Supplemental Material can be found at:

http://jn.nutrition.org/content/suppl/2010/02/19/jn.109.11493
4.DC1.html

The Journal of Nutrition

Nutritional Epidemiology

Dietary Protein Intake throughout Childhood

Is Associated with the Timing of Puberty13
Anke L. B. Gunther,4* Nadina Karaolis-Danckert,5 Anja Kroke,4 Thomas Remer,5 and Anette E. Buyken5
4
Department of Nutritional, Food and Consumer Sciences, Fulda University of Applied Sciences, 36039 Fulda, Germany; and 5Research
Institute of Child Nutrition, Rheinische Friedrich-Wilhelms-University Bonn, 44225 Dortmund, Germany

Abstract
Early puberty onset is associated with hormone-related cancers, but whether diet in childhood influences pubertal timing
is controversial. We examined the association of protein intake in early and mid-childhood with the ages at take-off of the
pubertal growth spurt (ATO), peak height velocity (APHV), and menarche in girls and voice break in boys using data from
the longitudinal Dortmund Nutritional and Anthropometric Longitudinally Designed Study. Among participants who
measurements between 6 and 13 y to allow estimation of ATO. Life-course plots were used to identify critical periods of
total, animal, and vegetable protein intake (percentage of total energy intake) for pubertal timing. At these ages, the
association between tertiles of protein intake (T1T3) and the outcomes was investigated using multiple linear regression
analysis. A higher total and animal protein intake at 56 y was related to an earlier ATO. In the highest tertile of animal
protein intake at 56 y, ATO occurred 0.6 y earlier than in the lowest [(mean, 95% CI) T1: 9.6, 9.49.9 vs. T2: 9.4, 9.19.7
vs. T3: 9.0, 8.79.3 y; P-trend = 0.003, adjusted for sex, total energy, breast-feeding, birth year, and paternal university
degree]. Similar findings were seen for APHV (P-trend = 0.001) and the timing of menarche/voice break (P-trend = 0.02).
Conversely, a higher vegetable protein intake at 34 and 56 y was related to later ATO, APHV, and menarche/voice break
(P-trend = 0.020.04). These results suggest that animal and vegetable protein intake in mid-childhood might be
differentially related to pubertal timing. J. Nutr. 140: 565571, 2010.

Introduction
Contemporary data suggests considerable variation in pubertal timing (1) and a secular trend toward earlier sexual
maturation may exist (2). In this context, the importance of
environmental factors is increasingly acknowledged. Among
those early-life environmental exposures that may have the
potential to influence pubertal timing, protein intake is of
particular interest. High intake levels in infancy and early
childhood seem to be associated with later obesity (3), which is
itself related to pubertal timing in both boys and girls (47).
We have previously shown that animal protein intake
[expressed as a percentage of total energy intake (%En)6] at
1
Supported by a research grant from the World Cancer Research Fund
International. The Dortmund Nutritional and Anthropometric Longitudinally
Designed Study is funded by the Ministry of Science and Research of North
Rhine Westphalia, Germany.
2
Author disclosures: A. L. B. Gunther, N. Karaolis-Danckert, A. Kroke, T. Remer,
and A. E. Buyken, no conflicts of interest.
3
Supplemental Table 1 and Supplemental Figure 1 are available with the online
posting of this paper at jn.nutrition.org.
6
Abbreviations used: APHV, age at peak height velocity; ATO, age at take-off;
DONALD, Dortmund Nutritional and Anthropometric Longitudinally Designed; %
En, percentage of total energy intake; FMI, fat mass index; IGF-1, insulin-like
growth factor 1; SDS, SD score.
* To whom correspondence should be addressed. E-mail: anke.guenther@he.
hs-fulda.de.

age 12 mo, as well as during the adiposity rebound period

(56 y), might be unfavorably related to body composition at the
age of 7 y (8). Similar critical windows of early protein
consumption may exist for pubertal timing. Previous studies,
however, yielded inconsistent results regarding an association
between protein intake and puberty markers (916). Typically,
these studies addressed later pubertal stages, such as menarche,
but an earlier menarche could theoretically represent shorter
duration of puberty only (4). Thus, it might take early markers
of puberty onset, such as age at take-off of the pubertal growth
spurt (ATO), to establish associations between childhood diet
and the actual onset of puberty.
We investigated whether protein intake at 12 and/or
1824 mo might be related to pubertal timing, or whether
diet in later childhood may be more important. Furthermore,
we were interested whether protein from different sources
could be decisive. Data were from a contemporary cohort of
healthy, free-living girls and boys for whom 1 early (ATO) and
2 later pubertal markers [age at peak height velocity (APHV)
and age at menarche/voice break in girls/boys, respectively]
could be estimated and who provided repeated dietary assessments in early and mid-childhood. Data for 24-h urinary
nitrogen excretion, a biomarker for protein intake (17), was
available in a subsample to corroborate the results from dietary
records.

0022-3166/08 $8.00 2010 American Society for Nutrition.

Manuscript received August 31, 2009. Initial review completed October 5, 2009. Revision accepted December 1, 2009.
First published online December 30, 2009; doi:10.3945/jn.109.114934.

