Inhibition of Physiologic Myocardial FDG Uptake in Normal Rodents: Comparison of Four Pre-Scan Preparation Protocols ()
1. Introduction
Under baseline and fasting conditions, cardiomyocytes primarily utilize fatty acids for metabolism; however, they readily consume glucose, pyruvate, ketone or lactate, depending upon their availability [1,2]. Cardiomyocytes physiologically accumulate FDG, a D-glucose analogue, which is the most commonly used PET radiotracer. The transport of D-glucose and hence FDG in to cardiomyocytes is mediated through several different transporters of the GLUT family, predominantly via GLUT-4, an insulinsensitive transporter [3].
Myocardial sequestration of FDG decreases the sensitivity and specificity of FDG PET imaging, specifically in patients with thoracic malignancies with or without myocardial involvement. Therefore, it is a major barrier to the growing potential of FDG PET imaging for identification of pathologic processes involving myocardium, pericardium and coronary arteries [4,5].
To overcome this physiological barrier, either new radiotracers with different myocardial pharmacokinetics are required or physiologic FDG uptake by myocardium needs to be suppressed. Fasting the patients for 6 - 12 hours remains an accepted preparation protocol to reduce serum glucose levels in order to minimize myocardial FDG uptake [6,7]; however, short to moderate periods of fasting (6 - 12 hours) have been found to inconsistently and variably suppress physiologic myocardial FDG utilization [8,9]. Long term fasting (greater than 18 hours) appears to be considerably more reliable [10], likely because of elevated serum free fatty acids in addition to low glucose, but prospective evidence is lacking and compliance remains an issue. Carbohydrate restriction alone followed by short term fasting results in further reduction in serum glucose and serum insulin [11]. This reduces serum glucose levels, but in the presence of low serum fatty acids, may not be very effective in suppressing physiologic myocardial glucose or FDG uptake.
Inhibition of glucose oxidation by elevated serum fatty acid concentrations, and vice versa are well established phenomena “the glucose-fatty acid cycle [12]. Cook et al [13] suggested consumption of a fatty meal or administration of a pharmacological fatty acid agent prior to FDG dosing to maximize fatty acid metabolism, and thereby minimize glucose metabolism and restrict myocardial FDG uptake, specifically for mediastinal imaging. This has recently been studied by William et al. [14] with effective suppression of physiologic myocardial FDG uptake in oncology patients, to optimize pathologic sequestration of FDG. Direct comparison of protocols using different duration of fasting with or without high fat diet consumption has not been studied.
Rats have been extensively studied to provide insight in to various patho-physiological processes involving the human heart. We have prospectively examined the effects of different preparation protocols in an animal model, with each rodent (rat) serving as its own internal control.
2. Material and Methods
2.1. Subjects
The subjects were 8 - 20 weeks old male (n = 5) and female (n = 4) Sprague Dawley rats, which weighed between 257 - 396 g (389.57 ± 122.91) at the inception of the experiment and 188 - 596 g (440 ± 148.15) at the end of the experiment. They were purchased from Charles River Laboratories International, Inc. (Wilmington, MA 01887). Prior to the experiments, all the subjects were quarantined and acclimated for 7 days in a vivarium (set at 12 hrs light and dark cycle) in a dedicated small animal imaging center. At all times, the subjects received continuous food and water in their home-cages, except during the fasting phase and the high-fat-diet phase. All the experimental methods were approved by our Institutional Animal Care and Use Committee.
2.2. FDG dose Preparation
FDG was synthesized at an offsite cyclotron facility, managed and operated by PETNET.Inc./Siemens located in Columbia, SC. Prior to injection, the FDG doses were calibrated and decay-corrected based on the institutional PET center’s well-counter. After dose calibration and decay-correction, an injected dose averaging to 3.6 mCi (millicurie) or 133.2 MBq (megabacquerel) ranged from 2.9 to 4.1 mCi (107.3 - 151.7 MBq) across different preparation protocols. Each injected dose was diluted and suspended in a 0.5-0.7 ml of Dulbecco’s Phosphate Buffered Saline (pH 7.4; Invitrogen, Carlsbad, CA 92008). All the decay corrections, during and after the scans, were performed by the scanner’s software (IAW 2.2).
2.3. Experimental Method
A within-subject design was used in which each subject served as its own internal control. Figure 1 summarizes the experimental design and depicts the four preparation protocols completed for each rodent. All nine rodents received four preparation protocols, each separated by 2 - 3 weeks: baseline/non-fasting (NF), 18-hrs/prolonged fasting (PF), 12-hrs/Short fasting combined with high fat diet (SF-HFD) and 18-hrs/Prolonged fasting combined with high fat diet (PF-HFD). For 12-hrs/short fasting combined with HFD (SF-HFD) protocol, each subject underwent 12-hrs/short fasting followed by 3-hrs period of consuming the high-fat sunflower seeds, followed by another 2 - 3 hours of fasting before the FDG tracer injection. A similar protocol was followed for 18-hrs/prolonged fasting combined with HFD (PF-HFD) except for the duration of fasting. The high-fat diet consisted of sunflower seeds containing protein (16%) fat (40%), and fiber [35%] (www.bio-serv; product #S5137) with each rat consuming 4.53 g ± 1.40 g of sunflower seeds. Sun flower seeds were chosen due to their high fat and negligible absorbable carbohydrate content out of the other commercial diets available by vendors supplying animal diets for our small animal imaging facility. A few minutes prior to FDG administration, each subjects’ blood glucose level (mg/dl) was measured with an Abbott’s Precision Xtra Glucometer (calibrated according to the vender’s glucose standard strip; Almeda, CA 94502).
