JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1991 VOL. 6 647 Determination of Zinc in Human Milk by Electrothermal Atomic Absorption Spectrometry Josiane Amaud and Alain Favier Laboratoire de Biochimie C Centre Hospitalier Regional et Universitaire de Grenoble BP 21 7X 38043 Grenoble Cedex France Josette Alary Laboratoire de Chimie Analytique Faculte de Pharmacie Universite Joseph Fourier Domaine de la Merci 38700 La Tronche France An electrothermal atomic absorption spectrometric method has been developed for determining zinc in human milk. Milk samples were analysed after 1 +99 dilution with 0.1% Triton X-100. The within-run precision was found to be 4.6%. The accuracy was ascertained by recovery of standard additions and was found to be 96+ 10% (n-85 milk samples). The accuracy was also checked against National Institute of Standards and Technology Standard Reference Material 1549 Non-Fat Milk Powder.The results were 713+35 pmol g-l (n=38) (certified value 705,+34 pmol g-l). The detection limit was found to be 0.052 pmol I-'. The calibration graph was linear up to 3 pmol I-'. The choice of experimental parameters (diluent dilution rate background correction graphite furnace tube etc.) are discussed. Normal values for colostrum and transitional milk varied from 45 to 318 and from 30 to 146 pmol I-' respectively. Keywords Determination of zinc; electrothermal atomic absorption spectrometry; human milk; lactation During the first three months of life milk is the principal source of zinc. However the concentration of zinc in human milk decreases rapidly with progressive lactation.Premature and low birth weight infants are at risk from zinc deficiency owing to their limited gastric capacity imma- ture intestinal kidney and liver function reduced body stores and rapid growth.I4 Severe zinc deficiencies are easy to detect and have been reported in infants receiving intravenous nutrition or formulae with inadequate zinc concentration (less than 31 pmol 1-l) or in premature infants receiving mature human milk from a Cases of zinc deficiency in premature or full term infants fed with their own mothers' milk have also been reported.'-16 In these cases the concentration of zinc in the milk was particularly low ( 1 A7.6 pmol 1-l). Zinc is commonly determined in milk samples by flame atomic absorption spectrometry.The sensitivity of the method is inadequate for determining very low zinc concentrations however with electrothermal atomic absorption spectrometry it is possible to determine low zinc concentrations. In this work a suitable technique for the determination of zinc in human milk using a direct electrothermal atomic absorption spectrometric method is presented. Experimental Instruments and Apparatus A Perkin-Elmer Model 560 atomic absorption spectro- meter was used With a zinc hollow cathode lamp as the light source (intensity- 15 mA). The instrument was fitted with an AS 40 autosampler and an HGA 500 graphite furnace. The absorbance was measured at 2 13.9 nm with a slit-width of 0.7 nm. A three-step graphite furnace programme was used; drymg 20 s at 1 10 "C with a ramp time of 20 s; ashing 30 s at 650 "C with a ramp time of 1 s and atomization 5 s at 2200 "C with a ramp time of 2 s.The internal gas was nitrogen and the flow rate during the atomization step was reduced to 250 ml min-l. The injection volume in either standard or pyrolytic graphite coated graphite furnace tubes All of the plasticware was made of polycarbonate polyethylene or polystyrene and was soaked for 16 h in 10% v/v nitric acid then again in 10% v/v hydrochloric acid for was l0pl. a further 16 h. Prior to use the plasticware was rinsed in de- ionized water and dried in a stainless-steel oven. The subjects who gave oral consent were enrolled during the first week of lactation. A total of 112 lactating mothers providing 203 individual milk samples were selected.Of these 97 mothers lived in the urban Grenoble area while 15 lived in rural areas 82 of the mothers who agreed to participate