Why you don’t need vegetable oil in your carrot cake and milk shake to protect your arteries.
By Chris Masterjohn
Published August 20, 2006
Newspapers the world over have recently declared that a single meal rich in saturated fats will disrupt the functioning of your arteries and contribute to the inflammation of your blood vessels, following an Associated Press story by Joe Milicia.1 Milicia reported on a recent study2 published by a team of researchers led by Dr. Stephen J. Nicholls of the Australian Heart Research Institute in the Journal of the American College of Cardiology entitled, “Consumption of Saturated Fat Impairs the Anti-Inflammatory Properties of High-Density Lipoproteins and Endothelial Function.”
The news article quoted the Kansas City cardiologist Dr. James O’Keefe as claiming the study showed that “when you eat [saturated fat], inflammation and damage to the vessels happens immediately afterward.” Of course, the study showed no such thing.
Dr. Nicholls, the lead author of the study, was quoted in the article as saying the study showed the “need to aggressively reduce the amount of saturated fat consumed in the diet.” The AP article then clarified for us that this meant reducing our intake of beef, pork, lard, poultry fat, butter, milk, cheeses, coconut oil, palm oil and cocoa butter, and replacing them with safflower oil, sesame oil, sunflower seeds, corn and soybeans. Wow! This study had the power to make sweeping conclusions about over 15 different foods! But in reality, of course, the study showed no such thing.
The study has already been widely criticized on the internet. Some of the criticism has been good; some has been rather poor. My own view is that this was a well-designed and interesting study; the authors of the report, however, unfortunately made unjustified conclusions from their data in the report itself, and the press articles further sensationalized the story and distorted the study’s findings, making rather hysterical claims, unfortunately with the support of the study’s lead author.
You may be surprised to find out that arterial function was actually better after the coconut oil meal than the safflower oil meal! Or that, contrary to the claims of the Associated Press article, the authors never measured inflammatory components in the subjects’ blood. Or further, that they provided absolutely no evidence that different types of fatty acids, such as saturated or unsaturated, had anything to do with their findings!
In fact, they completely overlooked an alternative explanation that has substantial evidence in the scientific literature to support it: the differences they observed between the anti-inflammatory effects of the different diets may have been due largely or entirely to the difference in vitamin E content of the diets rather than the type of fatty acids present in the oils.
So, let’s take a look at what the researchers actually found, and what it might actually mean.
One High-Saturated Fat Meal: The Real Story
The researchers fed fourteen adults a meal of carrot cake and a milk shake on two separate occasions. In half of the meals, the food was made with coconut oil, which is about 90% saturated fat, while in the other half, the food was made with safflower oil, which is about 75% polyunsaturated fat.
Both oils were non-hydrogenated, organic, unrefined and virgin (David Celermajer and Jason Harmer, personal communication).
Each subject received each of the two types of meals on separate occasions after an overnight fast. Half of them received the safflower oil meal first and the coconut oil meal second, while the other half received the coconut oil meal first and the safflower oil meal second, to ensure that the order in which they received the meals did not affect the result. The two meals were separated by a month, to ensure that the first meal had as little an effect as possible on the second meal. Finally, the researchers were blinded to which meals the subjects were receiving, to minimize the effect of bias on the collection of the data.
The researchers took three types of measurements before the meals were fed and at 3 hours and 6 hours after the meals were fed. The first type of measurement they took was the levels of various constituents in the subjects’ blood: total cholesterol, LDL, HDL, triglycerides, insulin, and free fatty acids.
The second type of measurement they took was of various parameters of blood flow. For example, they tested the amount of blood flowing through the subjects’ forearms at each point, and they tested the subjects’ “vascular reactivity.” That is, they used pressure to stop blood flow through an artery in the arm and then tested how quickly and to what extent the artery reacted once the pressure was released by dilating to increase the return of blood to the blood-deprived area. With this type of test, the more the blood vessel dilates when pressure is released, the better shape it’s believed to be in.
