3. Trace E[em. Med. BioL. Vo[. 15, pp. 103-108 (2001) http://www, urba nfischer.de/journa[s/jtracee[rn © 2001 by Urban & Fischer Vertag Influence of the glass packing on the contamination of pharmaceutical products by aluminium. Part I1: Amino acids for parenteral nutrition Denise Bohrer*, Pau[o Cicero do Nascimento, Regina Binotto, and RocheLe CarLesso Departamento de Qu~mica, Universidade Federal de Santa Maria, Santa Maria, RS, Brazil *Correspondence to: Dr. Denise Bohrer, Departamento de Quimica, Universidade Federal de Santa Maria, 97110-900 Santa Maria, RS, BraziL, Phone/Fax: 055/2208870, E-maiL: ndenise@quimica.ufsm.br Abstract The presence of aLuminium in amino acids parentera[ nutrition solutions can be related to the affinity of the amino acids for atuminium present in glass containers used for storage. For this study soLutions of 19 amino acids used in parentera[ nutrition were stored individuaLLy in glass flasks and the a[uminium measured at deter- mined time intervals. SoLutions of comp[exing agents for aLuminium, as ethy[ene-diaminetetraacetic acid, nitriLo- triacetic acid, citrate, oxa[ate and fluoride ions were also stored in the same flasks and the aLuminium measured during the same time intervaL. The measurements were made by eLectrothermaL atomic absorption spectrometry. The a[uminium content of the glass containers was aLso measured. The results showed that the gLasses have from 0.6% to 0.8% A[. OnLy solutions of cysteine, cystine, aspartic acid and g[utamic acid became contaminated by a[uminium. As the same occurred with the compLexing agents, aluminum can be reLeased from glass due to an affinity of the substances for a[uminium. Comparing the action of comp[exing agents and amino acids for which the stability constants of a[uminium complex are known, it is possible to relate the magnitude of the stabiLity constant with the aLuminium Leached from glass, the higher the stability constant, the higher the aLuminium released. The analysis of commercial formulations with and without cysteine, cystine, g[utamic acid or aspartic acid stored in gLass containers confirms that the presence of these amino acids combined with the age of the solution are, at Least partiaLLy, responsibLe for the aLuminium contamination. The results demonstrated that the contamination is an ongoing process due to the presence of aLuminium in glass combined with the affinity of some amino acids for this element. Key words: aLuminium contamination, glass containers, amino acids, parenteraL nutrition (Received September 1999 - Accepted April 2000) Introduction The presence of aLuminium as contaminant in parenteraL nutrition (PN) solutions is weLL-known and has been very discussed in the Literature in [east years (1-5). Many stud- ies showed that PN solutions are contaminated with aLu- mJnium even when they have different compositions and are the same products but from different brands (6, 7). ALthough atuminium contamination of PN soLutions is suspected to cause impaired bone growth or neuroLogicaL development in preterm infants since at [east 15 years (8-14), the contamination of commerciaL intravenous- feeding solutions is stiLL today a trouble (15-18). Parentera[ preparations shouLd be stored Jn containers that do not interact physicaLLy or chemicaLLy with the preparations, and are made of a transparent materiaL that do not permit diffusion into or across the waLLs of the container. GLass complies these generaL requirements, and pharmacopoeias (19, 20) prescribe the type of gLass preferable for each parenteraL preparation in the indJvidu- 0946-672X/01/15/2-3-I03 $15.00/0
`
`Eton Ex. 1008
`1 of 6
`
`

`