in this study were Caucasian 25 were Arabian 4 were Asian and 1 was Negro. Of the total 20 of the mothers were from low-income 87 from middle-income and 5 from high-income families. While 64 of the mothers were primiparae 48 were multiparae. The mean age of the mothers was 26 years (range 19-39 years). The mean weights before pregnancy and before delivery were 56 kg (range 40-87 kg) and 67 kg (range 52-96 kg) respectively. The mean height was 16 1 cm (range 145- 176 cm). All of the mothers were healthy and apparently well-nourished women based on clinical observation. There was neither glucose nor albumin in their urine and the mean haemoglo- bin value at delivery was 1 14 g 1-l (range 97-1 37 g 1-I).All had uncomplicated pregnancies and delivered a single infant at term [ 39 A 2 weeks (mean ~f standard deviation) range 36-43 weeks]. Of the infants 53 were male and 59 female. All were healthy and growing well. The mean birth weight of the infants was 3293 g (range 2370-4300 g) and their mean height was 50 cm (range 46-54 cm). Samples The breast was cleaned with de-ionized water. Approxi- mately 10 ml of breast milk were hand expressed before one of the morning feeds (09.00-1 1.00 hours) into a 30 ml polystyrene bottle. Milk samples were transported to the laboratory on ice. Aliquots of the milk (1 ml) were transferred into 5 ml polystyrene tubes then frozen at - 20 "C prior to analysis.Reagents Hydrochloric acid (0.1 mol 1-l) was prepared from ultra- pure hydrochloric acid I I mo1 1-' (Prolabo Normatom).648 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 Nitric acid (0.1 moll-') was prepared from ultrapure nitric acid 15 mol 1-l (Merck). The following reagents were also used 0.05- 1 % m/v Triton X- 100 (Prolabo); 1 % m/v sodium lauryl sulphate (Sigma); 1% m/v benzyl dimethyl hexadecyl ammonium chloride (Fluka); and 0. l0h m/v histidine (Sigma). A zinc stock solution containing 75 pmol 1-l of zinc was also prepared. Procedure The preparation of the standard solutions and samples and the analyses were conducted in a filtered-air room (class 100 000). Preparation of standard solutions The stock solution was prepared from zinc powder (Merck) 4.9 mg were dissolved in 5 ml of 11 mol I-' hydrochloric acid.After total dissolution the solution was poured into a calibrated flask containing about 700 ml of de-ionized water. The volume was made-up with de-ionized water. Aliquots (5 ml) of the stock solution were transferred into 5 ml polystyrene tubes and stored at -20 "C. Working standard solutions were prepared from the stock standard solution using the same diluent as for the milk samples. The zinc concentrations used were 0.375,0.75 and 1.5 pmol l-l when the milk was diluted 1 +99. Preparation of samples A whole human milk sample was diluted in 0.1 Oh Tritron X- 100 ( 1 + 99) before injection into the graphite furnace tube. Statistical analysis The mean and standard deviation were calculated and Student's t-test was performed.The linear regression and correlation coefficient were also determined. Results and Discussion Graphite Furnace Programme The optimum conditions of the graphite furnace pro- gramme were determined. All measurements were made in triplicate. A temperature of 650 "C was the highest temperature that could be used without volatilization of the zinc. During the atomization step the temperature and ramp time affected the sensitivity. The results presented in Fig. 1 show that the lowest temperature required to achieve total volatilization of the zinc was 2200 "C. A ramp time of 2 s allowed the best sensitivity (Fig. 1). A gas flow of 250 ml min-l provided the greatest linear range. Saner and 1000 1500 2000 2500 500 Tern peratu re/"C Fig.1 Influence of atomization temperature and ramp time (RT) on the absorbance of zinc in solution (broken lines) or in milk sample (solid lines). A and B RT = 0 s; C and D RT = 1 s; E and F RT=2 s; and G and H RT=5 s Table 1 Amount of zinc in the selected diluents (n = 30 determi- nations) Diluent De-ionized water