Anti-sugar/fructose researcher and diet book author Robert Lustig has a new trial published in Obesity claiming to have reduced fructose in the diet of kids, and isocalorically replaced it with starch, leading to improvements in metabolic health. This would be a big deal if conducted really well, since fructose has some unique properties to its metabolism relative to just boring old glucose. As of now, it appears that fructose's detrimental effects are due to overconsumption of calories (1). There is, however, a theoretical concern that insulin sensitive vs insulin resistant individuals (2) respond to fructose differently, as fructose, according to rodent studies (3), appears a particularly good substrate for de novo lipogenesis and contributes to excess liver fat. Sugar is postulated to be terrible for all sorts of people, but as I understand, it's been theorized that high fructose consumption, even under eucaloric conditions, can increasee hepatic liver fat in individuals with insulin resistance (AKA insulin isn't suppressing de novo lipogenesis as it should in the liver); tracer studies in individuals who are insulin sensitive have generally shown that very little fructose is converted to fat (5). The argument gets at the theory of metabolic partitioning and that while all calories are 'equal' from an energy standpoint, their unique properties can alter their metabolism and subsequent partitioning in the body. Seeing the headline of Lustig's trial, I was pretty interested to see if he adequately tested this theory. Unfortunately, we've got a few major design/methodology issues:
1. No randomization/control group - this was a smaller pilot study and individuals were simply recruited from the real world, and placed on a fructose-starch substitution diet. Just being part of a diet study alone can change your behavior, let alone the switch from your normal diet of self-selected foods to now a prescribed-diet, where the food you are to consume is provided to you, you're recording it, meeting with a dietitian, etc. We really need a control group for comparison here. The authors give multiple reasons in their discussion as to why they don't have one, particularly noting that it would be unethical not to intervene in kids who are overweight but that simply doesn't change the fact that in dietary interventions, a lot of behaviors can be altered that confound the results; it's weird to ask a research question that you then claim to be an ethical concern - by way of reason, is it not also an ethical concern to perform an isocaloric study in children with obesity? Regardless, these kinds of studies outside of metabolic wards are incredibly difficult to do, but if we're trying to get at true metabolic mechanisms, we need to hold high standards and a control group/diet is a must (a crossover trial here would've been great assuming a sufficient lead in diet and wash-out period). Any claims of the effects seen being due to just a reduction of added sugar, even for the subgroup that didn't lose weight, are inappropriate interpretations of the data due to this major limitation.
2. Isocaloric? - The authors and the title claim that fructose was exchanged for starch in an isocaloric manner - this should not have affected weight very much if the study had a strong design and a high level of control. Unfortunately, this being an outpatient study, where caloric requirements were estimated from average equations and not indirect calorimetry, led to a pretty low level of control. Most of the participants clearly lost weight, as you can see from this graph:
1. No randomization/control group - this was a smaller pilot study and individuals were simply recruited from the real world, and placed on a fructose-starch substitution diet. Just being part of a diet study alone can change your behavior, let alone the switch from your normal diet of self-selected foods to now a prescribed-diet, where the food you are to consume is provided to you, you're recording it, meeting with a dietitian, etc. We really need a control group for comparison here. The authors give multiple reasons in their discussion as to why they don't have one, particularly noting that it would be unethical not to intervene in kids who are overweight but that simply doesn't change the fact that in dietary interventions, a lot of behaviors can be altered that confound the results; it's weird to ask a research question that you then claim to be an ethical concern - by way of reason, is it not also an ethical concern to perform an isocaloric study in children with obesity? Regardless, these kinds of studies outside of metabolic wards are incredibly difficult to do, but if we're trying to get at true metabolic mechanisms, we need to hold high standards and a control group/diet is a must (a crossover trial here would've been great assuming a sufficient lead in diet and wash-out period). Any claims of the effects seen being due to just a reduction of added sugar, even for the subgroup that didn't lose weight, are inappropriate interpretations of the data due to this major limitation.
