Friday, September 28, 2012

Omega 6, Obesity, and Endocannabinoids (Again)

One great mystery in this nutritional debate is how the Harvard School of Public Health and their epidemiologists keep finding such links between the omega 6 polyunsaturates and all manner of good health. My confusion comes in part because in all likelyhood the most omega 6 fats will be eaten by people eating a load of processed crap who by all accounts should not be the healithiest. If the epidemiologists take all those people out of the equation what are we even measuring? Just the leftover folks who eat salad and chicken and fish and walk their dogs every day. Hardly seems fair.

A Civil Twilight: River (right click to open in new window)

And yet, those crafty vegetable oils elude my ability to smack them down entirely. I've done some reading on the metabolism of the omega 6s and it doesn't make for very fun blog posts. Suffice it to say that *maybe* if you pour tons of O6 down the gullet there may be some compensatory reduction in the inflammatory pathways they ought to light up like a runway. It seems that the real key to staying healthy while eating commercial salad dressings, factory-farmed eggs, and chicken skin is to make sure you do NOT skimp on the omega 3s. Also, avoid trans fats like the plague (duh) because they can interfere with uptake of the omega 3s.

Earlier this week a paper from Nature tweeted by Stephan Guyenet and Mike Eades that adds more fuel to the anti-O6 argument.  Omega 6 in large amounts: fragile polyunsaturates, biologically active, evolutionarily novel, and not nearly as tasty as steak or olive oil. Is it a coincidence that the obesity epidemic began and peaked when enthusiasm for omega 6 was at it's highest? Come the late 90s and 00s, Mediterranean diets became more the rage. Veggie oil, king of low cholesterol, began to be edged out by the monounsaturates. Nevertheless, "during the 20th century, elevations in AA-PL have been estimated from the dramatic increase in dietary LA resulting from > 1000-fold increase in per capita consumption of soybean oil from 0.006 to 7.38% of energy." (Gah) (AA-PL = arachidonic acid phospholipids, or the amount of omega 6 derived compounds in cell membranes.  LA = linoleic acid, the primary dietary omega 6 found in soybean and corn oils, etc.)

We've heard the omega 6 obesity story before, and it has everything to do with the endocannabinoid system.  Here is one of several blog posts from 2011 where I broke it down into exrutiating detail.  In short, omega 6 fats are made into natural endocannabinoids, our own happy cannabis compounds. Smoking a ton of weed is associated with: hanging out listlessly on a couch in one's parents' basement watching Yo Gabba Gabba AND the munchies.

Central cannabis receptor activation is associated with increased eating and increased fat accumulation and fatty liver. So researchers thought they would take some happy mice and increase linoleic acid as a controlled dietary variable and see if it made the little guys fat via increased levels of endocannabinoids.

So the mice were fed pellets with and pastes with 20% protein, some carbs, and then mixtures of 7 different oils.  Lipids were extracted from the mice livers and brains and levels of endocannabinoids were measured.

Mice with 8% linoleic acid diets (comparable to modern human diets) had elevated levels of linoleic acid and arachdonic acid in the cell membranes (not surprising) compared to the historic 1% mice (ahem, human) diets.  Levels of the endocannabinoids were tripled in the 8% LA diets. Dietary LA increased body weight, food intake, and fat tissue in the mice.

Here's the key, however: Adding 1% EPA and DHA omega 3s to the mouse diets seemed to undo much of the problems caused by the gallons of omega 6. Omega 6 in the cell membranes dropped, as did the levels of endocannabinoids, as did the fatty tissue, weight, and overeating in the mice. These mice didn't look quite as nice metabolically or had as beautiful cell membranes as the 1% LA mice, but it was loads better than the 8% LA omega 3 deficient mice.

Dietary LA also increased leptin and decreased adiponectin.

In the human population, dietary consumption of soybean oil, poultry, shortening, and sugars (but not grains, beef, fish, eggs, dairy, or vegetables) were positively correlated with obesity in several epidemiology cohorts from 1909 to 1999.

