Reference values for selected trace elements in serum of term newborns from the urban area of Rome

Clinica Chimica Acta 292 (2000) 163–173 www.elsevier.com / locate / clinchim

Reference values for selected trace elements in serum of term newborns from the urban area of Rome Alessandro Alimonti a , Francesco Petrucci a , Francesco Laurenti b , b a, Paola Papoff , Sergio Caroli * a

Applied Toxicology Laboratory, Istituto Superiore di Sanita` , Viale Regina Elena 299, 00161 Rome, Italy Pediatric Department, University ‘‘ La Sapienza’’ of Rome, Viale Regina Elena 324, 00161 Rome, Italy

b

Received 5 July 1999; received in revised form 13 October 1999; accepted 24 October 1999

Abstract Reference values for Al, Cd, Co, Cu, Li, Mn, Mo, Ni, Rb, Se and Zn, and indicative intervals for Sb are proposed in serum from cord blood of 143 term newborns of the urban area of Rome. On the basis of the eligibility criteria adopted, only babies with gestational age . 37 weeks and body weight at the delivery . 2500 g, i.e., ‘‘normal’’ term infants, were included in this study. With the exception of Cd, Li, Ni and Sb, experimental data for each of the other analytes were found to approach a normal distribution. The estimated references values (in ng / ml) were the following: Al, 1.12–6.79; Cd, 0.10–0.52; Co, 0.20–0.43; Cu, 140–691; Li, 0.31–2.23; Mn, 0.79–3.26; Mo, 0.36–1.56; Ni, 0.20–3.15; Rb, 196–1302; Sb, 0.10–1.48 (indicative range); Se, 20.2–69.7; and Zn, 318–1405. For several elements, the information available in the relevant literature does not allow adequate comparisons to be performed. This was actually possible only for Cu, Se and Zn. The correlations between the weights at birth (BW), gestational ages (GA) and elemental concentrations were elucidated. As expected, significant positive correlations were found for Cu and Se with GA and BW, respectively. Strong mutual associations were observed for several other elements, but their interpretation is still debatable.  2000 Elsevier Science B.V. All rights reserved. Keywords: Trace elements reference values; Serum analysis; Newborns; Inductively coupled plasma mass spectrometry

*Corresponding author. Tel.: 1 39-6-4990-2052; fax: 1 39-6-4990-2366. E-mail address: [email protected] (S. Caroli) 0009-8981 / 00 / $ – see front matter  2000 Elsevier Science B.V. All rights reserved. PII: S0009-8981( 99 )00217-X

164

A. Alimonti et al. / Clinica Chimica Acta 292 (2000) 163 – 173

1. Introduction Reference values for trace elements in human clinical specimens facilitate the interpretation of data deriving from environmental and occupational monitoring as well as clinical practice [1]. In fact, it is expected that they are a reflection of natural concentrations influenced by age, sex, habits and living and working environments. In the assessment of the relationship between trace elements and health status, knowledge of the reference intervals of the former in human body fluids and tissues becomes a key parameter for both environmental contamination control and therapeutic treatments. By definition, reference values reflect the findings in a selected group of individuals [2–5]. Unfortunately, inhomogeneity of the reference population often makes the obtained results of poor scientific quality, useless for regulatory decisions and inadequate for studying trace element-related diseases. Furthermore, reference values obtained from a defined population sample may not coincide with those of another population subgroup even when all sampling criteria are homogeneous. Age is one of the parameters that can mainly affect the body burden of trace elements [6]. Children, as an example, metabolize trace elements at a different rate than adults. Reference values for adult populations generally are of no use for studies on youngsters. Moreover, due the difficulty in obtaining data for young populations, reliable information on trace element concentrations in infants and newborns is scarce and fragmentary. On the other hand, the availability of accurate trace element data in body fluids such as serum may represent an important indicator of newborns’ health status at the delivery, also in relation to a previous significant exposure of the mothers during the pregnancy. This can also be of some assistance in understanding the role of trace elements in certain neonatal disorders in which either accumulation or depletion of some trace elements may occur. In the last decade, the biochemical functions of trace elements in infants are more and more being considered as crucial. Most of these functions appears to be related to enzymatic activity, either as cofactors or components of the prosthetic group. Stages of the life of rapid growth and development, such as infancy, are associated with remarkable increases in the utilization of certain trace elements. These phases of life carry with them, thus, particular vulnerability to deficiency, imbalance or toxicity of specific elements. Examples of trace elements deficiency highlight the important role they play in newborns’ health status. The lungs of the infant may be vulnerable to oxidative damage. Glutathione peroxidase levels are regulated not only by development, but also by Se availability. Low Se levels are associated with low enzyme activity and consequent increase of the susceptibility to oxidative lung injury as bronchopulmonary dysplasia. Alkaline phosphatase is reported to reflect Zn status. In young infants the marked reduction of this enzyme can be rapidly cleared up by Zn