565

Downloaded from jn.nutrition.org by guest on September 19, 2015

provided 3-d weighed dietary records at 12 mo, 1824 mo, 34 y, and 56 y, 112 had sufficient anthropometric

Methods
Study population. The Dortmund Nutritional and Anthropometric
Longitudinally Designed (DONALD) Study is an ongoing, open cohort
study conducted in Dortmund, Germany, since 1985. Details of the study
protocol are provided elsewhere (18). The study was approved by the
Ethics Committee of the University of Bonn. All assessments, described
below in detail, were performed with parental consent.
Among children who had reached adolescence by the time of this
analysis, 411 term (3742 wk gestation) singletons with a birth weight
$2500 g provided height measurements at 6 and 13 y, as well as $5
measurements between these ages, to allow estimation of ATO. Among
these, ATO was considered plausible (determined by visual inspection of
individual growth curves and by using the cut-offs ATO $5 and ,13 y)
in 376 participants. Finally, plausible dietary records at 12 mo, 1824 mo
(at least 1 of possible 2), 34 y (at least 1 of 2), and 56 y (at least 1 of 2)
and information on potential confounders were provided by 112
individuals.
The sample resulting from applying all 3 criteria, i.e. 112 children, was
used for this analysis. A further subsample of 57 children had collected
24-h urine samples at ages 34 y (at least 1 of 2) and 56 y (at least 1 of 2).
APHV could be calculated for 106 children. Information on age at
menarche or voice break was available for 92 participants.

Puberty outcome variables. ATO and APHV were determined using

the parametric Preece and Baines model 1 (21). ATO was defined as the
age at minimal height velocity at the onset of the pubertal growth spurt
(22). Best fit was determined by graphical inspection of each childs
individual growth curve and a comparison of the residual SD (random
error had to be smaller than the expected measurement error for height),
and by considering the plausibility and the distribution of the pubertal
variables estimated. Based on these 4 criteria, data from the following
age ranges were selected and entered into Preece and Baines model 1: all
measurements from age 5 to 13 y for girls and from age 6 to 13 y for
boys.
In addition to growth monitoring, age at menarche in girls and at
voice break in boys were assessed and combined to form our 3rd
pubertal measure (2).
Dietary data. In the DONALD Study, dietary intake is assessed by 3-d
weighed dietary records, as described previously (18). Parents are asked
to weigh all foods and beverages consumed by their children to the
nearest 1 g for 3 consecutive days. Parents are instructed by trained
dieticians, and semiquantitative recording using household measures is
allowed when exact weighing was not possible (e.g. foods eaten away
from home). Information on recipes or the types and brands of food
items is also requested and a dietician visits the family and the record for
completeness and accuracy.
The records are analyzed using the in-house nutrient database
LEBTAB (23). For this study, total energy (kJ/d) and protein intakes (g/d)
between 12 mo and 6 y were derived for each individual from the mean
of the 3 recording days. The reported energy intake was related to the
566

Gunther et al.

Urinary nitrogen. Starting at age 3 y, 24-h urine collections are

performed on d 3 of the 3-d weighed dietary record (20). All micturitions
are stored immediately in preservative-free, Extran-cleaned (Extran, MA
03; Merck Darmstadt), 1-L plastic containers at #2128C before they are
transported. At the institute, the containers are stored at #2208C before
being analyzed. Urinary creatinine was measured according to the
kinetic Jaffe procedure (26) with a Beckman-2 creatinine analyzer
(Beckman Instruments).
For the purpose of this analysis, completeness of 24-h urines was
checked via sex- and age-specific, body weight-related reference values of
creatinine (27). Urinary nitrogen was measured by the Kjeldahl
technique (Buechi 430 Digestor and Buechi Distillation Unit B-324).
Protein intake (g/d) was estimated under the assumption that urinary
nitrogen reflects 80% of protein ingested (17). Age- and sex-specific
reference values for protein requirements for growth were added to
allow for protein retention in childhood (28). Consistent with the dietary
record data, individual means represented protein intake at 34 and
56 y.
Statistical analysis and power considerations. To examine when
protein intake was important for the different puberty markers, lifecourse plots were constructed (29). Total, animal, and vegetable protein
intakes (from diet records) at 12 mo, 1824 mo, 34 y, and 56 y were
expressed as %En and standardized by age for comparability. They were
then entered into separate multiple linear regression models as independent variables (adjusting for each other) and with ATO, APHV, or age at
menarche/voice break as outcomes. All models additionally included sex
and the childrens birth year to exclude a potential secular effect in
pubertal timing. Furthermore, we added mean energy intake from 1 to
6 y: the individual energy intakes were standardized by age (mean = 0,
SD = 1) and the mean was calculated for each child. The resulting
regression coefficients were then plotted against age. Both their values
(representing the strength of the associations at the single time points)
and their changes (representing the associations between puberty
markers and change in the dietary variables over the corresponding
time interval) were examined to identify important time points and/or
periods of protein intake for pubertal timing.
Subsequently, we concentrated on the critical ages identified by the
life-course plots. Protein intakes (%En) at those time points were
grouped into tertiles. Adjusted mean outcome levels were calculated for
each tertile and P-values for linear trends derived from multiple linear
regression models (separate model for each continuous protein variable).
The Dunnett test for post hoc comparisons was used to examine whether
a significant difference to the lowest tertile existed. Tests for interaction
were conducted to evaluate whether associations differed by sex. Except
for vegetable protein intake at age 56 y and ATO and APHV (P = 0.02
for both outcomes), no significant interaction existed for any of the
outcomes or protein sources at any time point (P $ 0.1). Hence, boys
and girls were pooled for analyses, but all models included sex. Similar to
the life-course plots, every model contained birth year. We further
considered early life and parental characteristics to evaluate confounding
by perinatal, lifestyle, and genetic factors: birth weight (,3000 g, yes/
no), breast-feeding (full breast-feeding $4 mo, yes/no), rapid weight gain
between birth and age 2 y [increase in weight SDS .0.67, yes/no (30)],
maternal overweight (BMI $25 kg/m2, yes/no), parental education
(university entrance qualification, yes/no; university degree, yes/no).
Similarly, we considered total energy and fat intake (%En) to evaluate
whether any association of protein was due to differences in these dietary
factors. Likewise, fiber intake was considered in the analyses on
vegetable and cereal protein. Only confounders that modified the
associations or predicted the outcomes substantially (P # 0.1) were

