The data that support the findings of this study are available from the corresponding author upon reasonable request.

Large Animal Model of Hindlimb Ischemia

Male Yorkshire pigs (N = 8; 10.1 ± 0.4 kg) approximately 4 weeks of age underwent surgical cutdown and permanent ligation of the right femoral artery to induce unilateral hindlimb ischemia. A full description of surgical preparation and techniques for this animal model have been previously described in prior publications [10, 12]. The animal protocol was reviewed and approved by the Institutional Animal Care and Use Committee of Nationwide Children’s Hospital and was in compliance with the Association for Assessment and Accreditation of Laboratory Animal Care International policies. All procedures followed the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The use of radioisotopes and all imaging procedures were approved by the Radiation Safety Committee of Nationwide Children’s Hospital. The ARRIVE guidelines and regulations were followed for reporting of animal use and study results [13].

18F-FDG PET/CT Imaging Protocol

Animals were fasted for a minimum of 8 h prior to PET/CT imaging. The jugular vein and carotid artery were percutaneously accessed, and 4-F polyethylene catheters (Cook Medical, Bloomington, IN) were placed to allow for venous administration of 18F-FDG for PET imaging and continuous arterial blood sampling during PET imaging. To maintain consistent anesthesia levels during PET/CT imaging, propofol was infused at 10–30 mg/kg/hour through an ear vein catheter. The depth of anesthesia was monitored every 15 min by evaluating blink reflex and jaw tone. Physiological parameters including blood pressure, heart rate, oxygen saturation, and body temperature were monitored continuously for the duration of PET/CT imaging sessions (IntelliVue MP30; Philips Healthcare, Andover, MA), with body temperature maintained above 37 degrees Celsius to control for the potential influence of temperature on measures of skeletal muscle perfusion and metabolism.

Dynamic 18F-FDG PET/CT imaging was performed on the day of femoral artery occlusion and 2 weeks after occlusion using a hybrid PET/CT system (Discovery 690, GE Healthcare, Chicago, IL). Seven of eight animals underwent serial dynamic PET imaging in list-mode for 60 min, and one of the eight animals underwent serial dynamic PET imaging for 2.5 min. CT images were acquired with voxel dimensions of 1.37 × 1.37 mm and slice thickness of 3.27 mm, at 120 kVp and 300 mA, for attenuation correction and calf muscle segmentation. The PET/CT imaging study design is outlined in Fig. 1.

Fig. 1The alternative text for this image may have been generated using AI.

Experimental workflow for performing dual monitoring of lower extremity skeletal muscle perfusion and metabolism using 18F-FDG PET/CT imaging. Representative schematic displaying the study timeline, representative time activity curves for arterial blood and calf muscle tissue for the first 2.5 min of PET data acquisition, calf muscle segmentation, and resulting skeletal muscle perfusion map after 1-tissue compartment modeling. The perfusion deficit induced by unilateral femoral artery occlusion and localized to the calf is shown on fused PET/CT imaging

The 60-min dynamic PET scan was started in list mode immediately before intravenous injection of 18F-FDG (196.8 ± 7.5 MBq), with 18F-FDG dose and saline flush administered over a 20-s duration. The animal’s lower abdomen and hindlimbs were positioned in the PET camera’s field-of-view (FoV) (~ 15.7 cm axial FoV) to ensure coverage of both the abdominal aorta and calf muscles. 1 ml of arterial blood was collected at baseline (prior to imaging) and then continuously sampled every 5 s for the first 2.5 min of PET imaging at a constant rate (12 ml/min) using a Harvard peristaltic pump (Harvard Apparatus, South Natick, MA). Approximately 0.5 ml of each sample was used to measure whole blood radioactivity while the other 0.5 ml of blood was centrifuged at 2000 g for 15 min to acquire plasma radioactivity. Whole blood and plasma samples were weighed, counted, and corrected for radioactive decay using a gamma well counter (WIZARD2, PerkinElmer, Waltham, MA). Whole blood and plasma time activity curves (TACs) were then obtained, as well as plasma-to-whole blood ratios for each arterial sample. In addition to arterial blood sampling for gamma counting, plasma samples were analyzed for glucose concentration at baseline (prior to initiation of PET imaging) and every 10 min during image acquisition (i.e., 10-, 20-, 30-, 40-, 50-, and 60-min time point) to evaluate stability of plasma glucose during the 60-min PET acquisition. All measurements of plasma glucose concentration were assessed using a commercially available blood glucose monitor (AlphaTRAK 2; Zoetis Inc., NJ, USA).

