Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
ABSTRACT carbohydrate diet [mean group difference: 2.47 kg (95% CI: 0.20,
Background: Which diet is optimal for weight loss and mainte- 4.75 kg); P = 0.03]. The addition of FI strengthened these associations.
nance remains controversial and implies that no diet fits all patients. Conclusion: Elevated FPG before treatment indicates success with
Objective: We studied concentrations of fasting plasma glucose dietary weight loss and maintenance among overweight patients con-
(FPG) and fasting insulin (FI) as prognostic markers for successful suming diets with a low glycemic load or with large amounts of fiber
weight loss and maintenance through diets with different glycemic and whole grains. These trials were registered at clinicaltrials.gov as
loads or different fiber and whole-grain content, assessed in 3 ran- NCT00390637 (DiOGenes) and NCT01195610 (SHOPUS), and
domized trials of overweight participants. at ISRNCT.com as ISRCTN25867281 (NUGENOB). Am J
Design: After an 8-wk weight loss, participants in the DiOGenes Clin Nutr 2017;106:499–505.
(Diet, Obesity, and Genes) trial consumed ad libitum for 26 wk a
diet with either a high or a low glycemic load. Participants in the Keywords: glucose, insulin, precision medicine, personalized nutrition,
Optimal well-being, development and health for Danish children weight, glycemic load, glycemic index, fiber, prediabetes, diabetes
through a healthy New Nordic Diet (OPUS) Supermarket interven-
INTRODUCTION
tion (SHOPUS) trial consumed ad libitum for 26 wk the New Nor-
dic Diet, which is high in fiber and whole grains, or a control diet. In most countries, current interventions and policies have failed
Participants in the NUGENOB (Nutrient-Gene Interactions in Hu- to stop increases in overweight and obesity, resulting in the need
man Obesity) trial consumed a hypocaloric low-fat and high- for more effective and affordable preventive and management
carbohydrate or a high-fat and low-carbohydrate diet for 10 wk.
On the basis of FPG before treatment, participants were categorized Supported by the European Commission Food Quality and Safety Priority of
as normoglycemic (FPG ,5.6 mmol/L), prediabetic (FPG 5.6– the Sixth Framework Program [for the Diet, Obesity, and Genes (DiOGenes)
6.9 mmol/L), or diabetic (FPG $7.0 mmol/L). Modifications of project; contract FP6-2005-513946]. The Optimal well-being, development and
the dietary effects of FPG and FI before treatment were examined health for Danish children through a healthy New Nordic Diet (OPUS) Su-
permarket intervention (SHOPUS) study was supported by the Nordea Foun-
with linear mixed models.
dation (grant 02-2010-0389) and by sponsors who provided foods to the shop. A
Results: In the DiOGenes trial, prediabetic individuals regained a
full list of food sponsors is available at the study website (www.foodoflife.
mean of 5.83 kg (95% CI: 3.34, 8.32 kg; P , 0.001) more on the dk/shopus). The Nutrient-Gene Interactions in Human Obesity (NUGENOB)
high– than on the low–glycemic load diet, whereas normoglycemic study was supported by the European Community (contract QLK1-CT-2000-
individuals regained a mean of 1.44 kg (95% CI: 0.48, 2.41 kg; 00618). The work reported in this article was funded by grants from Gelesis Inc.
P = 0.003) more [mean group difference: 4.39 kg (95% CI: 1.76, Supplemental Figures 1–4 and Supplemental Tables 1–4 are available from
7.02 kg); P = 0.001]. In SHOPUS, prediabetic individuals lost a the “Online Supporting Material” link in the online posting of the article and
mean of 6.04 kg (95% CI: 4.05, 8.02 kg; P , 0.001) more on the from the same link in the online table of contents at http://ajcn.nutrition.org.
