Changes of Branched-Chain Amino Acids and Ectopic Fat in Response to Weight-loss Diets: the POUNDS Lost Trial

Abstract

Context

Recent evidence has related circulating branch-chained amino acids (BCAAs) to ectopic fat distribution.

Objective

To investigate the associations of changes in plasma BCAAs induced by weight-loss diet interventions with hepatic fat and abdominal fat, and potential modification by different diets.

Design, Setting, and Participants

The current study included 184 participants from the 2-year Preventing Overweight and Using Novel Dietary Strategies (POUNDS Lost) trial with repeated measurements on plasma BCAAs, hepatic fat, and abdominal fat over 2 years.

Main Outcome Measures

Repeated measurements of hepatic fat, abdominal fat distribution, including visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT), and total adipose tissue (TAT).

Results

Over 2 years, a decrease in total plasma BCAAs was significantly associated with improvement in hepatic density (a marker for hepatic fat; P = 0.02) and reductions in abdominal fat, including VAT, SAT, and TAT (all P < 0.05) in the main analyses. Additionally, we observed that decreases in BCAAs were associated with decreased insulin, homeostasis model assessment of insulin resistance, and triglycerides, independent of weight loss (all P < 0.05). Moreover, we found that dietary protein intake significantly modified the relation between changes in total plasma BCAAs and hepatic density at 6 months (P_interaction = 0.01). Participants with a larger decrease in total BCAAs showed a greater increase in hepatic density when consuming a high-protein diet, compared with those with a smaller decrease or increase in total BCAAs.

Conclusions

Our findings indicate that weight-loss diet-induced decrease in plasma BCAAs is associated with reductions of hepatic and abdominal fat. In addition, dietary protein intake may modify these associations.

Compelling evidence indicates that overweight and obese individuals have elevated circulating levels of branched-chain amino acids (BCAAs; i.e., leucine, isoleucine, and valine) (1, 2). Previous studies also associated blood levels of BCAAs with various metabolic disorders, such as insulin resistance (3-6), type 2 diabetes (7, 8), and liver fat accumulation (9-12). BCAAs are known to interfere with insulin signaling via mediating the activations of several critical metabolic signaling pathways in the liver, such as insulin resistance and glucose regulation (1), which are closely related to ectopic fat distribution (13-16). Increased levels of circulating BCAAs may induce insulin resistance by the persistent activation of the mammalian target of rapamycin complex 1 and the serine phosphorylation of insulin receptor substrate 1 3-6). Subsequently, insulin resistance may elicit an impairment of BCAAs oxidative metabolism in some tissues, such as the liver (3, 17). However, previous studies investigating the association between BCAAs and ectopic fat were largely limited by either cross-sectional design or small sample sizes (9-12).

BCAAs are essential amino acids that can only be obtained from foods. It has been shown that circulating amino acids could be affected by dietary factors, such as macronutrients (18-21). However, few studies have investigated the association between changes in BCAAs and changes in ectopic fat in response to dietary weight-loss interventions (9-12). We hypothesized that weight-loss diets varying in macronutrient-induced changes in plasma BCAAs might be related to ectopic fat storage, particularly, hepatic and abdominal fat distribution.

In this study, we investigated the associations between diet-induced changes in plasma concentrations of BCAAs (total and individual component) and hepatic and abdominal fat distribution among overweight and obese adults who participated in the Preventing Overweight Using Novel Dietary Strategies (POUNDS Lost) Trial. Because BCAAs are abundant in the high-protein diet, we also tested whether a high-/average-protein diet could modify these associations.

Research Design and Methods

Study participants

The POUNDS Lost trial is a 2-year randomized clinical trial designed to compare the effects of 4 energy-reduced diets varying in macronutrient compositions of fat, protein, and carbohydrate on weight loss. It was conducted from October 2004 to December 2007 at 2 sites: the Harvard T.H. Chan School of Public Health and Brigham and Women’s Hospital, Boston, MA, and the Pennington Biomedical Research Center (PBRC) of the Louisiana State University System, Baton Rouge, LA. The study design and primary results have been described elsewhere (22). Briefly, 811 overweight or obese participants aged 30 to 70 years old were randomly assigned to 1 of the 4 diets. The target percentage of energy from fat, protein, and carbohydrate were: (1) 20%, 15%, and 65%; (2) 20%, 25%, and 55%; (3) 40%, 15%, and 45%; and (4) 40%, 25%, and 35%, constituting a 2-by-2 factorial design. Among the 811 participants, 714 (88%) were followed up at 6 months, and a total of 645 participants (80%) completed the trial after 2 years.

