Skip to main content
International Journal of Epidemiology logoLink to International Journal of Epidemiology
. 2016 Jul 13;45(5):1482–1492. doi: 10.1093/ije/dyw143

Cumulative consumption of branched-chain amino acids and incidence of type 2 diabetes

Yan Zheng 1, Yanping Li 1, Qibin Qi 2, Adela Hruby 1, JoAnn E Manson 3,4,5, Walter C Willett 1,3,4, Brian M Wolpin 6, Frank B Hu 1,3,4, Lu Qi 1,7,*
PMCID: PMC5100612  PMID: 27413102

Abstract

Background: Plasma branched-chain amino acids (BCAAs, including leucine, isoleucine and valine) were recently related to risk of type 2 diabetes (T2D). Dietary intake is the only source of BCAAs; however, little is known about whether habitual dietary intake of BCAAs affects risk of T2D.

Methods: We assessed associations between cumulative consumption of BCAAs and risk of T2D among participants from three prospective cohorts: the Nurses’ Health Study (NHS; followed from 1980 to 2012); NHS II (followed from 1991 to 2011); and the Health Professionals Follow-up Study (HPFS; followed from 1986 to 2010).

Results: We documented 16 097 incident T2D events during up to 32 years of follow-up. After adjustment for demographics and traditional risk factors, higher total BCAA intake was associated with an increased risk of T2D in men and women. In the meta-analysis of all cohorts, comparing participants in the highest quintile with those in the lowest quintile of intake, hazard ratios (95%confidence intervals) were for leucine 1.13 (1.07-1.19), for isoleucine 1.13 (1.07-1.19) and for valine 1.11 (1.05-1.17) (all P for trend < 0.001). In a healthy subsample, higher dietary BCAAs were significantly associated with higher plasma levels of these amino acids (P for trend = 0.01).

Conclusions: Our data suggest that high consumption of BCAAs is associated with an increased risk of T2D.

Keywords: Diet, branched-chain amino acids, type 2 diabetes, cohort study


Key Message

  • Our data indicate that higher consumption of dietary branched-chain amino acids, the blood levels of which have been related to risk of type 2 diabetes, is also associated with higher risk of type 2 diabetes.

Introduction

Type 2 diabetes (T2D), which is strongly linked to chronic diseases including hypertension, cardiovascular disease, and certain cancers,1,2 has reached epidemic proportions in the United States and globally.3 There is compelling evidence that healthy dietary habits and lifestyle choices can prevent T2D.4,5

In a recent prospective study, Wang et al. reported that blood levels of branched-chain amino acids (BCAAs)—a group of essential amino acids (that are not produced in the body; diet is the only source) comprising leucine, isoleucine and valine —predicted risk of T2D in predominantly Caucasian cohorts;6 and similar associations were observed in an Asian population.7 High blood BCAA levels were also related to obesity, elevated fasting and postprandial glucose, and insulin resistance in non-diabetic children and adults.8– 12 In a recent analysis, we found that a genetic marker related to blood BCAA levels was associated with changes in insulin resistance in response to a dietary intervention among overweight and obese adults.13 However, several feeding studies have shown conflicting data on the effects of BCAA intake on glucose metabolism and insulin resistance.14,15 To our knowledge only one study, conducted in a Japanese population, has investigated habitual BCAA intake and long-term risk of T2D in a prospective cohort.16 In that study, total BCAA, leucine and valine intakes were inversely associated with T2D risk in women, and no associations were found in men.

In the present study, we prospectively analysed the associations between long-term intakes of BCAAs and the incidence of T2D in three large cohorts of US women and men—the Nurses’ Health Study (NHS), the Nurses’ Health Study II (NHS II) and the Health Professionals Follow-up Study (HPFS).

Methods

Study population

The NHS cohort began in 1976 when 121 700 female nurses aged 30–55 years living in 11 US states responded to a questionnaire regarding medical, lifestyle and other health-related information. The NHS II began in 1989 by enrolling 116 430 female registered nurses aged 25–42 years by completing an initial questionnaire.17 The HPFS similarly enrolled 51 529 males aged 40–75 years in 1986. Detailed descriptions of these cohorts have been presented elsewhere.18,19 In all cohorts, questionnaires were administered at baseline as well as biennially after baseline, to collect and update information on lifestyle practices and occurrence of chronic diseases. The follow-up rates of the participants in these cohorts are all greater than 90%. In the current dietary analysis, we excluded men and women who had diagnoses of cardiovascular disease or cancer at baseline (1980 for NHS, 1991 for NHS II and 1986 for HPFS, when dietary information was first collected), and who reported unusual total energy intake (i.e. < 800 or > 4200 kcal/day for men and < 500 or > 3500 kcal/day for women). After exclusions, data from 39 385 men and 152 755 women were available for analysis. This study was approved by the Harvard Institutional Review Board, and all participants provided written informed consent.

Assessment of dietary intake

Dietary intake was updated every 4 years with a semi-quantitative food frequency questionnaire (FFQ) beginning in 1980 in NHS, 1991 in NHS II and 1986 in HPFS. In NHS, the FFQ food list began with 61 items, increased to 116 items in 1984 and to 131 items in later versions.20 In NHS II and HPFS, an FFQ with 131 items was used for all study cycles.20 Nutrient contents, including BCAAs, of foods and beverages were derived primarily from US Department of Agriculture sources and supplemented with data from food manufacturers and published research. The reproducibility and validity of the measurement of protein intake by the FFQ against dietary records were reported previously.21,22 The correlation coefficients between FFQs and diet records for protein intake were in the order of 0.4.22 All nutrient data were adjusted for total energy intake using the residual method.23 In addition to each individual BCAA, total BCAAs was defined as the sum of energy-adjusted dietary leucine, isoleucine and valine.

Covariate assessment

In the baseline questionnaire and every biennial questionnaire thereafter, responding participants provided information about age, anthropometrics, smoking, menopausal status and use of postmenopausal hormone therapy (women only), aspirin use, family history of diabetes, history of hypertension and hypercholesterolaemia and other lifestyle habits. Height was ascertained on the 1976 enrolment questionnaire in NHS, the 1989 enrolment questionnaire in NHS II and the 1986 enrolment questionnaire in HPFS. Physical activity was expressed as metabolic equivalent tasks (MET) per hour, which were calculated with data from a self-report questionnaire focused on types and durations of activities over the previous year. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in metres (kg/m2).

