am j clin nutr 1976 picciano 242 54.pdf hmc

original communications

242 The American Journal of Clinical Nutrition 29: MARCH 1976, pp. 242-254. Printed in U.S.A.

Copper, iron, and zinc contents

of mature human milk’3

Mary Frances Picciano,4 Ph.D., and Helen A. Guthrie,5 Ph.D.

ABSTRACT Daily, weekly, and within-day variations in copper, iron, and zinc contents of

human milk were investigated in order to determine whether one sample from an individual is

representative of these elements. Total solids, fat, and protein contents were also measured. Fifty

women in their 6th to 12th week of lactation each provided seven milk samples consisting of five

consecutive daily samples and two additional samples collected either within a single day or at

weekly intervals. Fat varied the most of all constituents and total milk solids reflected this

variability. Values ranged from 0.2 to 10.4 g/ 100 ml for fat and from 8.58 to 17.49 g/ 100 ml for total

solids. Protein varied from 0.76 to 2.04 g/l00 ml among individuals, with littlevariation within an

individual. Copper content varied considerably among women and within the same woman. With a

large proportion of low values, the range was 0.09 to 0.63 �g/ml. Iron content was also found to

vary within women as well as among women. Values ranged from <0.1 to 1.6 zg/ml with a

preponderance of low values. Zinc content was more evenly distributed over the range of 0.14 to 3.95

.eg/ml, and within an individual it did not vary widely. A representative estimate of copper and

iron contents would therefore require multiple samples, whereas only one sample may provide a

representative estimate of zinc content. Comparison of morning, midday, and evening values showed

that copper and zinc are higher in the morning and iron is lower at this time. Increased amounts of

copper, iron, and zinc were found in multiparous women whether or not they had previously

lactated. Milk from older women had lower iron and higher copper and zinc contents than that from

younger women. No differences were found in milk of women receiving dietary mineral and vitamin

supplements. Calculations indicated that fully breast fed infants under 3 months of age receive

approximately 0.35 mg/kg per day of zinc and 0.05 mg/kg per day of both copper and iron. Am.

J. C/in. Nutr. 29: 242-254, 1976.

The composition of human milk is ofinterest to those concerned with the nutritionof infants, since early feeding regimens arebased on its use or substitutes modified toresemble it. However, it has long been knownthat human milk composition varies widelyamong women and that variations in qualityeven occur within the same woman. If humanmilk is to be used as a standard, it isimportant to establish which nutrients vary,

the extent of the variations, possible factorsexerting an influence, and the effects suchvariations may have on the growth of theinfant. Early investigators (1-6) have identi-fied variations in the macronutrients of

human milk as well as some factors whichinfluence their composition. Considerably

1 From the Nutrition Program, Division of Biological

Health, Pennsylvania State University. Taken in part

from a thesis submitted by M. F. Picciano to the

Graduate School, Pennsylvania State University, inpartial fulfillment of the requirements for the Ph.D.

degree.2 Address reprint requests to M. F. Picciano, 457

Bevier Hall, School of Human Resources and FamilyStudies, University of Illinois, Urbana, Illinois 61801.

Authorized for publication on May 5, 1975 as paper

no. 4860 in the journal series of the Pennsylvania

Agricultural Experiment Station.

Assistant Professor of Nutrition, University of Il-linois. 6 Professor of Nutrition, Pennsylvania State

U niversity.

by guest on August 28, 2013

ajcn.nutrition.orgD

ownloaded from

http://ajcn.nutrition.org/

CU, FE, AND ZN IN MATURE HUMAN MILK 243

less information is available for the micronu-trients, especially the trace elements. Thatvoids exist in our knowledge of trace elementcomposition of human milk has recently beenstressed by the World Health OrganizationExpert Committee on Trace Elements inHuman Nutrition (7). The Committee recom-

mended that highest priority be given to the

analysis of human milk in an international

collaborative effort to obtain data on thetrace element content of foods.

The purpose of this investigation was todetermine the copper, iron, and zinc contentsof human milk and to identify sources ofvariation, since reported values vary widely.The range of reported mean values (zg/ml)

for copper is 0.22 to 1.5 (8-11); for iron, 0.44to 5.0 (9-13); and for zinc, 1.3 to 12.4(10-12). From an analysis of 22 samples, themost recent investigation (10) provided aver-age values for copper, iron, and zinc of 0.24,0.84, and 1.34 zg/ml, respectively. Whetherthese reported variations could be attributed

to individual differences among women or

differences within the same woman wereexamined. Within individuals, patterns of

variation throughout a day, from day to

day, and from week to week were studied.From such information, it is possible todetermine whether one sample of milk from aparticular individual may be considered rep-resentative of copper, iron, and zinc contents.In addition, influences of maternal age, pari-ty, and lactation history were examined.

Subjects and methods

Subjects

Fifty Pennsylvania mothers in the 6th to 12th week of

lactation each provided seven milk samples for analyses.All subjects had healthy full-term infants and were

feeding them exclusively on breast milk without any

supplementary food. Forty-two subjects were betweenthe ages of 20 and 30, while eight were over 30. Sixteen

subjects were primiparae and 34 were multiparae, of

whom four were lactating for the first time.

Maternal nutrient intake

To obtain an indication of the nutritional quality ofsubjects’ diets, 24-hr recalls were obtained at the time of

interview. Nutrient intakes were tabulated using data

cards from the U.S. Department of Agriculture based ontheir Home and Garden Bulletin 72 (14). Additional

cards were prepared as needed from data in AgriculturalHandbook 8 (15) and Food Values of Portions Com-

monly Used (l6). Analyses for kilocalories, protein,vitamin A, thiamin, riboflavin, ascorbic acid, calcium,

and iron were made. Since complete data are not

available for copper and zinc contents of foods, these

nutrients were not tabulated. Comparison with Recom-mended Dietary Allowances (RDA) (17) showed that

with the exception of kilocalories and iron, averageintake either met or exceeded the RDA. However,

average intakes did provide two-thirds of the RDA for all

nutrients tabulated, a level of intake judged to be

adequate.Thirty-seven subjects (74%) were also supplementing

their diets with vitamin-mineral preparations. All ofthese contained �30 mg of iron, and five women (10%)

were taking two such preparations. A few supplements

contained minimal amounts of copper and less than 10%

of the RDA for zinc.

