The Journal of Nutrition

Nutrient Physiology, Metabolism, and Nutrient-Nutrient Interactions

The Effects of Fat and Protein on Glycemic

Responses in Nondiabetic Humans Vary with
Waist Circumference, Fasting Plasma Insulin,
and Dietary Fiber Intake1
Elham Moghaddam, Janet A. Vogt, and Thomas M. S. Wolever2*

Downloaded from https://academic.oup.com/jn/article-abstract/136/10/2506/4746688 by guest on 10 June 2019

Department of Nutritional Sciences, University of Toronto, Toronto, Ontario M5S 3E2 Canada

Abstract
The effects of protein and fat on glycemic responses have not been studied systematically. Therefore, our aim was to
determine the dose-response effects of protein and fat on the glycemic response elicited by 50 g glucose in humans and
whether subjects’ fasting plasma insulin (FPI) and diet influenced the results. Nondiabetic humans, 10 with FPI $40
pmol/L and 10 with FPI .40 pmol/L, were studied on 18 occasions after 10 14-h overnight fasts. Subjects consumed 50 g
glucose dissolved in 250 mL water plus 0, 5, 10, or 30 g fat and/or 0, 5, 10, or 30 g protein. Each level of fat was tested with
each level of protein. Dietary intake was measured using a 3-d food record. Gram per gram, protein reduced glucose
responses ;2 times more than fat (P , 0.001) with no significant fat 3 protein interaction (P ¼ 0.051). The effect of protein
on glycemic responses was related to waist circumference (WC) (r ¼ 20.56, P ¼ 0.011) and intake of dietary fiber (r ¼
20.60, P ¼ 0.005) but was unrelated to FPI or other nutrient intakes. The effect of fat on glycemic responses was related
to FPI (r ¼ 0.49, P ¼ 0.029) but was unrelated to WC or diet. We conclude that, across the range of 0–30 g, protein and fat
reduced glycemic responses independently from each other in a linear, dose-dependent fashion, with protein having
;3-times the effect of fat. A large protein effect was associated with high WC and high dietary-fiber intake, whereas a
large fat effect was associated with low FPI. These conclusions may not apply to solid meals. Further studies are needed
to determine the mechanisms for these effects. J. Nutr. 136: 2506–2511, 2006.

Introduction
However, the effects of fat and protein on postprandial re-
It is generally accepted that adding fat and protein to carbohy- sponses have not been studied systematically and their effects
drate reduces glycemic responses by delaying gastric emptying in mixed meals cannot be predicted reliably. Normal meals
and stimulating insulin secretion (1,2). These effects have a contain both fat and protein, but it is not known if there is an
number of possible implications for human nutrition, such as interaction between their effects. Also, there is evidence that fat
supporting the role of high protein or high fat diets in the and protein may not have the same effects in all situations. For
management of diabetes (3,4) or being a source of criticism for example, studies in normal humans suggest that the dose re-
the application of the glycemic index to mixed meals (1,5). sponse is not linear (6–8); thus, adding more fat and/or protein
to a carbohydrate meal that already contains some may have
little or no effect. The glucose-lowering effects of fat and protein
1
are attenuated or absent in subjects with diabetes (9,10), sug-
Supported by the Natural Sciences and Engineering Research Council of
Canada (NSERC). E.M. was supported by an Industrial Research Scholarship
gesting that the same may occur in insulin-resistant subjects
from NSERC and Glycaemic Index Testing, Inc. without diabetes. A high-fat diet potentiates the effect of oral
2
T.M.S.W. is President and part-owner of Glycaemic Index Testing, Inc., a glucose on gastric inhibitory polypeptide secretion (11) and
corporation that provides laboratory services and educational and promotional influences gastrointestinal transit (12,13), suggesting that the
activities related to the glycemic index, and President and part-owner of
acute glucose-lowering effects of fat may be influenced by
Glycemic Index Laboratories, Inc., a corporation that does contract research.
T.M.S.W. is the coauthor of a range of books on glycemic index under the habitual fat intake.
general title of The Glucose Revolution: Authoritative Guide to the Glycemic Thus, the aims of this pilot, dose-finding study were to
Index, published by Marlowe & Co, NY; and is also author of a scientific book determine the dose-response effects of 0–30 g protein and
entitled The Glycaemic Index: A Physiological Classification of Dietary fat, alone and in combination, on the glycemic response elic-
Carbohydrate, published by CABI, UK. The other authors declare no conflicts
of interest.
ited by 50 g glucose in nondiabetic humans and to see whether
* To whom correspondence should be addressed. E-mail: thomas.wolever@ the effects were influenced by subjects’ insulin sensitivity and
utoronto.ca. diet.
2506 0022-3166/06 $8.00 ª 2006 American Society for Nutrition.
Manuscript received 1 May 2006. Initial review completed 2 June 2006. Revision accepted 26 July 2006.
Materials and Methods protein) to reduce postprandial glucose in each subject was defined as the
slope of the regression line of RGR on dose of fat (or protein). The
Healthy men and nonpregnant, nonlactating women aged 18–60 y, regression equations were based on 16 points in each subject (4 3 4
recruited from the University of Toronto campus, were screened by levels of fat and protein). It was considered valid to pool the data in this
having their height, weight, waist circumference (WC),3 blood pressure, way because there was no significant fat 3 protein or FPI 3 fat 3 protein
fasting plasma insulin (FPI) and glucose, serum total and HDL cho- interaction and no significant evidence of a nonlinear dose-response
lesterol and triglycerides measured (14,15); and LDL cholesterol cal- relation. Correlations between these slopes and other variables were
culated (16). Exclusion criteria were: history of diabetes; use of diuretics, determined by simple and multiple linear regression analysis (Lotus 123
corticosteroids or b-blockers; BMI . 30 kg/m2; fasting glucose $7.0 97 Edition, Lotus Development). Differences were considered signifi-
mmol/L; or triglycerides $ 10.0 mmol/L. Ten (10) hyperinsulinemic cant if 2-tailed P , 0.05. The significance of differences between
(hyper[I]) subjects and 10 controls completed the study. We classified individual means was assessed using Tukey’s test to control for multiple
subjects by FPI, because in nondiabetic subjects, FPI is related to insulin comparisons.
sensitivity as measured by the euglycemic, hyper[I] clamp (17) or the
minimal model (18). Hyper[I] was defined as FPI . 40 pmol/L, rep-
resenting ;75th percentile of FPI in healthy volunteers in our hands
(14,19). The protocol was approved by the University of Toronto Re- Results
search Ethics Board, and all subjects gave their informed consent to