2. Isocaloric? - The authors and the title claim that fructose was exchanged for starch in an isocaloric manner - this should not have affected weight very much if the study had a strong design and a high level of control. Unfortunately, this being an outpatient study, where caloric requirements were estimated from average equations and not indirect calorimetry, led to a pretty low level of control. Most of the participants clearly lost weight, as you can see from this graph:
For a 9-day 'isocaloric trial', having lost on average ~2lbs suggests that the attempt to keep things isocaloric failed. Again, given the outpatient nature, and inability to accurately determine individual caloric needs, this isn't surprising. Without constant body weight and a control group, we can't attribute any of the results, including the change in fat mass and ALT (a marker of liver fat/dysfunction) to just a reduction in fructose. I should note that since the body weight changes are important, the body composition data is hard to interpret - DXA is relatively insensitive to short term changes in body fatness - think back to Kevin Hall's recent metabolic ward study where DXA was too insensitive to determine fat mass changes on a 30 percent kcal reduced diet for 6 days (4).
3. Statistics- Statistically, this study is a bit of a mess. It's a pretty small trial, and there's no mention of the power needed to determine an effect on the outcomes measured. The trial was registered to look primarily at liver fat and at insulin resistance as a secondary outcome. But as we can see from the paper, the trial measures a whole host of metabolic parameters. These are all quite significant, several standing up to multiple test corrections, when you include individuals who lost weight, but we would expect weight loss to affect these markers - we were hoping to attribute the effects just to fructose. The researchers attempt to correct for this by splicing the group further in their sensitivity analyses to exclude those who lost significant weight (leaving us with a n=10), and try to claim metabolic improvements, but these p-values don't come close to standing up to multiple test corrections. These considerations are all playing along with the authors' paradigm of assuming the effects are just due to fructose, and forgetting that there is no control group and we can't attribute anything to just fructose restriction to begin with. I reckon when stats.org comments on the trial, they'll have even more statistical critiques (their piece here).
4. Liver fat - On the point of the primary outcome, which was a reduction in liver fat, I think it's important to note that ALT, one of the markers of liver fat, wasn't significantly reduced in their sensitivity analysis of those who didn't lose weight. Again, that wasn't well powered to really detect anything, but it does stand against their primary hypothesis that alterations in metabolic health were due to liver fat reduction. ALT was reduced in the group who lost weight, but that's what we'd expect in a weight loss group. A measure like magnetic resonance spectroscopy, the way that liver fat is measured in strong study designs, would've really helped this study out. It's a lot of guess work and attempted statistical analyses on under-powered/difficult to interpret data.
5. Co-interventions - it's a very small critique, but this question of fructose's effect specifically would be better tested if one were to just alter the sugar composition of the diet e.g. take regular Sugar Sweetened Beverage consumers and change the make up of fructose in that beverage. We're only told a bit about the actual food intervention use, and it leaves us unsure of what other changes occurred. Food-based interventions are inherently multiple changes in nutrients. If added sugars were replaced with whole grains, you'd not only have the reduction in fructose but the increase in fiber and magnesium, both of which might affect insulin sensitivity. We're given a brief statement that the intervention affected protein intakes as well as fiber. No information about micronutrients is given. Keep in mind that co-interventions are hard to avoid in non-metabolic ward trials in nutrition, because even a perfect swap out of fructose for glucose affects the sweetness/sensory aspects of food, that may alter food consumption behaviors.
To summarize: we've got an uncontrolled trial that provided food-based substitutions in an outpatient setting, that led to significant numbers of individuals losing weight, and saw accompanying metabolic improvements, with questionable statistical significance. The primary outcome was a change in liver fat, which wasn't measured by MRS, but by a biomarker that is sensitive to weight changes, and did not significantly change in the statistically-challenged sensitivity analysis. Normally, we'd be concerned about sub-group analyses in such a small trial because they violate randomization, but this trial wasn't randomized to begin with. None of the effects can be attributed uniquely to fructose reduction.