Outcome ascertainment

The study outcome was incident T2D that occurred between the return of the baseline FFQ and 31 January 2010 (HPFS), 30 June 2011 (NHS II) or 30 June 2012 (NHS). In these cohorts, men and women who reported a diagnosis of T2D in the biennial follow-up questionnaires were sent a supplementary questionnaire to confirm the diagnosis. In this supplementary questionnaire, information on symptoms, diagnostic tests and treatment was collected. For cases before 1998, we used the National Diabetes Data Group criteria to define diabetes.24 We used the American Diabetes Association diagnostic criteria for diabetes diagnosis from 1998 onward.25 The validity of this method has been confirmed.26,27 In NHS, of a random sample of participants reporting T2D in the supplementary questionnaire, 98% of diagnoses were confirmed after medical record review by an endocrinologist blinded to the supplementary questionnaire information.26 In another validation study conducted in HPFS, 97% of self-reported T2D cases were confirmed by medical record review.27

Plasma BCAAs measurement

Blood samples were collected in EDTA tubes from 18 225 men in HPFS in 1993–95 and in heparin tubes from 32 826 women in NHS in 1989–90. Blood samples were collected by participants, mailed overnight on cold packs, then spun to collect and store plasma (delayed processing). Plasma BCAAs were measured in a subsample of 157 men and 240 women who were free of chronic diseases when the blood sample was drawn, as peak areas by a targeted liquid chromatography-mass spectrometry (LC-MS) metabolomics platform directed by Dr. Clary Clish at the Broad Institute of the Massachusetts Institute of Technology and Harvard University (Cambridge, MA). Total plasma BCAA levels were calculated as the summation of the peak areas of individual plasma BCAAs (leucine, isoleucine and valine). The total plasma BCAA levels were log10 transformed before analysis. Spearman rank correlation coefficients between heparin- and EDTA-collected samples were 0.88 for leucine, 0.85 for isoleucine and 0.95 for valine.

Statistical analysis

For dietary measures, we used the cumulative average of BCAA intakes from baseline to the censoring events to best represent long-term diet and minimize within-person variation.28 We generated quintile categories of energy-adjusted intakes of leucine, isoleucine, valine and total BCAAs. We calculated the Spearman rank correlation coefficients between dietary BCAAs and dietary intakes of animal protein and vegetable protein, adjusted for age and BMI.

We calculated each individual’s person-time from the return of the baseline questionnaire to the date of diagnosis of T2D, date of death, date of loss to follow-up or the cut-off date (31 January 2010 in HPFS; 30 June 2011 in NHS II; and 30 June 2012 in NHS), whichever occurred first. We used Cox proportional hazards models to calculate hazard ratios (HR) and 95% confidence intervals (CI) for the associations between intakes of BCAAs (in quintiles) and risk of incident T2D. Multiplicative interaction terms for age and quintiles of dietary BCAAs did not indicate violation of the proportional hazards assumption (P for interaction > 0.05). We generated several models beyond the age-adjusted model: in model 1 we additionally adjusted for the updated covariates at each 2-year cycle including: smoking (never, past, current with cigarette use of 1–14/day, 15–24/day, > 25/day, missing); alcohol intake (g/day: 0, 0.1–4.9; 5.0–14.9; ≥ 15); physical activity (MET h/week: < 3; 3–8.9; 9–17.9; 18–26.9; ≥ 27; missing), total energy intake (kcal/day, in quintiles), family history of diabetes (yes or no), history of hypertension and high blood cholesterol (yes or no) and a diabetes dietary score (in quintiles) based on the sum of quintiles of dietary glycaemic load, intake of trans fat, dietary fibre and the ratio of polyunsaturated to saturated fat. Glycaemic load was calculated by multiplying the carbohydrate content of each food by its glycaemic index,29 multiplying this value by the frequency of consumption and summing these values for all foods. In women, we also adjusted for menopausal status and postmenopausal hormone use (premenopausal and postmenopausal, with never, past, or current hormone use). In model 2, we further adjusted for updated BMI (kg/m2, continuous) and a quadratic term of BMI (BMI2, continuous). In sensitivity analyses, we additionally adjusted for total meat or protein intake. We tested for potential effect modification by risk factors (age, BMI, physical activity and drinking status) by including cross-product terms with the exposure variables in our fully adjusted model. Tests for trend were conducted by assigning the median value to each quintile category and modelling this value as a continuous variable.

For the analyses of diet-plasma BCAA associations, we used the average of BCAA intakes from the two most recent FFQs until blood samples were drawn. Spearman rank correlation coefficients were calculated to measure the correlation between dietary intake and plasma levels of amino acids. We also estimated mean levels of total plasma BCAAs by quintiles of total BCAA intake.

To summarize the associations across the three cohorts, we conducted a fixed effect meta-analysis because no heterogeneity by Cochran’s Q test was apparent between cohorts (P-values for heterogeneity > 0.1). All P-values were 2-sided and an alpha level of < 0.05 was considered statistically significant. Data were analysed using the SAS package, version 9.3 (SAS Institute Inc., Cary, NC).

Results

At baseline, women were on average [mean (standard deviation; SD)] 46 (7.2) years old in NHS and 36 (4.7) years in NHS II, and men were 53 (9.5) years old. Women had a lower mean BMI than men, at 24.2 (4.3) kg/m2 in NHS, 24.5 (5.2) kg/m2 in NHS II and 25.4 (3.2) kg/m2 in HPFS, respectively. During up to 32 years of follow-up, 12 807 women reported a new diagnosis of T2D [11.5% of the NHS (N = 7584) and 6.0% of the NHS II baseline population (N = 5223)]. During 24 years of follow-up, 3290 men reported a new diagnosis of T2D (8.4% of the baseline population). The main food sources of BCAAs were meat (chicken, beef and pork: ∼ 37%), fish (∼ 8%) and milk (∼ 12%) in our study population. The pairwise unadjusted Spearman rank correlation coefficients between energy-adjusted dietary intakes of leucine, isoleucine and valine were all 0.99 (all P < 0.0001). The Spearman rank correlation coefficient of overall BCAA with animal protein is 0.74 (NHS) to 0.94 (HPFS and NHS II), and that with vegetable protein is −0.004 (NHS) to −0.04 (HPFS and NHS II) (all P-values < 0.001).

Table 1 presents selected baseline characteristics of participants from NHS, NHS II and HPFS according to the quintiles of total dietary BCAAs (sum of energy-adjusted dietary leucine, isoleucine and valine). Total BCAA intake was positively associated with BMI, protein intake, diabetes dietary score and family history of type 2 diabetes and inversely associated with current smoking and alcohol intake, consistently in three cohorts. Of note, the positive correlation between BCAA intakes and family history of diabetes might suggest shared dietary behaviours, which led to an increased risk of diabetes, within families.