A. Alimonti et al. / Clinica Chimica Acta 292 (2000) 163 – 173

165

supplementation. Molybdenum is an essential trace element required by at least, three enzymatic systems. Infants might be at risk of Mo deficiency when born before adequate stores are formed. On the other hand, trace elements could have toxic effects when present at concentrations higher than physiological levels. The mechanisms of toxicity include inhibition of enzyme activity, alterations in the nucleic acid functions and structure, alteration in protein synthesis, effects on membrane permeability and inhibition of phosphorylation. In light of the above considerations, this paper reports the results of determination of the reference ranges for 12 trace elements in serum of a sub-population of 143 term newborns in the urban area of Rome.

2. Materials and methods

2.1. Population characteristics This study includes 143 healthy infants born at the Pediatric Clinic of Rome University ‘‘La Sapienza’’. The newborns were selected on the basis of a questionnaire specifically developed for this project. This investigation has been approved by the Ethics Committee of the hospital and informed written consent was obtained from each mother. Infants with mean gestational age (GA) of 39.1 weeks (range, 37–42 weeks) and mean body weight (BW) of 3258 g (range, 2500–4430 g) were enrolled in this study. Subjects whose clinical examinations were out of the physiological range and with mothers receiving supplementation of the trace elements of interest during pregnancy were excluded.

2.2. Sample processing and analysis Twelve trace and ultra-trace elements were taken into account in this study, namely Al, Cd, Co, Cu, Li, Mn, Mo, Ni, Rb, Sb, Se and Zn. Strict precautions were taken throughout the analytical procedure to minimize sample contamination [7]. Briefly, 0.5–2-ml aliquots of cord blood were drawn after delivery by a PTFE catheter and serum was rapidly separated by centrifugation, transferred to polyethylene metal-free tubes and immediately frozen at 2 288C until the analysis. After 1:10 (v / v) dilution with 0.14 mol / l HNO 3 , determinations of the elements of interest were made by quadrupole inductively coupled plasma mass spectrometry (Q-ICP-MS) combined with an ultrasonic nebulization system. The entire analytical procedure as well as the performance of the method in terms of accuracy, precision and sensitivity were ascertained in a preliminary investigation [7]. In addition to this, further information on the accuracy inherent in the analytical method was attained by means of the ‘‘Second-Generation’’ Biological Reference Material (RM) Freeze-Dried Human Serum supplied by J.

166

A. Alimonti et al. / Clinica Chimica Acta 292 (2000) 163 – 173

Versieck (University Hospital, Ghent, Belgium), certified for the concentration of Al, Cd, Co, Cu, Mn, Mo, Rb, Se and Zn.

2.3. Statistical analysis Simple descriptive statistics such as mean, standard deviation (S.D.), 5% trimmed mean, geometric mean (GM), median, mode and percentile were applied to the analytical data obtained. The normality and log-normality of the results were tested by using the Kolmorogov–Smirnov statistics (K–S test) and skewness and kurtosis were evaluated. Sex of the infants was tentatively treated as an explanatory variable by the Fisher F-test on variances. The statistical correlations among the different elements and the GA and the BW were tested by Pearson and Kendall correlation coefficients. The statistical package SPSS Base (SPSS, Chicago, IL, USA) was used for the analysis.

3. Results and discussion

3.1. Analytical quality control Table 1 summarizes the performances of the analytical method in terms of precision and accuracy as determined by means of the ‘‘Second-Generation’’ Table 1 Analytical quality control for elements in serum by Q-ICP-MS Element

Al Cd Co Cu Li Mn Mo Ni Rb Sb Se Zn a

Precision a

Accuracy Values obtained (mean6S.D., ng / ml)

Confidence ranges expected (95%, ng / ml)

Within series (R.S.D., %)

Between series (R.S.D., %)

1.9660.18 0.2160.02 0.3060.02 977637 1.4560.08 0.7060.04 0.5660.04 0.2260.04 17063.9 1.1060.1 10063.4 821647.6

1.59–2.12 0.15–0.23 0.27–0.38 970–1045 1.59 b 0.67–0.73 0.61–0.75 0.227 b 138–198 1.00 c 90–100 850–890

