Dietary Intake of Total, Animal, and Vegetable Protein and Risk of

Text Preview:
Epidemiology/Health Services Research
 O R I G I N A L                       A R T I C L E




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
IVONNE SLUIJS, MSC1                                                 ANNEMIEKE M.W. SPIJKERMAN, PHD2                                    based on extreme low-carbohydrate,
JOLINE W.J. BEULENS, PHD1,2                                         DIEDERICK E. GROBBEE, MD, PHD1                                     high-protein contents with favorable ef-
DAPHNE L. VAN DER A, PHD3                                           YVONNE T. VAN DER SCHOUW, PHD1                                     fects on body weight and glucose ho-
                                                                                                                                       meostasis in short-term interventions
                                                                                                                                       (6,7). In contrast, a cross-sectional study
OBJECTIVE -- Dietary recommendations are focused mainly on relative dietary fat and car-                                               related long-term high-protein intake to
bohydrate content in relation to diabetes risk. Meanwhile, high-protein diets may contribute to                                        elevated glucose concentrations and insu-
disturbance of glucose metabolism, but evidence from prospective studies is scarce. We exam-                                           lin resistance in healthy individuals (8).
ined the association among dietary total, vegetable, and animal protein intake and diabetes                                                 Prospective studies addressing di-
incidence and whether consuming 5 energy % from protein at the expense of 5 energy % from
either carbohydrates or fat was associated with diabetes risk.
                                                                                                                                       etary protein and diabetes risk focused
                                                                                                                                       mainly on high-protein food groups, such
RESEARCH DESIGN AND METHODS -- A prospective cohort study was conducted                                                                as meat and soy. Red processed meat in-
among 38,094 participants of the European Prospective Investigation into Cancer and Nutrition                                          take was related to increased diabetes
(EPIC)-NL study. Dietary protein intake was measured with a validated food frequency ques-                                             risk, independent of fat intake (9 12),
tionnaire. Incident diabetes was verified against medical records.                                                                     whereas intake of legumes and soy de-
                                                                                                                                       creased diabetes risk in Asian women
RESULTS -- During 10 years of follow-up, 918 incident cases of diabetes were documented.
Diabetes risk increased with higher total protein (hazard ratio 2.15 [95% CI 1.772.60] highest
                                                                                                                                       (13), suggesting divergent effects of ani-
vs. lowest quartile) and animal protein (2.18 [1.80 2.63]) intake. Adjustment for confounders                                         mal and vegetable protein. Studies exam-
did not materially change these results. Further adjustment for adiposity measures attenuated the                                      ining the relation between dietary protein
associations. Vegetable protein was not related to diabetes. Consuming 5 energy % from total or                                        and diabetes are scarce. One study re-
animal protein at the expense of 5 energy % from carbohydrates or fat increased diabetes risk.                                         ported increased diabetes risk with higher
                                                                                                                                       animal protein intake and no association
CONCLUSIONS -- Diets high in animal protein are associated with an increased diabetes                                                  with vegetable protein intake (11). Under
risk. Our findings also suggest a similar association for total protein itself instead of only animal
                                                                                                                                       isocaloric conditions, higher protein in-
sources. Consumption of energy from protein at the expense of energy from either carbohydrates
or fat may similarly increase diabetes risk. This finding indicates that accounting for protein                                        takes will lead to lower intakes of other
content in dietary recommendations for diabetes prevention may be useful.                                                              macronutrients, which can be investi-
                                                                                                                                       gated using substitution models in which
                                                                                       Diabetes Care 33:4348, 2010                    other macronutrients are substituted for
                                                                                                                                       protein (14). The European Prospective
                                                                                                                                       Investigation into Cancer and Nutrition

