The study was approved by the Regional Ethics Committee in Stockholm and the Radiation Protection Committee at the Karolinska University Hospital, Stockholm. Written informed consent was obtained from each subject.
Subjects and design
Study participants (inclusion criteria: men, 20–50 years of age) were recruited at the PET Centre, Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden. Subjects were healthy according to medical history, clinical examination, and routine laboratory blood and urine tests. No medications were used at the time of the study. PET examinations were carried out at the Department of Nuclear Medicine, Karolinska University Hospital, Solna. Design of the study period included two imaging sessions 1–4 weeks apart. In each session a low-dose CT of the chest was followed by a PET measurement using the radioligand [11C]VC-002. The two sessions were performed at the same time of the day. To capture any possible adverse event the study was completed by a follow-up telephone call within 1 week after the last session.
Radiochemistry and PET/CT imaging procedures
[11C]VC-002 was prepared as previously described . The radiochemical purity of [11C]VC-002 exceeded 99% at time of injection. The mean radioactivity per single injection was 220 MBq (SD ± 20 MBq, range 188–239 MBq, N = 7) and the molar activity was 370 GBq/micromole (SD ± 209 GBq/micromole, range 124–737 GBq/micromole). The calculated mean mass of the radioligand injected per single measurement was 0.29 μg (SD ± 0.19, range 0.1–0.62 μg, N = 7). At such “tracer dose” conditions there is no significant effect of chemical mass on the radioligand binding parameters.
As part of preparatory activities before imaging an arterial cannula was inserted in the radial artery of one arm and a venous cannula was inserted in each arm. The arterial cannula was used for arterial blood sampling during PET measurement. One venous cannula was used for bolus injection of radioligand (duration of injection was 10 s), whereas the other venous cannula was used for venous blood sampling for comparisons between venous and arterial data (data not shown).
The imaging measurements were performed using a GE Discovery PET/CT 710 system. The subjects were positioned in supine position. Initially, a low-dose CT scan (7.5 mAs, 120 kVp) of the chest was performed for attenuation and scatter correction purposes. This was followed by I.V. bolus injection of the radioligand [11C]VC-002 dissolved in a sterile physiological phosphate buffer (pH 7.4). The cannula was immediately flushed with 10 ml. saline. Emission data were acquired in 3D list mode with no respiratory gating information and were reconstructed into a consecutive time-series of 3D PET images with the following sequence of time frames: 9 × 10 s, 2 × 15 s, 3 × 20 s, 4 × 30 s, 4 × 60 s, 4 × 180 s, and 12 × 360 s, i.e. 38 frames with a total duration of 93 min. The reconstructed 4D PET image was corrected for subject movement in a retrospective, image-data-driven process whereby inter-frame subject motion was estimated in the reconstructed images. The procedure is described in detail in the supplemental material.
Arterial blood sampling
To obtain an arterial input function, an automated blood sampling system (Allog, Sweden) was used to collect blood samples continuously during the first 10 min of each PET measurement. Furthermore, arterial blood samples (2 ml) were drawn manually at approximately 2.5, 5, 7.5, 10, 15, 20, 30, 45, and 90 min after [11C]VC-002 injection. Note that manual samples were taken already during automated blood sampling to have early information also on plasma radioactivity and radiometabolites. Radioactivity in 1 ml of the manually drawn samples was then immediately measured for 10 s in a well counter cross-calibrated with the PET system. After centrifugation, 0.2 ml plasma was pipetted, and plasma radioactivity was measured in a well counter. Blood samples for the measurement of radiometabolites were drawn at approximately 5, 10, 20, 30, 45, and 60 min in five of the PET-measurements and up to 30–45 min in the remaining eight PET-measurements.
Plasma radiometabolite analysis of [11C]VC-002
The fraction of plasma radioactivity corresponding to unchanged radioligand in plasma was determined as has been previously described for other PET radioligands . Briefly, the plasma samples were deproteinized with acetonitrile and analyzed by high-performance liquid chromatography (HPLC) with radiodetection. In detail, an ACE C-18 HPLC column (ACE, 5 μm, 50 × 250 mm) was eluted at 5 ml/min with a mixture of acetonitrile (A) and aqueous ammonium formate (0.1 M) (B) according to the following gradient: 0–4.0 min (A/B) 40:60; 4.0–6.0 min (A/B) 80:20; 6.1–7.0 min (A/B) 90:10; 7.1–8.1 min (A/B) 40:60.
Image processing and analysis
Image data processing and analysis consisted of the following overall steps: (1) PET and CT image pre-processing, (2) automated delineation of lung and other regions of interest, (3) derivation of the arterial plasma input curve, (4) inspection of time curves in blood and lungs, (5) quantitative analysis of radioligand binding in lungs, (6) evaluation of test-retest repeatability of binding parameters.
For each PET measurement, the series of consecutive images were integrated to obtain a summation PET image showing average radioactivity concentration during the entire measurement (0–93 min). To control for within subject positioning differences between baseline (test) and follow-up (retest) measurements, the summation PET images were used to obtain co-registration parameters. Images of the follow-up measurements were subsequently resliced using these co-registration parameters (see below). In preparation for combined PET-CT image processing the CT images were resliced from the original 0.98 × 0.98 × 3.27 mm voxel size to the 3.65 × 3.65 × 3.27 mm voxel size of the PET images.
Delineation of regions of interest (ROI)
For each PET/CT measurement, ROIs corresponding to the lungs were delineated through an automated procedure using both the PET and the resliced CT images. A detailed description of the delineation procedure is included in the supplemental material.