Downloaded from https://academic.oup.com/jn/article-abstract/136/10/2506/4746688 by guest on 10 June 2019

Hyper[I] subjects were similar to control with respect to sex,
participate by signing the approved consent form.
ethnicity, age, height, weight, BMI, and percent body fat (Table
Subjects were studied on 18 mornings after 10 14-h overnight fasts.
On each day, subjects gave a fasting capillary finger-prick blood sample,
1). The 2 groups had similar fasting glucose, but WC, FPI, and
consumed a liquid test meal within 10 min, and gave further capillary HOMA were significantly higher in hyper[I] than controls.
blood samples at 15, 30, 45, 60, 90, and 120 min after starting to eat. Mean HOMA in hyper[I], 14 pmol 3 mmol, was similar to
Blood (2–3 drops) was placed into fluro-oxalate tubes, mixed by multi-ethnic, first-degree relatives of people with type 2 diabetes
rotation, and frozen prior to analysis of whole blood glucose using a YSI with normal (n ¼ 240) or impaired (n ¼ 191) glucose tolerance,
Model 2300 STAT analyzer. 12, and 16 pmol 3 mmol, respectively (24). Hyper[I] had sig-
The 16 different test meals consisted of 50 g anhydrous glucose nificantly lower HDL cholesterol and higher total and LDL
(Sigma Chemical); 250 mL tap water; 0, 5, 10, or 30 g fat (0, 5, 10, or cholesterol, triglycerides, and total:HDL cholesterol ratio than
30 g corn oil; Mazola, ACH Food); and 0, 5, 10, or 30 g protein (0, 5.6, controls. None of the control or hyper[I] subjects had the
11.1, or 33.3 g soy protein concentrate; Supro, Swiss Herbal Remedies;
metabolic syndrome using the National Cholesterol Education
90 g protein, 0 g carbohydrate, and 2.7 g fat/100 g) blended to a smooth
drink prior to consumption. Subjects took 250 mL water with each test
Program Adult Treatment Panel III definition (25). One hyper[I]
meal. Test meals were administered according to a randomized block subject had metabolic syndrome according to the International
design; each block consisted of 1 level of fat (g of fat, 0, 5, 10 or 30, Diabetes Federation consensus definition (26).
respectively, added to 50 g glucose [F0, F5, F10, or F30]) with each of the Mean blood glucose after the different test meals did not
4 levels of protein (g of protein, 0, 5, 10 or 30, respectively, added to 50 g differ significantly between control and hyper[I] subjects (Fig. 1);
glucose [P0, P5, P10, and P30]). The order of the blocks and the order of however, AUC, RGR, and the PR differed significantly between
tests within blocks were randomized. Glucose alone (F0P0) was added to the individual test meals (Table 2). There were significant main
2 other blocks so that each subject tested F0P0 3 times. effects of fat and protein on AUC, RGR, and PR (Table 3). AUC,
Nutrient intakes were estimated using 3-d diet records (2 weekdays RGR, and PR after 30 g fat were significantly less than after 5 g.
and 1 weekend day), filled out during the experimental period, and
analyzed using Food Processor SQL edition, version 9.3.1m (ESHA
Research).
Results are presented as means 6 SEM. Insulin sensitivity was
estimated using the homeostasis assessment model (HOMA; 20). Body TABLE 1 Details of subjects studied1
fat percentage was estimated from age, sex, BMI, and ethnicity using
Controls, Hyper[I],
regression equations developed in different ethnic populations (21).
Adjustment of screening parameters for sex was done by the residuals FPI #40 pmol/L FPI .40 pmol/L
method using dummy variables (1 ¼ female; 0 ¼ male). Peak rise
Sex (males:females), n:n 4:6 6:4
(PR) was the maximum glucose concentration achieved minus fasting
Ethnicity (SA:EA:C)2, n:n:n 2:6:2 4:3:3
glucose. Incremental areas under the curve (AUC), ignoring area below
fasting, were calculated as previously described (22). AUC after each test Age, y 24.5 6 1.0 27.4 6 1.9
meal was expressed as a percentage of the mean AUC after F0P0 in each Height3, cm 169 6 2 170 6 1
subject and the resulting values termed relative glycemic response Wt3, kg 62.7 6 2.5 66.8 6 3.3
(RGR). RGR was the primary outcome. RGR, AUC, and PR values were BMI, kg/m2 21.5 6 0.7 23.1 6 1.0
subjected to ANOVA for 3-factor experiments with repeated measures Body fat3, % 25.3 6 0.9 27.4 6 1.2
on 2 factors (23), examining for the main effects of FPI (control vs. WC3, cm 76.7 6 1.3 84.4 6 3.0*
hyper[I]), fat dose, protein dose, fat 3 protein interaction, and FPI 3 Fasting plasma glucose, mmol/L 4.74 6 0.10 5.04 6 0.15
fat 3 protein interaction. The shape of the dose-response relations was FPI, pmol/L 27 6 3 76 6 10*
assessed by testing if there was significant reduction of residual variation
HOMA, pmol 3 mmol 4.9 6 0.5 14.0 6 1.6*
when a term for dose2 of fat (or protein) was added to the linear
Systolic blood pressure, mm Hg 110 6 1 114 6 2
regression model (Prism 4 for Windows, GraphPad Software). A
significant effect would indicate that a nonlinear (quadratic) model fit Diastolic blood pressure, mm Hg 59 6 1 66 6 3
the data significantly better than a linear model. The ability of fat (or Serum total cholesterol, mmol/L 3.83 6 0.12 4.63 6 0.25*
Serum HDL cholesterol3, mmol/L 1.58 6 0.10 1.32 6 0.06*
Serum LDL cholesterol, mmol/L 1.85 6 0.11 2.77 6 0.23*
3
Abbreviations used: AUC, incremental area under the curve; F0, F5, F10, F30, g Total:HDL cholesterol ratio3 2.54 6 0.20 3.67 6 0.29*
of fat (0, 5, 10 or 30, respectively) added to 50 g glucose; FPI, fasting plasma Serum triglycerides3, mmol/L 0.80 6 0.09 1.22 6 0.10*
insulin; GLP-1, glucagon-like peptide-1; HOMA, homeostasis assessment model
1
(an index of insulin resistance); Hyper[I], hyperinsulinemic (FPI .40 pmol/L); P0, Values ¼ means 6 SEM. * Different from controls, P , 0.05.
2
P5, P10, P30, g of protein (0, 5, 10 or 30, respectively) added to 50 g glucose; PR, SA ¼ South Asian; EA ¼ East Asian; C ¼ Caucasian.
3
peak rise; RGR, relative glycemic response; WC, waist circumference. Values adjusted for sex (significant effect of sex).