M
        any research efforts have focused                           (3,4), but dietary protein content in rela-
        on macronutrient intake in rela-                            tion to diabetes prevention received little                        (EPIC)-Potsdam study related consump-
        tion to type 2 diabetes risk (1,2),                         attention (3).                                                     tion of 5 energy % from carbohydrate at the
but mainly on relative carbohydrate and                                 In industrialized countries, dietary                           expense of 5 energy % from protein with
fat content. Effects of various protein con-                        protein intake has increased substantially                         decreased diabetes risk (1). However, the
sumption are less well documented. Both                             during the last few decades, exceeding                             Nurses Health Study II did not find such an
Dutch and U.S. nutritional recommenda-                              50% of the recommended dietary allow-                              association (15). Both studies made no dis-
tions provide information on optimal di-                            ance (5). Moreover, popular weight loss                            tinction between animal and vegetable pro-
etary protein content for diabetic patients                         diets, such as the Atkins diet, are often                          tein. The response to dietary protein
                                                                                                                                       content may be dependent on an individu-
From the 1Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht,                              al's degree of underlying insulin resistance
  the Netherlands; the 2Center for Prevention and Health Services Research, National Institute for Public                              (6,7), determined by adiposity.
  Health and the Environment, Bilthoven, the Netherlands; and the 3Center for Nutrition and Health,                                         We aimed to investigate whether
  National Institute for Public Health and the Environment, Bilthoven, the Netherlands.                                                higher dietary intakes of total, animal,
Corresponding author: Ivonne Sluijs, i.sluijs-2@umcutrecht.nl.
Received 20 July 2009 and accepted 25 September 2009. Published ahead of print at http://care.                                         and vegetable protein were associated
  diabetesjournals.org on 17 October 2009. DOI: 10.2337/dc09-1321.                                                                     with type 2 diabetes risk and whether
 2010 by the American Diabetes Association. Readers may use this article as long as the work is properly                              consumption of energy from protein at
  cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.                           the expense of the same energy percent-
  org/licenses/by-nc-nd/3.0/ for details.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby   age from fat or carbohydrate was associ-
   marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.                                      ated with type 2 diabetes risk. Moreover,

care.diabetesjournals.org                                                                                                DIABETES CARE, VOLUME 33, NUMBER 1, JANUARY 2010       43
Dietary protein intake and diabetes risk

