The impact of fluoridated milks on the availability of trace elements in milk

Authors

  • Vida Zohoori,

    1. School of Dental Sciences, University of Newcastle, Framlington Place, Newcastle upon Tyne NE2 4BW, UK
    Current affiliation:
    1. School of Health and Social Care, University of Teesside, Middlesbrough, Teesside, UK.
    Search for more papers by this author
  • Christopher J Seal,

    1. Human Nutrition Research Centre, Newcastle University, Newcastle upon Tyne NE2 4HH, UK
    2. School of Agriculture, Food and Rural Development, Agriculture Building, Newcastle University, Newcastle upon Tyne NE1 7RU, UK
    Search for more papers by this author
  • Paula J Moynihan,

    1. School of Dental Sciences, University of Newcastle, Framlington Place, Newcastle upon Tyne NE2 4BW, UK
    2. Human Nutrition Research Centre, Newcastle University, Newcastle upon Tyne NE2 4HH, UK
    Search for more papers by this author
  • Ian N Steen,

    1. Institute of Health and Society, Newcastle University, Newcastle upon Tyne NE2 4AA, UK
    Search for more papers by this author
  • Anne Maguire

    Corresponding author
    1. School of Dental Sciences, University of Newcastle, Framlington Place, Newcastle upon Tyne NE2 4BW, UK
    • School of Dental Sciences, Newcastle University, Framlington Place, Newcastle upon Tyne NE2 4BW, UK.
    Search for more papers by this author

Abstract

BACKGROUND: Milk is a nutritious food and also used as a vehicle for fluoride (F) administration. However, the impact of added F on milk's nutritional profile is unknown. In vitro simulated gastrointestinal digestion with enzymatic steps was used to measure and compare the availability of trace elements (Fe, Zn, Cu, Cr, Mo and Se) in pasteurised skimmed (0.3% fat) and whole (4% fat) milk samples with four concentrations of F (0, 2.5, 3.75 and 5.0 ppm) as well as in non-F and F ultrahigh-temperature (UHT)-processed 4% fat milks. Post-centrifugation supernatant trace element concentrations were measured after each stage of digestion by inductively coupled plasma mass spectroscopy.

RESULTS: F showed a negative effect on Cu availability in cow's milk. Fat removal increased the availabilities of Cu, Zn, Cr and Se but decreased the Mo availability. There was a greater Cr availability in the UHT milk sample compared with pasteurised samples.

CONCLUSION: These initial data suggest that adding F to milk does not have a marked effect on its trace element profile, with the exception of reduced Cu availability. However, these findings would benefit from further studies both in vitro and in vivo. Copyright © 2009 Society of Chemical Industry

INTRODUCTION

The inverse relationship between fluoride (F) exposure and prevalence of dental caries is well established. Although systemic F is important for the developing dentition in young children, the main beneficial actions of F occur through topical action in the mouth.1

Milk, as a nutritious food,2 has been suggested as a vehicle for administration of topical and systemic F to individuals living in non-fluoridated, socially deprived communities. However, it is important to know the impact of added F on a milk's nutritional profile.

Milk is also an important source of trace elements and makes a significant contribution to the trace element content of diets in industrialised countries.3 Food-derived trace elements are important at all stages of the human life cycle; they are essential for optimal growth and development and play a role in the maintenance of optimal health. Apart from the inverse relationship between F intake and dental caries, several other claims for cariostatic as well as cariogenic effects of other trace elements have been reported, although these are all less potent than the positive effects of F on oral health. These other trace element effects include a negative correlation between molybdenum intake and dental caries and a positive correlation between copper intake and dental caries.4 However, limited information is available on the sources and contents of trace elements in foods, as they mostly occur in nanogram concentrations, and there are difficulties in measuring these elements when they occur at such low concentrations in foods.

Zinc (Zn) is an essential cofactor for more than 300 metalloenzymes5 and an important component of DNA-binding proteins. Other important trace elements in milk include: iron (Fe), which is essential for synthesis of haemoglobin and for cytochrome activity; copper (Cu), which is an essential cofactor for oxidoreductase enzymes, including Cu/Zn superoxide dismutase (SOD);6 selenium (Se), another antioxidant with activity as a free radical scavenger; molybdenum (Mo), a constituent of an essential enzyme, xanthine oxidase; and chromium (Cr), which is associated with glucose metabolism.7

The best dietary sources of trace elements are: oysters, shellfish and liver for Zn; beef, liver, shellfish, kidney and nuts for Cu; brewer's yeast, oysters, liver and corn oils for Cr; wholegrains, onions and meats for Se; liver, meat and egg yolks for Fe; and legumes, cereal grains and dark green leafy vegetables for Mo.7 However, in certain individuals, e.g. vegetarians, milk may be a primary dietary source of some trace elements such as Zn.