3.2 4.3 3.9 2.2 3.9 3.8 2.9 2.5 2.3 2.5 2.2 1.5

3.6 5.8 4.3 2.6 4.3 4.1 3.3 3.6 3.1 3.2 3.1 1.8

Twelve repeated measurements. Informative values. c Amount of element spiked to the serum sample. b

A. Alimonti et al. / Clinica Chimica Acta 292 (2000) 163 – 173

167

Biological RM. The experimental concentration of the elements of interest during the entire analytical procedure fall inside the certified 95% confidence interval, with the exception of Mo and Zn, whose levels are close to the given interval. The relative standard deviation (R.S.D.) between the series for all the analytes was always better than 6%. These findings, as well as the S.D. of the measurements of between 1.5 and 4.3% (within-series) and 1.8 and 5.8% (between-series), were fully satisfactory for the purposes of the present investigation.

3.2. Assessment of the reference values Table 2 summarizes the basic information necessary for assessing the distribution of the raw data. In a few cases, i.e., Cd, Li and Ni, only natural log-transformed variables provided good approximations to the normal distribution in terms of K–S test (D-max. values . critical values, at P $ 0.01) and in terms of symmetry and kurtosis (ratio between skewness or kurtosis and their Table 2 Parameters of distribution of the trace elements in serum Element

Al Cd ln-Cd Co Cu Li ln-Li Mn Mo Ni ln-Ni Rb Sb ln-Sb Se Zn a

No. of subjects

78 123 123 131 143 126 126 119 140 136 136 109 143 143 143 138

K–S test D-max.a

Critical value b

0.083 0.401 0.106 0.099 0.104 0.196 0.070 0.057 0.133 0.236 0.130 0.135 0.251 0.220 0.058 0.050

0.184 0.147 0.147 0.142 0.136 0.145 0.145 0.149 0.137 0.139 0.139 0.156 0.136 0.136 0.136 0.139

Symmetry (skew. / S.E. skew.)c

Kurtosis (kurt. / S.E. kurt.)d

4.1 6.4 2.7 5.7 5.6 6.8 0.4 2 0.6 5.7 9.5 2.6 5.7 7.5 2.3 2 0.9 0.3

4.8 4.4 2.2 9.0 2.8 5.1 2 2.3 2 2.0 3.2 8.0 1.6 4.4 5.4 2 2.7 2 1.5 2 1.1

The most extreme absolute difference of observed values from the normal curve. From the tables, at P $ 0.01. c Skewness divided by standard error of the skewness. Critical values: 0, symmetry; . 2.6, right skewness; , 2 2.6, left skewness. d Kurtosis divided by standard error of the kurtosis. Critical values: 0, regular Gaussian form; . 2.6, too-peaked distribution; , 2 2.6, too-flat-topped distribution. b

A. Alimonti et al. / Clinica Chimica Acta 292 (2000) 163 – 173

168

standard error higher than, or close to, the critical value). On the other hand, normality tests for Sb gave no evidence of adequate normal description, even after log-transformation. The strong and positive skewness of the Sb data can explained by the fact that rather often the concentrations of this analyte were close to the detection limit afforded by Q-ICP-MS for the matrix considered (0.1 ng / ml). For this reason, neither reference intervals nor uncertainty ranges were calculated for this analyte. More generally, the trace element composition of the serum was satisfactorily described by a normal distribution, with no need for log-transformation, although the skewness toward higher values and the deviation of the kurtosis became statistically significant. In particular, only Mn, Zn and Se showed excellent approximation to the normal distribution. Consequently, all statistical analyses of Cd, Li, Ni and Sb were carried out on log-transformed values, whereas for all other analytes analytical values were used as obtained. Table 3 gives the basic statistical treatment of data. The 5% trimmed mean, i.e., the arithmetic mean of the data in the 5–95% fractiles, is also reported. The comparison between means (normal and 5% trimmed), GM and median for each analyte further corroborate the evidence of a good central tendency for all elements, with the exception of the four analytes, Cd, Li, Ni and Sb. Table 4 summarizes the results obtained for each element in terms of experimental, fractile, reference and uncertainty ranges. In general, reference ranges are based on the fractiles concept. Here the 5% trimmed mean62 S.D. was preferred [3–5]. The trimmed mean is preferred to the conventional mean because it is less affected by the influence of the skewness in both directions. When the mean 2 2 S.D. value turned out to be lower than the low limit of the experimental range, this last value was included in the reference range, as in the Table 3 Descriptive statistics for trace elements (ng / ml) in serum of newborns Element

Mean

S.D.

