Radiation Dosimetry of ⁸²Rb in Humans Under Pharmacologic Stress

Subjects

Ten healthy volunteers (5 male, 5 female; mean age ± SD, 33 ± 10 y; age range, 18–50 y) were included in this study; subject details are given in Table 1. All volunteers were prescreened before inclusion in the study. Medical history was taken, concurrent medications assessed, physical examination performed, vital signs checked, 12-lead electrocardiography performed, and blood and urine sampled to check for hematologic and clinical pathologic abnormalities and for substance abuse. Anyone with evidence of clinical disease, a history of organ-removal surgery (e.g., cholecystectomy, hysterectomy, or splenectomy), or a history of substance abuse was excluded. In women, pregnancy was ruled out. Volunteers were asked not to consume xanthines (e.g., caffeine or aminophylline) for at least 12 h before the test because these agents directly block the effects of adenosine and dipyridamole. The protocol was approved by the Johns Hopkins Institutional Review Board. All volunteers gave written informed consent.

⁸²Sr/⁸²Rb Generator

The commercially available ⁸²Sr/⁸²Rb generator, CardioGen-82, was used to elute ⁸²Rb activity. Details of the quality control procedures for the generator and cross calibration of end-of-infusion activity between the generator and the PET/CT scanner are described elsewhere (2).

PET/CT

All ⁸²Rb PET/CT (Discovery Rx; GE Healthcare) acquisitions were performed in 2-dimensional mode. The 76.38-s half-life of ⁸²Rb and the corresponding requirement for pharmacologic vasodilation require that image acquisition be completed within 10 min of the ⁸²Rb infusion. The following protocol was used to collect the pharmacokinetic data needed for dosimetry while remaining within the required constraints. First, a whole-body transmission CT scan (120 kVp; 20–200 mA, automatically adjusted; pitch, 0.5; and rotation time, 0.5 s) was obtained. A series of 3 whole-body PET scans was subsequently acquired after an infusion of 1,534 ± 118 MBq of ⁸²Rb at rest. Stress was induced by administration of the vasodilator, dipyridamole. Four minutes after a 4-min infusion of dipyridamole (0.56 mg/kg), about 1,502 ± 158 MBq of ⁸²Rb were infused and 3 serial whole-body PET scans were acquired. The heart rate, blood pressure, and electrocardiogram were monitored continuously throughout pharmacologic vasodilation. Each whole-body PET scan was obtained with a maximum of six 20-s bed positions starting at the level of the femurs and extending to the base of the skull. The 3 serially collected whole-body PET scans were completed in approximately 9 min. The mean infusion time for the ⁸²Rb activity from the ⁸²Sr/⁸²Rb generator was 25 s (range, 18–32 s). The first scan was started from the base of the femurs 10 s after the end of ⁸²Rb infusion. Attenuation-corrected whole-body PET images were reconstructed for subsequent region-of-interest analysis.

Source-Organ Contouring

All PET/CT images were transferred into MIMvista image analysis software (version 4.2; MIMvista Corp.). The source organs were delineated on the CT images with the help of fused PET/CT images. Except for muscle, thymus, and bone (cortical and trabecular), all the source organs expected for dose estimation were delineated. The gastrointestinal tract source organs were delineated into 4 regions: stomach contents, small intestine contents, upper large intestine (ascending and transverse colon) contents, and lower large intestine (descending and sigmoid colon) contents. The contour for heart contents (blood pool in cavities) was obtained using the difference between contours drawn for the whole heart and those drawn for the heart wall (visualized ventricular myocardium). Red bone marrow was delineated on the femur bone. Each bed position in the whole-body PET scan had a different factor applied for decay correction. To obtain the uncorrected counts (Bq/cm³) for each source organ at each time point, the source-organ contours that spanned more than 1 bed position were split into 2 or more contours according to their bed positions in the whole-body PET scan.

