Measurements of 44Ca543Ca and 42Ca543Ca Isotopic Ratios in Urine Using High Resolution Inductively Coupled Plasma Mass Spectrometry† STEFAN STU� RUP*ab , MARIANNE HANSENc AND CHRISTIAN MØLGAARDc aPlant Biology and Biogeochemistry Department, Risø National L aboratory, P.O. Box 49, DK-4000 Roskilde, Denmark bDepartment of Chemistry, T he T echnical University of Denmark, Building 207, DK-2800 Lyngby, Denmark cResearch Department of Human Nutrition, T he Royal Veterinary & Agricultural University, Rolighedsvej 30, DK-1958 Frederiksberg C, Denmark A method is described for the measurement of calcium isotopic the same nominal masses as the analyte ions are often seen using FABMS.Many of those interferences can be resolved ratios (42Ca+543Ca+ and 44Ca+543Ca+) in human urine using high resolution inductively coupled plasma mass spectrometry with a higher resolution setting. Increasing the resolution reduces the sensitivity owing to a lower ion transmission (HR-ICP-MS).Relative standard deviations (RSD) of 0.33 and 0.41% were found for 44Ca+543Ca+ and 42Ca+543Ca+, through the instrument. For the measurement of calcium isotopic ratios in plasma and urine by FABMS, a resolution respectively. Using a mass spectrometric resolution setting of 4000, the calcium peaks were resolved from interfering setting of 5000 has been used to separate the calcium ions from other interfering ions.12 Less frequently, secondary ion polyatomic ions (40Ar2H+ and 40ArH2+ on 42Ca+ and 28Si16O+ and 12C16O2+ on 44Ca+).Interferences from Sr2+, mass spectrometry (SIMS),13 resonant ionization mass spectroscopy (RIMS)14 and laser desorption time-of-flight mass which cannot be resolved even using high resolution, were corrected for mathematically using the 43.5Sr2+ peak. The spectrometry15 have been used for the measurement of calcium isotopic ratios. The use of quadrupole inductively coupled sample preparation step consisted of a simple 50-fold dilution of the urine sample with 0.14 M HNO3, which ensured a plasma mass spectrometry (Q-ICP-MS) for measurement of calcium isotopic ratios is very limited due to the fact that the relatively high sample throughput of 6 samples per hour. signals from the calcium isotopes are severely interfered with Keywords: Calcium isotopic ratios; urine; high resolution by polyatomic ions and doubly charged ions.The 44Ca+543Ca+ inductively coupled plasma mass spectrometry; polyatomic ions; and the 48Ca+543Ca+ ratios have, however, been successfully doubly charged ions measured in human urine using Q-ICP-MS with RSD below 1% following a preconcentration step in which calcium was precipitated as oxalate.16,17 In high resolution inductively Calcium is the fifth most abundant element in the earth’s crust, the calcium concentration is approximately 3%.1 Calcium coupled plasma mass spectrometry (HR-ICP-MS), a doublefocusing magnetic sector mass spectrometer is used instead of is essential to humans and a daily intake of 800 mg is recommended for adults in order to maintain calcium homeo- the quadrupole filter of ICP-MS.The mass spectrometric resolution of the HR-ICP-MS can be varied up to a maximum stasis.2 Calcium is, however, one of the nutrients for which requirements are least agreed on. The ability of the human of 10000. The resolution is defined as m/Dm; Dm being the mass dierence between two ions of an average mass, m, which body to adapt to lower calcium intakes3 and, in turn, the influence of various dietary factors on calcium availability4 is give rise to equally intense peaks which overlap at 10% of the maximum peak height.A high resolution setting can sub- currently being debated. Radioisotope methods have been used successfully to determine calcium absorption in humans,5 sequently be used to separate analyte peaks from otherwise overlapping polyatomic interferences, e.g., 42Ca+ can be however, this technique is less suitable for children and pregnant/ lactating women as radioisotopes are a potential hazard separated from 40ArH2+ using a resolution setting of 2300.HR-ICP-MS has previously been used for the measurement