Derivation of arterial plasma input curve
The ROI of the aortic arch was applied to extract the time-(radio)activity curve (TAC) representing radioactivity in whole blood. The curve was interpolated to 1 s resolution. In addition, the whole blood and plasma radioactivity concentration, measured in the collected manual arterial samples, was used to calculate the ratio of plasma to whole blood radioactivity. Preliminary evaluation indicated that this ratio was essentially constant throughout the PET measurement. By consequence, the average ratio across all time points with manual blood sampling was used to multiply the continuous, interpolated image-derived whole blood TAC with the average plasma-over-blood ratio to obtain the TAC of total radioactivity in plasma to be used as input function for quantification. Worth noting is that radiometabolite correction was not performed, i.e. the total plasma TAC was used in the quantification of radioligand binding based on the assumption that radioactively labelled metabolites also enter the lung tissue .
Description of the time curves for [11C]VC-002 in blood and lungs
The basic examination of [11C]VC-002 binding in lungs involved visual inspection of the distribution of radioactivity, inspection of TACs in arterial whole blood, plasma, and lungs, as well as calculation of the ratio between the TACs of lung and plasma, or, between plasma and whole blood. Besides the thereby obtained plasma-to-blood ratios, estimates based on measured haematocrit values were also obtained and compared to the ones based on direct plasma measurements. Additionally, the metabolism of [11C]VC-002 was assessed from the results of HPLC analysis of arterial plasma.
Quantitative analyses of [11C]VC-002 binding
The lung ROI was applied to the series of PET images to extract the corresponding TAC. The TAC was analysed and interpreted according to a compartment model describing the pharmacokinetics of radioligand in lung tissue. For details on the underlying compartment model used in lungs please refer to the supplement that also shows a schematic view of the model (see Supplemental Fig. 5). The main outcome measure was the total distribution volume (VT), which is equivalent to the ratio of the radioactivity concentration in the lung to the total plasma concentration at equilibrium. VT has the unit of ml/cm3: i.e. it gives the volume of blood (in milliliter) required to match the amount of radioligand molecules present in one unit of the lung volume (in cm3). Multiplying the VT value with the total volume of the target organ will give the milliliter of plasma necessary to account for the total amount of radioligand in that organ. Total binding (VT) of [11C]VC-002 in lung was estimated using the one- and two-tissue compartmental model (1TCM and 2TCM, respectively), and the multi-linear variant of Logan’s linear graphical approach (MLLogan) was employed [19,20,21]. The 1TCM provided estimates of 2 kinetic rate constants (K1, k2) and VT was derived using the expression VT = K1/k2. The 2TCM provided 4 kinetic rate constants (K1, k2, k3, k4) and VT was derived using the expression VT = K1/k2 × (1 + k3/k4).
To account for the radioactivity contained in blood vessels in lungs the whole lung TACs were fitted using a version of 2TCM that, besides the kinetic rate constants also estimated the fractional blood content (Vb) as well as the time-shift between blood/plasma and tissue radioactivity curves. Subsequently, Vb and the time-shifted input curve was used to match the tissue signal. Note that the PET images and/or TACs were not corrected for fractional air content since respiratory gating information and the necessary specific CT sequences were not acquired.
Besides ROI-based quantification, binding parameters were also obtained at the voxel level. For this purpose, data-driven estimation of parametric images based on compartmental theory (DEPICT) was employed as described in the literature and using the software toolbox available from the author . Traditional kinetic modelling, such as the 1TCM and 2TCM described above, operates on the basis of a priori fixed number of tissue compartments to describe the observational data. In contrast, DEPICT does not work with an a priori fixed number of kinetic tissue compartments but estimates the optimal number of compartments, referred to as model order (MO), from the data itself; hence, it is a data-driven approach. DEPICT allows for a more detailed analysis of lung data that exhibits inhomogeneity in terms of tissue composition, fractional air and blood content. The approach was confined to perform the voxel-wise estimation only in voxels within the body yet excluding those in or near the heart and liver (for details, see the “Description of the derivation of ROI masks” in the supplemental section). The range of the exponents used for the derivation of the basis function table for DEPICT were logarithmically spaced between 0.0034 1/min (i.e. 10% of the isotope half-life for 11C) and 0.6 1/min . The DEPICT approach provided detailed parametric images of the fractional blood content (Vb), the rate constant describing transfer from plasma to tissue (K1), total binding (VT), as well as the MO for each voxel. Finally, the lung ROIs were applied to the parametric images to obtain the mean (or in case of MO, the median) parameter for all voxels within the organ.
Evaluation of repeatability of binding estimates
Absolute variability (VAR) was calculated as the difference between the test and retest parameters (e.g. VT) values divided by the mean of these two and multiplied by 100 to get percentage values . The VAR was first calculated for each subject and then summarized across subjects to obtain descriptive statistic parameters. In addition, the intra-class correlation coefficient (ICC) was calculated .
Evaluation of the time-stability of binding estimates
The time stability of [11C]VC-002 binding parameters, and test-retest repeatability of VT was calculated for different durations of the TAC data entered into the quantification. The values for the total, 90-min data were used as a point of reference to evaluate results for shorter durations. VAR was summarized across subjects using median statistic due to the low number of subjects.
Descriptive statistics, quantitative analyses and computations for test-retest analysis were performed using the MATLAB, version R2014b (www.mathworks.com). The Akaike information criterion  and F test were used to identify the compartment model that provided the statistically preferred interpretation of the data. The F test was 1-tailed (right). Pearson’s correlation coefficient was used to analyse the association between VT values obtained by different approaches. In all analyses, the statistical significance (alpha level) was set at p = 0.05.