Strontium 90 in Urine: Evaluation of Radiotoxicological Techniques by International PROCORAD Comparison Exercises

Strontium 90 in Urine: Evaluation of Radiotoxicological Techniques by International PROCORAD Comparison Exercises R Fottorino 1, M. Ruffin 1 and Ph. Bérard 2 1 CEA Cadarache, LABM, Bâtiment 102, 13108 Saint Paul Les Durance, France 2 PROCORAD Scientific Council, CEA Saclay, LABM, Bâtiment 601, 91191 Gif sur Yvette Cedex, France Abstract Strontium 90 (halflife = 28.79 years) is a «leader» of fission products. Through β disintegration, it leads to yttrium 90 which is a shortterm radioactive β emitter (64 h). Strontium 90 does not exist in a natural state: it is a product of the nuclear fuel cycle. The 90 Sr 90 Y couple makes up one of the means used in monitoring workers in the nuclear industry. Its quantification in urine serves as a means on confirming the operational radioprotection of nuclear facilities. In France, the Clinical Biochemistry Laboratories (CBL) are in charge of monitoring nuclear sites workers. These laboratories must be accredited and have ministerial certification. The accreditation of radiobioassay analyses notably requires comparison exercises between laboratories. One of the most important is that organized by the PROCORAD association. The specific activity of this association is to provide biological samples overloaded in various radio nuclides to international CBL and to exploit the results obtained by the different laboratories: both statistically and anonymously thanks to a speciallydesigned software entitled Procostat. The urinary Strontium 90 comparison exercise was initiated at the inception of PROCORAD. It answers a definite demand from laboratories in charge of monitoring workers, since every year more than 40 participants from all over the world sign up to participate. The authors present an overall assessment of the Strontium 90 comparison exercises reviewed over the past seven years (20002007) by comparing the mean geometric value of laboratories with a target value, all techniques included. The metrological difficulty leads to a tendency to underestimate its quantification. Analysis of the data through comparison of the couple, techniques and results, enables us to assess performances according to the techniques used. Certain techniques are very welladapted to the dosage of Strontium 90 in terms of exactitude and reproducibility of the results provided. A specific technique associating chromatography and measurement using a gas beta detector as proportional counter presents an even narrower bias. The gross beta techniques, nonspecific, are influenced by gamma emitters that might possibly be present in the samples. The study also deals with other techniques and reviews their performances. Apart from an assessment of the performances of the different laboratories, the urinary Strontium 90 comparison exercises organized by the PROCORAD association enable the Clinical Biochemistry Laboratories to evaluate the techniques implemented in terms of accuracy, reproducibility and specificity. KEYWORDS: Strontium 90; Procorad; Comparison exercises; Techniques; Performances. 1. Introduction Strontium 90 (half life = 28.79 years) is a leader of fission products. Through β disintegration, it leads to yttrium 90 which is a shortterm radioactive β emitter (64 h). Strontium 90 does not exist in a natural state: it is a product of the nuclear fuel cycle. The 90 Sr 90 Y couple makes up one of the means used in monitoring workers in the nuclear industry. Its quantification in urine serves as a means to confirming the operational radioprotection of nuclear facilities. In France, the CBL (Clinical Biochemistry Laboratories) on site at nuclear facilities are involved in monitoring of workers exposed to a risk of contamination. They must be accredited according to the ISO CEI 17025 standard, and authorized by the Nuclear Safety Authority (ASN). One of the key point of the accreditation is to participate in external quality controls which are comparison exercises between laboratories that allow an objective appreciation of analytical performances of the laboratory. One of the most important comparison exercises is that organized by the PROCORAD association. 1

