Advertisement

Dietary protein intake and subsequent risk of type 2 diabetes: a dose–response meta-analysis of prospective cohort studies

  • Jianhong Ye
  • Qixin Yu
  • Weihua Mai
  • Peiling Liang
  • Xiaoxia Liu
  • Yunnan WangEmail author
Open Access
Original Article

Abstract

Aims

Dietary proteins, including those obtained from animal and plant sources, have inconsistently been correlated with type 2 diabetes mellitus (T2DM) risk. Therefore, a meta-analysis was conducted to evaluate the association between dietary proteins and the risk of T2DM.

Methods

Prospective cohort studies published until November 2018 were systematically searched in PubMed, Embase, and the Cochrane library. The pooled relative risks (RRs) were calculated with 95% confidence intervals (CIs) using the random-effects model.

Results

Ten articles involving a total of 21 cohorts were included in the final meta-analysis. A total of 487,956 individuals were recruited in these studies and 38,350 T2DM cases were reported. Analysis of the pooled RRs indicated that high total protein intake was associated with an increased risk of T2DM (RR 1.10; P = 0.006), whereas moderate total protein intake was not significantly associated with T2DM risk (RR 1.00; P = 0.917). Moreover, a higher risk of T2DM was observed with high animal protein intake (RR 1.13; P = 0.013), whereas moderate animal protein intake had little or no effect on T2DM risk (RR 1.06; P = 0.058). Finally, high intake of plant protein did not affect T2DM risk (RR 0.93; P = 0.074), whereas moderate intake was associated with a reduced risk of T2DM (RR 0.94; P < 0.001).

Conclusions

The results of this study indicate that high total protein and animal protein intakes are associated with an increased risk of T2DM, whereas moderate plant protein intake is associated with a decreased risk of T2DM.

Keywords

Dietary proteins Energy intake Type 2 diabetes mellitus Meta-analysis 

Introduction

Type 2 diabetes mellitus (T2DM) is an important public health problem. In the USA, approximately 2.9 million individuals have diabetes and 90% of them are patients with T2DM [1]. These patients have an increased risk of atherosclerosis, cardiovascular diseases, chronic kidney disease, and cancer, all of which can reduce the life expectancy by nearly 10 years [2]. Many individuals have intermediate hyperglycemia, which can progress to diabetes [3, 4, 5]. Therefore, an effective preventive strategy should be explored by focusing on individual characteristics. Studies have already identified several risk factors for T2DM, such as adiposity, low hip circumference, certain serum biomarkers, unhealthy dietary patterns, low educational and conscientiousness levels, decreased physical activity, high sedentary time and duration of watching television, alcohol intake, smoking, air pollution, and several medical conditions [6].

Several studies have focused on the association between dietary protein intake and T2DM risk, but data are limited and remain inconclusive. Short-term trial results indicated that high dietary protein intake could improve glucose homeostasis [7, 8]. Nevertheless, T2DM risk seems to depend on the type of dietary protein, as plant proteins appear to exhibit a protective effect against T2DM [9]. One study showed that high intake of total and animal protein was associated with an increased risk of T2DM, whereas T2DM risk was reduced in individuals with high plant protein intake [10]. Nonetheless, the effect of moderate dietary protein intake on T2DM risk has not been investigated in previous studies. Definition of an optimal intake of dietary protein to reduce the risk of T2DM is particularly important for the general population. Therefore, this study aimed to evaluate the relationship between dietary protein intake and T2DM risk based on a large-scale examination of prospective cohort studies.

Methods

Data sources, search strategy, and selection criteria

This review was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement issued in 2009 [11]. Prospective cohort studies investigating the association between dietary protein intake and risk of T2DM were included in our study, without restrictions of publication language and status. We systematically searched for relevant studies published until November 2018 in the PubMed, Embase, and Cochrane library databases using the following search terms: “dietary protein” OR “protein intake” OR “plant protein” OR “animal protein” OR “food” AND “diabetes” OR “diabetes mellitus” OR “T2DM” AND “prospective”. A manual search in the reference lists from relevant original and review articles was also conducted to identify further studies.

The literature search and study selection were conducted by two reviewers according to a standard approach, and any disagreement was resolved by the primary author. Studies were included if they: (1) had a prospective cohort design, (2) investigated total, animal, and plant protein intakes, (3) defined T2DM incidence as the outcome, and (4) reported the effect estimates [risk ratio (RR), hazard ratio (HR), or odds ratio (OR)] and 95% confidence intervals (CIs) to compare high or moderate protein intake with the lowest protein intake relative to the risk of T2DM. Studies with a retrospective observational design, such as traditional case–control and retrospective cohort studies were excluded due to various confounding factors that could result in bias.

Data collection and quality assessment

The collected data included the first author’s name, publication year, study period, country, study cohort, sample size, number of T2DM cases, age range, percentage of male subjects, diet assessment, follow-up duration, adjusted factors, and effect estimate with corresponding 95%CIs. The effect estimate was selected as the maximally adjusted potential confounder if the study reported several multivariable adjusted effect estimates. The quality of retrieved studies was evaluated using the Newcastle-Ottawa Scale (NOS), based on the selection (4 items: 4 stars), comparability (1 item: 2 stars), and outcome (3 items: 3 stars) data [12]. A “star system” rating the study quality ranged from 0 to 9, and studies with 8 or 9 stars were regarded as good quality. Data collection and quality assessment were carried out by two reviewers, and any inconsistencies were resolved by an additional author by referring to the original studies.

Statistical analysis

The association between dietary protein intake and T2DM risk was evaluated based on the effect estimate and its 95% CI in each study. The high protein intake category was defined as the highest protein intake category in each study, and the moderate protein intake categories were pooled using a fixed-effect model if ≥ 2 categories were reported [13]. Then, the pooled RRs with their 95% CIs were calculated using the random-effects model [14]. I2 and Q-statistic were used to evaluate the heterogeneity among included studies, and P < 0.010 was considered significant heterogeneity [15]. Sensitivity analyses were conducted to evaluate the stability of pooled results by excluding individual studies [16]. Meta-regression and subgroup analyses were performed to explore the potential impact factors and evaluate the relationship between dietary protein intake and T2DM risk in patients with specific characteristics [17, 18]. Publication biases were assessed using Funnel plots, and Egger’s [19] and Begg’s [20] tests. All P values obtained for pooled results were two-sided, and P values of < 0.05 were considered statistically significant. Statistical analyses were performed using the STATA software (version 10.0; Stata Corporation, College Station, TX, USA).

