Serial noninvasive targeted imaging of peripheral angiogenesis: validation and application of a semiautomated quantitative approach

Abstract

Previous studies by our group have demonstrated the feasibility of noninvasive imaging of alpha(v) integrin to assess temporal and spatial changes in peripheral and myocardial angiogenesis. In this study, we validate the reproducibility, accuracy, and applicability of a new semiautomated noninvasive approach for serial quantitative evaluation of targeted micro-SPECT/CT images of peripheral angiogenesis in wild-type and endothelial nitric oxide sythase (eNOS)-deficient (eNOS-/-) mice subjected to hindlimb ischemia.

Methods: Mice (n = 15) underwent surgical ligation of the right femoral artery to induce unilateral hindlimb ischemia. One week after ligation, a (99m)Tc-labeled cyclic-Arg-Gly-Asp peptide targeted at alpha(v) integrin (NC100692, n = 10) or a (99m)Tc-labeled negative control (AH-111744, n = 5) was injected, and 60 min later in vivo micro-SPECT/CT images were acquired. Mice were euthanized, tissue from proximal and distal hindlimb was excised for gamma-well counting (GWC) of radiotracer activity, and ischemic-to-nonischemic (I/NI) ratio was calculated. Micro-SPECT/CT images were analyzed using a new semiautomated approach that applies complex volumes of interest (VOIs) derived from segmentation of the micro-CT images onto micro-SPECT images to calculate I/NI activity ratios for the proximal and distal hindlimb. Studies were reprocessed for determination of intra- and interobserver variability. To compare 3-dimensional (3D) VOI analysis with traditional manual 2-dimensional region-of-interest (ROI) analysis of maximum-intensity-projection images, micro-SPECT images were summed onto a single anterior-posterior projection. Rectangular ROIs were manually drawn and I/NI ratio calculated. Our new 3D analysis approach was applied to additional groups of mice (eNOS-/-, n = 5; wild-type, n = 3) imaged before and 1 and 4 wk after femoral artery resection.

Results: Our new semiautomated approach for the evaluation of images of alpha(v) integrin targeted with micro-SPECT/CT demonstrated both a high intra- and interobserver variability (R(2) = 0.997) and an accuracy (R(2) = 0.780) for estimation of relative radiotracer activity relative to GWC. Analysis of serial micro-SPECT/CT images demonstrated a significant increase in relative NC100692 retention in the ischemic hindlimb of both wild-type and eNOS-/- mice at 1 wk after surgery. There was a significant (approximately 25%) decrease in radiotracer uptake in eNOS-/- mice relative to wild-type animals, which was not observed at baseline or 4 wk after ligation.

Conclusion: A new semiautomated analysis of images of alpha(v) integrin targeted with micro-SPECT/CT provides a noninvasive approach for serial quantitative evaluation of peripheral angiogenesis. The reproducibility and accuracy of this approach allows for quantitative analysis of serial targeted molecular images of lower extremities, has applicability to other targeted SPECT or PET radiotracers, and may have implications for clinical imaging in patients with peripheral arterial disease.

Figure 1. Method for analysis of microSPECT-CT images

Planes were interactively positioned over the lower body of mice to segment the microCT to generate multiple volumes-of-interests (VOIs) (A). The contours of these VOIs are illustrated superimposed on a representative microCT image (B). K-means clustering-based segmentation was performed to eliminate bones from the VOIs (C). This complex irregular objectmap was applied to the registered microSPECT images to determine mean counts in each VOI (D).

Figure 2. Validation of microSPECT-CT image analysis approach

Correlation between ischemic-to-nonischemic (I/NI) ratios of counts calculated from image analysis and gamma well counting for both proximal and distal regions (A). The correlation was poor (R²=0.05) when all proximal and distal regions were included (A). The correlation coefficient between imaging analysis and gamma well counting improved significantly (R²=0.63) after proximal regions contaminated with scatter from radioactivity in the bladder have been removed from the analysis (B). Shown below are representative ^99mTc-NC100692 microSPECT-CT images from three mice at 7 days after surgical ligation of right femoral artery with superimposed VOIs. Figure 2 C illustrates a mouse with radioactivity within bladder which falls within the right proximal VOI, resulting in overestimation of counts in ischemic (right) relative to non-ischemic (left) proximal leg. Figure 2 D illustrates a second mouse with bladder located centrally in the body and no significant contamination of proximal hindlimb VOIs. Figure 2E illustrates images from third mouse following removal of the radioactive urine from the bladder by needle aspiration immediately before SPECT acquisition.

Figure 3. Validation of analysis of distal hindlimb

There was a good correlation (R=0.78) between ischemic-to-nonischemic counts ratio calculated from 3D volumetric analysis of microSPECT-CT images (A), and 2D analysis of maximum projection images (C) relative to gamma well counting for the distal hindlimb. Linear correlation (solid line) and 95% confidence intervals (dashed line) are shown (B & D). The quality of the fit was confirmed by residuals plot for microSPECT-CT images (B), 2D maximum projection images (D) analysis approaches.

Figure 4. Reproducibility of image quantification

The intra-observer (A) and inter-observer (B) reproducibility of microSPECT-CT image analysis were excellent.

Figure 5. Analysis of wild-type and eNOS knockout mice

Representative microSPECT-CT images of wild-type (eNOS+/+) and eNOS-knockout (eNOS-/-) mice injected with ^99mTc-NC100692 at baseline, 7 days, and 4 weeks post right femoral artery ligation (A). Yellow arrows indicate ischemic regions with increased ^99mTc-NC100692 retention. Less retention is seen in eNOS-/- mice. Serial microSPECT-CT images were analyzed and I/N ^99mTc-NC100692 activity ratios calculated (B). There was a significant (P<0.05) increase of ^99mTc-NC100692 retention in ischemic leg at 7 days post surgery in both groups. However, there was significantly less retention in the eNOS-/- mice at 7 days compared with wild-type mice. * P < 0.05 vs. wild-type # P < 0.05 vs. baseline

Figure 5. Analysis of wild-type and eNOS knockout mice

Representative microSPECT-CT images of wild-type (eNOS+/+) and eNOS-knockout (eNOS-/-) mice injected with ^99mTc-NC100692 at baseline, 7 days, and 4 weeks post right femoral artery ligation (A). Yellow arrows indicate ischemic regions with increased ^99mTc-NC100692 retention. Less retention is seen in eNOS-/- mice. Serial microSPECT-CT images were analyzed and I/N ^99mTc-NC100692 activity ratios calculated (B). There was a significant (P<0.05) increase of ^99mTc-NC100692 retention in ischemic leg at 7 days post surgery in both groups. However, there was significantly less retention in the eNOS-/- mice at 7 days compared with wild-type mice. * P < 0.05 vs. wild-type # P < 0.05 vs. baseline