#429: Kevin Hall, PhD & Stephan Guyenet, PhD – Carbohydrate-Insulin Model vs. Energy Balance Model

1. Key Ideas
2. Detailed Study Notes
3. Transcript

You are currently not signed-in as a Premium subscriber.

To view our Premium content, please log-in to your account or subscribe to Premium.

Not a Premium subscriber yet? Become one here…

Danny’s Key Ideas in this Episode:

1. Primary differences between the models: Direction of causality
2. The energy balance model has much more convincing evidence as a model of obesity

1: Primary differences between the models: Direction of causality

The Energy Balance Model of obesity (EBM) suggests that the primary player here is the brain. And specifically, regulation of body mass is controlled by the brain. This is below our conscious awareness. So the set of processes that the brain is orchestrating, that in turn controls our food intake, occurs in response to both our energy demands and the food environment. So external factors can impact the various internal signals and hormones that impact our food intake. And the brain is the master controller here.

Therefore, to understand the increased prevalence in obesity over the decades, the EBM points to changes in the food environment. So the “increased availability and marketing of a wide variety of inexpensive, convenient, energy-dense, ultra-processed foods that are high in portion size, fat, sugar, and low in protein and fiber.”

However, it is still not completely understood how the food environment alters the brain circuits that regulate food intake. Understanding the exact mechanisms would allow us to better understand why some people are more susceptible to obesity than others. But Dr. Hall did mention in our discussion that a number of areas of fascinating research are uncovering potential mechanisms that could be at play, albeit that these are in rodent models primarily at the moment. For example, one of those lines of research he mentioned was work that has identified a bidirectional circuit between the neurons that control hunger, and the hormone dopamine, which affects food reward. This data suggests that feeding rodents a high-fat diet alters this brain circuit and leads them to eat more calories. (Study linked in detailed study notes section).

Another area Dr. Hall has pointed to is the identification of specific gut-brain pathways that are able to sense nutrients, i.e. fat and carbohydrate, and impact the activity of neurons relating to hunger and food reward.

Finally, it’s worth re-emphasising the difference between the principle of energy balance (as a law of physics) versus the EBM as a theoretical model of obesity.

Turning our attention to the carbohydrate-insulin model of obesity (CIM), and more recent iterations of it in particular, in 2018 in a paper by Ludwig & Ebbeling, we see them highlighting one of the key differences compared to the EBM: the reversal of “the direction of causality”. Meaning… fat accumulation in fat cells is the driver of positive energy balance, rather than positive energy balance (or a calorie surplus) driving fat accumulation.

The most recent version of the model is published in Ludwig et al., 2021. From that paper: “the CIM proposes a reversal of causal direction: over the long term, a positive energy balance does not cause increasing adiposity; rather, a shift in substrate partitioning favoring fat storage drives a positive energy balance. Among modifiable factors, dietary glycemic load (GL) has central importance.”

So, accoring to the CIM, the aspect of our diet that is perhaps most responsible is high GL (glycemic load) diets. And such diets lead to fat being preferentially stored and trapped in fat cells, which then leads to positive energy balance.

2: The energy balance model has much more convincing evidence as a model of obesity

So as Dr. Hall outlined in our discussion, their paper on the EBM wasn’t a proposal for a model the authors came up with, but rather a model based on the synthesis of obesity research to this point. And it is robust and comprehensive in nature. Of course, as Dr. Hall acknowledged, it’s difficult to directly test this in one go, but the predictions can be tested.

So if we then have other researchers putting forth an alternative model, as has been the case with the CIM, it’s important to determine:

1. What the new model suggests is wrong with conventional thinking.
2. How the new model better fits current evidence.

My conclusion thus far is that the CIM fails on both of these accounts. First, consider what it outlines as incorrect about the EBM. As both Guyenet and Hall alluded to in this episode, the paper presenting the new CIM model, essentially compares it to a strawman version of the EBM that no serious scientist that I know of is putting forward. It didn’t really contend with the brain regulation in detail, and it presents this characture of the EBM of caring only about calorie intake, and that neither nutrient profiles nor food quality make too much difference.