`104 D. Bohrer et at. a[ monographs. Containers composed of glass should meet the requirements for chemical stability (19, 20), but this chemical stability is expressed only by the hydrolytic resistance, that is evaluated by tritating the alkalinity released into water under prescribed conditions of con- tact. According to their hydrolytic resistance glass con- tainers are classified in Type I, II, III or IV, being Type I or II, due to their high hydrolytic resistance, suitable for aqueous preparations for parenteral use. This high chemi- cal resistance is however obtained by addition of oxides to glass, mainly boric and a[uminium oxides (21); and consequently turns glass a source of a[uminium, and increases the possibility of contamination. The presence of a[uminium in PN solutions could be related to an interaction of these solutions with the atu- minium present in the glass container. The release of alu- minium from glass into acidic or alkaline solutions can be explained by the action of the medium on the glass sur- face; low pH favours exchange of metal ions from glass, whereas high pH solutions promotes the dissolution of the glass surface itself (21). Solutions of amino acid do not show pH at extreme values and therefore their action could not be related to processes similar to those of acids and alkalis. Considering that they form complexes with some metals and therefore act as ligands for metal ions (22), this complex formation could be responsible for the extraction of a[uminium from glass by some amino acids. In this work the action of amino acids used for PN on the a[uminium present in glass containers was studied and compared with the action of substances that complex a[uminium. The aim is to show that the affinity of some amino acids for aluminium is the main factor in their teaching action on glass, and therefore that the glass con- tainer can be source of contamination of amino acids PN solutions by this metal Material and Methods Apparatus A Marian Spectra AA200 atomic absorption spectrometer equipped with a GTA-100 graphite furnace and an autosampler (Melbourne, Australia), a Trox class 100 clean bench (Curitiba, Brazil), Digimed pHmeter D-20 (S~o Pau[o, Brazil) and a Phi[co microwave oven (S~o Pau[o, Brazil) were used. Reagents ALl chemicals were of analytical-reagent grade. All aque- ous solutions were prepared with disti(ted and deionized water, that was further purified by a Milti-Q high purity water device (MiL[ipore, Bedford, USA). An aLuminium stock standard solution containing 1000 mg/L A[ was pre- pared from aLuminium nitrate nonahydrate (Merck, Darm- stadt, Germany) and working standard solutions by suit- able dilutions of the stock solution. The amino acids were from different suppliers and the solutions were prepared by dissolution of the amino acid in Milli-Q purified water at the concentration showed in Table 1. Contamination contro[ To avoid contamination, only plastic materials were used. All Laboratory ware (pipette tips, volumetric flasks, etc.) was immersed for at Least 48 h in a 10% HNO 3 in ethanol (v/v) mixture and shortLy before the use washed with MiL[i-Q purified water. To avoid contamination from the air, aLl steps in the sample and reagents preparation were carried out in a class 100 clean bench. Glass ana[ysis AlL glasses used throughout this work were analysed to determine their hydrolytic resistance (19) and atuminium content. The glass containers were crushed into fragments about 1 mm in size, and 0.1 g of these was placed in a PTFE vessel with 5 mt of HF (48% m/m) (Merck, Darm- stadt, Germany) and 5 mt of water, and heated in a domestic microwave oven ata minimal power (174 W) for 10 rain, the procedure was repeated twice. After the total dissolution of the sample the volume was completed to 200 m[ with water and the atuminium measured by flame atomic absorption spectrometry (FAAS). First expedment Solutions of 19 amino acids were stored separately in 10 m[ type II glass flasks. The amino acids concentrations (Table 1) were the same of Soramin 10% (Darrow Labo I rat6rios S.A., S~o Pau[o, Brazil) or Aminon 20 (J.P. IndOs- Table 1. Atuminium content of amino acids Amino acid Brand Aa conc. A[ At solution 1 substance (g/L) (pg/a) (pg A[/g Aa) ALanine Arginine Aspartic acid Cysteinium chloride Cystine G[utamic acid G[ycine Histidine Isoleucine Leucine Lysine ch[or- hydrate Methionine Phenyla[anine Pro[ine Serine Threonine Tryptophan Tyrosine Va[ine Riede[ 15.0 8.9 _+ 0.3 0.59 de HaEn Merck 12.0 3.6 + 0.4 0.30 Labsynth 2.7 2.6 _+ 0.4 0.96 Merck 0.72 4.0 _+ 0.03 5.55 Merck 0.3 9.8 _+ 0.2 32.67 Ecibra 4.6 10.1 _+ 0.2 2.20 Merck 8.0 3.8 _+ 0.02 0.48 Verp 5.0 14.8 _+ 1.3 2.96 Sigma 5.1 < i < 0.2 Merck 9.8 2.8 _+ 0.3 0.29 Riede[ 6.6 16.2 _+ 1.9 2.45 de Hahn Merck 5.3 < i < 0.2 Merck 5.4 10.3 +_ 0.7 1.91 Merck 15.0 1.8 _+ 0.2 0.12 Sigma 2.5 < i < 0.4 Sigma 4.9 4.5 _+ 0.6 0.92 Vetec 2.0 2.5 _+ 0.3 3.65 Merck 1.6 < I < 0.6 Merck 6.2 < 1 < 0.2 in=3 J. Trace E[em. Med. BioL 15/2-3 (2001)