we examined whether an interaction with         and summed the values over all food             measurements were performed twice in
measures of adiposity was present.              items. Intakes of nutrients were adjusted       the supine position on the right arm using
                                                for total energy intake by the regression       a Boso Oscillomat (Bosch & Son, Jungin-
RESEARCH DESIGN AND                             residual method and by using nutrient           gen, Germany) (Prospect) or on the left
METHODS -- EPIC-NL consists of                  densities (percentage of total energy in-       arm using a random zero sphygmoma-
the two Dutch contributions to the EPIC         take, only for macronutrients) (14).            nometer (MORGEN), from which the
study, the Prospect-EPIC and MORGEN-                                                            mean was taken. Hypertension was de-
EPIC cohorts. These cohorts were set up         Diabetes                                        fined as being present when one or more
simultaneously in 19931997 and                 Occurrence of diabetes during follow-up         of the following criteria were met: dia-
merged into one Dutch EPIC cohort. The          was self-reported in two follow-up ques-        stolic blood pressure 90 mmHg, sys-
design and rationale of EPIC-NL are de-         tionnaires with 3- to 5-year intervals. Par-    tolic blood pressure 140 mmHg, self-
scribed elsewhere (16). The Prospect-           ticipants were asked whether diabetes           reported antihypertensive medication
EPIC study includes 17,357 women aged           was diagnosed, in what year, and by             use, or self-reported presence of hyper-
49 70 years living in Utrecht and vicinity     whom and what treatment was received.           tension. Waist circumference, height, and
(17). The MORGEN-EPIC cohort consists           In the Prospect study, incident cases of        weight were measured, and BMI was cal-
of 22,654 adults aged 21 64 years se-          diabetes were detected via a urinary glu-       culated. All measurements were per-
lected from random samples of the Dutch         cose strip test, sent out with the first fol-   formed according to standard operating
population in three Dutch towns (18). All       low-up questionnaire, for detection of          procedures. Weight during follow-up was
participants provided informed consent          glucosuria. Diagnoses of diabetes were          derived from mailed follow-up question-
before study inclusion. The study com-          also obtained from the Dutch Centre for         naires or physical examination (Doet-
plies with the Declaration of Helsinki and      Health Care Information, which holds a          inchem part). Weight change was defined
was approved by the institutional board         standardized computerized register of           as the difference between weight at base-
of the University Medical Center Utrecht        hospital discharge diagnoses. In this reg-      line and follow-up. Because the follow-up
(Prospect) and the Medical Ethical Com-         ister, admission files have been entered        period varied, we calculated annual
mittee of TNO Nutrition and Food Re-            continuously from all general and univer-       weight change by dividing weight change
search (MORGEN). After exclusion of             sity hospitals in the Netherlands from          by the years of follow-up.
those with prevalent diabetes (n  615)          1990 onward. All diagnoses were coded
and of individuals with abnormal energy         according to the ICD-9-CM. Follow-up            Data analysis
intake (kcal 600 or 5,000) (n  108),            was complete on 1 January 2006. Poten-          Protein intake, adjusted for total energy
missing nutritional data (n  213), and          tial cases identified by any of these meth-     intake by the regression residual method
missing follow-up (n  981), 38,094 par-         ods were verified against participants'         (14), was categorized into quartiles. Cox
ticipants were left for analysis.               general practitioner or pharmacist infor-       proportional hazard models were used to
                                                mation through mailed questionnaires.           calculate crude and adjusted hazard ratios
Intake of protein and other nutrients           Diabetes was defined as being present           (HRs) and 95% CI for the associations be-
Daily nutritional intake was obtained           when either of these confirmed the diag-        tween quartiles of protein intake and dia-
from a self-administered food-frequency         nosis. For 89% of participants with po-         betes. We estimated Ptrend by including
questionnaire (FFQ) containing ques-            tential diabetes, verification information      median protein intakes per quartile as
tions on the usual frequency of consump-        was available, and 72% were verified as         continuous variables in the Cox regres-
tion of 79 main food items during the year      having type 2 diabetes and were used for        sion models. In addition, we analyzed as-
preceding enrollment. This questionnaire        the analysis.                                   sociations between protein per 10 g of
allows estimation of the average daily                                                          intake and diabetes risk. In the multivar-
consumption of 178 foods. The FFQ was           Other measurements                              iate analysis, we first included sex (male
administered once at baseline and sent to       At baseline, participants filled in a mailed    or female) and age at recruitment (contin-
the participants by mail. Participants re-      general questionnaire containing ques-          uous). In the second model, we added nu-
turned the FFQ during the physical ex-          tions on demographics, the presence of          tritional factors: energy-adjusted intake of
amination screening, where difficulties in      chronic diseases, and risk factors for          saturated fat, monounsaturated fat, poly-
filling out the questionnaire were dis-         chronic diseases. Smoking was catego-           unsaturated fat, cholesterol, vitamin E,
cussed. A registered dietitian checked the      rized into current, past, and never smoker      magnesium, fiber, and glycemic load (con-
FFQ for inconsistencies, which were re-         and parental history of diabetes into none,     tinuous). In the third model, we addition-
solved by contacting the participant. The       one, and both parents. Physical activity        ally corrected for diabetes risk factors:
FFQ has been validated against 12 24-h          was assessed using a questionnaire vali-        energy-adjusted alcohol consumption
dietary recalls (19) with Pearson correla-      dated in an elderly population (20) and         (four categories), physical activity (four
tion coefficients for protein intake of 0.67    categorized after calculating the Cam-          categories), mean systolic and diastolic
in women and 0.71 in men (19). The gly-         bridge Physical Activity Score. Because         blood pressure (continuous), education
cemic index of foods, a measure of the          we could not calculate a total physical ac-     level (three categories), and parental his-
extent to which foods raise the blood glu-      tivity score for 14% of all participants, we    tory of diabetes (three categories). In the
cose level, was obtained from the Foster-       imputed missing scores using single lin-        fourth model, BMI (four categories) and
Powell international table. We calculated       ear regression modeling. Participants           waist circumference (continuous) were
glycemic load by multiplying the glyce-         could return the questionnaire at the           included. To examine the influence of
mic index of a food with its carbohydrate       physical examination screening. During          weight change during follow-up, we ad-
content and then multiplied this value by       the baseline physical examination screen-       ditionally corrected the analysis for an-
the frequency of consumption of this food       ing, systolic and diastolic blood pressure      nual weight change (continuous).