of because of internal radiation exposure. By a double stable isotope method calcium absorption can be estimated from the the 25Mg+526Mg+ and 206Pb+5207Pb+ ratios with RSD below 0.1% using a resolution setting of 300.18 This is similar to or ratio of the isotopes in urine.6 A number of mass spectrometric techniques are potentially useful for the measurement of stable even better than RSD obtained by TIMS.With a resolution setting of 3000, 63Cu+ and 65Cu+ peaks were separated from calcium isotopes. Thermal ionization mass spectrometry (TIMS) is the technique that oers the best precision [relative interfering polyatomic ions and the 63Cu+565Cu+ ratio was measured with RSD of 0.3–0.6% in human serum and in standard deviations (RSD) in the range of 0.2–0.5%]7 for the measurement of calcium isotopic ratios and is the most com- Antarctic sediments.19 This paper presents the development of an HR-ICP-MS method for the measurement of calcium monly used technique.8–11 One disadvantage of the TIMS technique is that the analyte has to be separated from the isotopic ratios in human urine.The impact of signal suppression from sample matrix and interferences from polyatomic sample matrix.The sample preparation is therefore often very tedious and time-consuming. Another technique frequently and doubly charged ions on the method performance are also discussed. Throughout this paper the term precision means used for the measurement of calcium isotopic ratios is fast atom bombardment mass spectrometry (FABMS).6,11,12 The relative standard deviation (RSD) unless otherwise stated. precision achieved by this technique is poorer than for TIMS, but the technique has the advantages that liquid samples, like EXPERIMENTAL urine and serum, can be applied directly without prior matrix Instrumentation separation.7 Interference from other ions (often hydrides) with All measurements were performed on a PlasmaTrace2 High Resolution Inductively Coupled Plasma Mass Spectrometer † Presented at the 1997 European Winter Conference on Plasma Spectrochemistry, Gent, Belgium, January 12–17, 1997.(Micromass Ltd, Manchester, England) in operation at Risø Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (919–923) 919National Laboratory, Denmark.This instrument was equipped investigated. The aim was to use the lowest possible dilution factor in order to obtain the largest possible calcium signal with a double-focusing magnetic sector mass spectrometer of reverse Nier–Johnson geometry. The resolution can be varied and thereby also the best theoretical precision. Using dilution factors higher than 200, the experimental precision was limited between 400 and 10000.Throughout this study, a resolution of 4000 was used. At this resolution setting, the ion transmission by the theoretical precision. Using dilution factors lower than 50, particulate matter was deposited inside the injector tube is approximately 20% of that at low resolution (resolution setting 400). All measurements were made with the standard after analysis of only a few samples resulting in decreasing sensitivity and an eventual extinction of the plasma. A dilution configuration of the Plasmatrace2, that is, with a Minipuls 2 Gilson peristaltic pump (Gilson, Villers, France), a concentric factor of 50 was chosen.Before a measurement sequence was started, the instrument was carefully mass calibrated using a nebulizer (AR35–1-F04, Glass Expansion Pty., Australia) and a Scott-type spray chamber maintained at 5 °C. The instrument 2 mg l-1 Ca standard solution. For every six human urine samples, a 2 mg l-1 Ca standard solution was analysed.The settings are summarised in Table 1. analysis time was 9 min per sample solution leading to a sample throughput of approximately six samples per hour. Standard Solutions and Reagents The isotopic ratios measured in the standard solutions were used to correct for instrumental mass bias and to check for All samples and standard solutions were prepared by dilution instrumeal drift. The counted ions under the whole peak with 0.14 M nitric acid.The nitric acid