Gary Debunks Taubes ~ It's the Fructose Stupid!

In his second blog installment in a series that says nothing about its purported purpose (to debunk the food reward hypothesis so we can all return to eating our high reward Science Krispies), Gary Taubes cites the following study:

C L Cox, K L Stanhope, J M Schwarz, J L Graham, B Hatcher, S C Griffen, A A Bremer, L Berglund, J P McGahan, P J Havel* and N L Keim
The results of short-term studies in humans suggest that, compared with glucose, acute consumption of fructose leads to increased postprandial energy expenditure and carbohydrate oxidation and decreased postprandial fat oxidation. The objective of this study was to determine the potential effects of increased fructose consumption compared with isocaloric glucose consumption on substrate utilization and energy expenditure following sustained consumption and under energy-balanced conditions.

As part of a parallel arm study, overweight/obese male and female subjects, 40–72 years, consumed glucose- or fructose-sweetened beverages providing 25% of energy requirements for 10 weeks. Energy expenditure and substrate utilization were assessed using indirect calorimetry at baseline and during the 10th week of intervention.

Consumption of fructose, but not glucose, led to significant decreases of net postprandial fat oxidation and significant increases of net postprandial carbohydrate oxidation (P<0.0001 for both). Resting energy expenditure (REE) decreased significantly from baseline values in subjects consuming fructose (P=0.031) but not in those consuming glucose.

Increased consumption of fructose for 10 weeks leads to marked changes of postprandial substrate utilization including a significant reduction of net fat oxidation. In addition, we report that REE is reduced compared with baseline values in subjects consuming fructose-sweetened beverages for 10 weeks.
Here's Gary's summary:
[This] paper reported that overweight and obese older adults (40 to 72 years-old) getting a quarter of their calories from fructose-sweetened beverages used less fat for fuel and actually expended less energy than did the same subjects when they were getting the equivalent calories from glucose drinks. So the calories were the same; the metabolic effects were different. For the fructose, the effects were what we’d expect (okay, what I’d expect) if the fructose beverages were causing or exacerbating insulin resistance, an observation that Havel et al had published earlier. The more insulin resistant these people became, the less fat they used for fuel—hence, we can assume, the more fat they stored—and the lower their basal metabolic rate.
See that linkey loo in there?  It takes you to the full text of a study entitled:  Consuming fructose-sweetened, not glucose sweetened, beverages increases visceral adiposity and lipids and decreases insulin sensitivity in overweight/obese humans.  Cue Tim Allen questioning grunt:   Aahrrrouh?  Did Gary just debunk Taubes?  Damn those parallel arms!!

For those who don't like to rely on abstracts when they don't have to, I was able to find the full text of the first study free online HERE.   Gotta hand it to Gary though ... had he not committed a referencing faux pas by calling the Cox study Havel, I may not have even bothered to take a look and found both Cox et. al. and Stanhope  That Gary is gooood at debunking the Taubes!  (*see footnote re:Havel)

In any case, here's a different link to that parallel arm ... hee hee ... plus, I think the title is worth repeating!  Consuming fructose-sweetened, not glucose sweetenedbeverages increases visceral adiposity and lipids and decreases insulin sensitivity in overweight/obese humans.  This is a well reported study, I'm only about 1/10th of the way through looking at the mass of supplemental data they provide.

The two papers are actually related to the same experimental protocol.  Unlike the play, let's keep this clean.  By that I mean, I do not want to argue the validity of food reward when we're trying to discuss TWICHOO.  Because (1) they are not complementary hypotheses, and (2)  TWICHOO does not rise to the level of being proposed based on observation, let alone acceptable as a null hypothesis**.  This is about what Gary's own references do to support (or rather refute) TWICHOO.  It doesn't look good folks.

Do you see it there in the title of the parallel arm?  Fructose sweetened beverages reduce insulin sensitivity, glucose sweetened ones do not.  Because context is everything, let's discuss what this study involved.  Surely these were those bulletproof young paleo-types that have yet to have pummeled their pancreata into dysfunctional submission ... subjects with pre-pre-deranged metabolisms if that??   Ahh ... no.  While relatively healthy, the subjects were overweight/obese (average BMI just under 30, weight stable for 6 months prior to study inception) and middle aged (avg. age around 50 -- oh, and they made it a point to mention that the women were all postmenopausal).  In the end the groups were rather small for the interventions (15 glucose, 17 fructose).  They split them by gender in the pre-study measures (M/F:  7/8 glucose, 9/8 fructose), but most of the analyses of the results of the intervention group them together.