44     DIABETES CARE, VOLUME 33, NUMBER 1, JANUARY 2010                                                             care.diabetesjournals.org
                                                                                                                            Sluijs and Associates

Table 1--Baseline characteristics of the study population, according to quartiles of daily nutritional total protein intake

                                                   Quartile 1 (low)                Quartile 2              Quartile 3             Quartile 4 (high)
Participants                                            9,523                        9,524                   9,524                      9,523
Male sex                                            2,643 (27.8)                 2,685 (28.2)            2,450 (25.7)               1,962 (20.6)
Age (years)                                            48  12                       48  12                  50  12                     51  11
Energy intake (kcal/day)                            2,036  645                   2,106  618              2,078  602                 1,998  622
Animal protein intake (g/day)                        35.2  7.0                    44.5  5.2               51.3  5.2                  62.9  8.3
Vegetable protein intake (g/day)                     27.0  5.5                    27.6  4.8               27.6  4.8                  26.9  4.7
Saturated fat intake (g/day)                         31.2  6.2                    32.5  5.6               33.0  5.6                  33.4  5.9
Polyunsaturated fat intake (g/day)                   15.3  4.3                    15.3  3.8               14.9  3.7                  14.2  3.6
Monounsaturated fat intake (g/day)                   29.2  5.5                    29.7  5.0               29.6  5.0                  29.2  5.2
Cholesterol intake (mg/day)                         190.0  53.0                  210.4  51.1             223.4  53.5                245.3  64.2
Carbohydrate intake (g/day)                         231.5  34.3                  223.7  29.0             219.3  28.2                213.6  28.8
Glycemic load intake (g/day)                        120.8  24.7                  114.7  20.1             110.8  18.9                106.0  18.3
Fiber intake (g/day)                                 22.1  5.2                    23.2  4.6               23.8  4.6                  24.2  4.7
Vitamin C intake (mg/day)                           101.4  49.7                  106.6  43.1             111.6  42.6                118.0  44.0
Vitamin E intake (mg/day)                            12.9  3.6                    12.5  3.2               12.1  3.0                  11.5  3.0
Magnesium intake (mg/day)                           304.9  46.2                  327.2  41.7             343.9  40.9                365.9  43.3
Iron intake (mg/day)                                 10.8  1.6                    11.3  1.5               11.7  1.6                  12.1  1.7
Heme iron intake (mg/day)                             1.6  0.7                     1.9  0.7                2.1  0.7                   2.5  0.9
Alcohol intake (g/day)                               15.8  24.4                   11.0  15.9               9.7  13.9                  8.0  12.6
BMI (kg/m2)                                          24.7  3.7                    25.3  3.7               25.8  3.9                  26.7  4.3
Waist circumference (cm)                             83.1  11.2                   84.6  11.2              85.5  11.2                 87.1  11.7
Current smoker                                      3,655 (38.4)                 3,014 (31.7)            2,541 (26.8)               2,420 (25.5)
Physically inactive*                                3,983 (41.8)                 3,609 (37.9)            3,482 (36.6)               3,413 (35.9)
Higher education                                    1,993 (20.9)                 2,104 (22.1)            2,030 (21.3)               1,703 (17.9)
Parental history of diabetes                        1,559 (16.4)                 1,615 (17.0)            1,767 (18.5)               1,922 (20.2)
Systolic blood pressure (mmHg)                      124.3  18.6                  125.3  18.8             126.6  18.7                127.8  19.0
Diastolic blood pressure (mmHg)                      77.2  10.6                   77.5  10.5              78.0  10.6                 78.4  10.7
Hypertension                                        3,108 (32.6)                 3,307 (34.7)            3,491 (36.7)               3,762 (39.5)
Data are n (%) or means  SD. *Inactive according to Cambridge physical activity index.