Downloaded from jn.nutrition.org by guest on September 19, 2015

Anthropometry, parental, and birth characteristics. As described in

detail in the following, DONALD Study participants are measured at
each visit according to standard procedures (19), dressed in underwear
only. The measuring-trained nurses undergo an annual quality control.
From the age of 2 y onwards, standing height is measured to the nearest
0.1 cm using a digital stadiometer. Weight is measured to the nearest
0.1 kg using an electronic scale (753 E; Seca). Skinfold thicknesses are
measured from the age of 6 mo onwards on the right side of the body to
the nearest 0.1 mm using a Holtain caliper. Percentage body fat 1 y
before ATO was calculated using the Slaughter equations for prepubertal
children (20). Percentage body fat was then converted to values of fat
mass index (FMI, in kg/m2) and FMI SD scores (FMI SDS) were obtained
by internally standardizing logged FMI values (mean = 0, SD = 1, by age
and sex).
At their childs admission to the study, parents provide information
about family and socioeconomic characteristics (i.e. school education
and professional degree) and are also weighed and measured.

basal metabolic rate (24), and age- and sex-specific cutoffs (25) were
used to exclude potentially implausible records (,3%). To represent diet
at 1824 mo, 34 y, and 56 y, we calculated the mean of the single
standardized energy intakes (mean = 0, SD = 1) or the mean nutrient
intakes at the respective time points. Besides total protein intake, we
considered animal and vegetable protein, which were further divided
into protein from dairy (excluding infant formula), meat, and cereal, as
previously described (10).

retained in the final models. Prepubertal FMI SDS (1 y before ATO) was
considered as a potential mediator in an attempt to address that body
fatness might lie on the pathway between animal, in particular dairy,
protein intake and pubertal timing (8).
Finally, to corroborate our results for the ages 34 and 56 y, we
compared how ATO was associated with total protein intake (g/d)
estimated from diet records or 24-h urinary nitrogen. For both protein
variables, ATO was divided into tertiles adjusted for sex and birth year.
Due to the limited number of children who provided at least 1 complete
24-h urine sample both at 34 and 56 y (n = 57, 159 sets of records and
urine overall), these analyses were performed on a descriptive basis only.
The correlation coefficient between the protein variables was r = 0.6 (P ,
0.0001) when all sets of records and urines were considered (n = 159)
and the mean difference between the 2 methods was 20.1 g/d (95% CI,
21.5; 1.4 g/d).
Statistical analyses were carried out using SAS (version 9.1.3). A Pvalue , 0.05 indicated significance. The sample size of 112 children was
sufficient to detect a difference of 0.46 y in ATO or 0.6 y in age at
menarche/voice break between 2 equal groups (a = 0.05, 80% power, 2tailed).

Results

TABLE 1

Early life, puberty, and family characteristics of the

total study sample1

Variable
n (%)
Early life characteristics
Birth weight, g
Birth length, cm
Gestational age, wk
Year of birth, n (%)
#1988
$1992
Breast-fed, n (%)
Ever fully
Fully $4 mo
Puberty characteristics, y
ATO
APHV
Menarche
Voice break
Maternal characteristics
Age at birth of child, y
Overweight, n (%)
Qualification for university entrance,2 n (%)
University degree, n (%)
Paternal characteristics
Age at birth of child, y
Qualification for university entrance,2 n (%)
University degree, n (%)
1
2

Boys

112

55 (49.1)

112
112
112
112
27
36

3603 6 457
53 (51; 54)
40 (39,41)
11 (20.0)
18 (32.7)

16 (28.1)
18 (31.6)

112
112

51 (85.0)
36 (60.0)

44 (73.3)
34 (56.7)

112
106
47
45

10.1 6 0.9
13.2 6 0.9

13.3 6 1.2

8.7 6 1.0
11.6 6 1.1
12.8 6 1.2

112
112
112
112

30 (27,33)
17 (28.3)
35 (58.3)
19 (34.6)

30 (28,32)
19 (31.7)
26 (56.7)
18 (31.6)

112
112
112

32 (29,37)
42 (70.0)
37 (61.7)

33 (30,36)
32 (53.3)
28 (46.7)

Girls
57 (50.9)
3403 6 413
51 (50; 53)
40 (40,41)

Values are means 6 SD, medians (25th; 75th percentile), or frequencies, n = 92112.
At least 12 y of school education.