Then, if we consider if the CIM is a better fit of the current data, as Stephan highlighted a number of times, it massively downplays the regulation by the brain. And so in turn, leaves large amounts of the evidence base not dealt with.

In the episode we touched on the Endocrine Society’s Scientific Statement, and its follow-up response to the Ludwig paper that had criticised it, and that statement makes the point that obesity is a complex disorder of energy homeostasis, characterized by the biological defense of elevated body fat mass in our obesogenic environment. It points to a large literature base that now exists in support of this concept, something echoed by the Hall paper.

The Hall et al., 2022 paper reviews data from a wide variety of studies that address the validity of each model and demonstrates clearly, to me, that the EBM is a more robust theory of obesity than the CIM.
And finally, as I stated in the discussion, these are not two equally valid competing models that split opinion; but rather one is a comprehensive model that is based on all obesity research to date, and the other is a recent model proposing something different, that doesn’t have the same weight of evidence. It’s not some sort of 50-50, coin-toss, as to which is correct, nor is it an equal head-to-head battle.

You are currently not signed-in as a Premium subscriber. Key Ideas are for Premium subscribers only.

To view our Premium content, please log-in to your account or subscribe to Premium.

What is a Valid Model?

The purpose of a model is to integrate your knowledge of a system (in this case a biological system); explain the observations that have already been made in an integrated and consistent way and then to be able to design experiments to test the validity of that model.

The model is your “best guess of how the system works”.

A valid model of obesity should help explain a couple of questions:

1) Why has obesity prevalence increased over time? Particularly the rapid increase in the past number of decades?

2) Why are some people more susceptible to gaining more body fat than others?

Genetic differences explain much of the individual variability. But for population-level obesity prevalence, the primary driver most likely seems to be changes in the food environment. However, the debate between proponents of the different models of obesity, relates to which changes in the food environment are responsible.

History of the Carbohydrate-Insulin Model (CIM)

• Originally a 1924 paper by Harris raised the hypothesis of high insulin driving hunger that leads to overeating.
• The modern CIM was perhaps originally suggested by Gary Taubes. Taubes proposes a model of obesity based on carbohydrates and insulin, that focused on regulation of fat moving in and out of adipose (fat) cells. Taubes’ original hypotheses seemed to suggest this was independent of caloric intake
• From Taubes:
  1. “carbohydrates are singularly responsible for prompting insulin secretion;
  2. insulin is singularly responsible for inducing fat accumulation;
  3. dietary carbohydrates are required for excess fat accumulation;
  4. both Type 2 diabetics and the obese have abnormally elevated levels of circulating insulin.”
• Some of the authors on the most recent CIM paper, notably lead author David Ludwig, have previously supported earlier versions of the CIM, such as that put forth by Taubes. But more recently the models put forward have gotten gradually further from Taubes’ proposal that carbohydrates are the “ultimate cause of obesity”.
• In 2018 a review paper, David Ludwig and Cara Ebbeling put forth their proposal for the CIM; what we can call the “adipo-centric” CIM. From that paper:
  • “High-carb diets…produce postprandial hyperinsulinemia, promotes deposition of calories in fat cells instead of oxidation in lean tissues, & thereby predisposes to weight gain through increased hunger, slowing metabolic rate, or both”
  • This adipo-centric CIM “considers fat cells as central to the etiology of obesity”. Meaning that all factors that are obesity-causing “affect insulin secretion or adipocyte biology directly” and that both increased calorie intake and decreased energy expenditure are just downstream consequences of the effects on insulin or adipocytes.
  • The key thing about this model (compared to the prevailing Energy Balance Model) is that it “reverses the direction of causality”; i.e. fat accumulation in adipose tissue is the primary driver of positive energy balance, not the other way around.
• In May 2021, Kevin Hall and John Speakman published a paper in Science that essentially highlighted how the model put forth by Ludwig and Ebbeling was overly simplistic in how it focused in on insulin and adipose tissue alone.
• Then in December 2021 Ludwig and colleagues published the most recent iteration of the CIM, in a paper appearing in the American Journal of Clinical Nutrition. This paper was the focus of much of this podcast discussion.
• This paper was a revision of the initial adipo-centric CIM, and seems like a relatively significant departure from some aspects of it, with a number of key differences to previous hypotheses put forward. More details in sections below.