Effect of fat and protein on glycemic responses 2507

 TABLE 2 AUC, RGR, and peak rises in whole blood glucose
in nondiabetic humans after they consumed 50 g
glucose plus different levels of fat and protein1

Fat, g Protein, g AUC, mmol 3 min/L RGR, % Peak Rise, mmol/L

0 0 301 6 21a 100 6 2a 4.5 6 0.2ab

5 0 314 6 25a 106 6 6a 4.7 6 0.3a
10 0 297 6 26a 100 6 6a 4.3 6 0.3ab
30 0 292 6 24a 99 6 5a 4.4 6 0.3ab
0 5 283 6 23a 96 6 5a 4.2 6 0.3ab
5 5 307 6 21a 106 6 6a 4.5 6 0.3ab
10 5 261 6 24abc 91 6 7abc 3.9 6 0.3bc
30 5 261 6 26abc 88 6 7abc 3.8 6 0.4bcd
0 10 276 6 18ab 94 6 5ab 4.2 6 0.3ab
5 10 287 6 28a 98 6 8a 4.2 6 0.4ab

Downloaded from https://academic.oup.com/jn/article-abstract/136/10/2506/4746688 by guest on 10 June 2019

10 10 257 6 21abcd 87 6 5abc 3.9 6 0.3bc
30 10 217 6 24bcde 72 6 6bcd 3.1 6 0.3cde
0 30 199 6 17cde 68 6 4cd 3.1 6 0.2cde
5 30 192 6 19de 63 6 3d 3.0 6 0.2def
10 30 182 6 20e 61 6 5d 2.9 6 0.2ef
30 30 166 6 19e 57 6 7d 2.2 6 0.2f
LSD2 65 23 0.8
1
Values are means 6 SEM, n ¼ 20. Means in a column without a common letter
differ, P , 0.05.
2
LSD ¼ Tukey’s Least Significant Difference.

FPI (Fig. 3) or HOMA. Fat and protein slopes were not related to
percent body fat.
Fat slope was not related to dietary fat intake. Protein slope,
Figure 1 Blood glucose responses after 50 g oral glucose plus 0, 5, 10, or 30 g expressed as DRGR/g, was significantly related to dietary fiber
fat and 0, 5, 10, or 30 g protein in nondiabetic humans with FPI # 40 pmol/L intake, whether expressed in g (r ¼ 20.67, P ¼ 0.0014) or g/kJ
(control) and FPI . 40 pmol/L (hyper[I]). Values are means 6 SEM, n ¼ 10. Error
(Fig. 4). In a multiple regression model, 75% of the variation in
bars not shown if they are smaller than the symbol or overlap other bars or
symbols. protein slope was explained by fiber (F, g/kJ) and WC (cm) as
follows: DRGR/g protein ¼ 1.59 2 0.34 3 F 2 0.028 3 WC.
The effects of F and WC were independent and significant (both
AUC, RGR, and PR after 30 g protein were significantly less P , 0.0001).
than after 0, 5, and 10 g (Fig. 2). Dose of fat was linearly
correlated with RGR (r ¼ 20.33, P ¼ 0.0025) and the nonlinear
(quadratic) model was not a significantly better fit (P ¼ 0.56).
Similarly, dose of protein was correlated with RGR (r ¼ 20.70,
P , 0.0001) and the quadratic model was not significantly better TABLE 3 Results of ANOVA for AUC, RGR, and PR values
(P ¼ 0.94). Mean changes in glycemic response per g protein in whole blood glucose in nondiabetic humans
were 2–3 times greater than those per g fat (P , 0.001; Fig. 2). (n ¼ 10 controls and n ¼ 10 hyper[I]) who
There were no significant differences in AUC, RGR, or PR consumed test meals consisting of 50 g glucose
between control or hyper[I] subjects (Table 3). Although there plus different levels of fat and protein.
were significant fat 3 subject and protein 3 subject interactions,
AUC RGR PR
the group 3 fat, group 3 protein, and group 3 fat 3 protein
interactions were not significant, indicating that differences P values
Main effects
between subjects were not explained by their classification into
Fat 0.029 0.017 0.0018
control and hyper[I] groups.
Protein ,0.001 ,0.001 ,0.001
The ability of fat or protein to reduce glycemic responses is
Group1 0.53 0.93 0.39
termed fat slope or protein slope, respectively. Fat and protein
Subjects2 ,0.001 , 0.001 ,0.001
slopes in all 20 subjects were not related to each other whether
expressed as RGR (r ¼ 20.11, P ¼ 0.65) or AUC (r ¼ 20.23, Interactions
P ¼ 0.33). Fat slope was related to FPI but not WC, whether Fat 3 protein 0.71 0.51 0.11
expressed in relative (DRGR/g fat, Fig. 3) or absolute terms Fat 3 group 0.78 0.71 0.74
(DAUC/g vs. FPI, r ¼ 0.52, P ¼ 0.018; DAUC/g vs. WC, r ¼ 0.27, Protein 3 group 0.80 0.92 0.72
P ¼ 0.25). The correlations between HOMA and DRGR/g fat Fat 3 protein 3 group 0.11 0.34 0.11
(r ¼ 0.44, P ¼ 0.051) and between HOMA and DAUC/g fat (r ¼ Fat 3 subjects 0.001 0.010 ,0.001
0.48, P ¼ 0.034) were weaker than those for FPI. Protein slope Protein 3 subjects ,0.001 ,0.001 ,0.001
was related to WC when expressed as DRGR/g (Fig. 3) but not as 1
Control group vs. hyper[I] group.
DAUC/g (r ¼ 20.30, P ¼ 0.19); protein slope was not related to 2
Individual subjects vs. each other, n ¼ 20.

