For this study, we used doubled-distilled nitric acid (GFS Chemicals) to prepare solutions of various concentrations and oxalic acid dihydrate (Fisher Scientific) to prepare the mixture of 0.05M oxalic acid and 3M nitric acid. 1000 μg/mL strontium standard in 0.1% nitric acid (Inorganic Ventures) was used to prepare calibrators and quality control (QC) samples with stable Sr-88 for ICP-MS analysis while 1000 μg/mL rhodium standard in 2% nitric acid (Inorganic Ventures) was used to prepare the diluent for ICP-MS analysis. We also used strontium nitrate, 99% (Fisher Scientific) to prepare 12.1 g/L solutions in deionized (DI) water; and aluminum nitrate (GFS Chemicals) to prepare a 2M aluminum nitrate solution in DI water. ≥18 MΩ∙cm DI water (Aqua Solutions) was used for all solutions. Ultima Gold AB liquid scintillation cocktail (UGAB) (PerkinElmer) was used for liquid scintillation counting (LSC) analysis.

‘Base urine’ was collected through anonymous human donations (in accordance with Centers for Disease Control and Prevention Institutional Review Board protocol 3994) and acidified to 1% v/v with double-distilled nitric acid. All radioactive source solutions were traceable to the National Institute of Standards and Technology (NIST) (Gaithersburg, MD, USA). Base-Sr90 is non-spiked base urine purified for Sr-90. LU-Sr90, HU-Sr90, Low-Sr90, and High-Sr90 are urine gross alpha/beta quality control materials purified for Sr-90. These quality control materials are base urine spiked with NIST-certified Am-241 and Sr-90/Y-90 nuclides at different levels (see Table 1 for characterization data before purification). They were obtained from Eckert & Ziegler Analytics on different dates. For the limit of detection (LOD) estimation, four Sr-90 LOD samples in the activity range of 0 – 65 Bq/L were prepared in the laboratory by spiking base urine with a Sr-90/Y-90 standard reference solution purchased from Eckert & Ziegler Analytics.

For our analysis, we used Sr resin cartridges (2 mL) acquired over several years (see Table 2) and a vacuum box that can accommodate 24 samples (Eichrom Technologies). Three Quantulus1220 ultra-low-level liquid scintillation counters (PerkinElmer) were used for LSC Sr-90 analysis in alpha/beta mode and a NexION^® 300 inductively coupled plasma dynamic reaction cell mass spectrometer (ICP-DRC-MS) (PerkinElmer) was used for Sr-88 ICP-MS analysis as a recovery method. We also used a high-precision analytical balance accurate to 0.0001 g (Mettler Toledo) and a hot plate (Fisher Scientific) for drying eluted samples after purification for gravimetric recovery. Additional supplies included 20 mL LSC plastic vials (PerkinElmer, Inc.); 15 mL and 50 mL conical polypropylene tubes (Beckton Dickinson Labware) for solution preparation, a bottle top dispenser with capacity from 5 mL to 25 mL (Brinkman Instruments) for LSC cocktail dispensing, and a set of four pipettes with total volume range from 5 μL to 5 mL (Eppendorf).

Sr-90/Y-90 and Am-241 urine spikes (5 mL) were mixed with concentrated nitric acid (5 mL), 2M aluminum nitrate (1 mL), and stable Sr standard 1000 μg/mL (100 μL) in 15 mL polypropylene tubes [4]. We performed the steps for the Sr-90 purification process as described in Figure 1.

Sr-90/Y-90 and Am-241 urine spikes (5 mL) were mixed with concentrated nitric acid (5 mL), 2M aluminum nitrate (1 mL), and strontium nitrate aqueous solution 12.1 gm/L (1 mL) in 15 mL polypropylene tubes [4]. Figure 2 shows the sequence of steps performed for Sr-90 purification for the gravimetric method.

For Sr-90 analysis we used LSC in alpha/beta mode. PSA optimization and quench curves for Sr-90/Y-90 were built as previously described [5]. We set count time for samples at 5 minutes (min) with an additional 1 min for the external standard, for a total analysis time of approximately 7.5 min per sample. This analysis time provides reasonable minimum detectable activity and good counting statistics [5]. For both recovery methods, the final sample volume is 5 mL; therefore, 15 mL of UGAB cocktail will work for 20 mL vial geometry.

Stable Sr has 4 isotopes: Sr-84 (0.56%), Sr-86 (9.86%), Sr-87 (7%), and Sr-88 (82.58%), and its standard atomic mass is 87.67 u. We use Sr-88 as the target for ICP-MS analysis because it is the most abundant isotope without significant interference.

The diluent in the ICP-MS method for stable Sr recovery is an aqueous solution of 10 μg/L internal standard (rhodium) in 1% v/v nitric acid [4]. We add this diluent to all samples and calibrators during the dilution process prior to analysis. The method rinse uses an aqueous solution of 1% v/v nitric acid. We pump this rinse solution through the sample introduction system between samples to prevent carry-over of strontium and internal standard from one sample measurement to the next. With each set of samples, the instrument software collects data on four external calibrators (in the total range of 0 – 200 μg/L) prepared in 0.1% v/v nitric acid and generates a simple linear calibration curve for Sr-88.