Of all the participants, a random sample of ~25% of participants was selected to undergo computed tomography (CT) scans. Of the ~25% participants (N = 194), the follow-up rates were comparable to the overall trial: 90% (175/194) at 6 months and 84% (163/194) at 2 years. In the current analysis, we included 184 participants at baseline who underwent CT scanning and who had available data on BCAAs. During the follow-up, 140 records were available at 6 months and 104 at 2 years. The study was approved by the human subjects committee at each institution and by data and safety monitoring board appointed by the National Heart, Lung, and Blood Institute. All participants provided written informed consent.

Measurements

BCAA assessment.

Fasting plasma samples were stored at –80°C since collection. The profiling of amino acids was measured by electrospray tandem mass spectrometry at the Institute of Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics, University Hospital Leipzig, Leipzig, Germany. The amino acids leucine and isoleucine were measured as 1 composite, indicated as “leucine/isoleucine” in the current study. Total plasma BCAAs were calculated as the sum of leucine/isoleucine and valine. Details of the profiling procedure have been described elsewhere (23).

Hepatic fat and abdominal fat distribution assessment.

CT scanning was performed with GE High-Light Computed Tomographic scanner (Milwaukee, WI) at Brigham and Women’s Hospital, and with GE LightSpeed–VCT (Milwaukee, WI) at PBRC at baseline, 6 months, and 2 years. The accuracy of the 2 instruments was verified by be a CT phantom; no correction was applied because the 2 instruments were comparable and stable over time (24). All the CT scans were analyzed at PBRC using Analyze Direct software. As a direct measure of hepatic fat, images from 8-slice CT scans were used to derive the liver attenuation, which is the difference between the hepatic density and the spleen density (25). A higher hepatic density indicates a lower hepatic fat content. This method showed a high correlation with the histologically confirmed macrovesicular hepatic steatosis (25).

Cross-sectional areas of the visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) were analyzed on a series of 8 images acquired every 10 cm across the abdomen with two below L4-5 0-point landmark, 1 at the L4-5 0 point, and 5 above the L4-L5 vertebral interspace (26, 27). Total adipose tissue (TAT) was calculated as the sum of each individual slices (24, 28). Reader variability (coefficient of variation) was 0.9% on average (24).

Body composition.

A random sample of ~50% of the total participants from the POUNDS Lost trial were selected to undergo dual-energy X-ray absorptiometry scans for body composition at baseline, 6 months, and 2 years (24). Lean body mass, whole body total percent lean mass, whole body total fat mass, whole body total percent fat mass, and percent trunk fat were measured.

Covariate assessment.

Body weight and waist circumference were measured in the morning before breakfast on two days at baseline, 6, 12, 18, and 24 months. Height was measured at the baseline examination. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters (kg/m²). Blood samples were collected in fasting state and stored at –80°C at baseline, 6 months, and 2 years. The analyses of glucose, insulin, and serum lipids were performed at the PBRC. Glucose and insulin were measured by an immunoassay with chemiluminescent detection on the Immulite analyzer (Diagnostic Products, Los Angeles, CA). Insulin resistance was estimated by the homeostasis model assessment of insulin resistance (HOMA-IR) using fasting insulin and fasting glucose (29). Hemoglobin A1c, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and triglycerides were measured on Synchron CX5 (Beckman Coulter, Brea, CA). Regular use of medications to control cholesterol was collected through the medical history questionnaire at baseline.

Statistical analysis

The primary outcomes in the current study were hepatic density and abdominal fat distribution (SAT, VAT, and TAT). Secondary outcomes included the markers of lipids and glucose metabolism. BCAAs, insulin, HOMA-IR, and triglycerides were log-transformed before analysis to improve normality. General linear models for continuous variables and χ ² test for categorical variables were performed to compare the difference of baseline characteristics across tertiles of total plasma BCAAs. Linear mixed-effect models with 3-times repeated measurements (baseline, 6 months, and 2 years) were applied to examine the association of BCAAs (total and individual component) with hepatic density and abdominal fat distribution during the 2 years, after adjustment for age, sex, ethnicity, diet group in model 1. We further adjusted for baseline BMI and the use of cholesterol-lowering medications in the main analyses model 2. To further explore whether such a relationship was independent of weight loss, we additionally controlled for concurrent weight loss (at 2 years) in model 3. Similar models were applied to analyze the associations of BCAAs (total and individual component) with markers of glucose and lipids metabolism. We also tested the main effect of diet-induced changes in BCAAs from baseline to 6 months on the concurrent changes in hepatic density by a general linear model, adjusted for age, sex, ethnicity, diet group, baseline BMI, use of cholesterol-lowering medications, baseline values of hepatic density, and baseline values of BCAAs. To test the potential modification effect by dietary protein intake, we included a ∆BCAAs × (high-/average-protein diet) in the previous model, leveraging the factorial design of POUNDS Lost trial (2 diets were high-protein and 2 were average-protein; 2 were high-fat and 2 were low-fat).