Results

Literature search

The electronic search identified 2542 records in PubMed, Embase, and the Cochrane library (Fig. 1). Among the identified studies, 2497 were excluded due to irrelevant topics and duplicate titles. The remaining 45 studies were retrieved for further evaluation, and 35 further studies were excluded because they reported an assessment other than dietary intake (n = 21), had a retrospective design (n = 8), or were reviews or meta-analyses (n = 6). Finally, ten prospective cohort studies that met the inclusion criteria were included in the final quantitative meta-analysis [21, 22, 23, 24, 25, 26, 27, 28, 29, 30]. The manual search of the reference list in retrieved studies did not identify any additional suitable studies.

Fig. 1

Flow diagram of the literature search and study selection processes

Study characteristics

The 10 identified studies included 21 cohorts with a total of 487,956 recruited participants and 38,350 reported T2DM cases. The follow-up duration was 5–24 years, and 2,332–74,155 individuals were included from individual studies. Three studies were conducted in the USA, 4 in Europe, 2 in Australia, and 1 in Asia (Japan). The study quality was evaluated using NOS; 4 studies had nine stars, 4 had eight stars, and the remaining 2 had seven stars. Table 1 summarizes the characteristics of the included studies.

Table 1

Baseline characteristics of included studies

Study

Publication year

Study period

Country

Study cohort

Sample size

T2DM cases

Age range (years)

Percentage male (%)

Diet exposure assessment

Follow-up (year)

Adjusted factors

NOS score

Song [21]

2004

1993

US

WHS

37,309

1558

≥ 45

0

FFQ

8.8

Age, BMI, TEI, smoking, PA, alcohol, FHD, dietary fiber intake, GL, magnesium, and TFI

9

Sluijs [22]

2010

1993–1997

The Netherlands

EPIC-NL

38,094

918

49–70

25.6

FFQ

10.1

Sex, age, energy-adjusted intake of saturated fat, monounsaturated fat, polyunsaturated fat, cholesterol, vitamin E, magnesium, fiber, and GL, energy-adjusted alcohol consumption, PA, mean SBP and DBP, education, FHD, BMI, and waist circumference

9

Tinker [23]

2011

1993–1998

US

WHI

74,155

3319

50–79

0

FFQ

8.1

Age, hormone therapy, PA, ethnicity, education, income, history of cardiovascular disease, smoking status, alcohol consumption, hypertension, FHD, hormone use, glycemic index, GL, and BMI

9

Ericson [24]

2013

1991–1996

Sweden

MDC

27,140

1709

45–74

38.9

FFQ

12.0

Age, method version, season, TEI, education, smoking, alcohol intake, leisure-time PA and BMI

9

Alhazmi [25]

2013

2002–2006

Australia

ALSWH

8370

311

45–50

0

FFQ

6.0

Area of residence, education, current smoking status, PA, self-rated health as good, menopausal status, BMI, alcohol consumption, TEI, and SFA and MUFA intakes for total carbohydrate; SFA, MUFA and fiber intakes for total protein; and fiber intake for TFI

8

van Nielen [26]

2014

1991–1998

France, Italy, Spain, UK, The Netherlands, Germany, Sweden, and Denmark

EPIC-InterAct

28,557

12,403

20–78

49.7

FFQ

12.0

Age, energy, center, sex, smoking, education, PA, alcohol, fiber, SFA, MUFA, PUFA, cholesterol, soft drinks, tea and coffee, BMI, and waist circumference

8

Nanri [27]

2015

1990, 1993

Japan

JPHC

64,674

1191

45–75

43.0

FFQ

5.0

Age, study area, BMI, smoking status, alcohol consumption, FHD, PA, history of hypertension, TEI, coffee, fat, and carbohydrate

7

Malik [28]

2016

1984

US

NHS

72,992

7214

38–63

0.0

FFQ

24.0

FHD, smoking, alcohol intake, PA, race/ethnicity, TEI, postmenopausal hormone use, percentages of energy from trans fat, SFA, MUFA, and PUFA, dietary cholesterol, dietary fiber, glycemic index, and BMI

8

Malik [28]

2016

1991

US

NHS II

72,088

5032

24–42

0.0

FFQ

18.0

FHD, smoking, alcohol intake, PA, race/ethnicity, TEI, postmenopausal hormone use, percentages of energy from trans fat, SFA, MUFA, and PUFA, dietary cholesterol, dietary fiber, glycemic index, and BMI

8

Malik [28]

2016

1986

US

HPFS

40,722

3334

40–75

100.0

FFQ

22.0

FHD, smoking, alcohol intake, PA, race/ethnicity, TEI, postmenopausal hormone use, percentages of energy from trans fat, SFA, MUFA, and PUFA, dietary cholesterol, dietary fiber, glycemic index, and BMI

8

Shang [29]

2016

1990–1994

Australia

MCCS

21,523

929

31–76

38.3

FFQ

13.0

Age, sex, ethnicity, socioeconomic status, PA, smoking, alcohol intake, glycemic index, consumption of energy, fiber, SFA, MUFA, PUFA and trans fat; and baseline plasma glucose, blood pressure, and BMI

8

Virtanen [30]

2017

1984–1989

Finland

KIHD

2332

432

42–60

100.0

FFQ

19.3

Age, examination year, TEI, marital status, income, use of hypertension medication, FHD, pack-years of smoking, education, leisure-time PA, serum ferritin, alcohol intake, glycaemic index, and dietary intakes of fiber, Mg, coffee, cholesterol, and SFA, MUFA, PUFA and trans-fatty acids, BMI, fasting plasma glucose and fasting serum insulin