     We used a multivariate nutrient den-           including continuous interaction terms.                During a mean follow-up of 10.1 
sity model by including total energy in-            The proportionality assumption was                1.9 (mean  SD) years, 918 incident
take and energy percentages of protein              checked visually using log-minus-log              cases of type 2 diabetes were docu-
and other macronutrients in the multivar-           plots, with no deviations detected. Data          mented. Diabetes risk increased signifi-
iate Cox regression model. Macronutrient            were analyzed using SPSS for Windows              cantly over the quartiles of total protein
intakes were entered into the model per 5           (version 14.0).                                   intake. Adjustment for age, sex, dietary
energy %. Total energy intake was entered                                                             factors, and diabetes risk factors yielded
into the model to keep energy intake con-           RESULTS -- Mean protein intake was                an HR in the highest versus lowest quar-
stant, essential for creating an isocaloric         75.7 g/day; animal protein accounted for          tile (HRQ4) of 1.67 (95% CI 1.29 2.16).
model (14). By leaving out energy per-              the majority. The main contributors to            After further adjustment for adiposity
centages from carbohydrate in the regres-           protein intake were meat (39%), milk              measures, this association was no longer
sion model, we created a model in which             (products) (29%), and cheese (18%) for            significant (HRQ4 1.18 [0.911.53]) (Ta-
the difference in diabetes risk associated          animal protein and bread (43%), fruit and         ble 2). Removing either BMI or waist cir-
with consumption of 5 energy % from                 vegetables (14%), and potatoes (9%) for           cumference from model 4 yielded
protein at the expense of 5 energy % from           vegetable protein. Moderate correlations          comparable, nonsignificant associations
carbohydrate, while total energy intake is          were present between meat intake and to-          (excluding BMI, HRQ4 1.22 [0.94 1.59]).
kept constant, is presented. Similarly, by          tal (r  0.30) and animal (r  0.36) pro-           For animal protein, we observed similar
leaving out energy percentages from fat,            tein, and between intake of milk                  results. Vegetable protein intake was not
we presented the difference in diabetes             (products) and total (r  0.46) and ani-           related to diabetes. Analyzing protein per
risk associated with consumption of 5 en-           mal (r  0.50) protein. Over the quartiles         10 g of intake showed comparable results,
ergy % from protein at the expense of 5             of total protein intake, mean age, BMI,           with significantly increased diabetes risk
energy % from fat, while energy intake is           waist circumference, and intakes of satu-         for higher total and animal protein intake
held constant.                                      rated fat and carbohydrates increased,            in all models (Table 2).
     Interactions of protein intake with            whereas mean intakes of polyunsaturated                Adjustment for weight change did not
BMI (25 or 25 kg/m2) and waist cir-                 fat and fiber and percentages of men,             change these findings (model 3, HRQ4
cumference (84 or 84 cm) were esti-                 smokers, and physically inactive individ-         1.67 [1.28 2.16]). Moreover, additional
mated using a likelihood ratio test and by          uals decreased (Table 1).                         correction for meat and poultry intake in

care.diabetesjournals.org                                                                   DIABETES CARE, VOLUME 33, NUMBER 1, JANUARY 2010    45
Dietary protein intake and diabetes risk

Table 2--Univariable and adjusted HRs (95% CI) for the association between intake of protein, in quartiles and per 10 g, and incident type 2
diabetes

