JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 109 Precise Determination of Iron Isotope Ratios in Whole Blood Using Inductively Coupled Plasma Mass Spectrometry* Invited Lecture Paul G. Whittaker and Jon F. R. Barrett University Department of Obstetrics Princess Mary Maternity Hospital Ne wcastle upon Tyne NE2 380 UK John G. Williamst NERC ICP-MS Facility Department of Geology Royal Holloway and Bedford New College Egham Surrey TW200EX UK The feasibility of measuring Fe absorption by incorporation of stable Fe isotopes into red blood cells has been assessed. The clinical protocol necessitated giving 10 mg of 57Fe orally and 0.5 mg of 58Fe intravenously with only two blood samples needed one basal and one after 14 d when most of the tracer is in the circulating red blood cells.Iron absorption was determined by comparison of 57Fe:56Fe and 58Fe:56Fe enrichments which required great confidence in the reliability of 56Fe measurements. Estimation of required precision by theoretical calculations suggested that a relative standard deviation of (0.9% would be required to detect the enrichment in isotope ratios which necessitated finding the optimum Fe signal in relation to the interfering polyatomic species present in both aqueous standards and whole blood samples diluted in buffer. It was found that Fe solutions of 10 ppm or greater routinely gave a precision of (0.4% for 57Fe:56Fe and <O.6% for 58Fe:56Fe approaching counting statistics. Assessment of dead time correction ensured that the concentration of Fe had a negligible effect on the isotope ratios and bias correction by running standards ensured comparability within and across assay occasions correcting for minor variations in blank subtraction for the less abundant isotopes.The use of a range of enriched aqueous and spiked whole blood samples showed that measured and calculated abundances correlated with a slope of unity. Blood samples from two subjects showed that after incorporation of enriched isotopes isotope ratios of 57Fe:56Fe and s8Fe:56Fe were clearly distinguishable (standard deviation >9) from the baseline. Inductively coupled plasma mass spectrometry with conventional aqueous sample introduction can be optimized to give precise measurement of all Fe isotope ratios in whole blood permitting clinical studies of Fe absorption. Keywords lnductively coupled plasma mass spectrometry; iron; blood; isotope ratios; absorption Stable isotopes are increasingly employed as tracers for studies of mineral metabolism in man.Methods have been developed for investigating elements such as Zn and Cu,' Mg2 and Ca.3 They have been applied to studies of Fe absorption in men,' n~n-pregnant'.~ and pregnant6 women and pre-term infant^.^ In recent years isolation of the factors that influence Fe absorption by individuals believed to be at the greatest risk of Fe deficiency (children and women of child-bearing age) have been hindered by ethical considerations that prohibit the use of radioisotopes. However the availability of stable isotope tracer methods should greatly facilitate research with these vulnerable groups.In a study of absorption of Fe during human pregnancy,8 inductively coupled plasma mass spectrometry (ICP-MS) was used with sample introduction by electrothermal vaporization (ETV) to determine Fe isotope ratios in blood serum without prior sample preparation. The ratios s4Fe:56Fe and 57Fe:56Fe were determined in serum taken from non-pregnant women following oral and intravenous administration of enriched 54FeS04 and 57FeS04 respec- tively. Sample introduction by ETV significantly reduces the levels of certain polyatomic ions in particular those associated with the interference of 54Fe ('OAr14N) and 56Fe (40Ar160). In addition ETV is an ideal method of sample introduction requiring minimal sample volume only 5 pl of sample are required for each analysis.A logical progression from the work of Whittaker et a1.8 is * Presented in part at the XXVII Colloquium Spectroscopicum Internationale (CSI) Bergen Norway June 9- 14 199 1 and the 4th Surrey Conference on Plasma Source Mass Spectrometry Guild- ford UK July 15-18 1991. t Invited Lecturer. to assess whether Fe absorption can also be measured by incorporation of stable Fe isotopes into red blood cells (RBC). A few recent studies have used ingestion of a single stable isotope such as 58Fe779 or 54Fe10 to determine Fe