Discussion
The present analysis suggests that protein intake during midchildhood might be differentially related to pubertal timing.
Dietary protein and pubertal timing

567

Downloaded from jn.nutrition.org by guest on September 19, 2015

In girls, ATO and APHV took place ~1.5 y earlier than in boys
(Table 1). The children were born between 1984 and 1994, but
ATO did not differ between those born before 1989, between
1989 and 1991, or after 1991 (P = 0.1; adjusted for sex). Table 2
summarizes the childrens energy and protein intakes.
According to the life-course plots, protein intake (%En) at 5
6 y was of particular importance for pubertal timing. Children
with a higher total protein intake at this age tended to have an
earlier ATO (P = 0.06) (Fig. 1). Apparently, animal protein

intake (%En) at 56 y was responsible for this association (P =

0.048). By contrast, a higher vegetable protein intake (%En) at
age 34 y was associated with a later ATO (P = 0.02). Because
for all protein variables, only 1 regression coefficient was
significantly different from zero, there was no indication of a
period in which a change in protein intake was of importance,
because this would have made a switch of signs of the
coefficients mandatory (8).
The life-course plots with APHV and age at menarche/voice
break as the dependent variables yielded entirely similar results
(data not shown). Therefore, we then investigated the relation
between protein intake (%En) at 34 and 56 y and ATO,
APHV, and age at menarche/voice break in depth. In the highest
tertile of animal protein intake at age 56 y, ATO took place 0.6 y
earlier than in the lowest tertile, even after year at birth as well as
additional confounders were adjusted for (Table 3, model 1).
When examined independently, however, there was a similar
association for animal protein intake at 34 y.
Vegetable protein (%En) at 34 y was associated with a later
ATO. Considering FMI SDS 1 y before ATO did not fully
explain these associations (model 2), indicating that differences
in FMI were not a main mediating factor between the different
protein sources and pubertal markers.
With respect to protein intake from different food groups (%
En), protein intake from cow milk and dairy products at age 56 y,
but not meat, was associated with an earlier ATO both in
model 1 (mean ATO, 95% CI, tertile 1: 9.5, 9.29.8; tertile 2:
9.5, 9.39.8; tertile 3: 9.1, 8.89.4 y; P-trend = 0.04) and in
model 2 (P = 0.03). Cereal protein intake was not related to ATO
at either age (P . 0.1) (data not shown).
Similar to ATO, children with a higher animal protein
consumption (%En) at 34 and 56 y had an earlier APHV
(Table 4, model 1), whereas those with a higher vegetable
protein intake at these ages experienced a later APHV. Adjustment for prepubertal FMI SDS only partly explained the findings
(model 2). Among the different protein sources, dairy, but not
meat protein, at 56 y mirrored the results for animal protein
(P = 0.02) (data not shown).
A higher animal protein intake at 34 y also tended (P = 0.06)
to be, and at 56 y of age was (P = 0.02), related to an earlier
menarche/voice break (Supplemental Table 1, model 1). Children with a higher vegetable protein intake experienced later
menarche/voice break, respectively (P = 0.020.03, model 1).
These associations were not fully explained by prepubertal FMI
SDS (model 2). Similar to ATO and APHV, dairy, but not meat,
tended to be decisive (P = 0.06 for age 34 y, P = 0.2 for age 56 y)
(data not shown).
Further adjustments for birth weight (categories), rapid
weight gain between 02 y (yes/no), or maternal overweight
(yes/no) did not change any of these results. Also, fat or fiber
intake did not explain the associations.
Finally, we evaluated our findings using protein intake (g/d)
estimated from 24-h urinary nitrogen excretion. Not all children
had provided urine samples and, hence, results were not
significant for urinary data (P . 0.1). However, ATO decreased
across tertiles of total protein intake (g/d) at both 34 and 56 y
regardless of whether protein intake stemmed from dietary
records or the urinary biomarker (Supplemental Fig. 1).

TABLE 2

Dietary data from 3-d weighed dietary and from 24-h urine samples in children (DONALD
Study)1
Age

Variable

3,476 (3,122,3,826)
50.9 (48.3; 55.4)
10.3 (7.9; 12.1)
35.2 (32.5; 37.9)
27.8 (24.3; 31.2)
13.7 (12.3; 14.9)
8.8 (7.4; 10.1)
4.9 (3.8; 5.7)
5.7 (2.7; 7.3)
1.8 (0.9; 2.4)
2.5 (1.7; 3.1)

1824 mo

3,969
49.2
10.0
37.1
32.8
13.8
9.6
4.2
6.1
2.1
2.2

(3,620,4,238)
(44.5; 52.2)
(8.3; 11.8)
(34.1; 40.6)
(28.8; 36.3)
(12.7; 14.9)
(8.2; 10.9)
(3.7; 4.8)
(4.6; 7.5)
(1.6; 3.2)
(1.6; 2.7)

34 y

56 y

4,791 (4,390,5,266)
49.9 (46.2; 53.4)
11.1 (9.1; 13.0)
37.3 (34.1; 39.8)
36.2 (32.6; 41.1)
12.9 (11.5; 14.1)
8.8 (7.3; 9.7)
4.2 (3.7; 4.9)
4.5 (3.3; 5.6)
2.4 (1.5; 3.6)
2.3 (1.9; 2.8)

5,663
51.0
13.6
36.3
43.0
12.8
8.0
4.4
4.0
2.5
2.5

(5,276,6,264)
(47.3; 53.2)
(11.7; 17.0)
(34.6; 39.8)
(37.8; 48.5)
(11.3; 13.9)
(7.0; 9.2)
(3.9; 5.1)
(2.7; 5.1)
(1.6; 3.4)
(2.1; 3.2)

0.47 (0.38; 0.59)

2.3 (2.1; 2.8)
312.7 (288.7; 373.3)
35.5 (31.9; 41.2)

0.55
3.1
383.7
42.7

(0.42; 0.67)
(2.7; 3.4)
(335.9; 438.7)
(37.2; 48.5)

Values are medians (25th; 75th percentile), n = 112 (diet records) and 57 (urine).
Excluding human milk protein.
3
Estimated from urinary nitrogen excretion (see text).
1
2