Table 1.

Baseline age-adjusted characteristics of participants in the NHS (1980), NHS II (1991) and HPFS (1986) according to quintiles of intake of total BCAAs (sum of energy-adjusted leucine, isoleucine, and valine)a

Characteristics NHS (N = 66125)
NHS II (N = 86630)
HPFS (N = 39017)
Q1 Q3 Q5 Q1 Q3 Q5 Q1 Q3 Q5
N 13247 13270 13222 17326 17352 17308 7873 7882 7887
Age, years 46.1 45.8 46.7 36.0 36.0 36.2 52.8 52.6 53.5
BMI, kg/m2 23.5 24.0 25.2 23.8 24.4 25.4 25.0 25.4 25.9
Physical activity, METs/week 12.7 14.2 15.9 20.5 20.1 22.8 21.3 21.0 21.9
Current smoking, % 33.7 27.0 24.7 14.9 11.5 10.8 11.9 9.4 8.3
Alcohol intake, g/day 8.3 6.5 4.9 4.0 3.2 2.2 17.0 11.1 7.2
Postmenopausal, % 33.0 32.7 32.7 3.2 3.0 3.3
Current PMH users, % 20.7 21.1 23.0 4.3 4.3 4.7
History of hypertension, % 13.9 13.9 16.5 3.1 3.0 3.6 18.3 18.2 20.2
History of hypercholesterolaemia, % 4.4 4.5 5.6 9.0 8.7 10.1 8.9 10.0 11.4
Family history of T2D, % 26.7 28.0 31.3 15.0 15.5 18.1 17.1 18.5 19.5
Energy intake, kcal/day 1534 1584 1539 1781 1816 1744 1976 2031 1963
Diabetes dietary scoreb 11.7 12.0 12.3 10.9 11.9 13.2 11.4 11.9 12.9
Glycaemic loadc 93.8 84.7 78.5 139.4 120.1 106.3 137.1 124.8 110.8
Trans fat intake, g/day 4.1 4.1 3.7 3.5 3.4 2.7 3.0 2.9 2.5
Dietary fibre intake, g/day 2.5 2.5 2.4 5.6 5.7 5.4 5.7 6.0 5.7
Ratio of polyunsaturated to saturated fat intake 0.4 0.4 0.3 0.6 0.5 0.5 0.6 0.6 0.6
Protein intake, g/day 66.8 75.3 85.8 65.9 86.0 107.9 70.8 91.1 115.0

PMH, postmenopausal hormone use; Q, quintile.

aValues are presented as means or percentages. All characteristics except for age are adjusted for age.

bSum of quintiles of dietary glycaemic load, intake of trans fat, dietary fibre and the ratio of polyunsaturated to saturated fat.

cGlycaemic load was calculated by multiplying the carbohydrate content of each food by its glycaemic index, multiplying this value by the frequency of consumption and summing these values for all foods. The measure has no units.

The age- and multivariable-adjusted associations of intakes of leucine, isoleucine and valine with incident T2D in the NHS, NHS II and HPFS are shown in Table 2. Comparing the highest quintiles of intake with the lowest quintiles in model 2, the HRs (95% CI) of T2D for leucine, isoleucine and valine were: 1.12 (1.04-1.21), 1.13 (1.05-1.22) and 1.11 (1.03-1.20) in NHS, respectively; 1.10 (1.01-1.21), 1.09 (1.00-1.20) and 1.07 (0.98-1.17) in NHS II, respectively; and 1.19 (1.06-1.33), 1.17 (1.04-1.31) and 1.15 (1.03-1.28) in HPFS, respectively. In the meta-analysis of three samples in fully adjusted models including BMI (model 2), the HRs (95% CI) of T2D comparing the highest vs the lowest intake of leucine, isoleucine and valine were 1.13 (1.07-1.19), 1.13 (1.07-1.19) and 1.11 (1.05-1.07), respectively (Figure 1A–C). All tests for linear trend across increasing quintiles were highly significant in the meta-analysis and in NHS, NHS II and HPFS separately (all P < 0.005).

Table 2.

Hazard ratios (95% CI) for risk of type 2 diabetes according to quintiles of energy-adjusted intakes of individual BCAAs in the NHS, NHS II and HPFS

