closely regulated [e.g. 42–44]. Campbell et al. [45] allowed 63 subjects to freely compose a diet from foods containing 10, 15, and 25% protein for 3 days (Fig. 6.4b). Subjects closely tracked a mean 14.7% protein intake, which differed highly statistically significantly from the null expectation of no selection (16.7%). It is significant that the target intakes suggested by population studies and experimental studies converge within a narrow margin of 15–17%. Interestingly, this is not dissimilar for non‐human apes studied in the wild, which range between 10% (orangutan) and 20% (mountain gorilla), with our closest living relative, the chimpanzee, selecting a diet of 13% protein [35].
At a physiological level, there have been significant advances in understanding the mechanisms controlling macronutrient appetites [46]. Most notable has been the discovery that fibroblast growth factor (FGF)‐21 is the circulating signal of low‐protein status in humans and rodents. FGF‐21 is produced mainly in the liver and acts in the brain to stimulate protein appetite, guiding mice either to select protein‐rich foods if available or to increase intake of low‐protein diets to ensure increased protein intake, with associated increased energy intake on low‐protein, high‐energy diets [47,48]. FGF‐21 is also implicated in the inhibition of carbohydrate intake under low‐protein, high‐carbohydrate feeding in mice and humans [49–52].
Response to variation in dietary macronutrient balance: protein leverage
The fact that dietary percent protein is far more consistent across populations than fats and carbohydrates [e.g. 40] suggests that humans, like most other primate species studied to date, regulate protein intake more strongly than either fat or carbohydrate. An important question is how the strong appetite for protein interacts with the appetites for fat and carbohydrate in the face of variation in dietary macronutrient balance. Testing this interaction requires examining how the absolute amounts of macronutrients eaten vary with variations in dietary macronutrient ratios.
Figure 6.4 Dietary macronutrient regulation in humans. (a) Data from the FAOSTAT global nutrient supply database for the United States indicate that compared with fat and carbohydrate, which have increased since 1960, protein supply has remained stable, as expected if humans regulated protein intake most strongly. (b) Daily protein vs. non‐protein energy intake by 63 adult Jamaican volunteers, averaged for each subject over the 3‐day experimental period clustered tightly around a dietary macronutrient ratio of 14.7% energy from protein. Red symbols are females and blue symbols males. The solid radial lines are nutritional rails representing the composition of the three experimental menus from which the diets could be freely selected (10, 15, and 25% protein by energy). The shaded region represents the range of diets that could potentially be selected from these menus, and the dashed line represents the expected outcome if the subjects mixed the diets randomly (16.7% P). The selected macronutrient ratio was significantly different from the random outcome (P < 0.001).
Source: Adapted from Campbell et al. [45].
Global data from the FOASTAT nutrient supply database for the United States indicate that percent protein in the food supply has fallen by ~1% since 1960 (Fig. 6.5a), and concurrently energy supply has risen (Fig. 6.5b), yet absolute protein supply has stayed remarkably stable (Fig. 6.4a) [53,54]. Hence, paradoxically, although protein has made little direct contribution to excess calorie intake throughout the obesity epidemic, its tight regulation may have played a major role in driving excess calorie intake in the form of fats and carbohydrates. The proposal that a decline in percent protein in the diet has interacted with the strong human protein appetite to drive increased energy intake and the development of overweight and obesity is termed “the protein leverage hypothesis” [53,55].
That total energy intake is indeed leveraged by protein in humans has now been demonstrated in several controlled experimental studies [45,56–60] and in secondary analysis of compiled literature data [61,62] (Fig. 6.5c). The studies of Gosby et al. [57], conducted in Sydney, and Campbell et al. [45], in Jamaica, disguised the macronutrient composition of experimental foods and controlled for palatability, variety, and availability differences between treatments. Subjects were provided with menus containing 10, 15, or 25% protein for 4 or 5 days. Some foods and snacks were savory and others sweet in flavor characteristics, but all were of the same macronutrient composition for a given experimental period. Although the nature of the experimental foods and menus differed between Sydney and Jamaica for cultural reasons [63], the outcomes in terms of nutrient and energy intakes were closely similar between the two studies, with subjects ingesting most calories on the 10% protein diet. Notably, in the Sydney study, subjects ingested 12% more calories on the 10% protein diet than on the 15% protein diet, with these excess calories coming mainly from increased snacking between meals on savory‐flavored food options [57]. This behavior was associated with elevated FGF‐21 levels on the 10% protein diet [48]. Hence, subjects on the 10% protein treatment diet demonstrated behavioral and physiological characteristics of protein‐seeking behavior. The increased salience of savory (umami) flavor cues when in a state of protein deficit has also been shown in brain imaging studies [64,65].
The studies of Campbell et al. [45], Gosby et al. [57], and Hall et al. [60] all reported increased energy intake with the dilution of dietary protein. Two experimental studies by Martens et al. [58,59] reported increased calorie intake on experimental diets containing 15% relative to 30% protein but did not show further increase in energy intake on a 5% protein diet. However, 5% protein is not physiologically viable, and lack of increased energy intake most likely indicates that protein leverage has regulatory limits [55]. An analysis of a compilation of 38 published experimental trials measuring ad libitum intake in subjects confined to menus differing in macronutrient composition by Gosby et al. [61] and Raubenheimer et al. [62] indicated that protein leverage operates most strongly across a range from 10 to 30% protein, which reflects the normal range for human diets [see 55].
Figure 6.5 Response of human appetite systems to variation in dietary macronutrient ratios. FAOSTAT supply data for the United States show that the percentage of energy from protein decreased over several decades (a), and this decrease was associated with an increase in total energy intake (b) as predicted by protein leverage. (c) Experimental evidence for protein leverage. Data are absolute protein (x‐axis) and non‐protein (fat + carbohydrate, y‐axis) ad‐libitum energy intakes by subjects restricted to one of 138 experimental diets [61,62]. The black dashed radials represent the nutritional rails for the diets with the highest (54%) and lowest (5%) proportional protein content. The area between these radials is the region of the nutrient space within which points for nutrient intakes are constrained to lie, with the exact pattern of actual intakes being determined by the ways that appetites for protein, fat, and carbohydrate interact. The blue, red, and green lines represent the protein prioritization, NPE prioritization and equal‐weighting models from