Collection of samples and experimental design

Sample collections were made between April and

August of 1972. As members of nursing mother groups,subjects were instructed to use both breasts at a feeding.The breast used first at any one feeding was emptied andused second at the following feeding. While mothers

adhered to this practice, all samples were collected with apolyethylene hand breast pump or by manual expression

from the second breast before it was offered to the infant.

In order to minimize the possibility of contamination,

sample collection equipment was acid-washed, rinsed inglass-redistilled water, and enclosed in plastic. Subjectsalso received explicit verbal and written instructions

concerning necessary precautions while collecting sam-ples. Samples were frozen immediately after collectionand kept frozen until analyzed.

Subjects each provided one 40-mI sample of milk aday, at the early morning feeding, for 5 consecutive days

(daily period). Two additional samples were provided

either at weekly intervals, also collected at the early

morning feeding (weekly period), or within a single day,collected at midday and evening feedings (within-day

period). All 50 participants therefore each supplied seven

samples of human milk according to the experimentaldesign outlined in Table I. Early morning samples were

collected between the hours of 5:00 and 8:00 AM with a

4- to 8-hr lapse from previous feedings. Midday sampleswere collected around the noon hour and evening samples

at approximately 6:00 PM. Since all subjects practiced

the demand method of nursing, there was only a 2- to

3-hr lapse from previous feeding in the within-day period.

A nalttical methods

Duplicate 5-mI portions of whole milk were placed in

a beaker which had been dried to constant weight. Themilk was then dried over a steam bath and in an oven at

97 to 100 C for 3 to 4 hr. After the samples had beencooled in a desiccator, total milk solids were determined

gravimetrically (18). Subsequently, the same sample wasdigested with concentrated nitric and sulfuric acids in a

4:1 ratio (v/v). Additional nitric acid was added until the

digest was clear. The clear sample was brought to avolume of 5 ml with glass-redistilled water and directly

aspirated into a Perk in-Elmer model 403 atomic absorp-tion spectrophotometer for determinations of copper,

iron, and zinc contents. The spectrophotometric method

of Nakai and Li (19) was used to measure fat and protein

contents.


ajcn.nutrition.orgD

ownloaded from


244 PICCIANO AND GUTHRIE

TABLE I

Experimental design

Time of collection

Early morning Midday Evening

Week I

Day I

Day II

Within-day IllS Day 1110r-�I I

ii 1 � (____.Day lV0.SH8.. Within-day IV

I8.1.�;‘I.�I

�

Day I� �Weck II

Week IIIDayl

Within-day III

IWithin-day Ill

I

Within-day IV

I

Within-day period (n =

Within-day IV

10)

Within-day period (n = 15)

Within-day samples were collected on either day III or day IV of the daily period.

Statistical methods

The total variance of each milk constituent wasquantitatively partitioned among the sources of variationand expressed as percentage of the total. Sources ofvariation were among mothers, among samples within

mothers, and among duplicates within samples (experi-mental error). An analysis of variance with a randomeffects model and a hierarchical classification of data wasused to measure the relative intensities of sources ofvariation. Differences among means in the within-day

period were determined by Duncan’s modified least

square difference test, and influences of dietary supple-

ments, maternal age, parity, and lactation history wereassessed with Student’s t test (20).

Results

Analyses of variance for total solids, fat,and protein contents are presented in Table 2for daily, weekly, and within-day periods.Total solids and fat content varied similarlyfrom day to day and week to week. In eachcase, variations among subjects accounted for50 to 60% and variations among samplesaccounted for 40 to 50%. In contrast, approx-imately 80% of total within-day variances forthese constituents is attributable to variationsamong samples. In all experimental periods,only 5% of total variances in protein contentis observed among samples, while among-subject differences accounted for 93%.

Analyses of variance for copper contentexpressed as micrograms per milliliter ofmilk, micrograms per gram of milk solids,

and micrograms per gram of solids-not-fatare presented in Table 3 for daily, weekly,and within-day periods. All ratios of parti-tioned variances are significant, yet variationpatterns differed in each experimental period.Regardless of expression basis, copper con-tent in the daily period varied in a similarfashion, with approximately 70% of totalvariance occurring among subjects and 22%

occurring among samples within subjects. Inthe weekly period, variation in copperamong samples was about 28% and variationamong rubjects was 55% when expressed asmicrograms per milliliter of milk or micro-grams per gram of solids-not-fat. However,when considered on the basis of total milk

solids, weekly variance among samples onlyaccounted for 18% of total and among sub-jects for 67%. In contrast, within-day coppervariance was greatest among samples when

expressed as micrograms per gram of milksolids. Approximately 21 % is ascribable tovariations among samples within subjects and77% to variations among subjects. Micro-

grams of copper per milliliter of milk or pergram of solids-not-fat varied similarly, withvariations among subjects accounting for

85%, and with variations among samples ac-counting for only 12%.

Under our conditions of analysis previouslydescribed, the smallest detectable amount of


ajcn.nutrition.orgD

ownloaded from


TABLE

Means and analysis of variances for total solids, fat, and protein contents of human milk

Milkconstituent(g/lOOml)

Mean ± SD Range Source of variance df Meansquare

Fva1ue

Variance

%ofObserved total

Daily period

Total solids

Fat

Protein

Weekly periodTotal solids

Fat

Protein

11.85 ± 1.43 8.58- 17.49 Total 499 2.03 100.00Amongsubjects 49 13.63 8.500 1.20 59.16Among samples 200 1.60 29.390 0.77 38.06Among replicates 250 0.06 0.06 2.78