HCl 0.1 mol 1-1 HN03 0.1 mol 1-1 Triton X-100 0.1% m/v Triton X-100 1% m/v Sodium lauryl sulphate 1% m/v Benzyl dimethyl hexadecyl ammonium chloride 1% m/v Histidine 1% m/v *ND = not detectable. Amount of zinchmol 1-1 ND* 46+ 13 77+ 15 ND ND 1381- 11 92+ 19 ND Yuzbasiyanl' determined zinc in human milk by electroth- ermal atomic absorption spectrometry but prior to analy- sis the milk samples were dry ashed. They used ashing and atomization temperatures of 450 and 2000 "C respectively with a 230 ml min-l gas flow during the atomization step.Influence of Diluent and Dilution Rate De-ionized water two acids (0.1 mol 1-I nitric and hydrochloric acids) various wetting agents (1% sodium lauryl sulphate Triton X-100 and benzyl hexadecyl di- methyl ammonium chloride) and a chelating agent (0.1Oh histidine) were selected for trial. De-ionized water is the most commonly used diluent for zinc determination in milk by flame atomic absorption spectrometry. Nitric and hydrochloric acids might influence the ashing and atomiza- tion steps. Furthermore nitric acid has been used by Murphy et a!.'* and Moran et all8 However Arpadjan and Nakova19 have shown that a wetting agent is necessary for direct determination of zinc in milk by flame atomic absorption spectrometry because of the high concentra- tions of lipids in the sample.Three wetting agents were selected according to their polar properties one anionic (sodium lauryl sulphate) one cationic (benzyl hexadecyl dimethyl ammonium chloride); and one neutral (Triton X- 100). The zinc present in milk is chelated by citric acid amino acids and proteins,20-22 as a result a chelating agent was also selected. The contamination from these diluents was calculated trom an aqueous calibration graph (Table 1). The zinc concentrations of the blanks were undetectable for Triton X- 100 and 0.1 % histidine. Zinc contamination in the other diluents varied from 46 k 13 (0.1 mol 1-l HCl) to 138 2 1 1 nmoll-l(lo/o sodium lauryl sulphate) (n= 30).Zinc contam- ination was lowest in the acids and highest from the 1% sodium lauryl sulphate and 1 O/o benzyl hexadecyl dimethyl ammonium chloride. Nevertheless these contaminations were sufficiently low to allow the determination of zinc in normal human milk. The results of the recovery experiment are given in Table 2 and the within-run precision in Table 3. The recovery of standard additions increased with the dilution rate. The recovery of standard additions varied from 65 (1 +49) to 144Oh (1 + 499) (de-ionized water) from 50 (1 + 49) to 1 12% (1 + 499) (0.1 moll-' hydrochloric acid) from 37 (1 +49) to 1OOoh (1 +499) (0.1 moll-' nitric acid) from 84 (1 + 49) to 200% ( 1 + 999) ( 1 % Triton X- loo) from 37 ( 1 + 49) to 107% (1 + 999) (1 96 benzyl hexadecyl dimethyl ammonium chlo- ride) from 42 ( 1 + 49) to 143% (1 + 999) (1 Oh sodium lauryl sulphate) and from 66 (1 +49) to 126Oh (1 +999) (0.1% histidine). With a dilution of 1 +49 the linearity was insufficient for accurate recoveries.However with dilutions of 1 +499 or 1 +999 (results not shown) the recoveryJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 649 Table 2 Influence of the diluent and dilution rate on the recovery of standard additions (n = 10) replicate determinations of the same sample analysed on different days). Values between 90 and 110% are considered satisfactory Dilution rate Diluent De-ionized water HC1 0.1 mol 1-1 HN03 0.1 mol 1-I Triton X-100 1% m/v Benzyl dimethyl hexadecyl ammonium chloride 1% m/v Sodium lauryl sulphate 1% m/v Histidine 0.1% m/v *Values given + 1SD.1 +49 65+5 50+ 1 37+4 84+8 42+2 66+3 37+ 1 1 +99 82+6 74+7 57+ 1 111+7 69+ 1 66+3 70+4 1 + 199 98+ 13 9728 69+5 122+ 12 71 + 7 77+7 74+9 Table 3 Influence of the diluent and the dilution rate on the within-run precision expressed as the RSD= [ B D ] x 100 for n = 30 determinations of different dilutions of the same milk sample. Values of less than 5% are considered satisfactory Dilution rate Diluent De-ionized water HCl 0.1 mol 1-1 HN03 0.1 rnol 1-I Triton