Finally, the researchers extracted HDL from the subjects’ blood at each time point. Then, they incubated endothelial cells from human umbilical veins with the HDL at various concentrations. After the incubation period, they added an inflammatory chemical called TNF-alpha to the cells, which stimulates the production of adhesion molecules such as ICAM-1 and VCAM-1, which are believed to play a role in the adhesion of plaque to arteries. HDL has been shown to inhibit the expression of these inflammatory molecules, and the researchers conducted this part of the study to see if how you eat can affect how much potential HDL has to inhibit the expression of presumably harmful adhesion molecules. (See note 1 for brief comments on the sample size and study design.)
The researchers claimed to generate two findings:
- Flow-mediated dilation, or the ability of blood vessels to dilate and return blood flow after being occluded with pressure, decreased more strongly after the coconut oil meal than after the safflower oil meal. From this, they concluded that “consumption of saturated fat impairs . . . endothelial function.”
- When cells were incubated with HDL taken from subjects after they ate the coconut oil meal, the expression of the inflammatory adhesion molecules ICAM-1 and VCAM-1 in response to TNF-alpha stimulation was increased compared to cells incubated with HDL taken from fasting subjects. By contrast, when cells were incubated with HDL taken from subjects after they ate the safflower oil meal, the expression of inflammatory molecules in response to TNF-alpha stimulation was decreased compared to cells incubated with HDL taken from fasting subjects. From this, the authors concluded that “consumption of saturated fat impairs the anti-inflammatory properties of high-density lipoproteins.”
Although both of these conclusions are more conservative than the statements written in the Associated Press article, neither of them are justified by the study.
Let’s take a closer look at each.
Coconut Oil and Flow-Mediated Dilation: Harmful or Helpful?
The researchers claim that consumption of the saturated fat meal impaired flow-mediated dilation — that is, it hurt the ability of blood vessels to dilate and return blood flow to an area from which blood flow had been stopped with pressure. The researchers did indeed show that at the 3-hour mark the decline in flow-mediated dilation was almost twice as great in the coconut oil group as it was in the safflower oil group. (See note 2 for a brief comment on statistical significance.)
Yet as Anthony Colpo, author of The Great Cholesterol Con has already pointed out,4 the flow-mediated dilation was actually higher in the coconut oil group than in the safflower oil group at every point along the way!
The reason? When the subjects were fasting, those who were about to eat the coconut oil meal had 33% better flow-mediated dilation than those who were about to eat the safflower oil meal. Even at the 3-hour point, when flow-mediated dilation had declined the most, it was still 9% higher in the coconut oil group than the safflower oil group!
There are two ways we could look at this. Figure 1 shows the changes that took place in flow-mediated dilation three and six hours after the meals, relative to the flow-mediated dilation before the meals (called “baseline”). (See note 3 for why I’m presenting it this way.) You can see for both the coconut oil meal and the safflower oil meal, flow-mediated dilation declined substantially at the 3-hour mark.
Figure 1. Percent change in the degree of flow-mediated dilation compared to baseline values.
|Group||Baseline||3 hours||6 hours|
|Safflower Oil||No change.||17% lower.||8% lower.|
|Coconut Oil||No change.||32% lower.||10% lower.|
Now let’s look at it another way. Figure 2 compares the relative degree of flow-mediated dilation between the coconut oil group and the safflower oil group. Surprise, surprise — the flow-mediated dilation is higher (a good thing) in the coconut oil group at every single time point during the study!
Figure 2. Comparison of the degree of flow-mediated dilation in the coconut oil group to that in the safflower oil group at three time points.
|Baseline||3 hours||6 hours|
|33% higher in coconut oil group.||9% higher in coconut oil group.||29% higher in coconut oil group.|
Thus, we have to ask: is consumption of coconut oil rather than safflower oil the reason for the greaterdecline of flow-mediated dilation in the coconut oil group? Or is the reason for this decline the simple fact that the people who ate the coconut oil started out with a higher value of flow-mediated dilation in the first place, and therefore, so to speak, had more to lose?
There are two reasons that the latter might be true: first, the decline in flow-mediated dilation after a meal might not be a function of the flow-mediated dilation before the meal; second, a randomly high sampling error for the flow-mediated dilation before the meal could result in what’s called “regression to the mean,” which is explained below.