3.05 ± 1.91 0.20-10.00 TotalAmong subjects

Among samples

Among replicates

49949

200250

23.48

3.220.02

7.300145.660

3.65

2.031.60

0.02

100.00

55.5743.82

0.61

1.28 ± 0.24 0.76-2.04 Total

Among subjectsAmong samples

Among replicates

49949

200250

0.552

0.0070.009

76.670

7.680

0.0590.054

0.0030.001

100.0093.05

5.351.60

11.70 ± 1.35 8.80-14.67 TotalAmong subjectsAmongsamples

Among replicates

149

2450

75

7.781.55

0.02

5030

l0l.8I#{176}

1.82

1.040.77

0.02

100.0057.11

42.05

0.832.91 ± 1.68 0.20 7.20 Total

Among subjects

Among samples

Among replicates

149

24

50

75

11.34

2.80

0.01

4.060

205.620

2.83

1.42

1.39

0.01

100.00

50.34

49.19

0.48

1.25 ± 0.28 0.79-2.04 Total

Among subjects

Among samples

Among replicates

149

2450

75

0.429

0.0 10

0.001

43.150

9.980

0.0754

0.0699

0.00450.0010

100.0092.74

5.94

1.32

8.58-17.49 Total

Amongsubjects

Among samples

Among replicates

149

24

5075

4.40

2.590.02

1.70

11.650

1.61

0.30

1.290.02

100.00

18.71

79.901.38

3.72 ± 1.90 0.20-10.00 Total


Among replicates

149

24

5075

9.983

5.7980.003

1.72

1840.690

3.598

0.698

2.8970.003

100.00

19.39

80.530.09

1.30 ± 0.22 0.76-2.04 TotalAmong subjects

Among samples

Among replicates

14924

5075

0.2658

0.00490.0011

53370

4540

0.04650.0435

0.0039

0.0011

100.00

93.474.18

2.36

Within-day period

Total solids 12.34 ± 1.27

Fat

Protein

�P < 0.001.


iron was 0.1 �zg/ml, and 40 samples werefound to contain less than this amount. Thisdoes not necessarily imply that no iron waspresent, and could be explained by insensitiv-ity of the analytical technique. When nodetectable amounts were found, the value0.05 �zg iron/ml of milk was assigned for

purposes of computation. Data showingmeans and analyses of varia.pce for ironcontent of daily, weekly, and within-dayperiods are presented in Table 4. Even though

variations in each experimental period arepatterned differently, all ratios of partitionedvariance are significant. Most of the variation

in iron content during the daily period isascribable to variations among sampleswithin subjects (53%) regardless of expressionbasis. Among-subject variations account forapproximately 39%. In the weekly period,variances among samples ranged from only9% when iron was expressed as microgramsper gram of solids-not-fat to 21 % when


ajcn.nutrition.orgD

ownloaded from


TABLE 3Means analysis of variances for iron content of human milk samples

VarianceMeanMilk constituent Mean ± SD Source of variances df �uare

Fvalue0 Observed oftotal

Daily periodCu (pg/mI)

Cu (pg/g milk solids)

Cu (pg/g solids-not-fat)

Weekly period

Cu (pg/ml)

Cu (j.eg/gmilk solids)


Within-day period

Cu (pg/mI)

Cu (pg/g milk solids)


0.245 ± 0.077 Total 499 0.0060 100.00

Among subjects 49 0.0464 15.99 0.0045 72.19

Among samples 200 0.0029 6.44 0.00 12 20.34Among replicates 250 0.0005 0.0005 7.47

2.08 ± 0.64

2.81 ± 0.92

Total

Among subjects

Amongsamples

Among replicates

TotalAmong subjects

Among samples

Among replicates

499

49

200

250

49949

200250

3.06

0.21

0.04

6.08

0.500.08

14.346.02

12.29

6.01

0.41

0.280.09

0.040.850.56

0.210.08

100.0069.5921.75

8.66100.00

65.94

24.349.71

0.232 ± 0.069

1.98 ± 0.56

2.66 ± 0.81

TotalAmong subjects

AmongsamplesAmong replicatesTotal

Among subjects

Among samples

Among replicates

Total

Among subjects

Among samplesAmong replicates

14924

50

75149

2450

75

149

24

5075

0.0201

0.0032

0.0006

1.400.16

0.05

2.55

0.530.13

62.34

5.15

8.853.37

4.86

3.94

0.00470.0028

0.0013

0.00060.31

0.210.060.05

0.66

0.34

0.200.13

100.0059.57

27.66

12.77100.00

66.8717.9815.15

100.00

50.63

29.3620.01

0.228 ± 0.077

1.85 ± 0.62

2.72 ± 0.97

TotalAmong subjects

Among samplesAmong replicatesTotal

Among subjects

Among samplesAmong replicatesTotal

Among subjects

AmongsamplesAmong replicates

1492450

75149

24

5075

149

24

5075

0.03230.00 14

0.0001

1.9470.170

0.009

5.01

0.280.02

23.299.64

11.4318.78

18.12

1.2.76

0.0059

0.0052

0.0006

0.00010.386

0.2960.081

0.0090.938

0.789

0.1270.022

100.0087.0710.50

2.43100.00

76.7420.90

2.35100.00

84.11

13.582.31

0� <0.001.


expressed per gram of milk solids. Variationsamong subjects ranged from 59% (micro-grams per gram of milk solids) to as high as78% (micrograms per gram of solids-not-fat).Thirty-nine percent of the variation in ironcontent is attributable to among-sample vari-ances and 50% is attributable to among-sub-ject variances in the within-day period.