7

ADA American Diabetes Association, ALSWH the Australian Longitudinal Study on Women’s Health, BMI body mass index, DBP diastolic blood pressure, EPIC-NL the Prospect-EPIC and MORGENEPIC cohorts, FFQ food-frequency questionnaire, FHD family history of diabetes, GI glycemic load, HPFS Health Professionals Follow-Up Study, JPHC Japan Public Health Center, KIHD The Kuopio Ischaemic Heart Disease Risk Factor Study, MCCS Melbourne Collaborative Cohort Study, MDC Malmo Diet and Cancer, NHS Nurses’ Health Study, PA physical activity, SBP systolic blood pressure, T2DM type 2 diabetes mellitus, TEI total energy intake, TFI total fat intake, WHI Women’s Health Initiative, WHS Women’s Health Study

Total protein intake

A total of ten studies reported an association between high total protein intake and subsequent T2DM risk. The pooled RRs indicated that high total protein intake was associated with an increased T2DM risk (RR 1.10; 95% CI 1.03–1.17; P = 0.006; Fig. 2), with significant heterogeneity across the studies (I2 48.8%; P = 0.020). The results of the sensitivity analysis indicated that the pooled results were stable and were not altered by sequentially excluding any of the studies (Supplemental file 1). Meta-regression analysis indicated that country (P = 0.014) and NOS (P = 0.002) might affect the relationship between high total protein intake and T2DM. Subgroup analyses revealed that high total protein intake was significantly associated with T2DM risk in studies conducted in the USA (RR 1.11; 95% CI 1.05–1.17; P < 0.001) and Europe (RR 1.18; 95% CI 1.07–1.31; P = 0.001), in those that included men (RR 1.14; 95% CI 1.03–1.26; P = 0.008) or both men and women (RR 1.19; 95% CI 1.05–1.34; P = 0.005), in those with a follow-up duration of ≥ 10.0 years (RR 1.14; 95% CI 1.07–1.21; P < 0.001), in those not adjusted for family history of DM (RR 1.19; 95% CI 1.08–1.31; P = 0.001), and in those with a high quality (RR 1.12; 95% CI 1.07–1.17; P < 0.001, Table 2). No publication showed a significant relationship between high total protein intake and T2DM risk (P value for Egger’s test: 0.429; P value for Begg’s test: 0.274; Supplemental file 2).

Fig. 2

Association between high total protein intake and the risk of type 2 diabetes mellitus

Table 2

Subgroup analyses on the risk of type 2 diabetes mellitus

Factor

Groups

RR and 95% CI

P value

Heterogeneity (%)

P value for heterogeneity

P value for meta-regression

High versus low total protein intake

Country

     

 Asia

0.80 (0.51–1.26)

0.330

80.6

0.023

0.014

 Australia

1.07 (0.78–1.48)

0.676

55.1

0.136

 US

1.11 (1.05–1.17)

< 0.001

19.3

0.292

 Europe

1.18 (1.07–1.31)

0.001

0.0

0.643

Gender

     

 Men

1.14 (1.03–1.26)

0.008

0.0

0.409

0.269

 Women

1.05 (0.95–1.17)

0.346

69.7

0.003

 Both

1.19 (1.05–1.34)

0.005

0.0

0.944

Follow-up duration (years)

     

 ≥ 10.0

1.14 (1.07–1.21)

< 0.001

0.0

0.444

0.207

 < 10.0

1.01 (0.89–1.15)

0.823

70.5

0.005

Adjusted family history of diabetes mellitus

     

 Yes

1.06 (0.98–1.16)

0.145

59.1

0.012

0.109

 No

1.19 (1.08–1.31)

0.001

0.0

0.510

NOS

     

 8 or 9

1.12 (1.07–1.17)

< 0.001

0.0

0.441

0.002

 7

0.83 (0.62–1.12)

0.219

62.8

0.068

Moderate versus low total protein intake

Country

     

 Asia

0.94 (0.67–1.33)

0.736

89.9

0.002

0.206

 Australia

0.92 (0.76–1.11)

0.395

47.3

0.168

 US

0.99 (0.91–1.09)

0.894

88.7

< 0.001

 Europe

1.06 (0.99–1.13)

0.080

9.0

0.355

Gender

     

 Men

0.99 (0.89–1.10)

0.825

64.5

0.038

0.047

 Women

0.99 (0.90–1.09)

0.848

84.2

< 0.001

 Both

1.04 (0.96–1.12)

0.324

0.0

0.580

Follow-up duration (years)

     

 ≥ 10.0

1.00 (0.94–1.07)

0.970

62.8

0.009

0.003

 < 10.0

1.00 (0.90–1.11)

0.972

79.7

0.001

Adjusted family history of diabetes mellitus

     

 Yes

1.00 (0.93–1.08)

0.969

81.9

< 0.001

0.469

 No

1.01 (0.93–1.11)

0.753

51.4

0.083

NOS

     

 8 or 9

1.01 (0.95–1.07)

0.759

75.5

< 0.001

0.591

 7

0.97 (0.78–1.22)

0.807

80.6

0.006

High versus low animal protein intake

Country

     

 Asia

0.82 (0.51–1.30)

0.395

82.2

0.018

0.013

 Australia

1.29 (0.99–1.68)

0.056

 US

1.19 (1.06–1.33)

0.002

64.1

0.039

 Europe

1.19 (1.06–1.33)

0.004

0.0

0.707

Gender

     

 Men

1.17 (1.01–1.36)

0.032

20.9

0.283

0.164

 Women

1.06 (0.87–1.29)

0.585

84.2

< 0.001

 Both

1.22 (1.09–1.36)

0.001

0.0

0.802

Follow-up duration (years)

     

 ≥ 10.0

1.19 (1.11–1.27)

< 0.001

0.0

0.672

0.066

 < 10.0

1.03 (0.80–1.32)

0.844

84.2

< 0.001

Adjusted family history of diabetes mellitus

     

 Yes

1.10 (0.98–1.24)

0.120

70.0

0.002

0.149

 No

1.24 (1.09–1.40)

0.001

0.0

0.712

NOS

     

 8 or 9

1.19 (1.10–1.28)

< 0.001

37.0

0.146

0.002

 7

0.88 (0.63–1.21)

0.427

69.5

0.038

Moderate versus low animal protein intake

Country

     

 Asia

0.98 (0.76–1.27)