availability. However although the patient protocols were simple requiring only two samples of erythrocytes (one basal and one 10-14 d after ingestion when incorporation of a tracer Fe into RBC is complete) single isotope erythrocyte analysis requires an assumption to be made about the level of oral Fe incorporation into erythrocytes this level in fact alters from between 86'O and 93%'' in non- pregnant individuals to 65% in subjects during late preg- nancy.6 The double isotope method'* overcomes the need for these assumptions.It is based on the simultaneous adminis- tration of two isotopes one by the intravenous route and the other by the oral route. This method provides greater accuracy since the comparison of oral with intravenously administered (absorbed 100%) tracer allows compensation to be made for redistribution of the tracer in the body. The RBC method used extensively with two radioactive tracers has been well quantified against whole blood counting. I 3 The double isotope approach is used in the present study to assess Fe absorption.This procedure uses the two least abundant stable Fe isotopes and has potential for safe serial studies both during pregnancy and in the newborn requiring only two blood samples. Iron absorption can be determined by comparison of 57Fe:56Fe and 58Fe:56Fe isotope ratio enrichments. Therefore great confidence in the reliability of 56Fe measurements is required. By using ETV-ICP-MS it was possible to determine Fe isotope ratios in the small volumes of serum available ( I ml or less) containing approximately 1 pg ml-l of Fe by reducing to background levels the polyatomic species that occur at 54 56 and 57 mlz. In the present study Fe concentration and110 JOURNAL OF ANAL,XTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL.7 sample volume were not limitations. The feasibility of using a conventional nebulizer/spray chamber for liquid sample introduction into an ICP-MS instrument for the determina- tion of 57Fe:56Fe and 58Fe:56Fe ratios in the presence of interfering polyatomic species in both simple aqueous standards and whole blood samples is described. Theory Estimation of Required Precision for Dual Isotope Studies of Fe Absorption A normal adult non-pregnant female has a total circulating Fe mass of about 1.56 g; 92 mg as 54Fe 35 mg as 57Fe and 5 mg as 58Fe. Enhancement of these isotopes is limited by the cost of the enriched stable isotopes (54Fe $5 per mg 57Fe f 15 per mg and 58Fe f 100 per mg approximately) and also by the desire to give physiological doses of Fe comparable to dietary intake (about 12 mg per day).The natural abundance of 54Fe is 2.6 times that of 57Fe which means that 2.6 times as much 54Fe would be required to yield the same enrichment as using 57Fe. While the cost benefits are marginal the use of 54Fe would require too much extra Fe in the daily intake which cannot itself be reduced to less than about 5 mg per day from other dietary sources. Conversely the use of 58Fe for oral administration would be very expensive at least in the adult. The theoretical calculations are based on giving the patient an oral dose of 10 mg of 57Fe (average non-pregnant rate of absorption is and an intravenous dose of 500 pg of 58Fe (absorption 100%) and assuming that 93% of both tracers are in the RBC after 2 weeks.'' With this addition the basal 57Fe:56Fe ratio of 0.02389 should change to 0.02454 an enrichment of 2.7% and the basal 58Fe:56Fe ratio of 0.003606 should change to 0.003931 an enrich- ment of 9%.The required relative standard deviation (RSD) for detection [>3 standard deviations (SD) from basal] of these enrichments is <0.9% for 57Fe:56Fe and (3% for 58Fe:56Fe. If an RSD of 0.5% is obtained then the minimum detectable enrichment is 1.5% i.e. 0.02425 for 57Fe:56Fe and 0.003660 for 58Fe:56Fe. This is equivalent to 0.53 mg of 57Fe absorbed and incorporated or a minimum detectable absorption of 5%. Estimation of Fe Signal Required to Obtain the Necessary Precision In order that isotope ratios of Fe can be determined with optimum precision in the presence of polyatomic ions the signal from an isotope suffering from interference has to be significantly greater than the signal of the underlying polyatomic species. In order to obtain such a differential in a conventional ICP-MS system the polyatomic signal would have to be reduced with respect to a given isotope signal or the isotope signal would