Means and analyses of variance for zinccontent are presented in Table 5. As withcopper and iron, all ratios of partitioned

variances for zinc content are significant.Comparable patterns of variation were ob-served in daily and weekly periods. Most ofthe total variance (85%) is derived fromamong subjects, while only 13% is derivedfrom among samples. During the within-dayperiod, zinc expressed per milliliter of milk orper gram of solids-not-fat displayed parti-tioned variances similar to daily and weeklyperiods. However, nearly twice as much vari-ation (24%) was observed among samples


ajcn.nutrition.orgD

ownloaded from


TABLE 4Means and analysis of variances for iron content of human milk samples

VarianceMeanMilk constituent Mean ± SD Source of variances df square

FValue#{176}Observed % of

total

Daily periodFe (pg/mI)

Fe (pg/g milk solids)

Fe (pg/g solids-not-fat)

Weekly period

Fe (pg/mI)



Within-day period

Fe (pg/mI)



0.202 ± 0.17 Total 499 0.030 100.00

Among subjects 49 0.158 4.73 0.013 41.32

Among samples 200 0.034 16.73 0.0 16 52.06Among replicates 250 0.002 0.002 6.62

1.75 ± 1.38 Total

AmongsubjectsAmongsamplesAmong replicates

499

49200250

9.282.270.14

4.10

16.29

1.90

0.70

1.060.14

100.0036.85

55.857.31

2.35 ± 2.03 Total

Amongsubjects

Amongsamples

Among replicates

499

49200

250

21.314.64

0.27

4.5917.37

4.12

1.67

2.19

0.27

100.0040.44

53.07

6.49

0.232 ± 0.190 Total

AmongsubjectsAmong samples

Among replicates

149

245075

0.1660.0 170.005

9.643.20

0.036

0.025

0.0060.005

100.00

68.68

16.4114.91

2.02 ± 1.46 Total

Among subjectsAmongsamples

Among replicates

1492450

75

8.811.26

0.38

6.99

3.29

2.121.26

0.440.38

100.0059.25

20.6418.04

2.76 ± 2.50 Total


Among replicates

149

24

5075

31.41

1.940.79

16.18

2.47

6.28

4.91

0.580.79

100.00

78.26

9.21

12.54

0.19 ± 0.35 Total

Among subjects

Among samples

Among replicates

149

24

50

75

0.048

0.0 10

0.001

4.70

8.44

0.012

0.0060.005

0.001

100.0052.4437.48

10.08

1.54 ± 0.75 Total

Among subjects

Among samples

Among replicates

14924

50

75

2.30

0.42

0.08

5.46

5.46

0.056

0.031

0.0 17

0.008

100.00

55.6630.60

13.732.25 ± 1.31 Total

Among subjects


149

2450

75

6.141.80

0.17

3.40

10.54

1.7160.725

0.8190.172

100.0042.25

47.7410.01


0P <0.001.

when zinc was expressed per gram of milksolids, with less variability attributable toamong subjects (75%).

Frequency distributions (Figs. 1 to 3) ofcopper, iron, and zinc contents were similar inall experimental periods regardless of expres-sion basis. Therefore, data were pooled andpresented only on the basis of volume of milk(milliliters). Since a preponderance of lowervalues for copper and iron content werefound, maximum frequencies of 0.20

�.tg/ml and 0.10 .tg/ml are slightly lowerthan mean values of 0.24 �g/ml and 0.21

�ig/ml, respectively. Zinc content was moreevenly distributed over the reported range,with the maximum frequency not different

from the mean value of 1.62 zg/ml.Mean values for milk constituents of morn-

ing, midday, and evening samples are shownin Table 6. Lower total solids content wasfound during the morning than during eithermidday or evening feedings. Fat content


ajcn.nutrition.orgD

ownloaded from


TABLE 5Means and analysis of variances for zinc content of human milk samples

VarianceMeanMilk constituent Mean ± SD Source of variances df square F

value0 Ob served % oftotal

Daily period

Zn (pg/mI)

Zn (pg/g milk solids)

Zn (pg/g solids-not-fat)

Weekly periodZn (pg/mI)



Within-day period

Zn (pg/mi)



1.68 ± 0.78 Total 499 0.602 100.00Amongsubjects 49 5.399 35.92 0.525 87.19

Amongsamples 200 0.150 40.75 0.073 12.18Among replicates 250 0.004 0.004 0.61

14.15 ±6.26 Total

Among subjects

Among samplesAmong replicates

49949

200250

342.93

11.690.34

29.35

34.60

39.1433.12

5.670.34

100.00

84.64

14.50

0.86

19.25 ± 9.37 Total

Among subjects


499

49

200

250

769.45

26.14

0.98

29.43

26.79

87.8874.33

12.58

0.98

100.0084.58

14.31

1.11

1.59 ± 0.84

13.47 ± 6.69

18.27 ± 9.65

TotalAmong subjects

Among samplesAmong replicatesTotalAmong subjectsAmongsamplesAmong replicates

Total


Among replicates

14924

5075

149245075

149

2450

75

3.7860.1890.009

239.8513.310.66

500.17

26.57

1.69

20.0720.55

18.0220.21

18.83

15.74

0.699

0.6000.090

0.00944.74

37.766.33

0.66

93.07

78.93

12.44

1.69

100.00

85.83

12.85

1.32

100.00

84.3914.14

1.47

100.0084.81

13.37

1.81

1.58 ± 0.81 Total

Among subjects

Among samples

Among replicates

149

24

50

75

3.5750.174

0.004

20.5640.45

0.656

0.5670.085

0.004

100.00

86.4212.93

0.66

12.81 ± 6.37 Total

Among subjects

Among samples

Among replicates

149

24

50

75

202.56

20.05

0.37

9.85

54.03

40.63

30.42

9.84

0.37

100.00

74.87

24.22

0.91

18.82 ± 10.02 TotalAmong subjects

Among samples

Among replicates

14924

50

75

535.04

31.85

1.06

16.80

30.13

100.32

83.87

15.39

1.06

100.00

83.60

15.35

1.05

0� <0.001.


varied significantly at each feeding, with thehighest value observed in the evening and thelowest in the morning. Protein concentrationswere lower in morning than in evening milk.Morning milk was higher in copper and zincand lower in iron compared to other feedingsduring the day. These differences were notapparent when minerals were expressed pergram of solids-not-fat.

No effects of vitamin and mineral supple-ments were detected. The influences of age,

parity, and lactation history on milk compo-sition are presented in Table 7. No differenceswere observed in total solids. Milk fromwomen over 30 years old had a lower fat

content than that from women under 30.Samples from women with additional chil-dren had higher fat contents than those fromprimiparae. Milk from older mothers had agreater protein concentration than that fromyounger mothers.

Mothers over 30 had increased copper


ajcn.nutrition.orgD

ownloaded from


80

U)

-J

a.U)

66.