0.900

83.2

0.015

0.549

 Australia

1.05 (0.91–1.21)

0.502

 US

1.07 (0.97–1.18)

0.154

87.3

< 0.001

 Europe

1.09 (1.01–1.17)

0.019

0.0

0.837

Gender

     

 Men

1.05 (0.95–1.15)

0.329

37.0

0.205

0.605

 Women

1.05 (0.93–1.19)

0.412

89.1

< 0.001

 Both

1.07 (1.01–1.15)

0.031

0.0

0.863

Follow-up duration (years)

     

 ≥ 10.0

1.03 (0.99–1.07)

0.128

0.0

0.530

0.267

 < 10.0

1.08 (0.92–1.26)

0.343

88.8

< 0.001

Adjusted family history of diabetes mellitus

     

 Yes

1.06 (0.98–1.14)

0.144

77.4

< 0.001

0.344

 No

1.08 (1.01–1.16)

0.036

0.0

0.656

NOS

     

 8 or 9

1.07 (1.00–1.14)

0.044

75.6

< 0.001

0.583

 7

1.03 (0.85–1.24)

0.788

72.0

0.028

High versus low plant protein intake

Country

     

 Asia

0.80 (0.66–0.96)

0.017

0.0

0.896

0.040

 Australia

1.02 (0.73–1.42)

0.907

 US

0.91 (0.84–0.97)

0.008

0.0

0.999

 Europe

1.06 (0.81–1.38)

0.674

58.5

0.090

Gender

     

 Men

0.85 (0.75–0.97)

0.018

0.0

0.438

0.006

 Women

0.90 (0.83–0.97)

0.007

0.0

0.903

 Both

1.15 (0.99–1.34)

0.063

0.0

0.678

Follow-up duration (years)

     

 ≥ 10.0

0.98 (0.87–1.12)

0.807

47.5

0.090

0.146

 < 10.0

0.88 (0.81–0.96)

0.005

0.0

0.680

Adjusted family history of diabetes mellitus

     

 Yes

0.90 (0.84–0.96)

0.001

0.0

0.532

0.012

 No

1.16 (0.96–1.39)

0.122

0.0

0.379

NOS

     

 8 or 9

0.96 (0.89–1.05)

0.370

31.3

0.189

0.041

 7

0.78 (0.66–0.93)

0.005

0.0

0.881

Moderate versus low plant protein intake

Country

     

 Asia

0.95 (0.86–1.06)

0.350

0.0

0.632

0.073

 Australia

0.99 (0.85–1.15)

0.896

 US

0.93 (0.90–0.96)

< 0.001

0.0

0.900

 Europe

1.02 (0.96–1.10)

0.494

0.0

0.819

Gender

     

 Men

0.92 (0.87–0.98)

0.011

0.0

0.932

0.032

 Women

0.93 (0.90–0.96)

< 0.001

0.0

0.823

 Both

1.02 (0.96–1.09)

0.482

0.0

0.881

Follow-up duration (years)

     

 ≥ 10.0

0.96 (0.92–1.00)

0.074

13.5

0.328

0.204

 < 10.0

0.93 (0.89–0.96)

< 0.001

0.0

0.856

Adjusted family history of diabetes mellitus

     

 Yes

0.93 (0.91–0.96)

< 0.001

0.0

0.853

0.031

 No

1.02 (0.95–1.10)

0.607

0.0

0.656

NOS

     

 8 or 9

0.95 (0.92–0.98)

0.002

24.0

0.246

0.846

 7

0.95 (0.87–1.05)

0.317

0.0

0.889

A total of 9 studies reported an association between moderate total protein intake and subsequent T2DM risk. No significant association was observed between moderate total protein intake and T2DM risk (RR 1.00; 95% CI 0.95–1.06; P = 0.917; Fig. 3), with a significant heterogeneity across the studies (I2 74.7%; P < 0.001). Sensitivity analysis indicated that the results remained the same after each study was sequentially excluded from the overall analysis (Supplemental file 1). The results of the meta-regression analyses indicated that gender (P = 0.047) and follow-up duration (P = 0.003) play an important role in the relationship between moderate total protein intake and T2DM risk. Subgroup analyses showed no significant associations in all subsets based on predefined factors (Table 2). No significant publication bias was observed (P value for Egger’s test: 0.891; P value for Begg’s test: 0.669; Supplemental file 2).

Fig. 3

Association between moderate total protein intake and the risk of type 2 diabetes mellitus

Animal protein intake

Seven studies reported an association between high animal protein intake and elevated T2DM risk. High animal protein intake was associated with an increase in T2DM risk (RR 1.13; 95% CI 1.03–1.25; P = 0.013; Fig. 4), with significant heterogeneity among the evaluated studies (I2 64.7%; P = 0.002). The conclusion was stable and not affected by the exclusion of any specific study (Supplemental file 1). Country (P = 0.013) and NOS (P = 0.002) were significant confounders to determine the relationship between high animal protein intake and T2DM based on meta-regression analyses. Subgroup analyses indicated significant association between high animal protein intake and T2DM risk in most subsets, whereas high animal protein intake was not associated with T2DM risk when the study was conducted in Asia or Australia, included women, had a follow-up duration of < 10.0 years, was adjusted for family history of DM, or had a low quality (Table 2). The Egger’s (P = 0.721) and Begg’s (P = 0.592) test results showed no evidence of publication bias (Supplemental file 2).

Fig. 4

Association between high animal protein intake and the risk of type 2 diabetes mellitus

Seven studies reported an association between moderate animal protein intake and increased T2DM risk, but without reaching statistical significance in the pooled analysis (RR 1.06; 95% CI 1.00–1.12; P = 0.058; Fig. 5) and with significant heterogeneity among the included studies (I2 71.9%; P < 0.001). Sensitivity analyses indicated that moderate animal protein intake might play an important role in elevated T2DM risk based on the marginal 95% CI (Supplemental file 1). Meta-regression analysis revealed that no predefined factors could affect the relationship between moderate animal protein intake and T2DM risk. Subgroup analysis demonstrated an increased T2DM risk after moderate animal protein intake when the study was conducted in Europe (RR 1.09; 95% CI 1.01–1.19; P = 0.019), included both men and women (RR 1.07; 95% CI 1.01–1.15; P = 0.031), made no adjustment for family history of DM (RR 1.08; 95% CI 1.01–1.16; P = 0.036), and had a high quality (RR 1.07; 95% CI 1.00–1.14; P = 0.044). No evidence of publication bias was observed (P value for Egger’s test: 0.539; P value for Begg’s test: 0.858; Supplemental file 2).