2508 Moghaddam et al.

Figure 2 Main effects of fat (black circles and bars) and protein (white circles
and bars) on glucose responses expressed as AUC (A), RGR (B), and PR (C) after
nondiabetic humans consumed 50 g glucose plus different levels of fat and
protein. Means not sharing a common superscript letter differ, P , 0.05 (xy for Figure 4 Relation between dietary fiber intake and protein slope in individual
comparisons of different levels of fat and ab for comparisons of different levels subjects. Protein slope is the extent to which protein, added to 50 g glucose,
of protein). Insets: slopes of regression lines of AUC, RGR, and PR on g of fat or reduces RGR expressed as a percentage of the response elicited by glucose

Downloaded from https://academic.oup.com/jn/article-abstract/136/10/2506/4746688 by guest on 10 June 2019

protein. *Mean slope for protein differs from slope for fat, P , 0.001. Values are alone. The regression line is shown.
means 6 SEM, n ¼ 20.

A number of our conclusions are based on using linear

Discussion regressions as quantitative estimates of the effects of fat and
protein on glycemic responses in individual subjects. The linear
The results showed that both protein and fat reduced the
model was considered valid because there was no evidence for
glycemic response elicited by oral glucose in normal humans.
nonlinear relations. The validity of pooling the results for the
The effects of protein and fat were independent of each other,
different levels of 1 nutrient (i.e. protein or fat) to determine the
but gram-for-gram, protein had a 2 to 3 times larger effect than
effect of the other (i.e. fat or protein) depends on there being no
fat. Fat reduced glycemic responses to a greater extent in subjects
nutrient 3 nutrient interaction. The fat 3 protein interaction
with low FPI, whereas protein had more effect in subjects with a
was not significant, but interactions are more difficult to detect
high WC and a high intake of dietary fiber.
than main effects, and the lack of significant difference does not
Previous studies suggest that adding fat and protein to
mean that no effect exists. Thus, we cannot rule out the existence
carbohydrate reduces glycemic responses nonlinearly, with the
of a fat 3 protein interaction; but, if it exists, it is likely to
glycemic impact reaching a plateau as more and more protein
be small, and a larger number of subjects will be required to
and fat are added (6,8). However, we found no evidence for a
detect it.
nonlinear dose response over the range of doses used (0–30 g).
The effect of fat on glycemic responses did not differ in
This may have been due in part to a lack of statistical power or
hyper[I] vs. control subjects, but there was a significant cor-
to differences in study design. We previously found that adding
relation between the fat effect and FPI in individual subjects.
margarine to bread reduced glucose PR in a significantly non-
This suggests that our definition of hyper[I], i.e. FPI .40
linear fashion (8); by contrast, in this study, liquid test meals
pmol/L, was too low and that an abnormal fat effect may occur
were used. Because of differences in how solids and liquids
only when FPI . 80 pmol/L, which is the mean FPI in people
empty from the stomach (27), the shape of the dose-response
with newly discovered diabetes (24). This is consistent with
curve for added fat may differ for solid and liquid meals.
studies showing that fat has no effect on glycemic responses in
subjects with diabetes (9). Taken together, the results suggest
that insulin resistance rather than diabetes per se may be
responsible for the lack of fat effect. Our finding that fiber intake
and WC modulated the glucose-lowering effect of protein may