X-100 1% m/v Sodium lauryl sulphate 1% m/v Benzyl dimethyl hexadecyl ammonium chloride 1% m/v Histidine 0.1% m/v 1+49 1+99 5 10 1.5 5 4 4 2.5 5 3 3 2 4 3 3 1 + 199 10 9 6 7 4 4 4 experiments suffered from problems of contamination. Although a mean recovery of standard additions between 90 and 110% was observed with all of the diluents tested the dilution rate was different for each.Dilutions of 1 + 499 even if the recoveries of the standard additions were correct were not recommended because of the low signal- to-noise ratios and the special requirements needed to avoid contamination. As a result the relative standard deviation (RSD) (Table 3) increased with the dilution rate. The RSD varied from 5 (1 +49) to 39% (1 +999) (de- ionized water) from 1.5 (1 +49) to 30% (1 +999) (0.1 mol 1-1 HCl) from 4 (1 +49) to 24% (1 + 999) (0.1 mol 1-1 HN03) from 2.5 (1 +49) to 36% (1 +999) (1% Triton X- loo) from 3 (1 +49) to 5Oh (1 +999) (1% sodium lauryl sulphate) from 2 (l+49) to 8% (1 +999) (1% benzyl dimethyl hexadecyl ammonium chloride) and from 3 (1 +49) to 7% (1 + 999) (0.1% histidine).The precision achievable was better in the presence of wetting agents which allowed stable homogenization of the milk samples. l9 The detection limit and linearity are shown in Table 4. The detection limits were calculated according to the following criteria X+3SD where X is the mean of 30 replicate zinc determinations at the blank level and SD is the corresponding standard deviation. The detection limit depended on the diluent (Table 4) as did the accuracy and precision. The best detection limits were obtained with de- ionized water O.loh histidine and l0h Triton X-100. The linearity also depended on the diluent. The linearity was greater with de-ionized water and 1 % Triton X- 100 (Table 4).The slopes of the calibration graphs (results not shown) depended on the diluent which was in agreement with findings previously reported. l9 According to these different results Triton X-100 and a 1 + 99 dilution were chosen as optimum. This choice of diluent is in agreement with that of Arpadjan and Na- kova,19 who compared a variety of different diluents [de- ionized water 0.05-0.1 % Triton X- 100,0.05-0.2% Meriten Table 4 Influence of the diluent on the linear range. Detection limit defined as X+ 3SD for n = 30 determinations at the blank level and linearity value for triplicate determinations Detection limit/ Linearity/ Diluent nmol 1-1 pmol I-' De-ionized water HCl 0.1 mol 1-1 HN03 0.1 mol 1-1 Triton X-100 1% m/v Sodium lauryl sulphate 1% m/v Benzyl dimethyl hexadecyl ammonium chloride l0h m/v Histidine 0.l0h m/v Triton X- 100 0.1 O/o m/v 39 4 85 3 122 2 27 4 171 1.5 149 2 35 3 52 3 (nonyl phenyl polyglycolether) 0.05-0.2% saponin and 0.05-0.2% sodium dodecyl benzenesulphonate] for the determination of zinc in milk by flame atomic absorption spectrometry and selected 0.1% Meriten and 0.05% Triton X-100. According to Arpadjan and Nakova,19 a wetting agent was also necessary to homogenize and dissolve the lipids present in the milk. Effect of the Concentration of Triton X-100 The results are presented in Fig. 2. The recovery of the standard additions increased with concentration up to 0.1 % Triton X-100 and then reached a plateau. Consequently 0.1% Triton X-100 was used. Background Correction Deuterium background correction has been used by several groups of w ~ r k e r s .l ~ * * ~ t ~ ~ In this study zinc was determined L Q) 8 50 CT 0 Concentration of Triton X-100 (% m/v) Fig. 2 Influence of Triton X-100 concentration on the recovery of standard additions (mean f 1 SD) C B 1 I Time - Fig. 3 Absorbance of zinc in diluted milk with and without deuterium background correction. A background; B corrected absorbance; and C uncorrected absorbance650 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 in milk both with and without deuterium background correction (Fig. 3). The slopes of the calibration graphs were similar in diluted milk (54 with deuterium background correction and 55 without n= 