Among the participants who were randomly selected to undergo CT scan, 140 of 175 had available CT scanned data at 6 months and 104 of 163 at 2 years. To address the missingness of CT data, we performed sensitivity analyses. We conducted multiple imputations for those with missing CT data at 6 months or 2 years, assuming missing at random (30). We first filled the missing values to get 5 complete datasets; then, we applied linear regression models to analyze the 5 complete datasets, and, finally, the results were synthesized. The multiple imputation procedure was implemented using PROC MI and PROC MIANALYZE in SAS.

Statistical analyses were conducted with SAS, version 9.4 (SAS Institute Inc, Cary, NC). All P values were 2-sided and P < 0.05 was considered statistically significant.

Results

Table 1 shows the baseline characteristics of the 184 participants included in the current analysis from the POUNDS Lost trial. The average age of the study participants was 54.0 years; 58.7% of the participants were female and 91.9% were white. All the participants were overweight or obese with a mean BMI value of 32.4 kg/cm². There was no difference between the included 184 participants and the remaining participants from the entire POUNDS Lost, except for a slightly higher proportion of white and older participants in the included sample (all supplementary material and figures are located in a digital research materials repository) (31). Such imbalance could be explained by the large proportion of white people in POUNDS Lost and the relatively small sampling fraction. In addition, age, gender, and ethnicity were included in the models in our analyses.

Table 1.

Baseline Characteristics of the Study Participants (N = 184)

Characteristics	Values
Age, years	53.96 (8.16)
Female	108 (58.70)
Race or ethnicity
White	169 (91.85)
Black	11 (5.98)
Hispanic	3 (1.63)
Other	1 (0.54)
Dietary intake per day
Total energy, kcal	1965.81 (602.73)
Carbohydrate, %	45.00 (7.74)
Fat, %	37.10 (6.15)
Protein, %	18.20 (3.27)
Weight, kg	92.64 (16.91)
BMI, kg/m2	32.41 (3.96)
Waist circumference, cm	104.47 (13.91)
Systolic BP, mm Hg	121.98 (12.47)
Diastolic BP, mm Hg	76.85 (9.17)
Glucose metabolism
Fasting glucose, mg/dL	92.71 (12.54)
Fasting insulin	10.40 [10.10]
HOMA-IR	2.36 [2.50]
HbA1c, %	5.37 (0.40)
Plasma lipids
Triglycerides, mg/dL	138.50 [116.50]
Total cholesterol, mg/dL	207.29 (39.38)
LDL cholesterol, mg/dL	127.27 (34.62)
HDL cholesterol, mg/dL	49.18 (17.08)
Hepatic fat, HU	6.34 (12.21)
Abdominal fat distribution
Visceral adipose tissue, kg	11.11 (2.63)
Subcutaneous adipose tissue, kg	5.46 (2.53)
Total adipose tissue, kg	16.66 (4.04)

Values are mean (SD), or median [interquartile range] for continuous variables, and N (%) for categorical variables.

BMI, body mass index; BP, blood pressure; HDL, high-density lipoprotein; HU, Hounsfield unit; HOMA-IR, homeostasis model assessment index of insulin resistance; LDL, low-density lipoprotein.

Consistent with the entire POUNDS Lost trial, the reported dietary intakes of total energy, macronutrient composites (fat, protein, and carbohydrate), and changes in biomarkers of adherence (urinary nitrogen and respiratory quotient) in our study population confirmed that participants adjusted their diets in the direction of the intervention, although the targets were not fully achieved (22). There was no difference in nutrient intakes or biomarkers of adherence across the tertiles of changes in plasma BCAAs at 6 months or 2 years (all P > 0.05; Table 2).

Table 2.