help explain the inconsistent effects of protein reported in the
literature (6,7), although other factors, such as protein digest-
ibility (28) and differences in the ability of specific amino acids
to stimulate insulin (29), may also be important.
Adding fat and protein to carbohydrate is thought to re-
duce glycemic responses by similar mechanisms (1,2), namely,
delayed gastric emptying (3,30), mediated by gut hormones such
as gastric inhibitory polypeptide and glucagon-like peptide-
1 (GLP-1) (31–34), and increased insulin secretion mediated by
direct effects of fatty and amino acids (35,36) and/or by the
hormones of the entero-insulin axis (37). However, the lack of
correlation between fat slopes and protein slopes in the different
subjects and the 2–3 times greater effect of protein than fat,
suggest that the mechanisms by which fat reduces glycemic
responses differ from those for protein.
We expected the glucose-lowering effects of both fat and
protein to be smaller in hyper[I] subjects than controls, because
Figure 3 Relations between fat (top) and protein slopes (bottom) in individual these effects are thought to be mediated by gut hormones that
subjects and their FPI (left) and WC (right) at screening. Fat or protein slope is
the extent to which fat or protein, added to 50 g glucose, reduces RGR
are reduced in insulin resistance and obesity (38–40). Also,
expressed as a percentage of the response elicited by glucose alone. Lines are we anticipated that any effects related to FPI would also be re-
regression lines. lated to WC, a marker of abdominal obesity, because insulin
Effect of fat and protein on glycemic responses 2509
resistance and hyperinsulinemia are closely associated with 10. Wolever TMS. The glycaemic index: a physiological classification of di-
abdominal obesity (41). Thus, our findings that the fat slopes etary carbohydrate. Wallingford (UK): CABI Publishing; 2006:83–115.
were positively related to FPI but not WC, whereas the protein 11. Morgan LM, Hampton SM, Tredger JA, Cramb R, Marks V. Modifi-
cations of gastric inhibitory polypeptide (GIP) secretion in man by a
slopes were negatively related to WC but not FPI, were un- high-fat diet. Br J Nutr. 1988;59:373–80.
expected. Indeed, this suggests that abdominal obesity and hyper- 12. Morgan LM, Tredger JA, Hampton SM, French AP, Peake JC, Marks V.
insulinemia have distinct metabolic implications, an idea that The effect of dietary modification and hyperglycaemia on gastric
has been proposed from the positive association between FPI emptying and gastric inhibitory polypeptide (GIP) secretion. Br J Nutr.
and increased risk of cardiovascular disease independent 1988;60:29–37.
of variation in WC (41). Our results are hard to explain but 13. Cunningham KM, Daly J, Horowitz M, Read NW. Gastrointestinal
adaptation to diets of differing fat composition in human volunteers.
could be consistent with the hypothesis that fat reduces glucose
Gut. 1991;32:483–6.
responses via GLP-1 mediated effects on gastric emptying,
14. Wolever TMS, Campbell JE, Geleva D, Anderson GH. High-fiber cereal
whereas protein reduces glucose due to amino-acid mediated reduces postprandial insulin responses in hyperinsulinemic but not
effects on insulin secretion. However, the results of this pilot normoinsulinemic subjects. Diabetes Care. 2004;27:1281–5.
study do not allow anything but speculation as to the mecha- 15. Tsihlias EB, Gibbs AL, McBurney MI, Wolever TMS. Comparison of
nisms involved, and further studies are needed to investigate high- and low-glycemic index breakfast cereals versus monounsaturated

Downloaded from https://academic.oup.com/jn/article-abstract/136/10/2506/4746688 by guest on 10 June 2019