For a child aged 4–6 years, drinking 200 mL of whole (4% fat) cow's milk provides 11% of the Reference Nutrient Intake (RNI) for Zn, 2% of the RNI for Fe, 3% of the RNI for Cu and 10% of the RNI for Se.8 No Dietary Reference Values (DRVs) have been set for Cr or Mo.8 However, safe and adequate levels of intake between 0.1 and 1 µg kg−1 body weight day−1 for Cr and between 0.5 and 1.5 µg kg−1 body weight day−1 for Mo have been suggested for children up to 18 years of age.8 Therefore 200 mL of whole cow's milk can provide up to 22 and 112% of daily Cr and Mo requirements respectively.

Addition of F to milk may be a suitable public health measure for optimising F intake in children; however, it is important that such an addition does not impact the wider nutritional profile of milk. Currently, the literature provides no information on the impact of adding F to milk on the availability of trace elements in milk. Nor does it provide any information on whether the availability of trace elements is affected by the dose of F in milk through interaction between trace elements and F either in or during uptake from the gastrointestinal system. More knowledge of the biological plausibility for any interaction and insight into the effects of F in milks on trace element availability would be useful and would help to inform milk fluoridation policies.

In the UK the prevalence of obesity among 2–10-year-olds has increased from 9.9% in 1995 to 13.7% in 2003.9 Many cohort studies have shown a relationship between obesity in childhood and chronic disease in adulthood. Campaigns to prevent and manage obesity in children and young people have recommended limited consumption of high-fat foods and gradual introduction to the diet of reduced-fat dairy products, such as semi-skimmed milk in accordance with ‘The Eatwell Plate’.10

It is pertinent that any oral health preventive programme conforms to general public health goals such as obesity prevention. With this in mind it is also important to know whether any effect of F on trace element availability differs between milks of different fat content. Most of the previous research on the trace element content of milks has been limited to whole milks, although a study of Se bioavailability from milk showed its availability from skimmed milk to be significantly higher than that from whole milk.11 In view of the promotion of lower-fat milks in line with general health policy, it would seem important to investigate whether this difference between milks, seen for Se availability, extends to other trace elements in the presence and absence of F, and if the fat content of the milk influences the way in which F may affect trace element availability.

Therefore the main aims of this study were to investigate the effects of (i) added F and (ii) F dose in milks with varying fat contents on the availability of Fe, Zn, Cu, Cr, Mo and Se in milk. A secondary aim was to investigate the effect of heat treatment on the availability of these trace elements in whole milk.

EXPERIMENTAL

Reagents and materials

All glassware was washed and rinsed with distilled deionised water (DDi-H2O), then soaked overnight in 1.0 mol L−1 HCl and rinsed again with DDi-H2O. DDi-H2O was used throughout for preparation of reagents. Reagents and enzymes were prepared as described by Miller et al.12

Samples

The following ten different types of milk were prepared from a single batch of raw cow's milk:

  • non-fluoridated ultrahigh-temperature (UHT)-processed whole (4% fat) milk (NFUW);

  • non-fluoridated pasteurised skimmed (0.3% fat) milk (NFPS);

  • non-fluoridated pasteurised whole milk (NFPW);

  • fluoridated UHT-processed whole (4% fat) milk containing 2.5 ppm F as sodium fluoride (2.5FUW);

  • fluoridated pasteurised skimmed milk containing 2.5 ppm F as sodium fluoride (2.5FPS);

  • fluoridated pasteurised skimmed milk containing 3.75 ppm F as sodium fluoride (3.75FPS);

  • fluoridated pasteurised skimmed milk containing 5.0 ppm F as sodium fluoride (5.0FPS);

  • fluoridated pasteurised whole milk containing 2.5 ppm F as sodium fluoride (2.5FPW);

  • fluoridated pasteurised whole milk containing 3.75 ppm F as sodium fluoride (3.75FPW);

  • fluoridated pasteurised whole milk containing 5.0 ppm F as sodium fluoride (5.0FPW).

In vitro enzymatic digestion process

Samples were digested by a simulated in vitro gastrointestinal digestion procedure, which was a modification of earlier methods.12–14 Three replicates of each milk sample were analysed through a series of stepwise experiments designed to mimic digestion through the stomach and small intestine. The process comprised:

  • stomach digestion (experiment A);

  • stomach and duodenal digestion (experiment B);

  • stomach, duodenal and jejunal digestion (experiment C).

At the start of each simulated digestive procedure an aliquot of each milk (or blank) was taken before digestion (T0). Then 0.2 ( ± 0.001) L of each milk sample was placed in a shaking water bath at 37 °C, set at 170 rotations min−1, and 1 mL of 50 U α-amylase (Sigma A1031, Poole, Dorset, UK) was added. The pH of the sample was then adjusted to 2 with 6 mol L−1 HCl, before adding 3 mL of pepsin solution (32 g of pepsin (Sigma P7000) in 100 mL of 0.1 mol L−1 HCl). The mixture was then incubated for 2 h.