was prepared from 65% were integrated and used for the further calculations. The nitric acid, pro analysi, (Merck, Darmstadt, Germany) further integrated data were transferred to a spread sheet programme purified by sub-boiling (Berghof BSB-939-IR, Germany) and all calculations including correction for detector dead diluted with ultrapure water (>18.2 MV cm) from an Elgastat time, correction for instrumental mass bias and overlap from Maxima Analytical System (Elga, Blocks, England).Calcium doubly charged ions were performed in the spreadsheet. standard solutions were prepared from a 1000 mg l-1 commercially available standard (Merck, Darmstadt, Germany) by dilution with 0.14 M nitric acid. A 2 mg l-1 Ca standard Dead Time Correction solution was used for instrument optimisation and calibration. The 42Ca+543Ca+ and 44Ca+543Ca+ isotopic ratios decreased with increasing calcium concentration. This error was caused Samples by dead time in the ion counting circuitry, which showed a relatively lower sensitivity of the most abundant isotopes with Human urine samples with elevated concentrations of 42Ca increasing calcium concentration. All isotope ratio measure- and 44Ca were sampled following a double-label stable isotopes ments were corrected for detector dead time.The detector experiment (a double-isotope procedure for the measurement dead time was calculated by the following equation21 using Ca of calcium absorption20) at the Research Department of standard solutions of 0.5, 1, 2, 3, 4 and 5 mg l-1: Human Nutrition, at The Royal Veterinary & Agricultural University, Copenhagen, Denmark.All collected urine samples Rc=Rm/(1-Rmt) (49 ml ) were acidified with 1 ml of HNO3 (65%) after sampling where Rm and Rc are the measured and corrected counts in and kept frozen (-18 °C) until analysis. the integrated analyte peak, and t is the detector dead time(s).A detector dead time of #10 ns was found, which is somewhat Sample Preparation, Measurements and Calculations lower than values reported for Q-ICP-MS instruments.21 The sample preparation step was very simple. The acidified human urine was diluted 50 times with 0.14 M nitric acid and Mass Bias thereafter aspirated directly into the HR-ICP-MS instrument. The instrument mass bias was calculated from the following The use of dierent dilution factors (between 10 and 400) was equation:22 (A/B)m=(A/B)t(1+a)n Table 1 Instrumental operating conditions and signal measurement parameters where (A/B)m is the measured isotopic ratio, (A/B)t is the true ratio, n is the mass dierence between the two isotopes and a Rf power 1350 W is the bias per mass unit.The instrument mass bias (%) was Plasma gas flow 14 l min-1 Auxiliary gas flow 1.0 l min-1 found to vary between 1–2 u-1. Bias values of #0.3% u-1 Nebulizer gas flow 0.95 l min-1 (optimised daily) are more typical, but higher values are often found for the Sample uptake rate 0.6 ml min-1 lighter elements due to mass discrimination.22 Ion sampling depth Optimised daily for max.intensity Ion lens settings Optimised for max. intensity and optimum resolution RESULTS AND DISCUSSION Sampler/skimmer cone nickel Polyatomic Ions Resolution 4000 Sweeps 80 All the calcium isotopes are overlapped by polyatomic ions at Scans 1 low resolution. Since the 40Ca+ peak can not be separated Peak widths 3 Points per width 30 from the 40Ar+ peak even at high resolution (a resolution of Dwell times/ms— approximately 190 000 is needed), it cannot be measured by 42Ca+ 4 argon plasma ICP-MS.The 46Ca and 48Ca isotopes have a 43Ca+ 8 low natural abundance (0.004 and 0.187%, respectively) and 44Ca+ 2 the signals are isobaric overlapped by titanium isotopes. 