Overview of the Energy Balance Model (EBM)

The Energy Balance Model (EBM) of obesity, as recently described by Hall et al. (2022), posits that “the brain regulates weight mainly below our conscious awareness by controlling food intake under the influence of internal signals and in response to the body’s dynamic energy needs as well as the food environment.”

The central role of the brain in weight regulation is the primary aspect of the model, and is one of the key differences with the CIM.

In the discussion, Stephan Guyenet highlights how Dr. Bernard Mohr highlighted how obesity is associated with abnormalities of the hypothalamus more than 180 years ago (Note: The particular case Stephan references of Elisa Moser is discussed in his book, The Hungry Brain [pg. 113]).

As was mentioned in the discussion, the brain being the key organ here, and the rest of the aspects of the model discussed in the Hall paper is not a newly proposed hypothesis. Rather, the paper is just a refined explanation of a model based on lots of past obesity research to this point.

It is important to note that the Energy Balance model, as a theoretical model of obesity, is distinct from the principle of energy balance (the law of thermodynamics).

So the principle of energy balance (as a law of physics) simply describes energy conservation in a system; i.e. it doesn’t account for how an energy surplus leads to changes in fat mass versus other stores of energy in other body compartments.

However, the EBM of obesity is not the same thing as the law of thermodynamics. Rather it is a theroretical model of obesity pathogensis that attempts to explain the drivers of obesity through all the best current evidence.

The EBM points to changes in the food environment as drivers of increased obesity prevalence; i.e. the increased availability and marketing of a wide variety of inexpensive, convenient, energy-dense, ultra-processed foods that are:

• available in larger portion sizes
• high in fat
• high in added sugar
• low in protein
• low in fiber

How does the food environment impact food intake?

The exact mechanisms by which the food environment alters the brain circuits that control food intake (and therefore why some individuals are more susceptible to obesity), have yet to be completely worked out. However, there is a lot of interest in this field of research right now. Some promising work, mainly in rodent models (i.e. animal studies), has suggested a couple of potential answers. For example:

1. In rodent studies, a bidirectional circuit was identified between neurons controlling hunger & dopamine affecting food reward. High-fat diets seem to alter this circuit, leading to increased energy intake & resulting obesity.
2. Distinct gut-brain pathways, that relate to the sensing of both dietary fat and carbohydrate, have also been identified. These pathways can therefore potentially have a controlling influence on the activity of neurons that encode for hunger and food reward.

Food Reward

“Food reward causes us to select refined calorie, dense, low satiety foods and eat more of them than we otherwise would.” – see 38:16 – 40:34 of episode for full discussion.

Food reward = The momentary value of a food to the individual at the time of ingestion (Rogers & Hardman, 2015)

Food reward consists of three aspects:

1. Liking
2. Motivational drive
3. Learning or reinforcement

Liking = The pleasantness of taste of food in the mouth. (Note that some researchers distinguish this from the overall pleasantness of eating, which is usually called ‘palatability’.) (Rogers & Hardman, 2015)

Food reward is much more than concious liking of a food. It is much more multifaceted and complex. It includes habitual behaviors, forming habits, social norms and how those inform your opinions about things.

Degree of Processing

Hall et al., 2019 published this RCT on how consumption of diets high in ultra-processed foods affects calorie intake and body weight.