Children with a higher animal protein intake, particularly at

56 y, experienced an earlier ATO, APHV, and menarche/voice
break, whereas those with a higher vegetable protein at 34 and
56 y appeared to have a delayed puberty.
Both the greater relevance of diet closer to puberty onset and
the opposing effects of animal and vegetable protein are in line
with Berkey et al. (9) and their cohort of 67 girls born during the
1930s and 1940s. We were additionally able to demonstrate
an association of animal and vegetable protein with both early
and late pubertal stages, obtaining remarkably similar results.
Others have suggested a role for fat (12,16), energy (11,
15,16,31), vitamin C (16), and fiber intake (13) in pubertal
development in girls, but not for protein. Because critical periods
of protein intake may exist, it is a limitation that the aforementioned studies uniformly addressed diet after the growth spurt
has already begun in girls. Besides, none distinguished between
animal and vegetable protein. This is of interest in view of the
debate as to whether a vegetarian diet could be associated with a
delayed puberty. In 63 girls, de Ridder et al. (32) found that
higher intake of vegetable protein was associated with later
breast development and menarche, although fiber intake was
more important. Because in the present study fiber intake did not
explain the effects for vegetable protein, it could be speculated
that our results represent other favorable aspects of vegetarian
diets. Fiber intake might have been too low to represent a crucial
factor; however, it was even higher in our sample than current
recommendations for children (33). Furthermore, in a recent
analysis of DONALD Study data from a similar age range (34),
intake levels were comparable to those seen by de Ridder et al.
(32).
Protein intake during transitional feeding (i.e. age 12 mo)
was not related to the puberty outcomes. Arising from a sudden
increase once complementary foods are introduced, protein
consumption levels are high at this age, which, according to the
early protein hypothesis, might predispose to later obesity (3).
However, only the years directly preceding the pubertal growth
spurt were relevant for pubertal timing. This time window may
568

Gunther et al.

represent adiposity rebound (i.e. age 56 y), when BMI reaches a

nadir in childhood before increasing to adult levels. Adiposity
rebound could be a marker of maturation, because it was
associated with age at menarche (35). This period of rapid body
mass change is perhaps particularly receptive to nutritional
influences.
The effect of animal protein intake, which was associated
with an earlier puberty onset, might mainly be due to dairy. This
would correspond to results obtained by Wiley et al. (36) who
reported that women who drank more milk in childhood (512 y)
experienced an earlier menarche. In prepubertal boys, skim
milk, but not a similar amount of animal protein from meat,
stimulated secretion of insulin and insulin-like growth factor
1 (IGF-1) (37,38). Because insulin suppresses IGF binding
protein 1 (39), availability of free IGF-1 would be enhanced.
IGF-1 is a major regulator of human growth, but a role in
adipocyte proliferation and differentiation has also been
suggested (40). Indeed, previous analyses of the DONALD
Study population yielded positive associations for animal
protein (%En), in particular from dairy, at both 12 mo and 5
6 y with body fatness at the age of 7 y (8). But with respect to
pubertal timing, protein seems to operate via a pathway that is
(partially) independent of body fatness, because adjusting for
prepubertal FMI SDS did not fully account for our findings. In
fact, prepubertal body composition was critical for later
pubertal markers but not the initiation of the pubertal growth
spurt in the DONALD Study population (4).
Although frequently used as a reference instrument in
validation studies, weighed diet records themselves are prone
to measurement error and require a high degree of dedication.
Nevertheless, this method represents a method of choice in
younger children, such as those studied here (41). In addition, it
is a major strength of this analysis that in a subsample, total
protein intake could be estimated from 24-h urinary nitrogen.
Further strengths include the repeated, prospective data from
early childhood until adolescence, allowing us to examine the
relation between diet and pubertal timing from a life-course

Downloaded from jn.nutrition.org by guest on September 19, 2015

3-d weighed dietary records

Energy, kJ/d
Carbohydrates, %En
Fiber, g/d
Fat, %En
Total protein, g/d
Total protein, %En
Animal protein,2%En
Vegetable protein, %En
Dairy protein, %En
Meat protein, %En
Cereal protein, %En
24-h urine
Volume, L
Creatinine, mmol/d
Nitrogen excretion, mmol/d
Total protein intake,3 g/d

12 mo

TABLE 3

Association between dietary protein intake in childhood and ATO1

Tertiles of protein intake
2

9.5 (9.2; 9.8)

9.4 (9.1; 9.7)

9.5 (9.2; 9.8)

9.5 (9.2; 9.8)

9.2 (8.9; 9.5)

9.2 (8.9; 9.5)

0.2
0.6

9.7 (9.4; 10.0)

9.7 (9.4; 10.0)

9.2 (8.9; 9.5)*

9.2 (8.9; 9.5)*

9.2 (8.9; 9.5)*

9.2 (8.9; 9.5)*

0.02
0.03

9.7 (9.4; 10.0)

9.6 (9.3; 9.9)

9.5 (9.2; 9.8)

9.5 (9.2; 9.8)

9.0 (8.7; 9.3)*

9.0 (8.7; 9.3)*

0.01
0.07

9.6 (9.4; 9.9)

9.6 (9.3; 9.9)

9.4 (9.1; 9.7)

9.4 (9.1; 9.7)

9.0 (8.7; 9.3)*

9.0 (8.7; 9.3)*

0.003
0.01

9.1 (8.8; 9.4)

9.1 (8.8; 9.4)

9.4 (9.1; 9.7)

9.4 (9.1; 9.7)

9.6 (9.2; 9.9)

9.6 (9.3; 9.9)

0.01
0.02

9.4 (9.0; 9.7)

9.4 (9.0; 9.7)

9.3 (9.0; 9.6)

9.3 (9.0; 9.6)

9.5 (9.2; 9.8)