5% trimmed mean

GM

Median

Mode

Al Cd Co Cu Li Mn Mo Ni Rb Sb Se Zn

3.52 0.28 0.22 340 0.83 1.95 0.86 0.94 598 0.45 45.5 886

1.69 0.13 0.11 182 0.73 0.65 0.33 1.18 366 0.43 12.1 259

3.40 0.26 0.21 327 0.76 1.96 0.84 0.78 569 0.40 45.5 886

3.14 0.11 0.20 305 0.58 1.85 0.81 0.54 503 0.31 43.7 901

3.34 0.09 0.20 298 0.52 1.98 0.80 0.40 486 0.20 45.7 905

4.50 0.05 0.15 206 0.20 1.00 0.90 0.30 191 0.20 29.0 450

A. Alimonti et al. / Clinica Chimica Acta 292 (2000) 163 – 173

169

Table 4 Concentration ranges (ng / ml) of trace elements in serum of newborns Element

Al Cd Co Cu Li Mn Mo Ni Rb Sb f Se Zn

Experimental range 1.12–10.5 0.10–0.68 0.01–0.50 79–959 0.11–3.80 0.60–3.30 0.36–2.09 0.10–5.00 145–2120 0.10–2.08 20.2–80.1 318–1643

5–95% range 1.23–6.43 0.02–0.40 0.05–0.44 117–766 0.15–2.30 1.00–2.95 0.42–1.60 0.12–4.06 189–1326 0.10–1.48 23.2–63.9 449–1300

Reference range

Ranges of uncertainty Upper

Lower

1.12–6.79 0.10–0.52 a 0.20–0.43 140–691 0.31–2.23 0.79–3.26 0.36–1.56 a 0.20–3.15 196–1302

6.59–10.5 0.52–0.68 0.43–0.50 691–959 2.23–3.80 3.26–3.30 1.56–2.09 3.15–5.00 1302–2120

b

20.2–69.7 a 318–1405 a

69.7–80.1 1405–1643

b

a

b

Literature data 0.3–7.5 c 0.04–0.36 c 0.08–0.4 c 44–900 d

0.01–0.20 79–140 0.11–0.31 0.60–0.79

e

b

e

0.10–0.20 145–196

0.24–2.8 c 78–317 c 0.09–0.25 g 31–96 h 378–1130 i

b

0.3–0.9 c

a

The low-end value is the lowest value of the experimental range. Not applicable because there is no difference between the lowest reference value and the experimental one. c Reference values referred to the adult population. d Range including data from Refs. [10–12]. e Not available. f Neither reference nor uncertainty ranges were applicable to this element (see text). g From Ref. [16], referred to infants aged , 1 year. h Range including data from Refs. [11] and [12]. i Range including data from Refs. [10] and [11]. b

case of Al, Cd, Mo, Sb, Se and Zn. In general, these reference ranges are roughly equivalent to those obtainable with the 5–95% percentiles (fourth column vs. third column of Table 4), this providing further evidence of the nearly symmetric distribution. The effects of the sex variable on the serum elemental concentrations were also analyzed. In cord blood serum, no marked sex-related differences (at P , 0.05) between male and female newborns were found. However, according to the sex variable, Fisher F-test revealed significant differences on variances (at P , 0.003) for Co and Se concentrations. The means6S.D., grouped by sex, were 0.2260.08 and 0.2260.14 ng / ml for Co in males and females, respectively, and 47.568.67 and 43.8613.0 ng / ml for Se in males and females, respectively. Only for Se are some previous results available, although they are thought to be contradictory and inconclusive. In fact, whilst Halmesmaki et al. [8] have reported that Se in umbilical serum was unrelated to the sex of the infants, Perona et al. [9] found significantly different Se concentrations in cord blood of male and female newborns. These discrepancies could be explained by