Honestly, this paper is the strongest one yet I've seen maligning omega 6 fatty acid in vast quantities in the diet. One can't necessarily make the leap from the observational data in humans and the controlled data in the rats to an absolute causal relationship in humans, but hey, playing it safe with olive oil and avocados and rolling on the wild side with some saturated animal and tropical plant fats doesn't seem like it would be that unwise compared to toking it up on processed fried foods, no?


  1. Yay, good summary. It would have been nice of them to do a few more experiments. 2% n-6, 4%, 6%, maybe 1% with 1% omega-3. But it's a good paper nonetheless.

  2. Meta-analysis Reveals “Heart Healthy Omega-6 Fat” Increases Risk of Heart Disease
    Bottom Line: The research upon which the American Heart Association based their "eat-your-omega-6-fat" advisory, is fatally flawed, according to the results of a meta-analysis study, which showed that a steady diet of omega-6 polyunsaturated fatty acids increases the risk of heart disease and death, especially for women [1]. British J Nutr. Dec 2010.
    These NIH researchers were angered by the AHA advisory and uncovered very flawed Harvard analysis. Selective citation and incomplete data are the reasons N6 PUFA are erroneously "healthy"

  3. Emily, I have real concerns about this study because it focuses on AA-PL as the vehicle for destruction stemming from excess converted LA - in rats! This is a major flaw, because a review [1] of the literature in humans shows that dietary LA has very little affect at all on AA levels - again, in humans. Furthermore, humans at best [2] can convert about 5% of dietary LA down the cascade to AA and beyond. In fact, in a just released study [3], except for nearly all african populations, many humans have ancestral genes coding for very inefficient conversion of pufas (o3 & o6) from LA and ALA. For these people, AA, EPA, DHA, et al may as well be essential nutrients, because except from ludicrous overconsumption of LA/ALA (which *does* happen in our western diets), many people aren't likely to see much AA/EPA/DHA from it IMHO. In my mind this calls into question any study that bashes AA levels using dietary LA as a starting point (especially in rats of all things! :)

    [1] Increasing dietary linoleic acid does not increase tissue arachidonic acid content in adults consuming Western-type diets: a systematic review [2011]

    [2] (lazy wikipedia cite)

    [3] Mathias RA, Fu W, Akey JM, Ainsworth HC, Torgerson DG, et al. (2012) Adaptive Evolution of the FADS Gene Cluster within Africa. PLoS ONE 7(9): e44926. doi:10.1371/journal.pone.0044926

  4. Chris, my understanding is that in normal, healthy individuals there is not a lot of conversation of LA to AA. However, I have also read that high insulin levels increases delta-5 desaturase and this makes the conversion of DGLA (from the LA->GLA->DGLA pathway) to AA easier.

    If true, perhaps LA in veggie oils *is* problematic for a chunk of the overweight/obese population.

  5. Oh, and one more thing. If insulin does drive LA->AA, this might also be an interesting explanation for why obesity is so predictive of inflammatory diseases!

  6. Beth, that would be interesting if true, I will look for papers to that effect. The implications are interesting for the genetic populations that would be predisposed to efficient (for a human) conversion of LA->AA as mentioned in citation [3]. The point of [3] is that significant minorities in Europe (and majorities in Asia/native-americans) have FADS1+2 genes that code for very inefficient conversion along the LA->AA pathways and beyond, whereas Africans near completely obtained mutations that code for efficient conversion. (FADS1 = d5d and FADS2 = d6d.)

    However we still need to reconcile your statements with [1] which shows that AA levels are not influenced by LA intakes in humans (as opposed to rats). Time to check if any of those papers in that review controlled for insulin...

    1. I think it is premature to say the exact thing happens to humans. Would love to see homogenized liver and human brains and the omega 6 fractions and endocannabinoid fractions for sure…I also think the active endocannabinoids are a bit different in the CNS of humans as well. One main issue is that it would be difficult in the real world to make humans quite so omega 6 deficient as these experimental mice! We do know from human RBC studies that omega 3 supplementation rapidly changes the % O3 in the plasma membranes.

    2. Tell me about it! Earlier this year I had my O6:03 ratio tested after over a year of taking cod liver oil and shunning vegetables oils.