how fat and protein affect postprandial responses in normal and fat in the long-term dietary management of type 2 diabetes. Am J Clin
Nutr. 2000;72:439–49.
hyper[I] subjects.
16. Friedwald WT, Levy RJ, Frederickson DS. Estimation of concentration
The results did not support our hypothesis that the acute
of low-density lipoprotein cholesterol in plasma without the use of the
glucose-lowering effect of fat is reduced by a high fat intake. preparative ultracentrifuge. Clin Chem. 1972;18:499–502.
However, unexpectedly, there was a strong relation between 17. Laakso M. How good a marker is insulin level for insulin resistance?
dietary fiber intake and protein slope, such that a high fiber Am J Epidemiol. 1993;137:959–65.
intake was associated with a larger protein effect (Fig. 4). A po- 18. Clausen JO, Borch-Johnsen K, Ibsen H, Bergman RN, Hougaard P,
tential explanation, based on animal studies, is that a high fiber Winther K, Pedersen O. Insulin sensitivity index, acute insulin response,
diet upregulates GLP-1 secretion, due to colonic short chain and glucose effectiveness in a population-based sample of 380 young
healthy caucasians: analysis of the impact of gender, body fat, physical
fatty acids produced by the colonic fermentation of dietary fiber fitness and life-style factors. J Clin Invest. 1996;98:1195–209.
(42). Because GLP-1 increases b-cell mass (43), a high fiber 19. Dworatzek PDN, Hegele RA, Wolever TMS. Associations between
intake may be associated with increased insulin secretion in metabolic phenotypes vary with the codon 54 polymorphism of fatty
response to protein ingestion. However, it is not known if fiber acid-binding protein 2 gene. Can J Diabetes. 2002;26:349–55.
has these effects in humans. 20. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF,
We conclude that, across the range of 0–30 g, protein and fat Turner RC. Homeostasis model assessment: insulin resistance and b-cell
function from fasting plasma glucose and insulin concentrations in man.
reduce the glycemic response elicited by oral glucose indepen-
Diabetologia. 1985;28:412–9.
dently from each other, with a linear dose response and a 2–3
21. Deurenberg-Yap M, Schmidt G, van Staveren WA, Deurenberg P. The
times greater effect from protein than fat. High fiber intake and paradox of low body mass index and high body fat percentage among
high WC were associated with an increased effect of protein, Chinese, Malays and Indians in Singapore. Int J Obes. 2000;24:1011–7.
whereas a high FPI was associated with a reduced effect of fat. 22. Brouns F, Bjorck I, Frayn KN, Gibbs AL, Lang V, Slama G, Wolever
These conclusions may not apply to solid meals. Further studies TMS. Glycaemic index methodology. Nutr Res Rev. 2005;18:145–71.
are needed to determine the mechanisms for these effects. 23. Cody RP, Smith JK. Applied statistics and the SAS programming
language. 4th ed. Upper Saddle River (NJ): Prentice Hall; 1997.
24. Jensen CC, Cnop M, Hull RL, Fujimoto WY, Kahn SE, and American
Diabetes Association GENNID Study Group. b-cell function is a major
Literature Cited contributor to oral glucose tolerance in high-risk relatives of four ethnic