New Nordic Diet than on the control diet, whereas normoglycemic Address correspondence to MFH (e-mail: madsfiil@nexs.ku.dk).
individuals lost a mean of 2.20 kg (95% CI: 1.21, 3.18 kg; P , 0.001) Abbreviations used: DiOGenes, Diet, Obesity, and Genes; FI, fasting in-
sulin; FPG, fasting plasma glucose; NND, New Nordic Diet; NUGENOB,
more [mean group difference: 3.84 kg (95% CI: 1.62, 6.06 kg);
Nutrient-Gene Interactions in Human Obesity; OPUS, Optimal well-being,
P = 0.001]. In NUGENOB, diabetic individuals lost a mean of development and health for Danish children through a healthy New Nordic
2.04 kg (95% CI: 20.20, 4.28 kg; P = 0.07) more on the high-fat Diet; SHOPUS, OPUS Supermarket intervention.
and low-carbohydrate diet than on the low-fat and high-carbohydrate Received March 3, 2017. Accepted for publication June 9, 2017.
diet, whereas normoglycemic individuals lost a mean of 0.43 kg First published online July 5, 2017; doi: https://doi.org/10.3945/ajcn.117.
(95% CI: 0.03, 0.83 kg; P = 0.03) more on the low-fat and high- 155200.
Am J Clin Nutr 2017;106:499–505. Printed in USA. Ó 2017 American Society for Nutrition 499
500 HJORTH ET AL.
options. Advice to eat less and exercise more, or to count and or high–glycemic load (high carbohydrate and high glycemic
limit calorie intake, seems reasonable, but it has not been ef- index) weight-maintenance diet for 26 wk (3 other dietary reg-
fective in reversing the obesity epidemic. The struggle to find the imens from the study are described in Supplemental Figure 4).
best diet for weight management has largely failed, giving rise Dietary fat content was held constant (w30% of energy) be-
to an endless number of fad diets (1); this failure has merely tween the 2 diets. Before the initial weight loss phase, blood
taught us that amounts of specific macronutrients are of minor samples were drawn from participants in the fasted state, from
importance as long as a diet is taught with the enthusiasm and which FPG and FI were analyzed. Furthermore, study partici-
persistence to make individuals adhere to it (2, 3). This failure pants completed diaries recording amounts of weighed food for
also may imply that no diet fits all needs, which justifies a search 3 consecutive days at the end of the intervention. Height and
for biomarkers that can predict success in losing weight and weight were measured before the initial weight loss phase.
maintaining weight loss, and can allow the most efficient diet to During the weight-maintenance period, body weight was
be selected on an individual basis. measured at randomization and at weeks 2, 4, 6, 10, 14, 18, 22,
the baseline characteristics between glycemic groups were as- trial only) and random effects for subjects and sites (except for
sessed through the use of 2-sample t tests or 1-factor ANOVA SHOPUS). Results are shown as the mean weight changes from
(variables possibly transformed before analysis), or Pearson chi- baseline with 95% CIs. At the end of the study, differences in
square tests. Pearson correlations were calculated between FPG weight change from baseline between diets were compared within
and FI, and between FPG and weight change. and between each blood marker group through the use of pairwise
In each of the 3 studies separately, the differences in weight comparisons with post hoc t tests. The level of significance was set
change from baseline between glycemic and insulinemic groups at P , 0.05, and statistical analyses were conducted with the use of
(and the combination of the 2) were analyzed by means of linear STATA/SE software version 14.1 (StataCorp LLC).
mixed models, with the use of all available weight measurements
(including those from noncompleters). The linear mixed models
included the 3-way interaction diet 3 time 3 glycemic or in- RESULTS
sulinemic strata, as well as all nested 2-way interactions and the In all 3 trials, individuals with prediabetes were older and had a
main effects; the models comprised additional fixed effects in- higher body weight or BMI at baseline than did the normoglycemic
cluding age, sex, BMI, and initial weight loss (for the DiOGenes individuals (P , 0.05) (Table 1). The dietary intakes are described