9.4 (9.1; 9.7)

0.2
0.3

Values are adjusted means (95% CI), n = 112. *Different from the lowest tertile, P ,
0.05 (Dunnetts test for post hoc comparisons).
2
Multiple linear regression models (continuous variables). Model 1: adjustments for
sex, total energy intake, full breast-feeding for $4 mo (yes/no), birth year, and paternal
university degree (yes/no); model 2: model 1 + FMI SDS 1 y before ATO.
1

FIGURE 1 Life-course plots with ATO (y) as the dependent variable

and standardized total protein (A), animal protein (B), and vegetable
protein intakes (C) (%En) throughout childhood as the explanatory
variables. Values are regression coefficients (95% CI), n = 92112,
adjusted for each other, for mean energy intake from 16 y, sex, and
birth year, *P , 0.05.

perspective. Both early and later markers of puberty could be

estimated on the basis of the pubertal growth curve, i.e. based on
somatic maturation, and we were not reliant on either recall of
pubertal events or more subjective markers of puberty such as
Tanner stage, which are difficult to assess and characterized by a
high inter-observer variability (42). Also, we were able to
control for important potential confounders, e.g. parental and
early life characteristics.
Our study has several limitations. DONALD Study participants are characterized by a high socioeconomic status (18).
ATO and APHV, however, were in accordance with other
European studies (7) and a representative German survey found
similar median ages at menarche (12.8 y) and voice beak (13.5 y)
(43). Furthermore, data on parental pubertal characteristics are
not available in the DONALD Study; hence, genetic influences
on pubertal timing could not be adjusted for. A 3rd limitation is
the small sample size, which is why we could not reliably

examine potential sex differences and were also forced to create

a variable combining age at menarche and voice break. In the
present examination, statistical tests for interactions did not
support differences between boys and girls. Stratified analyses,
however, suggested that the associations of protein intake with
pubertal timing might be stronger in boys, but this finding could
be due to chance (data not shown). Recent studies suggested that
factors influencing pubertal timing, such as adiposity, may not
differ between sexes (47).
An earlier puberty onset has been related to an increased risk
for hormone-related cancers in adulthood (4446). For example,
a meta-analysis of 26 epidemiological studies reported a 9% risk
reduction for breast cancer with every additional year at
menarche (47). Additionally, recent study results demonstrated
that a 1-y delay in menarche was associated with a 2.4 (48) to
4.5% (49) lower total mortality. The difference of a 0.40.8 y
earlier puberty onset in the highest compared with the lowest
tertile of animal protein intake observed in our final models thus
suggests that the present findings, if confirmed, could have
substantial public health implications. The same might be true
for vegetable protein intake, where differences were only slightly
smaller and pubertal timing was delayed by 0.30.5 y in the
highest tertiles of intake compared to the lowest.
In conclusion, dietary protein in mid-childhood may differentially influence pubertal timing. Whereas higher animal
protein intake at 56 y might be related to an earlier ATO,
APHV and menarche/voice break, higher intakes of vegetable
protein at 34 and 56 y were associated with a delayed puberty.
These findings need to be confirmed in future studies with
sufficient power to examine differences between boys and girls
and the ability to consider hormonal status.
Dietary protein and pubertal timing

569

Downloaded from jn.nutrition.org by guest on September 19, 2015

Total protein, %En

Age 34 y
Model 1
Model 2
Age 56 y
Model 1
Model 2
Animal protein, %En
Age 34 y
Model 1
Model 2
Age 56 y
Model 1
Model 2
Vegetable protein, %En
Age 34 y
Model 1
Model 2
Age 56 y
Model 1
Model 2

P-trend2

TABLE 4

Association between dietary protein intake in childhood and APHV1

10.

Tertiles of protein intake

1

P-trend

11.

12.7)
12.6)

12.5 (12.1; 12.8)

12.5 (12.1; 12.8)

12.3 (11.9; 12.6)

12.3 (11.9; 12.6)

0.4
0.9

13.1)
13.1)

12.2 (11.8; 12.5)*

12.1 (11.8; 12.4)*

12.1 (11.8; 12.5)*

12.1 (11.8; 12.5)*

0.02
0.03

12.
13.
14.
15.

12.9)
12.8)

12.5 (12.2; 12.9)

12.6 (12.2; 12.9)

12.0 (11.7; 12.3) *

12.0 (11.7; 12.3)

0.05
0.2

13.1)
13.1)

12.3 (12.0; 12.6)

12.3 (12.0; 12.6)

12.0 (11.7; 12.3)*

12.0 (11.6; 12.3)*

0.001
0.002

12.5)
12.5)

12.3 (12.0; 12.7)

12.3 (12.0; 12.6)

12.6 (12.3; 13.0)

12.6 (12.2; 13.0)

0.02
0.04

12.6)
12.5)

12.3 (11.9; 12.6)

12.3 (11.9; 12.6)

12.6 (12.2; 13.0)

12.5 (12.2; 12.9)

0.04
0.09

16.

17.
18.

1
Values are adjusted means (95% CI), n = 112. *Different from the lowest tertile, P ,
0.05 (Dunnetts test for post hoc comparisons).
2
Multiple linear regression models (continuous variables). Model 1: adjustments for
sex, total energy intake, full breast-feeding for $4 mo (yes/no), birth year, and paternal
university degree (yes/no); model 2: model 1 + FMI SDS 1 y before ATO.

19.
20.

21.
22.

23.

Acknowledgements
A.L.B.G., N.K-D., A.K., and A.E.B. designed research. A.L.B.G.
analyzed data and wrote the manuscript. A.L.B.G. had primary
responsibility for final content. All authors made substantial
contributions to and approved the final manuscript.

