A new study was just published on the health effects of a low-carb/high-fat (LCHF) diet in mice. http://www.nature.com/nutd/journal/v6/n2/full/nutd20162a.html.
It has been getting a lot of media attention with headlines touting things like
“New study says Paleo diet ‘unhealthy and fattening’ angering ardent devotees”.
The above was from the Telegraph: http://www.telegraph.co.uk/news/worldnews/australiaandthepacific/australia/12164394/New-study-says-Paleo-diet-unhealthy-and-fattening-angering-ardent-devotees.html.
Given that a LCHF paleo style diet (PSD, more primal due to inclusion of dairy products), I was skeptical, so I downloaded the paper and supplementary table.
Turns out I had a right to be skeptical.
Exhibit A: From the methods section: “The carbohydrate content of the LCHFD was exclusively derived from simple sugar (sucrose: 106 g kg−1)”
Doh! what a major flaw in this study! A healthy LCHF PSD like the one I and many of my friends and colleagues follow minimizes natural or added sugars, especially from refined sources like sucrose, and the relatively small portion of carbohydrates we DO include in our diets are from whole-food sources and tend to be of the complex starch variety, such as tubers, plantains, root vegetables (beets, carrots, etc.), and some fruit (in moderation). That is, we follow a WHOLE FOODS diet from ingredients that receive relatively low amounts of processing.
The diet used by the researchers was made of very highly processed ingredients. I’ve published evidence in a wild-type rat model showing that it is the degree of processing, irrespective of macronutrient ratio, that seems to be heavily implicated in health outcomes, including cognitive function: http://www.sciencedirect.com/science/article/pii/S0031938414000833.
Here’s the supplementary table showing the actual ingredients they used in their study:
The ingredients were mostly derived from highly processed sources. For example, the protein comes completely from casein, which has been shown to make mice highly susceptible to aflatoxin-induced cancer: http://www.westonaprice.org/our-blogs/cmasterjohn/the-curious-case-of-campbells-rats-does-protein-deficiency-prevent-cancer/. Hat tip to Chris Masterjohn for his detailed analysis of this literature.
Exhibit B: the mice used in the study were genetically modified to develop obesity. This is the topic of a whole other blog post, but it is often the case that mice that have been genetically modified to mimic a disease or disease process found in humans, turn out not to be a good model for the mechanism of the disease in humans.
Exhibit C: perhaps even wild-type mice would be a bad model for studying the health effects of a diet for humans because they are adapted to quite a different dietary niche than are humans. A great argument for this has been made by my friend and colleague Richard Feinman: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3488544/.
They summarize in their abstract: “In sum, the response of mice to a carbohydrate-free diet was greater weight gain and metabolic disruptions in distinction to the response in humans where low carbohydrate diets cause greater weight loss than isocaloric controls. The results suggest that rodent models of obesity may be most valuable in the understanding of how metabolic mechanisms can work in ways different from the effect in humans.”
This statement matches my own observation in the effect of a LCHF PSD in myself and others (though not everyone, which is an important point that we should not discount individualized approaches to diet and health).
Of course the news stories everywhere are claiming “Paleo diet shown to be harmful!”.
*sigh*
Professor Aaron P. Blaisdell
aaron blaisdell's blog 
I’m curious as to why the LCHF diet in the study had ~60% higher energy density:
“The LCHFD contained 24 MJ kg−1 digestible energy” and
“The chow diet contained 13.5 MJ kg−1 digestible energy”
Since the mice were allowed to eat as much as they want, wouldn’t this skew the LCHF mice to ingest more calories total. Also isn’t the Paleo diet generally less energy dense than regular diets
LikeLike
Even in my own rodent studies of different diets, it’s very difficult to control for factors like energy density. I’m trying to find a way to create two diets that are matched as closely as possible on ingredients, with the only difference being that the whole-foods diet comes from whole-foods ingredients, and the refined diet comes from those same ingredients that have been refined, purified, and then reconstituted to rebuild the same “food”. This would allow the study of quality of food (in terms of amount of processing the ingredients have gone through) separately from all other factors, including vitamins, minerals, energy density (presumably), etc.
LikeLike
I think it’s worth noting too that the low carb arm got 20% calories from carbohydrate and in a 2000 calorie diet, that’s almost 100g of carbohydrate which isn’t low enough to be “low carb”.
LikeLike
Indeed. And ALL of the 100g of carbohydrate was from sucrose! Rats tolerate starch quite well. They’re granivores/omnivores, so they even handle whole grains such as wheat and corn quite well. Humans, on the other hand are animal-based omnivores. In the absence of underlying disease states such as autoimmunity and metabolic syndrom/diabetes, we actually handle whole foods starch quite well, and it was a significant part of our diet during evolutionary history following the habitual use of fire for cooking. But sugar is bad for both rodents and humans.
LikeLike
“the low carb arm got 20% calories from carbohydrate ” Huh??? “SUPPLEMENTAL TABLE 1” states: Carbohydrate, % energy 6.0
LikeLike
Hmmmm, it appears it’s being reported about differently– here the low carb arm is being reported at 20%. I didn’t have access to the full text when I went looking for it.
http://medicalxpress.com/news/2016-02-diabetes-expert-paleo-diet-dangerous.html
“They took two groups of overweight mice with pre-diabetes symptoms and put one group on the LCHF diet. The other group ate their normal diet. The mice were switched from a three per cent fat diet to a 60 per cent fat diet. Their carbs were reduced to only 20 per cent.”
LikeLike
If you turn to the reference 26, Kluth et al, in that paper you will (I think) find an example of how to control for dietary variables in this sort of experiment as perfectly as possible, and also a very good lesson in how not to draw outrageous conclusions from exciting results.
https://profgrant.com/2016/03/18/more-on-mice-humans-and-diets-for-diabetes/
LikeLike
Thanks for sharing. The nitty gritty matters in how we interpret the mechanism being identified in an animal model, especially one that has been selectively bred or genetically engineered to have a very specific phenotype. They can certainly be useful, but not often in the way that the experimenters wish them to be.
LikeLike