502 HJORTH ET AL.
in Supplemental Table 1. The correlations between FPG and FI in Participants with low FI regained 2.27 kg (95% CI: 1.22,
the 3 trials were low but significant (r2 = 0.04–0.08; P # 0.005). 3.32 kg; P , 0.001) more consuming the high– compared with
After a median weight loss of 10.3 kg in the DiOGenes study, the low–glycemic load diet, whereas no difference was observed
prediabetic participants consuming ad libitum the high–glycemic for participants with high FI (P = 0.24). Consequently, no dif-
load diet for 26 wk regained 5.83 kg (95% CI: 3.34, 8.32 kg; ference in responsiveness to the diets were found between par-
P , 0.001) more than the group consuming the low–glycemic ticipants with high and low FI (P = 0.14) (Figure 2 and
load diet, whereas normoglycemic participants regained only Supplemental Table 3). Prediabetic participants with low FI
1.44 kg (95% CI: 0.48, 2.41 kg; P = 0.003) more (Figure 1 and consuming the high–glycemic load diet regained 7.78 kg
Supplemental Table 2). Consequently, a 4.39-kg (95% CI: 1.76, (95% CI: 4.39, 11.18 kg; P , 0.001) more than those consuming
7.02 kg; P = 0.001) difference in responsiveness to the diets was the low–glycemic load diet, whereas no difference was observed
found between normoglycemic and prediabetic participants. for normoglycemic participants with high FI [1.17 kg (95% CI:
FIGURE 2 Relative changes in body weight between diets within each of the 3 studies when stratified by FI before treatment. Data are presented as the
estimated mean weight changes from baseline for each combination of the diet 3 time 3 FI strata interaction in the linear mixed models, which were also
adjusted for age, sex, BMI, and weight loss on the low-calorie diet (for DiOGenes only), subjects, and sites (except for SHOPUS). At the end of each study,
differences in weight change from baseline between diets were compared within and between FI groups with pairwise comparisons through the use of post hoc
t tests and are presented as mean weight changes from baseline with 95% CIs. The zero line indicates no difference between diets. Data points above the zero
line favor the low–glycemic load diet (DiOGenes), the New Nordic Diet (SHOPUS), and the high-fat and low-carbohydrate diet (NUGENOB). Cutoffs were
90.3 (DiOGenes), 72.9 (SHOPUS), and 79.2 pmol/L (NUGENOB), representing the median FI of prediabetic participants within each study. USignificant
difference between the insulinemic groups, P , 0.05. xSignificant difference from zero, P , 0.05. DiOGenes, Diet, Obesity, and Genes; FI, fasting insulin;
NUGENOB, Nutrient-Gene Interactions in Human Obesity; OPUS, Optimal well-being, development and health for Danish children through a healthy New
Nordic Diet; SHOPUS, OPUS Supermarket intervention.
BASELINE GLUCOSE AND INSULIN MODIFY WEIGHT LOSS 503
20.59, 2.93 kg); P = 0.19] (Figure 3 and Supplemental Table found between normoglycemic and diabetic participants (Figure
4). The correlation between FPG and weight gain during the 1 and Supplemental Table 2). Participants with low FI lost
intervention was 20.14 (P = 0.14) with the low–glycemic load 0.42 kg (95% CI: 0.01, 0.83 kg; P = 0.046) more consuming the
diet and 0.22 (P = 0.028) with the high–glycemic load diet. low-fat and high-carbohydrate diet than the high-fat and low
In SHOPUS, prediabetic participants lost 6.04 kg (95% CI: carbohydrate diet, whereas no difference was observed for par-
4.05, 8.02 kg; P , 0.001) more consuming the 26-wk ad libitum ticipants with high FI (P = 0.84). Consequently, no difference in
NND than the control diet, whereas normoglycemic participants responsiveness to the diets was found between participants with
lost only 2.20 kg (95% CI: 1.21, 3.18 kg; P , 0.001) more. high FI and those with low FI (P = 0.33) (Figure 2 and Sup-
Consequently, a 3.84-kg (95% CI: 1.62, 6.06 kg; P = 0.001) plemental Table 3). Participants with FPG $6.4 mmol/L and low
difference in responsiveness to the diets was found between FI lost 3.06 kg (95% CI: 0.40, 5.71 kg; P = 0.02) more consuming
normoglycemic and prediabetic participants (Figure 1 and the high-fat and low-carbohydrate diet than the low-fat and high-
Supplemental Table 2). Participants with low FI lost 4.09 kg carbohydrate diet. Participants with FPG ,6.4 mmol/L and low
(95% CI: 2.91, 5.27 kg; P , 0.001) more consuming the NND FI lost 0.49 kg (95% CI: 0.08, 0.91 kg; P = 0.02) more consuming
than the control diet, whereas participants with high FI lost only the low-fat and high-carbohydrate diet (Figure 3 and Supple-
1.61 kg (95% CI: 0.28, 2.94 kg; P = 0.02) more. Consequently, a mental Table 4). The correlation between FPG and weight gain
2.48-kg (95% CI: 0.70, 4.26 kg; P = 0.006) difference in re- was 0.06 (P = 0.30) with the low-fat and high-carbohydrate diet
sponsiveness to the diets was found between participants with and 20.03 (P = 0.55) with the high-fat and low-carbohydrate diet.