1. Pi-Sunyer FX. Glycemic index and disease. Am J Clin Nutr. 2002;76: groups in the US. Diabetes. 2002;51:2170–8.
Suppl:S290S–8. 25. Expert Panel on Detection, Evaluation, and Treatment of High Blood
2. Nuttall FQ, Gannon MC. Plasma glucose and insulin response to mac- Cholesterol in Adults. Executive summary of the third report of the
ronutrients in nondiabetic and NIDDM subjects. Diabetes Care. 1991; national cholesterol education program (ncep) expert panel on detec-
14:824–38. tion, evaluation, and treatment of high blood cholesterol in adults
(Adult Treatment Panel III). JAMA. 2001;285:2486–97.
3. Welch IM, Bruce C, Hill SE, Read NW. Duodenal and ileal lipid
suppresses postprandial blood glucose and insulin responses in man: 26. International Diabetes Federation [homepage on the Internet]. The IDF
possible implications for the dietary management of diabetes mellitus. consensus worldwide definition of the metabolic syndrome. City: In-
Clin Sci. 1987;72:209–16. ternational Diabetes Federation; 2005 [cited 2005 Sept 19]. Available
from: www.idf.org/webdata/docs/Metabolic_syndrome_definition.pdf.
4. Gannon MC, Nuttall FQ, Saeed A, Jordan K, Hoover H. An increase in
27. Mathus-Vliegen EM, Van Ierland-Van Leeuwen ML, Roolker W.
dietary protein improves the blood glucose response in persons with
Gastric emptying, CCK release, and satiety in weight-stable obese sub-
type 2 diabetes. Am J Clin Nutr. 2003;78:734–41.
jects. Dig Dis Sci. 2005;50:7–14.
5. Flint A, Moller BK, Raben A, Pedersen D, Tetens I, Holst JJ, Astrup A.
28. Nuttall FQ, Gannon MC. Metabolic response to egg white and cottage
The use of glycaemic index tables to predict glycaemic index of com-
cheese protein in normal subjects. Metabolism. 1990;39:749–55.
posite breakfast meals. Br J Nutr. 2004;91:979–89.
29. van Loon LJC, Saris WHM, Verhagen H, Wagenmakers AJM. Plasma
6. Spiller GA, Jensen CD, Pattison TS, Chuck CS, Whittam JH, Scala J. insulin responses after ingestion of different amino acid or protein
Effect of protein dose on serum glucose and insulin response to sugars. mixtures with carbohydrate. Am J Clin Nutr. 2002;72:96–105.
Am J Clin Nutr. 1987;46:474–80.
30. Lang V, Bellisle F, Alamowitch C, Craplet C, Bornet FRJ, Slama G, Guy-
7. Westphal SA, Gannon MC, Nuttall FQ. Metabolic response to glucose Grand B. Varying the protein source in mixed meals modifies glucose,
ingested with various amounts of protein. Am J Clin Nutr. 1990;52: insulin and glucagon kinetics in healthy men, has weak effects on subjective
267–72. satiety and fails to affect food intake. Eur J Clin Nutr. 1999;53:959–65.
8. Owen B, Wolever TMS. Effect of fat on glycaemic responses in normal 31. Rocca AS, Brubaker PL. Stereospecific effects of fatty acids on
subjects: a dose-response study. Nutr Res. 2003;23:1341–7. proglucagon-derived peptide secretion in fetal rat intestinal cultures.
9. Gannon MC, Ercan N, Westphal SA, Nuttall FQ. Effect of added fat on Endocrinology. 1995;136:5593–9.
plasma glucose and insulin response to ingested potato in individuals 32. Herrmann C, Göke R, Richter G, Fehmann H-C, Arnold R, Göke B.
with NIDDM. Diabetes Care. 1993;16:874–80. Glucagon-like peptide-1 and glucose-dependent insulin-releasing

