Unless otherwise stated, all numerical data are provided as the mean ± 1 standard deviation.
Radiopharmaceutical preparation
The target compound [
18F]AH113804 was prepared in a two-step route starting with synthesis and purification of 4-[
18F]fluorobenzaldehyde from 4-(trimethylammonium)benzaldehyde triflate, followed by the reaction of the labelled aldehyde with the deprotected peptide precursor (Fig.
1).
[18F]Fluoride anion was produced via the 18O(p,n)18F nuclear reaction using a Scanditronix MC17 cyclotron. The bombardment of enriched 18O water (98%, Rotem) gave an aqueous solution of 18F− which was transferred from the cyclotron target using helium. The 18F− (30–50 GBq) was trapped on a QMA column and eluted with kryptofix K222 and aqueous potassium hydrogen carbonate in acetonitrile (5.1 mg K222, 320 μL MeCN, 1.4 mg KHCO3, 80 μL water). This mixture was dried at 105–120 °C for 9.2 min. The aldehyde precursor (3.8 mg) dissolved in 2 mL of dry DMSO was added to the dried 18F− kryptofix complex. The reaction mixture was heated at 80 °C for 2 min. 4-[18F]Fluorobenzaldehyde was purified with an Oasis MCX plus extraction cartridge using 4 mL of 4% aqueous ammonia and eluted in 1 mL of ethanol.
Five-milligramme peptide in 2.2 mL of aniline hydrochloride (10 mg/mL) was added to the purified labelled aldehyde. After 6.9 min at 60 °C, the crude product was diluted with sterile water and purified on a preparative HPLC column (Waters Xbridge Shield, 5 μm, 10 × 100 mm column with MeCN in 50 mM ammonium acetate, flow 4 mL/min; 0–1 min 15% MeCN, 1–16 min gradient 15–40% MeCN, 16–20 min gradient 40–100% MeCN) connected to a Knauer UV detector and a Bioscan RAD detector. Radiosynthesis was carried out on a FASTlab synthesiser module (GE Healthcare) using a single-use disposable AH113804 (18F) Injection FASTlab cassette. The product was formulated in 21 mL of phosphate-buffered saline containing sodium p-aminobenzoate (2.38 mg/mL). Sterile filtration was done on a Fluorodyne® 25-mm syringe filter (0.2 μm) (Pall Corporation, USA). The activity of the product was approximately 250 MBq/mL. The tracer production, including the drying of the [18F]fluoride anion and formulation of the purified product, took about 1 h.
The product identification and purity check was done by analytical reversed-phase HPLC utilizing a Phenomenex Kinetex C18, 2.6 μm, 100 × 4.6 mm column equipped with a Security Guard Ultra C18 cartridge with 10 mM ammonium acetate buffer (A) and methanol/acetonitrile 70:30 (v/v) (B). The gradient method was as follows: 0–3 min 30% B, 3–5 min 30–40% B, 5–32 min 40% B.
Safety data
Safety data were collected up to 24 h after injection and included adverse events (AEs), vital signs (blood pressure, respiratory rate, heart rate and body temperature), physical examination (lungs, cardiovascular system and abdomen), electrocardiogram, laboratory parameters (serum biochemistry, haematology, coagulation parameters and urinalysis) and injection site status. Two-millilitre venous blood samples were collected through an indwelling catheter at a nominal 2-, 5-, 10-, 15-, 30-, 60-, 90-, 180- and 260-min post-injection, and 18F activity concentration in a single sample of whole blood and plasma was determined in a well counter (in-house design including 3MW3/3 detector, Saint-Gobain Crystals) that was cross-calibrated against the scanner and subject to daily quality control.
Image acquisition and reconstruction
Emission images were acquired in 3D mode on a GE Discovery ST PET/CT scanner with a 15.7-cm axial field of view (FOV). A whole-body CT image was acquired for attenuation correction prior to the administration of AH113804 (18F) Injection. The axial extent of the acquired emission images was from the crown of the head to approximately mid-thigh so as to ensure that the urinary bladder was included in the image. Hence, a whole-body emission scan consisted of contiguous static positions acquired at a variable number of bed positions, depending on the subject’s height. Any 18F activity outside the FOV was assumed to be uniformly distributed throughout the unimaged anatomy.
Eight serial 3D whole-body PET images were acquired for each subject beginning nominally at 2 min and ending up to 6 h post-injection (p.i.). In order to compensate for the physical decay of 18F activity during imaging, the acquisition time for each bed position was increased from 30 to 60 s at 26 min and to 120 s at 220 min p.i. Images were acquired in three separate sessions between which the subjects were allowed to leave the scanner bed. Prior to the second and third emission imaging sessions, additional whole-body CT attenuation correction scans were performed. Corrections for scatter events and random coincidences were performed as per the models provided by the scanner’s manufacturer.
Subjects were encouraged to void and were given a yoghurt drink between imaging sessions to prompt gallbladder drainage into the duodenum. All urinary voids were collected and the volumes and 18F activity concentrations measured.
Emission images were reconstructed with both ordered subset expectation maximisation (OSEM) with 2 iterations and 21 subsets with post-reconstruction smoothing using an isotropic Gaussian filter (4.29 mm full-width at half-maximum) and filtered back-projection (FBP). Slice thickness was 3.27 mm with a pixel size of 3.91 mm in the OSEM reconstruction and 5.47 mm in the FBP.