43.5Sr++ 5 Consequently the 42Ca, 43Ca and 44Ca isotopes were chosen Magnet rest mass 41 u Magnet scan region 41–45 u for this study.These isotopes are also interfered with by Hysteresis settle times 10000 ms polyatomic ions, but can be resolved using a resolution setting Large jump settle time 100000 ms >2700. Table 2 shows the polyatomic ions seen when analysing Small jump settle time 100 ms human urine as well as the resolution needed to resolve them Sampling time 9 min per sample from the analyte peaks. According to Table 2, the calcium 920 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12Table 2 Polyatomic interferences found in human urine conditions.23 For the PlasmaTrace2 HR-ICP-MS instrument run under standard conditions we found formation rates well Natural Polyatomic above 1%. The formation rates for barium, strontium and Isotope abundance (%) interference Resolution* cerium, measured in standard solutions, are shown in Table 3. 42Ca+ 0.647 40Ar2H+ 2350 Even by very careful optimisation of the argon gas flows, it 40Ar1H2+ 2165 was dicult to obtain formation rates lower than 3% for 43Ca+ 0.135 — — strontium, for which the doubly charged ions interfere with 44Ca+ 2.086 12C16O2+ 1281 the analysis of calcium isotopes.Under these conditions the 28Si16O+ 2688 calcium signals were reduced by approximately 30%, so these * The resolution, calculated as m/Dm, needed to separate the poly- gas flow settings were not used for the measurement of calcium atomic ion from the analyte signal. isotopic ratios. The reason for this higher formation rate is unknown, but it might be due to the presence of a secondary signals can be resolved from the polyatomic interferences using discharge behind the skimmer cone.a resolution setting of 2700. However, since baseline separation Strontium has four isotopes at 84, 86, 87 and 88 u and is required in order to achieve the best possible precision, a doubly charged strontium ions interfere with the 42Ca+, 43Ca+ resolution setting of 4000 was used. The mass spectra for the and 44Ca+ signals in urine samples.The strontium concencalcium isotopes in urine are shown in Fig. 1. The most severe tration in human urine is approximately 0.2 mg l-1.24 The polyatomic interferences are 40Ar1H2+ and 40Ar2H+ on 42Ca+. peaks from doubly charged strontium ions cannot be resolved It can be seen from Fig. 1 that the signal from these inter- from the calcium peaks even at high resolution (resolution= ferences give rise to a signal equal to that of approximately 10000) and must therefore be corrected for mathematically. 1.5 mg l-1 calcium in urine.Since the 2H to 1H ratio in nature Like strontium, rubidium also has an isotope at 87 u, but since is approximately 156500 it might be suggested that the inter- the second IP of rubidium is very high (27.5 eV) it does not ference from 40Ar2H+ is insignificant. The actual ratio between form doubly charged ions in the ICP. As seen in Table 3 we the two interferences cannot be measured, since they are very did not observe any doubly charged rubidium ions when close in mass and cannot be separated even using a resolution analysing standard solutions.The signal from doubly charged setting of 10 000 (a resolution setting of approximately 25 000 strontium at m/z=43.5 can then be used directly to correct for is needed). But the 2H+ to 1H2+ ratio can be measured at overlaps from Sr2+ on 42Ca+, 43Ca+ and 44Ca+. m/z=2. This ratio was found to be approximately 151 in both standard solutions and urine samples, i.e., the formation rate Non-spectral Interferences of 1H2+ from 1H atoms/ions in the plasma and/or interface region must be low given that the 1H concentration is relatively HR-ICP-MS instruments are prone to interference from the high in the plasma.Suggesting that 2H+ and 1H2+ combine sample matrix in the same way as Q-ICP-MS instruments. with 40Ar+ at the same rate, the 40Ar1H2+ to 40Ar2H+ ratio When calcium isotopic ratios are measured in urine, matrix would be approximately 151.Following this argument it is interferences have to be considered since urine contain high likely that both polyatomic ions contribute significantly to the levels of the major elements, (especially sodium) in concencombined peak. trations of approximately 2200 mg l-1.24 The average total calcium concentration in the urine sampled in the present study was 135 mg l-1. In order to investigate the eect of a Doubly Charged Ions high sodium content on the calcium isotopic ratio measurements, these