This study is quite small, and while having some interesting metabolic effects, leaves us unable to make any strong conclusions about fructose's role in metabolic health under isocaloric conditions - it definitely doesn't overturn the current abundance of evidence that shows no effect of fructose under isocaloric conditions. I still think this is a valid question to ask - it just requires a better study design, and greater control than the others were able to provide in this publication. On a clinical note, I don't really think this study tells us much - if clinicians have patients who are consuming high levels of sugar in their baseline diet, like those in this study, they're going to tell them to reduce their sugar intake regardless of whether there's some unique metabolic benefit to it - clinicians want to induce weight loss in children with obesity. It'd be nice if we had a definitive study design to say, hey, focus on reducing fructose for the best clinical outcomes, but we can't say that from this study.
As a small aside, the authors list no conflicts of interest, which is quite interesting given Robert Lustig's extensive history of campaigning against sugar, and warning of the metabolic evils of fructose. Are diet books, advocacy organizations and involvement in a group who have decided the conclusions on sugar/health not conflicts of interest? Apparently the editors at the journal Obesity don't think so.
An additional closing comment: this research question extends far beyond just the 'obvious' "eat less sugar" for the general public; while some will see this as just evidence reinforcing the 'drink less soda' mantra, keep in mind that many clinical RDs rely on fructose-containing oral nutritional supplements in the hospital/clinic. If fructose is poison, even under isocaloric or hypo caloric conditions as these researchers believe, it could require major changes throughout formulations of nutrition supplements eg Ensure. Answering this question is a very important one, and should be done with gold standard methods.
1. http://www.nature.com/ejcn/journal/v68/n4/full/ejcn20148a.html
2.http://ajcn.nutrition.org/content/77/1/43.full
3.. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4276992/
4. http://www.cell.com/cell-metabolism/abstract/S1550-4131(15)00350-2
5. http://www.nutritionandmetabolism.com/content/9/1/89
This is a nice summary of the methodological issues from the study. I was also surprised that the authors did not provide a control diet for 7 days or so to establish a baseline biomarker profile before launching into the 9-day reduced-fructose diets. I realize that this would have meant extra needle sticks for the children. However, you cannot fully replicate the home diet from self-report and RD recalls. Our institution has found that our participants will display improved metabolic profiles even on the "control" diets, and my colleagues try to provide a week or so of food to establish a baseline profile before launching into the experimental feeding protocol. The extra cost can be worth it to ensure that your results aren't due to a generally improved diet as opposed to your experimental aim.
ReplyDeleteThanks for your comment Mandy! Completely agree - a lead-in diet could've helped a lot.
DeleteWait, you don't think fructose restriction caused them to lose 1.5 kg in 4 days by turning their metabolism into a fat-burning machine?
ReplyDeleteI love this from Lustig
ReplyDeletehttp://well.blogs.nytimes.com/2014/02/19/learning-to-cut-the-sugar/
Q.A lot of the recipes in your book use fruit to add sweet flavors. Was this a way to limit refined sugar?
A. Exactly. People always say to me, “What about fruit? It has sugar.” But I have nothing against fruit, because it comes with its inherent fiber, and fiber mitigates the negative effects. The way God made it, however much sugar is in a piece of fruit, there’s an equal amount of fiber to offset it.
There’s only one notable exception: grapes. Grapes are just little bags of sugar. They don’t have enough fiber for the amount of sugar that’s in them. But I have nothing against real food, and that includes real fruit. Eat all the fruit you want. It’s only when you turn it into juice that I have a problem with it, because then it loses its fiber.
http://well.blogs.nytimes.com/2013/07/31/making-the-case-for-eating-fruit/?_r=0
Another nutrition expert, Dr. Robert Lustig, a professor of pediatrics at the University of California, San Francisco, who has called sugar “toxic” at high doses and fructose the most “actionable” problem in our diet, is still a fan of fruit. “As far as I’m concerned, fiber is the reason to eat fruit,” since it promotes satiety and the slow release of sugar. He adds a third benefit from fiber: it changes our “intestinal flora,” or microbiome, by helping different species of healthy bacteria thrive.
Neither doctor favors certain fruits over others. But Dr. David L. Katz, director of the Yale University Prevention Research Center, said that “to maximize the benefit, you actually want a variety” of fruits. He advises “eating the rainbow,” since different colors signal different types of antioxidants and nutrients.
So if I stick Metamucil in my Dr. Pepper, I'm good. Right?
Delete