Fig. 5

Association between moderate animal protein intake and the risk of type 2 diabetes mellitus

Plant protein intake

Seven of the assessed studies reported an association between high plant protein intake and subsequent T2DM risk, without reaching statistical significance in the pooled RR analysis (RR 0.93; 95% CI 0.86–1.01; P = 0.074; Fig. 6); heterogeneity among the included studies was not statistically significant (I2 31.6%; P = 0.156). Sensitivity analysis indicated that high plant protein intake might protect from T2DM progression (Supplemental file 1). Meta-regression analysis revealed that country (P = 0.040), gender (P = 0.006), adjustment for family history of DM (P = 0.012), and NOS (P = 0.041) could affect the relationship between high plant protein intake and T2DM risk. Subgroup analyses showed that high plant protein intake was associated with reduced T2DM risk when the study was conducted in Asia (RR 0.80; 95% CI 0.66–0.96; P = 0.017) or the USA (RR 0.91; 95% CI 0.84–0.97; P = 0.008), included men (RR 0.85; 95% CI 0.75–0.97; P = 0.018) or women (RR 0.90; 95% CI 0.83–0.97; P = 0.007), had a follow-up duration < 10.0 years (RR 0.88; 95% CI 0.81–0.96; P = 0.005), contained adjustments for the family history of DM (RR 0.90; 95% CI 0.84–0.96; P = 0.001), and had a low quality (RR 0.78; 95% CI 0.66–0.93; P = 0.005). No significant publication bias was observed (P value for Egger’s test: 0.918; P value for Begg’s test: 0.858; Supplemental file 2).

Fig. 6

Association between high plant protein intake and the risk of type 2 diabetes mellitus

Seven studies reported an association between moderate plant protein intake and subsequent T2DM risk. Overall, moderate plant protein intake was associated with reduced T2DM risk (RR 0.94; 95% CI 0.92–0.97; P < 0.001, without evidence of heterogeneity; Fig. 7). Sensitivity analysis indicated that the conclusion was not altered after each study was sequentially excluded from the overall analysis (Supplemental 1). Meta-regression analysis demonstrated that gender (P = 0.032) and adjustment for family history of DM (P = 0.031) affected the relationship between moderate plant protein intake and T2DM risk. Subgroup analyses indicated that this significant association was present in the study conducted in the USA (RR 0.93; 95% CI 0.90–0.96; P < 0.001), in studies that included men (RR 0.92; 95% CI 0.87–0.98; P = 0.011) or women (RR 0.93; 95% CI 0.90–0.96; P < 0.001), in studies with a follow-up duration of < 10.0 years (RR 0.93; 95% CI 0.89–0.96; P < 0.001), in studies with adjustment for family history of DM (RR 0.93; 95% CI 0.91–0.96; P < 0.001), and in studies with a high quality (RR 0.95; 95% CI 0.92–0.98; P = 0.002) (Table 2). No significant publication bias was observed (P value for Egger: 0.079; P value for Begg’s test: 0.210; Supplemental file 2).

Fig. 7

Association between moderate plant protein intake and the risk of type 2 diabetes mellitus

High-protein food

The association between specific high-protein foods and the risk of T2DM was also assessed (Table 3). High (RR 1.26; 95% CI 1.04–1.52; P = 0.016) and moderate (RR 1.11; 95% CI 1.02–1.20; P = 0.017) processed meat intake and high poultry intake (RR 1.25; 95% CI 1.03–1.52; P = 0.025) were associated with increased risk of T2DM in women. The risk of T2DM in men was significantly increased by a high intake of dairy products (RR 1.24; 95% CI 1.04–1.47; P = 0.018). Moderate fish intake was associated with a reduced risk of T2DM in men (RR 0.85; 95% CI 0.77–0.94; P = 0.002), and moderate intake of fiber-rich bread and cereals reduced the risk of T2DM in both men (RR 0.89; 95% CI 0.79–1.00) and women (RR 0.88; 95% CI 0.78–1.00; P = 0.045). Finally, no other significant association was observed for other types of high-protein foods, including red meat, eggs, soy, and refined cereals.

Table 3

Relative risk of type 2 diabetes according to types of high-protein food

High-protein food

Dose

Sex

RR and 95% CI

P value

Heterogeneity (%)

P value for heterogeneity

Red meat

High

Men

1.03 (0.86–1.23)

0.746

0.0

0.737

Women

1.11 (0.92–1.34)

0.265

7.0

0.300

Moderate

Men

0.97 (0.87–1.08)

0.578

0.0

0.785

Women

1.06 (0.95–1.18)

0.295

0.0

0.770

Processed meat

High

Men

1.17 (0.98–1.40)

0.077

0.0

0.509

Women

1.26 (1.04–1.52)

0.016

30.3

0.231

Moderate

Men

1.05 (0.95–1.17)

0.329

0.0

0.606

Women

1.11 (1.02–1.20)

0.017

6.2

0.371

Poultry

High

Men

1.04 (0.85–1.27)

0.696

Women

1.25 (1.03–1.52)

0.025

Moderate

Men

1.09 (0.95–1.25)

0.227

24.8

0.265

Women

1.02 (0.90–1.15)

0.751

0.0

0.409

Fish

High

Men

0.89 (0.75–1.05)

0.167

0.0

0.667

Women

1.20 (0.96–1.49)

0.104

Moderate

Men

0.85 (0.77–0.94)

0.002

0.0

0.779

Women

0.96 (0.85–1.10)

0.567

0.0

0.926

Eggs

High

Men

1.02 (0.58–1.79)

0.951

83.0

0.015

Women

1.11 (0.89–1.38)

0.351

Moderate

Men

1.04 (0.91–1.19)

0.574

36.8

0.176

Women

1.05 (0.92–1.19)

