Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Objectives: There is increasing evidence that breastfeeding has long-term effects on offspring biology and health,
which has heightened interest in understanding the extent of variation in breast milk composition and its underlying
determinants. Here, we report variation in milk macronutrient composition in a well-characterized cohort of young Filipino mothers and test underlying maternal predictors of this variation.
Methods: Morning breast milk samples, anthropometrics, dietary recalls, and other interview data were collected
in 102 Filipino young breastfeeding mothers (age range 24.625.4 years) living in Cebu City, Philippines. Milk samples
were analyzed for protein, fat, sugar, and milk energy density. Regression models were used to test associations
between milk macronutrient composition and maternal diet, body composition, breastfeeding duration, and feeding frequency.
Results: Consistent with past studies, milk fat and energy increased with duration of breastfeeding; there were no
associations between maternal diet or percent body fat and milk composition with the exception of a modest, inverse
association between maternal adiposity and milk sugar content.
Conclusions: The relative lack of associations between maternal diet or body composition and milk composition at
Cebu is consistent with past studies and suggests that milk composition may be buffered against uctuations in maternal dietary intake or nutritional status. We speculate that the tendency for milk composition to vary between populations faced with different nutritional ecologies, but to show minimal responsiveness to intake during lactation, may
enhance the reliability of milk composition as a stable intergenerational cue of typical local environmental quality. Am.
J. Hum. Biol. 00:000000, 2012.
' 2012 Wiley Periodicals, Inc.
INTRODUCTION
The benets of breastfeeding for infant health include
reduced infectious disease morbidity and mortality (Duijts
et al., 2009) and improved cognitive and physical development (Daniels and Adair, 2005; Isaacs et al., 2009). In
addition, there is increasing evidence that breastfeeding
is associated with improved long term health outcomes for
infants such as reduced risk of diabetes, obesity, cardiovascular disease, and other adult-onset metabolic disorders (Grummer-Strawn and Mei, 2004; Koletzko, 2006;
Owen et al., 2003, 2006). These ndings raise questions
about the biological basis for these differences. In particular, the nutrient and hormonal composition of breast
milk has emerged as a likely inuence on individual physiological development (Koletzko et al., 2009; Riva et al.,
2004).
Although once considered relatively homogenous (Hall,
1979), the variation within human milk has recently
received more research attention (Jensen, 1999; Prentice,
1995; Wojcik et al., 2009). Table 1 summarizes milk macronutrient composition from past studies in human populations. Average fat content of human milk across populations varies from 2.8 to 4.78 g/dl, with women in the same
population showing equal, if not greater amounts of variation than different populations (Prentice, 1995; Wojcik
et al., 2009). Carbohydrate and protein content of milk
show less variance, with carbohydrates generally ranging
from 6.528.0 g/dl and protein constrained from 0.91.5 g/dl
in mature milk (Table 1). Colostrum, the milk produced in
C 2012
V
Fat
Lactose
Protein
Reference
Philippines (Manila)
The Gambia
Australia
Bangladesh
Sweden
Guatemala
Zaire
USA (DARLING)
Mean
3.93
3.78
3.74
2.66
5.69
2.40
3.30
3.80
3.66
7.31
7.74
6.14
8.08
6.70
8.00
6.30
7.40
7.21
0.85
1.09
0.92
1.00
0.83
0.94
1.30
1.10
1.00
WHO, 1985
Prentice et al., 1981a
Mitoulas et al., 2002
Brown et al., 1986
WHO, 1985
WHO, 1985
WHO, 1985
Nommsen et al., 1991
Macronutrients are presented as grams per 100 ml and calculated where necessary using a nitrogen to protein conversion factor of 6.38.
and maternal adiposity typically show an inverse relationship. Milk composition has also been shown to change
over the course of lactation (Lubetzky et al., 2007; Mandel
et al., 2005; Mitoulas et al., 2002; Nommsen et al., 1991),
possibly reecting developmental maturation of the lactating breast or differences in nursing frequency as the
infant ages. These generally weak and inconsistent relationships between maternal characteristics and milk composition may be evidence for physiological buffering of
milk composition. Evidence for marked variation in milk
composition among individuals that does not relate to current diet or body composition would point to the likely importance of other factors, such as early life nutrition or
growth history of the mother (Lammers et al., 1999; Van
Amburgh et al., 1998), as potentially important inuences
on milk composition.
