Association of Modifiable Lifestyle Factors with Plasma Branched-Chain Amino Acid Metabolites in Women

Abstract

Background: Circulating branched-chain amino acids (BCAAs-isoleucine, leucine, and valine) are strongly associated with higher risk of incident type 2 diabetes (T2D); however, determinants of elevated fasting BCAA concentrations are largely unknown.

Objectives: We aimed to characterize the modifiable lifestyle factors related to plasma BCAAs.

Methods: We performed a cross-sectional analysis among n = 18,897 women (mean ± SD age: 54.9 ± 7.2 y) in the Women's Health Study, free of T2D and cardiovascular disease at baseline blood draw. Lifestyle factors, weight, and height were self-reported via questionnaire, including smoking status, alcohol, leisure-time physical activity (LTPA), diet quality scores [2010 Alternative Healthy Eating Index (without alcohol) (aHEI); alternate Mediterranean Diet (aMED)], and dietary sources of BCAAs. Plasma BCAAs were quantified via NMR spectroscopy. We calculated multivariable-adjusted percentage mean differences (95% CIs) and P values for linear trend of BCAAs stratified by categoric lifestyle factors. We estimated R2 from univariate cubic spline regression models to estimate the variability in BCAAs explained.

Results: Compared with women with BMI (in kg/m2) <25.0, BCAAs were 8.6% (95% CI: 8.0%, 9.3%), 15.3% (95% CI: 14.4%, 16.3%), and 21.0% (95% CI: 18.2%, 23.9%) higher for the BMI strata 25.0-29.9, 30.0-39.9, and ≥40.0, respectively (P-trend < 0.0001). Women with higher LTPA and higher alcohol intake compared with lower had modestly (∼1%) lower plasma BCAAs (P-trend = 0.014 and 0.0003, respectively). Differences in smoking status, aHEI, and aMED score were not related to plasma BCAAs. Women with higher dietary BCAAs had dose-response higher plasma BCAA concentrations, 3.4% (95% CI: 2.5%, 4.4%) higher when comparing the highest with the lowest quintile (P-trend < 0.0001). BMI explained 11.6% of the variability of BCAAs, whereas other factors explained between 0.1% and 1%.

Conclusions: Our findings among a large cohort of US women indicate that BMI, but less so diet, physical activity, and other lifestyle factors, is related to plasma BCAAs.This trial was registered at clinicaltrials.gov as NCT00000479.

Keywords: branched-chain amino acids; diabetes; diet; lifestyle factor; physical activity.

© The Author(s) 2022. Published by Oxford University Press on behalf of the American Society for Nutrition.

Figures

FIGURE 1

Percentage differences and geometric means of circulating total BCAAs in relation to lifestyle factors based on adjusted linear models (n = 18,897). Values are geometric mean (95% CI) BCAA concentrations (μmol/L). The graphs show covariate-adjusted least-square means and percentage differences with the healthiest groups serving as the references, based on linear regressions of log-transformed circulating total BCAAs by modifiable lifestyle factors. Models were adjusted for age at blood draw (continuous), race, family history of type 2 diabetes, menopausal status [premenopausal, postmenopausal (natural or nonnatural menopause), unsure], use of postmenopausal hormone therapy (never, past, current), parity as number of pregnancies lasting ≥6 mo (0, 1, 2, ≥3), the use of blood pressure–lowering drugs and cholesterol-lowering drugs, total calorie intake (kcal/d), total LTPA (quintiles), aHEI (quintiles), alcohol intake (never, 0.1 to <5 g/d, 5 to <15 g/d, 15 to <30 g/d, ≥30 g/d), and smoking status (never, ever, current), in addition to BMI (kg/m2; continuous). Models for LTPA, diet quality, and smoking habit were not adjusted for total LTPA, aHEI, and smoking status, respectively. We tested for linear trends across exposure categories using category medians, except for smoking (never/ever/current) status for which we performed a 2-df likelihood ratio test. P values were Bonferroni method–adjusted. aHEI, 2010 Alternative Healthy Eating Index (without alcohol); aMED, alternate Mediterranean Diet; BCAA, branched-chain amino acid; LTPA, leisure-time physical activity; Q, quintile.

FIGURE 2

Percentage differences of circulating total BCAAs across quintiles of aHEI component food category scores (n = 18,897). Values are the median scores of each quintile group in each food category. The graphs show covariate-adjusted percentage differences based on linear regressions of log-transformed circulating total BCAAs by the 10 scores of the aHEI component food categories (vegetables, fruits, whole grains, SSBs and fruit juice, nuts and legumes, red/processed meat, percentage energy of trans FAs, long-chain (n–3) FAs, PUFAs, and sodium) in quintiles. The highest quintiles (the healthiest dietary pattern) served as the references. Models were adjusted for age at blood draw (continuous), race, family history of type 2 diabetes, menopausal status [premenopausal, postmenopausal (natural or nonnatural menopause), unsure], use of postmenopausal hormone therapy (never, past, current), parity as number of pregnancies lasting ≥6 mo (0, 1, 2, ≥3), the use of blood pressure–lowering drugs and cholesterol-lowering drugs, total calorie intake (kcal/d), total leisure-time physical activity (quintiles), alcohol intake (never, 0.1 to <5 g/d, 5 to <15 g/d, 15 to <30 g/d, ≥30 g/d), and smoking status (never, ever, current), in addition to BMI (kg/m2; continuous). We tested for linear trends across exposure categories using category medians. P values were Bonferroni method–adjusted. aHEI, 2010 Alternative Healthy Eating Index (without alcohol); BCAA, branched-chain amino acid; FA, fatty acid; Q, quintile; SSB, sugar-sweetened beverage.

FIGURE 3

Percentage differences of circulating total BCAAs across quintiles of animal/plant protein and dietary BCAA intake (n = 18,897). Values are the median scores of each quintile group in each food category. The graphs show covariate-adjusted percentage differences based on linear regressions of log-transformed circulating total BCAAs by (A) animal and plant protein and (B) dietary BCAA intake in quintiles. The lowest quintiles (the least intake) served as the references. Models were adjusted for age at blood draw (continuous), race, family history of type 2 diabetes, menopausal status [premenopausal, postmenopausal (natural or nonnatural menopause), unsure], use of postmenopausal hormone therapy (never, past, current), parity as number of pregnancies lasting ≥6 mo (0, 1, 2, ≥3), the use of blood pressure–lowering drugs and cholesterol-lowering drugs, total calorie intake (kcal/d), total leisure-time physical activity (quintiles), alcohol intake (never, 0.1 to <5 g/d, 5 to <15 g/d, 15 to <30 g/d, ≥30 g/d), and smoking status (never, ever, current), in addition to BMI (kg/m2; continuous). Animal and plant protein were mutually adjusted in model A. We tested for linear trends across exposure categories using category medians. P values were Bonferroni method–adjusted. BCAA, branched-chain amino acid.