                                       Quartile 1 (low)         Quartile 2             Quartile 3          Quartile 4 (high)       Ptrend          Per 10 g
Total protein
  Cases/at risk (n)                       153/9,523             185/9,524              249/9,524              331/9,523
  Quartile median total protein
     (g/day)                                  64                   72                79                88
  Univariable                                  1            1.20 (0.971.49) 1.63 (1.331.99)* 2.15 (1.772.60)* 0.001 1.36 (1.281.44)*
  Model 1: age and sex                         1            1.17 (0.941.45) 1.50 (1.231.84)* 1.85 (1.531.25)* 0.001 1.28 (1.221.36)*
  Model 2: model 1  dietary
     intake                                     1           1.24 (0.991.54) 1.65 (1.322.07)* 2.16 (1.692.76)* 0.001 1.45 (1.341.56)*
  Model 3: model 2 
     diabetes risk factors                      1           1.16 (0.921.45) 1.45 (1.151.83)* 1.67 (1.292.16)* 0.001 1.33 (1.221.45)*
  Model 4: model 3  waist
     and BMI                                    1           1.03 (0.821.29) 1.20 (0.951.51)             1.18 (0.911.53)         0.15      1.16 (1.061.26)
Animal protein
  Cases/at risk (n)                       155/9,523             182/9,524              243/9,524              338/9,523
  Quartile median animal
     protein (g/day)                          35                   44                52                62
  Univariable                                  1            1.16 (0.941.44) 1.56 (1.281.91)* 2.18 (1.802.63)* 0.001 1.32 (1.251.39)*
  Model 1: age and sex                         1            1.08 (0.871.33) 1.35 (1.101.65) 1.73 (1.432.10)* 0.001 1.24 (1.171.30)*
  Model 2: model 1  dietary
     intake                                     1           1.17 (0.941.46) 1.54 (1.231.92)* 2.09 (1.642.67)* 0.001 1.40 (1.301.51)*
  Model 3: model 2 
     diabetes risk factors                      1           1.09 (0.871.36) 1.31 (1.051.65) 1.58 (1.232.04)* 0.001 1.28 (1.181.39)*
  Model 4: model 3  waist
     and BMI                                    1           0.99 (0.791.23) 1.11 (0.891.40)             1.14 (0.881.47)         0.22      1.13 (1.041.22)
Vegetable protein
  Cases/at risk (n)                       245/9,524             228/9,524              235/9,523              210/9,523
  Quartile median vegetable
     protein (g/day)                          22                   26                29                           33
  Univariable                                  1            0.92 (0.761.10) 0.95 (0.791.13)             0.84 (0.701.01)         0.10      0.87 (0.760.99)
  Model 1: age and sex                         1            0.94 (0.791.13) 1.02 (0.851.22)             1.02 (0.851.23)         0.64      1.01 (0.881.15)
  Model 2: model 1  dietary
     intake                                     1           0.89 (0.731.08) 0.96 (0.771.19)             0.91 (0.701.19)         0.63      0.85 (0.691.06)
  Model 3: model 2 
     diabetes risk factors                      1           0.95 (0.781.16) 1.03 (0.831.27)             1.05 (0.801.37)         0.63      0.97 (0.781.20)
  Model 4: model 3  waist
     and BMI                                    1           0.99 (0.821.21) 1.11 (0.891.38)             1.15 (0.881.50)         0.23      1.04 (0.831.29)
Data are HRs (95% CI). *Significant at P  0.001 level. Model 1: Corrected for sex (male or female) and age at recruitment (continuous). Model 2: model 1 
energy-adjusted intake of saturated fat, monounsaturated fat, polyunsaturated fat, cholesterol, vitamin E, magnesium, fiber, and glycemic load (continuous). Model
3: model 2  energy-adjusted alcohol consumption (11, 1125, 26 50, or 50 g/day), physical activity (not active, moderately inactive, moderately active, or
active), mean systolic and diastolic blood pressure (continuous), education level (high, middle, or low), and parental history of diabetes (no, one parent, or both
parents). Model 4: model 3  BMI (20, 20 25 reference group, 2530, or 30 kg/m2), and waist circumference (continuous). Significant at P  0.05 level.


model 3 did not substantially change as-               ergy % from protein exchanged for 5 en-                 intake (HRQ4 2.15 [95% CI 1.24 3.15]
sociations for either total or animal pro-             ergy % from fat in the final model. For                 and 2.36 [1.30  4.29] for low BMI and
tein (1.50 [1.14 1.98]) nor did                       consuming 5 energy % from protein at the                waist circumference groups, respec-
adjustment for dairy intake (1.62 [1.24               expense of 5 energy % from carbohydrate,                tively), whereas there was no relation in
2.11]). Excluding participants who fol-                we observed an HR of 1.28 (1.011.61) in                obese participants. Similar results were
lowed a diet did not change the results                the final model. Similar results were ob-               found for animal protein. In addition,
(model 3, total protein 1.51 [1.112.06])              served for animal protein. We observed                  similar results were obtained when these
nor did exclusion of participants with                 no associations with consuming 5 energy                 interactions were analyzed continuously
baseline cardiovascular disease, hyper-                % from vegetable protein (Table 3).                     (Pinteraction  0.05). Correction for annual
tension, or hyperlipidemia (1.68 [1.17                     We observed borderline significant                 weight change did not change the associ-
2.43]).                                                interactions with BMI and waist circum-                 ations (low BMI group, total protein,
     Consumption of 5 energy % from                    ference (Pinteraction  0.08 for both) in the            HRQ4 2.16 [1.253.75]).
protein at the expense of 5 energy % from              relation between total protein and diabe-
fat increased diabetes risk, with an HR of             tes. For lean individuals, diabetes risk in-            CONCLUSIONS -- In this study,
1.31 (95% CI 1.06 1.61) for each 5 en-                creased with increasing total protein                   high total and animal protein intake, but

46      DIABETES CARE, VOLUME 33, NUMBER 1, JANUARY 2010                                                                               care.diabetesjournals.org
                                                                                                                                    Sluijs and Associates