high and low FI (Figure 2 and Supplemental Table 3). Pre-
diabetic participants with low FI lost 6.27 kg (95% CI: 3.51,
9.02 kg; P , 0.001) more consuming the NND than the control DISCUSSION
diet, whereas no difference was observed for normoglycemic We identified FPG as an important biomarker associated with
participants with high FI [0.10 kg (95% CI: 21.37, 1.57 kg; successful dietary weight loss and weight loss maintenance while
P = 0.89)] (Figure 3 and Supplemental Table 4). The correlation consuming a range of different hypocaloric and ad libitum diets.
between FPG and weight gain was 20.29 (P = 0.005) on the Thus, overweight or obese participants with elevated FPG
NND and 0.01 (P = 0.92) on the control diet. (i.e., prediabetic individuals) are extremely susceptible to weight
In the NUGENOB study, normoglycemic participants lost gain (regain) when consuming a high–glycemic load diet. On the
0.43 kg (95% CI: 0.03, 0.83 kg; P = 0.03) more consuming the other hand, these individuals can achieve substantial weight loss
10-wk hypocaloric low-fat and high-carbohydrate than the hy- when consuming a diet with a low glycemic load or a diet high
pocaloric high-fat and low-carbohydrate diet, whereas no dif- in fiber and whole grains, even without restricting calories.
ference was observed for prediabetic participants (P = 0.41). We previously reported a relative modest difference in weight
Diabetic participants tended to lose 2.04 kg (95% CI: 20.20, regain (2.0 kg) between the high– and low–glycemic load diets
4.28 kg; P = 0.07) more consuming the high-fat and low- after 26 wk of weight maintenance in the DiOGenes trial (9).
carbohydrate diet. Consequently, a 2.47-kg (95% CI: 0.20-, However, we now report that this difference in weight mainte-
4.75-kg; P = 0.03) difference in responsiveness to the diets was nance between the diets was .4 times larger in prediabetic than
504 HJORTH ET AL.
in normoglycemic participants (5.8 compared with 1.4 kg). being able to provide strong evidence for one or the other (2, 3, 9, 11,
Likewise, we previously reported the overall difference between 12). From our results, it seems that failure to stratify by glycemic
the NND and control diets to be 3.2 kg (11). This difference was status is likely to underestimate (9, 11) or overlook (12) specific
almost 3 times larger in prediabetic than in normoglycemic effects among prediabetic and diabetic individuals, whereas it may
participants (6.0 compared with 2.2 kg). Furthermore, we previously mask (12) or overestimate (9, 11) the effects of a specific diet among
reported a nonsignificant difference of 0.3 kg between 10-wk low- normoglycemic individuals. Therefore, we strongly encourage the
fat and high-carbohydrate and high-fat and low-carbohydrate hy- investigation of FPG as a modifier of weight loss and maintenance in
pocaloric (2600 kcal/d) diets (12). Stratifying by FPG revealed a other large dietary clinical trials in order to help find the most
2.5-kg difference in response between the 2 diets among normo- appropriate diet for individuals with differing glycemic statuses.
glycemic and diabetic participants, resulting in a small but sig- Generating evidence to support precision medicine is challeng-
nificantly larger weight loss among normoglycemic participants ing, but in randomized clinical trials, interaction testing of in-
consuming the low-fat and high-carbohydrate diet and a border- tervention effectiveness provides a potentially efficient means to do