(Mis)use of ¹³³Ba as a calibration surrogate for ¹³¹I in clinical activity calibrators

Abstract

Using NIST-calibrated solutions of ¹³¹Ba and ¹³¹I in the 5 mL NIST ampoule geometry, measurements were made in three NIST-maintained Capintec activity calibrators and the NIST Vinten 671 ionization chamber to evaluate the suitability of using ¹³³Ba as a calibration surrogate for ¹³¹I. For the Capintec calibrators, the ¹³³Ba response was a factor of about 300% higher than that of the same amount of ¹³¹I. For the Vinten 671, the Ba-133 response was about 7 % higher than that of ¹³¹I. These results demonstrate that ¹³³Ba is a poor surrogate for ¹³¹I. New calibration factors for these radionuclides in the ampoule geometry for the Vinten 671 and Capintec activity calibrators were also determined.

1. Introduction

Iodine-131 is one of the most commonly used radionuclides in nuclear medicine, particularly in the developing world. Clinical applications include Na[¹³¹I] for diagnosing and treating thyroid disease, both malignant and benign (Sawka, et al. 2009; Ross 2011) ¹³¹I meta-iodobenzylguanidine (mIBG) for neuroendrocrine tumors (Riad 2011); ¹³¹I-ethiodized oil for hepatocellular carcinoma (Ahmadzadehfar, et al. 2011); and radiolabeled monoclonal antibodies for non-Hodgkins lymphoma (Calais and Turner 2012). Measurements of ¹³¹I prior to administration are usually made in the hospital or radiopharmacy using activity calibrators (commonly called “dose calibrators”). A traceable calibration for the activity calibrator is crucial for ensuring the accuracy of the dosage prior to administration.

Although most guidance documents, including those from the International Atomic Energy Agency (IAEA) (2006), standards groups (ANSI 2004), and professional societies (AAPM 2012), recommend that activity calibrators be directly calibrated against standards of the same radionuclide, the expense and limited availability of ¹³¹I standard solutions force some clinics, particularly in (but not limited to) developing countries, to use surrogate sources instead. The lack of direct prohibition against this practice in some regulatory documents (CNSC 2006) may even appear to give tacit approval for this practice.

The long half-life (10.540(6) a) and photon spectrum of ¹³³Ba (DDEP 2015) make it attractive for use as a surrogate, especially since many locations also utilize ¹³³Ba check sources as part of their Quality Assurance measurements for constancy and may already have a source on hand. The main photon energies ( 360 keV) and total photon energies per decay are about the same for both radionuclides. These properties make ¹³³Ba an especially interesting surrogate for ¹³¹I in physics studies for single photon emission computed tomography (SPECT) imaging, and in fact a set of calibrated ¹³³Ba sources has been successfully used in a recent international single photon emission computed tomography SPECT image quantification comparison organized by the IAEA (Zimmerman, et al. 2013).

However, the fact that ¹³¹I decays via β particle emission (and therefore will also produce bremsstrahlung) and ¹³³Ba undergoes electron capture (which results in substantial differences in x ray emission), as well as the presence of higher energy photons in the ¹³¹I decay scheme, means that activity calibrators will have very different responses for the same amount of activity of each radionuclide. Geometrical effects can also introduce significant differences in activity calibrator response for these radionuclides, especially when calibrations are made with solid check sources (in common use clinically) when the main measurement geometry is a liquid in a syringe or vial.

Attempts have been made by manufacturers to compensate for differences between the two decay schemes by introducing surrogate sources that use a combination of radionuclides (such as ¹³³Ba and ¹³⁷Cs), but differences in the half-lives of the radionuclides will cause their activity ratio to change over time, requiring additional corrections to be made.

This work was carried out to quantify the relative responses between standardized sources of ¹³³Ba and ¹³¹I in several clinical activity calibrator models in an attempt to demonstrate the inappropriateness of using ¹³³Ba as a surrogate for calibrating activity calibrators.

