The effect of physical activity level and exercise training on the association between plasma branched-chain amino acids and intrahepatic lipid content in participants with obesity

Elevated plasma branched-chain amino acids (BCAA) are a hallmark of type 2 diabetes (T2DM) and are associated with insulin resistance and mitochondrial dysfunction. Prior studies have shown positive associations between plasma BCAA and intrahepatic lipid (IHL) content, and animal data suggest BCAA may stimulate hepatic lipogenesis. However, the determinants of elevated BCAA are unclear and whether physical activity or exercise training modulate the relationship between BCAA and hepatic steatosis has not been investigated. This study asked: (1) Is the association between plasma BCAA and IHL content influenced by habitual physical activity level? (2) Does a structured exercise training program lower plasma BCAA in parallel with reductions in IHL in individuals with NAFL or T2DM? The authors hypothesized that higher physical activity and exercise training would be associated with lower plasma BCAA and decreased IHL.

The paper summarizes evidence that: (a) plasma BCAA are elevated in T2DM and even before disease onset; (b) BCAA are strongly associated with insulin resistance and mitochondrial dysfunction; (c) dietary protein intake does not consistently explain elevated BCAA; (d) mechanisms proposed include reduced insulin-mediated suppression of proteolysis and impaired mitochondrial BCAA metabolism; (e) cross-sectional human cohorts (e.g., Young Finns Study) found positive associations of plasma BCAA with IHL, and longitudinal data suggest changes in BCAA may predict NAFL development; (f) animal studies indicate BCAA may promote hepatic lipogenesis via activation of lipogenic genes (e.g., SREBP-1c, FAS, ACC); (g) physical activity has been associated with lower plasma BCAA in observational data, yet controlled training studies have reported improved insulin sensitivity without reductions in BCAA. Notably, previous studies rarely assessed IHL alongside BCAA and physical activity/exercise.

Cross-sectional study: Participants were 1983 adults (45–65 years) from the Netherlands Epidemiology of Obesity (NEO) study, with oversampling of individuals with overweight/obesity. Baseline data included demography, lifestyle, medical history, and fasting blood sampling. Physical activity was self-reported via the SQUASH questionnaire and categorized by frequency of at least 30 minutes of moderate intensity activity: less than two times per week vs two or more times per week. IHL content was measured by proton magnetic resonance spectroscopy (1H-MRS) on a 1.5-T whole-body scanner. Stored fasting plasma samples (−80°C) collected in 2008–2012 had amino acids quantified in 2015 using the Nightingale NMR platform; metabolite levels were standardized (mean 0, SD 1). Analyses were weighted to the BMI distribution of the general Dutch population. Linear regression estimated differences in amino acid levels by sex and physical activity group, and the association of standardized fasting amino acid levels with log-transformed IHL, adjusting for age, sex, total body fat, alcohol intake, energy intake, and leisure-time physical activity. Interaction terms tested effect modification by sex and physical activity frequency. Coefficients were back-transformed to express relative change in IHL per SD of amino acid. Intervention study: Secondary analysis of samples from a prior 12-week exercise trial. Participants were male: seven with NAFL (IHL ≥ 5.0% by 1H-MRS and fasting glucose < 7.0 mmol/L), seven with T2DM (on oral agents for ≥6 months), and seven BMI-matched controls (CON); sedentary at baseline and asked to maintain dietary habits. Exclusions included active cardiac disease and impaired renal/hepatic function. Exercise training comprised three sessions/week: two 30-minute cycling sessions at 70% Wmax and one resistance session (eight exercises for large muscle groups; three sets of 10 repetitions at 60% MVC). Warm-up and cool-down were 5 minutes at 45% Wmax. Wmax and MVC were reassessed at 6 and 4 weeks, respectively, to progress training. Testing occurred 4 days before training and 48–72 hours after the final session. IHL was measured on a 3-T scanner (Philips Achieva 3Tx). A primed continuous infusion of [6,6-2H2]glucose preceded a two-step hyperinsulinemic-euglycemic clamp: low-dose insulin 10 mU/m2/min for 4 hours (assessing hepatic insulin sensitivity) and high-dose insulin 40 mU/m2/min for 2 hours (assessing peripheral insulin sensitivity), with target glucose 5–5.5 mmol/L. Glucose kinetics (Ra, Rd) were computed using Steele’s non–steady state equations; endogenous glucose production (EGP) was Ra minus exogenous infusion. Hepatic insulin sensitivity was percent insulin-suppressed EGP during low-dose insulin; peripheral insulin sensitivity was insulin-stimulated glucose disposal (Rd) during high-dose insulin. Due to technical issues, Rd and insulin-suppressed EGP were available in six T2DM and six NAFL participants. Amino acids were quantified by LC-MS after acetonitrile deproteinization; calibration curves established linear ranges. Total BCAA was valine + isoleucine + leucine; total aromatic AAs (AAA) was phenylalanine + tryptophan + tyrosine. The insulin-suppressive effect on plasma BCAA was the percent reduction from basal to high-insulin phase. Statistics: Cross-sectional comparisons used weighted linear regression with specified covariates and interaction tests; significance p ≤ 0.05. Intervention baseline differences used one-way ANOVA; training effects used linear mixed models with Bonferroni post hoc tests; Pearson correlations assessed associations. A priori sample size for correlation r = 0.57 yielded 22 participants for 80% power at α = 0.05.

