OBJECTIVE

The long-term association between dietary protein and type 2 diabetes incidence is uncertain. We aimed to investigate the association between total, animal, and plant protein intake and the incidence of type 2 diabetes.

RESEARCH DESIGN AND METHODS

The prospective European Prospective Investigation into Cancer and Nutrition (EPIC)-InterAct case-cohort study consists of 12,403 incident type 2 diabetes cases and a stratified subcohort of 16,154 individuals from eight European countries, with an average follow-up time of 12.0 years. Pooled country-specific hazard ratios (HRs) and 95% CI of prentice-weighted Cox regression analyses were used to estimate type 2 diabetes incidence according to protein intake.

RESULTS

After adjustment for important diabetes risk factors and dietary factors, the incidence of type 2 diabetes was higher in those with high intake of total protein (per 10 g: HR 1.06 [95% CI 1.02–1.09], Ptrend < 0.001) and animal protein (per 10 g: 1.05 [1.02–1.08], Ptrend = 0.001). Effect modification by sex (P < 0.001) and BMI among women (P < 0.001) was observed. Compared with the overall analyses, associations were stronger in women, more specifically obese women with a BMI >30 kg/m2 (per 10 g animal protein: 1.19 [1.09–1.32]), and nonsignificant in men. Plant protein intake was not associated with type 2 diabetes (per 10 g: 1.04 [0.93–1.16], Ptrend = 0.098).

CONCLUSIONS

High total and animal protein intake was associated with a modest elevated risk of type 2 diabetes in a large cohort of European adults. In view of the rapidly increasing prevalence of type 2 diabetes, limiting iso-energetic diets high in dietary proteins, particularly from animal sources, should be considered.

Dietary proteins are advocated to have positive effects on weight loss and weight maintenance due to properties related to satiety and diet-induced thermogenesis (1). Doubling the relative protein content of the diet under ad libitum conditions for 12 weeks reduces food intake and can lower body weight by >6% (2). Therefore, increasing protein intake seems a promising approach to tackle the obesity epidemic and therewith to reduce the incidence of chronic diseases.

In contrast, long-term observational studies report an association of high protein intake with a higher risk of type 2 diabetes (3,4). Suggested beneficial acute effects of dietary protein on insulin secretion and glycemic control (5) do not seem to persist mid- and long-term (6,7). An iso-energetic high-protein diet compared with a low-protein high-fiber diet reduced insulin sensitivity after 6 weeks (6), and participants with a habitually high- compared with a normal-protein diet showed signs of reduced insulin sensitivity (7). The few epidemiological studies that addressed the association between protein intake and type 2 diabetes all found an increased type 2 diabetes risk with high protein and/or meat protein intake (3,4,810). However, most of these studies had small sample sizes, ranging from 140 (8,9) to 1,200 (10) participants. Of two large cohort studies with >35,000 participants (3,4), Sluijs et al. (3) did not observe a significant association for total protein after adjustment for BMI and waist circumference in the Dutch cohort of European Prospective Investigation into Cancer and Nutrition (EPIC)-InterAct. Based on subanalyses, the authors concluded that only for participants with a BMI <25 kg/m2 does high protein intake increase type 2 diabetes risk (3).

Besides total and animal protein, prior research suggests that the protein source could be of relevance. Type 2 diabetes risk is associated with higher meat consumption, particularly red and processed meat (1113). Type 2 diabetes risk is reported to be lower in subjects with high dairy use (1417) and/or high plant product consumption, especially of legumes (18) and nuts (19). Data on the relation between fish consumption and type 2 diabetes reached mixed conclusions (2022), and a recent study in the EPIC-InterAct case cohort reported no association (23). It is unclear whether it is the protein or other nutrients in such food groups that explain the association with type 2 diabetes.

Within the setting of EPIC-InterAct, we were able to study the association between protein intake and risk of type 2 diabetes in a large case cohort in eight countries in Europe: the largest cohort of type 2 diabetes so far (24). The characteristics of this study made it possible to explore the association between protein intake and type 2 diabetes according to plant or animal origin and by protein source.

Study Design

The participants, methods, study design, and measurements have previously been described (24). Briefly, the InterAct project was initiated to investigate how genetic and lifestyle behavioral factors, particularly diet and physical activity, interact to lead to type 2 diabetes. As part of the wider InterAct project, consortium partners have established a case-cohort study of incident type 2 diabetes (EPIC-InterAct case cohort) based on cases occurring in EPIC cohorts between 1991 and 2007 in 8 of 10 EPIC countries. A case-cohort study is comparable with a cohort study but it is more efficient, as it uses a random sample of the cohort to compare cases with. The case-cohort design combines the advantages of a prospective cohort with the efficiency and power of a large case-control study.

Type 2 Diabetes Case Ascertainment

We followed a pragmatic high-sensitivity approach for case ascertainment with the aim of 1) identifying all potential incident type 2 diabetes cases and 2) excluding all individuals with prevalent type 2 diabetes. Prevalent and incident type 2 diabetes was identified using multiple sources of evidence including self-report, linkage to primary care registers and secondary care registers, medication use (drug registers), hospital admissions, and mortality data. Further details have previously been published (24).

Subcohort

A subcohort of 16,835 individuals was randomly selected stratified by center. After exclusion of 548 individuals with prevalent type 2 diabetes, 129 individuals without information on reported diabetes status, and 4 individuals with postcensoring type 2 diabetes, 16,154 subcohort individuals remained.

Participants

We used a case-cohort design, including incident diabetes cases (n = 12,403) and a representative subcohort (n = 16,154, including 778 cases of incident type 2 diabetes) selected from the EPIC cohort (24). After exclusion of participants with missing information on dietary data (n = 117; 70 case subjects, 47 subcohort) or other missing covariates, i.e., physical activity, educational, and smoking status (n = 790; 357 case subjects, 433 subcohort), and participants who fell in the top or bottom 1% of the “energy intake/energy requirement ratio” (n = 619; 339 case subjects, 280 subcohort), our analysis included 26,253 participants (10,901 incident type 2 diabetes case subjects and a subcohort of 15,352 participants including 736 cases of incident type 2 diabetes).

All EPIC study participants gave written informed consent, and the study was approved by the national ethics committees and the International Agency for Research on Cancer.

Protein Intake and Other Variables

Dietary intake, over the 12 months before enrollment, was assessed by self- or interviewer-administered dietary questionnaires (mainly food frequency questionnaires [FFQ]), developed and validated within each country, to estimate the usual individual food intakes of the study participants (for more detail, see Riboli et al. [25]). Protein intake (g/day) was adjusted for total energy intake by the residual method (26) and categorized in quintiles according to the data of the subcohort. As part of EPIC, standardized information on lifestyle exposures was collected by self-administered national questionnaires at baseline (25). Physical activity during work and leisure time was classified in four categories according to the Cambridge Physical Activity Index (27). Weight, height, and waist circumference were recorded using a standard protocol during a visit at the research center, except in Oxford (U.K.) and France, where only self-reported height and weight were available (25).

Statistical Analysis

The association between energy-adjusted protein intake and type 2 diabetes risk was examined in hazard ratios (HRs) and 95% CIs using Cox proportional hazards models adapted to case-cohort designs according to the Prentice-weighted method (28). We stratified all analyses by country, mainly because of the large dietary heterogeneity between countries, specifically between northern and southern Europe, e.g., relatively high protein intake in Spain and low protein intake in Germany and Sweden. We used random-effect meta-analyses to pool the country-specific HRs. Between-country heterogeneity was tested by I2 statistic. Linear associations between protein intake and type 2 diabetes were estimated per 10-g increment of protein intake and by analyzing linear trends across protein intake categories using the median value of each quintile as a continuous variable. We adjusted for type 2 diabetes risk factors and nutritional factors using a stepwise approach, with age as the underlying time scale. Model 1 included protein intake, total energy (kilocalories per day), center, and sex. In model 2, we added type 2 diabetes risk factors, i.e., smoking status (never, former, or current), education (low, secondary, or high), physical activity (inactive, moderately inactive, moderately active, or active), and alcohol use (0, >0–6, >6–12, >12–24, or >24 g/day). Model 3 was additionally adjusted for soft drinks, tea, coffee, and the following residual adjusted nutrients: fiber, saturated fatty acids, monounsaturated fatty acids (MUFAs), polyunsaturated fatty acids (PUFAs), and cholesterol. It was not adjusted for carbohydrates to estimate the effect of increasing protein at the expense of carbohydrates. Model 4 was additionally adjusted for BMI (measured as weight in kilograms divided by the square of height in meters) and waist circumference (centimeters).