170

A. Alimonti et al. / Clinica Chimica Acta 292 (2000) 163 – 173

the characteristics of the populations studied as well as by the weak differences found. Due the lack of available data in this field, only a few comparisons with analogous populations were possible, namely in the cases of Cu, Sb, Se and Zn. In this study Cu showed a reference interval narrower than the one reported in the literature (140–691 ng / ml vs. 44–900 ng / ml), whereas the Se reference range was shifted to a lower concentration (20.2–69.7 ng / ml vs. 31–96 ng / ml) [10–12]. However, the findings of both analytes displayed neonatal serum levels significantly lower than those of the corresponding reference ranges for adults, i.e., 601–1373 ng / ml and 56–105 ng / ml for Cu and Se, respectively [13,14]. This evidence further confirms the physiological lower contents of Cu and Se at the moment of the birth which are completely overcome only after the first few months of life [11,15]. As concerns Sb, the informative range was found to exceed the literature interval as regards the upper end. However, it is worth mentioning that the literature Sb range refers not to newborns, but to infants aged 1 year or less [16]. In the case of Zn, the present reference range is slightly larger than the corresponding interval found in the literature, i.e., 318–1405 ng / ml vs. 378–1130 ng / ml [10,11]. Moreover, when no comparison with a child population is possible, the present findings were compared to the data of adult populations. The analytes can be roughly divided into two groups, respectively including elements (Al, Cd, Co and Ni) which show reference ranges substantially coinciding for both newborns and adults, and elements (Mn and Rb) displaying instead concentrations out of the reference ranges for adults. Table 4 also gives, in the fourth and fifth columns, the upper and lower ranges of uncertainty. The upper range of uncertainty was defined, according to Minoia et al. [13], as the interval between the highest reference value and the highest experimentally observed value. Whenever possible, a lower range of uncertainty, i.e., the interval from the low reference limit to the lowest experimental result, was also reported. On the other hand, the absence of any differences between these values did not allow the identification of a lower interval of uncertainty in the case of Al, Cd, Mo, Se and Zn. However, it must stressed that this approach is only an attempt to organize the data with the purpose of defining limits of concentration beyond which an indication of possible pathological states can be obtained. At present, data falling into the uncertainty intervals can be considered as informative values to which, in the future, particular attention could be paid in order to evaluate the factors affecting them. In fact, data exceeding the upper or the lower limit of uncertainty might be judged as values testifying to possible deficiency or toxicity phenomena.

3.3. Correlation analysis Table 5 summarizes the study on the mutual relationships among the different element concentrations as well as the associations of these with GA or BW at

A. Alimonti et al. / Clinica Chimica Acta 292 (2000) 163 – 173

171

Table 5 Kendall correlations between serum trace elements concentrations, gestional age (GA) and body weight (BW) in newborns GA BW Al

BW

Al

ra

0.52

Pa

, 0.001

r

ns b

ns

ns

ns

ns

ns

Cd

Co

Cu

Li

Mn

Mo

Ni

Rb

Sb

Se

P Cd

r

ns

P Co

r

Cu Li

r

0.27

P

0.007

r

0.49

ns

, 0.001

P

ns

ns

ns

ns

ns

ns

0.24

P Mn

r

0.42 , 0.001

ns

ns

ns

r

ns

ns

ns

0.58

0.47

, 0.001

, 0.001

0.25

0.25

0.02

0.015

0.45

0.22

, 0.001

0.03

ns

P Ni

r

ns

ns

0.41

P Rb

r

ns

0.003 ns

ns

2 0.27

ns

P Sb

r r

ns ns

P Zn

r P a b

ns

ns ns

0.41 , 0.001

ns

2 0.41

2 0.10

, 0.001

0.001

ns

ns

ns

ns

ns

ns

ns

2 0.40

ns

2 0.42

0.04 ns

P Se

0.30 0.005

ns

P Mo

ns

0.04

0.21

2 0.46

0.24

, 0.001

0.03

ns

ns

0.04 ns

ns

0.27 0.05

ns

2 0.36

, 0.001

0.001

ns

, 0.001

0.21

0.38

0.23

0.28

0.05

, 0.001

0.03

0.01

ns

0.29

0.29

0.36

ns

0.30

0.21

0.42

0.008

0.001

0.009

0.07

, 0.001

ns

ns

ns

2 0.23

0.005 0.03

0.44 , 0.001

r: Pearson correlation coefficient; P: significance. ns: Not significant (P . 0.05).

the delivery. No correlation between GA or BW and trace elements levels was found, with the only exceptions of Cu with GA (r 5 0.27, P 5 0.007) and Se with BW (r 5 0.21, P 5 0.04). On the other hand, as expected, a significant positive association was observed between newborn BW and GA (r 5 0.52, P , 0.001). These results might be related to the well-known fact that the Cu and Se body levels are a function of the growth of the fetus and infant. Moreover, Cu and Se concentrations in the cord blood serum were correlated (r 5 0.38, P , 0.001), thus confirming the findings of previous investigations [12]. Significant correlation, i.e., with r . 0.40 and P , 0.001, were found for many elements. Aluminum and Co showed a strong mutual relation as well as associations with Ni and Sb (negative correlation), and Mn and Ni, respectively. Positive interactions of Cu with Mn, of Mn with Mo and of Se with Zn were also found. In turn, Sb was negatively related to Al, Mn and Ni, the latter being