have to be large with respect to a given polyatomic signal or a combination of both.Fig. 1 shows the relationship between the integrated absorbance of 56Fe and theoretical counting errors of 57Fe:56Fe and 58Fe:56Fe ratios for natural isotopic abun- dance of Fe assuming no polyatomic interferences on the Fe isotopes. In order that counting errors of better than 0.5% could be obtained for both isotope ratios an integrated absorbance of over 1 x lo7 s-l above any background (polyatomic) peak would have to be obtained. The ion count rate can be increased to the required level simply by introducing Fe solutions of a sufficiently high concentra- tion.However if the background signal were large it would be impossible to count an isotope signal above this level as the ion detection system would become saturated and incapable of counting the total ion flux. Optimizing the 0 10 20 Integrated absorbance of 56Fe/106 s Fig. 1 Relationship between the integrated absorbance of 56Fe and the theoretical isotope ratio counting error; A 58Fe:56Fe; and B s7Fe 56Fe ICP-MS system according to the method of Gray and Mlilliams14 ensures that the polyatomic ion at 56 m/z has an equivalent concentration of t 2 0 ng ml-l i.e. the signal from the polyatomic ion is at a low level such that large 56Fe isotope signals can be counted without causing the detec- tion system to become saturated.To summarize in order to obtain Fe isotope ratios with a precision of <0.5% the ICP-MS instrument has to be optimized such that the 40Ar160 signal is reduced to a minimurn. In addition the concentration of Fe in the saLmples has to be sufficiently high to produce the signal necessary to obtain the required precision without saturat- ing the detection system. Experimental Inistrumentation A PlasmaQuad PQ2 -b (VG Elemental Winsford Cheshire UK) inductively coupled plasma mass spectrometer was used for these studies. Details of the instrumentation are given in Table 1. The analysis time for five blanks and ten sample replicates was 40 min.Preparation of Blood Samples and Standards Aqueous solutions of whole blood (1 +24) (which were colllected in a heparinized tube) were prepared according to the following method (modified from the method of Delves and Campbell15). To a 50 ml calibrated flask were added 10 ml of doubly distilled de-ionized water 2 ml of a chemical modifier (0.14 mol dm-3 ammonia solution 0.003 rnd dm-3 disodium dihydrate ethylenediaminetetracetate and 0.029 mol dm-3 ammonium dihydrogen phosphate in water) 2 ml of whole blood 10 ml of Triton X- 100 (5% v/v in water) solution and doubly distilled de-ionized water to vdume. Iron concentrations were generally between 10 and 201 pg ml-l. Blanks and Fe standards (Sigma Poole Dorset UIK) were also prepared using this method.Enriched 57Fe and 58Fe standards were obtained from the UIK Atomic Energy Authority Harwell and Techsnabex- port London UK and made up as iron(@ sulfate by the pharmacy at Northwick Park Hospital (Harrow UK). A miISS analysis is shown in Table 2. Known amounts of these enriched standards were added to aqueous solutions of normal whole blood to give blood solutions of known enrichment.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 111 0.025 0) LL 3% LL u) 0.024 Table 1 ICP-MS instrument details and operating conditions - - Nebulizer Spray chamber Sampling cone Skimmer cone Operating conditions- Forward r.f. power/W Reflected power/W Coolant gas flow/l min-I Auxiliary gas flow11 min-' Nebulizer gas flow Sample solution pump ratelm1 min-l Mass scan range No.of sweeps No. of channels Dwell time/,us Total scan time/s (mass flow controlled)/l min-' 0.0036 0.0035 0) LL 0.0034 e 0.0033 0.0032 De Galan V-groove Single pass in-house design Nickel 1 mm orifice Nickel 0.7 mm orifice - - - - - - 1300 (5 14 0.5 0.75 0.8 50.94-64.92 1500 1024 80 123 Table 2 Mass analysis of enriched isotope preparations At.