  Individual BCAA intake
P for trendb
  Q1 Q2 Q3 Q4 Q5
Leucine
NHS
Median (g/day) 4.5 5.1 5.5 6.0 6.7
No. of cases/no. of person-years 1203/349533 1261/350823 1468/349996 1651/350137 2001/349622
Age-adjusted 1.00 1.04 (0.96-1.13) 1.21 (1.12-1.31) 1.36 (1.26-1.46) 1.64 (1.52-1.76) < 0.001
Model 1a 1.00 1.10 (1.01-1.19) 1.26 (1.16-1.36) 1.42 (1.31-1.53) 1.68 (1.56-1.81) < 0.001
Model 2b 1.00 0.99 (0.91-1.07) 1.04 (0.97-1.13) 1.07 (0.99-1.16) 1.12 (1.04-1.21) < 0.001
NHS II
Median (g/day) 5.3 6.1 6.6 7.1 7.9
No. of cases/no. of person-years 859/320700 852/322187 966/321685 1147/321043 1399/319592
Age-adjusted 1.00 1.00 (0.91-1.11) 1.14 (1.04-1.26) 1.36 (1.24-1.49) 1.64 (1.50-1.78) < 0.001
Model 1a 1.00 1.04 (0.95-1.15) 1.17 (1.06-1.28) 1.39 (1.27-1.53) 1.61 (1.47-1.76) < 0.001
Model 2b 1.00 0.92 (0.83-1.01) 0.95 (0.86-1.05) 1.07 (0.97-1.17) 1.10 (1.01-1.21)  0.001
HPFS
Median (g/day) 5.6 6.3 6.9 7.4 8.3
No. of cases/no. of person-years 530/154422 584/154897 602/154943 708/154641 866/153662
Age-adjusted 1.00 1.10 (0.98-1.24) 1.13 (1.01-1.27) 1.33 (1.19-1.49) 1.62 (1.45-1.80) < 0.001
Model 1a 1.00 1.10 (0.98-1.24) 1.12 (0.99-1.26) 1.29 (1.15-1.44) 1.59 (1.42-1.77) < 0.001
Model 2b 1.00 1.01 (0.90-1.14) 0.97 (0.86-1.09) 1.07 (0.95-1.20) 1.19 (1.06-1.33) < 0.001
Isoleucine
NHS
Median (g/day) 2.7 3.1 3.3 3.6 4.0
No. of cases/no. of person-years 1204/350349 1278/350491 1483/348948 1622/350635 1997/349688
Age-adjusted 1.00 1.06 (0.98-1.14) 1.23 (1.14-1.32) 1.33 (1.24-1.44) 1.63 (1.52-1.75) < 0.001
Model 1a 1.00 1.12 (1.03-1.21) 1.28 (1.19-1.39) 1.42 (1.31-1.53) 1.71 (1.58-1.84) < 0.001
Model 2b 1.00 1.01 (0.94-1.10) 1.06 (0.98-1.15) 1.08 (1.00-1.17) 1.13 (1.05-1.22) < 0.001
NHS II
Median (g/day) 3.1 3.6 3.9 4.3 4.8
No. of cases/no. of person-years 873/321491 869/321912 958/320602 1114/322124 1409/319078
Age-adjusted 1.00 1.01 (0.92-1.11) 1.12 (1.02-1.23) 1.30 (1.19-1.42) 1.63 (1.49-1.77) < 0.001
Model 1a 1.00 1.04 (0.95-1.15) 1.15 (1.04-1.26) 1.35 (1.23-1.48) 1.61 (1.47-1.76) < 0.001
Model 2b 1.00 0.92 (0.84-1.02) 0.94 (0.86-1.04) 1.03 (0.94-1.13) 1.09 (1.00-1.20)  0.002
HPFS
Median (g/day) 3.3 3.8 4.1 4.5 5.0
No. of cases/no. of person-years 550/154685 569/154931 599/154422 705/155157 867/153371
Age-adjusted 1.00 1.04 (0.92-1.17) 1.09 (0.97-1.22) 1.27 (1.14-1.42) 1.56 (1.40-1.74) < 0.001
Model 1a 1.00 1.03 (0.92-1.16) 1.08 (0.96-1.21) 1.24 (1.11-1.39) 1.55 (1.39-1.73) < 0.001
Model 2b 1.00 0.95 (0.84-1.07) 0.94 (0.84-1.06) 1.03 (0.92-1.16) 1.17 (1.04-1.31) < 0.001
Valine
NHS
Median (g/day) 3.0 3.4 3.7 4.0 4.4
No. of cases/no. of person-years 1230/350458 1270/349760 1471/349485 1653/351177 1960/349231
Age-adjusted 1.00 1.03 (0.95-1.11) 1.19 (1.10-1.28) 1.33 (1.23-1.43) 1.57 (1.46-1.68) < 0.001
Model 1a 1.00 1.09 (1.01-1.18) 1.25 (1.16-1.35) 1.42 (1.31-1.53) 1.65 (1.53-1.78) < 0.001
Model 2b 1.00 0.99 (0.92-1.07) 1.04 (0.97-1.13) 1.08 (1.00-1.17) 1.11 (1.03-1.20) < 0.001
NHS II
Median (g/day) 3.5 4.0 4.4 4.7 5.3
No. of cases/no. of person-years 890/320854 866/323113 954/320842 1136/321023 1377/319375
Age-adjusted 1.00 0.97 (0.88-1.07) 1.09 (0.99-1.19) 1.29 (1.18-1.41) 1.55 (1.42-1.68) < 0.001
Model 1a 1.00 1.02 (0.93-1.12) 1.13 (1.03-1.24) 1.35 (1.23-1.48) 1.56 (1.42-1.70) < 0.001
Model 2b 1.00 0.89 (0.81-0.98) 0.93 (0.84-1.02) 1.04 (0.95-1.14) 1.07 (0.98-1.17) 0.004
HPFS
Median (g/day) 3.7 4.2 4.6 4.9 5.5
No. of cases/no. of person-years 552/154366 586/155142 592/154510 714/154830 846/153717
Age-adjusted 1.00 1.06 (0.95-1.19) 1.07 (0.96-1.21) 1.29 (1.15-1.44) 1.52 (1.36-1.69) < 0.001
Model 1a 1.00 1.07 (0.95-1.20) 1.07 (0.95-1.20) 1.27 (1.13-1.42) 1.52 (1.36-1.69) < 0.001
Model 2b 1.00 0.98 (0.87-1.10) 0.93 (0.83-1.05) 1.06 (0.94-1.18) 1.15 (1.03-1.28)  0.002

Q, quintile; no., number.

aModel 1: adjusted for age (continuous), smoking status (never, past, current cigarettes/day: 1–14, 15–24, ≥ 25, missing), alcohol intake (g/day: 0, 0.1–4.9, 5.0–14.9, ≥ 15), physical activity (metabolic equivalent task hours/week: < 3, 3–8.9, 9–17.9, 18–26.9, ≥ 27, missing), menopausal status and postmenopausal hormone use in women (premenopausal, and postmenopausal with never, past, current hormone use), family history of diabetes (yes/no), history of hypertension (yes/no), hypercholesterolaemia (yes/no), total energy intake (kcal/day: in quintiles) and diabetes diet score (in quintiles).

bModel 2: adjusted for covariates in Model 1 and BMI (kg/m2, continuous) and BMI2.

Figure 1.

Figure 1.

Relative risk of type 2 diabetes according to quintiles of intakes of energy-adjusted BCAAs based on meta-analysed hazard ratios from NHS, NHS II and HPFS (A: leucine; B: isoleucine; C: valine; D: total BCAAs; all P for trend across quintiles < 0.0001). Hazard ratios were adjusted for age (months, continuous), smoking status (never, past, current cigarettes/day: 1–14, 15–24, ≥ 25, missing), alcohol intake (g/day: 0, 0.1–4.9, 5.0–14.9, ≥ 15), physical activity (metabolic equivalent task hours/week: < 3, 3–8.9, 9–17.9, 18–26.9, ≥ 27, missing), menopausal status and postmenopausal hormone use in women (premenopausal, and postmenopausal with never, past, current hormone use), family history of diabetes (yes/no), history of hypertension (yes/no), hypercholesterolaemia (yes/no), total energy intake (kcal/day: in quintiles), diabetes diet score (in quintiles), BMI (kg/m2, continuous) and BMI2.