2.4. Micro-PET Specifications
All subjects were imaged in a Siemens’ Quicksilver Inveon PET/CT capable of acquiring PET only or CT only 3D volumetric data and docking into an easily co-registered PET/CT 3D-data acquisition mode. For this study, only 3D-PET acquisitions were obtained and no cardiac gating was performed. The Inveon scanner has a 12.7- and 16.1- cm axial and transaxial center field of view (FOV), respectively [15]. The PET camera contains 16 blocks of cerium doped lutetium oxyorthosilicate crystals [16] with a ring geometry construction. Each LSO block is composed of 20 × 20 crystal matrix and each has a dimension of 1.5 × 1.5 × 10 mm3. At 3D-mode, the LSO detectors have an intrinsic spatial resolution between 1.3- and 1.5- mm FWHM (Full Width at Half Maximum) and 10% - 12% detection efficiency. The PET scanner utilized a pair of build-in dual Co-57 point sources for immediate serial attenuation correction. Acquisition and operation of the PET scanner was interfaced by an external host and embedded Dual 3.20 GHz Intel Xeon processor computers under the Microsoft Window XP pro SP2 operating system, which have 4 GB RAM for viewing and reconstruction.
2.5. Anesthesia Details
Forty-five minutes after the FDG administration (Figure 1), each subject was anesthetized with inhalation anesthetic agent Isoflurane (Phoenix Pharmaceuticals, Inc., St. Joseph, MO 64507). Specifically, anesthesia was induced in a temperature controlled, Lucite induction chamber with 1% Oxygen and 3 - 4 MACs of Isoflurane. Once an acute anesthetic plane was established, the subjects were transferred to the scanner bed and maintained with 1% Oxygen and 1.5 - 2 MACs of Isoflurane via a nose cone in a prone position for the duration of the 60-min scan. After each scan, the subjects were allowed to recover from the anesthesia and monitored until they were fully mobile and conscious.
2.6. Data Acquisition and Reconstruction
Forty-five minutes after FDG administration, each rodent in a prone position was scanned for 60-minutes, while under Isoflurane anesthesia. Decay and residual corrections were implemented based on the initial dose assayed. Immediately after the emission scan, a 10 min Blank-corrected transmission scan was performed for attenuation correction. The raw emission and transmission data were in list-mode form and this data was then transformed into a sinogram for reconstruction purposes. For volumetric reconstruction, the emission sinogram (at Nyquist frequency cutoff) in conjunction with attenuation and scatter correction sinograms, along with component-based normalization coefficients [17] were entered into an iterative 3D-Ordered-Subsets Expectation Maximization (3D-OSEM) reconstruction algorithm in which the missing pixels due to detector gaps were estimated using Fourier rebinning [18,19]. The 3D-OSEM reconstruction algorithm parameters were set at 4 iterations with 16 subsets [20]. The voxel data were reconstructed into a 128 × 128 × 128 matrix with reconstructed spatial resolution of 0.776 mm3. We utilized 3D-OSEM rather than Filtered Back Projection (FBP) as this 3D-OSEM recon-
Figure 1. Describes the four preparation protocols as detailed in sub-heading “Experimental method” of section “Material and Methods”. NF: non-fasting; PF: prolonged fasting; SF-HFD: short (12 hour) fasting with high fat diet; PF-HFD: prolonged (18 hour) fasting with high fat diet; SF seeds: sun flower seeds.
struction algorithm has been shown to provide superior spatial resolution with less pixel variation and lower coefficients of variation [19]. The reconstructed images were in IMG format. The IMG images were converted into DICOM format using Siemens’ Inveon Research Workstation software (IRW, version 2.2) for quantification and 3D-visualization.
2.7. Data Processing and Analyses
The reconstructed volumetric, DICOM data were visualized and quantified using Siemens’ IRW quantification software. Region of interests (ROIs) were manually drawn on left ventricular myocardium and ascending aorta lumen (close to origin of aorta) to obtain respective heart and blood pool SUVm activity. The quantified activities for these ROIs were in the units of distribution volume (DV = Bq/ml) per pixel. Since the data were acquired with the 3D-static mode, the data of interest were the maximum DV pixel values (DVmax) in the myocardium and the blood pool ROIs, instead of average pixel values. The DVmax pixel values were converted into SUVm [21,22] using Equation 1, with the injected dose (in Bq) and bodyweight (in grams) :
(1)
Maximum standard uptake value of heart (H-SUVm) and blood pool (BP-SUVm), and glucose values obtained across four different preparation protocols were subjected to one-way repeated analysis of variance (ANOVA). Pair-wise differences were performed using the least square planned contrasts using the total error term of the repeated ANOVA. The pair-wise planned contrasts were conducted via F-ratio statistics. The significant level was set at alpha of 0.05.
3. Results
Figure 2 demonstrates that H-SUVm is significantly lower in PF-HFD and SF-HFD preparation protocols compared to NF and PF protocols. H-SUVm is slightly lower in PF-HFD compared to SF-HFD protocol; however, the difference is non-significant. Similarly, H-SUVm in PF protocol is lower than NF protocol, but the difference is non-significant, secondary to significant overlap in H-SUVm among these protocols. In two subjects (A4, B3), the H-SUVm increased in PF protocol compared to NF protocol, which may be explained by the rodents not consuming enough carbohydrates during the non-fasting state, but under both PF-HFD and SF-HFD conditions, anticipated myocardial FDG suppression is observed. In subjects A5 and B4, significant H-SUVm suppression is noticed in PF protocol, essentially similar to SF-HFD and PF-HFD protocols, compared to NF protocol. This does endorse the significance of PF protocol in suppressing cardiac FDG uptake; however, it is less pronounced and inconsistent in other subjects.
Table 1 summarizes the H-SUVm of all subjects following each of the four different preparatory protocols. Overall, H-SUVm decreases significantly with fasting and