      The website for the test said that "4:1 is extremely good, and 2:1 would be close to our hunter-gather ancestors." Well, my ratio tested at 1.8:1. Total O3 of 13% (8% is good.)

      Can one have too much O3 in the body?

    3. Meant omega 3, not omega 6 in my reply, of course

  7. 1% EPA + DHA is more than a gram of each on a 2,000 calorie diet (and someone with the LA-AA munchies will eat more). Eating salmon every day, or overdosing on fish oil...
    Note that acetaminophen also has endocannibinoid activity, and childhood use has skyrocketed since 1980 (Reye's syndrome warning on aspirin).
    Endocannabinoid enhances fructose reward...

    1. Yes, I think you do need to get a considerable amount of fish oil (2 grams seems reasonable) on a high omega 6 diet…RDA is more like 400mg though

  8. We may have ALL inhaled..........

  9. Yehuda et al have done painstaking research suggesting that optimal AA:ALA intake is 4:1 to support membrane integrity and downstream eicosanoid production. It is argued that the omega 6 in commercial foods is primarily distorted/trans fats because of processing and that these 6s are not absorbed and integrated into the membrane as LA - this is a confounder in the implication of the modern SAD diet and more of a guilt by association issue than a clarification of the role of omega 6.
    The concern for the fish oil praises are that, in excess, it can competitively inhibit desaturase enzymes, AA and GLA production. AA is 12% of the brain cell membrane, and in the setting of controlled insulin, does important stuff!

  10. I'm aware 2 of corroborating finds
    1) in NASH, 3:6 ratio may be as great as 144:1

    2) obesity in the US correlates with a high gamma- to alpha-tocpherol ratio, showing high intakes of corn or soy oil which have this unusual tocopherol ratio.
    "gamma-Tocopherol levels were significantly higher in obese individuals (P < .05), whereas alpha-tocopherol levels did not differ among BMI subgroups."
    also, it may takes a lifetime, and have some gender bias:
    "In older women gamma-tocopherol and gamma-tocopherol:alpha-tocopherol ratios were directly related to indices of obesity. In young men alpha- and gamma-tocopherols were directly correlated with obesity, but gamma-tocopherol:alpha-tocopherol ratio was not."

    Conversion of LA to AA and anandamide may be stepped up after many years exposure, and of course there is more genomic variation in humans (and between races and genders) than in mice. Not everyone fattens on this diet (and some will die of other diseases first) but of those who do fatten, a substantial proportion may indeed fatten for this reason.
    What I really like about this hypothesis is that it suggests a way to reconcile FRH, CIH, CICO, and food toxins into one united theory of obesity.

  11. (6:3 ratio of 144:1 in first reference; I can see that's the wrong way round. But 144:1 - yikes!)
    I think many of the questions raised around whether this research applies to humans can be answered by Susan Allport's book on omega-3, "The Queen of Fats".
    She's not up to anandamide (the book's from 2006), but it's obvious that, as the 6:3 ratio increases, conversion of LA to ARA increases step-wise, because inhibition from ALA is lost.
    So whether humans "can" convert LA to ARA, or ALA to DHA, is probably a matter of ratios, more than genetics. Elongase, desuratase and phosphlipidase are not optional enzymes; they are non-specific but somewhat prefer ALA as a substrate.
    Many interesting details in the book; leafy greens can be an important dietary source of ALA omega 3. The highest 3:6 ratio was found in Africans eating lots of greens and cooking with palm oil, yet eating little fish.

    1. George, please read my cite above - a study published just recently shows it really is a matter of genetics, particularly for Africans - Africans tend to be near homozygous for alleles in FADS1+2 genes the encode for the most efficient conversion of LA/ALA to AA/EPA. Those genes are not as well entrenched elsewhere in the world as shown in the study.

      [3] Mathias RA, Fu W, Akey JM, Ainsworth HC, Torgerson DG, et al. (2012) Adaptive Evolution of the FADS Gene Cluster within Africa. PLoS ONE 7(9): e44926. doi:10.1371/journal.pone.0044926

  12. What are your thoughts on consuming chicken skin? Healthy or unhealthy? From a Paleo perspective, it seems that our paleo ancestors would have eaten the skins.

