Reply to A Drewnowski et al, O Devinsky, D A Booth and E L Gibson, and D J Millward

Dear Editor:

We are gratified to see that our Perspective (1) on the carbohydrate-insulin model (CIM) has stimulated debate, as illustrated by the numerous Letters to the Editor, and appreciate the opportunity to respond.

Drewnowski et al., co-authors of an Endocrine Society Scientific Statement on Obesity Pathogenesis (2), argue that our Perspective contains misleading and factually incorrect statements. We disagree on 4 points. First, we do not state that there are “only” 2 major competing models for obesity pathogenesis, nor did we intend to imply so. Important context is the opinion papers we cited, by advocates of the energy balance model (EBM), claiming to have disproven the CIM. Our Perspective responds to these claims, presents the first full academic formulation of the CIM, delineates testable hypotheses to promote constructive discourse, and recognizes the possibility of common ground. Consistent with these EBM advocates, we compared these 2 major explanatory models, which have neatly opposing views of causal direction. In the EBM, overeating (energy intake > energy expenditure) causes obesity. According to this view, dietary composition, independent of calorie content, has no meaningful effect on fat storage or energy expenditure. Therefore, special diets can produce weight loss only by reducing energy intake. In contrast, the CIM argues that increased deposition of body fat, driven by dietary and other effects on substrate partitioning, lies upstream in the causal pathway, with the positive energy balance a downstream effect. Therefore, the primary focus of dietary intervention should be what one eats, not how much.

Second, Drewnowski et al. criticize our Perspective for “[ignoring their paper's] main thrust, which views obesity as a complex disorder of energy homeostasis that is characterized by the biological defense of elevated body mass….” However, our purpose in considering their paper was not to appraise its strengths, but rather to identify a major point of disagreement. In stating that “a calorie is a calorie” and disregarding the endocrine and metabolic effects of macronutrients, their paper dismisses a fundamental feature of the CIM supported by many lines of experimental evidence, as considered in our Perspective. Furthermore, while downplaying the biological mechanisms affecting fat synthesis, storage, and mobilization, the Scientific Statement rebrands the EBM in terms of body mass homeostasis and retains its central focus on energy intake and energy expenditure as the primary determinants of fat accumulation. To be clear, we do not question the fundamental scientific observations, accumulated over more than a century, on the biological influences on body weight; the CIM rests on it. We challenge unproven assumptions regarding a singular causal direction and a simplistic translation of the physical law of energy conservation, as exemplified in Supplemental Table 1 of our paper.

Third, we do not believe that the characterization of our Perspective as unscholarly promotes collegial discourse. Constructive debate about the obesity pandemic requires that opposing sides clarify their experimental models, which we have done, supported by extensive cited literature. However, we were unable to find any mechanistically oriented, environmentally focused, testable model for the EBM, and therefore assumed the task of defining it (i.e., in Supplemental Table 1 and other citations) for comparative purposes. To advance this debate, EBM proponents should clarify their model, specifying experiments that would demonstrate causal direction involving the chronic control of body weight.

Fourth, we are criticized for a press release published by the American Society for Nutrition accompanying our paper. However, societies commonly publish press releases for papers with broad potential interest. For instance, the American Association for the Advancement of Science and the journal Science published a press release accompanying a short opinion paper claiming to have disproven the CIM (3). The ASN press release accurately summarized the premise of our paper, that overeating is not a primary cause of obesity but rather a downstream effect in the mechanistic pathway. The press release considered the adolescent growth spurt, to illustrate how a positive energy balance could result from, not cause, a physiological state of increased body energy stores.

Admittedly, in any protracted debate on complex biological problems such as obesity, each side may feel misrepresented by the other, or that the vast body of evidence has been selectively cited. In the interests of constructive discourse, we do not claim our formulation of the CIM is proven, nor do we categorically reject other explanatory models. We explicitly recognize that our model is unlikely to be entirely accurate or precise, and we address potential areas of common ground that may already exist. We call on all sides of this debate to recognize that collegial discourse among opposing scientific camps may lead to new solutions for this intractable problem.

We thank Devinsky for the commendation. We agree that, for many people, refined carbohydrates (i.e., those with high glycemic load or high free-sugar content) comprise the main problem and do not take issue with his alternative title “refined carbohydrate-insulin model (RCIM).” For individuals with genetic or acquired predisposition, manifest as high endogenous insulin secretion or insulin resistance, a reduction in total carbohydrate may be optimal. (Other dietary exposures than carbohydrate may also affect adiposity through the proposed causal pathway.) We agree that the CIM and EBM may not be mutually exclusive, as mentioned in our Conclusion and further addressed elsewhere (4). We generally agree with Devinsky's other comments regarding the nutrition transition, dual-burden populations involving obesity and undernutrition, interpretation of the laws of thermodynamics, and the pitfalls of using short ward studies (e.g., <1 mo) to elucidate obesity, a chronic disease that manifests over years.

Booth and Gibson take issue with our paper, although on 1 point we agree, “physics does not state that body energy gain comes from an accumulation of the amount of energy intake exceeding the amount expended.” Supplemental Table 1 in our paper provides what we consider examples of tautological confusion in public health recommendations and clinical guidelines involving translation of this law of physics. We aimed to highlight, not propagate, this confusion. However, we disagree that “there are no clear physiological mechanisms by which increased fat deposition could drive weight gain,” as extensively considered in our Perspective. Notably, Booth's work dating to the 1970s provides early support for fundamental tenets of the CIM, by emphasizing metabolic fuel supply in the control of food intake and the contribution of adipose tissue uptake and mobilization of fat to that control (5, 6).

Although the effect of dietary composition on energy expenditure remains unproven, we believe current evidence supports this possibility. In the largest published trial of this question (7), total energy expenditure was ∼200 kcal/d greater with weight-maintenance diets with 20% versus 60% carbohydrate. Whereas Booth and Gibson cite a reanalysis, involving post hoc exclusion of 50% of the data, that challenges this conclusion, they disregard 2 subsequent papers that address this challenge and show the original findings are robust to a wide range of experimental assumptions (8, 9). Furthermore, the findings are consistent with published trials of ≥2.5-wk duration—i.e., those with sufficient duration to allow for physiological adaptation to a major change in macronutrients (10). In addition, our Perspective addresses the challenges in making causal inference from ecological comparisons, such as those with subsistence farming societies.

Finally, we recognize that Millward strongly disagrees with many aspects of our model. Space does not allow for considering these differences in necessary detail. Instead, we invite the reader to compare the strengths of the evidence on both sides. We would simply reiterate our conclusion, that “[r]igorous research using complementary designs will be needed to resolve the debate, clarify a middle ground, or point the way to new explanatory models that better encompass the evidence.”