Measurement Techniques for Tracer Kinetic Studies with Stable Isotopes of Zirconium M

Home , Isotopes of zirconium

Radiation Protection Dosimetry (2007), Vol. 127, No. 1–4, pp. 266–269 doi:10.1093/rpd/ncm347 Advance Access publication 17 October 2007

MEASUREMENT TECHNIQUES FOR TRACER KINETIC STUDIES WITH STABLE ISOTOPES OF ZIRCONIUM M. Greiter1,*, K. Abbas2, M. C. Cantone3, W. Carli4, A. Geisler1, U. Gerstmann1, A. Giussani1,3, R. Hertenberger4, U. Holzwarth2, O. Meisenberg1,V.Ho¨llriegl1, U. Oeh1, I. Veronese3 and H. G. Paretzke1 1GSF—National Research Center for Environment and Health, 85764 Neuherberg, Germany 2European Commission JRC, Institute for Health and Consumer Protection, 21020 Ispra (Va), Italy 3Universita` degli Studi di Milano, Dipartimento di Fisica and INFN, Sezione di Milano, 20133 Milano, Italy 4MLL—Maier-Leibnitz Laboratory, 85748 Garching, Germany Downloaded from Biokinetic models are used in radiation protection to assess internal radiation doses. Experiments with stable isotopes as tracers can be performed to obtain characteristic parameters of these models. Two methods for the measurement of zirconium isotopes in human biological samples are presented—thermal ionisation mass spectrometry (TIMS) and proton nuclear activation analysis (PNA). Descriptions include sample preparation, operating conditions, relative uncertainties and method detection limits as well as important properties of both methods. http://rpd.oxfordjournals.org/

INTRODUCTION (2) Any suitable method must be able to measure low amounts of the target substance in human Internal radiation doses are impractical to measure, ﬂuid samples, i.e. 1 ng per sample in the case therefore, biokinetic models are used to assess these of zirconium. In a tracer experiment, the tracer doses. In order to build such models, the distribution concentration should be as low as possible in of radionuclides in the human body must be measured order not to disturb the normal metabolism of directly or estimated from animal experiments and the naturally occurring substance. other sources of information. Since metabolism often (3) Sources of interference must be avoided or at

differs between species, the former option is generally at GSF Zentralbibliothek on January 4, 2017 least be controllable. preferred, but can be difficult to accomplish. (4) Methods should be readily available with an ade- One method capable of delivering data directly quate sample throughput. from human subjects, without exposing them to ionis- ing radiation, is the application of stable isotopes as The first criterion is satisfied by mass spectrometry tracers(1). Different isotopes of the same element are and by techniques based on nuclear reactions. chemically similar, i.e. stable isotopes can be used to The wide range of different techniques in these fields investigate the distribution of their unstable counter- gets narrowed by the addition of the second criterion, parts. For this purpose, solutions of enriched stable which limits the use of nuclear reactions to those with isotopes are administered to test persons, who donate high reaction cross sections for zirconium. blood and urine samples at scheduled times. These Two methods that satisfy all four criteria—proton samples are analysed for their element content and nuclear activation analysis and thermal ionisation isotopic distribution. If two tracers are administered mass spectrometry—are presented below. simultaneously, one orally (e.g. 96Zr) and one intrave- nously (e.g. 90Zr), information can be derived about the absorbed fraction from the gut, clearance from Proton nuclear activation analysis (PNA) plasma and urinary excretion from the body. Description This method was introduced in 1982 for a variety of METHODS elements(2) and was later adopted for zirconium(3). Requirements One of two setups is in use at JRC Ispra, Italy; the second setup was re-established recently at the Stable tracer experiments set high demands for Maier-Leibnitz-Laboratory (MLL) in Garching, measurement methods. Germany. The data presented in this work are from (1) The selected method must be capable of isotope ongoing measurements at the MLL. separation to distinguish between one or more Samples are prepared by tracers and natural element content in samples. (1) mixing defined amounts of urine or plasma with an internal standard (51V), followed by *Corresponding author: [email protected] incubation,