Nutrient Intake and Biomarkers of Adherence According to the Tertile Categories of Changes in BCAA at 6 Months and 2 Years

		Dietary Intake per Daya				Biomarkers of Adherenceb
		Energy, kcal	Carbohydrate, %	Fat, %	Protein, %	Urinary Nitrogen, g/day	Respiratory Quotient
Baseline	T1	2131 (657)	45 (7)	38 (6)	17 (3)	0.85 (0.04)	12.4 (4.7)
	T2	1867 (535)	45 (8)	36 (6)	19 (4)	0.84 (0.04)	12.3 (3.9)
	T3	2014 (660)	45 (8)	37 (6)	18 (3)	0.84 (0.05)	12.2 (4.1)
	P values	0.24	0.78	0.26	0.16	0.54	0.68
6 months	T1	1653 (561)	55 (10)	28 (9)	19 (4)	0.84 (0.04)	10.4 (3.6)
	T2	1581 (521)	49 (11)	32 (9)	21 (5)	0.84 (0.04)	11.5 (5.1)
	T3	1581 (452)	51 (10)	32 (8)	20 (4)	0.85 (0.04)	11.9 (6.2)
	P values	0.22	0.08	0.07	0.41	0.17	0.14
2 years	T1	1839 (601)	52 (7)	31 (6)	19 (3)	0.83 (0.04)	11.2 (4.7)
	T2	1372 (504)	49 (10)	32 (8)	20 (6)	0.84 (0.04)	12 (4)
	T3	1522 (412)	50 (9)	31 (8)	20 (5)	0.83 (0.04)	10.1 (3.5)
	P values	0.09	0.58	0.84	0.72	0.55	0.8

Data were shown as mean (SD).

The median (25th, 75th) values of changes in total BCAAs in each category were: T1: –73.5 (–102.9, –55.9) µmol/L; T2: –15.8 (–30.2, –6.1) µmol/L; T3: 35.2 (19.3, 56.2) µmol/L, respectively.

^a Sample size for dietary intake at baseline: T1 = 53; T2 = 54; T3 = 53; 6 months: T1 = 50; T2 = 50; T3 = 52; 2 years: T1 = 24; T2 = 25; T3 = 26.

^b Sample size for biomarkers of adherence at baseline: T1 = 53; T2 = 54; T3 = 53; 6 months: T1 = 51; T2 = 53; T3 = 48; 2 years: T1 = 38; T2 = 44; T3 = 38.

P values were calculated by F statistics in general linear models after adjustment for age, sex, ethnicity.

In the study samples, the participant had a mean (SD) weight loss of 7.5 (5.9) kg at 6 months and 4.8 (8.1) kg at 2 years. The plasma concentrations of BCAAs (leucine/isoleucine and valine) were significantly decreased from baseline to 6 months (Mean [SD]: 324.5 [72.7] vs. 301.4 [54.0] µmol/L, P < 0.001), and such reduction was related to weight loss (P < 0.001, P = 0.014, and P < 0.001 for leucine/isoleucine, valine, and total plasma BCAAs, respectively). Incorporating the 3-times repeated measurements by the mixed effect model, we observed significant associations of plasma concentrations of BCAAs (total and individual component) with hepatic density and abdominal fat distribution during the 2-year course (Table 3). Individuals with a decrease in total plasma BCAAs were associated with an increase in hepatic density (P = 0.02), a reduction in SAT (P = 0.02), VAT (P < 0.001), and TAT (P = 0.001), after adjustment for age, sex, ethnicity, diet interventions, baseline BMI, and use of cholesterol-lowering medications (model 2). Similar associations were also found between individual BCAAs, including leucine/isoleucine and valine, and the outcomes including hepatic density, SAT, VAT, and TAT (all P < 0.05; model 2). After further adjustment for concurrent weight loss (weight loss at 2 years), the relationship between BCAAs (total and individual) and hepatic density attenuated to nonsignificant during the 2-year course (all P > 0.05, model 3), whereas the relationships with abdominal fat distribution did not change materially for most of the associations (all P < 0.05, except for SAT and valine, model 3). We also found that the decreases in BCAAs (total and individual component leucine/isoleucine) were associated with significantly increased whole body total % lean mass, and decreased whole body total fat mass, whole body total % fat, and truck fat percentage (P < 0.05 for all) (31). The association between valine and the previous outcomes were in the same direction but did not reach the statistical significance (31).

Table 3.