-% Enriched isotope 54 56 57 58 57FeS04 0 3.00 95.10 1.90 57FeS04 0 0.57 95.93 3.50 58FeS04 0 0.21 6.56 93.20 Clinical Protocol Two normal women attended the research unit following an overnight fast and 10 ml of blood were taken into a lithium heparin tube for determination of basal isotope ratios. Then 250 pg of 58FeS04 were given by intravenous injection followed by 5 mg of 57FeS04 administered orally with 50 ml of fresh orange juice.No food tea or coffee was allowed for 2 h. The following morning the protocol was repeated. After 14 d a 10 ml sample of blood was taken from which the enriched isotope ratios were measured. Results Counting error is a major contributor to precision and it varies inversely with the square root of the number of ions collected. Theoretical calculations of counting errors can be made for both 57Fe:56Fe and 5sFe:56Fe ratios and are shown in Fig. 1. An integrated absorbance of 1 x 1 O7 s-l is required to achieve a counting error of 0.5% for 58Fe:56Fe. In practice this was achieved with about 10 pg ml-l of the Fe standard solution and the calibration line for Fe was found to be linear between 1 and 20 pg ml-l.Eight repeated runs of groups of five replicates of 10 pg ml-l Fe standard samples showed that precision for 57Fe:56Fe varied between 0.13 and 0.29% with an average of 0.21% close to the theoretical counting error of 0.19%. Precision for 58Fe:56Fe varied between 0.30 and 0.90% with an average of 0.64% close to the theoretical counting error of 0.51%. Figs. 2 and 3 show that appropriate dead time corrections of 20-25 ns could be made so that both isotope ratios appeared independent of concentration. 0.026 * 0.023 1 1 I I I 5 10 15 20 25 Dead time/ns Fig. 2 Effect of concentration and dead time on the 57Fe:56Fe ratio A 5; B 10; and C 15 ,ug ml-* of Fe. Error is 3SD 5 10 15 20 ' 25 Dead time/ns Fig. 3 Effect of concentration and dead time on the 58Fe:56Fe ratio A 5; B 10; and C 15 ,ug ml-l of Fe.Error is 3SD The effect of increasing Fe concentration on the precision of both isotope ratios (Table 3) showed that concentrations of greater than 5 pg ml-l of Fe were optimum and equally precise and accurate isotope ratios could be obtained for diluted blood and aqueous standards.112 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 Table 3 Effect of concentration on precision of isotope ratios n= 15 Fe concentration/ Standard- pg ml-I 57Fe:56Fe RSD (O/O) 58Fe:56Fe RSD (%) 1 0.023 1 1 0.79 0.003 2 2 1.90 5 0.02303 0.34 0.00323 0.72 10 0.02322 0.35 0.0032 1 0.34 15 0.0233 1 0.2 1 0.00325 0.35 Blood- 9 0.02320 0.32 0.00322 0.60 18 0.02320 0.20 0.00322 0.38 Measurement of solutions of known isotope ratios showed that results were consistently accurate for all isotope ratios over the whole range of enrichment (Table 4) with correlation of measured and expected abundances giving a slope of unity.In practice minor adjustments in bias were made for unknowns by comparing standards with accepted natural abundances.16 Thus the average bias for 56Fe was 0.19'0 for 57Fe 4.l0h and for 58Fe -2.9% all significant and reflecting minor differences in the accuracy of blank subtraction for the less abundant isotopes. For the isotope ratios this translated into an average of 3.8% for 57Fe:56Fe and - 3.1 Oh for 5*Fe:56Fe. Table 5 gives results for sample 1 (measured 15 times) which was an aliquot of a blood sample taken before the test began (Le. basal). Sample 2 was an aliquot of a blood saimple from the same patient two weeks after ingestion of 10 mg of 57Fe and injection of 500 pg of 58Fe.An Fe standard was then measured before sample 3 (basal) and saimple 4 (enriched) from a second patient were measured. Einrichment of both isotope ratios were clearly significant and therefore estimates of Fe absorption could be derived; about 20% of the given oral dose in these patients. The method of calculating the final absorption figure in a larger group of subjects will be the topic of a further paper. Table 4 Measurement of solutions of known abundance 57Fe abundance s8Fe abundance Sample Enriched standard 1 Enriched standard 2 Enriched standard 3 Enriched blood 1 Enriched blood 2 Basal standard Regression equation* *x=expected y= measured ratio. _ _ _ ~ ~ Expected Measured 95.1 95.13 95.1 95 3 6 95.1 94 78 95.93 95..69 95.93 95.68 6.56 6.60 6.56 6,34 7.01 6,82 2.82 2.84 2.19 2.23 ' Expected Measured 1.9 1.87 1.9 1.82 1.9 1.84 3.5 3.62 3.5 3.65 93.23 93.22 93.23 93.20 0.40 0.40 0.34 0.34 0.33 0.32 y=0.0102+ 1.002x Table 5 Iron isotope ratios in blood from two test patients n = 15 Detection Detection Sample type 57Fe:56Fe RSD (%) limit* 58Fe:56Fe RSD (Oh) limit* 1 Basal 0.02379 0.24 0.023915 0.00353 0.35 0.00 3 5 7 2 Enriched 0.02496 0.13 - 0.00398 0.40 standardt 0.02390 0.23 - 0.00360 0.78 3 Basal 0.0238 1 0.16 0.0239 I 0.00359 0.48 0.00364 4 Enriched 0.02529 0.15 - 0.00409 0.34 * Basal mean+ 3SD.t Standard had 5 replicates.