Ui

z

60

40

U)Lii

-J

a..eU)

Li.

0

Ui

z

20

120

80

60

40

20

0

FIG. 2. The distribution of iron in mature humanFIG. I. The distribution of copper in mature human

milk.

i.Lg/mI


MEAN: 0.24±0.08RANGE: 0.09-0.63

N : 350

contents per gram of milk solids. However,this difference was not apparent when copperwas expressed per milliliter of milk or per

gram of solids-not-fat. Women with addi-tional children and lactation histories hadmilk with higher copper contents than primi-parous women. Iron content was lower inolder than in younger women. In contrast,women with more than one child and lacta-

tion histories had a higher iron content thanprimiparae. Increased age, multiparity, andlactation history are all associated with in-creased amounts of zinc in milk.

Discussion

As a basis for interpreting results, certaincharacteristics of the subjects and the sam-pling procedure should be considered. Sub-jects volunteered to take part in this study;therefore, willingness to provide samples be-came a selection criterion along with abilityto lactate successfully. All analyses weremade on foremilk-the first milk secreted ata feeding. It is generally agreed that fat

milk.

MEAN : 0.21 ±0.17RANGE :(0I0 - .6

N : 350

content is at its lowest in foremilk and at theearly morning feeding when the majority ofsamples were taken; therefore, data on fat arenot representative of the amount in total milksecretion. Furthermore, the fact that thecaloric content of milk is largely dependenton its fat content precludes the use of thesedata for calculations of energy intakes.

Fat varied the most of all constituentsstudied among women and among samplesfrom the same woman. Total solids patterns

reflected the wide variability of fat content.Values ranged from 0.20 to 10.4 g/ 100 ml forfat content and from 8.58 to 17.49 g/l00 mlfor total milk solids. The most striking obser-�ation was that during the within-day period,80% of the total variance for fat content wasascribable to variations among sampleswithin women, masking differences amongwomen. Kon and Mawson (21) observed that

the mean fat content was high when theinterval from the last feeding was short andlow when the interval was long. Since allsubjects fed their infants on demand, intervalsbetween feedings varied between 2 and 4 hr


ajcn.nutrition.orgD

ownloaded from


80MEAN I 63 ± 0.78

70

60

30

20

l0

U)Ui-Ja.4U)

Li.

0

Lii

D

z

milk.

TABLE 6

Comparison of morning, midday. and evening

mean values of milk constituents

0

FIG. 3. The distribution of zinc in mature human

11


Milk constituentTime of day

Morning Mtdda� Evening

Totalsolids(g/l00ml) ll.70ab l2.Sla l2.80b

Fat(g/lO0rnl) 2.96a 3.99a 4.Sla

Protein(g/lOOml) l.29a 1.30 l.3la

Copperpg/mI milk 0.231 0.22� 0.224

pg/g milk solids I.97ab I.84a I.Thb

pg/g solids-not-fat 2.68 2.70 2.76

Iron

pg/mI milk 0.l53ab 0.200a 0.21 lb

pg/g milk solids l.35ab l.61a l.67b

pg/gsolids-not-fat 2.59 2.38 1.79

Zincpg/mI milk l.7lab l.54a l.49b

pg/g milk solids l4.56ab l2.25a 11.6 lb

pg/g solids-not-fat 20.00 18.44 18.03

0 Means followed by the same letter arc significantI�

different (P < 0.05).

during the day, possibly contributing to thelarge within-day variations in fat content.However, previous workers (2, 4, 24) have re-

ported variations equally as large throughout

the day, even when all of the milk wassampled at scheduled intervals. Thus, mostinvestigators stress that a complete 24-hrsample must be analyzed for a representativeestimate of fat content. These data clearlysupport this conclusion.

Daily and weekly variations in fat contentcan be attributable to individual differences(58%). Even though all samples were ob-tamed in the early morning at least 4 hr fromthe last feeding, substantial variations (40%)among samples within subjects were ob-served. Large daily and weekly within-subjectvariations suggest that even a 24-hr milksample may not be representative of the fat

content for an individual. In support of thisview, Salmi (25) found differences of about 2

g/ 100 ml for fat content of 24-hr samples ineach of his six subjects over a 14-day period.

Comparison of morning, midday, and eve-ning values confirmed the work of earlyinvestigators that milk fat is lowest in the

morning (2, 4). However, previous investi-gators observed maximum values midmorn-ing with a steady decline thereafter, in con-

trast to this study, where maximum valuesoccurred in the evening. A midmorning peakcould not have been detected, since no sam-ples were taken at that time, whereas a highevening value may have resulted from shortintervals between feedings.

Most of the variations that were observedin protein content were attributable to indi-vidual differences among subjects (95%).With an average protein content of 1.28g/lOO ml, values ranged from 0.76 to 2.04g/ 100 ml. However, within an individual littlevariation occurred. Although a slightly lowerprotein concentration occurs in the morningthan at the evening feeding, it is not of amagnitude to be nutritionally significant.These observations are consistent with earlierfindings (4, 24) that one sample of humanmilk from an individual would furnish arepresentative estimate of protein content.

Before discussing general trends observedfor copper, iron, and zinc contents, considera-tion must be given to the manner in whichthese elements are expressed. Examination ofour results revealed that partitioned variancesfor minerals expressed per gram of milksolids often showed different patterns fromthose expressed per milliliter of milk or pergram of solids-not-fat. Data expressed on the


ajcn.nutrition.orgD

ownloaded from



bly be obtained from analysis of only one

TABLE 7Influence of maternal age, parity, and lactation history

Milk constituentA )

ge(yr(no. of p

Parityrevious children)

Lactation history(no. of times)

20-30 >30 0 lormore 0 lormore

Totalsolids(g/IOOml)

Fat(g/lOOml)

Protein(g/lOOml)

Copper (pg/mI)

Copper(pg/g milk solids)

Copper (pg/g solids-not-fat)

Iron(pg/ml)Iron (pg/g milk solids)

Iron(pg/gsolids-not-fat)

Zinc(pg/ml)Zinc(pg/g milk solids)Zinc (pg/g solids-not-fat)