Association of Decreases in Leucine/Isoleucine, Valine, and BCAA with Changes in Hepatic Fat and Abdominal Fat Distribution during 2 years (N = 184)

Outcomes	Leucine/Isoleucine		Valine		Total BCAAs
	β (SE)	P	β (SE)	P	β (SE)	P
Model 1
Hepatic density, HU	0.88 (0.39)	0.02	0.77 (0.35)	0.03	0.89 (0.37)	0.02
Abdominal fat distribution
SAT, kg	-0.25 (0.08)	0.003	-0.19 (0.08)	0.01	-0.23 (0.08)	0.004
VAT, kg	-0.24 (0.06)	<0.001	-0.12 (0.05)	0.02	-0.19 (0.05)	<0.001
TAT, kg	-0.48 (0.13)	<0.001	-0.35 (0.12)	0.003	-0.43 (0.12)	<0.001
Model 2
Hepatic density, HU	0.89 (0.39)	0.02	0.78 (0.35)	0.03	0.91 (0.37)	0.02
Abdominal fat distribution
SAT, kg	-0.18 (0.08)	0.02	-0.15 (0.07)	0.04	-0.17 (0.07)	0.02
VAT, kg	-0.24 (0.05)	<0.001	-0.12 (0.05)	0.01	-0.19 (0.05)	<0.001
TAT, kg	-0.38 (0.11)	<0.001	-0.28 (0.10)	0.005	-0.35 (0.10)	0.001
Model 3
Hepatic density, HU	0.67 (0.40)	0.09	0.63 (0.36)	0.08	0.70 (0.38)	0.07
Abdominal fat distribution
SAT, kg	-0.19 (0.08)	0.03	-0.11 (0.08)	0.14	-0.15 (0.08)	0.05
VAT, kg	-0.22 (0.06)	<0.001	-0.10 (0.05)	0.03	-0.16 (0.05)	0.003
TAT, kg	-0.37 (0.12)	0.001	-0.24 (0.11)	0.02	-0.32 (0.11)	0.005

β (SE) represents changes in the outcome traits per SD decrease in leucine/isoleucine, valine, or BCAA.

Model 1: mixed-effect model adjusted for age, sex, ethnicity, and diet intervention group.

Model 2: model 1+ baseline body mass index, medication for controlling cholesterol.

Model 3: model 2+ weight loss at 2 years (31 missingness on 2-year weight-loss).

BCAA, branched-chain amino acid; HU, Hounsfield unit; SAT, subcutaneous adipose tissue; TAT: total adipose tissue, VAT: visceral adipose tissue.

When we examined the association between plasma BCAAs (total and individual) and the markers of glucose and lipids metabolism (Table 4), we found that decreases in total plasma BCAAs were significantly associated with decreases in fasting glucose (P = 0.02), fasting insulin (P < 0.001), HOMA-IR (P < 0.001), and triglycerides (P < 0.001) after adjustment for age, sex, ethnicity, diet intervention, baseline BMI, and use of cholesterol-lowering medications (model 2). Further adjustment for 6-month weight loss did not materially change those associations, except for fasting glucose (P = 0.06; model 3). Similar associations were also observed between individual components (leucine/isoleucine and valine) and fasting insulin, HOMA-IR, and triglycerides.

Table 4.

Association of Decreases in Leucine/Isoleucine, Valine, and BCAA with Changes in Markers of Glucose and Lipids Metabolism during 2 years (N = 184)

Outcome	Leucine/Isoleucine		Valine		Total BCAAs
	β (SE)	P	β (SE)	P	β (SE)	P
Model 1
Glucose metabolism
Fasting glucose	-1.19 (0.47)	0.01	-0.78 (0.41)	0.06	-1.05 (0.44)	0.02
Insulina	-0.08 (0.02)	<0.001	-0.07 (0.02)	<0.001	-0.08 (0.02)	<0.001
HOMA-IRa	-0.10 (0.02)	<0.001	-0.08 (0.02)	<0.001	-0.10 (0.02)	<0.001
HbA1c	-0.02 (0.02)	0.16	-0.02 (0.01)	0.17	-0.02 (0.02)	0.14
Lipids metabolism
Total cholesterol	-1.67 (1.58)	0.29	-1.46 (1.39)	0.30	-1.69 (1.50)	0.26
HDL-C	0.74 (0.39)	0.06	0.23 (0.34)	0.49	0.50 (0.37)	0.17
LDL-C	-0.89 (1.38)	0.52	-0.70 (1.21)	0.57	-0.86 (1.31)	0.51
Triglyceridesa	-0.07 (0.02)	<0.001	-0.05 (0.02)	0.003	-0.07 (0.02)	<0.001
Model 2
Glucose metabolism
Fasting glucose	-1.18 (0.47)	0.01	-0.76 (0.41)	0.06	-1.03 (0.44)	0.02
Insulina	-0.08 (0.02)	<0.001	-0.07 (0.02)	<0.001	-0.08 (0.02)	<0.001
HOMA-IRa	-0.10 (0.02)	<0.001	-0.08 (0.02)	<0.001	-0.10 (0.02)	<0.001
HbA1c	-0.02 (0.02)	0.16	-0.02 (0.01)	0.18	-0.02 (0.02)	0.14
Lipids metabolism
Total cholesterol	-1.62 (1.55)	0.30	-1.55 (1.37)	0.26	-1.71 (1.47)	0.25
HDL-C	0.79 (0.38)	0.04	0.23 (0.33)	0.50	0.52 (0.36)	0.15
LDL-C	-0.80 (1.36)	0.56	-0.68 (1.20)	0.57	-0.80 (1.30)	0.54
Triglyceridesa	-0.07 (0.02)	<0.001	-0.05 (0.02)	0.004	-0.07 (0.02)	<0.001
Model 3
Glucose metabolism
Fasting glucose	-0.97 (0.50)	0.05	-0.66 (0.42)	0.12	-0.86 (0.47)	0.06
Insulina	-0.07 (0.02)	0.002	-0.06 (0.02)	0.001	-0.07 (0.02)	<0.001
HOMA-IRa	-0.08 (0.03)	0.001	-0.07 (0.02)	0.001	-0.08 (0.02)	<0.001
HbA1c	-0.02 (0.02)	0.20	-0.01 (0.02)	0.37	-0.02 (0.02)	0.25
Lipids metabolism
Total cholesterol	-0.08 (0.03)	0.78	-0.45 (1.44)	0.76	-0.49 (1.57)	0.76
HDL-C	0.99 (0.40)	0.01	0.36 (0.34)	0.30	0.69 (0.38)	0.07
LDL-C	0.06 (1.47)	0.97	0.22 (1.27)	0.86	0.16 (1.38)	0.91
Triglyceridesa	-0.07 (0.02)	<0.001	-0.07 (0.02)	0.009	-0.06 (0.02)	0.002