0.497

0.0

0.647

Total dairy product consumption

High

Men

1.24 (1.04–1.47)

0.018

0.0

0.568

Women

0.88 (0.71–1.10)

0.258

Moderate

Men

0.99 (0.89–1.10)

0.839

0.0

0.710

Women

0.95 (0.84–1.07)

0.411

0.0

0.824

Soy

High

Men

1.00 (0.77–1.29)

1.000

Women

1.00 (0.75–1.34)

1.000

Moderate

Men

1.08 (0.91–1.27)

0.394

22.8

0.274

Women

0.96 (0.81–1.14)

0.651

0.0

0.982

Fiber-rich bread and cereals

High

Men

0.84 (0.68–1.04)

0.108

Women

0.85 (0.68–1.06)

0.151

Moderate

Men

0.89 (0.79–1.00)

0.047

0.0

0.395

Women

0.88 (0.78–1.00)

0.045

0.0

0.520

Refined cereals

High

Men

1.02 (0.82–1.26)

0.857

Women

1.07 (0.87–1.32)

0.525

Moderate

Men

1.06 (0.94–1.20)

0.313

0.0

0.444

Women

0.94 (0.83–1.06)

0.310

0.0

0.461

Discussion

The present systematic review and meta-analysis of 21 cohort studies reported in 10 articles provide comprehensive evidence for an association between dietary protein intake and increased T2DM risk. A combined total of 487,956 individuals with various characteristics participated in these studies and 38,350 T2DM cases were reported. This study indicated that a high total protein intake was associated with an increased T2DM risk, whereas there was no effect of moderate total protein intake on T2DM risk. Moreover, a high animal protein intake increased the T2DM risk, whereas moderate intake of animal protein was not significantly associated with a risk of T2DM. Furthermore, moderate plant protein intake was linked to a reduction in T2DM risk, while no such effect was observed for high plant protein intake. In addition, the intake of processed meat, poultry, and total dairy products correlated with a greater risk of T2DM, whereas moderate intake of fish or fiber-rich bread and cereals appeared to protect from T2DM. Finally, these associations might differ according to country, gender, follow-up duration, family history of DM, and study quality.

A previous meta-analysis has already illustrated the relationship between dietary protein intake and T2DM risk. The authors demonstrated that the highest total and animal protein intakes were associated with an increased T2DM risk in both men and women, whereas plant protein consumption was associated with reduced T2DM risk in women. Moreover, the T2DM risk was increased in individuals who consumed high amounts of red and processed meat, whereas high intake of soy, dairy, and dairy products protected from T2DM. In contrast, egg and fish intakes were not linked to T2DM risk [10]. Nevertheless, the previous meta-analysis did not determine whether the relationship between dietary protein intake and T2DM differs based on individual characteristics. Moreover, we identified a significant mistake in this meta-analysis as it included the study conducted by Bao et al., which specifically evaluated the association between dietary protein intake and gestational DM risk [31]. Although women with gestational DM have an increased risk of developing T2DM in the years following pregnancy, this study might have overestimated the association between dietary protein intake and T2DM risk [32]. Therefore, the current study was conducted to provide an updated meta-analysis and elucidate the relationship between dietary protein intake and T2DM risk.

The mechanisms behind the association of dietary protein intake with T2DM risk are not well established. The positive relationship between total protein intake and T2DM risk might reflect the predominant protein source, as most of the dietary protein consumed was derived from animal sources, particularly in the USA and Europe [26]. Moreover, high animal protein intake is significantly correlated with high fat, low fiber, and low vitamin intakes, which can also have an impact on T2DM progression. Furthermore, amino acid metabolism plays a pivotal role in the development of metabolic disorders and could, therefore, increase the T2DM risk [33, 34]. Glycine and methionine are positively associated with T2DM risk and these amino acids are primarily obtained from animal food sources [35]. In addition, plant protein intake inversely correlates with T2DM risk. This could be due to the high intake of fiber, magnesium, and vitamin with plant-based food sources, which could slow down T2DM progression. Moreover, the balanced amino acid composition of plant proteins has been shown to exert beneficial effects on amino acid metabolism and could diminish the risk of metabolic disorders [36].

The results of this study indicated that the relationship between the dietary protein intake and T2DM risk differs according to country, gender, follow-up duration, adjusted family history of DM, and NOS. We propose the following reasons for this variation: (1) the source of protein intake differs between countries, and the protein intake was higher in Western countries than in Eastern countries; (2) women have a higher percentage of body fat, which could explain the differences between men and women in total and animal protein intake and associated T2DM risk [37, 38]; (3) the follow-up duration correlated with high event rates, which were associated with body weight in the overall analysis; (4) a family history of T2DM is significantly associated with diagnosis at a younger age and higher body mass index and waist circumference, which in turn may lead to complications; and (5) the study quality correlated with the evidence level, and in turn the results differed based on the study quality.

This study has several strengths that should be highlighted. The results were based on a prospective cohort study, which may prevent potential selection and recall biases observed with retrospective observational studies. Moreover, all included studies had fully adjusted models, and hence the results were adjusted for the most important factors. Furthermore, this study had a large sample size, and the pooled results are more robust than those of any individual study. Finally, analyses were performed to determine whether these associations differed according to country, gender, follow-up duration, adjusted family history of DM, and study quality.

The limitations of this study should be acknowledged. First, the included studies were inconsistent for the adjusted factors that may have an impact on T2DM progression. Second, the cut-off values of dietary protein intake were different across studies, which might introduce potential bias and high heterogeneity of analyzed data. Third, an analysis of the association between specific high-protein foods and the risk of T2DM stratified by country and other characteristics was not performed because only a few of the included studies reported specific high-protein foods. Fourth, we could not determine if any associations between dietary proteins and the risk of T2DM are due to interactions with other macronutrients such as fat, fiber, and vitamin, as data on these factors were not available. Fifth, publication bias was inevitable because this study is based on published studies. Finally, the studies contained no individual data and, therefore, a detailed analysis of specific characteristics of the study participants was restricted.

