Animal model
The regional Ethics Committee granted approval to conduct this study (no. N/02-10-09/18/10-12). All procedures performed in this study involving animals were in accordance with the ethical standards of the regional research committee and with the 1964 Helsinki Declaration and its later amendments.
In two experiments, 16 4-week-old female nude rats (Harlan Laboratories, Indianapolis, IN, USA) bearing subcutaneous SKOV3 human ovarian tumors were used. Ten rats were treated with BEZ-235, a dual Phosphoinositide 3-kinase/mammalian target of rapamycin inhibitor (Selleck Chemicals, Houston, TX, USA) from day 0 to day 3, with treatment being discontinued on day 4. Six rats were used as controls. Each animal had four tumors: two in the shoulders and two in the upper thighs, providing 64 lesions for evaluation.
In the first experiment, five rats underwent PET imaging on days 0, 3, and 7. This cohort consisted of two groups: untreated controls (n = 2) and treated rats (n = 3). A second cohort was treated at the same time points and used for molecular analysis. In the second experiment, a cohort of 11 animals, consisting of untreated controls (n = 7) and rats receiving BEZ-235 (n = 4), was used for both PET imaging and molecular analysis. For the latter aim, one rat each from the treated and untreated groups was sacrificed on days 0, 3, and 4, and all remaining animals were sacrificed after the last PET examination had been performed on day 5. For general anesthesia, heated inhaled isoflurane was administered with an anesthesia device dedicated to small animals (Minerve, France).
Molecular analyses were performed in both experiments, including cell proliferation assessment (Ki-67 immunostaining), and Phosphoinositide 3-kinase/mammalian target of rapamycin target expression studies (p4E-BP1 immunostaining), which were taken as gold standards for therapeutic assessment in the present work.
Cross-calibration
A cross-calibration among the SA-PET/CT system and the dose calibrator was performed. A 10-MBq 18F-FDG solution (as assessed by the dose calibrator) was used to fill a vial of exact known volume, which resulted in a solution of known concentration. This solution was used to fill a cylinder phantom that was scanned for 20 min on the SA-PET/CT scanner. A large volume of interest (VOI) was used to determine the mean activity concentration as assessed by the SA-PET/CT scanner. Cross-calibration factors were then derived and used to synchronize counts/measurements for the two pieces of equipment.
SA-PET/CT acquisitions, reconstructions, and analysis
Animals were kept fasting for 6 h. As detailed above, SA-PET/CT (Inveon system, Siemens Medical Solution, Knoxville, TN, USA) examinations were performed on days 0, 3, and 7 in the first experiment and days 0, 3, 4, and 5 in the second experiment. The same individual (NA, with a 15-year experience in tail vein injection) performed tracer injections. Injections were performed intravenously through the tail vein, under general anesthesia, using a 29-gauge needle. Injected volume was always kept below 0.4 mL. The average calibrated activities were 38 ± 7 and 39 ± 5 MBq with uptake times of 91 ± 8 and 105 ± 14 min following injection, respectively, for the first and second experiments. Animals were imaged in prone position with the tail positioned on their right side. Reconstructions were performed using a NEMA NU 4-Optimized Maximum A Posteriori (MAP) Reconstruction [14] with scatter and attenuation corrections.
SA-PET/CT analysis
Intravenous injections were qualitatively evaluated from reconstructed images. When the observer concluded to the presence of extravasation, a three-dimensional volume of interest (VOI) with a visually adapted isocontour was drawn over the tail injection site (Fig. 1). Activity in the tail (Bq and Bq/cc) was recorded. Two independent observers made these analyses to determine the inter-observer variability. The activity of tail vein injection extravasation was corrected for decay assuming that there was no interstitial absorption of the tracer in the tail between the injection and the imaging session. Corrected injected activity was calculated by the following formula:
$$ \begin{array}{l}\mathrm{Corrected}\ \mathrm{injected}\ \mathrm{activity}=\\ {}\mathrm{calibrated}\ \mathrm{activity}-\mathrm{decay}\ \mathrm{corrected}\ \mathrm{residual}\ \mathrm{syringe}\ \mathrm{activity}-\mathrm{decay}\ \mathrm{corrected}\ \mathrm{tail}\ \mathrm{extravasated}\ \mathrm{activity}\end{array} $$
For therapeutic response assessment in each experiment, a cylindrical VOI was drawn over each tumor. The uncorrected and corrected mean pixel values were extracted for each VOI and SUVmean were computed as follows, assuming a 1 g/mL density:
$$ \mathrm{S}\mathrm{U}\mathrm{V}=\frac{\mathrm{tumor}\ \mathrm{activity}\kern0.5em \left(\frac{\mathrm{Bq}}{\mathrm{mL}}\right) \times \mathrm{body}\ \mathrm{weight}\kern0.5em \left(\mathrm{g}\right)}{\mathrm{injected}\ \mathrm{dose}\ \left(\mathrm{Bq}\right)} $$
Data were processed with a dedicated Siemens station (Inveon Research Workplace 2.2, Siemens Molecular Imaging, Knoxville, TN, USA).
Statistical analysis
These analyses aimed to determine (1) if tail vein extravasation corrected data performed better than uncorrected data for predicting target inhibition and (2) if there was inter-observer variability in the tail vein injection correction process. Statistical analyses were performed on a per-lesion basis without taking into account intra-rat correlation. Fisher’s exact probability tests, with the Freeman–Halton extension when necessary [15], were performed to test if the occurrence of tail vein extravasation was different according to groups (treated or not) and protocol time points. Inter-observer agreement was assessed by means of Cohen’s kappa coefficients (qualitative analysis) and Lin’s concordance coefficient (quantitative analysis). Tail vein activity measurements (Bq) of observers A and B were compared using Bland–Altman method comparisons. Wilcoxon tests and Bland–Altman plots were used for paired comparisons of corrected and uncorrected data amongst treated and control groups. To assess the ability to discriminate between control and treated groups, relative SUVmean values of day 0 for each group were compared over the different time points (day 3, 4, 5, and 7), for corrected and uncorrected data, using Mann–Whitney tests. For each test, the alpha risk was set at 0.05. Statistical analysis, graphs, and plots were performed with GraphPad Prism version 5.0 for Mac (GraphPad Software, La Jolla, CA, USA; www.graphpad.com).