`
`Eton Ex. 1008
`2 of 6
`
`

`

`Atuminium in parenterat amino acid formulations 105 tria Farmac6utica S.A., S~o PauLo, Brazil) parenteraL solu- tions. The flasks were closed and Left to shake at room temperature during all the time of the experiment. The aluminium content of these solutions was measured by electrothermal atomic absorption spectrometry (ETAAS) 15, 60 and 90 days after they have been stored. The blank was the aLuminium measured in these solutions just before their storage. The solutions were prepared in polyethylene volumetric flasks previously decontaminated as above described. The samples were assayed in triplicate. All glass flasks for storage of amino acid solutions were placed before the use in a muffle oven at 560 °C for 12 h to make their surface free of contamination from adsorbed substances. Second experiment The second experiment was carried out with the amino acids that extracted atuminium from the glass containers in the first experiment. 500 m[ of soLution 0.017 moL/[ of cystine, cysteine, aspartic acid and glutamic acid were placed separateLy in PN glass bottles (500 m D, that were previously heated at 560 °C as above described. The fLasks were shaken during at[ the time of the experiment. The aluminium content of these soLutions was measured 15, 30 and 60 days after the storage at room temperature by ETAAS. The pH of the solutions was measured at the beginning and at the end of the experiment. The same procedure was carried out with solutions of citric acid, sodium fluoride, oxalic acid, ethylene-diaminetetraacetic acid (EDTA) and nitriLotriacetic acid (NTA), aLL from Merck (Darmstadt, Germany), at the same concentration of 0.017 mo[/l. ALl samples were assayed in duplicate. As the aluminium concentration in these soLutions after storing was too high, the analysis was done either by FAAS or ETAAS. Extraction rates Solution 0.028 moL/[ of cystine, cysteine, aspartic acid, glutamic acid, arginine, [ysine g[ycine and alanine were stored in glass bottles for PN solutions (500 m[) at room temperature and Left to shake. The aluminium in these solutions was measured by ETAAS at intervals of 10 days for 60 days and after that at intervals of 30 days for a year. The samples were assayed in duplicate. AnaLysis of commercial amino acid parenterat solutions Seven different formulations of amino acids for parenteral nutrition were analysed. At Least 3 boLttes of each batch were analysed. The formulations showed different compo- sitions, but only one of them had no cysteine, cystine, aspartic or glutamic acid; from this one three sets of 3 bottles with different storage times were analysed. The other six formulations had different storage times. From one of these formulations ten bottles of the same batch were analysed. The aluminium determination was carried out by ETAAS; when necessary the samples were diluted to have their aluminium concentration within the range of the analyti- cal curve. To check the accuracy of the a[uminium measurement in these amino acid formulations recovery experiments were carried out. Three different formulations were spiked with 25 IJg AL/L and the aluminium measured by ETAAS as above. Table 2. Aluminium Leached from glass container by different amino acid solutions as a function of time of storage at room temper- ature. The concentration of amino acids is described in Table 1 Amino acid pH AI leached (IJg AL/g Aa/L) _+ RSD pH initial final 15 days 60 days 90 days ALanine 5.5 * * < 1 6.0 Arginine 8.5 3 ± 0.5 4 _+ 1 4 ± 1 8.5 Aspartic acid 1.7 35 _+ 4 75 _+ 5 106 ± 9 2.5 Cysteine 1.3 230 ± 22 725 _+ 67 1056 _+ 99 1.0 Cystine 6.0 458 _+ 24 1661 _+ 75 3026 ± 246 6.0 Glutamic acid 2.8 27 ~ 2 68 +_ 8 87 ± 10 3.0 Glycine 5.9 * * * 5.g Histidine 8.1 < I I ± 0.1 3 _+ 0.3 8.1 IsoLeucine 6.1 * * * 6.2 Leucine 5.9 * * * 6.0 Lysine 7.0 2 ± 0.2 4 + 0.2 6 _+ 0.4 7.0 Methionine 5.6 * * * 5.6 Phenylalanine 5.5 * * * 5.5 Proline 6.2 * * * 6.3 Serine 7.0 * * * 7.0 Threonine 6.8 * * * 6.8 Tryptophan 5.8 * * * 5.8 Tyrosine 7.0 * * * 6.9 Maline 5.8 < 1 < 1 ] + 0.2 5.9 * No significant difference (n=3) between the mean value and the blank. For blank values see Table 1 3. Trace Elem. Med. Biol. %5/2-3 (2001)