    Also what about GLA gamma-linolenic acid which is an omega-6? Good or no good? I've seen it as a supplement for reducing inflammation? Which is confusing because omega-6 is inflammatory.

  13. @ Blissful, a second-hand copy of The Zone Diet will tell you all you need to know about GLA. It's good as a one-off fix, but not for longer use, in my experience.

    Thanks Chris, I've read that the opposite applies to Pima peoples too. It shouldn't affect the influence of diet on the ratios, but it would make animal sources of AA, EPA and DHA more or less important for different alleles.
    This helps to explain differentials in obesity and disease epidemiology between races in the US.

    Your ref 1: short-term clinical trials don't answer the effect of LA accumulation. It wouldn't happen till membranes were saturated, which might take 6 months or years in humans.
    Also, it is the inhibition of DHA synthesis as well as increase of AA that matters.
    Some people might be more or less resistant to the effect; which helps explain why not everyone gets food-addicted by eating fast food, and why it happens sooner or later with different people.
    Let's look and see if these genes are associated with obesity in a high-LA food environment.

    FADS gene polymorphisms in Koreans: association with _6 polyunsaturated fatty acids in serum phospholipids, lipid peroxides, and coronary artery disease

    Recent insights into the relation of Δ5 desaturase and Δ6 desaturase activity to the development of type 2 diabetes

    Fatty acid desaturase (FADS) gene polymorphisms and insulin resistance in association with serum phospholipid polyunsaturated fatty acid composition in healthy Korean men: cross-sectional study

    1. Your #3 paper conflicts with #2, however #3 reflects a healthy group whereas #2 appears not to be(can't access that paper to confirm). However, in both cases, the cascade was impaired at the d5d, which meant buildups of GLA/DGLA not getting converted to AA.

      It was the AA that was protective to the Koreans, who were healthy and had high AA/DGLA ratios; and while not stated in the paper, that AA is presumably obtained from the diet.

      While on the other hand, if one is NOT getting sufficient AA from the diet, a strong d6d (acting at the end of the cascade as well as the beginning) mentioned in your cite #2 abstract may be depleting dietary AA away. I would love to get at that paper to see what the AA levels were for those in the study.

      I am really beginning to think AA is not a bad thing, its a good thing, and one either needs a strong d5d (thanks to FADS1 genetics) or needs AA supplied in the diet (and that latter might be the preferred). A strong d5d also means getting EPA (and eventually DHA) from dietary ALA to balance it out - for those fortunate enough to have a good FADS1 gene.

  14. Ok I am running across papers that discuss d5d and d6d being upregulated in the presence of insulin, and that the presence of insulin is a factor in the creation of eicanosands. This again supports my budding theory that dietary AA == good & necessary & preferred vs AA synthesized in the body via the d5d pathway which appears to be costly. Interestingly, in my cite [1] in my original post above, the authors expressed the same hypothesis of such being costly (having a genetically upregulated d5d/d6d), since those polymorphisms were actually a mutation that swept near completely (homozygously) through africa but didn't take complete hold beyond africa where the ancestral alleles held on quite well - it must have been evolutionarily costly. The map of the globe showing presence in populations elsewhere is fascinating (note, the native Americas show a complete absence of this mutation - where are they getting their AA/EPA/DHA?).

    George, I agree with you that we may actually start seeing a unification of all of these hypotheses of obesity via the pufa metabolism.

    @Emily and George, One more thought: could the "addictive" endocannabinoid's actually be nature's way of driving people to consume LA in order to produce necessary AA even if such comes at a cost? It may be what the body really wants is dietary AA, but will settle for producing its own more "costly" form.

  15. It might be that AA produced from LA concentrates, at least at first, in different cell partitions from dietary AA and that they have somewhat different localised effects. Also, LA>AA competes with ALA>EPA, while AA would not, but would still compete with EPA going into and out of membranes.