I am going to do at least one more blog post, perhaps several, on these studies with a more thorough analysis of the implications vis a vis fructose per se, IR and obesity.  But TWICHOO says:
  • Carbs, by driving insulin, drive fat accumulation and make you fat 
  • It's the hormones, insulin in particular, that make you fat, not the calories***
  • Insulin causes fuel partitioning to fat vs. lean tissue by trapping fat in your fat cells.
  • Postprandial insulin causes insulin resistance
So in this study, they looked at what happens to already fat people when you try to replace one-quarter of their normal daily caloric intake with simple sugars in liquid form.   They took their SAD's and made them SADder with a daily consumption of roughly 600 cal/day of liquid sugar cals.  I note that habitual sweet beverage drinkers were excluded from the study so you're talking habituated people suddenly switching to taking in a large portion of their calories as liquids.  I also note that the  macronutrient composition of the diets was 55% carb, 30% fat, 15% protein.  This is in contrast to the documented intake of obese subjects that tend to put their SAD ratios closer to 40-50% carb, 35-45% fat and 10-15% protein.    In any case, their baseline measurements were following the 55/30/15 formulary.  The study had 3 phases:  
  1. Two weeks inpatient baseline:  Weight maintaining calories 55/30/15 as just described
  2. Eight weeks free-living intervention:  During this period subjects consumed 25% of their baseline energy requirements in beverage form and ad libitum from self-selected foods.
  3. Two weeks inpatient intervention:  Baseline level caloric intake with 25% from glucose or fructose beverage.
It is difficult to put exact numbers to intakes here ... despite extensive supplemental info provided in Stanhope, too much is reported in percent differences, etc.  However in that paper the fructose consumption was reported as averaging 617 calories/day.  From this we can figure that the weight maintaining caloric intakes of these subjects averaged 2468 cal/day.  We are told that during the free-living phase, caloric intake increased and the subjects gained weight!  The supplemental tables tell us that this increase averaged 7.9% total calories (8.4% glucose, 7.4% fructose), bringing intake to 2663 cal/day, a difference of 195 cal/day.

Obviously Gary read the abstract from Cox, saw all he needed to see, and rattled off his blog post.  Clearly between fructose and glucose we do see metabolic differences, but for 8 of the 10 weeks, the calories were NOT the same as baseline.  Sloppy science journalism there Gary.  But, yes, there were differences in the results between glucose and fructose.   Let's look at the body weight and fat distribution:

Not looking good for TWICHOO here.  (Note that I've edited this section on 11/27 after seeing a graphic in Stephan's post here, he has a little error as well, but I originally thought the graphic above right was for all fat but it is for abdominal fat only).  Though not significant, the body weight gain was a tad smaller for fructose vs. glucose, consistent with the slightly lower increase in intake.  Using values from Table 3, I've calculated the mass gains in absolute values. The glucose group gained 1.55 kg = 3.4 lb total body weight vs. 1.2 kg = 2.6 lb for the fructose group.  The glucose group gained 0.98 kg = 2.2 lb  total fat mass vs. 0.81 kg = 1.8 lb for the fructose group.  Therefore, the glucose group gained 0.57 kg = 1.25 lb fat free mass vs. 0.39 kg = 0.86 lb for the fructose group.  Further, gains were split lean/fat 37/63%  for glucose vs. 32/68% for fructose.  Now, none of this was statistically significant, but consider the Mt. Everest sized insulin spikes the 150g of pure glucose must have elicited.  According to TWICHOO, we would expect this to not only be the other way around, favoring more fuel partitioning to fat, but there should have been significantly more fat accumulation and fat compared to lean mass gains.  The results fly in the face of TWICHOO because insulin, if anything, promoted greater fuel partitioning to lean than did its absence.  

The glucose group, per graphic above right, gained significantly less abdominal fat, thus insulin partitions fuel to the less metabolically problematic subcutaneous (SAT) vs. visceral (VAT) fat and to non-abdominal fat vs. the more metabolically problematic abdominal accumulation.  When one looks at the metabolic havoc fructose wreaks, it seems that it is NOT insulin's actions on the fat cells that are causing them to "go wild" (and ultimately lead to the dysregulation of glucose homeostasis that is insulin resistance and type 2 diabetes).