2. PROCORAD association Created in 1995 by the biologists of AREVA and Atomic Energy Commission (CEA), Procorad is the Association for the Promotion of Quality Controls in Radiotoxicological Bioassays. Its purpose is to organize annual comparison exercises between laboratories throughout the world in different analytical fields and under the technical responsibility of the organizer laboratories from CEA and AREVA sites. The participant laboratories may be environmental or biological laboratories. Six comparison exercises are organized: Tritium in urine Actinides in urine Actinides in fecal ashes Carbone 14 in urine Strontium 90 and Gamma X emitters in urine Surprise urine The annual schedule of this preparation is very strict: From October to February: preparation of the biological samples overloaded by certified radioactive standard solutions December: inscriptions of the different laboratories via Procorad web site and communication of a specific anonymous number. End of February: sending of samples to participant laboratories End of April: sending of results and answers to technical questionnaire by participant laboratories. From May to mid June: statistical treatment of results via a specially designed software entitled Procostat and preparation of the presentation by the organizer laboratories. Mid June: annual congress and presentation of results by number of anonymity. End of October: sending of a CD Rom with presentations, minutes of each exercise and positioning of each laboratory (own result versus target value). The best methods, those whose bias in regard with the target value is minimal, are also published in a collection of techniques, always keeping anonymity. 3. Organization of Strontium 90 comparison exercises This comparison exercise is organized by the CBL of Cadarache, a nuclear research facility of CEA. Associated with GammaX emitter comparison exercise, the preparation of that comparison begins with the collection of urine from non exposed persons, e.g., Figure 1. Approximately 100 liters of urine are collected on sulfamic acid and nitric acid, filtered and shared in 3 parts, equivalent in volume. The part A, a non radioactive sample, corresponds to sample A. The parts B and C are overloaded by a determined mass of certified standard Strontium 90 and X and gamma emitter solutions and are nominated sample B and C. Many checks are made by the organizer laboratory: Check the activity of the certified standard solution used for overloading Check a sample after overloading. After conformity of these checks, the organizer laboratory sends the samples to the participant laboratories. Every laboratory receives approximately 500 ml of each sample whose activity in 90 Sr is less than 10 Bq.l 1. Each laboratory has to analyze the samples with its own technique and to express their results in term of measured activities associated with the global uncertainties. 2

Figure 1: Preparation of 90 Sr and gammax comparison exercise Non contamination check Inactive urine A urine B urine C urine Emitter(s) γ 1...γ n source Or γ 1...γ n Source sol. Activity check Conform 90 Sr Source m'1 m1 m'2 m2 No spike Spike Activity check Conform Spike + carrier Aliquots approx. 500 ml Aliquots approx. 500 ml Aliquots approx. 500 ml Sample A Sample B Sample C After reception of each result, the statistical treatment with Procostat software is performed by using three or four statistical tests according to the number of results to determine which result will be kept or excluded: Grubbs, Student and Z Score tests if the number of laboratories is more than 30 Dixon test in addition for less than 30 laboratories Arithmetical and geometrical mean values are calculated according to the retained results and a first diagram of repartition by anonymous number is made, e.g., Figure 2. Figure 2: Repartition of laboratories 66 53 52 25 60 51 23 46 28 49 12 45 43 27 68 48 5 36 37 8 57 6 39 31 3 19 17 1 69 13 3,90E+00 > 4,20E+00 > 4,50E+00 > 4,80E+00 > 5,10E+00 > 5,40E+00 > 5,69E+00 > 5,99E+00 > 6,29E+00 > 6,59E+00 <= 4,20E+00 <= 4,50E+00 <= 4,80E+00 <= 5,10E+00 <= 5,40E+00 <= 5,69E+00 <= 5,99E+00 <= 6,29E+00 <= 6,59E+00 6,89E+00 Column with target value 3

Then, two other diagrams are made, e.g., Figure 3. One corresponds to the result and uncertainty of each laboratory according to the target value and its uncertainty. The second one presents the bias in percentage in relation with the target value. Figure 3: Diagrams of results with uncertainties and bias (%) regarding to the target value 9,00E+00 30 8,00E+00 20 7,00E+00 10 Bq per litre 6,00E+00 5,00E+00 BIAS (%) 0 10 4,00E+00 20 3,00E+00 30 2,00E+00 57 68 6 39 49 31 48 53 66 23 25 52 12 5 3 19 36 45 46 60 37 43 17 51 1 8 27 28 69 13 N LABORATORY 40 57 68 6 39 49 31 48 53 66 23 25 52 12 5 3 19 36 45 46 60 37 43 17 51 1 8 27 28 69 13 N LABORATORY The annual meeting in June allows the participant laboratory to assess his accuracy according to the results sent. Moreover, the organizer presents the comparisons of performances between different chemical methods and counting equipments. 4. Different Strontium 90 methods used The information about the different kinds of methods is provided by the technical questionnaire linked to the results which is filled in by the participants (the answer to this questionnaire is optional). 4.1 Different kinds of chemical methods The volume of urine used for the test goes from 50 to 500 ml. Every method always associates a concentration of Strontium 90 in urine by precipitation in the form of oxalate or phosphate. At this level of the analysis, we can differentiate two kinds of methods: the direct methods and the indirect ones. 4.1.1 The direct methods After the first step of concentration, the precipitate is collected, chemically treated and counted. These methods are named gross beta methods. 4.1.2. The indirect methods After concentration, these methods include an additional step of strontium separation isolation before counting. This step can be: a chromatography: affinity chromatography (SrSpec Eichrom resins) or ion exchange chromatography (Dowex anionic resins) a solvent extraction of 90 Y by Di Ethyl Hexyl Phosphate (DEHP) toluene with reextraction with 3N nitric acid a separation of 90 Sr by fuming nitric acid, followed by a purification of 90 Sr. 4

