rat hepatocytes 22% of linoleic acid is converted to cholesterol. Even at low intakes. Feed enough linoleate (or soy oil) the liver is awash with unesterified cholesterol and free fatty acids. This tips it into the NASH version of NAFLD.
Big price to pay for lowering serum cholesterol.

Edward Edmonds said...

"Acceleration of essential fatty acid deficiency by dietary cholesterol"

George Henderson said...

@ Edward,
"The basic diet used in these studies contained 18% of casein, 70% of sucrose,
4% of cellulose, 4% of Wesson salt mixture, 1% of fat, and 3% of urea".
Wow! that's some diet.
Interesting about the Cholesterol esters incorporating EFAs.
In some versions of the cholesterol-accumulation hypothesis, it's the esters accumulating in membranes that impede mitochondrial matrix function, whereas the unesterified cholesterol promotes mitochondrial apoptosis.
Thus disease effects of excess linoleate - whether NAFLD or NASH - might depend on the particular allele or activity of the cholesterol esterification gene ACAT, as well as any dietary cholesterol thrown into the mix.

Jack Kruse said...

Lipid health or un-health parallels 3 issues tied to the Earth history. The earth’s life history is divided sharply into two periods – an anaerobic and an aerobic phase. The third phase occurred at the Cambrian explosion when environmental chemistry drove massive expansion of DHA in the sea. From that point in time all complexity on this planet arose in 32 phylla. It is really not controversial. I think modern beliefs and literature muddy the water. Why you guys use this data is beyond me when the history of lipids is clear from the evolutionary perspective. Starting about 600 million years ago, oxygen tension in the atmosphere rose above the Pasteur point, at which point aerobic metabolism became thermodynamically possible. Because of this one effect biochemistry shifted up eightfold in efficiency with aerobic metabolism giving rise to the 32 phyla. This uptick in energy also gave the power of DNA to mutate faster. Electrons from oxygen created the class of PUFA's and how they have been handled should be clear from this view point.
Right after this flowering plants and mammals showed up mammals began immediately collecting DHA in their cell membranes and brains. This has been a constant for 600 million years with respect to all PUFA's. In an anaerobic world there is no way to make long chain PUFA's. Aerobic metabolism took the world of Archea and prokaryotes from an alphabet of a few lipids to one with over 1200. This gave life the ability to respond quickly to environmental pressures. Lipid development dictated to DNA. This has been conserved even today. The influence of environment as the key driver in evolution is seldom given full thought. It was exemplified by the fact that chimps have the same genome as humans yet morphologically are mismatches. This has been repeated in modern western human populations. We have all changed in shape, size, and disease patterns in just one century. Yet, the genome remains well conserved in time and the proteins dictated by the nucleic acids. This all leads to the logically conclusion there has to be a variable to explain evolution at a quantum, chemical and molecular level. Darwin does not have that mechanism.
Peroxisomal proliferator-activated receptors (PPAR’s), the vitamin D receptors, and the retinoid X receptors (RXR), and the retinoic acid receptors (RAR) are all examples of the nuclear receptors that interact with the brain cell membranes to control inflammation and metabolism all over our bodies. These are all loaded with DHA. DHA selected these protein receptors long ago. DHA had little to do with it. It turns out that PPAR’s are the receptors that are at the crossroads of where inflammation and metabolism actually cross. DHA is critical in both pathways. These are specialized lipid sensors that pay attention to our balance of omega 6 and 3 levels. It took more 02 in the atmosphere to make DHA in the ocean. This allowed life to take advantage of the electron.
I'd like to point out here because you do not explicit state it, DHA in humans, is not burned for fuel. It is reserved. It is the most unsaturated of cell membrane fatty acids (Jump, 2002). Dolphin and other marine mammals have difficulty in obtaining sufficient AA from the marine food chain to serve the needs of the brain and reproduction. Humans had access to both in excess and collected more DHA than any other mammal. It is associated with health in every organ system ever studied. The context of how a mammal gets to use it in its neural circuitry is critical to what dietary lipids it should eat.
So who cares what mice and rats do with it.

Jack Kruse said...

PUFA's represent less than 1% of the fatty acids in body fat at birth, of which DHA is nearly half. It seems like a trivial amount, but when multiplied by the normal 5000-6000 g of fat at birth, it amounts to a 1000 mg reserve of DHA.

George Henderson said...

Great thought Jack.

Rats and mice react to linoleate and ALA in much the same way humans do - of course these are mainly just placeholders and weak precursors for the real deal long chain PUFA.
And LCPUFA oxidise easily, DHA being so unsaturated is fragile, hence excess fish oil is especially toxic in presence of alcohol and probably other drugs. And immunosupressive, could be a good thing, depending on the context, but worth watching.
Darwin did allow for non-Darwinian evolution; he quite clearly stated that his system didn't explain everything he had observed to his satisfaction.
Not all aspects of human replication and reproduction rely on DNA; some essential parts of cellular structures are duplicated directly when cells divide, so mutations could alter function without changes in DNA.

George Henderson said...

This is interesting - omega 3 transgenic mice

Using transgenic (Tg) mice over-expressing the C. elegans fat-1 gene encoding an enzyme that converts endogenous omega-6 to omega-3 polyunsaturated fatty acids (n-3 PUFAs), we showed that fat-1 Tg mice with chronically elevated brain levels of n-3 PUFAs exhibited less brain damage and significantly improved long-term neurological performance compared to wild type littermates.

Jack Kruse said...

George, Darwin was more forth right with where he went off the rails with his morphologic observations. But the Neo-Darwinist are not so kind in this regard because they focus on natural selection and not conditions of existence. Darwin said in his original text of the 2, the latter was far more important. Neo-Darwinists like Dawkins or Jerry Coyne only focus on the weaker of the two, namely natural selection, because the legacy of Darwin is only based on morphology. He put forth no mechanism to back his theory. In fact, Pangenes almost sunk his entire theory in the 1880's

As for my last few comments here is some more fuel.

DHA in humans is predominantly found in tissues with high metabolic rates; This would seem counterintuitive because of its allylic double bonds make is very susceptible to lipid peroxidation. One would think it would be burned for its carbon backbone but it is not burned for energy at all in humans. Hence, once acquired, it is avidly retained. (Brenna and Diau, 2007)

Moreover, when one considers the brain is only about 2% of total body weight in adult humans, it consumes 18-22% of whole body energy we begin to see DHA provides a huge energy advantage to the brain. This benefit is 100% tied to mass equivalence of the electron. It is not constrained by mass as a proton is. Peter is developing his ideas on this but it is the near massless electron that is the key. It has 1/2000th the mass of a proton, and has better abilities to carry energy and information to build complexity. This is why mitochondria use protons for signaling and the input is 100% electrons from foods.

They key point is much of this extra energy expense is provided by the thermodynamic advantage of the electron. This energy is directly devoted to phospholipid recycling, where proteins have been selected by their marriage with DHA to have thiols residues (from cysteine) to provide DHA with even more electrons to interact with its pi electron clouds. This is really true in the retina where recycling failures lead to macular degeneration. This symbiosis specifically maintains membrane voltage maintenance to maintain high performance. It also requires iodine's presence to protect the entire lipid raft from oxidation at the most metabolically active sites in neurons (Purdon and Rapoport, 2007).

shitmypresidentsays said...

MCT oil is not horribly expensive.