Sunday, May 31, 2015

Beyond Macronutrients

The Academy of Nutrition and Dietetics (AND) recently responded to the Dietary Guidelines Advisory Committee’s (DGAC) scientific report.  In their response, the AND praised the DGAC in general and for removing cholesterol from a ‘nutrient of concern’ and mentioned the problems related to population wide recommendations to lower salt and the limitations related to SFA and CHD.  They instead recommend that carbohydrate should be replaced by PUFA.  This is certainly a more defensible position from the evidence in observational studies and is likely to translate to better health outcomes if added sugars and refined starches are the carbohydrates that are being replaced* [1]

I imagine some people will challenge the recommendation to replace carbohydrate with PUFA, arguing that ‘not all carbohydrates are equal’ or that ‘there are good carbs and bad carbs’, because this is exactly what happened when studies were published showing no difference in CHD between carbohydrate and SFA.  Yet this argument very rarely gets applied to SFA, PUFA or other nutrients

In a previous post I suggested that the current RCTs aren’t very good and that a valid and fair test of replacing SFA with PUFA could help resolve any remaining debate (blog).  But what does this look like in practice?  In the table below I have 6 different hypothetical scenarios of RCTs to test whether carbohydrate (CHO), SFA, MUFA or PUFA are healthier, and the likely winner in the right column (don’t get too caught up with the examples, that’s not the point)


CHO
SFA
MUFA
PUFA
Winner
Scenario 1
White rice
Butter
Olive oil
Corn oil
?
Scenario 2
Fruit
Butter
Olive oil
Corn oil
CHO
Scenario 3
White rice
Cheese
Olive oil
Corn oil
SFA
Scenario 4
White rice
Butter
Avocado
Corn oil
MUFA
Scenario 5
White rice
Butter
Olive oil
Nuts
PUFA
Scenario 6
Fruit
Cheese
Avocado
Nuts
?

These scenarios remind me of the debate whether chronic (not postprandial) SFA intake impairs endothelial function.  There were two trials: one which concluded SFA impair endothelial function, but half the intervention was not a fair comparison for the SFA group (butter vs. almonds vs. walnuts vs. sultanas), so it’s hardly surprising it did worse [2]; and the other trial which had a fairer comparison and concluded SFA don’t impair endothelial function [3] (blog)

The diet heart trials were mainly in scenario 1**.  The meat, full fat dairy and added fats in the control group were replaced by filled meats, filled milk/cheeses and vegetable oils in the experimental group.  This aspect of the trials is a fair comparison, as the other nutrient/chemical differences between the fat sources is relatively minor (vitamin A, butyrate and CLA vs. vitamin E and phytochemicals that make it through processing).  However, it could be argued that it wasn’t valid comparison because we should base our diet on whole foods and shouldn’t have much added fat anyway (which was where most of the replacement occurred) (which I kind of agree with)

Alternatively, let’s say to improve validity we would by conduct a clinical trial with one group having extra cheese and another group having extra nuts to test the relative effects SFA and PUFA have on CHD (scenario 6).  We are certainly comparing cheese with nuts, at least in the context of the background diet and lifestyle, but are we really comparing SFA with PUFA?  Surely the sum of all the other nutrients and other chemicals in cheese and nuts has a far more meaningful impact on CHD than any potential effect of SFA or PUFA (though I doubt either has much effect within reasonable limits).  In this example it’s quite likely that the other nutrients/chemicals in cheese or nuts will bias the results in favour of one of the groups, therefore not making it a fair comparison.

This presents a dilemma for nutrition research of this kind.  If we have a hypothesis that replacing SFA with PUFA reduces CHD, then a clinical trial replacing butter with margarine could be accused as having poor validity, while a trial comparing cheese with nuts could be accused as not being a fair comparison.  Either way, the trial has questionable generalisability to other whole foods (due to the unique mix of other nutrients/chemicals in the whole food likely having a greater effect than SFA or PUFA)

Perhaps the solution here is: (1) to simply base our diets on whole, minimally proceed foods, which in almost all cases are superior to refined foods; and (2) for nutrition research and recommendations to be based on food rather than nutrients.  This simplicity is also likely to improve adherence (ego depletion and decision fatigue (blog)) and circumvent the processed food industry who can capitalise on reformulating and marketing products to meet the ‘nutrient of the month’

* The response also discusses the limitations of ‘eating patterns’, which is a very important part of the response that I think has been quite underappreciated

** Excluding differences in TFA intake (all of them), reducing baked goods/junk food as a means to reduce SFA intake (RCOT, MRCT, DART), multifactorial diet interventions (ODHS, STARS), vitamin E deficient control groups (LAVAT) and differences in cardiotoxic medication (FMHS)

4 comments:

  1. If you excluded rapidly absorbed carbs and sugar, you'd find that nothing in the whole food category was harmful.

    ReplyDelete
    Replies
    1. Are you suggesting that fruit and high GI root vegetables are harmful?

      Delete
  2. Here's a good example of what you mean - it is, I think, the first ever low fat diet trial, by Lester Morrison in 1951.
    It significantly restricts fat, cholesterol, calories, and refined carbohydrates.
    https://www.dropbox.com/s/ekn8emrvpca9d7g/morrison1951.pdf?dl=0

    https://www.dropbox.com/s/rwa4411pkpapkaq/morrison1960.pdf?dl=0

    Of note is the baseline weight of these patients - 75 Kg for men, 64 Kg for women, which dropped by 8-9 Kg over the trial. Which, unless they were midgets, is a simple trip across the "normal" weight range today!

    You might have some idea how many calories, and how much carbohydrate, the controls were likely eating in Los Angeles in 1951.
    It seems to me that much of the CHO in the test group's diet was fibre. 3 teaspoons of sugar was allowed at breakfast.
    Lester Morrison can't have been satisfied with the low-fat diet, because he went to experiment with chondroitin sulfate A for secondary prevention and got even better results than these.

    https://www.dropbox.com/s/17vsix1d1qbctft/morrison1973.pdf?dl=0

    This is plausible, if you think that the glycocalyx of the coronary vasculature might be important in CHD etiology. Morrison certainly did a great deal of animal experimental groundwork to support his later hypothesis. I wonder what the modern view of this is. It fits in with a paleo prescription that includes connective tissue and bone broth, sources of CSA.

    http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1915585/

    ReplyDelete
    Replies
    1. From the text it seems like the calorie target was 1500 and 1200 for those with high cholesterol. Which makes sense because 75/64 kg is generally not overweight. I agree, it looks like almost all the carbohydrate was fruit, veg and whole grains. Fibre intake could have been 25-35g, which is much higher than the (current) US average of ~15g

      If you thought fat and cholesterol were 'the' cause of atherosclerosis, then surely that almost no fat no cholesterol diet should have achieved more. Not to mention the difficultly in getting people to adhere to that kind of diet and calorie restriction

      That endothelial glycocalyx review is interesting. The response to retention hypothesis never sat comfortably with me; surely the endothelium would protect itself from penetrating LDL particles. This is probably one the factors behind paradoxes whereby metabolically people with high LDL-P show no signs of atherosclerosis

      Delete