Sunday, June 26, 2011

Diet and Dementia - A New Study Part 2

Last week I introduced us to a new diet and dementia paper.  For the most part I presented the results and fumbled around with the interpretation, but didn't quite know what to make of it, given a lack of details about the actual diets in the paper.  I contacted the researchers, who were kind enough (in fact, they were very kind in their emails!) to send me a bit of information about the methods and a sample daily diet.  So now I can fumble around some more.

Let's start with Big Picture context.  I tend to forget sometimes that thoughtful and well-meaning people still believe that Animal Fat is Bad and Simple Carbs are Bad and that we should all be eating cardboard (er, I mean fiber-rich foods), skim milk, olive oil (in moderation), and skinless chicken breasts.  I'm also perhaps a little too comfortable with the fact that not only do thoughtful and well-meaning people have that view, but that they think my view (that Animal Fat is Good and Macronutrients Aren't as Important as You Think Absent Insulin Resistance or Other Issues Such as Dementia In Which Case You Should Be Eating More Fat) is lunacy.  I mean, my kids get whole milk yogurt still, and they are both over two, the age when fat becomes Bad according to the USDA.  (Okay, I'll stop with the capital letters now…)  

Part of the reason that I'm comfortable being thought a lunatic is that I think most people secretly believe psychiatrists are lunatics anyway.  There is almost always a bit of a double-take when I tell people my profession.  In medical circles, I will get a "oh, she's one of those" sort of look.  Among lay folk (at a party, for example), people will suddenly clam up or begin telling me their life stories and all about the different psychiatric medicines they've tried.  Psychiatrists are a bit of a puzzle - what is it we do, anyway?  I think of what psychiatrists do thusly: we look at big, complex puzzles comprised of experience, world view, psychology, environment, and biology and try to figure out how to help someone (typically presenting with emotional issues, though this day and age it could be anything) manage all the factors that could be contributing to the problem.  That holistic view of problem-solving health problems makes a psychiatrist somewhat uniquely suited to examining Evolutionary Medicine, but perhaps that is my bias.  :-)

Speaking of bias, this week I read with great interest (as always) Paul and Shou-Ching Jaminet's Perfect Health Diet blog, particularly Blood Lipids and Infectious Disease Part 1.  In the comments I linked to an commentary on a study about cholesterol and all-cause mortality in Honolulu from the Lancet, the money quote being here:

Sir—The central, surprising, finding from our study was that low serum cholesterol, persisting for 20 years, increased subsequent all-cause mortality. We have no logical explanation for this finding, nor, apparently, do any of our correspondents.

Here is a graph of their data.

And another money quote (these are both from the author's letter replying to other commentary about the findings):

We agree that the association between low cholesterol and mortality is only that; we are sorry that Cuchel and Rader infer a causal relation, since that was not our intent.
Yes, a correlation is merely a correlation.  But when correlations are always pretty darn consistent (and the most recent PHD post links to 50 studies showing the same correlation, especially true in the elderly), it is important to put our thinking caps on and try to figure out why (correlatively-speaking)  higher cholesterol doesn't seem to be killing people as fast as having low cholesterol does, despite the fact that health-minded people have been busy eating good diets and taking meds to lower that cholesterol for a generation.  Instead of putting on the thinking caps, the researchers in the Lancet conclude "we have no logical explanation" and timidly suggest that the elderly not be treated quite so aggressively with cholesterol-lowering medications.  Which is nice and conservative of them, and fine, but how about a sentence with some thought about why cholesterol might be good for us, since our bodies go through all that trouble to make it, after all, and how we might go about proving causality from these correlations?

So, with the world view of the conventional nutritional researcher firmly in mind, let's go back and look at the Diet and Dementia paper and the actual diets.

A reminder, in this study, 20 healthy older adults (mean age 69) and 29 adults with mild cognitive impairment (mean age about 68) were put on a LOW (low fat, low glycemic carbohydrate) diet or a HIGH (high fat, high saturated fat, high glycemic carbohydrate) diet for 4 weeks.  A number of biomarkers related to cholesterol and dementia were measured, and the researchers felt it came out as a clear win for the LOW diet.

I listed most of the results in my previous post, and what I think is the most interesting finding of the paper is how the mild cognitive impairment patients responded quite differently (typically, in more extreme fluctuations of biomarkers, and sometimes opposite trends) to the dietary changes than the healthy controls.  For simplicity's sake, I will focus just on the mild cognitive impairment patients in this post.

First, more details about the diets, about which we were told the macronutrient ratios and the fact that food was delivered twice-weekly to the participants. Here was my question to the researchers:

Would you be able to be more specific about the diets themselves?  For example, was the HIGH diet carbohydrate primarily sugar or starch?  Were the foods more heavily processed (for example, the equivalent of Jenny Craig meals delivered twice weekly) or more whole foods?  Were the saturated fats animal fats, trans fats, or coconut or palm oil? 
And the initial reply:

The HIGH diet carbohydrates were a combination of sugars and starches. Some of the foods were processed, but many were not. The saturated fats were primarily animal-derived.