2510 Moghaddam et al.

 polypeptide plasma levels in response to nutrients. Digestion. 1995;56: 39. Meier JJ, Hückling K, Holst JJ, Deacon C, Schmiegel W, Nauck MA.
117–26. Reduced insulinotropic effect of gastric inhibitory polypeptide in
33. Morgan LM. The role of gastrointestinal hormones in carbohydrate and patients with type 2 diabetes mellitus. Diabetes. 2001;50:2497–504.
lipid metabolism and homeostasis: effects of gastric inhibitory polypep- 40. Ranganath LR, Beety JM, Morgan LM, Wright JW, Howland R, Marks
tide and glucagon-like peptide-1. Biochem Soc Trans. 1998;26:216–22. V. Attenuated GLP-1 secretion in obesity: cause of consequence? Gut.
34. Layer P, Holst JJ. GLP-1: a humoral mediator of the ileal brake in 1996;38:916–9.
humans? Digestion. 1993;54:385–6. 41. Gaudet D, Vohl MC, Perron P, Tremblay G, Gagne C, Lesiege D,
35. Beysen C, Belcher AK, Karpe F, Fielding BA, Clarke A, Levy JC, Frayn Bergeron J, Moorjani S, Despres JP. Relationships of abdominal ob-
KN. Interaction between specific fatty acids, GLP-1 and insulin esity and hyperinsulinemia to angiographically assessed coronary artery
secretion in humans. Diabetologia. 2002;45:1533–41. disease in men with known mutations in the LDL receptor gene. Cir-
36. van Loon LJC, Saris WHM, Verhagen H, Wagenmakers AJM. Plasma culation. 1998;97:871–7.
insulin responses after ingestion of different amino acid or protein 42. Reimer RA, McBurney MI. Dietary fiber modulates intestinal proglu-
mixtures with carbohydrate. Am J Clin Nutr. 2002;72:96–105. cagon messenger ribonucleic acid and postprandial secretion of gluca-
37. Collier GR, Greenberg GR, Wolever TMS, Jenkins DJA. The acute effect gon-like peptide-1 and insulin in rats. Endocrinology. 1996;137:
of fat on insulin secretion. J Clin Endocrinol Metab. 1988;66:323–6. 3948–56.
38. Ross SA, Brown JD, Dupré J. Hypersecretion of gastric inhibitory 43. Wang Q, Brubaker PL. Glucagon-like peptide-1 treatment delays the
polypeptide following oral glucose in diabetes mellitus. Diabetes. 1977; onset of diabetes in 8-week-old db/db mice. Diabetologia. 2002;45:

Downloaded from https://academic.oup.com/jn/article-abstract/136/10/2506/4746688 by guest on 10 June 2019

26:525–9. 1263–73.

Effect of fat and protein on glycemic responses 2511