(2) partial pressure digestion with HNO3 and H2O2 range of 0.2 to 0.7 mm aluminium is not correlated (only urine samples), with detection limit. (3) drying and grinding, In the following measurement cycles, zirconium (4) pressing into tablets and isotopes were detected in 41 measurements per- (5) encasing tablets into sample holders with 0.2 to formed on 28 experimental samples. Double deter- 0.7 mm aluminium shielding. minations were made on different gamma spectrometers with different waiting and measure- Standard samples are prepared in a similar way, ment times. The measured amount of zirconium in with known amounts of the tracer solutions added. these samples ranges from 1.4 to 40.5 ng, with 1.4 to Up to 40 samples, including standard samples, are 38.5 ng 90Zr in urine and plasma samples and 3.8 to then mounted onto a disc. The disc is cooled to 40.5 ng 96Zr in plasma samples. The associated rela- 2208C during irradiation and rotated to ensure tive uncertainties are approximately lognormally dis- similar irradiation parameter values for all samples. tributed; characteristic values for both isotopes are Irradiation is performed with 20 MeV protons and shown in Table 2. The uncertainties are derived from Downloaded from a beam current of 1.7 mA for up to 48 h, but discon- the counting uncertainties of the experimental and tinued overnight. After eight hours of cooling, samples standard samples and an assumed 5% uncertainty in are removed from the disc, dismantled and transferred experimental sample mass. No signiﬁcant difference to gamma spectrometry. The amounts of 90Zr and in 90Zr relative uncertainty distribution is found 96Zr in experimental samples are calculated by com- between urine and plasma samples using the parison with both internal and external reference stan-

Kolmogorov–Smirnov test (test statistics D ¼ 0.27 http://rpd.oxfordjournals.org/ dards, taking into account the times of irradiation, with sample sizes of 8 and 11). delay until measurement and measurement itself. Irradiations are scheduled every two to three months to provide enough time for all measurements Discussion as well as for the decay of activated parts in the setup. The PNA method has other properties, which have With this method, two isotopes of zirconium an inﬂuence on its suitability for zirconium measure- can readily be measured, namely 90Zr and 96Zr, ments and stable tracer experiments. An essential whose characteristic data are listed in Table 1. advantage of PNA is the simple preparation

method. Especially for plasma samples, the risk of at GSF Zentralbibliothek on January 4, 2017 contamination or significant loss of sample material Results is low. Furthermore, addition of natural stable sub- A set of 25 reference plasma samples was activated stances to samples after irradiation does not change and analysed to optimise the procedure at the MLL. measurement results. Detection limits are calculated according to the defi- One critical issue for good results is measurement nition of Currie(4) from the background uncertainty planning. Samples with expected low amounts of at the above-mentioned peak energies. In the case of tracer must be measured at optimal times to ensure 90Zr, measured on the 141 keV line of 90Nb, the detection; other samples must be delayed or detection limit calculated from the reference samples measured quickly with a high background count is 1.8 ng per sample of 0.9 ml plasma. It is greatly rate. If numerous samples are activated, gamma influenced by both waiting time between irradiation spectrometric measurements must be spread over and measurement, which should be shorter than 4 d, multiple detectors, which introduces additional as well as by the ratio of waiting time over measur- sources of uncertainty. ing time, which is lesser than three for optimal Another issue is the interpretation of measurement results. The detection limit of 96Zr is 1.6 ng per results. Isotope concentrations per sample have to be sample of 0.9 ml plasma. Optimal conditions are a interpreted concerning the mixture of tracer(s) and waiting time of 30 to 50 d and a measuring time of natural zirconium content in the sample. Gamma 5 to 7 d. In both cases, shielding thickness in the Table 2. Relative uncertainties (%, coverage factor 1) of Table 1. Characteristic data of reactions used in PNA of zirconium isotopic amounts in experimental samples as zirconium. measured with PNA; GM 5 geometric mean, GSD 5 geometric standard deviation. Reaction Product Main gamma half-life energy (keV) 90Zr (n ¼ 19) 96Zr (n ¼ 22)

90Zr(p,n)90Nb 14.6 h 141, 1129 Range 15.7–63 10.2–39 96Zr(p,2n)95Nb 35.0 d 766 GM 33.7 16.0 51V(p,n)51Cr 27.7 d 320 GSD 1.43 1.44