In conclusion, high total protein intake is notably associated with an increased risk of T2DM, whereas moderate total protein intake has little or no effect on T2DM progression. Moreover, high or moderate animal protein intake might increase the T2DM risk. Conversely, plant protein intake might offer protective effects against T2DM. Further large-scale prospective studies should be conducted to verify the stratified results of this study.

Notes

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.

Informed consent

For this type of study formal consent is not required.

Supplementary material

592_2019_1320_MOESM1_ESM.docx (1.9 mb)
Supplementary material 1 (DOCX 1995 KB)
592_2019_1320_MOESM2_ESM.docx (2 mb)
Supplementary material 2 (DOCX 2029 KB)

References

  1. 1.
    American Diabetes A (2018) Standards of Medical Care in Diabetes-2018 Abridged for Primary Care Providers. Clin Diabetes 36(1):14–37.  https://doi.org/10.2337/cd17-0119 CrossRefGoogle Scholar
  2. 2.
    Radaelli MG, Martucci F, Perra S, Accornero S, Castoldi G, Lattuada G et al (2018) NAFLD/NASH in patients with type 2 diabetes and related treatment options. J Endocrinol Invest 41(5):509–521.  https://doi.org/10.1007/s40618-017-0799-3 CrossRefGoogle Scholar
  3. 3.
    Alberti KG (1996) The clinical implications of impaired glucose tolerance. Diabetes Med 13(11):927–937  https://doi.org/10.1002/(SICI)1096-9136(199611)13:11%3C927::AID-DIA279%3E3.0.CO;2-E CrossRefGoogle Scholar
  4. 4.
    de Vegt F, Dekker JM, Jager A, Hienkens E, Kostense PJ, Stehouwer CD et al (2001) Relation of impaired fasting and postload glucose with incident type 2 diabetes in a Dutch population: the Hoorn Study. JAMA 285(16):2109–2113CrossRefGoogle Scholar
  5. 5.
    Tabak AG, Herder C, Rathmann W, Brunner EJ, Kivimaki M (2012) Prediabetes: a high-risk state for diabetes development. Lancet 379(9833):2279–2290.  https://doi.org/10.1016/S0140-6736(12)60283-9 CrossRefGoogle Scholar
  6. 6.
    Bellou V, Belbasis L, Tzoulaki I, Evangelou E (2018) Risk factors for type 2 diabetes mellitus: an exposure-wide umbrella review of meta-analyses. PLoS One 13(3):e0194127.  https://doi.org/10.1371/journal.pone.0194127 CrossRefGoogle Scholar
  7. 7.
    Nuttall FQ, Schweim K, Hoover H, Gannon MC (2008) Effect of the LoBAG30 diet on blood glucose control in people with type 2 diabetes. Br J Nutr 99(3):511–519.  https://doi.org/10.1017/S0007114507819155 CrossRefGoogle Scholar
  8. 8.
    Weickert MO, Roden M, Isken F, Hoffmann D, Nowotny P, Osterhoff M et al (2011) Effects of supplemented isoenergetic diets differing in cereal fiber and protein content on insulin sensitivity in overweight humans. Am J Clin Nutr 94(2):459–471.  https://doi.org/10.3945/ajcn.110.004374 CrossRefGoogle Scholar
  9. 9.
    Pounis GD, Tyrovolas S, Antonopoulou M, Zeimbekis A, Anastasiou F, Bountztiouka V et al (2010) Long-term animal-protein consumption is associated with an increased prevalence of diabetes among the elderly: the Mediterranean Islands (MEDIS) study. Diabetes Metab 36(6 Pt 1):484–490.  https://doi.org/10.1016/j.diabet.2010.06.007 CrossRefGoogle Scholar
  10. 10.
    Tian S, Xu Q, Jiang R, Han T, Sun C, Na L (2017) Dietary protein consumption and the risk of Type 2 diabetes: a systematic review and meta-analysis of cohort studies. Nutrients.  https://doi.org/10.3390/nu9090982 Google Scholar
  11. 11.
    Moher D, Liberati A, Tetzlaff J, Altman DG, Group P (2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 6(7):e1000097.  https://doi.org/10.1371/journal.pmed.1000097 CrossRefGoogle Scholar
  12. 12.
    Wells G, Shea B, O’connell D, Peterson J, Welch V, Losos M, Tugwell P (2016) The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. Ottawa Hospital Research Institute, Ottawa (Available in March) Google Scholar
  13. 13.
    DerSimonian R, Laird N (1986) Meta-analysis in clinical trials. Control Clin Trials 7(3):177–188CrossRefGoogle Scholar
  14. 14.
    Ades AE, Lu G, Higgins JP (2005) The interpretation of random-effects meta-analysis in decision models. Med Decis Making 25(6):646–654.  https://doi.org/10.1177/0272989X05282643 CrossRefGoogle Scholar
  15. 15.
    Higgins JP, Thompson SG, Deeks JJ, Altman DG (2003) Measuring inconsistency in meta-analyses. BMJ 327(7414):557–560.  https://doi.org/10.1136/bmj.327.7414.557 CrossRefGoogle Scholar
  16. 16.
    Tobias A (1999) Assessing the influence of a single study in the meta-analysis estimate. Stata Tech Bull 47:15–17Google Scholar
  17. 17.
    Deeks JJ, Altman DG, Bradburn MJ (2001) Statistical methods for examining heterogeneity and combining results from several studies in meta-analysis. In: Egger M, Smith GD, Altman DG (eds) Systematic reviews in health care: meta-analysis in context, 2nd edn. Wiley, New York, pp 285–312CrossRefGoogle Scholar
  18. 18.
    Thompson SG, Higgins JP (2002) How should meta-regression analyses be undertaken and interpreted? Stat Med 21(11):1559–1573CrossRefGoogle Scholar
  19. 19.
    Egger M, Davey Smith G, Schneider M, Minder C (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ 315(7109):629–634CrossRefGoogle Scholar
  20. 20.
    Begg CB, Mazumdar M (1994) Operating characteristics of a rank correlation test for publication bias. Biometrics 50(4):1088–1101CrossRefGoogle Scholar
