9 feb. 2017

Has Gary Taubes' sugar hypothesis been disproven?

Here is my (Bas) reply to claims that Gary Taubes' sugar hypothesis has long been falsified. The claims are quoted in grey with replies below them in regular text.

TLDR: Sugar is not single handedly and uniquely fattening. The method by which this is claimed to work does not work as pictured by Taubes. The primary causes of the diseases that follow from eating sugar according to Taubes are already established (and it is not sugar).

TLDR: The evidence you provide for the first statement does not show that. The method (de novo lipogenesis) works amazingly well. Zero evidence is provided for the last statement.

The claim is sugar is single handedly and uniquely fattening.
It would mean you cannot find another food group more fattening, but:
in these mice studies fat is more fattening then sugar and fat+sugar is more fattening then sugar:

What works for mice does not necessarily work for humans. Note that these mice are specifically bred to get fat and sick. So these results say nothing about normal healthy mice, let alone normal healthy humans.

Also the fact that foods exist that are more fattening than sugar is hardly evidence of sugar's innocence.

People eating lots of sugar would have to be fat otherwise it would not be uniquely fatterning, but https://www.ncbi.nlm.nih.gov/pubmed/24746602 (observational study lean ppl with high sugar diet)

According to that study the Hadza eat about 14% of calories in honey. Honey is about 70% sugar and the rest mostly water. 70% of 14 is about 10%, which is the amount the World Health Organisation recommends not to exceed. I would not consider that to be very much. And of course not everybody is fat on a high sugar diet, just like not everybody who smokes has lung cancer.

Bonus if you like rodent studies, in rats honey seems to be less bad than sugar:

Randomized controlled intervention studies with the Kempner white rice, fruit juice, and sugar diet where people lose weight while eating a normal amount of calories

Absolutely NOT a randomised controlled trial, as clearly stated in the article linked. This was apparently such a boring diet that there were a few cases where Kempner had to whip his patients to stay on it. I don't think the overweight patients on it ate a "normal amount of calories". See also here: https://intensivedietarymanagement.com/thoughts-on-the-kempner-rice-diet/ (don't miss the update at the end of the linked article)

http://diabetes.diabetesjournals.org/content/55/12/3566.long (Randomized controlled trial high-sucrose diet result: no change in weight)

In six weeks. In 13 healthy male subjects. So a short study on a few people, no wonder no significant changes in weight were seen. Non-significant average changes in weight: 10% sugar -0,2 kg, 25% sugar +0,2 kg. And the sugar in the "low sugar" diet of the study was 10% of calories, which in the Hadza observational study above you describe as "lots of sugar".  The study was funded by The Sugar Bureau and Suikerstichting, who I am sure, have been very happy with the results.

https://www.researchgate.net/publication/235384801_Comparison_of_5_versus_15_sucrose_intakes_as_part_of_a_eucaloric_diet_in_overweight_and_obese_subjects_Effects_on_insulin_sensitivity_glucose_metabolism_vascular_compliance_body_composition_and_lipid_ (randomised controlled trial. 5% or 15% sucrose diet no effect on weight)

Study from the same university, with the same length and three of the same authors of the previous study. This time with 17 subjects and slightly different amounts of sugar. With similar non-significant weight changes as before, so perhaps they should pool the results in a meta-analysis and find significant weight changes? This one was funded by Sugar Nutrition UK. Did they think something was wrong with the previous study that they had to do it again? Very strange.

So, animal, observational and intervention diets all pointing in the same direction. Sugar can do a lot of things, but on it own it is not uniquely fattening.

That is not a conclusion I would draw from the evidence provided.

De novo lipogenesis, which is the preferred method operandi of the villain in Taubes fantasy books, is also virtually impossible:

Here's what happens: The body preferably metabolises sugar immediately or stores it as glycogen. Largely shutting down fat oxidation while it does so, so any fat also ingested will be stored. Only after the glycogen stores are full and the body can't use the glucose for energy anymore, de novo lipogenesis starts ramping up. DNL is a method of last resort, not "the preferred method operandi" as you say.