267 M. GREITER ET AL. spectrometry allows the identification of decaying Results nuclei according to their characteristic energy emis- Measured ratios of zirconium isotopes in 107 exper- sion. Provided that the energy resolution of the spec- imental samples varied from 0.013 to 1. Setting at trometer is sufficient to rule out interferences from 1000 cps the minimum count rate level for the most other nuclei, different production reactions still have abundant isotope, the minimal count rate of the to be considered. 90Nb can also be produced from least abundant isotope is 13 cps. The maximum 91Zr through a (p,2n) reaction. At the effective channeltron dark noise of 0.1 cps then accounts for proton energy in the sample with the given setup, the less than 1% of the signal. cross section for this reaction is significantly lower Relative standard deviations of isotope ratios, as than for 90Zr(p,n)90Nb, but its contribution can only measured in the 107 experimental samples, are given be neglected if the sample is not enriched with 91Zr. in Table 3. Due to the skewness in data distribution, This fact limits the choice of tracer isotopes. geometric means (GM) and geometric standard deviations (GSD) are given, even if the distributions are not strictly lognormal in every case. Downloaded from Thermal ionisation mass spectrometry (TIMS) Each zirconium type contains all of the naturally- Description occurring isotopes in non-negligible amounts. This makes internal mass fractionation correction(8) The following sample preparation was developed and impossible, because none of the zirconium isotope optimised for biological samples containing nano- ratios are known in mixed samples. The maximal gram amounts of zirconium. Some aspects are based http://rpd.oxfordjournals.org/ effect on any isotope ratio at the beginning of a on earlier TIMS measurements of different measurement can be estimated by the square root of samples(5,6). In general, TIMS requires pure elemental the associated mass ratio, i.e. about 3% in the case of samples, and therefore sample preparation includes 96Zr and 90Zr or less for other combinations of zirco- (1) mixing defined amounts of urine or plasma with nium isotopes. Repeated measurements of laboratory a reference tracer (94Zr enriched to 92.7 %), fol- standards, with 2 ng natural zirconium per sample, lowed by incubation, have shown that the change of isotope ratio due to (2) microwave-powered acid digestion, mass fractionation is usually negligible compared (3) drying in Teflon beakers and redissolving, with the random fluctuations of the isotope ratio in

(4) separation of zirconium from the sample by the course of a typical measurement procedure. at GSF Zentralbibliothek on January 4, 2017 extraction chromatography, The amounts of the intravenous and oral tracers (5) drying in Teﬂon beakers and transferring zirco- have been measured with isotope dilution in 46 nium to pre-treated rhenium single ﬁlaments for samples, including blank samples. The maximum TIMS. amount measured was 40 ng. Absolute uncertainties (coverage factor 1) range from 7 pg to 3.7 ng with a The preparation of 21 samples, i.e. one TIMS median value of 0.11 ng. sample carrier, can be completed in 3–5 d. One Uncertainties are derived from the uncertainties in sample measurement including optimisation pro- isotope ratios of all zirconium types involved in the cedures takes 55 min. Samples are measured with IDMS measurement as well as from the uncertainty a thermal ionisation mass spectrometer (TRITON, of the concentration of the reference tracer. The Thermo Electron Corporation) utilising a set of six latter is the dominant source of uncertainty for channeltron detectors in multicollector mode on 60% of the results. mass numbers 90, 91, 92, 94, 96, as well as 95 for The detection limit of tracers strongly depends on isobaric correction for molybdenum. Isotope ratios the amount of natural zirconium present as back- are calculated from 50 cycles with 16.7 s as inte- ground signal. As long as high-purity ‘suprapure’- gration time, each. A dynamic evaluation is used, grade hydrochloric acid (Merck KGaA, Darmstadt, which accounts for slow changes in signal, which in Germany) was used during step 4 of the sample prep- turn could affect the online channeltron calibration aration, 0.1 mg of natural zirconium background measurements. Reference measurements of a laboratory standard solution of natural zirconium are made regularly to test stability and accuracy. Table 3. Relative standard deviations (%) of isotope ratios in Tracer concentrations in samples are measured by 107 experimental samples measured with TIMS. Correction for isobaric interference from molybdenum is denoted by (*). isotope dilution (ID), i.e. they are calculated from isotope ratios according to Yu et al.(7), but for four, Isotope ratio 91/90 92/90* 94/90* 96/90* instead of two, different types of zirconium. In the following, ‘type’ denotes a particular combi- nation of zirconium isotopes. The four types commonly Range 0.26–1.65 0.24–3.56 0.24–1.96 0.45–17.6 GM 0.58 0.61 0.65 1.30 appearing in this work are natural zirconium, refer- GSD 1.59 1.70 1.44 1.91 ence tracer, oral tracer and intravenous tracer.