3). Furthermore no differ- ence was found in precision or accuracy.The mean recovery of the standard additions measured in triplicate was 10 1 % with deuterium background correction and 100% without. The mean peak heights were found to be 96f6 mm (n=30) with deuterium background correction and 97.5 k 5 mm (n=30) without. Zinc concentrations were slightly lower with deuterium background correction (171.5+ 10.3 versus 174.5k8.7 pmol l-l n=30). This difference was not statistically significant. However the within-run precision was slightly improved without deuter- ium background correction ( 5 versus 6% n= 30). Conse- quently it was decided not to use the deuterium back- ground correction. The relatively high dilution used and the relatively low ion and low protein concentrations in milk probably explain the reduced interferences.Comparison of Standard (Non-coated) and Pyrolytic Graph- ite Coated Graphite Tubes Pyrolytic graphite coated graphite tubes for use in the furnace have a lower permeability to gases a lower porosity and a higher resistance to oxidation than standard graphite tubes. As a result pyrolytic graphite coated graphite tubes have extended lifetimes. However they are twice as expensive as standard graphite tubes. The detection limit was improved when using the standard graphite tubes (52 nmol 1-l) rather than the pyrolytic graphite coated graphite tubes (88 nmol 1-l). These results were different from those found by Sturgeon and Chakraba~=ti.~~ The slopes of the calibration graphs prepared in 0.1% Triton X-100 and in diluted milk were enhanced when using the standard graphite tubes [0.1% Triton X-100 5 1 & 2 versus 46 f 3 (n= 10); diluted milk 5 1 f 3 versus 49 + 3 (n= 30)].These results were in agreement with those found by Foote and Delves.26 No difference was observed in this work for the accuracy and precision. The recoveries of standard additions were 100 f 4% with a standard graphite tube and 107 + 5Oh with a pyrolytic graphite coated graphite 25c r L - K 3 .- 2 225 e N L 9 E k z # E 75 i2 - B 1 150 0 K 0 I I 1 75 150 225 250 Results from pyrolytic graphite coated graphite tube for zindpmol I-’ Fig. 4 Influence of the type of graphite tube on zinc concentra- tion. Regression line solid line; and line of identity broken line. The cross in the centre indicates the mean+ 1SD for n=75 milk samples tube (n=30 replicate determinations of the same milk sample).The within-run precision was slightly better with a standard graphite tube (3 versus 5% n= 30) which is in agreement with the previous The zinc concentrations were determined in 75 different milk samples using both the standard and the pyrolytic graphite coated graphite tubes. The results obtained (Fig. 4) showed good agreement between the two types of graphite tubes. A mean zinc concentration of 170 pmol 1-* with a standard deviation of 80 pmol l-l (range 46-283 pmoll-*) were obtained using the standard graphite tubes whereas with the pyrolytic graphite coated graphite tubes the mean zinc concentration was 169 k 75 pmol 1-l (range 46-296 pmol 1-l). The correlation coefficient (r) was found to be 0.90. The linear regression line had a slope of 0.96 and an intercept of 7 pmol 1-l.Nevertheless matrix interferences differed with the nature of the graphite tubes used for the analysis. The standard additions recoveries were found to be 97 k 11% (n=85) with the standard graphite tubes and 97 f 12% (n=85) with the pyrolytic graphite coated graph- ite tubes; but no correlation was found between the recoveries obtained (r=0.52). These results were in agree- ment with those found by Foote and Delves.26 According to these results standard graphite tubes were considered to be suitable for the determination of zinc in human milk. Calibration Zinc was determined in 85 milk samples by using either an external calibration procedure (standard solutions made up in 0.1% Triton X-100) or by the method of standard additions.The results presented in Fig. 5 show good agreement between the two calibration processes. Zinc