β (SE) represents changes in the outcome traits per SD decrease in leucine/isoleucine, valine, or BCAA.

Model 1: mixed-effect model adjusted for age, sex, ethnicity, and diet intervention group.

Model 2: model 1+ baseline body mass index, medication for controlling cholesterol.

Model 3: model 2 + weight loss at 2 years (31 missingness on 2-year weight-loss).

BCAA, branched-chain amino acid; HDL, high-density lipoprotein; HOMA-IR, homeostasis model assessment of insulin resistance; LDL, low-density lipoprotein.

^a Variables were log-transformed before analyses.

In the sensitivity analyses with multiple imputations for missing data, we found that decreases in plasma leucine/isoleucine, valine, and total BCAAs were significantly associated with increases in hepatic density (i.e., less hepatic fat) and reductions in VAT after adjustment for age, sex, ethnicity, diet intervention group, baseline BMI, medications for controlling cholesterol (model 2). However, no associations were found between BCAAs (total and individual component) and SAT and TAT (model 2) (31). When further adjusted for concurrent weight loss (model 3), the associations were largely the same as model 2. Such attenuated relationships might be explained by the uncertainty of multiple imputations. Consistent with the analyses using mixed-effect models, we found that decreases in leucine/isoleucine, valine, and total BCAAs were associated with decreases in fasting insulin, HOMA-IR, and triglycerides after adjustment for age, sex, ethnicity, diet intervention group, baseline BMI, medications for controlling cholesterol, and concurrent weight loss (31).

We also tested the relationships between 6-month changes in BCAAs (total and individual component) and concurrent changes in hepatic density, fat distribution, and markers of glucose metabolism (31). Consistent with our analyses on the 2-year changes, we found that 6-month decreases in total plasma BCAAs and leucine/isoleucine were significantly associated with decreases in VAT, TAT, fasting glucose, fasting insulin, and HOMA-IR at 6 months (all P < 0.05).

Furthermore, we observed that dietary protein (high vs. average) significantly modified the association between 6-month changes in total plasma BCAAs and hepatic density at 6 months after adjustment for age, sex, ethnicity, baseline BMI, baseline value of hepatic density, and baseline levels of total BCAAs (Fig. 1A, P_interaction = 0.01). The baseline characteristics according to tertile categories are shown in (31). There was no difference across the tertiles with regard to the markers of lipids or glucose metabolism after adjustment for age, sex, and ethnicity. Among those with an average-protein diet, the median (25th, 75th) levels of total plasma BCAAs was 309.2 (268.1, 355.7) µmol/L at baseline and had a change of –12.9 (–55.9, 25.6) µmol/L at 6 months; among those with a high-protein diet, total plasma BCAAs was 323.3 (287.3, 361.7) at baseline, and had a change of –16.2 (–51.2, 13.6) at 6 months. Among participants with a high-protein diet, a larger decrease in total plasma BCAAs (tertile 1) was associated with a higher hepatic density at 6 months, whereas no such relationship was found among those who consumed an average protein diet. Such interaction was attenuated to be borderline significant at 2 years (P_interaction = 0.06). A similar interaction pattern was also found between changes in plasma valine and hepatic density at 6 months (Fig. 1 B, P_interaction = 0.001). In the high-protein diet group, those with a larger decrease in plasma valine (tertile 1) were associated with a higher hepatic density, whereas no association was found in the average protein diet group. At 2 years, the interaction between plasma valine and diet protein on changes in hepatic density remained significant (P_interaction = 0.04). We did not find significant dietary protein modification between 6-month changes in leucine/isoleucine and hepatic density. We did not find significant interactions between BCAAs and abdominal fat distribution (SAT, VAT, TAT). The joint analysis combining tertiles of BCAA change and dietary protein intake showed similar patterns (31).