4.2 Different kinds of counting techniques The final counting step is made using three kinds of detection: Liquid scintillation counting (LSC) Cerenkov effect Gas beta counting (proportional counter) 5. Performances of 90 Sr methods 5.1 Statistical study with all methods The results of the different comparison exercises from 2000 to 2007 are represented in Table 1. The values correspond to the results given by the population included in the statistical tests (not the whole population). The bias between the geometric mean and the target value shows that an underestimation is obvious, except, generally, in case of gamma interferences where there is often an overestimation. Actually, the 90 Sr comparison exercise is coupled with the gammax comparison exercise. So, some betagamma emitters such as (especially), (at a lower level) can interfere. 5

Table 1: Results of comparison exercises from 2000 to 2007 All methods year 2000 2001 2002 2003 2004 2005 2006 2007 4.74 3.25 5.91 2.62 2.62 2.62 4.65 3.08 4.54 4.54 5.06 7.29 6.19 4.61 5.76 0.13 0.09 0.22 0.1 0.05 0.05 0.15 0.10 0.14 0.14 0.15 0.22 0.19 0.14 0.17 Target value uncertainty (Bq.l 1 ) Arithmetic mean 4.90 3.63 5.73 2.39 3.10 2.84 4.42 2.96 4.19 4.31 4.77 6.92 6.04 4.50 5.93 5.18 Geometric mean 4.77 3.48 5.63 2.36 3.05 2.81 4.39 2.95 4.15 4.26 4.66 6.80 5.95 4.44 5.90 5.15 Median 4.74 3.36 5.47 2.40 2.85 2.78 4.37 2.93 4.17 4.32 4.82 672 5.91 4.46 5.80 5.11 Minimum value 2.72 1.76 3.20 1.50 2.40 2.02 3.70 2.49 3.10 2.67 2.40 3.91 3.11 2.94 4.60 4.15 Maximum value 8.43 6.17 7.57 3.09 4.65 3.76 5.43 3.54 5.64 5.33 6.71 10.3 8.35 6.92 7.30 6.60 Bias geometric mean / target value (%) Possible gamma interferences + 0.6 +7.1 4.7 9.9 +16 +7.3 5.6 4.2 8.6 6.2 7.9 6.7 3.9 3.7 +2.4 +6.0 Population included in the statistical tests 4.86 0.15 6