PET/CT Image Reconstruction and Processing

All dynamic PET data was reconstructed using 2 iterations and 32 subsets of the ordered subset expectation and maximization (OSEM) algorithm with trans-axial full width at half maximum (FWHM) Gaussian filter equal to 4.5 mm. PET image volumes were preprocessed using Hounsfield-unit based attenuation correction. The PET image reconstruction matrix was 192 × 192 pixels, with voxel dimensions of 3.65 × 3.65 mm and a slice thickness of 3.27 mm. To address potential limitations of partial volume and spill-over effects for the image-derived arterial input function (AIF), arterial blood sampled during dynamic PET imaging was used to generate correction factors for the image-derived plasma AIF using recently published methods in the same porcine model of hindlimb ischemia [10]. All aspects of PET/CT image processing were performed using commercially available software (PMOD Technologies LLC, Fallanden, Switzerland).

For evaluation of skeletal muscle perfusion, the first 2.5 min of the PET list data was reconstructed in 3-s frames to obtain the AIF from the abdominal aorta, and 5-s frames for analysis of 18F-FDG kinetics in the bilateral calf muscles (Fig. 1). Calf perfusion values were calculated using a 1-tissue compartment kinetic model that incorporated the corrected image-derived plasma AIF, skeletal muscle TACs, and blood volume as a fitting parameter. First-pass extraction of 18F-FDG was assumed to be 100% in peripheral skeletal muscle with zero venous outflow. While it is understood that first-pass extraction of 18F-FDG is not 100% due to 18F-FDG not being freely diffusible, this assumption for a 1-tissue compartment model was previously used when validating dynamic 18F-NaF PET imaging as an approach for quantifying skeletal muscle perfusion [10] and by Mullani et al. [14] when validating dynamic 18F-FDG PET as a method for quantifying first-pass tumor perfusion. To adjust for variability in hemodynamic loading conditions, perfusion values were normalized by rate pressure product (RPP) by multiplying the perfusion value by the reference RPP for the study cohort and then dividing by the individual RPP at the time of PET perfusion imaging. Details of this PET image processing workflow were recently described and validated in the same porcine model of hindlimb ischemia using 18F-NaF [10].

For evaluation of skeletal muscle metabolism, the entire 60-min PET list data was reconstructed using the following sequence: 6 frames of 20 s each, followed by 3 frames of 60 s each, 3 frames of 300 s each, and 4 frames of 600 s each, as previously described by Pande et al. [11]. The overall rate of 18F-FDG uptake (Ki) of the ischemic and control calf was quantified from PET imaging with a 3-tissue compartment model developed by Bertoldo et al. [15] using an image-derived AIF and calf muscle TACs. The metabolic rate of glucose (MRGlu) in the calves was then calculated by multiplying the image-derived Ki by the plasma glucose concentration and correcting for the relative metabolism of 18F-FDG versus metabolism of true glucose using the lumped constant, which was assumed to be 1.2 for skeletal muscle based on the findings of Kelley et al. [16]. The MRGlu for was ultimately expressed as \(\mu\) mol/min/kg of tissue.

Quantification of Peripheral Microvessel Remodeling

Following completion of PET/CT imaging at 2 weeks post-occlusion, animals were euthanized via intravenous administration of euthasol (1 ml/4.5 kg), and muscle samples (~ 20 g) were harvested from the gastrocnemius muscle of the bilateral calves in 7 of 8 pigs to assess the relationship between calf muscle angiogenesis and recovery of calf muscle perfusion 2 weeks after arterial occlusion. Muscle tissue sampling sites were standardized for the ischemic and control hindlimbs. Samples were fixed with 4% paraformaldehyde, paraffin embedded and sectioned at 4-μm thickness. Sections were stained with primary Anti-CD31 antibody (ab28364, Abcam, Cambridge, U.K.) and secondary Alexa Fluor 580 IgG goat anti-rabbit antibody (A-11011, Thermo Fisher Scientific). All sections were then imaged on a fluorescent microscope (BZ- × 800, Keyence, Itasca, IL) and image quantification was performed as previously described using ImageJ (National Institutes of Health) [10]. Capillary density of the gastrocnemius calf muscle was expressed as capillary-to-muscle fiber ratios.

Statistical Analysis

Serial changes in PET-derived measures of muscle perfusion and muscle metabolism were evaluated using a repeated measures analysis of variance (ANOVA) with a linear mixed effect model, with animal included as random effect. Model residuals were assessed for normality using the Shapiro–Wilk test and Q-Q plots. For outcomes that satisfied normality assumptions, pairwise comparisons were performed using estimated marginal means with Tukey adjustment for multiple comparisons. Paired comparisons not meeting normality assumptions (i.e., MRGlu) were performed using the Wilcoxon signed rank test with Holm-Sidak adjustment for multiple comparisons. Differences in plasma glucose between imaging sessions, as well as differences in capillary density between control and ischemic hindlimb calf muscles were evaluated using paired t-tests. A p-value of < 0.05 was considered statistically significant for all analyses. Statistical analyses were performed using R, version 4.3.0 in R Studio, and Prism for macOS, version 9.3.0 (GraphPad Software, LLC).