11.93

3.26

1.26

0.24

1.99

2.76

0.211.83

2.51

1.5512.98

18.48

11.79

2.76#{176}

l.38C

0.25

2.lY’

2.73

0.18#{176}l.55c

2.00’

2.00’17.02’

20.98c

11.84

2.92

1.27

0.22

1.84

2.47

0.171.55

1.96

1.3911.8815.64

11.94

3.286

1.29

0.25c

2.lOi

2.89C

0.22Cl.89c

2.65’

l.75c14.57c

20.4 IC

11.87 11.93

3.03 3.26

1.29 1.27

0.23 0.25C

1.90 2.09C

2.57 2.89�

0.17 0.2Y1.54 l.94c

2.00 2.7lc1.46 l.75c

12.38 14.59’16.67 20.39’

0� <0.01. 6P <0.05.C� <0.001.

basis of milk solids may be misleading be-cause milk solids contain the wide variabilityof fat content. Thus, discussion of variationpatterns for copper, iron, and zinc is based onanalytical values expressed in terms of eitherwhole milk (micrograms per milliliter) orsolids-not-fat (micrograms per gram).

Copper content varied between 0.09 and0.63 .ig/ml with a mean value of 0.24�g/ml. The frequency distribution of allsamples analyzed deviates from a normalbell-shaped curve, with values skewed to theright and a clumping of lower values. Thisdeviation from a normal distribution suggeststhat there may be a physiological lower limitof copper content. That is, the lower valuesmay represent the minimal amount of copperthat is present in human milk.

Among samples from the same individual,copper varied more extensively over a periodof days and weeks than in a single day.Variations in individuals for the within-dayperiod (12%) were of such a small magnitudeas to indicate that one sample of milk wouldprovide a representative estimate of coppercontent on a particular day. However, parti-tioning of variances for daily and weeklysamples showed that approximately 25% ofthe total is attributable to variations withinindividuals. Thus, it appears that several milksamples taken over periods of days and weeksare necessary to furnish a reliable estimate ofcopper content for an individual.

Compared with copper, values for ironexhibited a much wider range (<0.1 to 1.6�g/ml), while the mean value was similar

(0.21 zg/ml). Like copper, the frequency

distribution for all samples deviates from a

normal curve with skewing to the right. Thus,the clustering of lower values may also beinterpreted to represent the minimal amountof iron in human milk.

Analysis of weekly samples showed thatvariations within individuals only accounted

for approximately 13% of the total variances,implying that little variation in iron content

occurs with the progression of lactation. Incontrast, daily and within-day analyses illus-trated that iron content varied considerablyamong samples of individuals. In theselater periods, as much as 53% of observedvariance is ascribable to variations occurring

within individuals. Results of these analysesare strongly suggestive that for a particular

individual one sample of milk does not pro-vide a representative estimate of iron content.

Zinc content, with a mean value of 1.63

zg/ml, is considerably higher than eithercopper or iron content. As can be seen fromthe frequency histogram, values vary widelyfrom 0.14 to 3.95 �g/ml. Unlike copper

and iron, zinc values are more evenly distrib-uted over the reported range. The amount ofzinc present in human milk appears to be anindividual characteristic. Irrespective of ex-perimental period, observed variance can bealmost totally accounted for by variationsamong mothers (85%). Since variationswithin mothers were of such a small magni-tude (14%), a representative estimate of zinccontent in mature human milk could proba-


ajcn.nutrition.orgD

ownloaded from



sample. However, this has only been shown tobe valid over a 2-week period.

When comparisons of morning, midday,and evening values were made, it was foundthat copper and zinc contents were higher inthe morning while iron was lower at this time.Since these differences were not evident whenminerals were expressed per gram of solids-not-fat, it appears that fat content may havean influence on the amounts of these elementspresent in human milk. In bovine milk, thelargest part of both copper and zinc is boundto the protein fraction (26, 27), whereasonly one-half of the iron is bound to thisfraction, the rest being associated with themilk fat (28). Assuming similarity inhuman milk, low-fat samples might be ex-pected to contain less iron and more copperand zinc than samples with a high fat content.That this may be the situation is inferredfrom these results since, as previously indi-cated, milk fat is lowest in the morning.

No differences in the trace element con-tent of milk from women receiving vitamin-mineral preparations were noted even thoughthese women were receiving at least 30 mg ofiron/day. This finding is in agreement withprevious investigators who have been unsuc-cessful in attempts to raise the iron level inmilk with dietary supplementation (29-31).Likewise, there is no evidence to indicate thatthe level of dietary copper influences theamount in milk (32); however, Berfanstam(33) was able to raise milk zinc content with

supplementation, suggesting that actual levelsmay depend upon maternal intakes.

A recent investigation by Murthy and Rhea(10) has provided data on copper, iron, andzinc contents in 22 samples of milk from 13women in the Cincinnati area. Mean copper

and zinc contents reported are essentially thesame as found in this study. This similarity ofvalues obtained from lactating women livingin two different geographic areas suggest thatthere may be a homeostatic mechanism oper-ating to maintain milk copper and zinc con-tents within narrow limits. However, zinclevels may only reflect a similarity in dietaryintakes. Mean iron content reported by theseinvestigators was 4 times greater than in ourstudy. It is not immediately apparent whythere is such a discrepancy with respect toiron content. Their mean value is within our

observed range and possibly lack of agree-ment is due to sample size differences.

The composition of milk is often linkedwith the age of the mother and the number of

pregnancies and lactations she has ex-perienced. However, these effects are usuallyoperating simultaneously because both multi-parity and lactation history are related toincreased age. Although age- and parity-related differences in fat content were ob-served, it is not possible to determine whetherthey are real or apparent because the samplesanalyzed were not representative of fat con-tent. Protein values were representative andfound to be higher in older than in youngerwomen. Recognizing that of the eight womenin this study over 30 years of age, two hadmilk with a relatively high protein content,this result may be an artifact of small num-bers. After critically reviewing the literatureon this phenomenon, Morrison (29) postu-lated that protein content of milk declineswith age. With respect to copper, iron, andzinc contents, both multiparity and lactationhistory appear to be associated with highamounts. Also, higher copper and zinc con-tents as well as a lower iron content werefound in older than in younger women. Whilethese data are suggestive that age, parity, andlactation history exert an influence on copper,iron, and zinc contents, a more reliableindication of the effects of these factors wouldprobably require longitudinal rather thancross-sectional data.