The researcher cc'd the nutritionist from the study, who very helpfully took time out of her day to send me a sample day's menu and some more about the methods.

Please see attached for a copy of a sample day from each diet. To keep participants as blinded as possible, we deliberately included food items on both diets that seemed healthy/unhealthy. 

And the menus:
LOW diet
Egg-white and lowfat cheese on a sprouted wheat English muffin
Fruit cocktail, canned in juice
Lowfat fruit yogurt
Whole grain granola
Grilled low-fat mozzarella, tomato, and basil sandwich on sourdough bread
Beef and barley soup Peanut M&Ms
Whole grain Chex with raisins, cashews, and chocolate chips
Roasted pork tenderloin with roasted apples and onions
Bulgur wheat with apricots and almonds 
Green beans
Blueberry crisp with lowfat whipped topping
HIGH diet
Buckwheat pancakes with butter and sugar- free syrup
Milk, 2%
Whole milk mozzarella cheese 
Saltine crackers
Grilled turkey and cheddar on white bread 
Cream of Tomato soup
Honey roasted Chex mix
Beef stroganoff with rice noodles Broccoli with butter
Nilla wafers 
Vanilla ice cream

Okay.   From a "paleo" perspective  the  difference I can see between the two diets (neither of which I would eat personally) is, indeed, more simple starches and sugars  in the HIGH diet and more animal fat in the HIGH diet (butter, milk-fat, and cream).  Both diets are chock full of grains and what I would consider "industrially processed" food rather than whole foods.   I would knock the HIGH diet for the sugar and the refined flour and sugar-free syrup.  I would knock the LOW diet for the use of fake fats (lowfat whipped topping) and grains.  I could see how both could be toxic to our delicate physiology.

With that in mind, let's take another look at the biomarker changes and interpretations for the cognitively impaired patients before and after four weeks on the diet:

The HIGH diet increased insulin resistance and that LOW diet decreased it.   The HIGH diet was 45% fat (25% saturated fat) and 35-40% high glycemic carbohydrates, compared to the LOW diet ( 25% fat (7% saturated fat), 55-60% carbohydrate).  I'm not sure what to make of that finding.  On the positive side for the LOW diet, eating more cardboard and less sugar increased insulin sensitivity.  Another interpretation is that lowering the carbs in the HIGH diet increased physiologic insulin resistance, but it certainly wasn't a "low carb" diet by any means.  

In the HIGH diet LDL increased by 20 points in the MCI patients, and on the LOW diet it dropped by 10 points.  Most modern nutritional researchers would call this finding a clear win for the LOW diet - keeping in mind the physiologic findings in dementia and the perspective and insight of Stephanie Seneff, however, I'm not so sure.  In dementia, lower lipid levels are found in the spinal fluid compared to non-demented controls.  To me, the jump in LDL (which was actually much higher than the jump in the healthy controls) could be the MCI patients way of trying to make the most of the animal fat and getting desperately needed saturated fat, fat soluble vitamins, CoQ10, and cholesterol into the besieged brain.  But then, everyone knows I'm a psychiatrist so I must be a lunatic…

On the LOW diet, HDL went down by 8 points in the MCI patients, and on the HIGH diet it went up by 4 points.  Everyone would have to agree this is a likely "win" for the HIGH diet.  I consider HDL a marker for rapid cholesterol and fat turnover, which I would consider good.  The less old rotten oxidized fats floating around, the better.

Then we get to some more specific biomarkers for dementia - and here is where you get to a very confusing picture comparing MCI patients to controls.  Some of these markers went down in controls (which the researchers considered bad) and up in MCI patients (which the researchers also considered bad.)  Their reasoning followed an "inflection point" model, where people developing dementia have increasing levels of certain biomarkers, to a point where dementia is inevitable, at which point the levels of these biomarkers begin to drop as dementia progresses.  Therefore, an increasing level in a healthy control would be bad (bringing you closer to the point of no return) and an increasing level in a dementia patient would be good (bringing you back closer to the inflection point and closer to health).  LDL is a good example.  LDL levels tend to be higher in carriers of ApoE4 (the risky genotype for developing Alzheimer's - Dr. BG has done some great posts about E4 recently), but LDL levels tend to drop before dementia develops.  

I think a higher LDL might be representative of the brain and body doing it's best to fill the brain with yummy delicious brain fat and nutrients.  In the conventional sphere, since high LDL is Bad, high LDL must be part of the pathologic process destroying the brain (probably through inflammation, though I haven't been sold on any reasonable physiologic method by which higher sat fat could cause inflammation - lipotoxicity seems to depend on hyperglycemia).  Conventional wisdom has no explanation for why lipids drop prior to the development of dementia, which is why it is considered reasonable to give someone with dementia statins.  Which I think is a bit of lunacy.  But once again… I'm the lunatic here.  Never forget that.

