Whole-Body Biodistribution and Radiation Dosimetry of ¹⁸F-GE067: A Radioligand for In Vivo Brain Amyloid Imaging

Healthy Subjects

Healthy subjects were recruited as part of a phase I PET study sponsored by GE Healthcare Ltd. and in response to an advertisement in a local community newspaper. Six healthy Caucasian subjects (5 men; 1 woman; age range, 51–74 y) were included in the study. Demographic data for these subjects are summarized in Table 1. The subjects did not have any clinically significant medical or neurologic condition, and they did not have any clinically significant abnormality on physical, neurologic, or laboratory examinations. In addition, they did not have a first-degree relative with a diagnosis of AD. The study was approved by the local Ethics Committee, and written informed consent was obtained from all the volunteers before the study started.

Radiotracer Characteristics and Preparation

The investigational medicinal product ¹⁸F-GE067 was batch-manufactured according to Good Manufacturing Guidelines, Eudralex volume 4, Annex 3 (11), at the Cyclotron Research Centre, University of Liege, Belgium, using a TracerLab-FX_F-N (GE Healthcare) chemistry platform. The chemical and radiochemical purity of the test item was assessed by analytic high-performance liquid chromatography system equipped with an ultraviolet and radioactivity detector. The total unlabeled GE067 chemical content was 0.7 ± 0.3 μg/mL, and the radiochemical purity was 98.1% ± 1.2%. The identity of the radiolabeled product was confirmed by coelution on high-performance liquid chromatography of a cold analytic standard of GE067.

¹⁸F-GE067 was supplied as a ready-to-inject solution with a maximum of 10 μg of cold GE067 per dose; the solution for injection contained 7% ethanol (v/v) and 0.5% polysorbate 80 (w/v) in phosphate buffer (0.015 M).

PET/CT Procedure

Subjects received on average 121 MBq (range, 96–147 MBq) of ¹⁸F-GE067 as a slow intravenous bolus injection (∼10 s) through a catheter in an antecubital vein. A dynamic series of whole-body PET emission scans was obtained as described previously (12).

All subjects were scanned with a PET/CT camera (HiRez Biograph16; Siemens) at 3 consecutive times. For the first 90 min, 8 sequential whole-body scans from the midfemoral position to the head were obtained. During the other 2 times, single whole-body scans were obtained at about 160 and 260 min after injection. A low-dose (tube potential, 80 kV; effective mAs, 11), no-contrast, whole-body CT scan was obtained at the beginning of each time. The additional radiation burden of this low-dose CT was 0.5 mSv per whole-body scan.

Urine samples were collected and counted between scanning times, up to 6 h after injection (n = 5/subject).

Images from the HiRez Biograph16 were reconstructed using a 3-dimensional ordered-subset expectation maximization iterative algorithm (5 iterations and 8 subsets) and postsmoothed with a 3-dimensional gaussian filter (6-mm full width at half maximum). Reconstruction included model-based scatter as well as attenuation correction based on the measured CT-attenuation map.

Volumes of Interest (VOI)

Data transfer, VOI definition, and processing to calculate time–activity curves were performed as described previously using PMOD software (version 2.9; PMOD Inc.) (12). The following organs showed significant tracer accumulation and were included in the VOI template: brain, gallbladder, small intestine, heart, kidneys, liver, lungs, axial bone marrow, spleen, thyroid, and urinary bladder.

Residence Time and Absorbed Dose Calculations

Residence times were computed and normalized to injected activities by calculating the area under the time–activity curve of each organ. This was done using curve fitting to the time–activity curves as the primary method. The uptake in the brain, heart, liver, kidneys, red marrow, lungs, spleen, and remainder could be modeled with the following function:where i = 1–3, A_i was a scaling parameter, τ_u,i represented the uptake time parameters, and τ_e,i represented the excretion time parameters. The model was fitted to the data by least-squares constraint minimization. Small intestine time–activity curves for 2 subjects and the gallbladder and thyroid time–activity curves in all subjects were not satisfactorily described by kinetic modeling. The curve fitting for these subjects and organs did not converge to an acceptable solution, even by changing the start estimates over a large representative range. For the intestine, accumulated activity entering the intestines at the last 2 times did not approximate the final fraction closely enough for unique fitting solutions. For the gallbladder, large variability in the measured values was encountered. In these subjects and organs, values based on the trapezoid rule were used, and the area under the curve after the acquisition of the last image was calculated by assuming only physical decay of ¹⁸F and no additional biologic clearance to be conservative. Previously, we have shown that the calculation of the area under the curve using the trapezoid method gives a good approximation of the residence time (12).

The absorbed doses reported were calculated using OLINDA (Vanderbilt University), and the effective dose (ED) was calculated from the individual organ doses on the basis of a predefined weighting factor for each of the source organs (13).

The residence times for the upper large intestine (ULI), lower large intestine (LLI), and small intestine were calculated in 2 ways. For the 4 subjects in whom the fraction of injected activity entering the small intestine could be estimated accurately, we used the International Commission on Radiological Protection (ICRP) 30 (14) gastrointestinal model as incorporated in OLINDA, using the fraction entering the small intestine as input (calculated from the kinetic modeling of the intestinal VOI time–activity curves). This value was set equal to the largest decay-corrected fraction of injected activity encompassed by the intestinal region of interest.

For the 2 subjects in whom no accurate estimate of the fraction injected activity could be made, we also used a predefined distribution of residence times—50% for the small intestine, 25% for the ULI, and 25% for the LLI—which is a good approximation for slowly excreted tracers (12). For the 4 subjects in whom small, ULI, and LLI parameter fitting was possible, the average retention time calculated with the trapezoid method was slightly lower than the kinetically modeled values and resulted in ED estimates that were 5% lower, indicating that the trapezoid approach is appropriate for quantifying organ ¹⁸F-GE067 uptake.

The updated dynamic bladder model (13) was used to calculate dose to the bladder wall, with a bladder voiding interval of 2 h. The fraction of activity excreted through the bladder and the corresponding biologic half-life were based on the accumulated activity excreted at all voiding moments between and after scans. For this purpose A × (1 − exp(−ln(2) × t/τ_u)) was fitted to the cumulative activity values, where τ_u was the biologic decay and A was the fraction of activity released from the body during voiding between scan sequences.