172

A. Alimonti et al. / Clinica Chimica Acta 292 (2000) 163 – 173

also inversely correlated with Mn. The majority of significant correlations was observed for Ni, which correlated positively with Al, Co, Mo and Zn, and negatively with Mn and Sb. If a lower significance level (P , 0.05) is considered, Co, Ni and Zn appear to be the most representative indicators of the elemental status in newborns, as they are involved in eight associations with other elements. However, the interpretation of these findings is still debatable. These results, in fact, suggest that trace element status of newborns and the mutual relationships of the elements are quite complex and unclear.

4. Conclusions These data provide some systematic information about the elemental status of infants at birth. Due the fact that the subjects included in the study were a large group of selected children, these findings can also be considered as indicative of reference ranges of the corresponding urban population. Moreover, the well known association of Cu and Se with newborn growth is confirmed. The elemental correlations found suggest interesting associations which, however, should be corroborated by further experimental evidence.

References [1] Vesterberg O, Alessio L, Brune D et al. International project for producing reference values for concentrations of trace elements in human blood and urine – TRACY. Scand Work Environ Health 1993;19(Suppl. 1):19–26. [2] Solberg HE. The theory of reference values. Part 5. Statistical treatment of collected reference values. Determination of reference limits. Clin Chim Acta 1984;137:97F–114F. [3] International Federation of Clinical Chemists (IFCC). Recommendations on theory of reference values. Clin Chim Acta 1987;170:1–12. [4] International Federation of Clinical Chemists (IFCC). Recommendations on theory of reference values. Clin Chim Acta 1987;170:13–32. [5] International Federation of Clinical Chemists (IFCC). Recommendations on theory of reference values. Clin Chim Acta 1987;170:33–42. [6] Rukgauer M, Klein J, Kruse-Jarres JD. References values for trace element copper, manganese, selenium and zinc in the serum / plasma of children, adolescents and adults. J Trace Elem Med Biol 1997;11(2):923–98. [7] Alimonti A, Petrucci F, Fioravanti S, Laurenti F, Caroli S. Assessment of the content of selected elements in serum of term and pre-term newborn by inductively coupled plasma mass spectrometry. Anal Chim Acta 1997;342:75–81. [8] Halmesmaki E, Alfthan G, Ylikorkala O. Selenium in pregnancy: effect of maternal drinking. Obstet Gynecol 1986;68:602–5. [9] Perona G, Guidi GC, Piga A et al. Neonatal erythrocyte glutathione peroxidase deficiency as a consequence of selenium imbalance during pregnancy. Br J Haematol 1979;42:567–74.

A. Alimonti et al. / Clinica Chimica Acta 292 (2000) 163 – 173

173

[10] Bogden JD, Thind IS, Kemp FW, Caterini H. Plasma concentrations of calcium, chromium, copper, iron, magnesium and zinc in maternal and cord blood and their relationship to low birth weight. J Lab Clin Med 1978;92(3):455–62. [11] Jochum F, Fuchs A, Cser A, Menzel H, Lombeck I. Trace mineral status of full-term infants fed human milk, milk-based formula or partially hydrolysed whey protein formula. Analyst 1995;120:905–9. [12] Arnaud J, Preziosi P, Mashako L et al. Serum trace elements in Zairian mothers and their newborns. Eur J Clin Nutr 1994;48:341–8. [13] Minoia C, Sabbioni E, Apostoli P et al. Trace element reference values in tissue from inhabitants of the European Community. I. A study of 46 elements in urine, blood and serum of Italian subjects. Sci Total Environ 1990;95:89–105. [14] Caroli S, Alimonti A, Coni E, Petrucci F, Senofonte O. The assessment of reference values for elements in human biological tissues and fluids: a systematic review. Crit Rev Anal Chem 1994;24(5&6):363–98. [15] Burns J, Forsyth JS, Paterson CR. Factors associated with variation in plasma copper levels in preterm infants of very low birth weight. Eur J Pediatr 1993;152:240–3. [16] Cullen A, Kiberd B, Matthews T, Mayne P, Delves HT, O’Regan M. Antimony in blood and urine of infants. J Clin Pathol 1998;51(3):238–40.