Dosimetry

The MIRD Committee schema (4) as implemented in the OLINDA/EXM 1.0 software (5) was used to perform the absorbed dose calculations. The time-integrated activity coefficient (TIAC) (or residence time) is required as input for these calculations. This coefficient is obtained by integrating the time-dependent activity curve for each source tissue. The ultra-short half-life of ⁸²Rb poses a challenge for obtaining such curves over the whole body. Usually, with longer-lived tracers dosimetry studies use serial whole-body images for measurement of organ uptake and clearance. The rapid decay of ⁸²Rb precludes imaging for more than 10 min after a single administration of activity and thus limits the number and count statistics of serial whole-body images. A maximum of 3 serial whole-body images could be obtained while maintaining adequate count statistics. A repeated-injection protocol as used in our previous study (2) could not be implemented because the imaging time would have exceeded the duration of pharmacologic vasodilation. Therefore, to obtain the pharmacokinetic data for dosimetry calculations under stress conditions, we used the 3-time-point whole-body rest–stress PET scans to obtain a ratio of stress to rest TIAC. This ratio was used to scale the mean TIAC obtained from the multibed multiinjection method used in the rest-only studies (2). The validity of this methodology was checked by comparing the TIAC obtained at rest from the 3-time-point study with that obtained from the detailed biokinetics (rest) study (2).

This approach was implemented as follows. Contours delineated on whole-body CT as described earlier in the section on source-organ contouring were overlaid onto each whole-body rest–stress PET image to extract activity concentrations. Organ volumes from CT were converted to mass using published density values (6,7). The mass of organ contents was obtained by multiplying the volume by the CT-measured average density. Decay corrections applied during image reconstructions were reversed using decay factors from image headers. The administered activities (A₀) listed in Table 1 were the end-of-infusion values obtained from the ⁸²Rb generator, after correcting for the cross-calibration factor between the PET/CT and the ⁸²Rb generator output as described previously (2). For each source organ (r_S), a monoexponential expression was fit to the time-dependent activity curves (3 time points) obtained from the matched whole-body rest–stress (RS) studies. The fitted expressions were analytically integrated to give the time-integrated activity (TIA) for rest (R) and also for stress (S), (designated A˜R,RS(rS) and A˜S,RS(rS), respectively). These values were divided by the administered activity to give the TIAC (analogous to residence time and designated a˜R,RS(rS) and a˜S,RS(rS), respectively). The ratio of stress to rest TIAC was used to scale the mean rest TIAC concentration ([a˜R(rS)]) obtained from the previously reported rest-only study for which we had detailed pharmacokinetic data (2). The resulting scaled rest-only TIAC concentration was set equal to the stress TIAC concentration used in the dosimetry calculations. This approach assumes that the administered activity–normalized stress-to-rest ratio observed at the 3 measured time points applies to all time points for all organs and that the shape of the mean normalized rest time–activity curve obtained from the previous study (2) is representative of all normalized rest curves. The expression for obtaining the stress TIAC concentration ([a˜S(rS)]) is shown below:[a˜S(rS)]=[a˜R(rS)]⋅a˜S,RS(rS)a˜R,RS(rS).Eq. 1To confirm that scaling in this manner is reasonable, we compared the TIAC concentration obtained from the rest curve of the 3-time-point rest–stress study, a˜R,RS(rS), with that obtained from the previous rest-only study, [a˜R(rS)]. Because it was not possible to resolve the rising portion of the time-dependent activity curve in the 3-time-point rest–stress study, we corrected for this factor as shown in Equation 2 before making the comparison. The correction essentially replaces the back-extrapolated portion of the RS curve with the rising portion obtained from the mean rest-only time-dependent activity curve:[a˜′R,RS(rS)]=∫0∞[aR,RS(rS,t)]−∫0T[aR,RS(rS,t)]+∫0T[aR(rS,t)].Eq. 2The first integral in Equation 2 is [a˜R,RS(rS)]; the last integral is calculated by the trapezoidal method. A graphical illustration of Equation 2 is shown in Figure 1.

For each subject, the TIAC concentrations obtained from Equation 1 were multiplied by the standard source-organ mass and were input into OLINDA/EXM 1.0 (5) for dose estimation. Because no activity was excreted by the subjects during imaging, the TIAC for the remainder of the body was calculated by subtracting the sum of the TIAC of all source organs from the reciprocal of the ⁸²Rb physical decay constant.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Graphical representation of Equation 2. Shaded regions in A, B, C, and D show respectively, where T represents peak uptake time.

⁸²Rb activity observed in the gastrointestinal tract source organs (stomach, small intestine, upper large intestine, and lower large intestine) is due to blood flow to these organ walls. OLINDA/EXM 1.0 calculates the gastrointestinal tract (stomach, small intestine, upper large intestine, and lower large intestine) wall doses by assuming that the activity is in the contents. Therefore, we corrected the OLINDA/EXM 1.0 dose estimates for gastrointestinal tract walls so that the estimates corresponded to wall self-absorbed doses and not absorbed doses from contents to the wall. The correction was implemented as described previously (2).