We evaluated the aqueous calibration curve versus a matrix matched calibration curve, by performing the following experiments. First, we analyzed purified base urine diluted 200-fold with DI water (matrix) which was not spiked with stable Sr tracer. Using an aqueous calibration curve, we did not find Sr-88. To establish matrix equivalency, we made 10 DI water based calibrators with concentrations spanning our calibration range, prepared in a ratio of DI water : calibrator : diluent of 50:50:4600 and 10 matrix matched calibrators based on the same 10 calibrator concentrations prepared in a ratio of matrix : calibrator : diluent of 50:50:4600. We analyzed both sets of 10 calibrators 3 times, and the difference between the average of the slopes for both calibration curves, aqueous and matrix matched, was 0.72%. This meets our definition of matrix equivalency. Analyses of several stable strontium spikes using both calibration curves produced similar results, further substantiating the equivalence of DI water to the analyte’s native urine matrix. This allows us to prepare samples and calibrators in the ratio of sample/calibrator: diluent of 50:4600. We prepared and characterized low and high QC for Sr-88 ICP-MS method (60 μg/L and 150 μg/L) which we analyzed with each run to confirm that the method and instrument is under control.

Normal stable Sr content in urine is in the range of 27 μg/L to178 μg/L [6]. A 200-fold dilution of the processed urine sample decreases this normal stable Sr content to 0.14 μg/L to 0.89 μg/L. With the final ICP-MS concentration of added Sr tracer at approximately 100 μg/L (after sample prep and 200-fold dilution), the analytical error introduced from naturally occurring stable Sr in urine is no more than 1%.

For Sr-90 LSC analysis, we suggest using a gross alpha/beta method [5] for several reasons. (1) If alpha-emitting radionuclides are not completely removed during column extraction, gross alpha/beta analysis allows separation of alpha and beta signals. (2) Optimized gross alpha/beta method parameters [5] (including Pulse Shape Analysis setting, LSC cocktail, quench curves, count time etc.) are developed using Sr-90/Y-90 as the beta emitter therefore, applying gross alpha/beta LSC analysis to this strontium specific method is appropriate. (3) Our characterized gross alpha/beta QC materials contain Sr-90/Y-90 as the beta emitter, allowing us to use the same QC for Sr-90 analysis (see Table 1).

Tables 2 shows the recovery results for both methods. We saw an obvious trend for both Sr recovery methods: the older the cartridges, the lower the recovery. However, for the ICP-MS method, Sr recovery, even for the older cartridges (2009), is still high (approximately 89–90%). For the gravimetric method, recovery for older cartridges is much lower (approximately 38%).

Tables 3 and 4 present observed and target Sr-90 specific activity correlations in known urine spikes for both recovery methods. The activities were found by LSC as gross beta in alpha/beta mode and accounted for Sr recovery and Y-90 in-growth. Y-90 in-growth was calculated with the Bateman equation for secular equilibrium [7] as shown in Eq. (2): (2)AY-90(t) = ASr-90×(1−e−0.693/64 × t) where A_Y-90 (t) – Y-90 activity at time t, A_Sr-90 – Sr-90 activity, 0.693/64 – decay constant for Y-90 (ln2/T_1/2 where T_1/2=64 hr is Y-90 half life time); t – time from purification to LSC analysis (in hours). We calculated Y-90 activity as percentage from Sr-90 activity as shown in Eq. (3): (3)AY-90/ASr-90= 1−e−0.693/64 ×t For example, in one hour after Sr-90 purification Y-90 in-growth will be 1.08% from Sr-90 amount; in 24 hours it will be 22.9%; in three weeks it will be about 100%.

Sr-90 activity could be found according to Eq (4): (4)A(Sr-90) = A(t) / (1+ AY-90/ASr-90) / Sr recovery Where A_(Sr-90) is current Sr-90 activity, A(t) is measured current gross beta activity, and A_Y-90/A_Sr-90 is the ratio from Eq. (3), Sr recovery was found as discussed above.

Knowing the reference date for Sr-90 spikes, we can calculate the original Sr-90 activity from the exponential decay formula [8] as shown in Eq. (5) and compare it with the target data: (5)Original A(Sr-90) = A(Sr-90)/e−0.693/28.9×t1 where original A_(Sr-90) is Sr-90 activity at the reference date, A_(Sr-90) is current Sr-90 activity from Eq. (4), 0.693/28.9 is decay constant for Sr-90 (ln2/T_1/2 where T_1/2=28.9 years - Sr-90 half life time); t₁ is time from sample reference date till the date of analysis (years).

The correlation between found and target Sr-90 specific activity results are in the range of ±5% is within our acceptability criteria for both Sr recovery methods. These experiments were done in 2015. At that time, the cartridges from 2009 were 6 years old, cartridges from 2012 were 3 years old, and cartridges from 2014 were 1 year old.

For comparison purposes we calculated the LOD for Sr-90 analysis with each recovery method. It was based on the Clinical and Laboratory Standards Institute (CLSI) method EP17-A2 [9]. Four Sr-90 LOD samples were prepared in the laboratory in the range of 0 – 65 Bq/L in base urine matrix and analyzed 60 times after purification on Quantulus 1220 series instruments. LODs were determined according to the Eq. (6): (6)Activity LOD = [M b + 1.645 (SD b + I)] / [1–1.645S] where M b is the blank average activity, and SD b is the standard deviation for blank average activity, I is the intercept and S is the slope of the curve of standard deviation (SD) versus LOD samples activity (see Figs. 3 and 4). All activities are in Bq/L. The LOD for the ICP-MS recovery method was 20 Bq/L while for gravimetric recovery it was 25 Bq/L. The higher LOD for gravimetric recovery method might be due to the use of glass vials for the gravimetric method versus the use of plastic vials for ICP-MS recovery. Glass vials produce higher background because of higher potassium content in glass.