were measured in 3 mg l-1 standard solutions Elements with a low second ionization potential (IP) can form containing 0, 20, 40, 60, 80 and 100 mg l-1 of sodium.These doubly charged ions in the ICP ion source. In Q-ICP-MS, the concentrations cover the concentration range in which sodium formation rate is generally <1% under normal operation is found in the sample solutions, since the urine was diluted 50 times before analysis.The measured 44Ca+543Ca+ ratios and the intensity of the 44Ca+ signal are shown in Fig. 2. As expected, the calcium sensitivity decreased with increasing sodium concentration. In the presence of 100 mg l-1 of sodium the calcium signal was depressed by 20%. The 44Ca+543Ca+ ratio was not aected by the increasing sodium concentration. It can therefore be concluded that matrix interferences from major elements do not aect the measurement of calcium isotopic ratios.The only eect is that signal intensities are depressed, which in theory can degrade the precision of the measurements owing to poorer counting statistics. Similar results were found for the 42Ca+543Ca+ ratio. Precision of the Isotopic Measurements The precision of measurements of isotopic ratios by ICP-MS is limited by counting statistics (Poisson statistics) but other Table 3 The formation rate of doubly charged ions found under standard conditions of the PlasmaTrace2 in standard solutions Element Second IP/eV M2+5M+ (%) Ba 10.0 14.9 Ce 10.9 6.5 Sr 11.0 6.5 Fig. 1 Mass spectra of (a) 42Ca+; (b) 43Ca+; and (c) 44Ca+ in a urine Rb 27.5 0 sample containing approximately 2 mg l-1 of calcium. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 921Table 4 The long-term RSD of the 42Ca+543Ca+ and 44Ca+543Ca+ ratios in human urine measured over 8 days Days 42Ca+543Ca+ 44Ca+543Ca+ 1 4.7734 15.4598 1 4.7500 15.4889 2 4.8140 15.5268 2 4.7579 15.3960 3 4.7966 15.5383 3 4.7754 15.3975 3 4.7907 15.5054 4 4.7806 15.4824 4 4.7628 15.5184 5 4.8256 15.5486 5 4.7722 15.4495 6 4.7987 15.5469 6 4.8113 15.6925 7 4.7830 15.6515 Fig. 2 Eect of sodium on the isotopic ratio measurements. All 8 4.7779 15.5065 solutions contain 3 mg l-1 of calcium. The 44Ca+543Ca+ ratio is 8 4.8086 15.5112 shown as the mean±standard deviation (s), s=0.33%. The 44Ca+ intensity is set to 100% in the standard solution with no sodium added.Average 4.7862 15.5138 Standard deviation 0.0216 0.0775 Relative standard deviation (%) 0.45 0.50 sources of error, like plasma flicker noise, noise from the Expected ratio 4.7926 15.4518 peristaltic pump during sample uptake and changes in the nebulization, ionization and extraction processes also contribute to the overall method precision. Precisions of 2–3 times of that imposed by counting statistics are most often found using ICP-MS.25,26 The minimum error in the ion counting process calculations, e.g., the calculation of human calcium absorption rates in nutritional studies.with the Plasmatrace2 is, as mentioned above, set by Poisson statistics. The RSD of a measurement is 1/ÓN, where N is the A precision of 0.33% corresponds with that of <1% found for the measurement of the 44Ca+543Ca+ and the 48Ca+543Ca+ total number of ions observed. The theoretical RSD of an isotopic ratio is therefore Ó1/N1+1/N2, if the precision is ratios following a sample preparation step in which calcium was isolated from the sample matrix by precipitation as limited only by counting statistics, where N1 and N2 are the number of ions observed for isotopes 1 and 2, respectively. oxalate.16,17 Comparable precisions (0.2–0.5%) have been obtained by FABMS for the measurement of the 42Ca+540Ca+ When the two isotopes are not equally abundant, longer measuring time should be spent on the less abundant isotopes.and 44Ca+540Ca+.7 As in HR-ICP-MS analysis, the sample preparation in FABMS analysis is very simple and plasma and The optimum ratio of counting times are t1/t2=ÓA2/A1, where t1, A1 and t2, A2 are the dwell time and natural abundance for urine can be applied directly if the elemental concentration is suciently high in the samples.HR-ICP-MS and FABMS isotopes 1 