Effect modification was examined by various dietary and lifestyle factors, i.e., sex, BMI, waist circumference, physical activity, smoking, menopausal status, hyperlipidemia, and hypertension. A country-specific multivariate Wald test was used to evaluate interactions by continuous interaction terms. In case of significant interactions, HRs were stratified. The association between protein intake and type 2 diabetes risk was estimated within each country by sex (11,241 men; 15,012 women) and by BMI (normal: BMI <25 kg/m2, n = 8,317; overweight: BMI 25–30 kg/m2, n = 10,951; obese: BMI ≥30 kg/m2, n = 6,985) and meta-analyzed. Models for sources of animal and plant protein and sequentially excluding main sources of protein, e.g., meat, were designed to explore the contribution of various protein sources to the associations of protein type with type 2 diabetes risk. Finally, sensitivity analyses were performed by excluding individuals who might have made dietary adjustments and/or lifestyle changes because of chronic disease at baseline (i.e., hypertension, hyperlipidemia, myocardial infarction, and/or stroke) and by excluding misreports of energy according to Goldberg criterion categories, defined as “under-reporters” with a ratio of energy intake to basal metabolic rate <1.14 and “over-reporters” with a ratio of >2.1 (29). Data analysis was performed using SAS 9.2, and the meta-analyses were conducted in STATA11.

Our analysis of this case-cohort study consisted of 10,901 incident type 2 diabetes cases and a subcohort of 15,352 participants (including 736 type 2 diabetes cases) with an mean ± SD follow-up time of 12.0 ± 2.3 years. FFQ-based median estimated energy-adjusted total protein intake was 90.4 g/day for men and 91.0 g/day for women, mainly consisting of animal protein. It was highest in Spain (102.5 g/day) and lowest in Germany and Sweden (respectively, 80.0 and 80.8 g/day) (Supplementary Fig. 1). Main animal protein sources in order of proportion were meat, dairy, and fish; main plant protein sources were bread, pasta and rice, potatoes, and vegetables (Table 1).

Table 1

Sex-specific characteristics and dietary consumption of the EPIC-InterAct subcohort by categories of total protein intake* (n = 15,352)