    LA> anandamide mimics cannabis in its effect on appetite. omega 3 protects against this.
    Cannabis increases propensity to develop psychosis. Omega 3 reduces incidence of psychotic episodes.
    Does omega 3 reduce cannabis munchies?

    @Chris, the absence of FADS mutations in Native Americans backs Paleo claims that these were originally highly carnivorous populations.

  16. Note that a sufficiency of EPA would reduce phospholipidase activity on AA, making less available for anandamide. There may be an EPA analogue of anandamide.
    Yes, there is:
    "In addition to anandamide (arachidonyl ethanolamide) perhaps the
    Eicosapentaenoyl ethanolamide should be tested."

  17. Hi, Emily,

    I don't know if this is happening to anyone else who has recently tried to access your blog, but Google Chrome is blocking access because it says your site has been infected, or may infect others who visit, because of being a known malware distributor.


    1. I had that same issue at one point. Not today.

    2. Her website was hacked twice, but it is fixed now.

  18. The fact that some cannabinoids are anti-psychotic while others, like delta-9 THC are pro-psychotic should warn us against seeing natural cannabinoids as all having similar effects on food reward. The EPA anandamide analogue may have oppositional effects.

    Does NAC make LA-plus-fructose less addictive?

  19. I read this and your other posts on the subject, and it seems to me, jargon aside, that what you seem to be saying is that food reward (a misleading term) is related to the omega 6/vegetable oils which, combining with the lack of satiety provided by grain consumption, leads to obesity. It doesn't have to be a large effect and easily proved, because the daily effect could be quite small. What would matter would be the cumulative effect. And most of us see that every day.

    Thus the "food reward" is not necessarily some scheme by evil corporations (although it might be) but instead could an unfortunate circumstance related to several changes in modern diet. We could propose that although wheat is a much less desirable food than some others, people might not eat so much wheat that it makes them obese were it not for the "munchies" related to the omega 6 consumption. So it isn't the carbs themselves, exactly, but rather the massively excessive consumption of vegetable oils leading to excessive consumption of carbs. (This would be a pernicious effect of people also being told to go on a low saturated fat diet. This leads of course to fewer animal fat calories and protein and more grains, etc.) When people stop eating the omega 6 fats to excess (and eat more sardines?) the system for signaling of satiety is restored to its proper functioning state. Maybe that is all we need to do? No need to go primal, etc., in terms of obesity?

    As someone amused by concepts of truth, I sometimes wonder at the hesitancy to take a risk with causality. You have pretty much investigated the matter and come to some interesting ideas. Is it proof? Concepts like proof and causality and truth are related, and are sort of ultimate and final terms, signaling the end of some quest and a statement of the stated facts matching "reality." (Whatever "reality" is, as opposed to human constructed systems, etc.) In this grail-like quest people are extremely cautious because they don't want to get it wrong. Correlation is not causation, the double blind "gold standard" etc. But a cautious statement regarding causality is justifiable at some point in this discussion. It would even be helpful.

    S. Guyenet went wrong in his statements about food reward, and became a cautionary tale, when he started to suggest that the food, or eating the food, by itself was "rewarding" and somehow made people happy. The proposal you seem to be advancing is that obesity is an unintended consequence of excessive omega 6 consumption that gives people the unthinking, hidden, desire to eat, and eat more, by interfering with satiety signaling. Neither the eating nor the food "rewards" the individual or makes them happy. If anything, such an unthinking, unknown desire is the conceptual opposite of a reward.

    It is one thing to smoke a joint and know you have done it, and know you will get the munchies, compared to consuming an omega 6 oil and not knowing you have eaten a food with certain chemical properties that make you feel hungry by interfering with proper satiety signals. Does that prove that omega 6 oils make you fat? No, but it is a link in chain of events.

    1. There's a difference between the questions I ask and statements I make on this blog, and what I do personally in life. I think it is utterly reasonable to most processed food most of the time.

      My research into food reward has been more along the lines of binge eating, which is not synonymous with obesity by any account, but is quite common in the populations I treat. In binge eating, there are small but clear randomized controlled trials showing that binge eaters have similar brain activation to those who are addicted to alcohol and drugs, for example, an excess of activation in reward areas in anticipation of eating certain foods and decrease in actual satisfaction compared to those without bingeing when eating the foods. Combine that and the "using" dreams people often have when starting a clean diet, and that's pretty interesting information.