2. Materials and Methods

During the experiments described in Zimmerman, et al. (2013), four 5 mL NIST ampoules were prepared, each containing nominally 4.4 MBq·g⁻¹ of ¹³³Ba in 5 g of a carrier solution of 5 mmol·L⁻¹ BaCl₂ in 0.5 mol·L⁻¹ HCl. Two of those ampoules, denoted Ba-D1-A3 and Ba-D1-A4, were also used in the present study (the solutions from the other two ampoules were used in experiments not related to this study). The total ¹³³Ba activities for Ba-D1-A3 and Ba-D1-A4 were calibrated by measurement in NIST “4π” γ ionization chamber “A” (IC “A”) as previously described (Zimmerman, et al. 2013) and found to be 22.57(15) MBq and 20.01(13) MBq at the reference time, respectively. The uncertainties are the combined standard uncertainties on the IC “A” measurements calculated as described in Zimmerman et al. (2013).

For the ¹³¹I measurements, a single NIST 5 mL ampoule containing Standard Reference Material 4401L-7, Lot 38 (NIST 2012) was used. This solution consisted of nominally 156 μg of KI, 0.234 μg of LiOH, and 70 μg Na₂SO₃ per gram of solution in distilled water. The activity of the solution was calibrated in IC “A” and found to be 24.34(11) MBq at the reference time. The uncertainty on the activity value is the combined standard uncertainty calculated from the quadratic addition of components due primarily to the IC “A” calibration factor (0.44 %), measurement precision (0.06 %), live-time determination (0.05 %), radium reference source positioning (0.05 %), and mass measurements (0.05 %).

Each of the ¹³³Ba ampoules was measured in 4 commercial activity calibrators maintained at NIST: a Capintec CRC-15R, a Capintec CRC-12, a Capintec CRC-25R, and a Vinten 671 connected to a Keithley 6517 electrometer (the system is denoted hereafter as “VIC”) ¹. The Capintec activity calibrator measurements were made with the ampoule in the bottom of the standard dipper that shipped with each chamber at both the “¹³¹I” and “¹³³Ba” dial settings recommended by the manufacturer, which are 151 and 591, respectively. The Capintec CRC-12 had been modified by the manufacturer to allow for direct current readings in addition to operating in its “normal” mode. For the VIC measurements, the vial holder was used with the ampoule sitting in the lowest position.

A total of 100 measurements of the ionization current were taken in the VIC at intervals of 2 seconds for each ampoule; 1000 measurements of the background current were made before and after the ampoule measurements. The ¹³¹I ampoule was measured in each chamber in the same manner as was done for the ¹³³Ba ampoules, except that the measurements were carried out only at the manufacturer’s recommended setting for ¹³¹I in the case of measurements made in the Capintec chambers.

For both radionuclides, the dial setting for the ampoules in each of the Capintec chambers was re-determined by changing the dial until the correct total activity was displayed on the readout.

3. Results and Discussion

For the VIC, the calibration coefficients for the ¹³¹I and ¹³³Ba in ampoules were measured to be 3.993(18) pA·MBq ⁻¹ and 4.287(30) pA·MBq ⁻¹, respectively, giving a ¹³³Ba:¹³¹I response ratio of 1.074(9). The uncertainties on the individual calibration coefficients are combined standard uncertainties calculated from components due to decay correction (0.045 % and 0.004 % for ¹³³Ba and ¹³¹I, respectively), the uncertainty on the ¹³³Ba and ¹³¹I activity calibrations (0.66 % and 0.45 %, respectively), and repeatability of the ionization current measurements (0.3 % and 0.03 % for ¹³³Ba and ¹³¹I, respectively). The large difference between the decay correction uncertainties originates from the differences in time between when the particular samples were calibrated and the time of the measurements made in this study. The ¹³³Ba solutions were calibrated well before these studies were performed, while the ¹³¹I solution was calibrated within one to two days of the experiments. The¹³³Ba calibration coefficient uncertainty also contains an additional component with magnitude 0.2 % to account for measurement variability between the two ampoules that is estimated from the standard deviation on the mean value of the calibration coefficients obtained from the two ampoules.