In view of the fact that human milk is stillconsidered the best food for the young infant,“it is believed that the assessment of thenutritional requirements of infants can bebased on the composition of milk of well-nourished mothers” (21). Analyses were

made on 350 samples of milk from motherswho were judged to be adequately nourished.

Since each woman provided seven samples,average values for copper, iron, and zinc wereconsidered representative assuming thatamounts of elements in foremilk are repre-sentative of the entire feed. Estimations ofinfant intakes were calculated utilizing a milkyield of 850 ml and an average infant weightof 4 kg. From such calculations, it wasestimated that these fully breast fed infantsless than 3 months of age were ingestingapproximately 0.05 mg/kg per day of both


ajcn.nutrition.orgD

ownloaded from



copper and iron and 0.35 mg/kg per day ofzinc. The Food and Nutrition Board pro-posed that infants require between 0.05 and0.1 mg/kg per day of copper and set the RDAat 0.08 mg/kg per day (17). This recom-mended intake corresponds to a copper con-

tent of about 0.4 jzg/ml milk, a level foundin approximately 15% of our samples. The

recommended iron intake of 1.66 mg/kg perday is about 30 times greater than thatreceived by our infants. For the totally breastfed infant to meet the RDA for iron, humanmilk would have to contain 7 �zg/ml, avalue considerably higher than any recorded

in this investigation. The recommended zincintake of 0.5 mg/kg per day would be met ifinfants consumed human milk containing 3 to4 �zg/ml. Only about 15% of our valueswere in this range: therefore, the majority of

our infants were receiving less than the rec-ommended intake of zinc.

Since most infants in the United States arenot breast fed (34, 35), it is of interest tocompare the levels of copper, iron, and zinc inhuman milk with those in cow’s milk andformula preparations commonly used in in-fant feeding. The range of reported values forcopper, iron, and zinc contents (microgramsper milliliter) of cow’s milk are 0.05 to 0.30,0.11 to 1.0, and 1.9 to 6.0, respectively (36,37). Formulas based on cow’s milk werefound to contain from 0.13 to 0.25 �g/mlcopper, 0.0 to 15.0 jzg/ml iron, and 1.8 to3.5 zg/ml zinc. Generally, soy-based for-mulas contained relatively high amounts,with values ranging (micrograms per milli-liter) from 0.32 to 0.5 for copper, 5.0 to 11.6for iron, and 2.2 to 3.8 for zinc (38).

It is difficult to assess the adequacy of theselevels of trace elements because little isknown about their availability. Moreover, itis becoming apparent that quantitative inter-relationships among various trace elementsmay play a more important role nutritionallythan the absolute amount of a particularelement. For example, Priev (39) reportedthat the amount of iron assimilated by thenewborn is dependent upon the amount ofcopper in the diet. Widdowson and co-work-ers (40) observed that full-term breast fedinfants retained more copper than preterminfants given either cow’s milk or a preparedformula which contained added copper mak-

ing the concentration similar to human milk.These authors suggested that poor copperretention in the preterm infants may haveresulted from high zinc intakes, a factorknown to depress copper absorption. Thezinc to copper ratio of human milk is 6-8,while it is 30-38 in cow’s milk. However,Walravens and Hambidge (41) recently re-ported that infants fed a zinc-supplementedformula (4 zg/ml) grew more rapidly andhad higher plasma zinc levels than infantsfed the same formula without added zinc(1.8 zg/ml). In this case, zinc supplementa-tion increased the zinc to copper ratio from 5to 17. Additional studies are obviouslyneeded to determine trace mineral needs dur-ing infancy.

Summary

Values for copper, iron, and zinc of 350

samples of mature human milk from 50women were found to vary between 0.09 and0.63 �tg/ml, <0.1 and 1.6 �tg/ml, and0.14 and 3.95 �zg/m1, respectively. Resultsfrom this study imply that with respect tocopper and iron content, analysis of multiplesamples of milk from an individual is re-quired to furnish a representative estimate ofthese trace elements. Conversely, only onesample of milk from an individual mayfurnish a representative estimate of zinc con-tent. Results were also suggestive that age,parity, and lactation history exert an influ-

ence upon amounts of these trace elementspresent. From these data on copper, iron, andzinc content, it was estimated that fullybreast fed infants receive 0.35 mg/kg per dayof zinc and 0.05 mg/kg per day of bothcopper and iron. fl

The authors gratefully acknowledge the Nursing

Mothers’ Committee of the Childbirth Education Associ-ation and La Leche League International for their help in

locating subjects and the cooperation of subjects makingthis study possible.

References

1. MACY, I. G. Composition of human colostrum andmilk. Am. J. Diseases Children 78: 589, 1949.

2. GUNTHER, M., AND J. E. STANIER. Diurnal variationin the fat content of breast milk. Lancet I: 235, 1949.

3. HYTTEN, F. E. Clinical and chemical studies inhuman lactation. II. Variation in major constituents

during a feeding. Brit. Med. J. I: 176, 1954.4. HYTTEN, F. E. Clinical and chemical studies in


ajcn.nutrition.orgD

ownloaded from



human lactation. III. Diurnal variation in major

constituents of milk. Brit. Med. J. 1: 179, 1954.

5. HYTTEN, F. E. Clinical and chemical studies in

human lactation. IV. Trends in milk composition

during course of lactation. Brit. Med.J. 1:249, 1954.

6. HYTTEN, F. E. Clinical and chemical studies in

human lactation. V. Individual differences in compo-

sition of milk. Brit. Med. J. 1: 253, 1954.7. Report of a WHO Expert Committee. Trace Ele-

ments in Human Nutrition. Geneva: World Health

Organ. Tech. Rep. Ser. 1973, No. 532.8. ARCHIBALD, J. C. Trace elements in milk: a review,

part I. Dairy Sci. Abstr. 20: 712, 1958.9. MACY, I. G., H. KELLY AND R. SLOAN. The

Composition of Milk. Washington, D.C.: National

Research Council, 1953.

