Background
The worldwide obesity epidemic represents a major health burden, with obesity leading to a spectrum of comorbidities that are the basis for cardiometabolic diseases (CMD). CMD are multifactorial metabolic diseases that can include insulin resistance (IR), type 2 diabetes (T2D), and cardiovascular disease (CVD) [
1,
2], with other conditions like hyperlipidemia and hypertension also related to CMD. Circulating concentrations of the branched-chain amino acids (BCAAs), isoleucine, leucine, and valine have consistently been associated with CMD and CMD risk factors in various studies [
3‐
5]. BCAAs are essential amino acids strictly sourced from diet that have several physiological and metabolic roles and have been established to be risk factors for IR and diabetes [
6‐
8]. Recent literature, however, has challenged the role of BCAAs as mere risk factors by suggesting that they also play an etiological role in CMD development. For IR and diabetes, multiple prospective and mechanistic studies have now produced evidence for BCAAs playing a causal role. For instance, in a prospective cohort of 6244 individuals, higher baseline concentrations of BCAAs were associated with IR and found to predict incident T2D in a 7.5-year follow-up, independent of other risk factors [
9]. Similarly, higher concentrations of BCAAs also predicted the development of diabetes in women with a history of gestational diabetes and IR [
10], even in young non-glycemic adults [
11].
Mechanistic studies using mouse models have shed light on the pathways potentially involved. One leading hypothesis suggests that BCAAs, or their breakdown products (branched-chain keto acids), have a direct influence on key factors involved in the pathogenesis of diabetes through their interaction with the mammalian target of rapamycin (mTOR) signaling pathway [
3,
12] and through induction of oxidative stress [
13‐
16], mitochondrial dysfunction [
17,
18], apoptosis [
19,
20], and IR and/or impaired glucose metabolism [
3,
21‐
28]. In line with this, White et al. were the first to show that BCAA supplementation to a high fat diet downregulated the AKT pathway (pAKT), a marker of insulin signaling in muscle, leading to impaired glucose tolerance via hyperactivation of mTOR signaling in muscle [
29]. This finding is further supported by a study showing that BCAA metabolites downregulated pAKT [
25] and by other studies showing that BCAA supplementation with a high fat diet or defective BCAA oxidation in mice induces IR [
3,
26].
Relationships between BCAAs and other CMD traits are also emerging, but it is still unclear whether they represent important indicators of CMD risk or even play a causal role. In this regard, some of the most extensively studied CMD traits are CVDs, a set of diseases including atherosclerosis and myocardial infarction, that affect the heart and the circulatory system. Several studies demonstrated that increased concentrations of various BCAAs were associated with increased CVD risk [
30] (e.g., for hypertension [
31]) and that their presence in plasma is an independent predictor for adverse cardiovascular events [
32]. Yet, the BCAA–CVD nexus is not as straightforward for other CVDs and clinical contexts, e.g., for atherosclerosis. While a strong correlation between atherosclerosis and BCAAs has been reported, the association was not as clear in obese individuals [
33]. Indeed, the influence of BCAAs in CVD in obese subjects, as well as their role in the severity of obesity, has yet to be properly understood. This is mostly because atherosclerosis and body fat distribution are closely related to CVD, but there is also mounting evidence that fat-tissue metabolism is key in determining blood concentrations of BCAAs [
34,
35].
One way to better understand the correlation between BCAAs and CMD would be to look at a wide range of CMD-related parameters, including CVD risk factors [
36]. The OLINK panel provides a large toolkit of known CVD-related proteins and could potentially unravel a correlation of BCAAs and CMD [
37]. Several other emerging CVD biomarkers, such as trimethylamine N-oxide (TMAO) and its precursors, should also be considered for this screening given their recently confirmed correlation in T2D patients [
38‐
40]. Likewise, several other important components of CMD, such as lipid metabolism, inflammation, and immunity, are also potential biomarkers, but their relationships with BCAAs are poorly understood. The interplay between BCAAs and lipids in the onset of various CMD, such as IR and chronic hyperinsulinemia, has been widely suggested [
3,
25,
26,
41,
42], but the relationship between BCAAs and lipids has not been systematically explored, even though it could shed light on a common pathway to CMD. Similarly, BCAAs can pose as donors of nitrogen and carbon skeletons for the synthesis of other amino acids that have major roles in immune cell function [
43]. BCAAs have also been shown to activate inflammatory signals and increase the release of inflammatory cytokines, which exacerbates IR by blocking insulin signaling transduction in adipocytes and skeletal muscle cells [
44]. Still, as a nascent field of research, many aspects of BCAAs and their effects on inflammation and immune function, along with their concomitant role in CMD, are poorly explored, and a first step toward resolving this knowledge gap would be to systematically explore the relation of BCAAs and inflammatory and immune parameters in large human cohorts.
Taken together, the relationship between BCAAs and some components of CMD, such as IR/diabetes, have been extensively studied, but the relationship of BCAAs and other CMD parameters has yet to be established, as most studies to date are either pre-clinical or performed for sex- and disease-specific conditions. In addition, while many of the mechanistic studies in animal models have provided evidence of individual-level causality, systematic evaluation in human cohorts is crucial to provide population-level evidence and identify the potential etiological roles of BCAAs in CMD. Moreover, previous results about the direction of the causal relationship between BCAAs and CMD are conflicting. For example, a causal role for BCAAs in IR was supported in a study by Lotta et al. [
45], while reverse causality of IR on BCAAs has been suggested by other studies [
46,
47]. Here, we present a systematic, cross-sectional association analysis between fasting plasma concentrations of BCAAs and a large panel of 537 parameters (including clinical CMD measures such as non-invasive measures of atherosclerosis, fat distribution, circulating CVD-related proteins, plasma metabolites and inflammatory cell counts and immune cytokines) in 1400 individuals from the general population–based LifeLines DEEP (LLD) cohort [
48] and 294 overweight/obese individuals from 300OB the cohort [
49]. In this study, we (1) establish association relationships between BCAAs and CMD-related traits that are independent of age, sex, BMI, and other potential covariates; (2) estimate and compare the association strength between different BCAAs; and (3) interrogate the potential causal direction of associations using a bi-directional Mendelian randomization (MR) approach.