In this study, we describe the extent of variation and
maternal predictors of the energy density and macronutrient composition of breast milk collected from 102 Filipino mothers participating in a longitudinal birth cohort
in Cebu City, Philippines. These women represent all original cohort members who were still participating in the
study in 2007 and were currently lactating at the time of
interview. Although the mothers in this study were drawn
from a longitudinal birth cohort, they were recruited into
the study at different time points in their lactations,
resulting in a cross sectional sampling of milk composition
and infant age. Breastfeeding behaviors vary by infant
age with older infants generally nursing less than
younger infants, particularly following the introduction of
complimentary foods. The decline in nursing frequency
and bout duration will result in a decrease in milk production. However, in Cebu, there is only a modest decline in
breastfeeding frequency for mothers nursing older
infants, and infant age and nursing frequency were tested
as independent predictors of milk macronutrient and
energy content. Other groups have reported age related
differences in milk macronutrients and energy without
consideration of age related changes in nursing frequency
(Michaelsen et al., 1990; Mitoulas et al., 2002). We
hypothesized that milk fat and energy would increase
while sugar would decline with increasing infant age in
this sample, following trends previously described in the
literature. Building from evidence for maternal buffering
of milk composition reported in other populations, we predicted that there would be few or no signicant associations between current maternal diet or body composition
and milk composition (Brown et al., 1986; Nommsen et al.,
mise between the ideal sampling method of a full mammary expression and concerns about the risks such a sampling method would impose on mothers and infants. We
were also unwilling to give a portion of the sampled milk
back to the infant in a bottle, both out of concern for reinforcing bottle feeding and limited access to appropriate
storage conditions for the milk. Small samples, either
paired before and after the infant nurses (Kent et al.,
2006; Mitoulas et al., 2002) or mid-feed samples (Xiang
et al., 2005), have been used elsewhere. Neville et al.,
(1984), comparing fore, mid, and pooled breast milk samples reported less than 13% coefcient of variation for
residuals of protein, lactose, and several micronutrients
between samples, and suggested that mid-feed samples
could be an alternative to full mammary expression for
analyses of these specic components. However, milk fat
does vary signicantly across a single feeding (Daly et al.,
1993; Neville et al., 1984) and fat measures may be biased
accordingly. It has been shown that milk composition
varies across a single feed (Daly et al., 1993), across a day
(Daly et al., 1993; Kent et al., 2006; Ruel et al., 1997) and
even between mammary glands (Daly et al., 1996; Mitoulas et al., 2002). There is some possibility that some of the
associations reported here may be the product of differences in sampling protocols. Kent et al. (2006) have shown
that while milk fat content is independent of nursing frequency, it related to time of day with morning samples
tending to be lower in overall fat. Pooled 24 h samples
reect the gold standard for collection; however, the nutritional impact and disruption of normal nursing behaviors
makes them problematic for eld use. Ruel et al. (1997),
as previously described, has validated samples collected
between 6 am and 8 am as the best time for a single sample approximation of milk composition, particularly for fat
and energy, although again, here we present early to midmilk samples rather than a full mammary expression as
done in the original study.
Statistical analyses
All statistical analyses were performed using Stata 10.0
(College Park, TX). Analyses rst included evaluating distributions and descriptive statistics, which allowed comparisons between the sample and values derived from
other human populations. A series of multivariate linear
regression models were then used to test for associations
between the outcome variables (human milk fat, protein,
sugar, and total energy) and several candidate predictor
variables suggested as predictors of milk composition in
prior studies (mothers BMI, percent body fat, dietary
energy and macronutrient intake, and infant age). Regression models testing for an association between milk composition and maternal parity and infant sex found no signicant association between these predictors and milk
macronutrients, which were not included in the models
reported below.