Simulated stomach digestion (experiment A)

In the simulated stomach digestion (experiment A) an aliquot was taken at 30 min (T1), 60 min (T2), 90 min (T3) and 120 min (T4).

Simulated stomach and duodenal digestion (experiment B)

The first step (i.e. stomach digestion) of simulated stomach and duodenal digestion (experiment B) was similar to simulated stomach digestion (experiment A), but, following the aliquot taken before digestion (T0), the next aliquot was taken at 120 min (T1). For the second step (i.e. duodenal digestion) the sample was then adjusted to pH 7.5 with NaHCO3, before adding 5 mL of bovine bile and pancreatin dissolved in sodium bicarbonate solution (3.4 g of pancreatin (Sigma P1750) and 21.2 g of bile (Sigma B8631) in 200 mL of 0.1 mol L−1 NaHCO3). An aliquot was taken at 150 min (T2), 180 min (T3), 210 min (T4) and 240 min (T5).

Simulated stomach, duodenal and jejunal digestion (experiment C)

The first 240 min of simulated stomach, duodenal and jejunal digestion (experiment C) was similar to simulated stomach and duodenal digestion (experiment B); aliquots were taken at 120 min (T1) and 240 min (T2). Then 1 mL of mucin solution (8 g of mucin (Sigma M2378) in 200 mL of 0.1 mol L−1 NaHCO3) was added and an aliquot was taken at 255 min (T3), 270 min (T4), 285 min (T5) and 300 min (T6).

All aliquots were centrifuged at 6900 × g for 5 min, then the supernatants were collected in 5 mL bijous and frozen at − 18 °C prior to analysis for F, Fe, Zn, Cu, Cr, Mo and Se.

Analysis of samples

The F concentration in digested samples was measured by a direct method using a fluoride ion-selective electrode (Orion, Beverly, MA, USA).15, 16 The concentrations of Fe, Zn, Cu, Cr, Mo and Se in digested samples were measured by inductively coupled plasma mass spectroscopy (ICP-MS).17 In addition, 25% of the 594 samples generated were reanalysed for all trace elements and F for quality assurance purposes. Test/retest reliability was assessed to determine the level of agreement between the concentration values recorded in the resampling exercise and the corresponding values recorded during the initial assays.

Calculation of trace element availability

The availability of each trace element in the samples was calculated from the concentration of the trace element in the supernatant in the digestion tube following centrifugation, after each stage of digestion.

Statistical analysis

The test/retest reliability of the analytical method used to determine trace element concentrations was assessed using the intra-class correlation coefficient based on a two-way random effects analysis of variance model. This method was employed to look at reliability when a single aliquot of each milk sample was analysed for trace element contents twice and when the trace element contents of different samples of milk subjected to the same experimental treatment were compared. A ρ value of ≥ 0.80 was considered as acceptable reliability.

Each experiment was statistically analysed separately. The concentration of trace elements was analysed using a multilevel model with occasions nested within ‘milk sample’ (during each experiment the concentration of the trace element in each milk sample was measured on repeated occasions). In general, the following analytical strategy was adopted. First, a reference model was fitted with differences between samples and differences between occasions included as random effects. The second step was to investigate the significance of the main effects of fat content (4 vs 0.3%), milk type (UHT vs pasteurised) and F concentration (as a continuous variable). The third step was to investigate changes in concentration of F and trace elements over time. The final step was to investigate whether the main effects varied over time. This procedure was implemented using the ‘xtreg’ procedure in Stata 9.18

RESULTS

The results of the test/retest reliability of the ICP-MS method used for trace element measurement were good: a range from ρ = 0.83 for Se to ρ = 0.99 for Cu. However, the reliability of the whole procedure from production of milk samples to analysing the digested samples for trace element contents was satisfactory only for Cu (ρ = 0.78), Cr (ρ = 0.81) and Mo (ρ = 0.79), while the reliability of Fe (ρ = 0.69) and Se (ρ = 0.54) measurement was not good and the reliability of Zn (ρ = 0.28) measurement was poor. Plots of the raw data revealed the presence of extreme values most likely to have arisen owing to contamination of the samples. Based on these plots, it was decided to discard values that exceeded the following thresholds: ≥ 30 × 10−5 g kg−1 Fe and ≥ 0.1 × 10−5 g kg−1 Cr at baseline and ≥ 50 × 10−5 g kg−1 Fe, ≥ 400 × 10−5 g kg−1 Zn, ≥ 20 × 10−5 g kg−1 Cu, ≥ 0.5 × 10−5 g kg−1 Cr, ≥ 4.5 × 10−5 g kg−1 Mo and ≥ 3.5 × 10−5 g kg−1 Se at all other time points. The results shown are based on the values for the remaining samples.

Concentration of trace elements in milk samples during different stages of digestion

The means and ranges (10−5 g kg−1) of trace element contents of milk samples before digestion are presented in Table 1 (n = 9) and show a difference in concentration of trace elements between milks and between F doses of the milks.