Men
Women
Q1: 68.9 g/day
(63–73)Q3: 89.0 g/day
(87–91)Q5: 113.5 g/day (109–121)Q1: 73.5 g/day
(69–77)Q3: 90.6 g/day
(89–92)Q5: 109.5 g/day (105–116)
n cases/N 45/1,162 76/1,162 105/1,161 51/1,909 83/1,909 96/1,909 
Characteristics       
 Age (years) 53.0 ± 9.4 53.5 ± 9.0 51.4 ± 7.8 51.1 ± 9.9 53.0 ± 8.9 51.3 ± 8.6 
 Follow-up (years) 11.8 ± 2.6 11.8 ± 2.7 12.2 ± 2.6 11.8 ± 2.1 11.9 ± 2.2 12.3 ± 2.1 
 BMI (kg/m225.6 ± 3.5 26.5 ± 3.5 27.9 ± 3.6 24.2 ± 3.9 25.7 ± 4.4 27.1 ± 4.8 
  <25 (%) 44.6 32.9 20.2 66.4 50.3 36.8 
  25–30 (%) 44.8 52.9 54.0 25.1 33.7 40.4 
  >30 (%) 10.7 14.2 25.8 8.5 16.0 22.7 
 Waist (cm) 92.8 ± 10.1 94.7 ± 9.6 97.9 ± 9.6 77.6 ± 10.0 81.1 ± 10.7 84.5 ± 11.6 
 Family history of diabetes (%)       
  Yes 15.8 17.1 11.7 20.4 22.1 21.4 
 Smoker (%)       
  Never 34.5 30.7 30.2 52.9 57.1 59.8 
  Former 36.8 37.9 33.4 23.5 21.3 20.3 
  Smoker 28.7 31.4 36.4 23.7 21.7 20.0 
 Hypertension (%)       
  Yes 18.1 18.9 18.1 16.9 18.8 18.5 
  Do not know 2.1 3.9 3.9 1.5 0.9 1.0 
 Hyperlipidemia (%)       
  Yes 22.6 20.9 23.2 16.1 14.1 13.8 
  Do not know 9.3 12.9 7.3 6.4 7.1 4.8 
 Educational level (%)       
  Long education (incl. university degree) 28.5 24.5 19.0 23.5 17.7 14.4 
 Physical activity (%)       
  Active 24.2 22.8 27.2 20.0 17.4 13.6 
 Postmenopausal (%) — — — 45.8 51.7 41.1 
Country, n       
 France, ♀ 526 (%) — — — 3.4 5.9 6.8 
 Italy, ♂ 639, ♀1,300 (%) 7.5 13.6 6.7 10.7 16.1 11.4 
 Spain, ♂ 1,333, ♀ 2,154 (%) 2.5 15.5 58.7 3.1 19.1 49.9 
 U.K., ♂ 423, ♀ 661 (%) 7.8 7.6 4.2 6.9 6.0 8.6 
 The Netherlands, ♂ 226, ♀ 1,139 (%) 2.4 5.3 3.0 8.3 15.0 9.8 
 Germany, ♂ 845, ♀ 1,176 (%) 26.7 15.6 4.1 28.7 9.2 1.6 
 Sweden, ♂ 1,231, ♀ 1,625 (%) 40.5 20.2 4.6 31.8 16.1 2.4 
 Denmark, ♂ 1,110, ♀ 964 (%) 12.6 22.3 18.7 7.0 12.5 9.6 
Dietary intake       
 Total energy (kcal/day) 2,564.2 ± 673.6 2,361.0 ± 594.2 2,664.3 ± 674.5 2,061.2 ± 567.5 1,857.7 ± 485.3 2,005.1 ± 525.2 
 Total protein (energy %) 12.8 ± 1.2 16.5 ± 0.6 20.7 ± 1.9 13.2 ± 1.2 17.2 ± 0.8 21.6 ± 2.4 
  Animal protein (energy %) 6.8 ± 1.7 10.2 ± 1.4 14.4 ± 2.5 6.8 ± 1.8 10.6 ± 1.5 15.1 ± 2.8 
   From red meat (g/day)* 6.6 (3–12) 13.7 (8–21) 19.0 (11–28) 6.6 (4–10) 11.9 (8–17) 16.3 (10–23) 
   From processed meat (g/day)* 5.4 (3–8) 5.8 (3–9) 6.2 (3–12) 4.8 (3–7) 5.3 (3–8) 6.0 (4–10) 
   From poultry (g/day)* 2.6 (1–4) 4.8 (3–8) 9.8 (5–16) 2.7 (2–4) 5.1 (3–8) 10.3 (6–16) 
   From milk and dairy (g/day)* 5.3 (1–11) 7.0 (2–14) 7.5 (3–15) 5.4 (2–10) 8.5 (5–14) 10.9 (6–17) 
   From cheese (g/day)* 5.2 (2–9) 6.8 (3–12) 7.0 (3–15) 7.2 (4–10) 9.0 (6–13) 9.5 (5–16) 
   From fish (g/day)* 2.6 (1–5) 5.6 (3–9) 11.8 (6–19) 3.2 (2–5) 5.8 (3–9) 10.3 (6–16) 
   From eggs (g/day)* 1.2 (0–3) 2.0 (1–3) 3.0 (1–5) 1.5 (1–3) 2.2 (1–3) 2.9 (2–4) 
  Plant protein (energy %) 4.4 ± 1.3 5.0 ± 1.3 5.4 ± 1.4 4.7 ± 1.4 5.2 ± 1.3 5.3 ± 1.3 
   From bread (g/day)* 10.0 (7–14) 11.0 (8–15) 11.4 (8–17) 10.4 (8–13) 11.4 (9–15) 11.2 (8–14) 
   From potatoes (g/day)* 2.0 (1–3) 2.2 (1–3) 1.7 (1–3) 1.5 (1–2) 1.6 (1–2) 1.4 (1–2) 
   From pasta and rice (g/day)* 1.2 (0–2) 1.6 (1–3) 1.8 (1–3) 1.5 (1–3) 1.9 (1–3) 1.9 (1–3) 
   From legumes (g/day)* 0.2 (0–1) 0.5 (0–2) 3.1 (0–6) 0.6 (0–1) 1.1 (1–2) 2.1 (1–4) 
 Total carbohydrates (energy %) 46.0 ± 7.7 43.1 ± 6.8 38.7 ± 6.6 48.0 ± 7.2 45.1 ± 6.1 40.9 ± 6.2 
  Starch (energy %) 24.0 ± 6.6 24.6 ± 6.3 23.3 ± 6.1 24.0 ± 6.4 24.1 ± 6.0 22.0 ± 5.8 
  Sugar (energy %) 21.0 ± 7.2 17.8 ± 5.8 15.0 ± 5.3 23.4 ± 7.1 20.5 ± 5.8 18.5 ± 5.9 
 Total fat (energy %) 34.4 ± 7.0 34.1 ± 5.5 35.4 ± 5.6 35.1 ± 6.4 34.9 ± 5.6 35.5 ± 5.9 
  Saturated fat (energy %) 14.1 ± 4.0 13.0 ± 3.4 11.9 ± 3.3  13.3 ± 3.1 12.6 ± 3.3 
  Monounsaturated fat (energy %) 12.3 ± 2.8 12.9 ± 2.9 14.7 ± 3.8 12.5 ± 3.0 13.1 ± 3.6 14.2 ± 4.0 
  Polyunsaturated fat (energy %) 5.5 ± 2.0 5.4 ± 1.8 5.8 ± 2.2 5.7 ± 1.9 5.6 ± 1.7 5.7 ± 2.2 
 Fiber (g)* 20.1 ± 7.0 22.2 ± 6.1 23.9 ± 7.5 23.0 ± 6.6 23.5 ± 5.5 24.1 ± 5.9 
 Cholesterol (mg)* 274.4 ± 118.4 334.9 ± 108.3 422.6 ± 138.0 288.0 ± 93.2 342.2 ± 91.6 405.5 ± 105.9 
 Calcium (mg)* 792.7 ± 286.0 941.9 ± 335.4 1,026.5 ± 501.6 923.2 ± 247.5 1,054.3 ± 265.9 1,188.8 ± 453.7 
 Magnesium (mg)* 383.7 ± 107.8 379.5 ± 98.8 429.3 ± 110.3 330.1 ± 93.6 315.8 ± 86.7 353.8 ± 96.5 
 Vitamin B1 (mg)* 1.4 ± 0.5 1.4 ± 0.5 1.8 ± 0.6 1.2 ± 0.4 1.2 ± 0.4 1.4 ± 0.5 
 β-Carotene (µg)* 2,232.9 ± 2,048.6 2,596.7 ± 2,124.2 2,817.9 ± 2,483.3 3,119.4 ± 2,490.5 3,194.8 ± 2,403.8 3,493.4 ± 2,917.0 
 Vitamin C (mg)* 109.8 ± 64.8 110.7 ± 59.1 140.7 ± 79.8 129.9 ± 79.9 121.9 ± 60.3 141.0 ± 70.8 
 Vitamin D (mg)* 4.6 ± 3.0 4.6 ± 2.9 5.1 ± 3.8 4.1 ± 2.2 4.1 ± 2.1 4.4 ± 2.2 
 Vitamin E (mg)* 12.5 ± 6.0 11.7 ± 5.2 14.7 ± 6.8 11.7 ± 5.4 10.4 ± 4.2 11.9 ± 5.1 
 Soft drinks (g/day)* 37.9 (0–150) 15.9 (0–86) 0.0 (0–29) 13.3 (0–90) 2.4 (0–57) 0.0 (0–28) 
 Coffee (g/day)* 429.8 (190–621) 311.1 (100–611) 130.6 (43–450) 357.1 (120–580) 261.3 (89–500) 160.0 (52–450) 
 Tea (g/day)* 10.3 (0–150) 4.9 (0–146) 0.0 (0–12) 21.4 (0–250) 6.6 (0–238) 0.0 (0–119) 
 Alcohol (g)       
  0 (%) 4.7 4.2 5.4 7.3 9.0 12.9 
  ♂ 0–12, ♀ 0–6 (%) 42.0 39.1 39.6 46.6 53.6 55.4 
  ♂ 12–24, ♀ 6–12 (%) 18.1 22.4 20.4 17.2 15.0 13.4 
  ♂ >24, ♀ >12 (%) 35.3 34.3 34.5 28.9 22.4 18.4 
Men
Women
Q1: 68.9 g/day
(63–73)Q3: 89.0 g/day
(87–91)Q5: 113.5 g/day (109–121)Q1: 73.5 g/day
(69–77)Q3: 90.6 g/day
(89–92)Q5: 109.5 g/day (105–116)
n cases/N 45/1,162 76/1,162 105/1,161 51/1,909 83/1,909 96/1,909 
Characteristics       
 Age (years) 53.0 ± 9.4 53.5 ± 9.0 51.4 ± 7.8 51.1 ± 9.9 53.0 ± 8.9 51.3 ± 8.6 
 Follow-up (years) 11.8 ± 2.6 11.8 ± 2.7 12.2 ± 2.6 11.8 ± 2.1 11.9 ± 2.2 12.3 ± 2.1 
 BMI (kg/m225.6 ± 3.5 26.5 ± 3.5 27.9 ± 3.6 24.2 ± 3.9 25.7 ± 4.4 27.1 ± 4.8 
  <25 (%) 44.6 32.9 20.2 66.4 50.3 36.8 
  25–30 (%) 44.8 52.9 54.0 25.1 33.7 40.4 
  >30 (%) 10.7 14.2 25.8 8.5 16.0 22.7 
 Waist (cm) 92.8 ± 10.1 94.7 ± 9.6 97.9 ± 9.6 77.6 ± 10.0 81.1 ± 10.7 84.5 ± 11.6 
 Family history of diabetes (%)       
  Yes 15.8 17.1 11.7 20.4 22.1 21.4 
 Smoker (%)       
  Never 34.5 30.7 30.2 52.9 57.1 59.8 
  Former 36.8 37.9 33.4 23.5 21.3 20.3 
  Smoker 28.7 31.4 36.4 23.7 21.7 20.0 
 Hypertension (%)       
  Yes 18.1 18.9 18.1 16.9 18.8 18.5 
  Do not know 2.1 3.9 3.9 1.5 0.9 1.0 
 Hyperlipidemia (%)       
  Yes 22.6 20.9 23.2 16.1 14.1 13.8 
  Do not know 9.3 12.9 7.3 6.4 7.1 4.8 
 Educational level (%)       
  Long education (incl. university degree) 28.5 24.5 19.0 23.5 17.7 14.4 
 Physical activity (%)       
  Active 24.2 22.8 27.2 20.0 17.4 13.6 
 Postmenopausal (%) — — — 45.8 51.7 41.1 
Country, n       
 France, ♀ 526 (%) — — — 3.4 5.9 6.8 
 Italy, ♂ 639, ♀1,300 (%) 7.5 13.6 6.7 10.7 16.1 11.4 
 Spain, ♂ 1,333, ♀ 2,154 (%) 2.5 15.5 58.7 3.1 19.1 49.9 
 U.K., ♂ 423, ♀ 661 (%) 7.8 7.6 4.2 6.9 6.0 8.6 
 The Netherlands, ♂ 226, ♀ 1,139 (%) 2.4 5.3 3.0 8.3 15.0 9.8 
 Germany, ♂ 845, ♀ 1,176 (%) 26.7 15.6 4.1 28.7 9.2 1.6 
 Sweden, ♂ 1,231, ♀ 1,625 (%) 40.5 20.2 4.6 31.8 16.1 2.4 
 Denmark, ♂ 1,110, ♀ 964 (%) 12.6 22.3 18.7 7.0 12.5 9.6 
Dietary intake       
 Total energy (kcal/day) 2,564.2 ± 673.6 2,361.0 ± 594.2 2,664.3 ± 674.5 2,061.2 ± 567.5 1,857.7 ± 485.3 2,005.1 ± 525.2 
 Total protein (energy %) 12.8 ± 1.2 16.5 ± 0.6 20.7 ± 1.9 13.2 ± 1.2 17.2 ± 0.8 21.6 ± 2.4 
  Animal protein (energy %) 6.8 ± 1.7 10.2 ± 1.4 14.4 ± 2.5 6.8 ± 1.8 10.6 ± 1.5 15.1 ± 2.8 
   From red meat (g/day)* 6.6 (3–12) 13.7 (8–21) 19.0 (11–28) 6.6 (4–10) 11.9 (8–17) 16.3 (10–23) 
   From processed meat (g/day)* 5.4 (3–8) 5.8 (3–9) 6.2 (3–12) 4.8 (3–7) 5.3 (3–8) 6.0 (4–10) 
   From poultry (g/day)* 2.6 (1–4) 4.8 (3–8) 9.8 (5–16) 2.7 (2–4) 5.1 (3–8) 10.3 (6–16) 
   From milk and dairy (g/day)* 5.3 (1–11) 7.0 (2–14) 7.5 (3–15) 5.4 (2–10) 8.5 (5–14) 10.9 (6–17) 
   From cheese (g/day)* 5.2 (2–9) 6.8 (3–12) 7.0 (3–15) 7.2 (4–10) 9.0 (6–13) 9.5 (5–16) 
   From fish (g/day)* 2.6 (1–5) 5.6 (3–9) 11.8 (6–19) 3.2 (2–5) 5.8 (3–9) 10.3 (6–16) 
   From eggs (g/day)* 1.2 (0–3) 2.0 (1–3) 3.0 (1–5) 1.5 (1–3) 2.2 (1–3) 2.9 (2–4) 
  Plant protein (energy %) 4.4 ± 1.3 5.0 ± 1.3 5.4 ± 1.4 4.7 ± 1.4 5.2 ± 1.3 5.3 ± 1.3 
   From bread (g/day)* 10.0 (7–14) 11.0 (8–15) 11.4 (8–17) 10.4 (8–13) 11.4 (9–15) 11.2 (8–14) 
   From potatoes (g/day)* 2.0 (1–3) 2.2 (1–3) 1.7 (1–3) 1.5 (1–2) 1.6 (1–2) 1.4 (1–2) 
   From pasta and rice (g/day)* 1.2 (0–2) 1.6 (1–3) 1.8 (1–3) 1.5 (1–3) 1.9 (1–3) 1.9 (1–3) 
   From legumes (g/day)* 0.2 (0–1) 0.5 (0–2) 3.1 (0–6) 0.6 (0–1) 1.1 (1–2) 2.1 (1–4) 
 Total carbohydrates (energy %) 46.0 ± 7.7 43.1 ± 6.8 38.7 ± 6.6 48.0 ± 7.2 45.1 ± 6.1 40.9 ± 6.2 
  Starch (energy %) 24.0 ± 6.6 24.6 ± 6.3 23.3 ± 6.1 24.0 ± 6.4 24.1 ± 6.0 22.0 ± 5.8 
  Sugar (energy %) 21.0 ± 7.2 17.8 ± 5.8 15.0 ± 5.3 23.4 ± 7.1 20.5 ± 5.8 18.5 ± 5.9 
 Total fat (energy %) 34.4 ± 7.0 34.1 ± 5.5 35.4 ± 5.6 35.1 ± 6.4 34.9 ± 5.6 35.5 ± 5.9 
  Saturated fat (energy %) 14.1 ± 4.0 13.0 ± 3.4 11.9 ± 3.3  13.3 ± 3.1 12.6 ± 3.3 
  Monounsaturated fat (energy %) 12.3 ± 2.8 12.9 ± 2.9 14.7 ± 3.8 12.5 ± 3.0 13.1 ± 3.6 14.2 ± 4.0 
  Polyunsaturated fat (energy %) 5.5 ± 2.0 5.4 ± 1.8 5.8 ± 2.2 5.7 ± 1.9 5.6 ± 1.7 5.7 ± 2.2 
 Fiber (g)* 20.1 ± 7.0 22.2 ± 6.1 23.9 ± 7.5 23.0 ± 6.6 23.5 ± 5.5 24.1 ± 5.9 
 Cholesterol (mg)* 274.4 ± 118.4 334.9 ± 108.3 422.6 ± 138.0 288.0 ± 93.2 342.2 ± 91.6 405.5 ± 105.9 
 Calcium (mg)* 792.7 ± 286.0 941.9 ± 335.4 1,026.5 ± 501.6 923.2 ± 247.5 1,054.3 ± 265.9 1,188.8 ± 453.7 
 Magnesium (mg)* 383.7 ± 107.8 379.5 ± 98.8 429.3 ± 110.3 330.1 ± 93.6 315.8 ± 86.7 353.8 ± 96.5 
 Vitamin B1 (mg)* 1.4 ± 0.5 1.4 ± 0.5 1.8 ± 0.6 1.2 ± 0.4 1.2 ± 0.4 1.4 ± 0.5 
 β-Carotene (µg)* 2,232.9 ± 2,048.6 2,596.7 ± 2,124.2 2,817.9 ± 2,483.3 3,119.4 ± 2,490.5 3,194.8 ± 2,403.8 3,493.4 ± 2,917.0 
 Vitamin C (mg)* 109.8 ± 64.8 110.7 ± 59.1 140.7 ± 79.8 129.9 ± 79.9 121.9 ± 60.3 141.0 ± 70.8 
 Vitamin D (mg)* 4.6 ± 3.0 4.6 ± 2.9 5.1 ± 3.8 4.1 ± 2.2 4.1 ± 2.1 4.4 ± 2.2 
 Vitamin E (mg)* 12.5 ± 6.0 11.7 ± 5.2 14.7 ± 6.8 11.7 ± 5.4 10.4 ± 4.2 11.9 ± 5.1 
 Soft drinks (g/day)* 37.9 (0–150) 15.9 (0–86) 0.0 (0–29) 13.3 (0–90) 2.4 (0–57) 0.0 (0–28) 
 Coffee (g/day)* 429.8 (190–621) 311.1 (100–611) 130.6 (43–450) 357.1 (120–580) 261.3 (89–500) 160.0 (52–450) 
 Tea (g/day)* 10.3 (0–150) 4.9 (0–146) 0.0 (0–12) 21.4 (0–250) 6.6 (0–238) 0.0 (0–119) 
 Alcohol (g)       
  0 (%) 4.7 4.2 5.4 7.3 9.0 12.9 
  ♂ 0–12, ♀ 0–6 (%) 42.0 39.1 39.6 46.6 53.6 55.4 
  ♂ 12–24, ♀ 6–12 (%) 18.1 22.4 20.4 17.2 15.0 13.4 
  ♂ >24, ♀ >12 (%) 35.3 34.3 34.5 28.9 22.4 18.4 