Highlights (verbatim from paper):

1. 20 inpatient adults received ultra-processed and unprocessed diets for 14 days each
2. Diets were matched for presented calories, sugar, fat, fiber, and macronutrients
3. Ad libitum intake was ~500 kcal/d more on the ultra-processed vs unprocessed diet
4. Body weight changes were highly correlated with diet differences in energy intake

Does the EBM say all calories are the same?

In the Ludwig et al. paper, the authors state: “the EBM considers all calories metabolically alike for practical purposes”. The supporting reference given in this case is a 2017 review (Endocrine Society statement) from Schwartz and colleagues.

From that review by Schwartz et al.: “The impact of diet on obesity risk is explained largely by its effect on calorie intake, rather than by changes of either energy expenditure or the internal metabolic environment. Stated differently, ‘a calorie is a calorie.’ Thus, habitual consumption of highly palatable and energy-dense diets predispose to excess weight gain irrespective of macronutrient content.”

That same group, i.e. the authors of the Endocrine Society 2017 statement, responded to Ludwig et al. in February 2022 with a Letter to the Editor, stating:

• “In mischaracterizing the Endocrine Society’s Scientific Statement, Ludwig et al. ignore its main thrust, which views obesity as a complex disorder of energy homeostasis that is characterized by the biological defense of elevated body fat mass in our obesogenic environment. A large literature now exists not only in support of this concept but in the identification of underlying mechanisms; none of this work is acknowledged. By sidestepping this crucial and well-developed aspect of obesity pathogenesis, Ludwig et al. provide a selective and one-sided view of the issue.”

In the podcast discussion Hall explained that the EBM does not postulate that all calories are metabolically alike inside the body. Rather it acknowledges different diets can result in markedly different internal signals (e.g., circulating hormones and fuels) that influence food intake and metabolism.

From Hall et al., 2022: “The EBM recognizes that individual differences in energy partitioning can result in different degrees of adiposity, even when energy intake is not different”

The EBM also outlines how both the quality and composition of the diet are important in the prevention and treatment of obesity. And it also highlights why “eat less and move more” is ineffective advice.

Hall et al., 2022: “the [CIM] Perspective conflated and confused the principle of energy balance, a law of physics which is agnostic as to obesity mechanisms, with the EBM as a theoretical model of obesity that is firmly based on biology. In doing so, the authors presented a false choice between the CIM and a caricature of the EBM that does not reflect modern obesity science.”

Criticisms of the EBM by Ludwig

In the Ludwig et al. (2021) paper, the Carbohydrate-Insulin Model is presented as a better explanation of obesity pathogenesis. That argument has to be built on the premise there are problems with the prevailing model, namely the Energy Balance Model. Ludwig et al. raise a number of criticisms…

One of the critques relates to the fact that the EBM puts the availability (and overconsumption) of hyper-palatable, energy dense, highly processed foods at the centre of increased obesity prevalence.

One example that I (Danny) brought up in the episode was a section of this lecture that Dr. Ludwig gave at the Harvard School of Public Health (publish date of Feb 1st 2022). About 10 minutes in to the lecture, while raising his criticisms of the EBM, Dr. Ludwig suggests there is little value on focusing on palatability and processing by means of a comparison; he compares: olive oil, sugary drinks, and baked potato chips.

In reference to the olive oil and sugary drinks, Ludwig states: “Both of them are considered palatable, very tasty, and they are both extensively processed, they are both single macronutrient extractions from plants. …. And olive oil actually has 25 times the energy density of the sugary beverage” but the sugary beverage is much more strongly correlated with a host of negative outcomes.

He then goes on to compare them to baked potato chips, demonstrating how this is a food with minimal processing (sliced and baked), and has lower energy density than olive oil, but has adverse outcomes associated with it.

Push back:

• Hall points out how Ludwig et al. claim that palatability and liking of foods are under concious control and awareness. However, this is too reductionist. (See the above notes on food reward).
• A better comparison is refined sugar vs. soybean oil – as both are ingredients used in cooking

Guyenet: John de Castro’s work has shown in free living people that palatability is actually pretty strongly correlated with food intake at a meal.