Back to the paper!  ApoE levels go down on the HIGH diet and increased on the LOW diet in MCI folks - again, this would fit the model that ApoE takes cholesterol into the brain, so you have to be more aggressive about scavenging cholesterol on a low fat diet.  The researchers don't believe cholesterol goes from the circulation into the brain and link some papers with that view.  Senoff believes differently and links other papers and describes the mechanism by which ApoE helps the astrocytes escort LDL from the circulation across the blood brain barrier.  Given that autopsy studies show that cholesterol levels in the CSF are similar to those in the circulation, it seems logical that these two systems have ways of communicating and equalizing.  Well.  We'll have to let the separate sets of researchers duke it out on that one.  

On the LOW diet, amyloid levels went up.  On the HIGH diet, they stayed the same.  (I think this is a win for more fat, but the researchers interpreted that differently as one of those markers related to the inflection point, and they went so far as to explain that for this biomarker, the HIGH diets probably couldn't do more damage in an "already extant" process in the brain, whereas in the controls, where amyloid went up on the HIGH diet, the HIGH diet was clearly damaging).  

F2-isoprostanes (derived from omega 6 fats and a marker of inflammation) decreased in the LOW diet and stayed the same on the HIGH diet in the MCI folks.  The HIGH diet increased F2-isoprostanes in the controls.  Since a 45% fat diet designed by a nutritionist probably has more omega 6 than a 25% fat diet (but who knows, the 45% fat was 25% saturated fat, the 25% diet only 7% sat fat, so maybe it is a wash. Depends how much mono they ate) - and there is also all that sugar in the HIGH diet, so I don't even know how to look at inflammation in these diets.

So, what does it all mean?  What I see, with my biases and knowledge base, is some evidence showing that increasing fat in the diets of MCI patients is probably a good idea, though I can see how both diets could be inflammatory and damaging so I wouldn't personally pick either.  The researchers, with their biases and knowledge base, feel that the HIGH diet is a disaster and that this study presents strong preliminary evidence that the LOW diet is preferable for everyone, MCI patients and controls.  The researchers have PhDs in nutrition - I'm just a clinical psychiatrist with a hobby (oh, wait, I'm a professional nutrition writer now too!  I keep forgetting that…).  You decide whom you want to believe.


  1. Fist, congratulation for your blog. It's very interesting.

    To the topic, I think high fat diets has a problem and it is that food fat in convetional food, food that you can easily get in any market, isn't very good. You know too much omega 6 fatty acids, oxided fatty acids, trans... For this reason, in order to plan a high fat diet is more difficult and it is easy to introduce unhealthy fats.

    There are a lot of studies that fail in this topic and conclusions are not favorable to high fat diets but all of these problems are caused by mistakes in the selection of the food, specially, fat.

    Sorry if I don't write so good. English isn't my fist language...


  2. Hi Emily,

    Actually, I'm surprised they're so mystified, it's easy to understand the epidemiology when you consider the immune functions of LDL, which I'll be blogging about in the current series.

    Briefly, LDL is designed to pick up fat-soluble toxins from pathogens, such as lipopolysaccharide. When it picks up these endotoxins it gets oxidized so that it is no longer picked up by the LDL receptor but instead by scavenger receptors on macrophages. The macrophages extract the endotoxins from the oxLDL particle and present them on MHC molecules for antibody generation.

    When infectious burden becomes large enough, LDL particles increasingly get oxidized and destroyed in this pathway, and LDL levels drop.

    Since dementia largely results from brain infections, high infectious burden produces both lower serum LDL and dementia simultaneously.

    Dietary manipulations of LDL level are probably mostly tangential to this process, but alterations in serum LDL and HDL will affect immunity to some degree.

    Best, Paul

  3. David, I agree that a major problem with a conventional low carb diet is the lack of attention to massive overages of omega 6.

    Paul - you are an exceptionally thoughtful person. Sometimes you have to put on the boots of someone mired in the lipid hypothesis to understand the interpretations of these papers.

  4. See an excellent post by Ned Kock on the relationship between heart disease and cholesterol at

  5. If science hasn't figured it out, I'm siding with nature.

  6. Thanks! I think both diets look pretty dreadful, and I think I would be tending towards some type of mental issue if I followed either of them.
    (AND, thanks for bringing UP the CAPS issue - you must have read my blogpost on the practice.)

  7. The omega six content is a huge factor for CNS degenerative disease. Since a great part of my practice is spent taking care of these folks I do omega six/three ratio's on them all and they are horrible across the board. The interesting thing arguing against Paul's theory is I do draw HS CRP on them all with many other labs.....and 35% come back below 2.0. But their clinical disease state mirrors the omega six levels. It is something I now pay a great deal of attn to and make sure my neurologists are aware of the findings. They too are now putting all of our degenerative neuro patients on high fat ketogenic diets we use in brain tumor patients. Nice job Emily!

  8. Putting aside the simple focus on macronutrients and a bit of a disagreement about what I would consider unprocessed foods - a major weakness of this paper is a lack of attention or commentary on O6/O3. There is a reasonable literature on that subject and dementia, with review articles and many RCTs for MCI. Even mainstream nutrition researchers should not miss that one.


Tired of receiving spam comments! Sorry, no new comments on the blog

Note: Only a member of this blog may post a comment.