RESULTS
This study relied on a single sample of human milk, collected between 6 am and 10 am, as discussed above. This
time has been shown elsewhere to be the most representative of 24 milk averages (Ruel et al., 1997). A 10 ml sample
of milk was collected after 2 min of nursing, reecting an
early to mid-feed sample of milk. This reects a compro-
Sample characteristics are shown in Table 2. Most participants were in the average BMI range with relatively
low percent body fat. Because the sample was drawn from
a one year birth cohort, all mothers were within 1 year of
age when measured (age range 24.625.4 years). Infants
averaged 8.4 6 5.5 months of age (range 2 weeks to 18
American Journal of Human Biology
Mean (SD)
Range
23.8 (0.3)
150.5 (5.9)
46.3 (8.6)
20.4 (3.5)
24.9 (3.9)
24.5%
53.9%
9.8 (2.6)
2324
138163.8
29.174.5
14.535.9
9.634.8
1,411.3 (720.1)
19.3% (13%)
50.0% (10%)
19.3% (5%)
257.73,478.1
257%
2266%
1138%
253.0 (165.4)
12 (3.6)
7,036 (1,823)
65.8 (8.4)
15.9 (1.5)
51% Male
3,037 (668)
9548
424
3,17010,215
46.979.7
12.919.5
1,3644,545
3.8 (1.5)
1.3 (0.48)
7.3 (0.58)
0.72 (0.15)
0.989.37
0.843.24
4.968.91
0.431.24
215
TABLE 3. Regression models testing associations between maternal predictors and milk energy content
Unadjusted B (SE)
Infant age (months)
BMI (kg/m2)
Body fat (%)
Energy intake (kcal)
% kcal from fat
% kcal from protein
% kcal from carbs
Model R2
0.005 (0.003)
20.009 (0.004)**
20.004 (0.004)
20.034**0.028)
20.038 (2.068)
0.131 (1.546)
20.142 (2.907)
R2
0.03
0.04
0.01
0.02
Model 1
0.005 (0.003)
20.009 (0.004)
0.079
Model 2
*
0.005 (0.003)
20.002 (0.004)
0.039
Model 3
*
0.005 (0.003)*
20.036 (0.029)
0.113 (2.043)
0.193 (1.527)
0.023 (2.872)
0.055
Unadjusted B (SE)
*
0.063 (0.026)
20.068 (0.042)
20.029 (0.040)
20.280 (0.298)
0.051 (21.354)
2.024 (15.971)
20.995 (30.020)
Model 1
Model 2
*
0.05
0.03
0.01
0.02
0.065 (0.026)
20.073 (0.041)**
0.082
0.061 (0.027)
Model 3
*
20.012 (0.040)
0.064 (0.027)*
20.302 (0.292)
1.844 (20.871)
2.834 (15.602)
1.169 (29.334)
0.073
0.055
TABLE 5. Regression models testing associations between maternal predictors and milk protein content
Unadjusted B (SE)
R2
Model 2
Model 3
Model 4
20.004 (0.009)
20.010 (0.014)
0.001 (0.013)
20.075 (0.093)
24.316 (6.674)
21.994 (4.992)
25.359 (9.383)
0.002
0.01
0.00
0.05
20.004 (0.009)
20.010 (0.014)
20.004 (0.009)
20.006 (0.009)
0.000 (0.013)
.007
0.002
20.073 (0.093)
24.496 (6.696)
22.071 (5.006)
25.565 (9.412)
0.0547
All models were adjusted for breastfeeding duration, in months. There were no signicant associations between any maternal predictors and milk protein content in
this sample.
TABLE 6. Regression models testing associations between maternal predictors and milk sugar content
R2
Model 2
0.03
0.08
0.03
0.07
20.016 (0.010)
20.044 (0.016)*
Unadjusted B (SE)
20.018 (0.010)
20.046 (0.016)*
20.026 (0.015)**
20.102 (0.113)
5.696 (8.09)
1.616 (6.051)
6.640 (11.374)
*
Model 3
**
0.099
Model 4
20.022 (0.010)
20.032 (0.015)
0.067
20.015 (0.010)
20.098 (0.112)
5.266 (8.056)
1.432 (6.022)
6.148 (11.323)
0.0844
Energy
Fat
Protein
Sugars
Unadjusted B (SE)
R2
1 BF Dur.
R2
20.009 (0.004)
20.109 (0.040)**
0.009 (0.013)
0.010 (0.016)
0.05
0.07
0.004
0.004
0.008 (0.004)
20.097 (0.040)*
0.008 (0.013)
0.006 (0.016)
0.07
0.11
0.00
0.03
Unadjusted models for self reported daily feeding frequency. Models were then
adjusted for breastfeeding duration. As shown in other studies, both breastfeeding duration and feeding frequency had independent, signicant associations
with milk fat.
*
P < 0.05, **P < 0.01.
DISCUSSION
In this sample of Filipino women living in the metropolitan area of Cebu city, the strongest predictor of variance
in breast milk macronutrients and energy was breastfeeding duration. Few maternal characteristics predicted the
composition of milk that a woman produced. Current
maternal diet was not associated with milk composition,
while BMI was signicantly inversely associated with
Power ML, Oftedal OT, Tardif SD. 2002. Does the milk of Callitrichid monkeys differ from that of larger anthropoids? Am J Primatol 56:117127.