Table 3--Multivariable HRs (95% CI) for the association between the consumption of 5%                         cused on consuming carbohydrate at the
energy from protein at the expense of 5% energy from fat or carbohydrates while keeping total                 expense of protein, reported similar find-
energy intake constant and incident type 2 diabetes                                                           ings (1). Yet, in that study, exchanging
                                                                                                              energy from protein for fat was not ac-
                                                            Model 3*                      Model 4             counted for, and no differentiation was
                                                                                                              made for total protein content and protein
Total protein                                                                                                 source.
  Substitution protein for fat                         1.72 (1.412.12)              1.31 (1.061.61)             Because the majority of protein intake
  Substitution protein for carbohydrates               1.91 (1.522.40)              1.28 (1.011.61)        in our study is from animal sources, one
Animal protein                                                                                                might think the association with total pro-
  Substitution protein for fat                         1.51 (1.261.82)              1.19 (0.991.44)         tein is merely driven by the association
  Substitution protein for carbohydrates               1.72 (1.392.12)              1.20 (0.971.49)         with animal protein. However, when we
Vegetable protein                                                                                             corrected the association between total
  Substitution protein for fat                         1.13 (0.671.92)              1.32 (0.822.13)         protein and diabetes for animal protein,
  Substitution protein for carbohydrates               0.97 (0.571.65)              1.17 (0.731.89)         the association attenuated but remained
Data are HRs (95% CI). *Model 3: corrected for sex (male or female), age at recruitment (continuous),         (model 3, HRQ4 1.46 [0.96 2.25]). Sim-
energy-adjusted intake of cholesterol, vitamin E, magnesium, fiber, and glycemic load (continuous), total     ilarly, adjusting total protein for several
energy intake (continuous), energy densities of fat, carbohydrates, and alcohol (per 5 energy %), physical    sources of animal protein intake, such as
activity (not active, moderately inactive, moderately active, or active), mean systolic and diastolic blood
pressure (continuous), education level (high, middle, or low), and parental history of diabetes (no, one      meat, did not explain the entire associa-
parent, or both parents). Model 4: model 3  BMI (20, 20 25 reference group, 2530, or 30 kg/m2)              tion. This finding indicates that part of the
and waist circumference (continuous). Significant at P  0.001 level. Significant at P  0.05 level.           association between total protein and di-
                                                                                                              abetes indeed seems to be explained by
                                                                                                              animal protein intake, but that a role for
not vegetable protein intake, was associ-              cannot conclude whether it is the protein              total protein cannot be excluded. For veg-
ated with increased diabetes risk. This re-            or other nutrients in meat, such as iron,              etable protein, we found an association in
lation was not explained by specific                   that promoted diabetes risk. Only one                  the same direction as that for animal pro-
protein sources such as meat or by weight              prospective study in women further in-                 tein, although this result did not reach
change during follow-up but was attenu-                vestigated which nutrients in meat (sev-               statistical significance. Different effects of
ated after adjustment for baseline adipos-             eral types of fat and protein, heme, and               amino acids in animal and vegetable pro-
ity measures. Consuming 5 energy %                     total iron) could promote diabetes (11).               teins on glucose metabolism may underlie
from protein at the expense of 5 energy %              These researchers observed no relation-                the difference found between animal
from carbohydrate or fat increased diabe-              ship with vegetable protein, consistent                and vegetable protein (22,23). Further
tes risk by 30%.                                       with our study. Animal protein intake sig-             studies addressing the effect of total
     Some aspects of the study need to be              nificantly increased diabetes risk. After              protein intake in populations with dif-
addressed. First, although we corrected                correction for BMI, this association atten-            fering intakes of protein sources are
for all possible available confounders, we             uated but remained significant, in con-                needed to establish the effects of total
cannot exclude unknown or unmeasured                   trast with our findings. Differences in                protein intake and specific protein
confounding. Second, the presence of di-               study population and range of protein in-              sources on diabetes risk.
abetes goes often undetected and may be                take might explain this difference. Unfor-                  Several mechanisms may explain the
preclinical up to 9 12 years (21). Indi-              tunately, the study did not address total              relationship between protein intake and
viduals with undetected diabetes may                   protein intake.                                        diabetes. Insulin resistance may arise, as
have been misclassified as nondiabetic in-                  We observed that both high total and              amino acids can inhibit glucose transport
dividuals, resulting in attenuated associa-            animal protein were associated with                    and phosphorylation, leading to impaired
tions. Strengths of our study include its              higher diabetes risk. Fat intake did not               glucose synthesis. Furthermore, amino
prospective design, large sample size, and             change much over the quartiles of protein              acids intervene with glucose metabolism
long follow-up. Use of validated cases of              intake, and the association was not altered            via stimulation of insulin and glucagon
diabetes minimized the presence of false-              after correction for fat intake. Moreover,             secretion and by serving as substrates for
positive cases of diabetes, reducing dilu-             after correction for meat or dairy intake,             gluconeogenesis. Although stimulation of
tion of associations.                                  the association between total and animal               insulin secretion is expected to prevent
     Thus far, it is unclear whether a po-             protein and diabetes remained, suggest-                hyperglycemia due to increased glucone-
tential harmful effect of protein on diabe-            ing a detrimental role for protein per se in           ogenesis, this process might not suffi-
tes is caused by high protein sources, such            diabetes risk. This association is further             ciently compensate in subjects with
as meat, or by protein per se. Several stud-           supported by the finding that consuming                impaired insulin secretion (6,7).
ies related higher red, mainly processed,              energy from protein at the expense of en-                   An individual's degree of insulin sen-
meat intake with increased diabetes risk               ergy from either fat or carbohydrate in-               sitivity is determined by the degree of ad-
(2,9 12). When corrections for fat intake             creased diabetes risk. We found no                     iposity. We therefore investigated
were made, associations remained (9                   difference in risk when energy from pro-               whether adiposity modified the relation
11), indicating that the association is not            tein was consumed at the expense of car-               between protein intake and diabetes. In
caused by fat intake. However, as most                 bohydrate or fat, suggesting that the                  contrast with our hypothesis, we only
studies did not further address which nu-              increase in protein itself and not the de-             found an association in lean individuals.
trients were responsible for the increased             crease in fat or carbohydrate caused this              In the EPIC-Potsdam study, a similar but
diabetes risk with high meat intake, one               effect. Only one previous study, which fo-             nonsignificant interaction with adiposity