Guyenet on palatability vs food reward: “I think there’s valid debate about the role of palatability per se, in kind of long term calorie intake and body fatness… So I think that’s a valid counter argument. However, I think the discussion doesn’t address the main point of food reward in my view, which is not liking, but motivation and learning. So food reward basically has three components. There’s the liking, there’s the motivational drive, and then there is the learning or reinforcement.”

Body Fat Set Point Theory

Another question posed in the Ludwig paper: explain “why the so-called body weight “set point” has increased rapidly among genetically stable populations.”

The proposed biological defense of an elevated level of body fat that occurs in obesity is discussed more here by Guyenet & Schwartz (2012).

This is a core reason why weight loss is so difficult to maintain in the long-term for those living with obesity. One part of this “biologocial defence” relates to changes in the circulating levels of the hormone leptin. When weight loss occurs, the levels of leptin drop, in turn activating a negative feedback loop. This response includes an increase in the drive to eat (hunger, food seductiveness) and a decrease in energy expenditure (decreased metabolic rate and physical activity thermogensis). [We discuss weight loss maintenance and the biological response to weight loss in episode 352].

Carbohydrate-Insulin Model (as per Ludwig et al., 2021)

One aspect of the CIM that the authors emphasize is the influence of diet on substrate partitioning. Specifically, they state: “In the CIM, a crucial effect of diet is metabolic, by influencing substrate partitioning. Rapidly digestible carbohydrates, acting through insulin and other hormones, cause increased fat deposition, and thereby drive a positive energy balance.”

“… the CIM proposes a reversal of causal direction: over the long term, a positive energy balance does not cause increasing adiposity; rather, a shift in substrate partitioning favoring fat storage drives a positive energy balance. Among modifiable factors, dietary glycemic load (GL) has central importance.” – Ludwig et al., 2021

Glycemic load (GL) is the glycemic index multiplied by the total carbohydrate content:

• GL = (GI x amount (in grams) of available carbohydrate per serving)/100

• Solid red arrows indicate sequential steps in the central causal pathway
• The dashed red arrows (—-) and associated numbers indicate testable hypotheses comprising multiple causal steps
• Black arrows indicate “other relations”.

As per the above diagram, the most recently updated version of the CIM seem to acknowledge the relationship of energy balance (intake vs. expenditure) to obesity. However, crucially, the CIM sees energy intake/expenditue as “proximate, not root, causes of weight gain”.

Also, and interestingly, it is clear that the updated CIM abandons the idea that all pathways to positive energy balance act downstream of adipose tissue fat storage. New parallel pathways have been introduced; for example, one pathway it highlights is a direct effect of dietary glycemic load on energy intake. There is also an effect here with “circulating fuels”, i.e. glucose, fatty acids, etc.

A simplified chain of events could be thought of as:

1. Consumption of a high-GL meal
2. Increases in insulin secretion and suppression of glucagon
3. This state promotes uptake of glucose into cells, and stimulates the process of lipogenesis (synthesis and storage of fats or triglycerides, from the conversion of fatty acids and glycerol into triglycerides).
4. After a few hours al nutrients have been absorbed.
5. But then the levels of glucose in the bloodstream goes from high in the early postprandial phase, to low in the late postprandial phase. And it is suggested this swift drop in glucose (and other fuels) is perceived by the brain as a big problem; it sees it as critical tissues and organs like the liver being deprived of energy.
6. A response to this perceived shortfall is then launched, causing increases in hunger and cravings for high-GL foods. Thus creating a “vicious cycle”, as the paper terms it.
7. Additionally, there are the suggestions that energy expenditure declines due to the decrease fuel availability.
8. And all this is what drive a positive energy balance.
9. So the changes in obesity prevalence at a population level can be explained by the increase in total carbohydrate intake and particularly the intake of high-GI of processed carbohydrates.
10. And then to take one step further to looking at interventions to reduce fat mass or body weight, it is suggested that “weight reduction produced by carbohydrate restriction would decrease the insulin-to-glucagon ratio, enhance lipolysis and fat oxidation, and result in lower spontaneous food intake.”