Data are means ± SD or median (25th percentile–75th percentile) unless otherwise indicated. ♂ men, ♀ women; incl., including; Q, quintile. Family history of diabetes was not collected in Italy, Spain, Heidelberg, or Oxford (missing n = 7,723).

*FFQ-estimated intake energy adjusted by the residual method.

In the subcohort, participants with high intake of total protein (highest vs. lowest quintile) had higher mean BMI and waist circumference and higher intake of MUFAs, dietary fiber, dietary cholesterol, calcium, and β-carotene, whereas educational level and mean intake of carbohydrates, saturated fatty acids (SFAs), coffee, tea, and soft drinks were lower (Table 1). Additionally, with increasing protein intake women were less physically active, drank less alcohol, and were less often smokers, whereas men were more often smokers. With increasing quintiles of total protein intake, the number of type 2 diabetes cases in the subcohort increased (Table 1).

High total protein intake was associated with a 13% higher incidence of type 2 diabetes (HR 1.13 [95% CI 1.08–1.19]) for every 10-g increment after adjustment for energy, center, sex, type 2 diabetes risk factors, and dietary factors (Table 2; Supplementary Fig. 2). Animal protein intake showed comparable results (1.12 [1.07–1.17]). Additional adjustment for waist circumference and BMI attenuated the associations to some extent for total protein to a 6% higher incidence of type 2 diabetes (1.06 [1.02–1.09]) and to 5% for animal protein (1.05 [1.02–1.08]). Analyzing total and animal protein intake by quintile (high vs. low) showed comparable results (1.17 [1.00–1.38], Ptrend < 0.001 and 1.22 [1.06–1.40], Ptrend < 0.001). Between-country heterogeneity was low (I2 0.0–45.3%) (Supplementary Fig. 2).