Short term trial (4 days) but "De novo lipogenesis was 2- to 3-fold higher after overfeeding by 50% than after the control treatment in all subjects." So there is definitely something working. This study only measures hepatic (of the liver) de novo lipogenesis and not in adipose tissue. But still the study finds 10 gram per day DNL, which is about 3,5 kilos yearly. That looks like enough to get fat to me.


Not a study, but an editorial on the study above.


Another similar study, with the same author. Why the duplication?


A study about the influence of 24 gram of alcohol on DNL. Interesting that alcohol does not seem to stimulate insulin secretion, while it does largely shut down fat oxidation, but otherwise seems unrelated to the subject at hand.


Study on non-overfed subjects consuming a single carbohydrate load. Conclusion: if you don't eat a lot, you don't have a lot of de novo lipogenesis. Seems logical.


Review article from the same author as the previous article. Unsurprisingly seems to come to the same conclusion.

Let me show you a study where they manage to stimulate DNL to a ridiculous 150 grams a day, using massive carbohydrate overfeeding:
Skinny to morbidly obese within a year, if they would continue beyond the 14 days of the study, and that is not even counting storage of dietary fat. Not "virtually impossible", possible almost beyond belief. But where else would the glucose go when it can't be oxidised and can't be stored as glucogen? It's either DNL or you end up a diabetic and pee it out.

There are no RCTs supporting Taubes' sugar claim at all AFAIK, but there are some observations studies linking sugar to weight gain. Looking closer they have a confounding factor 'total calories'.

Yes, it is not for nothing that Gary Taubes keeps asking for RCTs to be done.

Sugar among other things stimulates appetite, leading to overeating. You do need to consume some calories if you want to get fat. However I of course do not mean to imply it is a question of calories in, calories out. Well, it is, the same way you need to earn a lot of money if you want to be rich. Without money, no richness. All poor people should just earn more money. Earn more, spend less. Money in, money out. Obviously this isn't a solution for poverty. And neither is it for obesity.

http://www.bmj.com/content/346/bmj.e7492 (meta observational study. Conclusion its not sugar but total calories)

Also in the conclusion "Among free living people, advice to reduce free sugars was associated with an average 0.80 kg reduction in weight; advice to increase intake was associated with a corresponding 0.75 kg increase". Which is Taubes point.

This is also just observation (about observation), so to back it with an intervention study:
https://www.researchgate.net/profile/Richard_Surwit2/publication/14121677_Metabolic_and_behavioral_effects_of_a_high-sucrose_diet_during_weight_loss/links/0f3175320cba9ef86b000000.pdf (intervention study conclusion: its not sugar but total calories)

Poor women on an 1100 calorie starvation diet. Nobody denies that you lose weight if you starve yourself. Spend more money than you make, you get poor. Funded by The Sugar Association, Inc. and the Kellogg Company. What a coincidence that so many "sugar is innocent" studies are funded by sugar companies.

Also the very often cited 1970-2000 correlation of carb/sugar sales and obesity starts falling apart after 2000 http://wholehealthsource.blogspot.nl/2015/11/carbohydrate-sugar-and-obesity-in.html

The infamous graph with the highly misleading axes using questionable data. Smoking started to decrease from 1965, but lung cancer incidence kept increasing for 20 more years. And only began decreasing another 10 years later. No RCT's were done on smoking, so can we conclude cigarettes are innocent too?

The primary causes of the plethora of other diseases, where Taubes likes to blame sugar as the lone villain, have long been established by an immense amount of research, including several large multi-year randomised controlled trials. I'll leave the collection of these studies as an exercise for the reader 😃

The existence of other hypotheses and theories, does not prove that Taubes is wrong. Nor does he blame sugar as the lone villain, just the most important one. In my opinion he is pretty honest about it, acknowledging lack of evidence and admitting he only presents one side of the issue. I do think that by putting out a book about sugar, he puts the focus on the subject, thereby risking that other dietary factors that are almost as bad get ignored.