However, in this episode Hall and Guyenet raised concerns about the validity of placing glycemic load as such the primary causal factor.

Guyenet (01:14:27): “I don’t think it’s at all an unreasonable hypothesis, totally plausible. However, the thing that I think is much harder to support, which is what the paper is suggesting, is that this is the primary dietary determinant of body fatness. If you want to say: ‘this is a factor in a more complex landscape of factors’; I think that’s very reasonable. If you want to say: ‘this is the primary factor’, you run into some big empirical problems with the scientific literature”

Guyenet highlights the large number of randomized controlled trials that compare diets that differ greatly in glycemic load (e.g. low-carb vs low-fat), that don’t show differences. One such example he highlights is the DIETFITS trial by Gardner et al., 2018 (note: you can listen to the lead author Christopher Gardner discuss the study on episode 223 of the podcast).

Another example of contradictory data includes a tightly controlled feeding trial done by Hall and collegues (2019), which compared the impact of ultra-processed diet vs. unprocessed diet on intake and weight. The diet consisting of ultra-processed foods caused significantly more calories to be consumed and weight to be gained. However, the two diets were of relatively equivalent glycemic load. In addition, there were no differences in either average daily blood glucose concentrations or glycemic variability between the diets, as measured by daily continuous glucose monitor (CGM).

Stephan Guyenet’s Issues with the Ludwig Paper:

1. It misrepresents alternative models of obesity. Specifically it pits the CIM against an “obsolete model that virtually no one believes”
2. It is a “speculative new model of obesity that discards most of what we know about obesity biology.” Meaning that it “virtually ignores the brain, which is very hard to understand given the last 180 years of obesity science.”
3. The paper’s commentary on the genetics of obesity, doesn’t deal with the “overwhelming evidence that the brain is central”.

Conclusions

1. Agreement: Both models agree that diet quality and composition are important in prevention and treatment of obesity. Both models account for endocrine regulation, peripheral energy sensing, and energy partitioning.
   —
2. The EBM focuses mainly on the brain, the CIM focuses moreso on the adipose tissue. With both models examining how internal (hormonal) and external (environmental) cues impact the brain and adipose tissue, respectively.
   —
3. The most recently published form of the CIM by Ludwig et al., 2021, focuses on excess glycemic load of the diet as being a causal driver of obesity. It’s focus is on what stores fat in fat cells and keeps it there, and thus looks to interventions that prevent this. In contrast, the EBM proposes that a range of factors drive excess calorie intake and considers adipose tissue to be an “active endocrine organ evolved to coordinate safe & efficient storage & mobilization of energy”.
   —
4. Currently a large amount of evidence exists showing that several variables in the food environment can drive increased calorie intake. Although it’s possible high glycemic load meals/diets can contribute, there is little to suggest it it the primary driver, and other variables have evidence independent of GL. The EBM would propose that “obestity can arise if any one or more of these factors are in play.”
   —
5. The Hall et al., 2022 paper reviews data from a wide variety of studies that address the validity of each model and concludes that the EBM is a much more robust theory of obesity than the CIM.
   —
6. To me (Danny), it is an error to view these two models as “equals”, that both have evidence and just happen to split opinion. But rather it is more accuate to view the situation as one (the EBM) being a comprehensive model that is most coherent with overall obesity research, and the other (the CIM) is a recently proposed alternative model, that doesn’t have the same weight of evidence. And while some aspects are interesting and plausible, other aspects are difficult to consolidate with current evidence. (See the “key ideas” section for more commentary on this point).

You are currently not signed-in as a Premium subscriber. Detailed study notes are for Premium subscribers only.

To view our Premium content, please log-in to your account or subscribe to Premium.

You are currently not signed-in as a Premium subscriber. Episode transcripts are for Premium subscribers only.

To view our Premium content, please log-in to your account or subscribe to Premium.