Quantification of activity
Image analysis was performed on a MIM workstation (version 6.0, MIM Software Inc., Cleveland) which includes a tool for volume of interest (VOI) definition using a predefined VOI template and an image registration algorithm. A set of VOIs were initially drawn around organs that could be readily delineated on the CT component of the PET/CT scan. These organs included the brain, salivary glands, thyroid, lungs, heart, liver, spleen and kidneys. The defined VOIs were registered to a common CT template using a non-rigid deformation algorithm and stored in the MIM database. This database was augmented as delineation was performed on successive subjects.
Analysis regions were defined using the templatized VOI database and were then manually edited using the fused CT and OSEM PET images as a guide. Having defined a VOI set for imaging session 1, these were translated as required and applied to sessions 2 and 3. Finally, VOIs were applied to the FBP data and the resulting mean activity concentration (Bq/mL) per region, the region volume (mL) and the standard deviation of counts within each region at each time point were exported in spreadsheet format. It was assumed that organs did not change in size or shape during the course of the acquisition with the notable exceptions of the bladder and the intestinal contents.
As the contents of the small intestine varied with time, VOIs were drawn on each time frame, and where no significant uptake could be identified, the VOI from the nearest available time point was copied to provide a measure of background activity. Background activity was included in the fitted model and subtracted from the resulting calculation of normalised cumulated activity and injected fraction entering the small intestine.
The activity within the cardiac chambers was estimated from the product of the measured whole blood activity and the chamber volume (477 mL for males, 351 mL for females representing 9% of the total blood volume) [
6]. The activity within the cardiac wall was then estimated by subtracting the activity within the cardiac chambers from the activity within a whole heart VOI.
The presence of significant artefacts in the region of the bladder on both the FBP and OSEM reconstructed images complicated the analysis and bladder activity was estimated in two ways.
The first method used a 42% maximum intensity threshold VOI drawn on each time frame. The second method assumed the bladder was the only significant source within a rectangular VOI positioned to include the whole FOV over the maximum axial extent of the bladder. This whole-slice VOI was not resized as the bladder filled to ensure that the background component of the measured activity was as constant as possible. The change in background activity with time as a consequence of biological clearance was considered insignificant compared to the change in bladder activity per se. This background activity was then accounted for by including a constant term in the model used for curve fitting.
Administered activity not accounted for by the defined VOIs or excretion was assigned to the remainder category.
Measured activity data for each source region,
r
s, were decay-corrected to the time of injection and normalised to the administered activity. These data were then fitted to the generalised analytical function of Eq.
1,
$$ {A}_{r_{\mathrm{S}}}^{\mathrm{Corr},\mathrm{Norm}}(t)={C}_{r_{\mathrm{S}}}+{\displaystyle \sum_{i=1}^N{\alpha}_{r_{\mathrm{S}},i}\;{e}^{-{\lambda}_{r_{\mathrm{S}},i}t}} $$
(1)
where
\( {\alpha}_{r_{\mathrm{S}},i} \) and
\( {\lambda}_{r_{\mathrm{S}},i} \) are parameters extracted from a Simplex (GRG Nonlinear Solver, Microsoft Excel) fit minimising the weighted sum of squared difference between the model and the biodistribution data. The constant term,
\( {C}_{r_{\mathrm{S}}} \), was either fixed to zero or fitted alongside the other parameters. Mono- (
N = 1) and bi-exponential (
N = 2) fits were performed with and without the inclusion of the constant background term.
The selection of the appropriate equation to fit to the data was made using the Akaike Information Criterion (AIC) [
7]. Fitted functions were subsequently integrated, after first accounting for the effect of
18F physical decay, to yield the normalised cumulative activities (NCAs) of the VOI [
8]. The NCA of the urinary bladder contents and voided urine was calculated from the analytical fit to the summed activities in the urinary bladder contents and voided urine using a dynamic urinary bladder model [
9]. As recommended by the International Commission on Radiological Protection (ICRP), a 3.5-h voiding interval was assumed [
10].
Internal radiation dosimetry
The internal radiation dosimetry for each subject was determined following the Medical Internal Radiation Dose (MIRD) schema [
11]. For each subject, the NCAs were used as input to the Organ Level Internal Dose Assessment/Exponential Modelling (OLINDA/EXM) software [
12] to calculate the absorbed doses to the 24 MIRD-specified target regions of the Cristy-Eckerman adult hermaphrodite male and adult female phantoms [
13].
Following the recommendations of Publication 103 of the ICRP [
14], these absorbed doses were then sex-averaged and the effective dose was evaluated using the tissue weighting factors of Publication 60 of the ICRP [
15]. Recommendations of the ICRP subsequent to this publication were followed in the effective dose evaluation: the absorbed dose to the thymus gland was used as a surrogate for that to the oesophagus; the absorbed dose to the colon wall was calculated as the mass-weighted sum of the absorbed doses to the walls of the upper and lower large intestines; and the gonadal absorbed dose was taken to be the mean of the absorbed doses to the testes and ovaries [
16].