24.
25.

26.

Literature Cited
1.

2.
3.
4.

5.

6.

7.

8.

570

Papadimitriou A, Pantsiotou S, Douros K, Papadimitriou DT, Nicolaidou P, Fretzayas A. Timing of pubertal onset in girls: evidence for nonGaussian distribution. J Clin Endocrinol Metab. 2008;93:44225.
Karlberg J. Secular trends in pubertal development. Horm Res. 2002;57
Suppl 2:1930.
Agostoni C, Scaglioni S, Ghisleni D, Verduci E, Giovannini M, Riva E.
How much protein is safe? Int J Obes (Lond). 2005;29 Suppl 2:S813.
Buyken AE, Karaolis-Danckert N, Remer T. Association of prepubertal
body composition in healthy girls and boys with the timing of early and
late pubertal markers. Am J Clin Nutr. 2009;89:22130.
Juul A, Magnusdottir S, Scheike T, Prytz S, Skakkebaek NE. Age at
voice break in Danish boys: effects of pre-pubertal body mass index and
secular trend. Int J Androl. 2007;30:53742.
Sandhu J, Ben-Shlomo Y, Cole TJ, Holly J, Davey Smith G. The impact
of childhood body mass index on timing of puberty, adult stature and
obesity: a follow-up study based on adolescent anthropometry recorded
at Christs Hospital (19361964). Int J Obes (Lond). 2006;30:1422.
Silventoinen K, Haukka J, Dunkel L, Tynelius P, Rasmussen F. Genetics
of pubertal timing and its associations with relative weight in childhood
and adult height: the Swedish Young Male Twins Study. Pediatrics.
2008;121:e88591.
Gunther ALB, Remer T, Kroke A, Buyken AE. Early protein intake and
later obesity risk: which protein sources at which time points throughout infancy and childhood are important for body mass index and body
fat percentage at 7 y of age? Am J Clin Nutr. 2007;86:176572.

Gunther et al.

27.

28.

29.
30.
31.

32.

33.
34.

Berkey CS, Gardner JD, Frazier AL, Colditz GA. Relation of childhood
diet and body size to menarche and adolescent growth in girls. Am J
Epidemiol. 2000;152:44652.
Maclure M, Travis LB, Willett W, MacMahon B. A prospective cohort
study of nutrient intake and age at menarche. Am J Clin Nutr.
1991;54:64956.
Meyer F, Moisan J, Marcoux D, Bouchard C. Dietary and physical
determinants of menarche. Epidemiology. 1990;1:37781.
Merzenich H, Boeing H, Wahrendorf J. Dietary fat and sports activity as
determinants for age at menarche. Am J Epidemiol. 1993;138:21724.
Koo MM, Rohan TE, Jain M, McLaughlin JR, Corey PN. A cohort study
of dietary fibre intake and menarche. Public Health Nutr. 2002;5:35360.
Moisan J, Meyer F, Gingras S. Diet and age at menarche. Cancer Causes
Control. 1990;1:14954.
Petridou E, Syrigou E, Toupadaki N, Zavitsanos X, Willett W,
Trichopoulos D. Determinants of age at menarche as early life
predictors of breast cancer risk. Int J Cancer. 1996;68:1938.
Britton JA, Wolff MS, Lapinski R, Forman J, Hochman S, Kabat GC,
Godbold J, Larson S, Berkowitz GS. Characteristics of pubertal
development in a multi-ethnic population of nine-year-old girls. Ann
Epidemiol. 2004;14:17987.
Bingham SA. Urine nitrogen as a biomarker for the validation of dietary
protein intake. J Nutr. 2003;133 Suppl 3:S9214.
Kroke A, Manz F, Kersting M, Remer T, Sichert-Hellert W, Alexy U,
Lentze MJ. The DONALD Study. History, current status and future
perspectives. Eur J Nutr. 2004;43:4554.
Lohman TG, Roche AF, Martorell R, editors. Anthropometric standardization reference manual. Champaign (IL): Human Kinetics; 1988.
Slaughter MH, Lohman TG, Boileau RA, Horswill CA, Stillman RJ,
Van Loan MD, Bemben DA. Skinfold equations for estimation of body
fatness in children and youth. Hum Biol. 1988;60:70923.
Preece MA, Baines MJ. A new family of mathematical models
describing the human growth curve. Ann Hum Biol. 1978;5:124.
Molinari L, Gasser T. The human growth curve. In: Hauspie R,
Cameron N, Molinari L, editors. Methods in human growth research.
Cambridge: Cambridge University Press; 2004. p. 2754.
Sichert-Hellert W, Kersting M, Chahda C, Schafer R, Kroke A. German
food composition data base for dietary evaluations in children and
adolescents. J Food Compost Anal. 2007;20:6370.
Schofield WN. Predicting basal metabolic rate, new standards and
review of previous work. Hum Nutr Clin Nutr. 1985;39 Suppl 1:541.
Sichert-Hellert W, Kersting M, Schoch G. Underreporting of energy
intake in 1 to 18 year old German children and adolescents.
Z Ernahrungswiss. 1998;37:24251.
Bartels H, Cikes M. Ueber Chromogene der Kreatininbestimmung nach
Jaffe. [Chromogens in the creatinine determination of Jaffe]. Clin Chim
Acta. 1969;26:110.
Remer T, Neubert A, Maser-Gluth C. Anthropometry-based reference
values for 24-h urinary creatinine excretion during growth and their use
in endocrine and nutritional research. Am J Clin Nutr. 2002;75:5619.
Garlick PJ. Protein requirements of infants and children. In: Rigo J,
Ziegler EE, editors. Protein and energy requirements in infancy and
childhood. Nestle Nutrition Workshop. Ser Pediatr Programm. 58th
vol. Nestec Ltd, Vevey/S. Basel: Karger AG; 2006. p. 3950.
Cole TJ. Modeling postnatal exposures and their interactions with birth
size. J Nutr. 2004;134:2014.
Monteiro PO, Victora CG. Rapid growth in infancy and childhood and
obesity in later lifea systematic review. Obes Rev. 2005;6:14354.
Koprowski C, Ross RK, Mack WJ, Henderson BE, Bernstein L. Diet,
body size and menarche in a multiethnic cohort. Br J Cancer.
1999;79:190711.
de Ridder CM, Thijssen JH, Van t Veer P, van Duuren R, Bruning PF,
Zonderland ML, Erich WB. Dietary habits, sexual maturation, and
plasma hormones in pubertal girls: a longitudinal study. Am J Clin Nutr.
1991;54:80513.
Williams CL, Bollella M, Wynder EL. A new recommendation for
dietary fiber in childhood. Pediatrics. 1995;96:9858.
Cheng G, Karaolis-Danckert N, Libuda L, Bolzenius K, Remer T,
Buyken AE. Relation of dietary glycemic index, glycemic load, and fiber
and whole-grain intakes during puberty to the concurrent development
of percent body fat and body mass index. Am J Epidemiol.
2009;169:66777.