`
`Eton Ex. 1008
`3 of 6
`
`

`

`!06 D. Bohrer et aL Results and discussion The glass flasks used to store the individual amino acid solutions had an aluminium content of 0.6% and the glass bottles of commercial amino acid parenteral solutions about 0.8% AL In Table 1 it is shown the aluminium already present as impurity in the assayed amino acids. Whereas isoleucine, methionine, serine, tyrosine and vaLine showed no con- AI (P-g/D 1500 1000 500 0 Fluoride EDTA NTA Citrate Cys Oxalate Gtu Asp Figure 1. ALuminium released from PN glass bottles by action of 0.0!7 mol/[ solution of amino acids and complexing agents after 60 days storage at room temperature. The numbers above the bars are the stability constants (log K) of the aLuminium com- ptexes (24, 25). AI (pg/I) 12oo ~Cys 900- ~~Cystine 600 300 /~/.~ Arg o 0 1 O0 200 300 400 time (day) Figure 2. A[uminium leached from glass containers by amino acids as function of time at room temperature. Amino acids con- centration: 0.028 mol/L. tamination, cysteine and cystine were the highest con- taminated: ca 6 and 33 pg Al/g amino acid respectively. The analysis of other cystine samples of different batches and from different brands showed also high aluminium content: two samples from Merck had 29 and 63 pg Al/g, and one sample from Sigma 45 pg Al/g. The results in Table 2 show that solutions of cystine, cysteine, aspartic acid, glutamic acid, lysine and arginine extracted aluminium from glass, whereas other amino acids did not. The higher releasing occurred in solutions of cystine, cysteine, aspartic acid and glutamic acid. The action of these amino acids is still stronger than salts (showed in Part I of this work). As the pH of the solutions of these amino acids is between 1 and 6, and they are no ionic substances, their action cannot be related to pH, as occurred with HC[ and NaOH, nor to an ion exchange pro- cess as with salts. In Figure 1 it can be seen that all investigated com- plexing agents extract an elevate amount of aluminium from glass and the action of amino acids can be compared with the action of these substances. As the molar concen- tration of the substances, nature of the glass, temperature and time of contact are the same, it could be possible to relate the leaching action of the substance to its affinity for aluminium. Amino acids complex metals (22); the aluminium com- plexes of glutamic and aspartic acids have known stability constants, and although the stability constant for cys-Al is not found in the literature, cysteine must interact with A[ as it interacts with other metals; this could explain its great leaching action on the aluminium present in glass. Whether the stability constants (23-26) are placed in decreased order, the aluminium in the solutions of these substances arising from glass is in decreased order too (Figure 1). Cysteine is a ligand for aluminium as stronger as citrate or oxalate. The pH of the solution could play a small role in the action of arginine, lysine and histidine. They do not form any known complex with aluminium but due to their alka- line character they act on the glass surface and the alu- minium is released into the solution. Figure 2 shows that an equilibrium between the alu- minium in solution and on the glass surface is estab- lished, but it depends on the nature of the substance and the time of contact. Whereas for arginine, lysine, aspartic and glutamic acid this equilibrium is reached in nearby 90 At (p,g/I) t 6OO ................. . ................................................................ ................. i ................................. 400 1 24 months I ~--.::-!~"12 months 0 months Formulation without Formulations containing at least one of these amino acids cysteine, cystine aspartic acid glutamic acid Figure 3. A[uminium mea- sured in different commercial amino acid parenteral formu- [ations. The samp[es were shared according to the age, 6, ]2 or 24 months old. Each bar corresponds to a mean value of three samples. Each bar type corresponds to a dif- ferent formulation. J. Trace Elem. IVied. BioL 15/2-3 (2001)