concentrations were not statistically significantly different. With the external calibration the mean zinc concentration was 149f64 pmol 1-l (range 48-274 pmol I-l) whereas with the method of standard additions the mean zinc concentration was 159 f 74 pmol 1-l (range 46-289 pmol 1-I). The correlation coefficient was found to be 0.90. The linear regression line slope was 1.02 and the intercept 6 pmol l-l. Nevertheless the recoveries of standard additions varied from 8 1 to 120% (Fig. 6). Considerable variability in mineral content of human milk has been reported2’ and might explain these results. 2 0 75 150 225 250 [Zinc] by external calibration/pmol I-’ Fig. 5 Influence of the type of calibration procedure on the zinc concentration obtained from milk.Regression line solid line; and line of identity broken line. The cross in the centre indicates the mean f 1SD of n= 85 milk samplesJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 65 1 ~~ ~ Table 5 Zinc concentrations (in pmol 1-l) in human milk samples in the first days of lactation Data source Days post-partum 1 122k87 Ref. 28 n=7 Ref. 23 n= 13 Ref. 35 n= 10 This work n=6 131 k86 119+20 128k34 2 n=8 177+72 n=24 168+69 - - n=27 167k61 3 n=9 144+30 n-34 133+41 n= 10 69 + 20 n=46 134+51 4 n=9 98+ 10 n= 33 99 f 23 n = l l 88 f 26 - - 5 n=8 83k 17 n=33 82k 18 n= 10 60k 19 n=25 83 k 29 6 7 8 - n= 10 - - 73_+ 16 - - - - - - - - n= 10 - - 51 + 19 n=41 n=44 n=3 75k25 74k17 64+17 - 80 90 100110120 Recovery (%I Fig.6 Distribution of recoveries of standard additions in 85 milk samples The external calibration procedure was retained for practical reasons. This choice was in agreement with previous studies.1a~2a Precision Accuracy and Linear Range The electrothermal atomic absorption spectrometric method selected used standard (uncoated) graphite tubes and no background correction. The diluent was a 0. l0h m/v Triton X-100 solution and the dilution rate was 1 + 99. The working standard solutions were prepared in the same diluent and contained zinc concentrations of 0.375 0.75 and 1.50 pmoll-l. The precision expressed as the RSD for 30 analyses of different dilutions of the same sample was 4.6%.The accuracy was evaluated by the recovery of standard addi- tions. By using 85 different milk samples recoveries of different amounts of zinc which had been added to the milk were 97 2 10%. The accuracy was also determined by analysing National Institute of Standards and Technology Standard Reference Material 1549 Non-Fat Milk Powder. The mean value was found to be 7 13 k 35 pmol g-l (n= 38) (certified value 705 f 34 pmol g-l). The detection limit was calculated according to the criteria X+3SD where X was the mean of 30 replicate determinations at the blank level (0.1% Triton X-1 00) and SD the corresponding standard deviation. The detection limit was 52 nmol 1-l. The linear limit determined over a period of 3 d was up to 3 pmol l-I. Normal Values Zinc was determined in 203 milk samples collected in the first week post-partum.The major factor influencing zinc concentration in human milk is the stage of lactation. Factors such as age parity and anthropometric measure- ment do not influence the zinc con~entration.~~-~~ Zinc concentrations in the milk of women living in urban or rural areas are ~ i m i l a r . j ~ - ~ ~ The results are presented in Table 5. There was an increase in the concentration of zinc from day 1 to day 2. After day 2 the zinc concentration in milk declined. This instability of the zinc concentration during the first post-partum week has been reported by other workers,23*28*35 and reflects the rapidly changing physiological state of the mammary gland. The results given here are similar to those reported by Casey et af.23*28 but the zinc concentrations found are higher than those reported by Hibberd e? al.35 Differences amongst individual women were substantial.These results were also in agreement with those previously r e p ~ r t e d . ~ ~ * ~ ~ y ~ ~ Conclusion The validity of the proposed method was