In the first case, it could be that eating carrot cake and drinking a milkshake, regardless of whether it is made with safflower oil or coconut oil, depresses flow-mediated dilation to a certain point regardless of fasting levels of flow-mediated dilation. For example, eating the meal might depress flow-mediated dilation to about 5%, regardless of whether the person’s fasting level of flow-mediated dilation was 6% or 9%, in which case a person with a higher fasting level would experience a greater decline simply by virtue of the higher fasting level.
Thus, the coconut oil group, who by random chance had a 33% higher fasting rate of flow-mediated dilation, would exhibit a greater relative decline than the safflower oil group for no other reason than that they started off with substantially better flow-mediated dilation in the first place!
The authors themselves admitted a very similar explanation in the journal article, writing that “it is possible that ‘regression to the mean’ may have contributed to some of the FMD [flow-mediated dilation] reduction observed after consumption of the saturated fat.” The concept of “regression to the mean” is essentially this: if by random sampling error an initial value tends to be higher than the mean, a second value will tend to be closer to the mean. Thus, a decline in values could result simply from the first value being randomly high.
And of course that’s exactly what we saw here. Yet was this caveat noted in the press? Of course not. Instead, we were told that when we eat saturated fat, “damage to the vessels happens immediately afterward,” and thus we must “aggressively reduce the amount of saturated fat consumed in the diet.”
No one warned us that if when fasting, by random sampling error we happen to have a higher-than-average value of flow-mediated dilation, “damage to the vessels happens immediately” after we eat due to “regression to the mean.” No one warned us that we must “aggressively reduce the amount of random sampling error” lest we suffer statistical arterial dysfunction with one, single meal.
Does Coconut Oil Cause Inflammation?
Contrary to the Associated Press report’s claim that “fewer inflammatory agents were found in the arteries” after the safflower oil meal than before it, the researchers did not measure any type of inflammation in the people consuming the meals. Instead, they incubated isolated umbilical vein endothelial cells with HDL taken from these subjects at various time points before and after the meals, and then stimulated these isolated cells to produce inflammatory adhesion molecules by adding a compound called TNF-alpha to the cells, and measured whether the HDL isolated after the different meals had a different ability to lower the amount of adhesion molecules released after stimulation with the TNF-alpha.
The researchers found that cells incubated with the HDL isolated from subjects after they had eaten the coconut oil meal produced more adhesion molecules (ICAM-1 and VCAM-1) after stimulation with TNF-alpha than cells incubated with HDL isolated from fasting subjects, and that cells incubated with HDL isolated from subjects after they had eaten the safflower oil meal produced fewer adhesion molecules after stimulation than cells incubated with HDL isolated from fasting subjects.
There are a number of problems with the large leap of logic it takes to conclude from this that a meal rich in saturated fat causes inflammation. First, others4, 5 have already questioned how relevant this finding with isolated cells is to how our arteries actually function within us. After all, we are neither test tubes nor Petri dishes, but complex organisms with many different chemical and electrical feedback systems that do not exist in laboratory dishes. The researchers could have directly measured the levels of ICAM-1 and VCAM-1 in the subjects’ blood, but that is not what they chose to study.
Second, the researchers only studied the anti-inflammatory potential specifically of HDL. The researchers could have incubated the cells with whole plasma to measure the total anti-inflammatory capacity of the blood, but they chose not to, for the simple reason that they were only trying to answer one small question about HDL and not look at the bigger picture (David Celermajer, personal communication). Virgin coconut oil is rich in very powerful polyphenols,6 some types of which have been shown to decrease expression of TNF-alpha and adhesion molecules,7, 8 and which are carried by water-soluble proteins in the blood and not by HDL.9 Thus, virgin coconut oil’s contribution to the anti-inflammatory capacity of the blood could be primarily in the non-HDL fraction, whereas safflower oil’s contribution to the anti-inflammatory capacity of the blood might be primarily in the HDL fraction. We simply do not have enough knowledge at this point to say for sure.
