Influence of Animal Heating on PET Imaging Quantification and Kinetics: Biodistribution of ¹⁸F-Tetrafluoroborate and ¹⁸F-FDG in Mice

Impact of Ambient Temperature on Physiologic Parameters

To study the impact of ambient temperature on heart rate, respiratory rate, and body temperature, we first studied 3 groups of 6 normal female BALB/c nude mice (body mass, 22.3 ± 0.7 [mean ± SD] g). Animals were maintained under continuous anesthesia inside a dedicated animal holder (Minerve imaging cell; Minerve SAS), and their physiologic parameters were continuously monitored during the entire procedure. Heart rate was measured with 3 carbon-tube electrodes positioned on the animals’ extremities (12), and body temperature was assessed with a dedicated intrarectal probe. The anesthesia gas mixture was kept constant to avoid any influence of the isoflurane concentration on the animals’ homeostatic capabilities. All groups were maintained under a mixture of 1.5% isoflurane and oxygen.

All animals were warmed at the same temperature for 15 min before any recording to ensure the stabilization of physiologic rates. After this initial delay period, the ambient temperature in the holder was set to 33°C, 35°C, and 37°C for the first, second, and third groups, respectively, and individual heart rates and body temperatures were measured during a 60-min observation period.

From these first monitoring observations and as a preamble to the kinetic measurements, we defined and retained for the forthcoming imaging procedures 2 sets of heating conditions. The optimal heating condition was defined as a holder maintained at a constant temperature of 35°C, and the suboptimal heating condition was defined as a temperature of 33°C. These conditions were applied successively to our animal groups during the imaging procedures.

A higher temperature for the holder (37°C) led to overheating, as attested to by an increase in measured mouse body temperatures, and unstable anesthesia, with uncontrolled limb movements and undesired awakening. Therefore, this heating condition was not considered further for the imaging procedures.

Imaging of Animals Under Different Physiologic Conditions

A pool of 12 adult female BALB/c nude mice (body mass, 18.7 ± 0.3 g) was selected for tumor implantation.

Wild-type anaplastic thyroid cancer cells and human sodium iodine symporter–transfected cancer cell xenografts were implanted subcutaneously in all animals. Implanted cells were obtained and prepared as described by Gholami et al. (13). Transfected xenografts were implanted between the left shoulder and the flank of the animal, and wild-type xenografts were implanted between the right shoulder and the flank of the same animal (14).

Imaging Protocol

After 4 d of tumor growth (both implanted tumors were 5–7 mm), animals were prepared for the imaging procedures. Animals were provided with food and water ad libitum before being imaged with both tracers. Mice were maintained under anesthesia inside the same animal holder attached to an R4 microPET scanner (Concorde Microsystems). Mice were kept warm by the circulating air in the holder and stabilized before tracer administration. The tracers (¹⁸F-TFB and ¹⁸F-FDG) were injected separately directly in a tail vein inside the holder just before the beginning of imaging acquisition. ¹⁸F-TFB was produced as described by Jauregui-Osoro et al. (15).

Each animal underwent 4 tracer injections followed by dynamic PET acquisitions, 2 with ¹⁸F-FDG on day 1 and 2 with ¹⁸F-TFB on day 2. A 4-h free interval was set between the 2 successive acquisitions on the same day. During this interval and during the night, mice had access to food and water ad libitum.

Circadian rhythm is known to influence homeostatic and metabolic regulation systems and therefore might have influenced tracer uptake in our experiments (5, 9), depending on the time of injection and on the time of the animal’s last food intake. Tumor growth between the first and the last acquisitions (24 + 4 h) also might have influenced tracer uptake and confounded intraindividual comparisons. To limit the influence of these circadian effects (including feeding habits) as well as tumor growth, we divided the animals into 4 groups of 3 animals each and, for each group, cycled the imaging sequence between optimal and suboptimal heating conditions for both tracers as well as the time of injection to avoid any systematic influence or risk of systematic error.

Mice were injected twice with 3.1 ± 0.2 MBq of ¹⁸F-TFB and twice with 3.2 ± 0.4 MBq of ¹⁸F-FDG. Dynamic PET images were acquired during the first 45 min after injection. The optimal or suboptimal heating condition for the holder was kept stable during the entire acquisition process. Mouse heart rates were monitored and recorded throughout the 45-min process. In total, 48 datasets were collected for the 12 mice with bilaterally implanted tumors.

Image Reconstruction and Data Collection

Ninety successive frames were collected for each dataset during the first 45 min (30 s/frame), with a single bed position covering the entire mouse body (full axial field of view, 7.2 cm). Data were reconstructed by use of the vendor’s OSEM2d algorithm (Concorde Microsystems) with 16 subsets and 4 iterations. ¹⁸F decay correction was applied, but no attenuation correction method was selected. The final image reconstruction matrix was 128 × 128 × 95 pixels for a field of view of 11 × 11 × 8 cm.

After data reconstruction (3-dimensional datasets plus time) and to assess time–activity data to ensure reproducible region-of-interest (ROI) positioning, we generated a time series of 2-dimensional projection images; 90 successive single 2-dimensional ventral-to-dorsal projection images were calculated from each 3-dimensional frame. Ventral-to-dorsal projection was preferred over other axis projections for ROI positioning to ensure positioning reproducibility as well as limited overlapping of organs of interest.

Several ROIs were manually drawn on the ventral-to-dorsal projection frames and used to derive time–activity data. ROIs were placed over the thyroid, stomach, implanted tumors, and blood-pool compartment (heart ROI) for ¹⁸F-TFB image analysis (Fig. 1, top). ROIs were placed over the liver, brain, heart, and implanted tumors for ¹⁸F-FDG image analysis (Fig. 1, bottom). The sizes and shapes of the ROIs were the same for both heating conditions.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

¹⁸F-TFB (top) and ¹⁸F-FDG (bottom) dynamic image acquisitions. After each tracer injection, full sets of 90 images were acquired for 45 min. Optimal and suboptimal heating acquisitions for same animal are shown at left. Nine reformatted frames (5 min/frame) are presented as ventral-to-dorsal projections. Single 45-min projection is shown at right. Circular ROIs used for time–activity curve extraction are plotted. For ¹⁸F-TFB, ROIs were as follows: 1, thyroid gland; 2, heart (blood pool); 3, right implanted tumor; 4, left tumor; 5, stomach area; 6, bladder. For ¹⁸F-FDG, ROIs were as follows: 1, brain area; 2, right implanted tumor; 3, liver area; 4, heart; 5, left tumor; 6, kidney; 7, bladder.

Curve Fitting and Kinetic Measurements

Raw time–activity curves were obtained from the ROIs for each animal for both optimal and suboptimal heating acquisitions and for both injected tracers. Image-derived time–activity data for organs and tumors were analyzed with a biexponential model:where A and B are fitting constants (reported as tracer concentrations) and λ_washout and λ_uptake are washout and uptake time constants (reported per minute), respectively. Time–activity curve raw data points collected from the blood-pool compartment were fitted with a biexponential decreasing function:where A and B are fitting constants and λ_{fast washout} and λ_washout are fast initial apparent washout and slow late washout time constants, respectively. Dedicated modeling software (proFit; Quantumsoft) was used to determine the uptake and washout time constants.

Statistical Analysis

A paired Student t test was used for direct comparison of uptake and washout time constants for each compartment of the model—comparing optimal and suboptimal heating regimens. P values of less than 0.01 were considered significant.