Here, we use multidimensional nutritional modelling, together with established techniques in meta-analysis and meta-regression, to disentangle the complex relationship between diet, BCAAs and metabolic health. The first issue we addressed was the relationship between dietary BCAAs and blood levels of BCAAs. Overall, there was a positive association between dietary and blood levels of BCAAs in both the fasting and fed state, supporting the notion that circulating BCAA levels are likely to reflect long-term protein intake [1, 7, 10, 17]. The relationship, however, is more nuanced. In animals restricted to diets containing different amounts of BCAAs, there was a curvilinear relationship between dietary content of BCAAs and the blood levels of BCAAs, a finding consistent with our previous experimental work. When the dietary background level of BCAAs was lower than standard mouse chow (0.52 kJ/g), increasing dietary BCAAs resulted in elevated blood levels of BCAA. At higher levels dietary BCAA, however, adding more to the diet had little effect. We have previously seen a plateau in circulating BCAAs when blood levels reach about 40 μg/mL [10, 22], occurring consistently when either total dietary protein content or dietary BCAA levels are increased beyond a point. This relationship reflects the network of mechanisms that influence BCAA levels. Blood levels of BCAAs are primarily regulated by BCKDH, a mitochondrial enzyme complex found in the liver and muscle that catabolizes the ketoacid metabolites of BCAAs. Because insulin and BCAAs both activate BCKDH which acts to reduce BCAA levels [17], the mechanism for the plateau in BCAA blood levels when dietary content is high may be explained by a compensatory increase in BCKDH activation. This plateau occurs once food content of BCAA is greater than in standard diets and when blood levels of BCAA are about 40 mg/L [10, 22], suggesting that in these conditions, the catabolic capacity of BCKDH cannot be further downregulated.
Another mechanism by which an animal can regulate blood levels of BCAAs is by altering dietary intake. As essential amino acids, BCAAs are primarily acquired through dietary sources. In our meta-analysis, animals only had access to a single diet; therefore, the only option for increasing or decreasing BCAA intake is by changing feeding behaviour to consume more, or less food. Evidence for this response was apparent in the analysis of the relationship between food intake and dietary BCAAs. Animals on diets high in BCAAs ate 94% of the amount eaten by animals on low BCAA diets. However, the effect of BCAAs food intake that we observe is small, even in these experimental animals where dietary BCAA levels are often dramatically manipulated (e.g. ranging from 20 to 200% of standard amounts of BCAA [10]). The impact of BCAAs on food intake is, however, complex. While the general trend showed that BCAAs reduced food intake of animals on diets with high BCAA, low non-BCAA ratios, it is important to note that many studies did not experimentally control for protein content when manipulating BCAA levels. It remains uncertain whether this reduction in food intake is attributable to the satiating effect of increasing total dietary protein.
While high amounts of dietary protein can suppress food intake and protein intake is prioritized over intake of fat and carbohydrates [23], the role of individual amino acids and their mixtures on protein appetite and food intake is complex and not yet fully understood. We found an effect of dietary BCAAs on food intake consistent with animals having the capacity to regulate food intake according to BCAA content; however, the effect is small and is likely confounded by the overall total protein content and balance of amino acids. Imbalance of amino acids is also known to influence feeding behaviour, with the effect of suppressing or increasing food intake dependent on the nature of the manipulation. For example, diets extremely deficient or devoid in one or more essential amino acids result in food aversion [24]. However, when the deficiency is small enough to be leveraged by compensatory feeding, hyperphagia is observed [25]. When compared to control groups, reducing dietary levels of single amino acids such as methionine, threonine or isoleucine [2, 26, 27] or groups of amino acids such as essential amino acids or the BCAAs [4, 5, 27] sufficiently increases food intake. In addition to dietary availability, the interaction between amino acids in circulation can regulate food intake by influencing whether the amino acid precursors necessary for neurotransmitter production are transported across the blood-brain barrier in sufficient quantities. A recent example showed that a diet high in BCAAs but low in tryptophan reduces uptake of tryptophan into the brain by competing for transport across the blood-brain barrier by the LAT1 amino acid transporter [10]. As tryptophan is the sole precursor for serotonin synthesis, a neurotransmitter involved in the control of food intake [28], reduced levels in the brain led to lower brain serotonin levels, greatly increased food intake, obesity and shortened lifespan. All these effects occurred without activation of canonical ageing pathways such as MTOR and IGF1 [10]. While this increase in food intake on high BCAA diets appear at odds with the findings of this meta-analysis, this effect may be explained by the interaction between dietary BCAAs and the total protein content of the diet. Many studies that supplement dietary BCAAs also increase the total protein content of the diet, an effect which will have important implications for promoting satiety. Solon-Biet et al 2019, however, use a unique design where BCAAs levels were doubled compared to the control group, while keeping total protein content constant. In experiments where this is not controlled, the effect of total dietary protein is likely to dominate any effect on appetite of dietary BCAAs.