Downloaded from jn.nutrition.org by guest on September 19, 2015

Total protein, %En

Age 34 y
Model 1 12.4 (12.0;
Model 2 12.3 (12.0;
Age 56 y
Model 1 12.8 (12.5;
Model 2 12.8 (12.4;
Animal protein, %En
Age 34 y
Model 1 12.5 (12.2;
Model 2 12.4 (12.1;
Age 56 y
Model 1 12.8 (12.5;
Model 2 12.8 (12.4;
Vegetable protein, %En
Age 34 y
Model 1 12.1 (11.8;
Model 2 12.1 (11.8;
Age 56 y
Model 1 12.2 (11.8;
Model 2 12.2 (11.9;

9.

35. Williams S, Dickson N. Early growth, menarche, and adiposity

rebound. Lancet. 2002;359:5801.
36. Wiley AS. Does milk make children grow? Relationships between milk
consumption and height in NHANES 19992002. Am J Hum Biol.
2005;17:42541.
37. Hoppe C, Molgaard C, Juul A, Michaelsen KF. High intakes of
skimmed milk, but not meat, increase serum IGF-I and IGFBP-3 in
eight-year-old boys. Eur J Clin Nutr. 2004;58:12116.
38. Hoppe C, Molgaard C, Vaag A, Barkholt V, Michaelsen KF. High
intakes of milk, but not meat, increase s-insulin and insulin resistance in
8-year-old boys. Eur J Clin Nutr. 2005;59:3938.
39. Wolk K, Larsson SC, Vessby B, Wolk A, Brismar K. Metabolic,
anthropometric, and nutritional factors as predictors of circulating
insulin-like growth factor binding protein-1 levels in middle-aged and
elderly men. J Clin Endocrinol Metab. 2004;89:187984.
40. Wabitsch M, Hauner H, Heinze E, Teller WM. The role of growth
hormone/insulin-like growth factors in adipocyte differentiation.
Metabolism. 1995;44:459.
41. Livingstone MB, Robson PJ, Wallace JM. Issues in dietary intake assessment
of children and adolescents. Br J Nutr. 2004;92 Suppl 2:S21322.
42. Euling SY, Herman-Giddens ME, Lee PA, Selevan SG, Juul A, Sorensen
TI, Dunkel L, Himes JH, Teilmann G, et al. Examination of US puberty-

43.

44.

45.
46.
47.

48.

49.

timing data from 1940 to 1994 for secular trends: panel findings.
Pediatrics. 2008;121 Suppl 3:S17291.
Kahl H, Schaffrath-Rosario A, Schlaud M. Sexuelle Reifung von
Kindern und Jugendlichen in Deutschland. [Sexual maturation of
children and adolescents in Germany]. Bundesgesundheitsbl - Gesundheitsforsch - Gesundheitsschutz. 2007;50:67785.
Velie EM, Nechuta S, Osuch JR. Lifetime reproductive and anthropometric risk factors for breast cancer in postmenopausal women. Breast
Dis. 2005;24:1735.
Berkey CS, Frazier AL, Gardner JD, Colditz GA. Adolescence and
breast carcinoma risk. Cancer. 1999;85:24009.
Garner MJ, Turner MC, Ghadirian P, Krewski D. Epidemiology of
testicular cancer: an overview. Int J Cancer. 2005;116:3319.
Clavel-Chapelon F, Gerber M. Reproductive factors and breast cancer
risk. Do they differ according to age at diagnosis? Breast Cancer Res
Treat. 2002;72:10715.
Jacobsen BK, Heuch I, Kvale G. Association of low age at menarche
with increased all-cause mortality: a 37-year follow-up of 61,319
Norwegian women. Am J Epidemiol. 2007;166:14317.
Jacobsen BK, Oda K, Knutsen SF, Fraser GE. Age at menarche, total
mortality and mortality from ischaemic heart disease and stroke:
the Adventist Health Study, 197688. Int J Epidemiol. 2009;38:24552.

Downloaded from jn.nutrition.org by guest on September 19, 2015

Dietary protein and pubertal timing

571