and 2, respectively. This equation was used to optimise the dwell times of 42Ca+, 43Ca+ and 44Ca+. methods show similar figures of merit with regard to sample preparation, need of interference corrections and precision.The theoretical precision function (calculated from Poisson statistics) and the experimentally measured RSD values of the One distinct advantage of HR-ICP-MS is that this technique has a higher sensitivity than FABMS. TIMS, on the other 44Ca+543Ca+ ratio are shown in Fig. 3. The short-term RSD (10 replicates in one day) was found to be in the range of hand, shows better precision for the measurement of calcium isotopic ratios, often >0.2%.7 Yet, the TIMS technique 0.3–0.4% (0.33% on average) in urine samples. The RSD found is approximately 3 times that of the theoretical RSD, requires a full separation of the analyte from the sample matrix.This sample preparation is demanding and often very time- hence there is a considerable contribution from other sources than counting statistics to the RSD. A similar RSD of 0.41% consuming and consequently TIMS leads to a considerably lower sample throughput than HR-ICP-MS methods.No was found for the 42Ca+543Ca+ ratio. The long-term RSDs (16 replicates in 8 days) in human urine are approximately other reports on the measurement of calcium isotopic ratios by HR-ICP-MS have been published previously, but in a 0.45–0.50%, as shown in Table 4 where also the day to day results are shown. Short-term RSDs of 0.33 and 0.41% corre- similar study the 63Cu+565Cu+ ratio in human serum and Antarctic sediments has been measured with RSDs in the spond to standard deviations of the mean of 0.10% and 0.13% (s/Ón, n=10, where n is the number of replicates).Thus, the range of 0.3–0.6% using a resolution setting of 3000.19 That study and the present report indicate that the HR-ICP-MS overall precision of the method can be improved by replication of single measurements. The RSD will be improved by a factor technique provides results with a precision better than 0.6%, for the measurement of isotopic ratios of elements in the lower of 1/Ón.Even though the overall analysis time will increase, if all samples have to be measured in replicates or triplicates, it mass region (<80 u), where the analyte signals often are interfered with by polyatomic ions using Q-ICP-MS. might be useful if high precision data are used for further Fig. 3 Short-term precision for the 44Ca+543Ca+ ratio in human urine: (—) theoretical precision induced by counting statistics; (D) precision measured for 10 replicates in one day for a human urine sample. 922 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12Table 5 Isotopic ratios measured in urine samples before and after enrichment with 42Ca and 44Ca: isotopic ratios are given as the measured value ±2 s, where s is the method standard deviation of 0.33% and 0.41% for the 44Ca+543Ca+ and 42Ca+543Ca+ ratio, respectively. A1 and A2 were given the low calcium isotope doses, B1 and B2 the high doses Volunteer 42Ca+543Ca+ (before) 42Ca+543Ca+ (after) 44Ca+543Ca+ (before) 44Ca+543Ca+ (after) A1 4.7966±0.0393 4.9837±0.0409 15.5383±0.1026 15.6955±0.1036 A2 4.7806±0.0392 4.9441±0.0405 15.4824±0.1022 15.8035±0.1043 B1 4.7722±0.0391 5.0915±0.0418 15.4495±0.1020 15.7845±0.1042 B2 4.7987±0.0393 5.0357±0.0413 15.5469±0.1026 15.7548±0.1040 Kastenmayer, P., Luten, J.B., and McGaw, B. A., Analyst, 1994, Analysis of Urine Samples Enriched in 44Ca and 42Ca 119, 2491. In the double-label stable isotopes experiment, volunteers were 8 Yergey, A.L., Vieira, N. E., and Hansen, J. W., Anal. Chem., 1980, 52, 1811. given 8.9 or 16.7 mg 44Ca orally with a milk-based meal 9 Yergey, A. L., Vieira, N. E., and Covell, D. G., Biomed. Environ. containing 400 mg calcium. Subsequently 1.7 or 3.4 mg 42Ca Mass Spectrom., 1987, 14, 603. was injected intravenously. Urine samples were collected before 10 Price, R. I., Kent, G. N., Rosman, K. J. B., Gutteridge, D. H., and after isotope administration. Table 5 shows the isotopic Reeve, J., Allen, J.P., Stuckey, B. G. A., Smith, M., Guelfi, G., ratios measured in the base and enriched urine samples from Hickling, C. J., and Blakeman, S. L., Biomed. Environ. Mass Spectrom., 1990, 19, 353. 