The value of 1.074(9) for the ratio of the ¹³³Ba and ¹³¹I VIC calibration coefficients obtained in this study for the NIST standard 5 mL ampoule is in agreement with the ratio of 1.066(8) obtained from the volume- (or more correctly, mass) corrected calibration coefficients published by the National Physical Laboratory (NPL) for the 5 mL British Standard ampoule (Keightley 2012). Although minor differences exist between the configurations of the two ampoules (BSI 1983, NIST 2012), this level of agreement gives confidence in the present result. Further confidence can be gained by calculating the average photon energy per decay for each radionuclide. Based on the available decay scheme data (DDEP 2015) we obtain values of 376.1(19) keV per decay for ¹³¹I and 398.2(11) keV per decay for ¹³³Ba, which gives a ratio of 1.059(6), which is in accord with the values given above to within uncertainties, assuming that the VIC response to those average decay energies is nominally equal. The agreement between the three ratios also suggests that for the VIC, the contribution to the response from bremsstrahlung in the decay of ¹³¹I is small, since otherwise a larger difference between the experimental NIST and NPL ratios and the one calculated solely from the photons would be observed.

The physical construction of the Capintec chambers and the VIC are quite different, namely the filling gas is at a higher pressure in the Capintec chambers and is nominally 1.22 MPa of argon instead of 1 MPa of nitrogen. As a result, the walls of the chamber are thicker in the Capintec calibrators in order to accommodate the higher pressure. The walls of the Capintec chambers are constructed from steel, whereas the Vinten chamber walls are aluminum. This will result in a different bremsstrahlung spectrum in each case. In addition, the outer walls of the Capintec calibrators contain additional shielding, as they are designed primarily to be used in radiopharmacies with relatively high levels of activity. The VIC, on the other hand, was designed to be a secondary standard instrument with emphasis on sensitivity at low energy and no additional shielding. This philosophy is also reflected in the inner wall construction, which results in a lower cutoff energy for the Vinten. Thus, it is expected that some differences in the relative responses for ¹³³Ba and ¹³¹I between the chambers are observed.

By reading the current from the Capintec CRC-12 chamber, we can derive calibration coefficients for both radionuclides in a similar manner as for the VIC. For this instrument, we obtain values of 11.01(8) pA·MBq⁻¹ and 3.76(2) pA·MBq⁻¹ for the ¹³³Ba and ¹³¹I ampoules, respectively. The uncertainties are standard uncertainties and were calculated from the components due to decay correction (0.045 % and 0.004 % for ¹³³Ba and ¹³¹I, respectively), the uncertainty on the ¹³³Ba and ¹³¹I activity calibrations (0.66 % and 0.45 %, respectively), and for ¹³³Ba, between ampoule measurement variability (0.2 %). This gives a ratio of 2.93(3) for the calibration coefficients measured in this particular chamber.

The activities reported by the various Capintec activity calibrators when measuring the ¹³³Ba and ¹³¹I ampoules at the manufacturer’s recommended settings for ¹³¹I, are given in Table 1. As can be seen in the first column, the activities reported by the activity calibrators for ¹³³Ba are consistently a factor of 3 higher than the calibrated activity when measured at the ¹³¹I setting. For the ¹³¹I solution, the activity reading at the ¹³¹I setting is about 2.5 % higher than the calibrated activity, as indicated by the second column. By dividing the values in the first two columns, we obtain a new ratio of the ¹³³Ba and ¹³¹I activities that is corrected for the error in the manufacturer’s recommended setting for ¹³¹I. These values are presented in the third column of Table 1.

Table 1.

Ratios of activity calibrator readout values to NIST-calibrated activity for ¹³³Ba and ¹³¹I in three NIST-maintained radionuclide calibrators at the manufacturer’s recommended setting for ¹³¹I. The third column gives the ratios of the first two columns to provide a ¹³³Ba:¹³¹I response ratio that is corrected for the error in the ¹³¹I activity introduced from use of the manufacturer’s setting. Uncertainties are standard uncertainties and are calculated as described in the text.

	R(Achamber/ANIST) 133Ba	R(Achamber/ANIST) 131I	R(133Ba:131I)
CRC-12, 131I setting	3.01(2)	1.018(5)	2.96(2)
CRC-15R, 131I setting	3.03(2)	1.024(2)	2.96(2)
CRC-35R, 131I setting	2.99(2)	1.029(5)	2.91(2)
Average	3.01(3)	1.024(8)	2.94(4)

Both the average ratio of the ¹³³Ba activity readout value at the ¹³¹I setting to the calibrated ¹³³Ba activity and the average ratio of the calibration coefficients obtained from the CRC-12 current measurements are in excellent agreement and indicate that the response of ¹³³Ba in these chambers is 2.94 times higher than that of ¹³¹I for an equivalent amount of activity. This striking difference in response ratios demonstrates how differences in chamber construction can have a profound effect on the energy response.