`
`Eton Ex. 1008
`4 of 6
`
`

`

`ALuminium in parenteraL amino acid formulations 107 days, for cysteine and cystine after 120 days the atumini- um in solution stiLL increases. When poLyethyLene containers were used in place of glass, no increase in the atuminium concentration was observed over the same time period. Commercial amino acid parenteral solutions Seven types of commerciaL amino acid parentera[ soLu- tions were analysed. OnLy one of these formulations con- tained no cysteine, cystine, aspartic acid or gtutamic acid. As the samples were supplied by the university hospi- taL, they had different storage times, and to simplify the presentation of the data, they were shared in three groups according to their age at the time of the analysis: 6, 12 or 24 months. From some formulations we had samples of three different batches (and different storage times) but from others samples onLy one batch (but three bottles) was obtained. The results of this analysis are in Figure 3. It can be seen that formulations with cysteine, cystine, aspartic acid or gLutamic acid showed more aLuminium than the formulation without these amino acids. The results also shown that the storage time could be respon- sibLe for higher contamination. Whereas in the formuLa- tion without cysteine, cystine aspartic acid or gtutamic acid the atuminium accumulation was not higher than 80 pg/L, even after two years storage, aLL the others showed accumulations above this value even after 6 months stor- age. In Figure 4 are the results of ten samples of the same batch. The aLuminium in these samples varied from 101 to 193 pg/L with a mean deviation of +18 pg/L It can be seen that the atuminium present in these samples is not random, within the same Lot of samples the atuminium LeveLs were reproducible, and the contamination always occurred. To assess the accuracy of measuring atuminium in com- mercial amino acid formulations, three different formuLa- tions were spiked with 25 pg At/L and the atuminium con- tent measured by ETAAS. Recoveries between 88 and 105% showed that the results of the aLuminium present in these solutions are reLiabLe. Conclusion Cysteine, cystine, aspartic acid and gtutamic acid are abLe to release aLuminium from g[ass whereas the other amino acids are not. This action can be related to their affinity for atumini- um like some comptexing agents, EDTA, oxa[ic acid, citric acid and fluoride ions, considering that they extracted a[uminium from glass in the same extension of these agents. A relationship between the stability constant of the a[uminium comp[exes and the atuminium released from glass was observed; the higher the stability the higher the aLuminium released. AlL analysed commercial formu[ations of amino acids showed atuminium, and Like occurred with prepared indi- vidual solutions, the composition and age of the solution are important factors for the atuminium contamination. Formulations containing cysteine, cystine, gLutamic acid and aspartic acid that have been stored for several months are the most contaminated. The difference of the aLuminium extracted by different formulations can be due to the composition and interac- tions between amino acids; with regard to the individual solutions of cystine, cysteine, aspartic acid and gLutamic acid, the difference could be attributed, besides the pres- ence of several amino acids in the formulations, to the maintenance of the pH at a value near the neutral in the commercia[ formu[ations. A[though aLuminium in amino acid parentera[ solutions is not very high, the contamination of these solutions occurs and could be, at Least partiaLLy, attributed to their storage in glass containers. This contribution is the most troubling because it is ongoing until the product is used; over many months of storage, the aEuminium contribution of the g[ass container predominates. Acknowledgments The authors are grateful to CNPq (ConseLho NacionaL de Desen- vo[vimento Tecnot6gico), Brazil and GTZ (GesetLschaft f[ir Tech- nische Zusammenarbeit), Germany for financiaL support and to the parentera[ nutrition service of the Santa Maria University HospitaL for the amino acid formuLations. 