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 113 Discussion The sample preparation method was simple requiring only dilution of whole blood in a buffer.A previous study of Fe in blood by ICP-MS7 used wet ashing with HNOJ before dilution to a similar concentration as that used in the present method. Because it is hoped that a clinical method applicable to human pregnancy can be developed assumptions of isotope incorporation into RBC could not be relied upon and therefore a protocol that compensated for tracer redistribu- tion was required i.e. the use of two isotopes. Janghorbani et aL7 used only one isotope 58Fe by the oral route and monitored 58Fe:57Fe to show that a precision below 1% could be achieved. It has been shown here that a precision of less than 0.6% for both 57Fe:56Fe and 58Fe:56Fe can be attained when Fe concentration and ion intensity are optimized.The high concentration of Fe in whole blood means that only 1 ml of whole blood is required (diluted 1 +24) for 10 replicate analyses at 2 min each. The long measurement times can be accommodated by use of an automatic sampler. Information about the accuracy of the method is also important. It has been shown here that the independence of the isotope ratios from Fe concentration is determined by the use of appropriate dead time correction which compen- sates for counting losses of 56Fe at the upper end of the concentration range. In addition there is close agreement between measured and calculated abundances over the entire range available. Minor correction of bias when measuring clinical samples is achieved by running Fe standards every third sample and using accepted natural abundances since a reference material with certified iso- tope ratios or abundances does not exist for Fe in blood. Thus one can overcome day to day variation in measured natural isotope ratios and thereby use the basal isotope ratio of each subject rather than an average of all subjects as used by Janghorbani et aL7 The difference in Fe isotope ratios before and after incorporation of isotopes into RBC are small but signifi- cant.The enriched samples have an SD39 from the basal ratios. It is therefore clear that the use of 57Fe administered orally gives sufficient RBC enrichment to permit quantita- tive studies of Fe absorption in adults. The reduced blood volume and body Fe stores in children and infants would mean that the protocol could be applied to these groups with resulting greater enrichment. Alternatively the use of 57Fe and 58Fe could be applied to the simultaneous within- subject comparison of food and aqueous Fe absorption.In conclusion ICP-MS with conventional aqueous sam- ple introduction can be optimized to give precise measure- ments of all Fe isotope ratios in whole blood permitting clinical studies of Fe absorption. The authors are grateful to the Royal Society for financial support of the project. The ICP-MS facility is supported by the Natural Environment Research Council. References I King J. C. Raynolds W. L. and Margen S. Am. J. Clin. Nutr. 1978 31 1198. 2 Schuette S. Vereault D. Ting B. T. G. and Janghorbani M. Analyst 1988 113 1837. 3 Fairweather-Tait S. J. Johnson A. Eagles J. Ganatra S. Kennedy H. and Gurr M. I. Br. J. Nutr. 1989 62 379. 4 Janghorbani M. Ting B. T. G. and Young V. R. J. Nutr. 1980 110 2190. 5 Cantone M. C. Molho N. Pirola L. Gambarini G. Hansen C. Roth P. and Werner E. Med. Phys. 1988 15 862. 6 Dyer N. C. and Brill A. B. in Nuclear Activation Techniques in the Life Sciences I.A.E.E. Vienna 1972 pp. 469-477. 7 Janghorbani M. Ting B. T. G. and Fomon S. J. Am. J. Hematol. 1986 21 277. 8 Whittaker P. G. Lind T. Williams J. G. and Gray A. L. Analyst 1989 114 675. 9 Fairweather-Tait S. J. and Minski M. J. Br. J. Nutr. 1986 55 279. 10 Lehmann W. D. Fischer R. and Heinrich H. C. Anal. Biochem. 1988 172 151. 1 1 Larsen L. and Milman N. Acta Med. Scand. 1975,198,271. 12 Saylor L. and Finch C. A. Am. J. Physiol. 1953 172 372. 13 Werner E. Roth P. Hansen C. and Kaltwasser J. P. in Structure and Function of Iron Storage and Transport Proteins ed. Urshizaki I. Elsevier Amsterdam 1983 pp. 403-408. 14 Gray A. L. and Williams J. G. J. Anal. At. Spectrorn. 1987 2 599. 15 Delves H. T. and Campbell M. J. J. Anal. At. Specctrorn. 1988 3 343. 16 Emsley J. The Elements Oxford University Press Oxford 1989. Paper 1/05006K Received September 30 I991 Accepted November 13 1991