The associations of total BCAA intake and risk of T2D are presented in Table 3. In the NHS, NHS II and HPFS, those with the highest intake of total BCAAs had 13%[HR (95% CI) 1.13 (1.05-1.22)], 8% [1.08 (0.99-1.18)] and 15% [1.15 (1.03-1.29)] higher risk of T2D, respectively, than those with the lowest intake. In the meta-analysis, total BCAA intake was associated with 12% [1.12 (1.06-1.18)] higher risk of T2D in the highest vs the lowest quintile of intake in the fully adjusted model (Figure 1D). When assessed as a continuous exposure, each 1-SD higher intake of total BCAAs was associated with 6% (95% CI: 4-7%) increased risk of incident T2D. Adjustment for total dietary protein or meat intake attenuated the associations between dietary BCAA intake and diabetes risk, although the positive relationship remained in the meta-analysis of the results from all three cohorts (Supplementary Figures 1 and 2, available as Supplementary data at IJE online). Adjustment for ‘total dietary protein intake minus BCAA intake’ generated similar results (data not shown).

Table 3.

Hazard ratios (95% CI) for risk of type 2 diabetes according to quintiles of total energy-adjusted BCAA intake in the NHS, NHS II and HPFS

Total BCAA intake
P for trendb
Q1 Q2 Q3 Q4 Q5
NHS
Median (g/day) 10.1 11.5 12.5 13.5 15.1
No. of cases/no. of person-years 1207/349691 1268/350328 1468/350410 1642/350192 1999/349489
Age-adjusted 1.00 1.05 (0.97-1.13) 1.21 (1.12-1.3) 1.35 (1.25-1.45) 1.63 (1.51-1.75) < 0.001
Model 1a 1.00 1.11 (1.02-1.20) 1.26 (1.17-1.36) 1.42 (1.32-1.53) 1.70 (1.57-1.83) < 0.001
Model 2b 1.00 1.00 (0.92-1.08) 1.05 (0.97-1.13) 1.08 (1.00-1.17) 1.13 (1.05-1.22) < 0.001
NHS II
Median (g/day) 12.0 13.7 14.9 16.1 18.0
No. of cases/no. of person-years 878/321018 854/321709 956/321880 1144/321115 1391/319484
Age-adjusted 1.00 0.98 (0.89-1.08) 1.11 (1.01-1.22) 1.32 (1.21-1.44) 1.60 (1.46-1.74) < 0.001
Model 1a 1.00 1.02 (0.93-1.13) 1.14 (1.03-1.25) 1.36 (1.24-1.50) 1.58 (1.44-1.73) < 0.001
Model 2b 1.00 0.90 (0.82-0.99) 0.93 (0.84-1.02) 1.05 (0.95-1.15) 1.08 (0.99-1.18)  0.002
HPFS
Median (g/day) 12.6 14.3 15.5 16.8 18.8
No. of cases/no. of person-years 550/154502 567/154961 596/154903 717/154516 860/153683
Age-adjusted 1.00 1.03 (0.92-1.16) 1.08 (0.96-1.22) 1.30 (1.16-1.45) 1.55 (1.39-1.72) < 0.001
Model 1a 1.00 1.03 (0.92-1.16) 1.07 (0.95-1.2) 1.27 (1.13-1.42) 1.53 (1.37-1.71) < 0.001
Model 2b 1.00 0.95 (0.84-1.07) 0.94 (0.83-1.05) 1.05 (0.94-1.18) 1.15 (1.03-1.29) < 0.001

Q, quintile; no., number.

aModel 1: adjusted for age (continuous), smoking status (never, past, current cigarettes/day: 1–14, 15–24, ≥ 25, missing), alcohol intake (g/day: 0, 0.1–4.9, 5.0–14.9, ≥ 15), physical activity (metabolic equivalent task hours/week: < 3, 3–8.9, 9–17.9, 18–26.9, ≥ 27, missing), menopausal status and postmenopausal hormone use in women (premenopausal, and postmenopausal with never, past, current hormone use), family history of diabetes (yes/no), history of hypertension (yes/no), hypercholesterolaemia (yes/no), total energy intake (kcal/day: in quintiles) and diabetes diet score (in quintiles).

bModel 2: adjusted for covariates in Model 1 and BMI (kg/m2, continuous) and BMI2.

In analyses stratified by diabetes risk factors (Table 4), only age stratification (< 60 vs ≥ 60 years old) and BMI stratification (< 25 vs ≥ 25 kg/m2) in NHS and BMI stratification (< 25 vs ≥ 25 kg/m2) in NHS II appeared to result in slightly modified risk. In NHS, the association of BCAA with diabetes risk was more pronounced among the younger and normal-weight women, compared with their counterparts, although risk nevertheless increased in all strata with increasing BCAA intake. In NHS II, across BMI quintiles the diabetes risk appeared non-linear among normal-weight women, whereas it increased among the overweight and obese women. Other than these, we did not observe modification effects of other risk factors or dietary factors on the association of BCAAs with diabetes risk in any cohort.

Table 4.

Stratified analysis of type 2 diabetes by quintiles of total energy-adjusted BCAA intake in the NHS, NHS II and HPFS