The only way to determine the effect of safflower oil and coconut oil on the total anti-inflammatory capacity of the blood is to perform the experiment by incubating the cells with whole plasma. The only way to determine the effect of safflower oil and coconut oil on the actual level of inflammation in the people consuming the oils is to measure the inflammatory compounds being directly produced in their blood. This study did neither.
Finally, and most importantly, the researchers provided no evidence whatsoever that the effects they observed were due to the type of fat. They simply assumed that the difference they observed between safflower oil and coconut oil was due to the fact that coconut oil is high in saturated fat and safflower oil is high in unsaturated fat. In doing so, they overlooked a very interesting hypothesis that could explain their results and that has substantial support in the scientific literature.
An Alternative Hypothesis: Vitamin E
The difference between safflower oil and coconut oil does not stop at the relative saturation of their fatty acids. Figure 3 shows the difference in vitamin E content between the two oils. Safflower oil is 77 times higher in alpha-tocopherol and 47 times higher in total tocopherols.10
Figure 3. Typical tocopherol (vitamin E) content of coconut oil and safflower oil. Source: (Enig, 2000).
|Tocopherol||Coconut Oil||Safflower Oil|
|Alpha-tocopherol||5 mg/kg||387 mg/kg|
|Delta-tocopherol||6 mg/kg||240 mg/kg|
|Total tocopherols||11 mg/kg||801 mg/kg|
Is it plausible that the difference in vitamin E content of the oils could account for the difference in the expression of adhesion molecules in the isolated cells? Absolutely.
A recent review of alpha-tocopherol’s role in regulating gene expression listed the suppression of the gene that codes for ICAM-1 as one of its functions.11 In fact, Chinese researchers performed a very similar experiment to the one we’ve been discussing, where they incubated endothelial cells taken from human umbilical veins with vitamin E instead of HDL. They found that incubating the cells with alpha-tocopherol, gamma-tocopherol and mixed tocopherols all inhibited the ability of oxidized LDL to induce ICAM-1 expression in the cells in a dose-dependent manner.12 Another group found vitamin E to reduce both ICAM-1 and VCAM-1 in the heart cells of rats.13
Vitamin E suppressed ICAM-1 and VCAM-1 levels in vivo in rabbits, although the effect on VCAM-1 was not statistically significant.14 In humans, the combination of vitamins E and C, but not vitamin C alone, decreased blood levels of ICAM-1 after six months. When the supplementation was stopped, blood levels of ICAM-1 returned to their initial levels. A similar effect was seen on VCAM-1, but it was not statistically significant. Unfortunately the researchers did not study the effect of vitamin E alone.15
Vitamin E travels in the blood associated with lipoproteins, including HDL.16 When endothelial cells are incubated with vitamin E-enriched HDL, they selectively take up vitamin E from the HDL at ten times the rate at which they take up the HDL particles themselves.17 It is therefore reasonable to suggest that the high vitamin E content of safflower oil led to an enrichment of the subjects’ HDL particles with vitamin E, which was then taken up by the endothelial cells where it suppressed the expression of adhesion molecules.
Yet one question remains: why would the HDL taken from subjects after they ate the coconut oil meal be less effective at suppressing the expression of adhesion molecules than HDL taken from subjects when they were fasting? From what I can find, data is very limited on the effects of eating a meal on the distribution of vitamin E in the blood. The one study I’ve found so far16 suggests that the fraction of vitamin E in HDL actually declines temporarily after a meal when the meal is relatively low in vitamin E, but rises if the meal is high in vitamin E. (See note 4.) It may be, then, that the vitamin E content of HDL declined after the coconut oil meal not because of the coconut oil itself but because any low-vitamin E meal reduces the amount of vitamin E in circulating HDL, while the safflower oil added enough vitamin E to the meal to make the vitamin E content of HDL rise.
The only way to actually know would be to directly measure the vitamin E content of the HDL particles after the meal. Although the researchers who conducted the study we’ve been discussing measured the amount of protein, phospholipid, triglyceride and cholesterol in the HDL particles that they extracted, they unfortunately did not measure the amount of vitamin E in these particles.
This is, of course, a hypothesis. I have not shown conclusively that the effects observed in the study must have been due to vitamin E; I have simply shown this is a plausible explanation. Further research would be needed to confirm or refute my hypothesis (see note 5.)