For completeness, the ratios of the readout in the Capintec activity calibrators when measuring the ¹³³Ba ampoules in each instrument using the manufacturer’s recommended setting for ¹³³Ba to the calibrated activity are presented in Table 2. The data indicate that the recommended ¹³³Ba setting gives a reading that is an average of 5.2 % high relative to the calibrated activity in the CRC-12 and CRC-15R and 3.7 % for the CRC-35R. The values in Table 2 are not intended to be used as correction factors, but are presented merely to give an indication as to the accuracy in the measurement of ¹³³Ba that can be expected when using the manufacturer’s recommended setting. The reason for the observed difference in response between the CRC-12 and CRC-15R and the CRC-35R is not known, but is consistent with other results published by our laboratory and is within the tolerances quoted by the manufacturer (Capintec 2004).

Table 2.

Ratios of activity calibrator readout values to NIST-calibrated activity for ¹³³Ba in three NIST-maintained radionuclide calibrators at the manufacturer’s recommended setting for ¹³³Ba. Uncertainties are standard uncertainties and are calculated as described in the text.

	R(Achamber/ANIST)133Ba
CRC-12, 133Ba setting	1.054(7)
CRC-15R, 133Ba setting	1.051(7)
CRC-35R, 133Ba setting	1.037(8)

The correct dial settings for both radionuclides, as measured in each of the activity calibrators used in this study, are given in Table 3. Because of the prevalence of low-energy photon emission and the possible contribution of bremsstrahlung to the response of ¹³¹I, users are always encouraged to evaluate the dial settings for these radionuclides for their specific measurement geometry and calibrator using a traceable standard of the actual radionuclide and not rely on surrogates.

Table 3.

Experimentally determined dial settings for ¹³¹I and ¹³³Ba in the NIST 5 mL ampoule geometry using three NIST-maintained Capintec activity calibrators. The uncertainties are standard uncertainties and are due primarily to the uncertainty in the activity calibration for the respective radionuclide.

Activity Calibrator	DS131I	DS133Ba
CRC-12	156(1)	628(2)
CRC-15R	156(2)	626(2)
CRC-35R	158(2)	618(2)

The results from this work clearly show that although both ¹³³Ba and ¹³¹I both have strong photon emissions at about 360 keV and may initially appear to be well-matched in terms of ¹³³Ba being able to act as a surrogate for ¹³¹I, the reality is that substantial errors will result if calibrations for ¹³¹I are made using a ¹³³Ba source. Even the smallest difference that we observed, namely 7.5 % for the VIC, is very close to the 10 % error generally accepted for therapeutic applications, and those measurements were made using a calibrated source with a small uncertainty on well-maintained systems. If a source with a larger uncertainty, which would typically be the case for check sources found in most clinics, is used, a 10 % error in administration becomes increasingly likely.

If ¹³¹I is used clinically, e very effort should be made to obtain a traceable ¹³¹I standard and calibrate the activity calibrator at least once. If a ¹³³Ba source is available, it can make a good long-term monitor of the stability of the ¹³¹I calibration if the relative response of the two radionuclides is established when the instrument is first calibrated and routinely measured as part of a quality control system.

5. Conclusion

We have measured the relative responses of standardized solution sources of ¹³³Ba and ¹³¹I in several models of activity calibrators from two manufacturers to evaluate the suitability of using ¹³³Ba as a long-lived surrogate for ¹³¹I. The results showed that significant errors in the assumed ¹³¹I activity, ranging from 7 % to almost 300 % can be realized from this practice, depending on the type of activity calibrator used. Despite the fact that local conditions do not always make it possible or convenient to obtain traceable ¹³¹I standards on a regular basis, at least one calibration should be made using such a standard. Barium-133 should not be used as a calibration surrogate for ¹³¹I, but it can be used to monitor the stability of a properly performed ¹³¹I calibration over time.