To summarize:
  • The subjects all gained weight because they ate more, they did not compensate for the calories consumed as sweetened beverages. 
  • Glucose partitioned more fat to SAT vs. VAT compared to fructose.
  • Glucose partitioned more fat to non-abdominal regions compared to fructose
  • Glucose partitioned proportionally more weight to fat free mass vs. fat mass compared to fructose (not statistically significant.  [/edit]
I think the reduction in lipid oxidation rates and resting energy expenditure do warrant at least one separate blog post.  But these metabolic effects are not proposed in TWICHOO as it does not address metabolic rate.  Still, again, the lower insulin-provoking diet was the one where this was seen.

In the end, we saw weight gain because .... drumroll please .... the subjects ended up eating more and perhaps in the case of fructose, their basal metabolic rates were reduced slightly (it works out to about 7.5%).   Sounds like that old conventional wisdom at play.

What of the insulin resistance?  Well, thanks to that parallel arm, we have the full OGTT results.

Yeah ... replace roughly half of the complex carbs with *the* most refined carb you can, glucose, and the blood glucose levels spike how much?  The insulin does what??  Nada.

Does carbohydrate consumption cause hyperinsulinemia leading to insulin resistance?  Not even in already overweight/obese middle aged people in energy surplus.  (I note these OGTT's were done in Week 9 so when caloric balance had been restored for one week on an inpatient basis)

[edit]Stephan noted something else I didn't, from Table 5, the fasting insulin increased by only about 3% in the glucose group while it increased by 10% in the fructose group. In other words, "hyperinsulinemia" didn't really occur in either of the groups. But to the extent that fasting insulin levels rose, it was not in proportion to postprandial insulin secretion that would have been considerably larger for the glucose vs. the fructose group. Yet it was the fructose group that experienced a 3-fold greater increase in fasting insulin. Could it be that hyperinsulinemia is the result of fat accumulation rather than the cause? Another TWICHOO smackdown.  [/edit]

Gary Taubes has just debunked Gary Taubes.  Thanks Gary!

*I had come across works by PJ Havel (perhaps the Stanhope paper, it sounds all too familiar to me) a while back when Kurt Harris brought my attention to this paper by Havel. Havel is another rather prolific researcher. Perhaps Gary Taubes should put his money where his mouth is as regards good scientific inquiry. Just saying.

**See this post for a discussion on hypotheses. In statistics, the null hypothesis is presumed true as a matter of the process. It does not even imply that the person doing the testing believes this to be the case. GT wants us to believe that his TWICHOO should be the one presumed true. Sheesh, is all I have to say about that nonsense.

***Although initially GT would discuss energy balance extensively and merely point to the direction of causality, he has more recently abandoned that to adopt a "calories don't matter at all" attitude.

LASTLY ... I do not know when exactly I get around to discussing other implications of these studies. They don't make fructose look too good. But let's put this into some context. The fructose group was:
  • Already overweight/obese 
  • Consuming 150g isolated fructose (e.g. 300g sucrose carb) daily for 10 weeks. 
  • In positive caloric balance for 8 weeks 
  • Consuming a diet for baseline that may well have differed from habitual diet composition 
I'm not ready to jump on the Lustig Fructophobe line on Amtrak anytime soon.


Sanjeev said…
countering point:

more realistic fructose consumption for comparison

> reduction in lipid oxidation rates and
> resting energy expenditure do warrant
> at least one separate blog post

And several face saving posts from certain bloggers who claim repeatedly that insulin's effects on appetite and REE work ONLY WHEN INSULIN IS INJECTED DIRECTLY INTO THE BRAIN.

Except for damaged mice where circulating insulin is 20 times normal and the mice stay lean ... THEN, (CONVEEEEEEEENIENTLY), the fat cells can ignore insulin (and circulating glucose) and the insulin can have an effect in the brain (without being injected)
Sanjeev said…
Great first act for Gary and Evelyn ... or is this just the prologue, with plenty of foreshadowing ... cue the Jaws riff (or the Keystone Kops "running madly about" piece as "Gary's theme").
Sanjeev said…
Sorry folks ... if the link doesn't come out again, here it is for copy and paste:

as a link

(it showed up as a clickable link in preview, so I hope "post comment" doesn't clobber it)
Sanjeev said…
note of interest: Google's doing something "interesting"

I could have sworn the first post up there did have a working clickable link but it's not now clickable, and in Google Reader that post shows up with the title

"countering point:<a rel="nofollow"> more realist..."