care.diabetesjournals.org                                                                         DIABETES CARE, VOLUME 33, NUMBER 1, JANUARY 2010       47
Download Link:
Share Link: Forum Link:

More on Health

  • Picture: English Language Arts and Mathematics Tests Preliminary

    English Language Arts and Mathematics Tests Preliminary

    File Size: 387.45 KB, Pages: 6, Views: 110,459 views

    English Language Arts and Mathematics Tests Preliminary Results Spring 2015 Administration Highlights The new Ohio State Tests in English language arts and mathematics were administered for the first time during the 2014-2015 school year. These tests were produced by the Partnership for Assessment of Readiness …
  • Picture: Medicare & Reimbursement Advisor Weekly - HCPro

    Medicare & Reimbursement Advisor Weekly – HCPro

    File Size: 292.11 KB, Pages: 6, Views: 88,487 views

    Medicare & Critical insight for Reimbursement managed markets Advisor Weekly professionals 1 2 3 Contents | Survey snapshot | Reporter's notebook | Contact us 4 5 6 Infusion reimbursement drop would shift Medicare July 15th, 2009 Vol. 5 No. 23 patients to hospitals In this …
  • Picture: Employment Application - Cummins

    Employment Application – Cummins

    File Size: 206.80 KB, Pages: 6, Views: 12,807 views

    Employment Application Cummins Location: Today's Date: Reviewed By: Date available to start work: Print name in full: LAST FIRST MIDDLE Social Security No.: Salary expected: Present address: City: State & Zip Code: Previous addresses over the past 10 years: Email address: Phone No. & Area …
  • Picture: Live your life, enjoy your life

    Live your life, enjoy your life

    File Size: 9,988.82 KB, Pages: 9, Views: 3,832 views

    Live your life, enjoy your life A guide to using your health plan In a hurry? Pages 45 shows where to find your medical plan information online quickly at myuhc.com. Welcome We're glad you're here. While no one can predict t the future, , you …
  • Picture: Fetal Pig Dissection Lab - Quia

    Fetal Pig Dissection Lab – Quia

    File Size: 807.23 KB, Pages: 11, Views: 11,554 views

    Fetal Pig Dissection Labs Dr. J. Lim Objective: In this exercise you will examine the organization of the many body systems studied this semester in the context of a single specimen, the fetal pig. Be sure to identify the major organs as you explore the …

Leave a Reply

Your email address will not be published. Required fields are marked *