      Is there a conspiracy to create and manufacture high reward food? Not a conspiracy, it's just business sense to make food that people will want to consume and buy more of. That omega 6 oils, cheap salt, and high fructose corn syrup make sugary-fat and salty foods cheap is just a bonus, and certainly trans fats interfering with omega 3 absorption and an excess of omega 6 could be especially disastrous.

    2. I read my post again, and maybe (obviously) what I said could be taken the wrong way, and for that I must apologize. I really like the points you were making, and the relationship to cannabis is a fascinating concept. I agree that the "food reward" business plan could exist, but I do not think that is the whole answer. Vegetable oil consumption to the exclusion of animal fats, combined with wheat consumption, is entirely in line with "official" dietary recommendations that are still in use.

      The nature of the problem is that somewhere along the way people got most of the dietary facts almost entirely backward.

      Guyenet seemed to propose, as I understood from his many posts, that people got fat because the food was "rewarding" to them. It wasn't just palatable, it was something more...something ineffable, perhaps. You on the other hand in your practice consider a specific group, binge eaters, with an addiction problem regarding food. So yes, clearly that group would have something going on in addiction centers of the brain. Maybe that really is food reward. But that is not what Guyenet was talking about, and I read many of his posts on the topic.

      Your post seemed to indicate that one likely cause of obesity was the interference of the omega 6 with the various feedback signals to stop eating, much as caused by the munchies. It is difficult for me to see how that impaired signaling could be characterized as a reward. And I do not think that is what you were trying to say. Of course, the term reward is not essential to the concept, but it was not chosen at random.

      The problem with the munchies is that nothing really makes you stop eating, except perhaps the dawning awareness that you are mindlessly eating. And eating caused by the munchies does not really satisfy. It is all a product of the satiety signaling systems not working properly. That is not a reward, and more like the opposite of a reward. The urge to eat when the system has been impaired is essentially involuntary and the feelings of hunger are real. How could we better describe the mindless over-consumption of the high fat and wheat food items you describe.

      What you were describing in you post, as I understood it, had far more to do with generalized population obesity than a small group of people addicted to food. I think the research you describe is important to understanding what is going on with obesity. Of course, it may also apply in some ways to the binge eaters you describe. What is disturbing is the stealth aspect of the vegetable oil consumption: what if people were told on the bottle (or package) that their cannabis receptors and satiety signaling systems were being influenced by the vegetable oil?

      Thank you for your posts on this topic.

  20. The "food reward" paradox might be that no food is intrinsically over-rewarding, but that some nutrients, eaten for a long time, and not especially "rewarding" in themselves, can prime us to find other nutrients "rewarding" to an extent that over-rides appetite.
    This makes sense in an evolutionary way; fructose and linoleate are "windfall" nutrients and an animal that responded to their abundance by eating more, fattening, and slowing its metabolism would be ready for winter.

  21. can someone help me to discover more details about how to find a way to having a great Flaxseed oil vegetarian with omega nutrition dishes ?

    i want to make a wonderful healthy food to my gest,
    in most cases they not much like organic food to say the list ,
    so please i need your help
    make me pursues them that health diner can go with the phrase "wonderfully lovely satisfy diner"

    flax seed benefits

  22. As for the conspiracy theory
    1) all businesses are conspiracies. That's how capitalism works; people conspire to take money from others within the law.
    2) linoleate was imposed on fast-food manufacturers in the US by the Center for Science in the Public Interest lobby group. The fast food restaurant chains did not think of it themselves; this has been an unintended consequence of the war on saturated fat. Without "Health" experts jumping the gun before the facts were known (we probably needed more omega 3s, if anything, not less SAFAs plus more linoleate) there would not be the juxtaposition of oil and fructose in MacDonalds today.
    It should have been a wonderful conspiracy, but it has the history of being a massive balls-up instead.