5.2 Statistical study for each method The study began in 2003, e.g., Table 2. Table 2: Statistical study by methods Mean of Bias (%) 5.17% 1.56 0.5% Arithmetic means standard deviation Bias (%) / target Year 2003 2004 2005 2006 2007 Target value uncertainty (Bq.L 1 ) Sr spec + LSC Sr Spec + beta proportional counter DEHP extraction + beta proportional counter Gross beta + beta proportional counter Possible gamma interferences 4.65 0.15 4.15 0.3 11% 4.64 0.6 0.2% 3.85 +0.2 17% 4.70 0.47 +11% 3.08 0.10 2.90 0.1 6% 2.94 0.3 5% 2.60 0.14 16% 2.90 0.2 6% 4.54 0.14 4.21 0.3 7% 4.11 9% 4.29 5.5% 4.15 9% 4.54 0.14 4.28 0.1 6% 4.13 0.8 9% 4.56 +0.4% 4.51 1% Population included in the statistical tests 5.06 0.15 4.95 2% 5.20 0.5 +2% 3.29 35% 4.74 0.9 6% 7.29 0.22 7.45 1.6 +2% 7.47 +2% 6.67 8.5% 6.79 1.0 7% 6.19 0.19 5.51 0.2 11% 6.10 0.4 1.5% 5.93 +0.3 4.2% 6.14 1.0 0.8% 4.61 0.14 4.11 0.5 11% 4.24 0.6 8% 4.47 0.2 3% 4.58 0.5 % 5.76 0.17 5.81 0.8 +0.9% 6.04 0.3 +4.9% 5.65 +0.2 1.9% 5.90 0.6 +2.4% 4.86 0.15 4.83 0.5 0.6% 5.24 0.4 +7.8 % 5.38 + +11 % 5.45 0.6 +12 % The narrowest bias (0.5%) is given by the gross beta methods. Nevertheless, the dispersion of these results is important (19%) as it goes from 7% to + 12%. So, the best method seems to be the association SrSpec + beta proportional counter. In Table 3, the 2008 results confirm that the SrSpec method is well adapted to 90 Sr determination, particularly when coupled with SLC. The use of SrSpec coupled with beta proportional counter is used by only one participant. The bias of it is wider than the other years but, the statistical impact of these results cannot be retained as significant. The gross beta method also gives good results as the bias regarding to the target is less than 4% in absolute value, but it is to be noted that, in the case of the presence of (sample C), the bias is higher than that of SrSpec coupled with LSC. 7

Table 3: 2008 results according to the methods 2008 Number of labs Sample B Sample C 5.48 0.16 Bq.l 1 5.52 0.17 Bq.l 1 Arithmetics means standard deviation bias SrSpec + LSC 7/8 5.39 Bq.l 1 0.8 1.6% 5.53 Bq.l 1 0.9 +0.2% SrSpec + beta proportional counter Gross beta + beta proportional counter DEHP extraction + beta proportional counter Possible gamma interference 5.3 Distribution of the results 1 6.11 Bq.l 1 +11.5% 6.17 Bq.l 1 +11.8% 7 5.28 Bq.l 1 0.6 3.7% 5.73 Bq.l 1 1.38 +3.8% 2 5.53 Bq.l 1 0.03 +0.9% 5.90 Bq.l 1 0.23 +6.9% Population included in the statistical tests In Table 4, this distribution,, shows that 94% of results are included in the interval 25% to +50%. Table 4: Distribution of results (means from 2000 to 2008 data) 10% to +10% 20% to +20% 25% to +50% (ANSI) 55% 80% 94% Population included in the statistical tests 6. Conclusion The gross beta methods are used by approximately half of comparison exercise participants. This method gives good results but presents interferences with gamma emitters which can eventually be present in the sample. So, this method can be used in first intention, for the initial testing, but needs to be confirmed by a more specific method in case of result above the detection limit to avoid an increase due to possible interferences. The extraction chromatography methods give good results. Used by a large number of laboratories, they have the advantage of being more specific and less sensitive to possible interfering gamma emitters. These two methods can be considered as complementary methods. Above all, it is to be noted that the global performances of the 90 Sr determination methods give good results as 94% of results are included in the 25% to +50% interval, which is the confident interval recommended by the American National Standards Institute (ANSI). Beyond the metrological evaluation indispensable to objectively prove the capability of the laboratory to perform analysis, the comparison of analytical performances regarding the accuracy and the reproducibility of the available methods is a real interest of PROCORAD s international comparison exercises, especially as the participation to comparison exercises is becoming a requirement for accredited laboratories. 8

REFERENCES 1. Journal of Radioanalytical and Nuclear Chemistry articles 197 N 2 357368 (1995) "Interlaboratory Comparisons : determination of actinides in excreta" P. Bérard, JC Harduin and D. Cavadore 2. Radiation Protection dosimetry Vol 79 1/4 509512 (1998) "International RadioBioassay quality controls " P. Bérard, JC Harduin, F Briot and D. Cavadore 3. Radiation Protection Dosimetry Vol 105, N 14 pp 447450 (2003) : Procorad s international intercomparisons highlight the evolution of techniques used to determine uranium in urine P. Bérard, A Montègue, F Briot 4. American National Standards Institute (ANSI) 1996, Performance Criteria for Radiobioassay, HPS N13.301996 5. ISO 127901 standard, Radiation Protection Performance Criteria for Radiobioassay Part 1: General Principles 9