Molecular imaging of metastatic potential

Abstract

If molecular imaging is to prove clinically useful it will have to surpass current, primarily anatomic techniques in terms of sensitivity and the ability to detect minimal changes in tissue. One of the most important tests for molecular imaging is to determine whether it can image the metastatic potential of tumors. Like all predictive endeavors, the imaging of such "potential" is a daunting task, but one that only molecular imaging--rather than standard, anatomic techniques--is likely to solve. Although difficult, imaging of metastatic potential is also arguably the most important task for molecular imaging of cancer because it is generally the dissemination of malignant tissue, not its prolonged residence in an inopportune site, which kills the patient. Below are examples of uses of molecular imaging of metastases as well as of metastatic potential, the former being a far more developed area of clinical inquiry.

FIGURE 1

Illustration of overall metastatic process. Tumor cells are believed to proceed through these sequential steps to form clinically detectable metastases. (Adapted with permission of (193).)

FIGURE 2

Metabolic Boyden chamber assay demonstrating differences in invasiveness among 6 different cancer cell lines that included 3 human breast cancer lines (MDA-MB-435 [A], MDA-MB-231 [B], and MCF-7 [C]), 1 rat prostate cancer cell line (MatLyLu [D]), and 2 human prostate cancer cell lines (PC-3 [E] and DU-145 [F]) at approximately 47 h. T1-weighted MR images show the bright Matrigel layer, which is significantly degraded by all cell lines except MCF-7. (Adapted with permission of (70).)

FIGURE 3

Coregistered MRI maps of vascular volume (red) and vascular permeability (green) for 6 different tumor xenograft models from cell lines shown in Figure 1. Paucity of yellow regions (representing areas concurrently exhibiting elevated vascular volume and permeability) indicates that regions with high vascular volume were significantly less permeable and regions of high permeability consistently exhibited lower vascular volumes. Data from this study collectively indicated that cancer cells expressing both highly invasive and elevated angiogenesis represented the most lethal phenotype. (Adapted with permission of (70).)

FIGURE 4

(A) Representative functional MRI maps of pooling and draining voxels for MDA-MB-231 and MCF-7–bearing animals. (B) Box-and-whisker plot comparing volume of extravasated fluid pooled or drained between MDA-MB-231 and MCF-7 xenografts (*P < 0.1 with 1-sided Mann–Whitney U test). (C) Phase contrast image obtained at ×40 from 5-μm-thick hematoxylin-stained MDA-MB-231 tumor section overlaid with same tumor section stained with LYVE-1 (lymphatic vessel marker), with transparency of latter adjusted to enable visualization of lymphatics packed with tumor cells. (D) Photomicrograph obtained at ×40 from 5-μm-thick hematoxylin- and eosin-stained axillary lymph node from MDA-MB-231–bearing animal, scored positive for presence of cancer cells (arrows). (E) Photomicrograph obtained at ×40 from 5-μm-thick hematoxylin- and eosin-stained lung from MDA-MB-231–bearing animal scored positive for presence of cancer cells (arrows). (F) Similar image obtained at ×40 from 5-μm-thick hematoxylin-stained MCF-7 tumor section overlaid with image of same tumor section stained with LYVE-1, with transparency of latter adjusted to enable visualization of lymphatics that have few, if any, tumor cells. (G) Photomicrograph obtained at ×40 from 5-μm-thick hematoxylin- and eosin-stained axillary lymph node from MCF-7–bearing animal, scored negative for presence of cancer cells. (H) Photomicrograph obtained at ×40 from 5-μm-thick hematoxylin- and eosin-stained lung from MCF-7–bearing animal scored negative for presence of cancer cells. (Adapted with permission of (76).)

FIGURE 5

Transaxial ¹⁸F-FMISO PET, transaxial ¹⁸F-FDG PET, and sagittal ¹⁸F-FDG PET images of patient with normoxic tumor and patient with hypoxic tumor. ¹⁸F-FMISO is taken up in hypoxic tumors (arrows) but not in normoxic tumors. (Adapted with permission from (95).)

FIGURE 6

¹⁸F-FAZA–based imaging of hypoxia in planning of radiation therapy for head and neck tumor: FAZA PET (A and C), CT (B), and PET/CT (D). In B, large primary tumor (yellow arrow) and retro-pharyngeal lymph node metastasis (red arrow) demonstrate heterogeneous ¹⁸F-FAZA uptake. In C, blue region of interest depicts highest radiopharmaceutical uptake whereas yellow region delineates lowest. Dotted red line in D outlines tumor anatomically, and solid red line shows tumor, with a tumor-to-muscle ratio of more than 1.5. (Adapted with permission of (114).)