  23. Just thinking; early cannabis use stops young people from learning to their full capacity;
    Omega 3 is a treatment for attention deficit disorder;
    Is attention deficit disorder a result of high anandamide levels?
    Dysfunction of the dopamine (DA) system has been proposed to explain some of the clinical manifestations of attention-deficit/hyperactivity disorder (ADHD), a diagnostic concept describing children with inattention, impulsivity, and hyperactivity.1

    A complex interaction between DA and the endocannabinoid system (ECS) has been found experimentally,2 and the ECS has already been implicated in a number of DA-related disorders, such as schizophrenia, Parkinson disease, Huntington disease, and drug addiction.3 Of note, the association between ADHD and drug abuse has been described, as well as that between ADHD and abnormal activity of reward-related brain areas. These findings further suggest that the DA system and DA/ECS interaction might be altered in this disorder.1 Based on these premises, we investigated the possible alterations of anandamide (AEA) metabolism in ADHD. Notably, the endocannabinoid AEA reduces the activity of the DA transporter,4 and might therefore be implicated in the dysfunction of DA uptake mechanisms involved in ADHD pathophysiology.1

    Of course, dopamine disregulation also relates to food reward.
    And ritalin and amphetamines are dopamine agonists.

  24. I found this longitudinal study on depression and dietary patterns interesting...

    "The highest quartile of low-fat, western, high snack and high fat-sweet diets in men and low-fat and high snack diets in women were associated with higher likelihood of depressive symptoms.."
    "Conversely, the highest quartile of traditional diet (characterized by fish and fruit consumption) was associated with a lower likelihood of depressive symptoms in women compared to the lowest quartile,... as the healthy pattern (characterized by vegetables consumption) ... in men and women, respectively. However, there was probably a reverse causality effect for the healthy pattern.

  25. Look at this:

    The rise and fall of CB1 receptor antagonists: possible future perspectives
    György Bagdy
    Department of Pharmacodynamics, Semmelweis University, 1089 Budapest,
    BMC Pharmacology 2011, 11(Suppl 2):A55

    Abstract: Cannabinoid type-1 (CB1) receptor antagonists were among the most promising drug targets in the last decade.
    They have been explored and found to be effective as therapeutic agents for OBESITY and related cardiometabolic problems, including e.g. dyslipidaemias, diabetes, and metabolic syndrome. However, the use of rimonabant, the first marketed CB1 receptor antagonist, has been suspended due to its anxiogenic and depressogenic side effects, which were present in about 20–30% of the patients, i.e. a 2.5–3-fold increase compared to placebo.

  26. Rimonabant was designed to block CB1 receptors in the brain and thus treat obesity by reducing appetite, however many of the benefits are now believed to be due to the drug's effects on peripheral organs such as the liver and adipose fat tissue. There is now an abundance of published scientific literature which supports the role of the peripheral CB1 receptor in the development of the cardiometabolic problems associated with obesity including T2 diabetes and liver damage. There is also evidence that these peripheral receptors are responsible for changes in energy balance that leads to much of the weight gain - but that is more contentious.
    Obesity is, in most cases, linked to hyperactivation of the cannabinoid system - that is increase in amounts of both the endocannabinoids and the receptors. The mechanisms are well described by which CB1 receptor in particular causes increased fat and glucose production, inflammation and cell death (apoptosis) leading to fibrosis and tissue damage in organs including the liver, heart, kidneys, pancreas and eyes.
    It is quite probable that hyperactivation of the cannabinoid system may be induced by high ratios of omega 6/omega 3 fatty acids in the diet, but this may not relate directly to the amount of arachidonic acid-containing phopholipids in cell membranes. It is possible that it is the production of reactive oxygen species during the metaboolism of different fats and sugars that is the determining factor. It is also known that high amounts of fructose and alcohol as well as hepatitis C virus can have similar effects - all resulting in similar liver pathology mediated by the CB1 receptor.
    Unfortunately, it seems likely that, following the withdrawal of rimonabant, the Pharma industry will be too nervous to revisit the CB1 area, even though the development of drugs which do not enter the brain should be possible. So sadly the real value of such a drug for so many people in the 21st century will probably never be realised.