2~ 100 50 0 1 2 3 4 5 6 7 8 9 10 sample Figure 4. ALuminium measured in ten samples of one batch of the same formulation. Pediamino PLM 10% (Lab. B. Braun S.A.), containing cysteine and gLutamic acid. Measurement done after 6 months storage. References I. KLein, G.L. (1995) ALuminum in parenteraL soLutions revisit- ed-again. Am. J. C[in. Nutr. 61 (3), 449-456. 2. Klein, G.L., ALfrey, A.C., Shike, M., eta[. (1991) ParenteraL drug products containing aluminum as an ingredient or a contaminant: response to FDA notice of intent. Am. 3. C[in. Nutr. 53, 399-402. 3. KLein, G.L., ALfrey, A.C., Shike, M., et aL. (1991) Parentera[ drug products containing aLuminum as an ingredient or a contaminant: response to FDA notice of intent and request for information. 3. Parenter. Entera[ Nutr. 15, 194-198. 4. Sedman, A.B., Klein, G.L., Merritt, R. 3., et aL (1985) Evi- dence of aLuminum Loading in infants receiving intravenous therapy. N. EngL ,I. Med. 312, 1337-1343. ,3. Trace ELem. Med. BioL. %5/2-3 (2001)
`
`Eton Ex. 1008
`5 of 6
`
`

`

`1_08 D. Bohrer et al. 5. Baydar, T., Aydin, A., Duru, S., eta[. (1997) Aluminum in entera[ nutrition formulas and parenteral solutions. 3. Toxi- coL. C[in. ToxicoL 35, 277-281. 6. Koo, W.W., Kaplan, L.A., Horn, J., et aL (1986) A[urninum in parentera[ nutrition solutions - sources and possible alter- natives. 3. Parenter. and Entera[ Nutr. :10, 591-595. 7. Recknage[, S., Bratter, P., Chfissafidou, A., eta[. (t994) Par- entera[ aluminum loading in critical care medicine. Part I: Aluminum content of infusion solutions and solutions for parentera[ nutrition. Infusionsther. Transfusionsmed. 21, 266-273. 8. Klein, G.L., ALfrey, A.C., Miller, N.L., eta[. (1982) Aluminum loading during total parentera[ nutrition. Am. 3. C[in. Nutr. 35, 1425-1_429. 9. Ott, S.M., Ma[oney, N.A., Klein, G.L., et aL (1983) Aluminum is associated with low bone formation in patients receiving chronic parenteral nutrition. Ann. Intern. Med. 98, 9t0-914. 10. Grajower, M.M. and Sas, N. (1986) More on aluminum in infants. N. Eng[. J. Med. 314, 923. 11. Vargas, J.H., Klein, G.L., Ament, M.E., et aL (1988) Metabolic bone disease of total parentera[ nutrition: course after changing from casein to amino acids in parentera[ solutions with reduced aluminum content. Am. J. C[in. Nutr. 48, 1070-1078. 12. Koo, W.W. (1992) Parentera[ nutrition-related bone disease. J. Parenter. Entera[ Nutr. 16, 386-394. 13. Klein, G.L and Coburn, J.W. (1994) Total parentera[ nutri- tion and its effects on bone metabolism. Crit. Rev. C[in. Lab. Sci. 31, 135-167. 14. Golub, M.S. and Domingo, J.L. (1996) What we know and what we need to know about developmental aluminum toxi- city. 3. Toxico[. Environ. Health 48, 585-597. 15. Bishop, N.3., Morley, R., Day, 3.P., et aL (1997) Aluminum neurotoxicity in preterm infants receiving intravenous-feed- ing solutions. N. EngL J. Med. 336, 1557-1561. 16. Klein, G.L. (1998) Metabolic bone disease of total parenter- a[ nutrition. Nutrition 14, t49-152. 17. Koo, W.W. (1996) Laboratory assessment of nutritional meta- bolic bone disease in infants. C[in. Biochem. 29, 429-438. 18. Driscol[, W.R., Cummings, 0.3. and Zorn, W. (1997) Aluminum toxicity in preterm infants. N. Eng[. O. Med. 337, 1090-1091. 19. British Pharmacopoeia, International Edition, HMSO, United Kingdom, 1993 20. The United States Pharmacopeia, USP 23, USPC, Inc., Rockville, 1995 21. Scho[ze, H. (1988) G[as: Natur, Struktur und Eigenchaften. (3 rd ed) Springer Verlag, Berlin, 305-329. 22. Cowan, J.A. (1993) Inorganic Biochemistry: an introduc- tion. VCH Publishers, New York, 8-11. 23. Orvig, C. (1993) The aqueous coordination chemistry of alu- minum. In: Coordination chemistry of aluminium. (Robin- son, G.H.) VCH Publishers, New York, 85-121. 24. Perrin, D.D. (1983) Stability constants of metal-ion com- plexes - Part B - Organic Ugands. In: IUPAC Chemical Data Series. Oxford, t52-153. 25. H6gfelddt, E. (1982) Stability constants of metal-ion com- plexes - Part A - Inorganic Ligands. In: IUPAC Chemical Data Series. Oxford, 194-195. 26. Ringbom, A. (1963) Comp[exaSon in Analytical Chemistry. Interscience Publishers, London, 293-360. O. Trace Elem. Med. Biol. 15/2-3 (2001)
`
`Eton Ex. 1008
`6 of 6
`
`