Cohorts Characteristic No. of cases BCAA HR (95% CI)a
Pinteraction
Q1 Q2 Q3 Q4 Q5
NHS Age, y 0.01
< 60 2655 1.00 1.14 (0.99-1.30) 1.08 (0.94-1.24) 1.23 (1.08-1.40) 1.25 (1.10-1.42)
≥ 60 4929 1.00 0.94 (0.86-1.04) 1.04 (0.94-1.14) 1.02 (0.93-1.12) 1.09 (0.99-1.20)
BMI, kg/m2 < 0.0001
< 25 1046 1.00 0.97 (0.79-1.17) 1.31 (1.09-1.58) 1.31 (1.08-1.59) 1.38 (1.12-1.68)
≥ 25 6538 1.00 1.05 (0.96-1.15) 1.10 (1.01-1.20) 1.20 (1.11-1.31) 1.35 (1.24-1.46)
Physical activity 0.26
< 18 METs/wk 5602 1.00 0.97 (0.88-1.06) 1.03 (0.94-1.12) 1.08 (0.99-1.17) 1.08 (0.99-1.18)
≥ 18 METs/wk 1749 1.00 1.14 (0.95-1.37) 1.16 (0.97-1.39) 1.12 (0.94-1.33) 1.31 (1.10-1.55)
Drinking status 0.90
< 5 g/d 5528 1.00 1.03 (0.93-1.13) 1.04 (0.95-1.14) 1.12 (1.02-1.23) 1.16 (1.05-1.26)
≥ 5 g/d 1221 1.00 0.93 (0.77-1.12) 1.15 (0.96-1.37) 0.98 (0.81-1.19) 1.20 (1.00-1.45)
NHS II Age, y 0.75
< 60 5042 1.00 0.89 (0.81-0.98) 0.91 (0.83-1.00) 1.02 (0.93-1.12) 1.08 (0.98-1.18)
≥ 60 181 1.00 1.34 (0.79-2.27) 1.32 (0.78-2.23) 1.80 (1.10 – 2.95) 1.18 (0.70 – 1.99)
BMI, kg/m2 < 0.0001
< 25 278 1.00 0.89 (0.62-1.29) 1.09 (0.77-1.56) 1.32 (0.93-1.88) 0.90 (0.60-1.36)
≥ 25 4944 1.00 0.94 (0.85-1.04) 1.00 (0.90-1.10) 1.15 (1.05-1.26) 1.28 (1.17-1.41)
Physical activity 0.60
< 18 METs/wk 3490 1.00 0.90 (0.80-1.01) 0.89 (0.79-0.99) 1.00 (0.90-1.12) 1.02 (0.91-1.13)
≥ 18 METs/wk 1061 1.00 0.84 (0.66-1.06) 1.08 (0.87-1.34) 1.17 (0.94-1.44) 1.10 (0.89-1.36)
Drinking status 0.59
< 5 g/d 3778 1.00 0.88 (0.79-0.99) 0.92 (0.82-1.02) 1.01 (0.91-1.12) 1.02 (0.91-1.12)
≥ 5 g/d 547 1.00 0.88 (0.67-1.16) 0.79 (0.60-1.05) 1.00 (0.76-1.31) 1.08 (0.82-1.43)
HPFS Age, y 0.64
< 60 1022 1.00 1.04 (0.84-1.29) 1.01 (0.82-1.25) 0.96 (0.77-1.18) 1.15 (0.94-1.41)
≥ 60 2268 1.00 0.91 (0.79-1.05) 0.91 (0.79-1.05) 1.10 (0.96-1.26) 1.15 (1.01-1.32)
BMI, kg/m2 0.46
< 25 542 1.00 1.06 (0.81-1.40) 1.04 (0.78-1.37) 1.28 (0.98-1.67) 1.50 (1.14-1.97)
≥ 25 2748 1.00 0.96 (0.84-1.09) 0.98 (0.86-1.12) 1.12 (0.99-1.27) 1.31 (1.16-1.48)
Physical activity 0.55
< 18 METs/wk 1749 1.00 1.04 (0.89-1.22) 0.93 (0.79-1.10) 1.09 (0.93-1.27) 1.19 (1.02-1.39)
≥ 18 METs/wk 1541 1.00 0.83 (0.69-0.99) 0.92 (0.78-1.09) 0.99 (0.84-1.17) 1.09 (0.92-1.28)
Drinking status 0.26
< 5 g/d 1462 1.00 0.94 (0.78-1.14) 1.01 (0.84-1.21) 1.03 (0.86-1.23) 1.21 (1.02-1.43)
≥ 5 g/d 1828 1.00 0.98 (0.84-1.14) 0.92 (0.79-1.08) 1.14 (0.98-1.32) 1.20 (1.03-1.39)

Q, quintile; y, years; wk, weeks; d, days; MET, metabolic equivalent.

aAdjusted for age (continuous), smoking status (never, past, current cigarettes/d: 1-14, 15-24, ≥ 25, missing), alcohol consumption (g/d: 0, 0.1-4.9, 5.0-14.9, ≥ 15), physical activity (metabolic equivalent task hours/week: < 3, 3-8.9, 9-17.9, 18-26.9, ≥ 27, missing), menopausal status and postmenopausal hormone use in women (premenopausal, and postmenopausal with never, past, current hormone use), family history of diabetes (yes/no), history of hypertension (yes/no), hypercholesterolaemia (yes/no), total daily energy intake (in quintiles), diabetes diet score (in quintiles), BMI (kg/m2, continuous) and BMI2.

In the subsample of NHS and HPFS (N = 397) with plasma measures of BCAA metabolites, we found direct associations between dietary intakes and plasma levels of total BCAA. The unadjusted Spearman rank correlation coefficients between consumption of individual BCAAs and their plasma levels ranged from 0.11 to 0.14 (all P < 0.03).

Discussion

In three large, prospective cohorts of US men and women, we observed consistent associations of long-term consumption of BCAAs, including leucine, isoleucine and valine, individually or in sum, with increased risk of incident T2D. These associations were independent of traditional diabetes risk factors, including BMI.

To our knowledge, only one previous study examined the association between dietary BCAA intake and risk of diabetes, which was conducted in a Japanese population.16 In the Takayama study, intakes of BCAAs were calculated as a percentage of total dietary protein, not as absolute intake as in our study. The percentages of total BCAAs, leucine and valine in protein intakes were inversely associated with T2D risk in women; no associations were observed in men. Whereas absolute intakes of total BCAAs were similar in both our study and the Takayama study, in contrast to the present US population in which the major food contributors to BCAA intake were meat (∼ 37%), milk (∼ 12%) and fish (∼ 8%), the major contributors in the Japanese diet were cereals, potatoes, and starches (23–25%), fish and shellfish (21–23%) and meats (14–15%).16 In addition, in the Takayama study population, the intake of BCAA was relatively homogeneous (2 − 3 g difference across BCAA tertiles) assessed by a baseline (single time point) assessment, whereas in our study, the range of cumulative averaged intake of BCAA across quintiles was about 5 − 6 g assessed by quadrennial repeated assessments.

In our populations, meat is the main source of dietary BCAA and dietary BCAAs were highly correlated with dietary total protein and animal protein; and thus distinguishing between effects of intakes of BCAA’s, meat, total protein and animal protein was difficult. However, the association between BCAAs and diabetes risk remained, although attenuated, after adjusting for total protein intake or especially total meat intake. Of note, adjustment for total protein intake or meat intake may be considered an over-adjustment because meat was a main source of dietary BCAAs in these cohorts.

Our findings are consistent with results from several recent studies in which high circulating levels of BCAAs or associated genetic markers were associated with diabetes risk, impaired fasting glucose or insulin resistance.6,7,12,13,30 In the prospective Framingham Offspring Study and Malmö Diet and Cancer Study, Wang et al. reported that individuals with the highest quartile of plasma BCAA had nearly 3-fold higher risk of T2D compared with those in the lowest quartile.6 The associations of blood BCAAs with diabetic risk in that study were stronger than those we observed for dietary BCAAs in the present study. Although diet is the only source of BCAAs and 80% of dietary BCAAs reach blood circulation,31 circulating levels of BCAAs are also affected by their catabolism.32 We observed positive but moderate correlations between dietary and plasma levels of BCAAs. These correlations were similar to that observed in another study (r = 0.14).33 As we discuss further below, higher blood concentrations of BCAAs may reflect an early disturbance of protein metabolism that worsens further if intake of BCAAs remains unchanged and/or high.