Table 2

Meta-analyzed (pooled) HRs (95% CI) for the association between protein intake and type 2 diabetes in the EPIC-InterAct case-cohort study

Q1 (low)Q2Q3Q4Q5 (high)PtrendPer 10 g
Total protein N (cases) 5,023 (2,048) 5,080 (2,144) 5,180 (2,265) 5,354 (2,425) 5,616 (2,755)  26,253 (11,637) 
 Median protein* 71.8 82.1 90.0 98.1 111.0  90.0 
 Model 1 1.08 (0.96–1.22) 1.20 (1.02–1.40) 1.40 (1.11–1.76) 1.63 (1.37–1.93) <0.001 1.14 (1.09–1.19) 
 Model 2 1.10 (0.97–1.26) 1.19 (1.01–1.40) 1.38 (1.09–1.75) 1.57 (1.33–1.84) <0.001 1.13 (1.09–1.18) 
 Model 3 1.12 (0.97–1.31) 1.19 (0.99–1.43) 1.39 (1.08–1.79) 1.55 (1.25–1.91) <0.001 1.13 (1.08–1.19) 
 Model 4 1.00 (0.85–1.19) 1.02 (0.86–1.20) 1.13 (0.91–1.42) 1.17 (1.00–1.38) <0.001 1.06 (1.02–1.09) 
Animal protein N (cases) 4,950 (1,972) 5,034 (2,073) 5,207 (2,315) 5,351 (2,436) 5,711 (2,841)  26,253 (11,637) 
 Median protein* 36.0 47.4 55.6 64.3 78.1  55.6 
 Model 1 1.06 (0.95–1.19) 1.22 (1.08–1.37) 1.35 (1.18–1.54) 1.62 (1.35–1.94) <0.0001 1.12 (1.08–1.17) 
 Model 2 1.07 (0.94–1.21) 1.24 (1.08–1.41) 1.35 (1.18–1.54) 1.61 (1.34–1.93) <0.0001 1.12 (1.08–1.16) 
 Model 3 1.05 (0.90–1.22) 1.20 (1.01–1.42) 1.31 (1.11–1.56) 1.51 (1.20–1.91) <0.001 1.12 (1.07–1.17) 
 Model 4 1.01 (0.85–1.20) 1.11 (0.96–1.29) 1.12 (0.99–1.27) 1.22 (1.06–1.40) 0.001 1.05 (1.02–1.08) 
Plant protein N (cases) 5,536 (2,614) 5,243 (2,320) 5,144 (2,205) 5,128 (2,197) 5,202 (2,301)  26,253 (11,637) 
 Median protein* 18.6 23.1 26.2 29.6 35.9  26.2 
 Model 1 0.92 (0.82–1.04) 0.88 (0.78–0.99) 0.93 (0.73–1.19) 0.91 (0.72–1.15) 0.507 0.92 (0.80–1.04) 
 Model 2 0.94 (0.84–1.05) 0.89 (0.80–0.99) 0.94 (0.75–1.18) 0.92 (0.75–1.12) 0.416 0.91 (0.82–1.02) 
 Model 3 1.03 (0.94–1.13) 1.02 (0.92–1.13) 1.15 (0.99–1.34) 1.18 (1.02–1.36) 0.006 1.02 (0.95–1.10) 
 Model 4 1.05 (0.92–1.19) 0.97 (0.83–1.15) 1.09 (0.91–1.31) 1.22 (0.98–1.52) 0.098 1.04 (0.93–1.16) 
Q1 (low)Q2Q3Q4Q5 (high)PtrendPer 10 g
Total protein N (cases) 5,023 (2,048) 5,080 (2,144) 5,180 (2,265) 5,354 (2,425) 5,616 (2,755)  26,253 (11,637) 
 Median protein* 71.8 82.1 90.0 98.1 111.0  90.0 
 Model 1 1.08 (0.96–1.22) 1.20 (1.02–1.40) 1.40 (1.11–1.76) 1.63 (1.37–1.93) <0.001 1.14 (1.09–1.19) 
 Model 2 1.10 (0.97–1.26) 1.19 (1.01–1.40) 1.38 (1.09–1.75) 1.57 (1.33–1.84) <0.001 1.13 (1.09–1.18) 
 Model 3 1.12 (0.97–1.31) 1.19 (0.99–1.43) 1.39 (1.08–1.79) 1.55 (1.25–1.91) <0.001 1.13 (1.08–1.19) 
 Model 4 1.00 (0.85–1.19) 1.02 (0.86–1.20) 1.13 (0.91–1.42) 1.17 (1.00–1.38) <0.001 1.06 (1.02–1.09) 
Animal protein N (cases) 4,950 (1,972) 5,034 (2,073) 5,207 (2,315) 5,351 (2,436) 5,711 (2,841)  26,253 (11,637) 
 Median protein* 36.0 47.4 55.6 64.3 78.1  55.6 
 Model 1 1.06 (0.95–1.19) 1.22 (1.08–1.37) 1.35 (1.18–1.54) 1.62 (1.35–1.94) <0.0001 1.12 (1.08–1.17) 
 Model 2 1.07 (0.94–1.21) 1.24 (1.08–1.41) 1.35 (1.18–1.54) 1.61 (1.34–1.93) <0.0001 1.12 (1.08–1.16) 
 Model 3 1.05 (0.90–1.22) 1.20 (1.01–1.42) 1.31 (1.11–1.56) 1.51 (1.20–1.91) <0.001 1.12 (1.07–1.17) 
 Model 4 1.01 (0.85–1.20) 1.11 (0.96–1.29) 1.12 (0.99–1.27) 1.22 (1.06–1.40) 0.001 1.05 (1.02–1.08) 
Plant protein N (cases) 5,536 (2,614) 5,243 (2,320) 5,144 (2,205) 5,128 (2,197) 5,202 (2,301)  26,253 (11,637) 
 Median protein* 18.6 23.1 26.2 29.6 35.9  26.2 
 Model 1 0.92 (0.82–1.04) 0.88 (0.78–0.99) 0.93 (0.73–1.19) 0.91 (0.72–1.15) 0.507 0.92 (0.80–1.04) 
 Model 2 0.94 (0.84–1.05) 0.89 (0.80–0.99) 0.94 (0.75–1.18) 0.92 (0.75–1.12) 0.416 0.91 (0.82–1.02) 
 Model 3 1.03 (0.94–1.13) 1.02 (0.92–1.13) 1.15 (0.99–1.34) 1.18 (1.02–1.36) 0.006 1.02 (0.95–1.10) 
 Model 4 1.05 (0.92–1.19) 0.97 (0.83–1.15) 1.09 (0.91–1.31) 1.22 (0.98–1.52) 0.098 1.04 (0.93–1.16) 

Model 1 includes age (= time scale) and covariates energy, center, and sex. Model 2, see model 1 plus covariates smoking, education, physical activity, and alcohol. Model 3, see model 2 plus covariates fiber, SFA, MUFA, PUFA, cholesterol, soft drinks, tea, and coffee (not adjusted for carbohydrates; i.e., a substitution model). Model 4, see model 3 plus covariates BMI and waist. Q, quintile.

*FFQ-estimated intake, adjusted for energy by the residual method.

For the association between total and animal protein intake and type 2 diabetes, effect modification by sex (P < 0.001) and by BMI among women (P < 0.001) was present. The association between 10-g increment of total and animal protein intake and type 2 diabetes was confirmed in women (HR 1.10 [95% CI 1.06–1.14] and 1.09 [1.05–1.14], respectively) in model 4 (Fig. 1; Supplementary Table 1). In men, no association was present (both total and animal protein, 1.02 [0.98–1.06]). Compared with overweight women (1.07 [1.01–1.14]) and normal-weight women (1.11 [0.99–1.25]), obese women had a stronger association between animal protein intake and type 2 diabetes (1.19 [1.09–1.32]) (Table 3). In the sensitivity analyses, exclusion of “under-reporters” (n = 8,096) and “over-reporters” (n = 1,206) did not change the overall and sex-specific associations for both total and animal protein intake and type 2 diabetes (data not shown). Excluding the effect of possible lifestyle changes as a result of a medical condition (i.e., baseline self-reported hypertension, hyperlipidemia, myocardial infarction, and stroke, n = 11,043; 3,641 case subjects and 7,402 subcohort) strengthened the associations between both total and animal protein intake and type 2 diabetes in women (1.15 [1.04–1.27] and 1.12 [1.03–1.22], respectively) (Supplementary Table 2). No specific group of protein sources was accountable for the positive association between animal protein and type 2 diabetes; excluding protein from dairy, fish, or meat from total animal protein did not alter the association.

Figure 1

Meta-analyzed HRs of type 2 diabetes associated with 10-g increments of total, animal, and plant protein. Meta-analyzed (pooled country specific) HRs and 95% CI. Protein intake is FFQ-estimated intake, adjusted for energy using the residual method. A: Women. B: Men. The HRs are adjusted for energy, smoking, education, physical activity, alcohol, fiber, SFA, MUFA, PUFA, cholesterol, soft drinks, tea, coffee, BMI, and waist (not adjusted for carbohydrates; i.e., a substitution model).

Figure 1

Meta-analyzed HRs of type 2 diabetes associated with 10-g increments of total, animal, and plant protein. Meta-analyzed (pooled country specific) HRs and 95% CI. Protein intake is FFQ-estimated intake, adjusted for energy using the residual method. A: Women. B: Men. The HRs are adjusted for energy, smoking, education, physical activity, alcohol, fiber, SFA, MUFA, PUFA, cholesterol, soft drinks, tea, coffee, BMI, and waist (not adjusted for carbohydrates; i.e., a substitution model).