Table 1. Concentrations of trace elements in milk samples before digestion
Milk typeF (10−5 g kg−1)Mean (range) of trace element concentration (10−5 g kg−1)
FeZnCuCrMoSe
Pasteurised, whole, no F3.00 (0.10)16.05 (5.99)318.57 (11.91)14.26 (3.17)0.08 (0)3.56 (0.09)1.44 (0.12)
Number of samples9694599
Pasteurised, skimmed, no F2.80 (0.10)13.46 (3.54)314.44 (24.85)16.04 (2.18)0.08 (0)2.89 (0.15)1.58 (0.66)
Number of samples9898698
UHT, whole, no F2.50 (0.10)17.76 (3.51)333.49 (32.68)10.91 (1.24)0.18 (0)3.70 (0.21)1.50 (0.23)
Number of samples9799399
Pasteurised, whole, 2.5 ppm F246.2 (3.90)16.32 (4.84)339.85 (21.37)6.56 (1.53)0.08 (0)3.77 (0.22)1.64 (0.35)
Number of samples9989789
Pasteurised, skimmed, 2.5 ppm F243.6 (0.50)12.91 (4.55)291.17 (36.79)11.55 (2.66)0.08 (0)2.93 (0.15)1.15 (0.17)
Number of samples9999999
UHT, whole, 2.5 ppm F242.0 (0.8)17.99 (4.49)334.00 (32.16)8.48 (1.31)0.08 (0)3.65 (0.28)1.40 (0.23)
Number of samples9798299
Pasteurised, whole, 3.75 ppm F375.7 (0.2)13.85 (4.42)312.03 (24.51)3.92 (1.34)0.08 (0)3.56 (0.22)1.57 (0.20)
Number of samples9999999
Pasteurised, skimmed, 3.75 ppm F374.9 (0.1)12.65 (2.40)322.53 (23.86)9.58 (0.61)0.08 (0)2.96 (0.20)2.25 (1.01)
Number of samples9999798
Pasteurised, whole, 5.0 ppm F491.5 (1.6)19.10 (4.02)332.00 (20.02)5.73 (1.87)0.08 (0)3.52 (0.25)1.49 (0.39)
Number of samples9789889
Pasteurised, skimmed, 5.0 ppm F486.1 (0.4)13.42 (3.19)328.25 (22.86)8.18 (0.70)0.08 (0)2.86 (0.13)1.30 (0.39)
Number of samples9999989
Distilled deionised water (blank)1.6 (0.0)3.97 (9.66)1.50 (0)0.38 (0)0.08 (0)0.08 (0)0.08 (0)
Number of samples9999999

Tables 2–4 present the means and ranges (10−5 g kg−1) of trace element contents of milk samples after stomach (experiment A, n = 9), duodenal (experiment B, n = 6) and jejunal (experiment C, n = 3) digestion respectively.