Power ML, Verona CE, Ruiz-Miranda C, Oftedal OT. 2008. The composition of milk from free-living common marmosets (Callithrix jacchus) in
Brazil. Am J Primatol 70:7883.
Prentice A. 1995. Regional variations in the composition of human milk.
In: Jensen RG, editor. Handbook of Milk Composition. Academic Press,
San Diego, CA. p 115221.
Prentice A, Prentice AM, Whitehead RG. 1981a. Breast-milk fat concentrations of rural African women. I. Short-term variations within individuals. Br J Nutr 45:483494.
Prentice A, Prentice AM, Whitehead RG. 1981b. Breast-milk fat concentrations of rural African women. II. Long-term variations within a community. Br J Nutr 45:495503.
Prentice AM, Prentice A, Lamb WH, Lunn PG, Austin S. 1983. Metabolic
consequences of fasting during Ramadan in pregnant and lactating
women. Hum Nutr Clin Nutr. 37:28394.
Prentice AM, Goldberg GR, Prentice A. 1994. Body mass index and lactation performance. Eur J Clin Nutr 48 (Suppl 3):S78S86; discussionS86S89.
Prentice AM, Roberts SB, Watkinson M, Whitehead RG, Paul AA, Prentice
A, Watkinson AA. 1980. Dietary supplementation of Gambian nursing
mothers and lactational performance. The Lancet 316:886888.
Rakicioglu N, Samur G, Topcu A, Topcu AA. 2006. The effect of Ramadan
on maternal nutrition and composition of breast milk. Pediatr Int
48:278283.
Riva E, Verduci E, Giovannini M, Agostoni C. 2004. Early protein intakes
and adiposity: reloaded or downloaded? Acta Paediatr 93:725726.
Rolls BA, Gurr MI, van Duijvenvoorde PM, Rolls BJ, Rowe EA. 1986. Lactation in lean and obese rats: effect of cafeteria feeding and of dietary
obesity on milk composition. Physiol Behav 38:185190.
Ruel MT, Dewey KG, Martinez C, Flores R, Brown KH. 1997. Validation of
single daytime samples of human milk to estimate the 24-h concentration of lipids in urban Guatemalan mothers. Am J Clin Nutr 65:439
444.
Saarela T, Kokkonen J, Koivisto M. 2005. Macronutrient and energy contents of human milk fractions during the rst six months of lactation.
Acta Paediatr 94:11761181.
Strode MA, Dewey KG, Lonnerdal B. 1986. Effects of short-term caloric
restriction on lactational performance of well-nourished women. Acta
Paediatr Scand. 75:2229.
Tigas S, Sunehag A, Haymond MW. 2002. Metabolic adaptation to feeding
and fasting during lactation in humans. J Clin Endocrinol Metab
87:302307.
Van Amburgh ME, Galton DM, Bauman DE, Everett RW, Fox DG, Chase
LE, Erb HN. 1998. Effects of three prepubertal feeding body growth
rates on performance of Holstein heifers during rst lactation. J Dairy
Sci 81:527538.
Villalpando S, Butte NF, Flores-Huerta S, Thotathuchery M. 1998. Qualitative analysis of human milk produced by women consuming a maizepredominant diet typical of rural Mexico. Ann Nutr Metab 42:2332.
Villalpando S, del Prado M. 1999. Interrelation among dietary energy and
fat intakes, maternal body fatness, and milk total lipid in humans. J
Mammary Gland Biol Neoplasia 4:285295.
Villalpando S, del Prado-Manriquez M, Stafford J, Delgado G. 1995. Diurnal variations in the fatty acid composition of milk fat from marginally
nourished women. Arch Med Res 26:S139S143.
Villalpando SF, Butte NF, Wong WW, Flores-Huerta S, Hernandez-Beltran
MJ, Smith EO, Garza C. 1992. Lactation performance of rural Mesoamerindians. Eur J Clin Nutr 46:337348.
Wojcik KY, Rechtman DJ, Lee ML, Montoya A, Medo ET. 2009. Macronutrient analysis of a nationwide sample of donor breast milk. J Am Diet
Assoc 109:137140.
World Health Organization Collaborative Study of Breastfeeding. 1985.
The Quantity and quality of breast milk: Report on the WHO Collaborative Study of Breastfeeding. WHO Library, Geneva.
Xiang M, Harbige LS, Zetterstrom R. Long-chain polyunsaturated fatty
acids in Chinese and Swedish mothers: diet, breast milk and infant
growth. Acta Paediatr. 2005 94:15439.