Dietary Linoleic Acid Does Not Reduce Diabetes Risk

Introduction

In a recent Twitter thread, Simon Hill, MSc. BSc. tweeted this about a recent podcast episode of The Proof featuring Bill Harris, PhD.

Did you listen to the debate on my show? Omega 6 being a driver of CHD is a very hard position to take.

— Simon Hill MSc, BSc (Hons) (@theproof) June 6, 2022

Obviously he is not a believer that seed oils that are high in linoleic acid are a driver of heart disease.

I’m pretty sure what he’s referring to is this clip at 38:00 of the full video:

He says in the clip (and I’m paraphrasing), “Are the people who have the most linoleic acid their blood more or less likely to develop heart disease and diabetes over time? What we found is that those with the highest levels of linoleic acid in their blood are the least likely to develop cardiovascular disease and diabetes.” He goes on to say (not in the clip), “The people I know who have a problem with Omega-6 will NEVER respond to that kind of data.”

I’m your huckleberry. I have a problem with Omega 6 and I’m happy to respond to your data.

The Harris and Willett paper can’t determine causality

There are two types of scientific papers: those which generate testable hypotheses and those which test the hypotheses.  Sometimes I wish there were more papers that generated testable hypotheses.  For instance, there was a recent paper that looked at circulating metabolites in obese vs. lean humans.  This is useful!

Once you’ve generated your hypothesis, it’s time to test it.  In 1992, Walter Willett published a paper based on data from the Nurses Health Study suggesting that consumption of vegetable oil was associated with decreased levels of diabetes^​1​.  The main component of most vegetable oils is linoleic acid.  So in 1992 we had a testable hypothesis: dietary linoleic acid reduces diabetes risk.  

The Nurses Health Study is a prospective Cohort Study, meaning that you find a bunch of people (cohorts) and you ask them what they eat, how much exercise they get, etc and then you follow them for a bunch of years to see how many develop diabetes, heart disease, etc.  Among the many confounders in this type of study is something called Healthy User Bias.  The people who are the most well off, who have the best education, who are the healthiest overall are the most likely to do the thing encouraged by the mainstream health industry, which in 1992 was most definitely to eat more linoleic acid.  What this type of study can never prove, especially if you let the cat out of the bag and told everyone to eat veg oil before you proved it was better for them, is whether or not vegetable oil is better for them.  Like I say, the whole point of the study is to generate a hypothesis.

29 years later – in 2021 – WIllett published the same testable hypothesis^​2​, adding nothing to global scientific knowledge.  In 30 years, Willett failed to test his hypotheses. With the three decades of additional research on the hypothesis, the benefit of maximal linoleic acid consumption has now dropped to a 6% reduction in diabetes risk.  6% strikes me as being well within the range of what would be expected from healthy user bias^​3​.  The 6% shows statistical significance but biological irrelevance. I wouldn’t even test it at this point.

That is why Dr. Harris is so careful with his language.  He never mentions diet, he only says that blood levels of linoleic acid are protective, which is the other finding of the Willett paper with which he is involved with.  He is correct, but we all already knew that.  One of the groups whose data Willett republished (its a meta-analysis) is the group who performed the EPIC Potsdam study, which I’ve referenced previously.  That group showed that linoleic acid in red blood cell membrane phospholipids do indeed correlate with lower risk of diabetes^​4​.  We’ve known this since 2011.

A less fortuitous finding of the Potsdam study (From Dr. Harris’ perspective) is that there appears to be a positive correlation between dietary linoleic acid and diabetes risk that falls just barely below the threshold of statistical significance (p=0.054). Even though there was a very weak overall correlation between diet and red blood cell fatty acid levels of 0.21 – meaning that 21% of the linoleic acid levels of the people could be explained by diet and 79% is explained by something else – the relationship between dietary linoleic acid and diabetes risk was none or maybe even a positive correlation. 

This all suggests that the reason for the correlation between high circulating linoleic acid and low diabetes risk has more to do with how individuals metabolize it and less to do with dietary intake, per se.  The EPIC Potsdam group is very good, they explained this in 2011 (FA = Fatty Acid):

the FA profile in biological tissues is also strongly dependent on the endogenous metabolism of FAs. Several FAs can be newly synthesized [saturated FAs (SFAs) can be synthesized from acetyl coenzyme A], elongated (lengthened by 2 carbon atoms), or desaturated (insertion of a double bond) endogenously (3). Three desaturases are known in humans. Stearoyl coenzyme A desaturase (SCD; also called D9 desaturase) catalyzes the synthesis of monounsaturated FAs (MUFA) from SFA,
whereas D5 desaturase (D5D) and D6 desaturase (D6D) are required for the synthesis of long-chain n26 (omega-6) and n23
(omega-3) polyunsaturated FAs (PUFAs)

Erythrocyte membrane phospholipid fatty acids, desaturase activity, and dietary fatty acids in relation to risk of type 2 diabetes in the European Prospective Investigation into Cancer and Nutrition (EPIC)–Potsdam Study, Janine Kroger, Vera Zietemann, Cornelia Enzenbach, Cornelia Weikert, Euge`ne HJM Jansen, Frank Do¨ring, Hans-Georg Joost, Heiner Boeing, and Matthias B Schulze

Unlike Willett, the Potsdam group proceeded to actually go out and test their hypothesis.  By the time Willett and Harris republished the EPIC Potsdam data, it was old news AND it had been thoroughly explained.  