10. MURTHY, G. K., AND U. S. RHEA. Cadmium,copper, iron, lead, manganese, and zinc in evapo-

rated milk, infant products, and human milk. J.

Dairy Sd. 54: 1001, 1971.

II. CAVELL, P. A., AND E. M. WIDDOWSON. Intakes

and excretions of iron, copper, and zinc in the

neonatal period. Arch. Diseases Childhood 39: 496,

1964.

12. ARCHIBALD, J. C. Trace elements in milk: a review,

part II. Dairy Sci. Abstr. 20: 801, 1958.

13. FEUELLEN, Y. M., AND M. PLUMIER. Iron metabo-

lism in infants. I. The intake of iron in breast feedingand artificial feeding. Acta. Paediat. 41: 138, 1952.

14. UNITED STATES DEPARTMENT OF AGRICULTURE.

Nutritive Value of Foods: Home and Garden Bul-

letin 72. Washington, D.C.: U.S. Govt. Printing

Office, 1971.

15. WATT, B. K., AND A. L. MERRILL. Composition of

Foods: Raw, Processed, Prepared. AgricultureHandbook 8, Washington, D.C.: U.S. Govt. Print-

ing Office, Rev. Dec. 1963.t6. BOWES, A., AND C. F. CHURCH. Food Values of

Portions Commonly Used (11th ed), edited by

C. F. Church and H. N. Church. Philadelphia:J. B. Lippincott, 1970.

17. NAIL. ACAD. SCI.-NATL. RES. COUNCIL. Recom-

mended Dietary Allowances (8th ed). Washington,

D.C.: Natl. Acad. Sci., 1973.18. ASSOCIATION OF OFFICIAL AGRICULTURAL CHEM-

ISTS. Official Methods of Analysis (11th ed). Wash-

ington, D.C., 1970.

19. NAKAI, S., AND A. C. LI. Spectrophotometric deter-

mination of protein and fat in milk simultaneously.

J. Dairy Sci. 53: 276, 1970.

20. STEEL, R. G. D., AND J. H. TORRIE. Principles andProcedures of Statistics. New York: McGraw-Hill,

1960.

21. KON, S. K., AND E. H. MAWSON. Human Milk:

Wartime Studies of Certain Vitamins and Other

Constituents. Medical Research Council Special

Report Series. London: His Majesty’s Stationery

Office, No. 269, 1950.

22. Deleted in proof.23. Deleted in proof.

24. NIM5, B., I. G. MACY, H. A. HUNSCHER AND M.

BROWN. Human milk studies. X. Daily and monthly

variations in milk components as observed in two

successive lactation periods. Am. J. Diseases Chil-

dren 43: 1062, 1932.

25. SALMI, T. Composition and fat content of humanmilk with reference to war-time food conditions.Acta Paediat. 32: 1, 1944.

26. PARKASH, S., AND R. JENNESS. Status of zinc incow’s milk. J. Dairy Sci. 50: 127, 1967.

27. DELLS, W. L., AND J. M. NELSON. Isolation of a

copper bearing protein from cow’s milk. J. Am.

Chem. Soc. 64: 1616, 1942.

28. SCHXFER, K. H., A. M. BREYER, W. HORST, H.KARTE AND W. LENZ. Die protein bindung des eisens

in der much. KIm. Wochschr. 34: 300, 1956.

29. MORRISON, S. D. Human Milk: Yield, Proximate

Principles and Inorganic Constituents. Technical

Communication of the Commonwealth Bureau ofAnimal Nutrition. Aberdeen: University Press, No.

18, 1952.

30. KARMARKAR, M. G., AND C. V. RAMAKRESHNAN.

Studies on human lactation: relation between the

dietary intake of lactating women and the chemicalcomposition of milk with regard to principal and

certain inorganic constituents. Acta Paediat. Belg.49: 599, 1960.

31. STURGEON, P. Studies on iron requirements in

infants. III. Influence of supplemental iron during

normal pregnancy on mother and infant. A. Themother. Brit. J. Haematol. 5: 31, 1959.

32. MUNCH-PETERSON, P. On the copper content of

mother’s milk before and after intravenous copper

administration. Acta Paediat. 39: 378, 1950.

33. BERFANSTAM, R. Studies on blood zinc: a clinical and

experimental investigation into the zinc content ofplasma and blood corpuscles with special reference

to infants. Acta. Paediat. 41: 105, 1952.34. MEYER, H. F. Current feeding practices in hospital

maternity nurseries. Clin. Pediat. 8: 69, 1969.

35. RIVERA, J. The frequency of use of various kinds of

milk during infancy in middle and lower income

families. Am. J. Public Health 61: 277, 1971.

36. KIRCHGESSNER, M., H. FRIESECKE AND G. KOCH.

Nutrition and the Composition of Milk. Philadel-phia: Lippincott, 1967.

37. UNDERWOOD, E. J. Trace Elements in Human andAnimal Nutrition (3rd ed). New York: Academic,

1971.

38. HUFF, B. B. (editor). Physicians’ Desk Reference(29th ed). New Jersey: Medical Economics Com-

pany, 1975.39. PRIEV, I. G. Effect of bottle-feeding on copper and

iron metabolism in infants during the first year oflife. Federation Proc. 24: T6l4, 1965. Translated

from Vapr. Med. Khim. 10: 352, 1964.

40. WIDDOWSON, E. M., J. DAUNCEY AND J. C. L.SHAW. Trace elements in foetal and early postnataldevelopment. Proc. Nutr. Soc. 33: 275, 1974.

41. WALRAVENS, P. A., AND K. M. HAMBIDGE. Sup-

plementation of an infant milk formula with zinc.

Am. J. Clin. Nutr. 28: 425, 1975 (abst.).


ajcn.nutrition.orgD

ownloaded from


am j clin nutr 1976 picciano 242 54.pdf hmc

Documents