Table 2. Concentrations of trace elements in milk samples after completion of stomach digestion
Milk typeF (10−5 g kg−1)Mean (range) of trace element concentration (10−5 g kg−1)
FeZnCuCrMoSe
Pasteurised, whole, no F2.7 (0.2)31.16 (10.66)320.55 (10.09)13.24 (1.14)0.20 (0.08)3.71 (0.11)1.86 (0.28)
Number of samples9999999
Pasteurised, skimmed, no F2.7 (0.0)15.06 (10.05)343.57 (12.41)17.66 (1.12)0.22 (0.08)2.65 (0.13)2.24 (0.79)
Number of samples9988787
UHT, whole, no F2.5 (0.2)35.11 (9.72)336.40 (34.59)11.12 (2.35)0.33 (0.10)3.68 (0.26)1.76 (0.37)
Number of samples9799589
Pasteurised, whole, 2.5 ppm F205.6 (2.0)32.25 (6.40)314.43 (16.59)6.22 (1.89)0.14 (0.04)3.54 (0.23)1.72 (0.24)
Number of samples9989999
Pasteurised, skimmed, 2.5 ppm F203.7 (3.8)13.17 (1.93)366.07 (18.08)14.82 (0.84)0.26 (0.03)2.77 (0.09)2.36 (0.46)
Number of samples9999999
UHT, whole, 2.5 ppm F198.7 (7.5)34.83 (8.60)323.74 (31.41)8.89 (1.04)0.25 (0.09)3.68 (0.44)1.78 (0.35)
Number of samples9799499
Pasteurised, whole, 3.75 ppm F313.9 (3.6)35.40 (5.20)309.93 (10.66)5.06 (4.93)0.21 (0.05)3.61 (0.10)1.81 (0.21)
Number of samples9999999
Pasteurised, skimmed, 3.75 ppm F304.1 (2.4)10.41 (3.04)339.10 (6.77)9.67 (0.74)0.20 (0.07)2.68 (0.01)1.79 (0.06)
Number of samples9988999
Pasteurised, whole, 5.0 ppm F395.3 (2.9)29.65 (15.36)325.57 (14.95)5.24 (2.12)0.19 (0.05)3.42 (0.29)1.41 (0.03)
Number of samples9999999
Pasteurised, skimmed, 5.0 ppm F410.1 (2.8)13.32 (8.66)363.81 (18.77)8.49 (0.92)0.21 (0.07)2.87 (0.32)1.86 (0.73)
Number of samples9999999
Distilled deionised water (blank)1.6 (0.0)17.58 (4.80)18.13 (4.06)0.96 (0.53)0.23 (0.06)0.28 (0.04)0.21 (0.11)
Number of samples9999999
Table 3. Concentrations of trace elements in milk samples after completion of stomach and duodenal digestion
Milk typeF (10−5 g kg−1)Mean (range) of trace element concentration (10−5 g kg−1)
FeZnCuCrMoSe
Pasteurised, whole, no F4.0 (0.0)30.07 (3.04)309.53 (16.27)13.71 (3.54)0.24 (0.03)3.78 (0.09)1.82 (0.25)
Number of samples6664666
Pasteurised, skimmed, no F4.6 (0.2)31.44 (1.06)328.00 (10.66)16.54 (0.96)0.32 (0.01)3.16 (0.11)3.06 (0.24)
Number of samples6666666
UHT, whole, no F3.5 (0.1)36.38 (7.15)315.15 (28.08)10.38 (2.39)0.42 (0.05)3.76 (0.25)1.89 (0.55)
Number of samples6566266
Pasteurised, whole, 2.5 ppm F242.8 (3.3)29.71 (2.82)308.15 (9.74)6.93 (0.95)0.23 (0.09)3.46 (0.11)1.57 (0.24)
Number of samples6666666
Pasteurised, skimmed, 2.5 ppm F227.6 (63.0)33.67 (2.78)340.17 (16.75)13.63 (0.75)0.35 (0.02)3.19 (0.09)2.84 (0.03)
Number of samples6666666
UHT, whole, 2.5 ppm F234.2 (11.8)39.37 (4.25)320.99 (23.51)9.30 (1.93)3.80 (0.19)1.68 (0.21)
Number of samples6566066
Pasteurised, whole, 3.75 ppm F365.4 (39.0)34.32 (7.89)309.12 (8.06)4.22 (1.76)0.23 (0.03)3.57 (0.08)1.65 (0.37)
Number of samples6566666
Pasteurised, skimmed, 3.75 ppm F333.8 (60.0)32.96 (4.31)333.67 (22.62)9.73 (0.76)0.33 (0.04)3.18 (0.09)3.14 (0.20)
Number of samples6666666
Pasteurised, whole, 5.0 ppm F465.3 (2.7)37.62 (6.63)328.51 (6.43)6.15 (0.22)0.31 (0.07)3.52 (0.06)1.81 (0.03)
Number of samples6566666
Pasteurised, skimmed, 5.0 ppm F446.7 (8.2)30.61 (1.30)332.75 (10.32)6.33 (1.63)0.29 (0.08)3.25 (0.08)2.45 (0.05)
Number of samples6666666
Distilled deionised water (blank)3.4 (0.2)18.03 (1.13)17.73 (3.55)0.89 (0.22)0.29 (0.02)0.46 (0.02)0.45 (0.06)
Number of samples6666666
Table 4. Concentrations of trace elements in milk samples after completion of stomach, duodenal and jejunal digestion
Milk typeF (10−5 g kg−1)Mean (range) of trace element concentration (10−5 g kg−1)
FeZnCuCrMoSe
Pasteurised, whole, no F3.9 (0.0)29.78 (1.23)312.36 (3.07)11.12 (0.46)0.25 (0.05)3.84 (0.05)1.98 (0.36)
Number of samples3332333
Pasteurised, skimmed, no F4.2 (0.2)34.27 (0.94)327.67 (14.64)16.67 (0.96)0.34 (0.01)3.16 (0.12)3.01 (0.21)
Number of samples3333333
UHT, whole, no F3.2 (0.1)41.33 (0.85)320.59 (2.15)12.10 (3.28)0.343.76 (0.12)2.32 (0.29)
Number of samples3333133
Pasteurised, whole, 2.5 ppm F241.2 (5.2)33.08 (1.95)313.24 (11.17)7.41 (1.43)0.25 (0.06)3.55 (0.09)1.76 (0.27)
Number of samples3333333
Pasteurised, skimmed, 2.5 ppm F232.8 (0.8)35.33 (0.73)342.83 (1.44)14.17 (0.40)0.36 (0.02)3.34 (0.06)3.03 (0.15)
Number of samples3333332
UHT, whole, 2.5 ppm F249.3 (2.1)35.65 (3.90)291.85 (3.23)7.63 (0.20)3.57 (0.08)1.67 (0.12)
Number of samples3233033
Pasteurised, whole, 3.75 ppm F366.8 (5.4)30.35 (0.61)302.69 (7.07)2.96 (0.37)0.25 (0.04)3.69 (0.03)1.99 (0.031)
Number of samples3333333
Pasteurised, skimmed, 3.75 ppm F318.8 (1.9)35.23 (0.11)343.67 (0.58)10.14 (0.28)0.36 (0.02)3.24 (0.06)3.40 (0.06)
Number of samples3333333
Pasteurised, whole, 5.0 ppm F456.0 (1.7)34.01 (1.35)317.21 (6.38)6.11 (0.32)0.36 (0.12)3.61 (0.10)1.81 (0.24)
Number of samples3233333
Pasteurised, skimmed, 5.0 ppm F460.7 (1.1)31.30 (0.56)348.33 (2.31)5.09 (0.16)0.32 (0)3.46 (0.04)3.33 (0.22)
Number of samples3333333
Distilled deionised water (blank)3.4 (0.1)23.70 (1.48)27.85 (0.52)1.01 (0.12)0.31 (0.01)0.51 (0.02)0.56 (0.01)
Number of samples3333333