Close modal
Table 3

BMI-specific meta-analyzed (pooled) HRs (95% CI) (per 10 g protein intake) for the association between protein intake and type 2 diabetes: results of the adjusted model (model 4)

OverallMenWomen
Animal protein N (cases) 26,253 (11,637) 11,241 (5,798) 15,012 (5,839) 
 BMI (kg/m2   
  <25 1.08 (1.00–1.17); 52.4 1.02 (0.88–1.17); 50.2 1.11 (0.99–1.25); 53.1 
  25–30 1.04 (1.00–1.07); 57.4 1.07 (0.99–1.15); 56.7 1.07 (1.01–1.14); 58.2 
  >30 1.06 (1.00–1.11); 60.7 0.96 (0.91–1.02); 62.1 1.19 (1.09–1.32); 60.0 
Plant protein N (cases) 26,253 (11,637) 11,241 (5,798) 15,012 (5,839) 
 BMI (kg/m2   
  <25 1.04 (0.89–1.23); 25.9 0.97 (0.76–1.24); 26.1 1.15 (0.90–1.46); 25.8 
  25–30 1.06 (0.96–1.17); 26.5 1.13 (0.99–1.29); 26.3 1.01 (0.84–1.21); 26.6 
  >30 1.08 (0.87–1.33); 26.8 1.17 (0.93–1.47); 26.2 1.12 (0.78–1.61); 27.2 
OverallMenWomen
Animal protein N (cases) 26,253 (11,637) 11,241 (5,798) 15,012 (5,839) 
 BMI (kg/m2   
  <25 1.08 (1.00–1.17); 52.4 1.02 (0.88–1.17); 50.2 1.11 (0.99–1.25); 53.1 
  25–30 1.04 (1.00–1.07); 57.4 1.07 (0.99–1.15); 56.7 1.07 (1.01–1.14); 58.2 
  >30 1.06 (1.00–1.11); 60.7 0.96 (0.91–1.02); 62.1 1.19 (1.09–1.32); 60.0 
Plant protein N (cases) 26,253 (11,637) 11,241 (5,798) 15,012 (5,839) 
 BMI (kg/m2   
  <25 1.04 (0.89–1.23); 25.9 0.97 (0.76–1.24); 26.1 1.15 (0.90–1.46); 25.8 
  25–30 1.06 (0.96–1.17); 26.5 1.13 (0.99–1.29); 26.3 1.01 (0.84–1.21); 26.6 
  >30 1.08 (0.87–1.33); 26.8 1.17 (0.93–1.47); 26.2 1.12 (0.78–1.61); 27.2 

Data are HR (95% CI); median protein intake unless otherwise indicated. Median protein intake: FFQ-estimated intake, adjusted for energy using the residual method. Model 4 includes age (= time scale) and covariates energy, center, sex, smoking, education, physical activity, alcohol, fiber, SFA,MUFA, PUFA, cholesterol, soft drinks, tea, coffee, BMI, and waist (not adjusted for carbohydrates; i.e., a substitution model). The pooled HRs per 10 g protein intake did not include risk scores for Dutch men and French women because of the low number of obese subjects in these groups.

Plant protein intake per 10 g was not associated with type 2 diabetes, with an HR of 0.92 (95% CI 0.80–1.04), Ptrend = 0.507, in model 1 and 1.04 (0.93–1.16), Ptrend = 0.098, in model 4 (Table 2; Supplementary Fig. 2) and 1.12 (0.98–1.29) in the sensitivity analysis additionally adjusting for possible lifestyle changes as a result of a medical condition (Supplementary Table 2). No effect modification by sex or BMI was present. Excluding specific groups of plant protein sources did not alter the overall absent association between plant protein and type 2 diabetes.

Our study, the largest of its kind in terms of sample size, number of cases, and follow-up years, is the first to investigate the association between type 2 diabetes incidence and protein intake at a general European level. We found that high total protein at the exchange of carbohydrates is associated with a small elevated risk of type 2 diabetes. This association was largely explained by animal protein intake. BMI and waist circumference attenuated the associations. Plant protein intake was not associated with type 2 diabetes incidence in our cohort.

In this current study, with low heterogeneity between the eight countries, we observed a positive association for total and animal protein and type 2 diabetes risk, independent of known type 2 diabetes risk factors and dietary factors including fat, saturated fat, and fiber intake. We observed that type 2 diabetes incidence was 17% higher in individuals with the highest total protein intake compared with individuals with the lowest intake and that type 2 diabetes incidence increased 6% per 10-g increment of total protein intake at the expense of carbohydrates. A 10-g increment represents ∼50 g meat or fish, a glass of milk, or 50 g nuts. We reviewed the associations with increasing protein intake at the expense of carbohydrates because this is the most suitable source to replace protein, which is reflected by the lower carbohydrate intakes of participants with high-protein intakes. Furthermore, in clinical trials carbohydrates are also the source of choice to replace protein. Analyses of protein intake at the expense of fat intake were comparable with the substitution of carbohydrates. The HRs with increasing intake of protein over the quintiles show a linear dose-response relation. The association between total protein intake and type 2 diabetes appears to be largely explained by animal protein, with a 22% higher type 2 diabetes incidence when comparing highest versus lowest quintile and 5% higher incidence per 10-g increment of animal protein intake. The magnitude of the increased type 2 diabetes risk associated with high total protein intake is comparable with the results of the earlier Dutch cohort study (3), a biomarker-calibrated cohort in the U.S. (4), two small cross-sectional studies in Asian populations (n <150) (8,9), and one small (n = 1,190) Greek population (10).

We observed that type 2 diabetes incidence was 38% higher in women with the highest animal protein intake compared with women with the lowest intake and that type 2 diabetes incidence increased 9% per 10-g increment of animal protein intake. In obese women, the association was even stronger, with a risk increase of 19% per 10-g increment of animal protein intake. In men, only a weak nonsignificant association was present. A difference between men and women has been observed in a prior study, though it as most evident in men (30). In our study, it cannot be explained by differences in total protein intake and/or protein sources. On average, women did have a 20% lower absolute intake of total, animal, and plant protein, but the energy percentage of protein in the diet was equal for men and women. Also, the contribution of protein sources did not differ between sexes. Most dietary and lifestyle factors, associated with protein intake, did not differ substantially between men and women; only women with high protein intake were less likely to be physically active and were more restricted alcohol consumers. Further, the association remained similar after adjustment for BMI and waist in women but was attenuated in men. So, it seems that measures of abdominal obesity largely explain the association of protein intake and type 2 diabetes for men, which is not the case in women. Further research is required to explore why animal protein intake was found to be positively associated with type 2 diabetes risk in women only.

In our study, the association between protein intake and type 2 diabetes was attenuated by measures of body composition, most evident in men. This is in line with earlier research (3,4) and could be explained by the strong independent effect of abdominal obesity on type 2 diabetes risk and the positive correlation of protein intake with overweight and obesity (31). We found that the association between protein intake and type 2 diabetes was strongest in obese women in contrast to prior research declaring weaker associations with increasing BMI (3). Our findings may be explained by the fact that higher total protein intake, and/or higher protein intake from animal sources, is associated with weight and weight gain (31,32). In our data, this effect would be stronger in women than in men. More research is needed to explain these mixed results. Unfortunately, it was not possible to consider weight change as a mediating factor between protein intake and type 2 diabetes incidence because data on weight change in our cohort were ascertained at fixed time intervals after baseline recruitment, so there is the potential for information bias.

In contrast to suggested beneficial short-term effects of dietary protein on glycemic control (5,33), our study found that habitually high intake of protein increases type 2 diabetes risk. This discrepancy between short- and long-term effects of protein intake can be explained by differences in energy content and/or in long-term and acute effects of dietary protein. In energy-restricted diets, high-protein content as a percentage of total energy is found to be beneficial, while absolute protein intake is similar or only modestly increased compared with protein intake in energy balance. We observed that in the general population, in energy balance or positive energy balance, a high absolute protein intake is associated with increased type 2 diabetes risk. The mechanism of the potential harmful effect of high dietary protein intake on type 2 diabetes is largely unknown. It could be driven by high protein sources, such as red or processed meat, and factors associated with protein intake or protein per se, e.g., based on amino acid profiles. Dietary proteins are known to increase glucagon, which could partially explain high blood glucose levels. Dietary proteins also increase insulin secretion, possibly leading to hyperinsulinemia, a risk factor for insulin resistance. A recent study suggested there could be a key role for the plasma amino acid levels of isoleucine, leucine, valine, tyrosine, and phenylalanine in the pathogenesis of type 2 diabetes (34), which have also been found to be associated with type 2 diabetes incidence in EPIC-Potsdam (35). High levels of these plasma amino acids predicted future diabetes, e.g., as found for single plasma amino acids, such as leucine, with an HR of 3.66 (95% 1.61–8.29), and for combinations of the amino acids isoleucine, tyrosine, and phenylalanine, with an HR of 5.99 (2.34–15.34) comparing the highest versus lowest quartile (34). This is in line with earlier experimental elevations of plasma amino acids by infusion, which resulted in impaired insulin-stimulated glucose disposal and insulin-mediated suppression of (hepatic) glucose production (36,37). The above-mentioned branched-chain amino acids and tyrosine and phenylalanine are mainly present in meat and dairy, though they are available in all protein-rich foods.