Figure 1.

Hepatic density at 6-month according to tertile (T) categories of changes in total plasma BCAAs and valine in response to the high-/average-protein diet group. Data are means ± SEs values after adjustment for age, sex, ethnicity, baseline BMI, medication for controlling cholesterol, baseline values of valine or total BCAAs when appropriate, and baseline values of hepatic density. For ΔBCAAs from baseline to 6 months, median (25th, 75th) values were T1: –73.5 (–102.9, –55.9) µmol/L; T2: –15.8 (–30.2, –6.1) µmol/L; T3: 35.2 (19.3, 56.2) µmol/L, respectively. For ΔValine from baseline to 6 months, median (25th, 75th) values were T1: –42.4 (–61.0, –32.6) µmol/L; T2: –8.0 (–16.1, 0.2) µmol/L; T3: 24.0 (13.1, 31.3) µmol/L, respectively. For the average-protein group: T1, n = 30; T2, n = 22; T3, n = 30. For the high-protein group: T1, n = 23; T2, n = 32; T3, n = 23. BCAA, branched-chain amino acid; BMI, body mass index.

Discussion

In this study, we found that weight-loss diet-induced decreases in leucine/isoleucine, valine, and total BCAAs were associated with reductions in hepatic and abdominal fat during the 2-year intervention course. We also found that dietary protein significantly modified the association of changes in valine and total BCAAs with hepatic fatness. In addition, we found that decreases in plasma BCAAs (total and individual component) were related to reductions in fasting insulin, HOMA-IR, and triglycerides, independent of weight loss.

To the best of our knowledge, our study for the first time showing that decreased levels of plasma BCAAs by weight-loss diets were associated with improvement in hepatic and abdominal fat distribution in a prospective study. In epidemiological studies, high circulating levels of BCAAs have been consistently associated with obesity and related metabolic diseases such as type 2 diabetes and cardiovascular disease. Recently, emerging data from observational studies also show that circulating BCAAs are related to the ectopic fat, such as in the liver or abdominal regions (10-12, 32-35). For example, a previous cross-sectional study reported significantly higher concentrations of plasma BCAAs among individuals with nonalcoholic fatty liver disease (11). However, most of the previous studies were cross-sectional or small in sample sizes. Our analyses incorporating 3-time repeated measurements from a prospective study suggests that, among patients who are overweight or obese, and who have high circulating BCAAs, low-calorie weight-loss diets may improve BCAA profile, and subsequently reduce hepatic and abdominal fat. Results from the current analyses showed that deceases in BCAAs were associated with improvement in ectopic fat (including abdominal fat and hepatic fat), most of which were independent of concurrent weight loss, except for hepatic fat. The attenuated relationship between BCAAs and hepatic fat after further adjustment for concurrent weight loss suggests that it may associated with hepatic fat differently, compared with abdominal fat distribution. More studies are warranted to further explore the relationship between BCAAs and hepatic fat.

BCAAs have been linked to hepatic and visceral fat accumulation (9, 35-37). A widely accepted pathway is that an increased level of plasma BCAAs, together with insulin, may activate the mammalian target of rapamycin complex 1 that leads to the serine phosphorylation of insulin receptor substrate-1 and insulin receptor substrate-2, resulting in insulin resistance (1, 38, 39). Insulin resistance has been consistently found to be associated with ectopic fat accumulations, especially hepatic and abdominal fat (40-45). Intriguingly, in our study, we also found that decreases in plasma BCAAs were significantly associated with improved insulin and HOMA-IR, in line with the BCAAs induced bidirectional effects between insulin resistance and hepatic/abdominal fat distribution. Of note, we found that the relationship between BCAAs and hepatic fat was no longer statistically significant after controlling for the concurrent weight loss, whereas the reductions in abdominal fat and the biomarkers (insulin, HOMA-IR, and triglycerides) were independent of weight loss. The plasma level of BCAAs in the current study was measured in fasting status, not the postprandial level. Thus, the plasma BCAAs in our sample represents the combination of food intake and metabolisms in the participant’s body. Taken together, our observations of the significant associations of decreases in plasma BCAAs with improvement in hepatic and abdominal fat, as well as the decreases in insulin and HOMA-IR, may lend support to potential interventions to lower hepatic and abdominal fat through reduction of plasma BCAAs among overweight and obese individuals, even though we acknowledge the observational nature of our study. Further investigations to examine the causality are warranted.