  21. 21.
    Alhazmi A, Stojanovski E, McEvoy M, Garg ML (2014) Macronutrient intake and type 2 diabetes risk in middle-aged Australian women. Results from the Australian Longitudinal Study on Women’s Health. Public Health Nutr 17(7):1587–1594.  https://doi.org/10.1017/S1368980013001870 CrossRefGoogle Scholar
  22. 22.
    Ericson U, Sonestedt E, Gullberg B, Hellstrand S, Hindy G, Wirfalt E, Orho-Melander M (2013) High intakes of protein and processed meat associate with increased incidence of type 2 diabetes. Br J Nutr 109(6):1143–1153.  https://doi.org/10.1017/S0007114512003017 CrossRefGoogle Scholar
  23. 23.
    Malik VS, Li Y, Tobias DK, Pan A, Hu FB (2016) Dietary protein intake and risk of Type 2 diabetes in US men and women. Am J Epidemiol 183(8):715–728.  https://doi.org/10.1093/aje/kwv268 CrossRefGoogle Scholar
  24. 24.
    Nanri A, Mizoue T, Kurotani K, Goto A, Oba S, Noda M et al (2015) Low-carbohydrate diet and type 2 diabetes risk in Japanese men and women: the Japan Public Health Center-Based Prospective Study. PLoS One 10(2):e0118377.  https://doi.org/10.1371/journal.pone.0118377 CrossRefGoogle Scholar
  25. 25.
    Shang X, Scott D, Hodge AM, English DR, Giles GG, Ebeling PR, Sanders KM (2016) Dietary protein intake and risk of type 2 diabetes: results from the Melbourne Collaborative Cohort Study and a meta-analysis of prospective studies. Am J Clin Nutr 104(5):1352–1365.  https://doi.org/10.3945/ajcn.116.140954 CrossRefGoogle Scholar
  26. 26.
    Sluijs I, Beulens JW, Spijkerman AM, Grobbee DE, Van der Schouw YT (2010) 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 33(1):43–48CrossRefGoogle Scholar
  27. 27.
    Song Y, Manson JE, Buring JE, Liu S (2004) A prospective study of red meat consumption and type 2 diabetes in middle-aged and elderly women: the women’s health study. Diabetes care 27(9):2108–2115CrossRefGoogle Scholar
  28. 28.
    Tinker LF, Sarto GE, Howard BV, Huang Y, Neuhouser ML, Mossavar-Rahmani Y et al (2011) Biomarker-calibrated dietary energy and protein intake associations with diabetes risk among postmenopausal women from the Women’s Health Initiative. Am J Clin Nutr 94(6):1600–1606CrossRefGoogle Scholar
  29. 29.
    van Nielen M, Feskens EJ, Mensink M, Sluijs I, Molina E, Amiano P et al (2014) Dietary protein intake and incidence of type 2 diabetes in Europe: the EPIC-InterAct Case-Cohort Study. Diabetes Care 37(7):1854–1862.  https://doi.org/10.2337/dc13-2627 CrossRefGoogle Scholar
  30. 30.
    Virtanen HEK, Koskinen TT, Voutilainen S, Mursu J, Tuomainen TP, Kokko P et al (2017) Intake of different dietary proteins and risk of type 2 diabetes in men: the Kuopio Ischaemic Heart Disease Risk Factor Study. Br J Nutr 117(6):882–893.  https://doi.org/10.1017/S0007114517000745 CrossRefGoogle Scholar
  31. 31.
    Bao W, Bowers K, Tobias DK, Hu FB, Zhang C (2013) Prepregnancy dietary protein intake, major dietary protein sources, and the risk of gestational diabetes mellitus: a prospective cohort study. Diabetes Care 36(7):2001–2008.  https://doi.org/10.2337/dc12-2018 CrossRefGoogle Scholar
  32. 32.
    Goueslard K, Cottenet J, Mariet AS, Sagot P, Petit JM, Quantin C (2017) Early screening for type 2 diabetes following gestational diabetes mellitus in France: hardly any impact of the 2010 guidelines. Acta Diabetol 54(7):645–651CrossRefGoogle Scholar
  33. 33.
    Wang TJ, Larson MG, Vasan RS, Cheng S, Rhee EP, McCabe E, Lewis GD, Fox CS, Jacques PF, Fernandez C, O’Donnell CJ, Carr SA, Mootha VK, Florez JC, Souza A, Melander O, Clish CB, Gerszten RE (2011) Metabolite profiles and the risk of developing diabetes. Nat Med 17(4):448–453.  https://doi.org/10.1038/nm.2307 CrossRefGoogle Scholar
  34. 34.
    Würtz P, Havulinna AS, Soininen P, Tynkkynen T, Prieto-Merino D, Tillin T et al (2015) Metabolite profiling and cardiovascular event riskclinical perspective: a prospective study of 3 population-based cohorts. Circulation 131(9):774–785CrossRefGoogle Scholar
  35. 35.
    Wittenbecher C, Mühlenbruch K, Kröger J, Jacobs S, Kuxhaus O, Floegel A et al (2015) Amino acids, lipid metabolites, and ferritin as potential mediators linking red meat consumption to type 2 diabetes. Am J Clin Nutr 101(6):1241–1250CrossRefGoogle Scholar
  36. 36.
    Young VR, Pellett PL (1994) Plant proteins in relation to human protein and amino acid nutrition. Am J Clin Nutr 59(5 Suppl):1203S–1212S.  https://doi.org/10.1093/ajcn/59.5.1203S CrossRefGoogle Scholar
  37. 37.
    Karastergiou K, Smith SR, Greenberg AS, Fried SK (2012) Sex differences in human adipose tissues–the biology of pear shape. Biol Sex Differ 3(1):13CrossRefGoogle Scholar
  38. 38.
    Kautzky-Willer A, Harreiter J, Pacini G (2016) Sex and gender differences in risk, pathophysiology and complications of Type 2 diabetes mellitus. Endocr Rev 37(3):278–316.  https://doi.org/10.1210/er.2015-1137 CrossRefGoogle Scholar

Copyright information

© The Author(s) 2019

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors and Affiliations

  1. 1.Department of EndocrinologyFoshan Hospital of TCMFoshanChina
  2. 2.Functional DepartmentFoshan Hospital of TCMFoshanChina

Personalised recommendations