Discussion
Valine, leucine, and isoleucine—the BCAAs—are essential amino acids that are used for protein synthesis but also play important roles in signaling pathways. Increased plasma BCAA concentrations have been found to increase the risk of IR and diabetes and some causal pathways have been proposed [
3,
12,
14‐
16], but their relation with other CMD-related parameters is not completely understood. Previous studies investigated the associations of BCAAs with CMD using only a limited number of CMD phenotypes, or in sex-, age- or disease-specific cohorts with small sample sizes, making it difficult to generalize their findings [
61‐
63]. Here, we made use of very detailed clinical and omics-guided phenotyping data available for 1694 individuals from two cohorts, a general population cohort and an overweight/obese cohort of older individuals with high cardiovascular risk, to uniformly profile BCAA relations with common CMD-related parameters. We systematically explored the relationships between BCAA concentrations and a large panel of 537 CMD parameters and provided an estimate of causality of these associations using MR.
We created an inventory of 827 significant BCAA associations with CMD-related parameters. We not only replicated previous findings of associations with glycemic-related factors and all three BCAAs [
4,
41], we also detected a large number of associations for a wide range of CMD-related parameters, particularly with lipoproteins and circulating protein levels. We identified fewer BCAA associations with cytokines, cell counts and TMAO metabolites.
BCAAs appear to have both common and specific associations. A large set of available CMD parameters allowed us to accurately compare the association effects between isoleucine, leucine, and valine. Under the presumption that these three BCAAs have similar functions, many studies have missed the specific effects of each BCAA and broaden their conclusions based on the study of single amino acids and/or of their individual effects to all BCAAs. It is becoming increasingly clear that the associations of specific BCAAs with CVD may differ substantially [
64,
65]. The largest currently recognized difference is for leucine, which is known to have slightly different cellular functions compared to valine and isoleucine [
66,
67]. Based on our findings, it appears that the associations with glycemic traits are shared across all three BCAAs; however, for other CMD parameters, we found a large number of specific associations, especially the associations between isoleucine and circulating lipoprotein and protein levels. These findings are consistent with a recent study proposing that both isoleucine and valine were responsible for the adverse metabolic effects of BCAAs in mice, with a particularly clear effect for isoleucine [
65]. Furthermore, in our study, only isoleucine was associated with the number of carotid artery plaques, suggesting that it would be prudent to specify BCAAs when evaluating the CMD etiology. Based on the associations demonstrated in the current study, it would be plausible to consider isoleucine in clinical and experimental research on atherosclerotic plaques.
Another important finding is the relationship of BCAAs, obesity, and lipid metabolism. As noted above, BCAAs appear to have particularly strong associations with lipid profile. We found that these associations depend on age and BMI. First, effect sizes were higher in the 300OB cohort of overweight/obese individuals, who were also considerably older. Second, valine showed a large number of associations with opposing directions across the two cohorts, a conflict that was resolved after matching for age and BMI of individuals from the two cohorts.
While BCAAs clearly play a role in IR and diabetes, previous studies reported conflicting results on the direction of causal relationship in these associations: Lotta et al. reported that changes in BCAA levels contribute to IR and the incidence of type 2 diabetes [
45], whereas more recent studies showed evidence of a BCAA effect on IR [
46,
47]. Our results support the potential causal effect of fasting insulin and glucose on BCAA levels. In addition, we see multiple lipid-related causal links, e.g., total cholesterol levels were potentially causally related to a decrease in BCAA levels. Additionally, we observed that many NMR-based lipoproteins showed a causal effect on BCAA levels. However, these results should be interpreted with caution because of a strong correlation between the lipid traits and a high heterogeneity of the resulting MR estimates. No significant causal links from BCAA levels to CMD-related parameters were detected, potentially due to the lack of strong genetic instruments for this analysis, which may indicate that altered plasma BCAA levels are more likely to be the outcome of the metabolic syndrome. Further studies in other datasets, including non-European populations, are required to estimate the generalizability of the study results.
We acknowledge several limitations. This cohort-based, epidemiological study primarily provided population-level association evidence, as well as future directions for investigating the role of BCAAs in CMD. However, the cross-sectional association design of the current study led to several limitations. First, the identified associations may not directly reflect the observations from animal-based, mechanistic biology studies, as experimental settings vary greatly. Second, our association approach assumes linear relationships between BCAAs and CMD parameters, which might not always be the case. Larger studies using non-linear models are needed to capture these complex relationships. Third, the identified associations do not imply causality. While we did employ MR to estimate directions of causality for these associations, only a few causal relationships were supported, and these were predominantly in the direction from CMD parameters to BCAA levels. However, the causal effects of BCAAs on CMD parameters were more difficult to estimate due to the lower number of BCAA-associated SNPs available for the analyses. Lastly, violation of assumptions for the MR analyses may occur even when performing a rigorous sensitivity analysis, potentially leading to false conclusions. Further studies are needed to elucidate the potential underlying mechanisms of the identified associations and causal links.
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.