Effect of fluoride on availability of trace elements in milk

The statistical analysis showed conflicting estimates of the effect of F on the availability of Fe and Se in milk between the three stimulated digestive processes. In the first two digestive steps there was a positive but non-significant association between Se and F concentrations, while in the simulated stomach, duodenal and jejunal digestion there was some evidence of a negative association (P < 0.05) between these variables. When the data from the three simulated digestions were pooled across all 12 time points, the association between Se and F was positive (P = 0.002; regression coefficient 0.94; 95% confidence interval (CI) 0.33, 1.54). There was a negative association between the concentration of Fe and the concentration of F in the simulated stomach digestion and a positive association in the simulated stomach and duodenal digestion. However, when the data from the three experiments were pooled, overall the association between the concentration of Fe and the concentration of F was not significant (regression coefficient − 2.34; 95% CI − 7.82, 3.14). The results of the present study showed no evidence of an association between the concentration of F and the availability of Zn, Cr and Mo. However, a significant negative association between the concentration of F and the availability of Cu in milks was found (regression coefficient − 17.1; 95% CI − 19.3, − 15.0). The mean ( ± standard deviation) concentration of Cu in the supernatant from pasteurised 4% fat milk with 5 ppm F was 5.73 ( ± 1.87) and 6.11 ( ± 0.32) × 10−5 g kg−1 before digestion and after stomach, duodenal and jejunal digestion respectively, while for the equivalent non-fluoridated milk the mean concentrations were 14.26 ( ± 3.17) and 11.12 ( ± 0.46) × 10−5 g kg−1 respectively. The concentration of Cu averaged across milk samples against time during digestion for each concentration level of F in milk is plotted in Fig. 1.

Figure 1.

Cu concentration in supernatant after centrifugation of all pooled (whole, skimmed and UHT) milk products during digestion for each concentration of F over time.

Effect of fat content on availability of trace elements

The results revealed a difference between Fe availability from whole milk and skimmed milk which was observed predominantly over the time frame of simulated stomach digestion. The difference was much less after stomach digestion and had more or less disappeared after duodenal digestion. In contrast, there was no statistically significant difference between Se availability from whole and skimmed milks during stomach digestion (experiment A), while the availability of Se in skimmed milk increased significantly (P < 0.005) more over the course of duodenal (experiment B) and jejunal (experiment C) digestion than it did in whole milk (Fig. 2). The availabilities of Zn, Cu and Cr were significantly (P < 0.005) lower in whole milk than in skimmed milk. However, the availability of Mo was significantly (P < 0.005) higher in whole milk than in skimmed milk, and this finding was consistent across the three simulated digestive processes.

Figure 2.

Se concentration in supernatant after centrifugation of all pooled whole and skimmed milk products during digestion over time.

Effect of heat treatment on availability of trace elements

There was evidence that the availability of Fe and Zn was greater in UHT milk than in pasteurised milk; however, this finding was not consistent across the three simulated digestions, being most evident in stomach (experiment A) and jejunal (experiment C) digestion. Heat treatment did not appear to have any effect on the availabilities of Cu, Mo and Se, while the availability of Cr was higher in UHT milk than in pasteurised milk, and this difference became more obvious during simulated duodenal (experiment B) and jejunal (experiment C) digestion.

DISCUSSION

This is the first study to investigate the effect of (i) added F and F dose and (ii) varying fat contents of milk on the availability of trace elements in milk.