Interestingly, we found a significant interaction between dietary protein and plasma BCAAs in relation to hepatic fat. Among participants with a high-protein diet, a greater decrease in BCAAs (total and individual valine) was associated with a higher value of hepatic fat. Our result showed that initial changes in BCAAs could interact with dietary protein and subsequently result in different changes in hepatic fat. Zhang et al conducted a comprehensive study in mice and showed that BCAAs could impair the liver by abnormal lipolysis and hyperlipidemia, which led to fat accumulation in the liver (32). No previous study has specifically investigated the interaction between BCAAs and dietary protein. Currently, the underlying mechanism of such interaction between BCAA and dietary protein remains unclear. Because BCAAs are essential amino acids abundant in a high-protein diet, we postulate that the potential beneficial effects of the reduction of BCAAs on hepatic fat might be more efficient when the amino acid load was high (46). Although the entire POUNDS Lost participants showed a significant difference in BCAAs concentrations between high-protein and average-protein group, there were suggestive but nonsignificant higher BCAAs among those with high-protein diet in the current analysis of the subsamples of the trial (19). We assumed that other mechanisms were likely to be involved in accounting for the modification effect of dietary protein intake on the association between change in BCAA and ectopic fat. Further functional studies are needed to explore the interactions between plasma BCAAs and dietary protein on hepatic fat, as well as the mechanisms.

Our study has several strengths. First, in our study, hepatic fat and abdominal fat distribution were assessed at multiple time points by CT scans, which have been validated against magnetic resonance spectroscopy (47). Second, we used 3-time repeated measurements, which capture the dynamic changes of both BCAAs and ectopic fat over time. Compared with other methods, such longitudinal analyses on dynamic changes may produce more robust, consistent, and biological plausible results (48). Analyzing dynamic changes in risk factors may also help translate findings into prevention and treatment strategies. If changes in metabolites are associated with the outcomes, these metabolites might then be used as potential targets in interventions in disease prevention and treatment. Our consistent observations across mixed-effect models, change analysis, and sensitivity analysis strengthened our conclusions. Third, the POUNDS Lost trial is one of the largest and longest diet intervention trials. However, several limitations of this study warrant consideration. First, we could not determine whether plasma BCAAs were the causal factor in relation to hepatic fat and abdominal fat distribution, or just the biomarkers of the underlying metabolic dysfunction. Second, we tested the associations between plasma BCAAs with multiple outcomes, raising the issue of multiple comparisons. However, the present study was based on a priori hypothesis, and those outcomes were not mutually independent. Thus, correcting the multiple comparisons would increase the possibility of a false negative (type 2) error. Third, measurements on protein synthesis/catabolism or rate of catabolism was not available in POUNDS Lost. Fourth, our result might not be generalizable to other populations as most of our study population are white and well educated. Further investigations are warranted to explore the potential mechanisms underlying the observations and validate our findings in other populations.

In conclusion, in this 2-year diet intervention trial, we found that weight-loss diet-induced decreases in plasma BCAAs were associated with reductions in hepatic and abdominal fat distribution; and dietary protein intake modified the associations of changes in plasma BCAAs with improvement in hepatic fat. Our results suggest that changes in BCAAs in response to diet interventions may affect changes in ectopic fat distribution among overweight and obese people. Notably, hepatic/visceral fat has been related to various cardiometabolic complications of obesity. Our findings may provide important information to the development of novel intervention strategies targeting BCAAs in the treatment of obesity and mitigation of the cardiometabolic complications.

Glossary

Abbreviations

BCAA

branched-chain amino acid

BMI

body mass index

CT

computed tomography

HOMA-IR

homeostasis model assessment of insulin resistance

PBRC

Pennington Biomedical Research Center

SAT

subcutaneous adipose tissue

TAT

total adipose tissue

VAT

visceral adipose tissue.

Additional Information

Disclosure Summary: The authors have nothing to disclose.

Data Availability

Restrictions apply to the availability of data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.