Epidemiology is the study of associations and not causality. It essentially began in 1854 when John Snow noticed that there seemed to be a higher concentration of cholera patients in a certain area in London during one of its many cholera epidemics in the 19th century. That’s an association. The real breakthrough for John Snow was to remove the pump handle on the suspected water source and then observe a significant reduction in the cases of cholera in that area. That’s called an intervention study based on epidemiology. Now in the 21st century we seem very reticent to do any type of intervention studies and rely more on epidemiology to guide our medical decisions. This is made even more confusing with the introduction of meta-analysis into the picture. Meta-analysis is taking a large number of studies (often done under very different conditions), pretending they are all valid and then coming up with a conclusion. When you do a meta-analysis on epidemiology studies, it’s like trying to separate a piece of filet mignon from intestines used to make sausage.
This month an article from the Annals of Internal Medicine suggested that there is no relationship of any type of fatty acid with heart disease (1). Well, if there is no association of any type of fatty acid with heart disease, why not just eat lard instead of salmon? If this sounds a little fishy to you (pardon the pun), it does to me too. As I stated earlier, the problem with meta-analysis is that good studies are added to bad ones. Here’s a dirty secret about medical research. There are a lot of bad studies that get published. Usually if you can’t get the funds to do original research, then you write a review paper, and if you can’t write a review paper, then you do a meta-analysis of all published studies and pretend it’s original research. The media might buy that, but I don’t.
The irony of this study is that one of the authors had actually published a good article using good controls in the same journal a year earlier indicating that the higher the levels of omega-3 fatty acids in the blood, the less heart disease death and the greater the longevity of the individuals (2). Maybe he forgot that article when publishing this new sausage publication (1).
That notwithstanding, the problem with these types of published studies is that they miss the point of what causes heart disease in the first place. It is not fatty acids or cholesterol, but inflammation. The best way to measure inflammation is the ratio of AA to EPA in the blood. This was first reported in the New England Journal of Medicine some 25 years ago (3). High-dose fish oil in healthy volunteers (5 grams of EPA and DHA per day) reduced the AA/EPA ratio from 21 to 2.5 within six weeks. During that time many of the additional markers of cellular inflammation also dropped. When they stopped the omega-3 fatty acid supplementation, the AA/EPA ratio gradually returned to its initial high level with a corresponding increase in the depressed inflammatory proteins to their initial levels. A very nice intervention study.
Then there is the disturbing fact that Japanese males have essentially the same LDL cholesterol levels as Americans, but Americans have 3.5 times the age-adjusted death rate. In fact, the LDL cholesterol levels of the Japanese having been rising since 1980, whereas American’s LDL cholesterol levels have been dropping. In addition, Japanese males in the study were about 7 times more likely to smoke than Americans. Let’s see, rising LDL cholesterol levels coupled with more smoking, but they have 72% fewer deaths from heart disease (4). Maybe the AA/EPA ratio as a marker of inflammation might be a key? The AA/EPA ratio of the Japanese in that study was 2.6, whereas the Americans were 11.1. Actually the Americans in this study were less inflamed than the general American population that has an AA/EPA ratio of 20 (5). But even in the above study, the Japanese AA/EPA ratio was 76% lower than the Americans (4). Let’s see, the Japanese had 76% lower inflammation and 72% lower mortality from heart disease compared to the Americans even through their LDL cholesterol levels were the same and they smoked like chimneys. If I was a betting man, I would put my money on doing an intervention study to see what the effect on heart disease would be if I lowered the AA/EPA ratio. That’s exactly what the Japanese did with the JELIS trial that was one of the largest cardiovascular trials ever undertaken with some 18,000 subjects (6). All of them had high cholesterol, so all of them were put on statins. The average AA/EPA ratio of these subjects was 1.6 compared to the 20 in Americans (5,6). Half the subjects were then given more omega-3 fatty acids. If the meta-analysis study recently published was valid (1), then these extra omega-3 fatty acids would have no benefit especially since everyone was getting a statin. Actually, just the reverse occurred after 3 ½ years. Those who lowered their AA/EPA ratio had 20% fewer cardiovascular events compared to those that didn’t see a change in the placebo group. Further sub-group analysis indicated that the change in the AA/EPA ratio was the overriding factor (7) behind these cardiovascular benefits. This is a complicated way of saying that if you lower inflammation, you lower cardiovascular risk.
So the next time you read about a meta-analysis study on the lack of effect of fatty acids on heart disease, ask to see a real intervention trial that lowers the levels of inflammation. When you do, then you see a very different picture of the role of fatty acids in heart disease than you do by reading more sausage studies (1,8). And if you do an intervention trial with omega-3 fatty acids, make sure that you lower the AA/EPA ratio to the level found in the Japanese. Based on published dose-response studies, this will take a minimum of 5 grams of EPA and DHA per day (9). Up to this point in time, no such cardiovascular studies have been conducted with that level of omega-3 fatty acids. If you are not using at least that level of omega-3 fatty acids to study cardiovascular disease, then you are probably using a placebo dose and should expect placebo results.