An in vitro method was selected as a useful tool for predicting the availability of trace elements by mimicking the digestion process and measuring the concentrations of elements which are available for absorption. The technique employed in this study has been used in studies of iron12, 13 and antioxidant14 availability and consisted of simulation of the different stages of digestion, with a change in pH from neutral to acid and to neutral again to simulate the intraluminal digestive conditions through stomach, duodenal and jejunal digestion. In the simulated gastric procedure, pepsin/HCl would result in chemical hydrolysis of proteins together with proteinase activity to release lower-molecular-weight peptide fractions. In the simulated intestinal procedure, pancreatin/NaHCO3 would return the partially digested material to neutral pH and at the same time cleave carbohydrates into smaller oligosaccharides and continue protein digestion. In addition, bile salts, through their emulsifying properties, would facilitate the digestion of fats. By centrifuging the aliquots taken at different time points, the insoluble fraction was removed from the system; the supernatants remaining contained the soluble forms of trace elements, which were assumed to be the fraction available for absorption.

Reliability of methods

The results of the test and retest of 25% of samples by the ICP-MS method indicated that the method fulfilled the requirement of adequate repeatability and reproducibility for measurement of all trace elements apart from Zn in the study. The unreliability of the values obtained by the ICP-MS method for measuring the concentration of Zn has also been observed by other trace element laboratories (Trace Element Centre, University of Surrey, personal communication). The reasons for this are not clear but are probably related to the rubber closure used with the vials of freeze-dried material. Rubber is a well-recognised source of Zn. However, this should not create any difficulty when attempting to compare these data with the work of others.

The reliability test for the total procedure showed some uncertainty for Fe and Se and poor reliability for Zn. The total procedure involved the preparation of milk samples, the transport of produced milk samples from the production site to the laboratory, the preparation of reagents and samples for the experiment, the conduct of the experiment itself, the transfer of aliquots into containers for storage and the measurements using the ICP-MS method. Therefore it is difficult to judge which part of the procedure was the source of the observed variability.

In general, a common problem in the study of trace elements is the risk of contamination during sampling and sample preparation. However, contamination in our laboratory was minimised by rinsing the containers and washable equipment in distilled deionised water, soaking them in weak acid and then rinsing with distilled deionised water again. All chemicals and reagents were of analytical grade. However, less control was possible with regard to the consistency in the trace element content of the disposable equipment used, such as pipette tips, bijous, etc. Based on plots of the raw data, concentration values that exceeded specified thresholds were discarded to try to minimise the impact of any contamination on the results of the analysis.

Availability of trace elements

Copper

Cu is absorbed from the stomach to some extent, but the major site of absorption is the duodenum.7 The bioavailability of Cu is reduced by high levels of Zn and the presence of ascorbic acid and certain carbohydrates such as fructose.7 Several studies have also suggested that Fe can reduce the bioavailability of Cu, but only when the concentration of Cu is low.19 The influence of F in reducing the availability of Cu might be similar to the negative effect of Fe on Cu. However, the mechanism(s) behind the interaction of F and Cu availability, which was found in the present study, needs to be investigated further.

The present study showed a higher concentration of Cu in skimmed milk compared with whole milk at baseline, which confirms earlier findings that only a small (2%) proportion of Cu in milk is associated with the milk fat fraction.3 In cow's milk, Cu is mainly bound to proteins (8% to whey proteins and 44% to casein), and 47% of Cu is in low-molecular-weight ligands.3

Iron

The concentration of Fe was lower in skimmed milk compared with whole milk at baseline as well as during the simulated stomach digestion. These results are in agreement with the fact that, in cow's milk, almost 14% of Fe occurs in milk fat, while about 24 and 29% of the Fe content is bound to casein and whey protein respectively and 32% of Fe is associated with a low-molecular-weight fraction.3 Fe may become soluble or ionised as ferrous and ferric salts when it comes into contact with HCl in the stomach. However, the main site of Fe absorption is the small intestine rather than the stomach. When Fe leaves the stomach and enters the duodenum, the increased pH in the duodenum results in Fe hydrolysis of the hydrated ferric and ferrous salts and the formation of high-molecular-weight polymers which eventually precipitate.12, 13 The results suggest that F has no effect on the solubility of Fe as measured by the concentration of Fe in the supernatant fraction, and so it is likely that F would not affect polymerisation and precipitation of Fe in the small intestine. However, since the reliability of data on Fe availability was moderate, these findings require further confirmation.

Zinc

The lack of consistency in the results for Zn availability across the whole experiment could be due to the low reliability associated with the Zn measurements. In the present study, no statistically significant differences in the concentration of Zn in whole compared with skimmed milk at baseline were obtained, which is in agreement with the fact that only 1% of the Zn is in the lipid fraction of cow's milk, with the remainder in the skimmed milk fraction.3 However, the availability of Zn in whole milk was lower than that in skimmed milk after starting the digestion process. The mechanisms of Zn absorption are not well understood; however, it seems that a protein-rich diet could promote absorption of Zn by forming Zn–amino acid chelates that present Zn in a more absorbable form. The results of the present study showed that F added to milk did not change the availability of Zn, and this might be due to the fact that over 95% of the Zn in the skimmed milk fraction is associated with the casein micelles,3 rendering it inaccessible for interaction with F. However, since the reliability of the Zn measurements was not strong and the differences between samples was very small, these results should be interpreted with caution.