Cross-sectional: • Mean plasma isoleucine, leucine, and valine in the NEO cohort were 50.3 ± 14.7, 66.0 ± 13.7, and 153.9 ± 27.6 μmol/L, respectively. • Dietary protein intake showed near-zero correlations with plasma amino acids. • Median IHL was higher in men than women; and lower in those reporting ≥2 sessions/week of ≥30 min moderate activity vs less active (2.0% [IQR 1.2–5.6%] vs 3.3% [1.6–7.7%]). • Plasma BCAA were higher in men than women. • Standardized plasma BCAA were positively associated with IHL after multivariable adjustment: valine 1.29× (95% CI 1.21–1.38), isoleucine 1.52× (1.43–1.61), leucine 1.54× (1.44–1.64) IHL per SD increase. Associations were somewhat stronger in women (sex interaction p < 0.05). • More active individuals had lower isoleucine and leucine than less active individuals, yet the BCAA–IHL associations were similar across activity strata (interaction p > 0.5 for all). Intervention: • At baseline, NAFL had the highest IHL; T2DM > CON; peripheral and hepatic insulin sensitivity were ~60% and ~38% higher in CON vs NAFL/T2DM. • Total AAA and total BCAA were higher in NAFL vs CON (ANOVA p = 0.015 and p = 0.006, respectively). Individual AA differences: tryptophan, tyrosine, phenylalanine higher in NAFL vs CON; valine, leucine, isoleucine higher in NAFL vs CON; isoleucine higher in T2DM vs CON. • Fasting total BCAA correlated positively with IHL (p < 0.05); correlated negatively with peripheral insulin-stimulated glucose disposal (p < 0.05). Hepatic insulin sensitivity (insulin-suppressed EGP) correlated with total BCAA and with isoleucine (r = 0.62, p < 0.01), valine (r = 0.54, p < 0.01), and leucine (r = 0.56, p < 0.05). • During clamps, insulin-mediated suppression of BCAA was blunted in NAFL and T2DM vs CON for isoleucine and leucine (ANOVA p = 0.004 and p = 0.001), and lower for valine in NAFL vs CON (ANOVA p = 0.01). • Exercise training did not change total plasma BCAA or AAA within or across groups (pooled BCAA 414 ± 11 to 425 ± 11 μmol/L; AAA 148 ± 5 to 152 ± 5 μmol/L). • Exercise reduced IHL in NAFL (11.6 ± 3.0% to 8.1 ± 2.0%, p < 0.05) and CON (2.4 ± 0.6% to 1.6 ± 1.4%, p < 0.05); improved peripheral insulin sensitivity in NAFL by ~23% (p < 0.05).

The study demonstrates a robust positive association between plasma BCAA and intrahepatic lipid content independent of habitual physical activity level, indicating that higher activity does not attenuate the BCAA–IHL link. In a controlled exercise intervention, hepatic fat content decreased and insulin sensitivity improved (notably in NAFL), yet plasma BCAA concentrations remained unchanged. These findings suggest that while physical activity and exercise beneficially impact hepatic steatosis and insulin sensitivity, the mechanisms underlying these improvements are not mediated by reductions in circulating BCAA. The blunted insulin-induced suppression of BCAA in NAFL and T2DM relative to controls points to impaired amino acid metabolic regulation in insulin-resistant states, potentially reflecting altered proteolysis or impaired BCAA catabolism. Overall, the results address the research questions by showing (1) physical activity frequency does not modify the BCAA–IHL association, and (2) exercise-induced reductions in IHL occur without concomitant decreases in plasma BCAA.

Across a large population-based cohort, plasma BCAA levels were positively associated with liver fat, and this association did not differ by habitual physical activity frequency. In a 12-week combined aerobic–resistance training program, individuals with NAFL and controls reduced IHL and NAFL improved peripheral insulin sensitivity without changes in circulating BCAA. Thus, exercise-induced improvements in liver fat and insulin sensitivity are not linked to reductions in plasma BCAA. Future studies should elucidate mechanisms connecting BCAA metabolism with hepatic steatosis and determine whether targeting BCAA catabolism can modulate IHL and insulin sensitivity in larger, diverse cohorts.

Cross-sectional analyses relied on self-reported physical activity and cannot infer causality; residual confounding may persist despite adjustment. Amino acids were measured in stored samples several years after collection. The cohort oversampled overweight/obesity, which may limit generalizability despite weighting. In the intervention, sample sizes were small (n = 7 per group), all participants were male, and due to technical issues clamp-derived measures were missing for some participants. Participants for amino acid analyses were selected from a prior trial, and selection based on exercise-induced IHL changes may introduce bias. The 12-week duration may be insufficient to detect changes in circulating BCAA, and dietary intake was not tightly controlled beyond instructions to maintain usual habits.