more than adequate for clinical and biochemical investigations. Zinc concentrations at early stages of lactation were in agree- ment with those reported by Casey et a1.,23*28 although the methods of analysis used were different. Casey e? al.23*28 used a flame atomic absorption spectrometric method and prior to analysis the milk samples were ashed in a low- temperature asher. The proposed method presents the advantage of only performing a simple dilution of the milk sample before injection into the graphite furnace. The detection limit was adequate for the determination of zinc in milk with a low zinc content.This direct method allowed rapid determinations. Problems of contamination were reduced by the use of only a single reagent before injection into the graphite furnace. Numerous determinations could be performed on the same day. References 1 Friel J. K. Gibson R. S. Balassa R. and Watts J. L. Acta Paediatr. Scand. 1984 73 596. 2 Pleban P. A. Numerof B. S. and Wirth F. H. in Clinics in Endocrinology and Metabolism ed. Taylor A. vol. 14 WB Saunders London 1985 p. 545. 3 Mendelson R. A. Bryan M. H. and Anderson G. H. J. Pediatr. Gastroenterol. Nutr. 1983 2 256. 4 Ziegler E. E. Am. J. Clin. Nutr. 1985 41 440. 5 Aggett P.J. Atherton D. J. More J. Davey J. Delves H. T. and Harries J. T. Arch. Dis. Child. 1980 55 547. 6 Ahmed S. and Blair A. W. Arch. Dis. Child. 1981 56 31 5. 7 Blom I. Jameson S. Krook F. Larsson-Stymme B. and Wraume L Br. J. Dermatol. 1981 104 459. 8 Connors T. J. Czamecki D. B. and Haskett M. I. Arch. Dermatol. 1983 119 319. 9 Courtney Moore M. E. Moran J. R. and Greene H. L. J. Pediatr. (St. Louis) 1984,105 600.652 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 10 Murphy J. F. Gray 0. P. Rendall J. R. and Hann S. Early Hum. D o . 1985 10 303. 11 Parker P. H. Helinek G. L. Meneely R. L. Stroop S. Ghishan F. Y. and Greene H. L. Am. J. Dis. Child. 1982 136 77. 12 Weymouth R. D. Kelly R. and Landsdell B. J. Aust. Paediatr. J. 1982 18 208.13 Zimmerman A. W. Hambidge K. M. Lepow M. L. Greenberg R. D. Stover M. L. and Casey C. E. Pediatrics 1982,69 176. 14 Bye A. M. Goodfellow A. and Atherton D. J. Pediutr. Dermatol. 1985 2 308. 15 Kuramoto Y. Igarashi Y. Kato S. and Tagami H. Acta Dermatol. Venereol. 1986 66 359. 16 Roberts L. J. Shadwick C. F. and Bergstresser P. R. J. Am. Acad. Dermatol. 1987 16 301. 17 Saner G. and Yuzbasiyan V. Nutr. Rep. Int. 1984,29 1181. 18 Moran J. R. Vaughan R. Stoop S. Coy S. Johnston H. and Greene H. L. J. Pediatr. Gastroenterol. Nutr. 1983 2 629. 19 Arpadjan S. and Nakova D. Nahrung 1981 25 359. 20 Blakeborough P. Salter D. N. and GUK M. I. Biochem. J. 1983,209 505. 21 Cousins R. J. and Smith K. T. Am. J. Clin. Nutr. 1980,33 1083. 22 Lonnerdal B. Am. J. Clin. Nutr. 1985 42 1299.23 Casey C. E. Hambidge K. M. and Neville M. C. in Truce Elements in Man and Animals eds. Mills C. F. Bremner I. and Chesters J. K. vol. 5 CAB 1985 p. 633. 24 Costantini S. Macri A. and Vernillo I. Riv. Suc. Ital. Sci. Aliment. 1981 10 231. 25 Sturgeon R. E. and Chakrabarti C. L. Anal. Chem. 1977,49 90. 26 Foote J. W. and Delves H. T. Analyst 1982 107 1229. 27 Atkinson S. A. Radde I. C. Chance G. W. Bryan M. H. and Anderson G. H. Early Hum. Dev. 1980 4 5 . 28 Casey C. E. Hambidge K. M. and Neville M. C. Am. J. Clin. Nutr. 1985 41 1193. 29 Feeley R. M. Eitenmiller R. R. Benton Jones J. and Barnhart H. Am. J. Clin. Nutr. 1983 37 443. 30 Higashi A. Ikeda T. Uehara I. and Matsuda I. Tohoku J. Exp. Med. 1982 137 41. 31 Picciano M. F. Calkins E. J. Gamck J. R. and Deering R. H. Acta Paediatr. Scand. 1981 70 189. 32 Rajalakshmi K. and Srikantia S. G. Am. J. Clin. Nutr. 1980 33 664. 33 Clemente G. F. Ingrao G. and Santaroni G. P. Sci. Total Environ. 1982 24 255. 34 Coni E. Falconieri P. Ferrante E. Semeraro P. Beccaloni E. Stacchini A. and Caroli S. Ann. 1st. Super. Sunita 1990 26 119. 35 Hibberd C. M. Brooke 0. G. Carter N. D. Haug M. and Haner G. Arch. Dis. Child. 1982 57 658. Paper 1 /O I2 781 Received March I8th 1991 Accepted August 16th I991