268 MEASUREMENT OF ZIRCONIUM ISOTOPES was present in each sample. If, before use, hydrochloric PNA is limited to two isotopic tracers, whereas any acid of the same grade was purified by distillation at isotopic tracer can be measured with TIMS. On the temperatures below its boiling point (‘distillacid BSB- other hand, natural zirconium can only be measured 939-IR’, Berghof Products þ Instruments GmbH, with PNA, if its amount in a sample exceeds 4 ng. Eningen, Germany), the amount of natural zirconium The TIMS method requires significantly more per sample dropped to 4 ng per sample. The results staff time per sample as PNA, since the sample of 20 blank samples, with 5.3 ng reference tracer preparation is more complex. In contrast, although added to each sample, are presented in Table 4. TIMS instruments with multiple ion counters are rare, such instruments are cheaper and easier to maintain and operate than a cyclotron or particle Discussion accelerator, which is needed for PNA. Teflon beakers are used repeatedly for sample prep- With TIMS and PNA relying on different physical aration. Initial samples and further tests showed that mechanisms, it is unlikely that the same error could the efficiency of their cleaning procedure is critical occur in both methods. Therefore, the results of each Downloaded from to eliminate memory effects due to a mixing of method can be used to test those of its counterpart. present and previous samples. For stable tracer experiments with zirconium, it is The time-consuming sample preparation method recommended that TIMS is used for routine measure- contains numerous steps in which the sample can be ments due to its superior sensitivity, accuracy and contaminated with natural zirconium, e.g. by acids sample throughput. PNA should be employed as an during extraction chromatography. Therefore, no independent means of quality control. http://rpd.oxfordjournals.org/ reliable measurement of natural zirconium content in a sample is possible. ACKNOWLEDGEMENTS Concentration calculations assume a similar quan- titative behaviour of the different zirconium types The authors would like to thank B. Wolf, F. Wagner present in the sample, with the same fractional loss and A. Khegai (all GSF) for sample preparation during preparation for all types. Thorough mixing and TIMS measurements, as well as the MLL staff and equilibration of these types is imperative, for their invaluable assistance in re-establishing the especially for the reference tracer, which is added in PNA setup.

a small volume of acidic solution to the plasma or at GSF Zentralbibliothek on January 4, 2017 urine samples. TIMS is not capable of absolute calibration, i.e. REFERENCES count rate is not directly linked to sample amount, 1. Werner, E., Roth, P.,Ho¨llriegl, V.,Hansen, C., Kaltwasser, but isotope dilution mass spectrometry can be J. P., Giussani, A., Cantone, M. C., Greim, H., Zilker, T. applied to calculate type concentrations in samples. and Felgenhauer, N. Assessment of intestinal absorption This is an advantage for the purpose of tracer of trace metals in humans by means of stable isotopes. experiments, since tracer concentrations, not isotope Isotopes Environ. Health Stud. 38, 7–15 (2002). concentrations, are the desired results. Another 2. Cantone, M. C., Molho, N. and Pirola, L. Trace advantage is that all isotopes can be measured and, elements analysis in biological samples by proton nuclear therefore, each can be used as tracer. activation. Clin. Phys. Physiol. Meas. 3, 67–77 (1982). 3. Veronese, I., Giussani, A., Cantone, M. C., Maggioni, T., Birattari, C., Groppi, F., Werner, E., Roth, P. and COMPARISON Ho¨llriegl, V. A re-evaluation of the biokinetics of zirconium in humans. Appl. Radiat. Isot. 58, 431–439 (2003). TIMS and PNA can be used as independent and 4. Currie, L. A. Limits for qualitative detection and quanti- complementary techniques. They both allow the tative determination. Anal. Chem. 40, 586–593 (1968). measurement of zirconium isotopes in biological 5. Ho¨llriegl, V., Louvat, P., Werner, E., Roth, P., Schramel, samples, with TIMS generally being more sensitive. P., Wendler, I., Felgenhauer, N. and Zilker, T. Studies of strontium biokinetics in humans Part 2: Uptake of strontium from aqueous solutions and labelled foodstuffs. Table 4. Effect of acid quality on TIMS detection limit. Radiat. Environ. Biophys. 41, 281–287 (2002). 6. Boswell, S. M. and Elderﬁeld, H. The determination of zirconium and hafnium in natural waters by isotope HCl suprapure HCl subboiled dilution mass spectrometry. Mar. Chem. 25, 197–209 (1988). Natural Zr per sample 0.1 mg4ng 7. Yu, L. L., Fassett, J. D. and Guthrie, W. F. Detection Mean uncertainty of 0.31 ng (n ¼ 12) 0.03 ng (n ¼ 8) limit of isotope dilution mass spectrometry. Anal. Chem. blank 74, 3887–3891 (2002). Detection limit 1.5 ng 0.14 ng 8. Habfast, K. Fractionation correction and multiple collec- tors in thermal ionization isotope ratio mass spectrometry. Edited by AR. Int. J. Mass Spectrom. 176, 133–148 (1998).

269