Likewise, it is an unconfirmed hypothesis that the effect observed was a result of the consumption of saturated fat. This unfortunately did not stop the researchers from titling their paper “Consumption of Saturated Fat Impairs the Anti-Inflammatory Properties of High-Density Lipoproteins and Endothelial Function” as if they had actually shown this to be the case. (My emphasis.)
So Which Oils Should We Eat?
If it turns out to be true that the difference in protective effect of HDL in the test tube was in fact due to the high vitamin E content of safflower oil and the low vitamin E content of coconut oil, that does not mean we should avoid coconut oil. It doesn’t even mean we should eat safflower oil!
It simply means that coconut oil is not a good source of vitamin E. Coconut oil is still the best source of medium-chain fatty acids that boost metabolism and support the immune system, and virgin coconut oil is rich in powerful antioxidant polyphenols.
Polyunsaturated fatty acids such as those found in safflower oil actually deplete the body of vitamin E and thereby increase the body’s need for vitamin E — this is basic textbook biochemistry.18 Safflower oil may raise the amount of vitamin E in lipoproteins immediately after a vitamin E-rich meal, but what is the long-term effect on vitamin E status of a high intake of polyunsaturated fats?
It makes sense then that the best way to obtain vitamin E would be from sources that are high in vitamin E but low in polyunsaturated fat. Palm oil is an excellent example of such a source.
Palm oil is only 9% polyunsaturated, compared to safflower, which is 75% polyunsaturated. In terms of absolute amount of vitamin E, palm oil has a somewhat lower level of alpha-tocopherol, more than double the gamma-tocopherol, and large amounts of tocotrienols, which are another important part of the vitamin E complex that are completely absent in safflower oil. The combined absolute value of tocopherol and tocotrienol forms of vitamin E is 46% higher in palm oil than safflower oil.
When one takes into account the high polyunsaturated fat content of safflower oil, which increases the need for vitamin E, the advantage of more saturated palm oil becomes obvious: the ratio of vitamin E to polyunsaturated fatty acids in palm oil is 12 times the same ratio in safflower oil!
Yet newspapers the world over carrying the Associated Press article told us to reduce our intake of palm oil and other saturated fats “aggressively.”
Drawing Conclusions: One Meal High In Saturated Fat is Not So Bad
We’ve been told that this study shows that when “you eat [saturated fat], inflammation and damage to the vessels happens immediately afterward.” We’ve been told that it shows we must “aggressively reduce the amount of saturated fat consumed in the diet.” We’ve been further told to throw out the beef, pork, lard, poultry fat, butter, milk, cheeses, coconut oil, palm oil and cocoa butter, replacing all these fats with safflower oil, sesame oil, sunflower seeds, corn and soybeans.
This is all on the basis of a study that couldn’t differentiate the effect of coconut oil from the effect of random sampling error on flow-mediated dilation and showed people consuming coconut oil to havebetter flow-mediated dilation at all time points than people consuming safflower oil.
It is on the basis of a study that could not differentiate between the effects of saturated fats and the effects of low-vitamin E meals on the capacity of HDL to prevent inflammation in a Petri dish.
It is on the basis of a study that told us nothing about the amount of inflammation going on within the people consuming the meals, who are much more complex than globs of cells in a Petri dish.
Further research should uncover whether the effects seen in the test tube are due to vitamin E, to saturated or unsaturated fats, or to other causes entirely, and what relevance these observations in the test tube have for real, living people.
In the mean time, I’m going to continue cooking with CLA-rich clarified butter, and continue eating vitamin E-rich red palm oil and polyphenol-rich virgin coconut oil and extra virgin olive oil. I will continue to get my essential fatty acids from animal sources including butterfat, egg yolks from pasture-raised chickens, organ meats, cod liver oil, and fatty fish, so I can obtain the most benefit from the hormone precursors and structurally useful essential fatty acids while not overdosing on peroxide-promoting, free radical-generating, vitamin E-depleting polyunsaturates from vegetable oils like safflower oil.
Whoever’s going to convince me to do otherwise has a bit more work to do.