No specific group of protein sources accounted for the positive association of animal protein and type 2 diabetes incidence. Protein from meat did not explain the association in our cohort, and neither did protein from dairy or fish. So, although the well-established association between meat consumption and type 2 diabetes is suggested to be mainly due to other nutrients, such as iron, nitrites, sodium, or advanced glycation end products (12,38), a direct effect of protein from meat cannot be excluded. In our analyses, protein from dairy and protein from cheese were not accountable for the reported reduced type 2 diabetes risk associated with dairy (1416) and cheese consumption. Fish consumption is not associated with type 2 diabetes (2023,39), so possibly our observed association between animal protein and increased type 2 diabetes risk is counterbalanced by potential risk-reducing nutritional components of fish. The findings of this study did not confirm the suggested reduced type 2 diabetes risk associated with protein intake from plant products (especially legumes [18] and nuts [19]). This could be related to the large proportion of bread, pasta and rice, and potatoes among the plant protein sources and relatively low intake of vegetables, legumes, and nuts, although no indication for a risk-reducing effect of protein from vegetables, legumes, and nuts was present in our analyses. To estimate protein intake, we used uncalibrated results of FFQs, so the intake is not equal to the 24-h protein intake in EPIC reported by Halkjaer et al. (40).

Our study with a large sample size from eight European countries and long follow-up had several strengths. The prospective design, with data collection before the occurrence of type 2 diabetes, and the use of validated FFQs at baseline reduce possible biased recall of diet, although it is possible that diets have changed during follow-up, which could influence the results. The strict validation of diabetes cases reduced the probability of misclassifying noncases as cases. In contrast, it cannot be ruled out that incident and prevalent type 2 diabetes cases have remained undiagnosed, which may lead to an underestimation of main effects and reduced power. Further, we were able to adjust our associations for a wide range of potential risk factors for type 2 diabetes and dietary factors, so the observed positive association between protein intake and type 2 diabetes is likely to be explained by proteins per se. The possibility of unmeasured or residual confounding cannot be ruled out, though. Information on trans-fatty acids was, for example, not available, but in Europe intake in non–margarine using, low–dairy using countries intake was probably low, and in margarine-using diary countries such as the Netherlands, trans-fatty acids correlate with PUFA intake. Because of the observational design, conclusions regarding causality cannot be drawn. The associations could for example relate to a less healthy diet and/or lifestyle, even though total and animal protein intake was not associated with known risk factors such as higher SFAs or lower fiber intake.

Overall, we conclude that a greater intake of total protein intake is associated with a higher type 2 diabetes incidence in European populations, but the effect of protein intake is small and known type 2 diabetes risk factors are also important. Our results show that protein of animal origin is largely responsible for the association—not plant protein. The association is confirmed in women, not in men, and is strongest in obese women. The association cannot be explained by a single food source. In view of the rapidly increasing prevalence of type 2 diabetes, limiting iso-energetic diets high in dietary proteins, particularly from animal sources, should be considered.

Acknowledgments. The authors thank all EPIC participants and staff for their contribution to the study. The authors thank Nicola Kerrison (Medical Research Council Epidemiology Unit, Cambridge) for managing the data for the InterAct project.

Funding. Funding for the InterAct project was provided by the EU FP6 programme (grant no. LSHM_CT_2006_037197). In addition, InterAct investigators acknowledge funding from the following agencies: the Alpro Foundation; Dutch Ministry of Health, Welfare and Sports; Netherlands Cancer Registry; LK Research Funds; Dutch Prevention Funds; Dutch ZON (Zorg Onderzoek Nederland); World Cancer Research Fund; Statistics Netherlands; Cancer Research UK; Medical Research Council U.K.; Swedish Research Council; Swedish Heart-Lung Foundation; Swedish Diabetes Association; Danish Cancer Society; Deutsche Krebshilfe; German Cancer Aid; German Cancer Research Center; Federal Ministry of Education and Research; Associazione Italiana per la Ricerca sul Cancro-Italy; Asturias Regional Government; Health Research Fund of the Spanish Ministry of Health; CIBER en Epidemiología y Salud Pública, Spain; Murcia Regional Government (no. 6236); AIRE-ONLUS Ragusa; AVIS Ragusa; Sicilian Regional Government; Innovation and the Health Research Fund of the Spanish Ministry of Health (RTICC DR06/0020/0091); and Basque Country Regional Government. Verification of diabetes cases was additionally funded by the Netherlands Agency (grant no. IGE05012) and an incentive grant from the Board of the Utrecht Medical Centre, Utrecht.

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Duality of Interest. InterAct investigators also acknowledge funding from Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. M.v.N. researched data, wrote the manuscript, contributed to study conception and design and interpretation of data, revised the article critically for important intellectual content, and gave final approval of the version to be published. E.J.M.F., M.M., I.S., E.M., C.L., S.J.S., N.G.F., and N.J.W. contributed to discussion, reviewed and edited the manuscript, contributed to study conception and design and interpretation of data, revised the article critically for important intellectual content, and gave final approval of the version to be published. P.A., E.A., B.B., J.W.J.B., H.B., F.C.-C., G.F., P.W.F., J.H., J.M.H., V.Ka., T.J.K., K.T.K., V.Kr., T.K., V.V.M.M., P.N., K.O., D.P., S.P., O.R., I.R., C.S., M.-J.S., M.B.S., A.M.W.S., A.T., R.T., D.L.v.d.A., A.M.L.W., R.Z.-R., and E.R. reviewed and edited the manuscript, contributed to study conception and design and interpretation of data, revised the article critically for important intellectual content, and gave final approval of the version to be published. M.v.N., E.J.M.F., and M.M. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 20th International Congress of Nutrition, Granada, Spain, 15–20 September 2013.