Indeed, there are biologically plausible mechanisms for the adverse effects of high intakes of BCAAs on diabetes risk. In humans, BCAAs not only serve as precursors in protein and peptide synthesis but also play regulatory roles in insulin and glucose metabolism.34 At the molecular level, BCAAs, especially leucine, can activate pathways which inhibit insulin signalling and insulin-stimulated glucose transport in muscle and fat. In several recent studies, BCAA deprivation improved overall and hepatic insulin sensitivity in animal models,35,36 and reduction in blood BCAAs was found to be associated with improvement in blood sugar regulation in humans.37 In the context of over-nutrition, rising circulating BCAAs lead to increased flux of these amino acids through their catabolic pathways. A ‘BCAA overload hypothesis’ posits that increased BCAA catabolic flux may contribute to increased gluconeogenesis and glucose intolerance.8

Recently, there has been growing interest in treating insulin resistance with dietary manipulation of micronutrients, including BCAAs.34 It is noteworthy that studies of dietary supplementation of BCAAs, especially leucine on insulin sensitivity, have generated mixed results. Whereas some studies found that increased oral intake of leucine might improve whole-body glucose metabolism in mice maintained on a high-fat diet,15 other studies found that increased serum levels of leucine either had no effect or might increase insulin resistance in humans or in animal models of obesity.8,38,39 The reasons underlying the conflicting observations remain unclear. Nevertheless, our results and the evidence from recent metabolomic studies6,7,12,13,30 lend support to potentially adverse, rather than beneficial, effects of high consumption of BCAAs on insulin resistance, hyperglycaemia and T2D.

Previous studies suggest the detrimental effects of BCAAs on insulin resistance might be strengthened in the context of a dietary pattern such as high fat consumption.8 However, we did not observe interactions between BCAA intakes and dietary fat (data not shown). Stratification by other traditional diabetes risk factors such as obesity, smoking, alcohol consumption and physical activity also did not modify the associations between BCAAs and diabetes risk in our cohorts.

Our study has strengths and limitations. The prospective nature of the study design minimized the likelihood of recall and selection biases, and the high follow-up rates largely reduced the concern that the results were affected by differential follow-up rates. Residual confounding is a common and unavoidable issue in observational studies. We sought to minimize the influence of the potential confounders by controlling for potentially confounding variables including major lifestyle and dietary risk factors. The association between BCAA intakes and diabetes risk might be confounded by adiposity. Adjustment for BMI substantially attenuated the associations, although the positive associations remained; the possibility that adjustment for a more perfect measure of adiposity might have further attenuated the associations cannot be excluded. However, when we further adjusted for waist circumference in model 2, the association between BCAA intakes and T2D risk did not materially change (data not shown). In addition, our dietary assessments were based on previously validated FFQs, and we reduced measurement error by using repeated measurements and cumulatively averaging intake. We only measured plasma BCAA levels in a small sample of participants, and the correlation of dietary and plasma BCAAs was moderate but comparable to other diet-plasma biomarker correlations. However, as we mentioned above, in our study we were not able to distinguish clearly between the effects of BCAA and those of total protein or animal protein. Considering the long follow-up periods in our cohorts, competing risk from non-response might be an issue. However, only ∼ 7% of all participants were lost to follow-up and these participants were censored at the time of loss to follow-up. Therefore, non-response might have very minor influence on our analysis. Furthermore, most of the participants of our study were Caucasian, US health professionals; this generates within-study homogeneity across many biological and social characteristics, but it compromises the generalizability of our findings such that their applicability to other populations may be limited.

In summary, we observed consistent associations between higher dietary intakes of BCAAs and increased risk of T2D in three cohorts. More prospective studies and randomized clinical trials are warranted to investigate the potential roles of dietary BCAAs in insulin resistance, hyperglycaemia and diabetes.

Supplementary Data

Supplementary data are available at IJE online.

Supplementary Data

Acknowledgements

We thank all the participants of the Nurses’ Health Study, Nurses’ Health Study II and the Health Professionals Follow-up Study for their continued cooperation.

Funding

This study was supported by the: National Institutes of Health P01 CA87969, UM1 CA186107, R01 CA49449, UM1 CA176726, UM1 CA167552, R01 DK091718, HL71981, DK58845 and HL60712; United States – Israel Binational Science Foundation Grant 2011036; American Heart Association Scientist Development Award; and Boston Obesity Nutrition Research Center (DK46200).

Conflict of interest: Q.Qi. was a recipient of the American Heart Association Scientist Development Award (0730094N). No other disclosures or conflicts of interest were reported.