I never ever have typed "nofollow" in a link, so Google's doing some kind of link re-writing. As I type this the "sorry folks" comment has a working click through link but it may not for much longer.
CarbSane said…
That's very strange Sanjeev. Google has changed up quite a few things lately with new looks on gmail and docs and such. Hope they don't try to reroute links through google or something :(
@sanjeev - in google's own FAQ pages, they only recommend nofollow if a link is paid or could land you in a "bad neighborhood". even the dude who invented wordpress admits nofollow has completely failed at deterring comment spam. so nice to see the Google-borg is now adding nofollow in as it sees fit, deciding these things for us puny humans.
MM said…

Wow, yet another study that debunks the insulin hypothesis. How many will it take? :) Obviously fructose isn't making people fatter by chronically raising their insulin levels. Do you have any idea how it does work then? Clearly fructose is doing something to change the body composition so dramatically. Have you blogged on this and I missed it?
Sanjeev, I think in the first comment you accidentally omitted the href="..." source for your link. It's actually not required, without it, the text becomes an anchor (folks use this to do within-page targets).

Many blog platforms automatically add the "rel=nofollow" to comment links (I know does it). This is meant to discourage spamming, as the nofollow code simply means that Google won't use it to build pagerank for the linked page. I.e., it's a feature, not a bug.
CarbSane said…
@Sanjeev: Yeah, I do love how all of a sudden certain fat cells develop those convenient, or rather as you say conveeeeeeeeeenient, aversions to fat accumulation.

@MM: I haven't blogged much on fructose. The postprandial fat oxidation rates after the study were rather alarming, and the depression of metabolic rate coupled with increased consumption would seem to account for this. But I can't get too worked up over that. That 150g fructose is equivalent to consuming 300g HFCS/sugar. Put into context, that's almost 8 cans/day or almost 3 two liter bottles of soda/day -- consistently, day in and day out for two months. Hmmmm ... put into context since they consumed the isolated simple sugars, these folks were drinking like 3 liters of sugary beverage per day!!
Sanjeev said…
As far as the fructose worsening the blood lipids, SO WHAT? There's NOTHING unique about fructose in that regard.

Over on Lyle's forums they've posted study after study after study over the years that showed almost every time the subjects gain weight the blood lipids go to h*ll.

Doesn't matter how it's done. Higher calories, no matter where they come from - sugar, oil, fat, starch. If it's excess calories the situation gets worse.

Conversely whenever subjects lose weight, no matter what the diet the tests come out better.
Sanjeev said…
> Put into context, that's almost 8 cans/day
> or almost 3 two liter bottles of soda/day -

And it's NOT PROOF AGAINST food reward.

Sweet taste by itself is not very self-reinforcing / recursive in the food reward sense ... as Seth Roberts has shown (though not in randomized clinical trials).

When distinctive flavours are not added the sugar water does not have the same kind of effect. It's like chocolate Ensure without the chocolate.

And Guyenet TOLD Taubes that at the AHS

(paraprahsed, Taubes asked about high sugar drinks ):
"... so how is it not palatable ..."
Sanjeev said…
> It's like chocolate Ensure without the chocolate.

another log on the fire: here's a case of non-caloric elements, NOT acting through the conscious mind causing obesity, yet Taubes, the self proclaimed "fat person's friend", who supposedly is taking the blame off the obese[0]

actually has a REAL chance to take the blame off the conscious decision-making processes, off conscious choice, and he's doing a whole series trying to debunk it.

[0] (he's not really, if I ever have a blog someday I'll do a long dissection)
Tomas said…

I've seen this posted at the Danny Roddy's:

Carbohydrate Analysis of High Fructose Corn Syrup (HFCS) Containing Commercial Beverages

"...the carbohydrate contents of beverages determined after acid hydrolysis were substantially (4–5 fold) higher than the listed values of carbohydrates. As fructose and glucose in HFCS may exist as monosaccharides, disaccharides and/or oligosaccharides, analysis of the carbohydrate content of HFCS containing samples may yield widely different results depending on the degree of hydrolysis of the oligosaccharides. With inclusion of mild acid hydrolysis, all samples showed significantly higher fructose and glucose content than the listed values of carbohydrates on the nutrition labels."
CarbSane said…
Welcome Tomas! This is very interesting. If I'm understanding this correctly, they're saying that HFCS contains oligosaccharides that would be broken down to usable carb by gastric acid? Interesting ...
CarbSane said…
Tomas, I've been looking into this some more. While I can't find anything definitive, I came up with a lot of info that bacterial metabolism of oligosaccharides is lower than claimed by those who advocate them as prebiotics. This tells me they are likely indeed essentially undigested by humans. If the difference were that substantial, I tend to think the HCFS v. sucrose studies would have shown *significant* differences.