Chromium

The chemical form of Cr in milk is unknown.3 The slightly higher concentration of Cr in skimmed milk products observed in the present study is in agreement with the only other study carried out on different milk slurries, which found a slightly higher concentration of Cr in skimmed milk (5.7 × 10−5 g kg−1) compared with whole milk (4.3 × 10−5 g kg−1).20 These results suggest that only a small fraction of the Cr in milk may be found in the fat fraction of milk. However, further studies are needed to confirm this.

There are some interesting but conflicting data on the Cr content of milks in the literature: (0.02–0.36) × 10−5 g kg−1 (mean 0.2 × 10−5 g kg−1),3 (0–17.7) × 10−5 g kg−1 (mean 5.7 × 10−5 g kg−1 for skimmed milk and 4.3 × 10−5 g kg−1 for whole milk)20 and (0–4.21) × 10−5 g kg−1 (mean 0.55 × 10−5 g kg−1).21 Our data are at the lower end of these reported Cr concentrations. These differences between studies could be due to the differences in the analytical methods used for the determination of Cr as well as the type of milk being analysed. However, the risk of measuring an elevated concentration as a result of contamination should not be ignored. For our interest we have analysed the trace element content of the source milk. The results suggested that the concentration of Cr in the products before centrifugation was higher than the baseline measurements after centrifugation. However, this was not found for other trace elements.

The present study showed a marked increase in the availability of Cr during/after intestinal digestion, an observation which has also been reported in a study by Cabrera et al.22 This group also investigated the absorbable fractions of Cr in dairy products using an in vitro method and reported that the solubility (availability) of Cr doubled from 1.0 ± 0.55% after gastric digestion to 2.0 ± 1.0% after intestinal digestion.

Molybdenum

The present study also showed that F had no impact on the availability of Mo; however, fat had a positive influence on the availability of Mo, suggesting the importance of Mo within the fat fraction.

The lower concentration of Mo in skimmed milk compared with whole milk for all milk products at baseline, found in the present study, shows again that a fraction of Mo is associated with the fat fraction of milk, although there is no information in the literature in this regard.

Selenium

While no clear relationship between the availability of Se and the concentration of F was found in the present study, the different rate of increase in Se availability for skimmed milk and whole milk was obvious. These results show that fat has more effect on the availability of Se in milk than the F content. The present study showed that the availability of Se was higher in skimmed milk than in whole milk. This result confirms the finding of a study by Shen et al.11 that removal of the fat fraction from milk significantly increased the Se availability. The negative influence of animal fat on Se availability is unclear and needs more extensive study. Since Se mainly exists in protein-bound forms in milk and almost 80% of Se in cow's milk is associated with casein,11, 23, 24 it has been suggested that protein digestibility is a main determinant for Se bioavailability from milk.11 Therefore the higher availability of Se from skimmed milk might be due to better protein digestibility of skimmed milk, although this would require further investigation.

Impact of negative effect of F on availability of Cu in cow's milk

In various parts of the world, including the UK, children are offered almost 200 mL of milk at schools, usually on schooldays only. Since the mean Cu concentration in cow's milk is low, around 9 × 10−5 g L−1,3 consumption of 0.2 L of milk on 200 days a year would provide only 0.02 mg Cu day−1 for a child aged 4–6 years, which is equivalent to 1.6% of the RNI for Cu based on consumption. Based on the results of the present study, adding 5 ppm F to milk would reduce the availability of Cu by almost half. Therefore 0.2 L of fluoridated milk containing 5 ppm F would provide 0.8% of the RNI for Cu rather than 1.6% of the RNI if the child consumed non-F milk on schooldays. In addition, since, under the current milk fluoridation schemes, F milk is not consumed with any types of food, and milk is not an important source of Cu, the negative impact of F on the availability of Cu in milk is not an important issue.

CONCLUSIONS

The present study investigated the effect of F on the availability of trace elements in milk as a single food item and found no negative effect of F on the availability of trace elements, with the exception of Cu. The present study also showed that the removal of fat from milk increased the availability of Cu, Zn, Cr and Se, although the removal of fat decreased the availability of Mo. There was a greater availability of Cr in the UHT milk sample compared with pasteurised samples. However, owing to the lack of consistency in results across the whole study for some trace elements, these findings would benefit from further investigation both in vitro and in vivo.

Acknowledgements

This work was funded by The Borrow Foundation. The views expressed are those of the authors and not necessarily of The Borrow Foundation. The authors declare that they have no conflicts of interests. AM was the principal investigator, supervised the study, chaired the steering group and constructed the reports and manuscripts. VZ conducted the experiment, reviewed the literature and constructed the reports and manuscripts. CJS and PJM took an active role in constructing the manuscript, commenting on all drafts of the manuscript. INS advised on statistical matters.

Ancillary