Enzymes involved in fat metabolism are crucial to our metabolism and they leave footprints.  Perhaps the most well known of these is SCD1, which converts saturated fats to monounsaturated fats.  When SCD1 levels are high, the level of palmitoleic acid (monounsaturated fat – the end-product of SCD1)  rises with respect to the level of palmitic acid (saturated fat – the starting product of SCD1).  This leads to a high desaturase index (DI16), which is an indirect indicator of SCD1 levels.

There is another desaturase enzyme which catalyzes the first step on the pathway of linoleic acid becoming arachidonic acid (AA) called D6D (delta-6 desaturase).  A third desaturase enzyme called D5D is also involved in this pathway.  When D6D activity is low, linoleic acid builds up at the expense of its end product (18:3 n6).  When D6D activity is high, linoleic acid is rapidly converted to arachindonic acid.  High D6D activity does not necessarily predict high AA since, like all fats, its level depends on quantity in minus quantity out.  AA in is controlled by D6D.  AA out is determined by phospholipase A2 and oxidases such as cytochrome oxidases and cytochrome p450s which convert AA into oxylipins.

In succession, the authors of the epic Potsdam paper showed that 1) a common genetic variant in the D6D/D5D locus was associated with lower activity of D6D, low risk of diabetes and higher levels of linoleic acid,^​5​ 2) age, sex, physical activity, smoking and dietary factors were only weakly associated with the estimated desaturase activities^​6​, 3) FA-ratios reflecting activity of Δ6-desaturase (D6D) and stearoyl-coenzyme A-desaturase (SCD) were positively associated with triglyceride and LDL-cholesterol concentrations, (in the journal Nature)^​7​, 4) the desaturase index (16:0/18:2n6) is linked to cardiometabolic diseases in the EPIC Potsdam (PLOS ONE)^​8​ and 5) “D6D is causally linked to cardiometabolic risk, which is likely due to downstream production of fatty acids and products resulting from high D6D activity”^​9​.

Points 4 and 5 need a bit of explaining.  Point 4: When De Novo Lipogenesis is highly active, the end product – palmitic acid (16:0) – is elevated and dilutes out linoleic acid (18:2n6).  So there are two potential reasons other than dietary intake that people might have low levels of linoleic acid: high D6D activity and/or high DNL activity.  Both high D6D and high DNL are bad!

Point 5:  It’s very difficult to do double blinded, randomized dietary controls on free living humans to establish causality.  The next best method we have for establishing causality is something called Mendelian Randomization (MR).   MR uses both genetic and lifestyle variations – in this case differences in linoleic acid consumption and the genetic variation in D6D enzymes – to assess cause and effect.  We can say that a high conversion rate of linoleic acid to AA likely CAUSES diabetes.

This is what it looks like when a good research group decides to test a hypothesis.  It’s true that circulating linoleic acid is associated with lower diabetes risk, but high levels of circulating linoleic acid aren’t associated with dietary linoleic acid, they’re associated with low levels of D6D and low levels of SCD1 (AKA lipogenic enzymes). Here is a table of the relative risks they found:

Reductive Stress, D6D and SCD1

Perhaps you’ve read this whole article with interest but thought, “OK, but what does this all have to do with reductive stress?”  In my last post I explained that dietary linoleic increases SCD1 expression through a buildup of mitochondrial acetyl-CoA and a decrease of the NAD+/NADH ratio, which inhibits Sirt1 and Sirt3, which are responsible for deacetylation.  Acetylation activates lipogenic gene expression.

Adding to this, the enzymes D5D and D6D consume NADH and run at faster rates the higher the NADH/NAD+ ratio is^​10​ (SCD1 does, too).  The more reductive stress you are in, the more linoleic acid will be converted to oxylipins.  Even though genetic factors and dietary linoleic acid converge to play a huge role in your risk of diabetes, you may have any of 100s of other risk factors or environmental toxins (such as Forever Chemicals) that could put you into reductive stress, raising your risks of diabetes by putting you into reductive stress and activating D6D, D5D and SCD1.

Conclusions

If you want to plant your flag with the research group whose been sitting on the same testable hypothesis since 1992, be my guest.  I’d recommend going with the group that is out there doing the real work and publishing in Nature and PLOS ONE.  I’d go with the group that did the research that is the closest thing we have today to proving causality – mendelian randomization.  The MR experiment, combined with the genetics results and everything we know about the regulation of D6D suggests that the association between diabetes and low linoleic acid levels has everything to do with how linoleic acid is metabolized and nothing to do with dietary linoleic acid.  

Diabetes is caused by reductive stress, which drives forward the D6D reaction and lowers linoleic acid levels.