References

  • 1.Beckman JA,, Creager MA,, Libby P. Diabetes and atherosclerosis:epidemiology, pathophysiology, and management. JAMA 2002;287:2570–81. [DOI] [PubMed] [Google Scholar]
  • 2.Nolan CJ, Damm P, Prentki M. Type 2 diabetes across generations:from pathophysiology to prevention and management. Lancet 2011;378:169–81. [DOI] [PubMed] [Google Scholar]
  • 3.Zimmet P, Alberti KG, Shaw J. Global and societal implications of the diabetes epidemic. Nature 2001;414:782–87. [DOI] [PubMed] [Google Scholar]
  • 4.Laakso M. Prevention of type 2 diabetes. Curr Mol Med 2005;5:365–74. [DOI] [PubMed] [Google Scholar]
  • 5.Delahanty LM, Nathan DM. Implications of the diabetes prevention program and Look AHEAD clinical trials for lifestyle interventions. J Am Diet Assoc 2008;108:S66–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Wang TJ, Larson MG, Vasan RS. et al. Metabolite profiles and the risk of developing diabetes. Nat Med 2011;17:448–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Xu F, Tavintharan S, Sum CF, Woon K, Lim SC, Ong CN. Metabolic signature shift in type 2 diabetes mellitus revealed by mass spectrometry-based metabolomics. J Clin Endocrinol Metab 2013;98:E1060–65. [DOI] [PubMed] [Google Scholar]
  • 8.Newgard CB, An J, Bain JR. et al. A branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance. Cell Metab 2009;9:311–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Tai ES, Tan ML, Stevens RD. et al. Insulin resistance is associated with a metabolic profile of altered protein metabolism in Chinese and Asian-Indian men. Diabetologia 2010;53:757–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.McCormack SE, Shaham O, McCarthy MA. et al. Circulating branched-chain amino acid concentrations are associated with obesity and future insulin resistance in children and adolescents. Pediatr Obes 2013;8:52–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Wurtz P, Tiainen M, Makinen VP. et al. Circulating metabolite predictors of glycemia in middle-aged men and women. Diabetes Care 2012;35:1749–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Scholtens DM, Muehlbauer MJ,, Daya NR. et al. Metabolomics reveals broad-scale metabolic perturbations in hyperglycemic mothers during pregnancy. Diabetes Care 2014;37:158–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Xu M, Qi Q, Liang J. et al. Genetic determinant for amino acid metabolites and changes in body weight and insulin resistance in response to weight-loss diets:the Preventing Overweight Using Novel Dietary Strategies (POUNDS LOST) trial. Circulation 2013;127:1283–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Macotela Y, Emanuelli B, Bang AM. et al. Dietary leucine - an environmental modifier of insulin resistance acting on multiple levels of metabolism. PloS One 2011;6:e21187. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Zhang Y, Guo K, LeBlanc RE, Loh D, Schwartz GJ, Yu YH. Increasing dietary leucine intake reduces diet-induced obesity and improves glucose and cholesterol metabolism in mice via multimechanisms. Diabetes 2007;56:1647–54. [DOI] [PubMed] [Google Scholar]
  • 16.Nagata C, Nakamura K, Wada K, Tsuji M, Tamai Y, Kawachi T. Branched-chain amino acid intake and the risk of diabetes in a Japanese community:the Takayama study. Am J Epidemiol 2013;178:1226–32. [DOI] [PubMed] [Google Scholar]
  • 17.Rich-Edwards JW, Goldman MB, Willett WC. et al. Adolescent body mass index and infertility caused by ovulatory disorder. Am J Obstet Gynecol 1994;171:171–77. [DOI] [PubMed] [Google Scholar]
  • 18.Colditz GA, Manson JE, Hankinson SE. The Nurses' Health Study: 20-year contribution to the understanding of health among women. J Womens Health 1997;6:49–62. [DOI] [PubMed] [Google Scholar]
  • 19.Rimm EB, Giovannucci EL, Willett WC. et al. Prospective study of alcohol consumption and risk of coronary disease in men. Lancet 1991;338:464–68. [DOI] [PubMed] [Google Scholar]
  • 20.Rimm EB, Giovannucci EL, Stampfer MJ, Colditz GA, Litin LB, Willett WC. Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol 1992;135:1114–26; discussion 27-36. [DOI] [PubMed] [Google Scholar]
  • 21.Rimm EB, Giovannucci EL, Stampfer MJ, Colditz GA, Litin LB, Willett WC. Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol 1992;135:1114–26; discussion 27-36. [DOI] [PubMed] [Google Scholar]
  • 22.Willett WC, Sampson L, Stampfer MJ. et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 1985;122:51–65. [DOI] [PubMed] [Google Scholar]
  • 23.Willett W, Stampfer MJ. Total energy intake:implications for epidemiologic analyses. Am J Epidemiol 1986;124:17–27. [DOI] [PubMed] [Google Scholar]
  • 24.National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 1979;28:1039–57. [DOI] [PubMed] [Google Scholar]
  • 25.Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 1997;20:1183–97. [DOI] [PubMed] [Google Scholar]
  • 26.Hu FB, Stampfer MJ, Solomon C. et al. Physical activity and risk for cardiovascular events in diabetic women. Ann Intern Med 2001;134:96–105. [DOI] [PubMed] [Google Scholar]
  • 27.Manson JE, Colditz GA, Stampfer MJ. et al. A prospective study of maturity-onset diabetes mellitus and risk of coronary heart disease and stroke in women. Arch Intern Med 1991;151:114–47. [PubMed] [Google Scholar]
  • 28.Hu FB, Stampfer MJ, Rimm E. et al. Dietary fat and coronary heart disease: a comparison of approaches for adjusting for total energy intake and modeling repeated dietary measurements. Am J Epidemiol 1999;149:531–40. [DOI] [PubMed] [Google Scholar]
  • 29.Jenkins DJ, Kendall CW, Augustin LS. et al. Glycemic index: overview of implications in health and disease. Am J Clin Nutr 2002;76:266–73S. [DOI] [PubMed] [Google Scholar]
  • 30.Menni C, Fauman E, Erte I. et al. Biomarkers for type 2 diabetes and impaired fasting glucose using a non-targeted metabolomics approach. Diabetes 2013;62:4270–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Rietman A, Schwarz J, Tome D, Kok FJ, Mensink M. High dietary protein intake, reducing or eliciting insulin resistance? Eur J Clin Nutr 2014;68:973–79. [DOI] [PubMed] [Google Scholar]
  • 32.Newgard CB. Interplay between lipids and branched-chain amino acids in development of insulin resistance. Cell Metab 2012;15:606–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Shah SH, Crosslin DR, Haynes CS. et al. Branched-chain amino acid levels are associated with improvement in insulin resistance with weight loss. Diabetologia 2012;55:321–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Wu G. Functional amino acids in nutrition and health. Amino Acids 2013;45:407–11. [DOI] [PubMed] [Google Scholar]
  • 35.Xiao F, Huang Z, Li H. et al. Leucine deprivation increases hepatic insulin sensitivity via GCN2/mTOR/S6K1 and AMPK pathways. Diabetes 2011;60:746–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Guo F, Cavener DR. The GCN2 eIF2alpha kinase regulates fatty-acid homeostasis in the liver during deprivation of an essential amino acid. Cell Metab 2007;5:103–14. [DOI] [PubMed] [Google Scholar]
  • 37.Laferrere B, Reilly D, Arias S. et al. Differential metabolic impact of gastric bypass surgery versus dietary intervention in obese diabetic subjects despite identical weight loss. Sci Translat Med 2011;3:80re2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Nairizi A, She P, Vary TC, Lynch CJ. Leucine supplementation of drinking water does not alter susceptibility to diet-induced obesity in mice. J Nutr 2009;139:715–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Tremblay F, Krebs M, Dombrowski L. et al. Overactivation of S6 kinase 1 as a cause of human insulin resistance during increased amino acid availability. Diabetes 2005; 54: 2674–84. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data

Articles from International Journal of Epidemiology are provided here courtesy of Oxford University Press

RESOURCES