1.
Westerterp-Plantenga
MS
,
Nieuwenhuizen
A
,
Tomé
D
,
Soenen
S
,
Westerterp
KR
.
Dietary protein, weight loss, and weight maintenance
.
Annu Rev Nutr
2009
;
29
:
21
41
[PubMed]
2.
Weigle
DS
,
Breen
PA
,
Matthys
CC
, et al
.
A high-protein diet induces sustained reductions in appetite, ad libitum caloric intake, and body weight despite compensatory changes in diurnal plasma leptin and ghrelin concentrations
.
Am J Clin Nutr
2005
;
82
:
41
48
[PubMed]
3.
Sluijs
I
,
Beulens
JWJ
,
van der A
DL
,
Spijkerman
AM
,
Grobbee
DE
,
van der Schouw
YT
.
Dietary intake of total, animal, and vegetable protein and risk of type 2 diabetes in the European Prospective Investigation into Cancer and Nutrition (EPIC)-NL study
.
Diabetes Care
2010
;
33
:
43
48
[PubMed]
4.
Tinker
LF
,
Sarto
GE
,
Howard
BV
, et al
.
Biomarker-calibrated dietary energy and protein intake associations with diabetes risk among postmenopausal women from the Women’s Health Initiative
.
Am J Clin Nutr
2011
;
94
:
1600
1606
[PubMed]
5.
Promintzer
M
,
Krebs
M
.
Effects of dietary protein on glucose homeostasis
.
Curr Opin Clin Nutr Metab Care
2006
;
9
:
463
468
[PubMed]
6.
Weickert
MO
,
Roden
M
,
Isken
F
, et al
.
Effects of supplemented isoenergetic diets differing in cereal fiber and protein content on insulin sensitivity in overweight humans
.
Am J Clin Nutr
2011
;
94
:
459
471
[PubMed]
7.
Linn
T
,
Santosa
B
,
Grönemeyer
D
, et al
.
Effect of long-term dietary protein intake on glucose metabolism in humans
.
Diabetologia
2000
;
43
:
1257
1265
[PubMed]
8.
Duc Son
NT
,
Hanh
TTM
,
Kusama
K
, et al
.
Anthropometric characteristics, dietary patterns and risk of type 2 diabetes mellitus in Vietnam
.
J Am Coll Nutr
2005
;
24
:
229
234
[PubMed]
9.
Wang
ET
,
de Koning
L
,
Kanaya
AM
.
Higher protein intake is associated with diabetes risk in South Asian Indians: the Metabolic Syndrome and Atherosclerosis in South Asians Living in America (MASALA) study
.
J Am Coll Nutr
2010
;
29
:
130
135
[PubMed]
10.
Pounis
GD
,
Tyrovolas
S
,
Antonopoulou
M
, et al
.
Long-term animal-protein consumption is associated with an increased prevalence of diabetes among the elderly: the Mediterranean Islands (MEDIS) study
.
Diabetes Metab
2010
;
36
:
484
490
[PubMed]
11.
Aune
D
,
Ursin
G
,
Veierød
MB
.
Meat consumption and the risk of type 2 diabetes: a systematic review and meta-analysis of cohort studies
.
Diabetologia
2009
;
52
:
2277
2287
[PubMed]
12.
Pan
A
,
Sun
Q
,
Bernstein
AM
, et al
.
Red meat consumption and risk of type 2 diabetes: 3 cohorts of US adults and an updated meta-analysis
.
Am J Clin Nutr
2011
;
94
:
1088
1096
[PubMed]
13.
InterAct Consortium
.
Association between dietary meat consumption and incident type 2 diabetes: the EPIC-InterAct study
.
Diabetologia
2013
;
56
:
47
59
[PubMed]
14.
Elwood
PC
,
Pickering
JE
,
Givens
DI
,
Gallacher
JE
.
The consumption of milk and dairy foods and the incidence of vascular disease and diabetes: an overview of the evidence
.
Lipids
2010
;
45
:
925
939
[PubMed]
15.
Malik
VS
,
Sun
Q
,
van Dam
RM
, et al
.
Adolescent dairy product consumption and risk of type 2 diabetes in middle-aged women
.
Am J Clin Nutr
2011
;
94
:
854
861
[PubMed]
16.
Fumeron
F
,
Lamri
A
,
Abi Khalil
C
, et al
Data from the Epidemiological Study on the Insulin Resistance Syndrome (DESIR) Study Group
.
Dairy consumption and the incidence of hyperglycemia and the metabolic syndrome: results from a french prospective study, Data from the Epidemiological Study on the Insulin Resistance Syndrome (DESIR)
.
Diabetes Care
2011
;
34
:
813
817
[PubMed]
17.
Tong X, Dong JY, Wu ZW, et al. Dairy consumption and risk of type 2 diabetes mellitus: a meta-analysis of cohort studies. Eur J Clin Nutr 2011;65:1027–1031
18.
Villegas
R
,
Gao
Y-T
,
Yang
G
, et al
.
Legume and soy food intake and the incidence of type 2 diabetes in the Shanghai Women’s Health Study
.
Am J Clin Nutr
2008
;
87
:
162
167
[PubMed]
19.
Kendall
CW
,
Josse
AR
,
Esfahani
A
,
Jenkins
DJ
.
Nuts, metabolic syndrome and diabetes
.
Br J Nutr
2010
;
104
:
465
473
[PubMed]
20.
Kaushik
M
,
Mozaffarian
D
,
Spiegelman
D
,
Manson
JE
,
Willett
WC
,
Hu
FB
.
Long-chain omega-3 fatty acids, fish intake, and the risk of type 2 diabetes mellitus
.
Am J Clin Nutr
2009
;
90
:
613
620
[PubMed]
21.
van Woudenbergh
GJ
,
van Ballegooijen
AJ
,
Kuijsten
A
, et al
.
Eating fish and risk of type 2 diabetes: A population-based, prospective follow-up study
.
Diabetes Care
2009
;
32
:
2021
2026
[PubMed]
22.
Patel
PS
,
Sharp
SJ
,
Luben
RN
, et al
.
Association between type of dietary fish and seafood intake and the risk of incident type 2 diabetes: the European prospective investigation of cancer (EPIC)-Norfolk cohort study
.
Diabetes Care
2009
;
32
:
1857
1863
[PubMed]
23.
Patel
PS
,
Forouhi
NG
,
Kuijsten
A
, et al
InterAct Consortium
.
The prospective association between total and type of fish intake and type 2 diabetes in 8 European countries: EPIC-InterAct Study
.
Am J Clin Nutr
2012
;
95
:
1445
1453
[PubMed]
24.
Langenberg
C
,
Sharp
S
,
Forouhi
NG
, et al
InterAct Consortium
.
Design and cohort description of the InterAct Project: an examination of the interaction of genetic and lifestyle factors on the incidence of type 2 diabetes in the EPIC Study
.
Diabetologia
2011
;
54
:
2272
2282
[PubMed]
25.
Riboli
E
,
Hunt
KJ
,
Slimani
N
, et al
.
European Prospective Investigation into Cancer and Nutrition (EPIC): study populations and data collection
.
Public Health Nutr
2002
;
5
(
6B
):
1113
1124
[PubMed]
26.
Willett
WC
,
Howe
GR
,
Kushi
LH
.
Adjustment for total energy intake in epidemiologic studies
.
Am J Clin Nutr
1997
;
65
(
Suppl.
):
1220S
1228S; discussion 1229S–1231S
[PubMed]
27.
Wareham
NJ
,
Jakes
RW
,
Rennie
KL
, et al
.
Validity and repeatability of a simple index derived from the short physical activity questionnaire used in the European Prospective Investigation into Cancer and Nutrition (EPIC) study
.
Public Health Nutr
2003
;
6
:
407
413
[PubMed]
28.
Barlow
WE
,
Ichikawa
L
,
Rosner
D
,
Izumi
S
.
Analysis of case-cohort designs
.
J Clin Epidemiol
1999
;
52
:
1165
1172
[PubMed]
29.
Goldberg
GR
,
Black
AE
,
Jebb
SA
, et al
.
Critical evaluation of energy intake data using fundamental principles of energy physiology: 1. Derivation of cut-off limits to identify under-recording
.
Eur J Clin Nutr
1991
;
45
:
569
581
[PubMed]
30.
Schulze
MB
,
Schulz
M
,
Heidemann
C
,
Schienkiewitz
A
,
Hoffmann
K
,
Boeing
H
.
Carbohydrate intake and incidence of type 2 diabetes in the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam Study
.
Br J Nutr
2008
;
99
:
1107
1116
[PubMed]
31.
Vergnaud
A-C
,
Norat
T
,
Mouw
T
, et al
.
Macronutrient composition of the diet and prospective weight change in participants of the EPIC-PANACEA study
.
PLoS ONE
2013
;
8
:
e57300
[PubMed]
32.
Halkjar J, Olsen A, Overvad K, et al. Intake of total, animal and plant protein and subsequent changes in weight or waist circumference in European men and women: the Diogenes project. Int J Obes (London) 2011;35:1104–1113.
33.
Tremblay
F
,
Lavigne
C
,
Jacques
H
,
Marette
A
.
Role of dietary proteins and amino acids in the pathogenesis of insulin resistance
.
Annu Rev Nutr
2007
;
27
:
293
310
[PubMed]
34.
Wang
TJ
,
Larson
MG
,
Vasan
RS
, et al
.
Metabolite profiles and the risk of developing diabetes
.
Nat Med
2011
;
17
:
448
453
[PubMed]
35.
Floegel
A
,
Stefan
N
,
Yu
Z
, et al
.
Identification of serum metabolites associated with risk of type 2 diabetes using a targeted metabolomic approach
.
Diabetes
2013
;
62
:
639
648
[PubMed]
36.
Krebs
M
,
Krssak
M
,
Bernroider
E
, et al
.
Mechanism of amino acid-induced skeletal muscle insulin resistance in humans
.
Diabetes
2002
;
51
:
599
605
[PubMed]
37.
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
2684
[PubMed]
38.
de Oliveira Otto
MC
,
Alonso
A
,
Lee
D-H
, et al
.
Dietary intakes of zinc and heme iron from red meat, but not from other sources, are associated with greater risk of metabolic syndrome and cardiovascular disease
.
J Nutr
2012
;
142
:
526
533
[PubMed]
39.
Nanri
A
,
Mizoue
T
,
Noda
M
, et al
Japan Public Health Center-based Prospective Study Group
.
Fish intake and type 2 diabetes in Japanese men and women: the Japan Public Health Center-based Prospective Study
.
Am J Clin Nutr
2011
;
94
:
884
891
[PubMed]
40.
Halkjaer
J
,
Olsen
A
,
Bjerregaard
LJ
, et al
.
Intake of total, animal and plant proteins, and their food sources in 10 countries in the European Prospective Investigation into Cancer and Nutrition
.
Eur J Clin Nutr
2009
;
63
(
Suppl. 4
):
S16
S36
[PubMed]

Supplementary data