What are the implications for human studies of the finding of this meta-analysis and the curvilinear relationship between dietary BCAAs and blood levels of BCAAs? First, it must be emphasized that these animal studies involved restriction to a single diet. Humans, on the other hand, have access to multiple foods with different contents of BCAAs, and other components such as tryptophan which can interact with BCAA to influence appetite. Second, unlike human studies, animal studies are undertaken with homogeneous genotypes and environments. Humans often have conditions and diseases unrelated to BCAA intake, but which may influence BCAA levels via their impact on various anabolic (insulin, IGF-1, GH) and catabolic (TNFα, cortisol, catecholamines, glucagon, inflammatory cytokines) factors that influence BCKDH activity. Even so, we predict that the curvilinear relationship between dietary BCAA and BCAA blood levels seen in animals will be apparent in human populations because it is a consequence of regulatory networks shared with humans. Thus, for human studies, we predict that there will the strongest association between individual dietary and blood levels in populations/groups with comparatively low levels of dietary BCAAs. While a positive correlation might be statistically significant over an entire range of blood levels and intakes, this may misrepresent the underlying curvilinear nature of the relationship. It must be noted, however, that BCAA levels are tightly regulated in the fasting period, so it is not simply a case of more dietary BCAAs entering the blood and increasing BCAA levels. If an association was found between dietary BCAAs and blood levels (when the dietary BCAAs are high), this may be explained by an indirect or confounding association that impacts on the regulatory network—in particular, BCKDH. For example, people with obesity may consume a diet with higher amounts of BCAAs but also have insulin resistance which impairs BCAA catabolism [19].
Here, we studied the relationship between blood levels of BCAAs and dietary BCAAs, but not total dietary protein. A weak association between dietary protein and blood levels of BCAAs has been reported in humans, and stronger associations in animal studies where protein intake and content can be strictly controlled [7, 29]. Although BCAAs are only found in dietary protein, the amount of BCAAs varies substantially depending on the source and type of protein, which makes evaluating any association more uncertain.
The second question we addressed with this meta-analysis was whether there are effects of dietary BCAAs on glucose metabolism, and if so, are these moderated by nutrient background? There were four metabolic outcomes assessed (insulin, glucose, glucose AUC and HOMA). Only glucose AUC had a significant overall association with dietary BCAAs, but not any other traits related to glucose homeostasis. In human studies, it has usually been reported that there is a strong association between dietary BCAAs and /or blood levels of BCAAs with impaired insulin and glucose metabolism, metabolic syndrome and diabetes [13, 19]. Although ascertaining the direction of causality in epidemiological studies is difficult, the most widely accepted conclusion is that elevated BCAAs are a consequence of insulin-resistant states—rather than elevated BCAAs contributing directly to insulin and glucose dysmetabolism, although there is evidence supporting both hypotheses [19]. The results of our meta-analysis are consistent with that interpretation. That is, we found in otherwise healthy animals (i.e. not obese or diabetic) on diets with sufficient levels of BCAAs/protein, changes in dietary BCAAs alone were not associated with overall significant metabolic disturbance. Model fitting, however, showed that these results are more nuanced and can be influenced by background nutrition. For glucose AUC, we found that the largest effects of dietary BCAAs occurred when there was a simultaneous increase in non-BCAA content (i.e. increasing protein content), a finding consistent with studies in humans where it has been shown that people not-subject to protein restriction have higher fasting blood glucose [5]. Although these results are complicated, the unifying theme is that when increased dietary BCAAs reflect increasing dietary protein, there is an increased association with glucose dysmetabolism. An association between excess dietary protein, particularly from animal sources, and cardiometabolic disorders has been widely reported [22, 30]. Thus, any association between BCAA and metabolic disease is more likely to be a result of BCAA being a biomarker for the amount and type of dietary protein, rather than being an independent risk factor.
Important outstanding questions raised by our study are (1) how quickly the effects of dietary protein/BCAAs on glucose metabolism take hold and (2) the degree to which any effects are reversible. On the first question, it seems unlikely that all outcomes respond similarly quickly to dietary BCAAs, yet our analyses detected few moderating effects of study duration. However, it is important to point out that our search and analysis did not explicitly target longitudinal experiments on the effects of dietary BCAAs. Regarding reversibility, this question requires examination of the responses to a diet switching experiments, which was also beyond the scope of the current synthesis. Nonetheless, some such studies have been performed. For example, Cummings et al. [8] found that reducing BCAAs and total amino acids, after animals had been exposed to a ‘western diet’ reduced fat mass and glucose AUC implying a degree of plasticity. However, Hahn et al. [31] found that a late-life switch to dietary restriction, which involves a reduction of all nutrients in the diet (including amino-acids), did not result in the expected improvements in survival, implying irreparable damage from the preceding diet. The GFN-based meta-regression approach that we present allows the user to identify the key nutritional dimensions of major effect, and thus may help to unify the results of different diet-switch experiments.
Finally, we addressed the issue of body composition and BCAAs. Overall, dietary BCAAs were not associated with body mass or body fat in this meta-analysis. However, there were associations when the underlying diet was considered. Increased BCAAs were associated with increased bodyweight when the background diet was low in BCAAs. This is likely a result of the relationship we found with food intake, where BCAAs were associated with increased food intake when the background diet was low in BCAAs, reflecting behavioural mechanisms of animals to reach intake targets of limiting nutrients [20]. It is important to note, however, that the balance of amino acids in the diet, in addition to macronutrient background, may exert different effects on food intake. For example, reducing levels of other specific amino acids such as tryptophan, while simultaneously increasing BCAAs may impair central appetite signalling mechanisms and promote hyperphagia [10]. Our meta-analysis also found that increased dietary BCAAs were associated with elevated body fat when the diet was high in protein and low in carbohydrates. This is consistent with amino acid biochemistry whereby excess amino acids above those required for protein synthesis can either be utilized via gluconeogenesis or ketogenesis for energy production or indirectly via acetyl coA converted to fat and glycogen [1, 14].