4 of the volunteers [two that were given the low doses (A1 11 Sandstro�m, B., Fairweather-Tait, S. J., Hurrel, R., and Van and A2) and two that were given the high doses (B1 and B2)]. Dokkum, W., Nutr. Res.Rev., 1993, 3, 71. The 42Ca+543Ca+ ratios in the enriched urine samples were 12 Smith, D. L., Anal. Chem., 1983, 55, 2391. significantly larger than that of the base urine for volunteers 13 Roy, S., Gillen, G., Conway, W. S., Watada, A. E., and Wergin, given both the low and high calcium doses, whereas the W. P., Protoplasma, 1995, 189, 163. 14 Nicolussi, G. K., Pellin, M. J., Calaway, W. F., Lewis, R. S., 44Ca+543Ca+ ratio in the urine samples from volunteers given Davis, A.M., Amari, S., and Clayton R. N., Anal. Chem., 1997, the low 44Ca doses are not significantly larger than that of the 69, 1140. base urine. The conclusion is therefore that in future experi- 15 Koumenis, I. L., Vestal, M. L., Yergey, A. L., Abrams, S., Deming, ments volunteers should be given the low 42Ca dose and the S. N., and Hutchens, T. W., Anal. Chem., 1995, 67, 4557. high 44Ca dose. Alternatively, all measurements should be 16 Luten, J. B., Muys, T., and Dokkum, W., Proceedings of Bioavailability ’93, Part II, ed.Schlemmer, U., Bundes- repeated (e.g., replicates or triplicates) and the mean used for forschungsanstalt fu� r Erna�hrung, Karlsruhe, 1993, pp. 161–168. further calculations. This would reduce the overall standard 17 Van Dokkum, W., De La Gueronniere, V., Schaafsma, G., Bouley, deviation by a factor of Ón. In this case, the low dose of 44Ca C., Luten, J., and Latge, C., Brit. J. Nutr., 1996, 75, 893. would probably be sucient to create a statistical significant 18 Vanheacke, F., Moens, L., Dams, R., and Taylor, P., Anal.Chem., enrichment of the urine samples. Even though each sample 1996, 68, 567. 19 Vanhaecke, F., Moens, L., Dams, R., Papadakis, I., and Taylor, should be analysed more than once this would reduce the total P., Anal. Chem., 1997, 69, 268. cost of future experiments, since stable isotopes are very 20 Van Dokkum, W., Fairwether-Tait, S. J., Hurrell, R., and expensive. Sandstro�m, B., in Stable Isotopes in Human Nutrition, Inorganic Nutrient Metabolism, ed. Mellon, F. A., Sandstro�m, B., Academic Press. London, 1996. pp. 23–42. 21 Koirtyohann, S. R., Spectrochim. Acta, Part B., 1994, 49, 1305. REFERENCES 22 Price Russ III, G., in Application of Inductively Coupled Plasma Mass Spectrometry, ed. Date, A. R., Gray, A. L., Blackie. Glasgow 1 Dilena, B. A., Larsson, L., and O� hman, S., in Handbook on metals and London, 1989. pp. 90–110. in clinical and analytical chemistry, ed. Seiler, H. G., Sigel, A., and 23 Jarvis, K. E., Gray, A. L., and Houk, R. S., in Handbook of Sigel, H., Marcel Dekker, Inc, New York, 1994, pp. 299–310. Inductively Coupled Plasma Mass Spectrometry, ed. Jarvis, K. E., 2 Nordiske Na�ringsrekommendationer (1996). Nord 1996528 © Gray, A. L., and Houk, R. S., Blackie, Glasgow and London, Nordiske Ministerra°det, Ko� penhamn 1996. ISBN 92 9120 930 9 1992, pp. 143–145. (Swedish). 24 Caroli, S., Alimonti, A., Coni, F., Petrucci, F., Senofonte, O., and 3 Lee, W. T., Leung, S. S., Fairweather-Tait, S. J., Leung, D. M., Violante, N., Crit. Rev. Anal. Chem., 1994, 24, 363. Tsang, H. S., Eagles, J., Fox, T., Wang, S. H., Xu, Y. C., and 25 Ting, B. T. G., and Janghorbani, M., J. Anal. At. Spectrom., 1988, Zeng, W. P., Brit. J. Nutr., 1994, 72, 883. 3, 325. 4 Miller, D. D., Adv. Food Nutr. Res., 1989, 103. 26 Begley, I. S., and Sharp, B. L., J. Anal. At. Spectrom., 1994, 9, 171. 5 Shipp, C. C., Maletskos, C. J., and Dawson-Hughes, B., Calcif. T issue Int., 1987, 41, 307. Paper 7/04079B 6 Fairweather-Tait, S. J., Johnson, A., Eagles, J., Ganatra, S., Received June 11, 1997 Kennedy, H., and Gurr, M. I., Brit. J. Nutr., 1989, 62, 379. 7 Crews, H. M., Ducros, V., Eagles, J. E., Mellon, F. A., Accepted July 22, 1997 Journal of Analytical Atomic Spectrometry, September 1