FIGURE 7

¹⁸F-galacto-RGD PET images of woman with lymph node metastasis from melanoma (white arrow). Intense radiopharmaceutical uptake is seen in lesion (black solid arrow). Minimal uptake is seen in urinary bladder (red arrow) and gallbladder (dotted arrow). (Adapted with permission of (130).)

FIGURE 8

Tenascin aptamer imaging. U251 glioblastoma tumor is faintly visible at 10 min, prominent at 3 h, and brightest structure at 18 h. MDA-MB-235 tumors could also be visual-ized. (Adapted with permission of (143).)

FIGURE 9

Comparison of detection of tdTomato fluorescence and GFP fluorescence in mouse cadaver phantom with Xenogen IVIS 200 system. (A) Upper left image shows fluorescence from tubes packed with 100 × 10⁶ cells expressing MDA-MB-231-tdTomato (red) or MDA-MB-231-GFP (green). Below these are false-color-overlay images (0.01-s exposure time) using either GFP filter set or DsRed set. White tubes in these images indicate that tdTomato did not fluoresce when GFP was being imaged and GFP did not fluoresce when tdTomato was being imaged. Mouse false-color-overlay images (regions of interest circled; 1-s exposure time) at center and right show that only tdTomato fluorescence could be detected from implanted tubes. (B) Panel showing similar results to those in A, only number of tdTomato cells used was 45 × 10⁶ while number of GFP-expressing cells remained same. Implanted tubes are shown on left; red and green fluorescence was easily detected (0.01-s exposure times). (C) Unprocessed fluorescent images with 2.5-s exposure times show that autofluorescence from fur of SCID mouse depends on emission wavelength filter used. Tube implants were those shown in A (left image), B (center image), and tube containing 9.25 × 10⁶ cells expressing tdTomato (right image). Regions of interest are circled in first 2 cases. Left and center images show intense fur autofluorescence that masks detection of implanted fluorescence signals. Right image indicates that use of 620-nm emission filter allowed lowest number of tdTomato cells implanted to be detected (fluorescent signal indicated by arrow) above fur autofluorescence. (D) Images illustrating excellent sensitivity of optical imaging. False color overlay at left, with region of interest circled, shows detection of fluorescence signal from 9.25 × 10⁶ implanted tdTomato-expressing cells using 0.01-s exposure and 620-nm emission filter. Faint fluorescence from implant has been made visible by enhancing central image with PhotoShop (Adobe), as indicated by arrow in image at right.

FIGURE 10

Selection of breast cancer cells metastatic to lung. Representative lungs harvested at necropsy and BLI of indicated cell lines are shown after tail-vein injection. Color scale depicts photon flux (photons/s) emitted from xenografted mice. (Adapted with permission of (38).)

FIGURE 11

Comparison of expression of CMV-mtfl-mrfp1-wttk and CMV-mtfl-jred-wttk vectors by fluorescence and BLI in living mice. (A) Two and 4 million 293T cells were transiently transfected with CMV-mtfl-jred-wttk (left) and CMV-mtfl-mrfp1-wttk (right) vectors, respectively, and implanted on dorsal side of nude mouse (sites 1 and 2 have 2 million cells and sites 3 and 4 have 4 million cells) and imaged for fluorescence using Maestro system. Fluorescence signal from cells expressing CMV-mtfl-mrfp1-wttk was clearly visible (2 and 4). However, cells expressing CMV-mtfl-jred-wttk plasmid (1 and 3) show faint fluorescence signal. (B) Same mouse was then injected with D-luciferin and imaged for bioluminescence in Xenogen IVIS 200 optical system. Bioluminescence signal showed equal signal intensity by both groups of cells (2 and 4 vs. 1 and 3). (Adapted with permission of (183).)

FIGURE 12

Image reconstruction to estimate penetration depth of fluorescent signal from tdTomato protein. Three-dimensional (3D) reconstruction is shown in center panel. False-colored contour map represents photon densities that have been assigned coloration on basis of multicolored bar spectrum (photons/mm³ × 10⁻³) shown at right of reconstruction. To obtain this map, program fills structure image (left panel) with cubic voxels and scores fluorescence density of each voxel. Red and dark circles within reconstruction indicate sources of fluorescence that lay on selected coronal (violet), sagittal (blue), and transaxial (green) planes. As seen from black-to-red bar spectrum at lower right of reconstruction, intensity of fluorescence from these sources is on order of 10¹² photons/s. Enlarged sagittal and transaxial sections indicate that dorsal–ventral depth of animal was about 2.5 cm. Center (red dot) of fluorescent source from within rib cage cavity was about half way through animal or approximately 1.0 cm from ventral surface. 3D reconstructions were performed using 3D analysis software (Living Imaging; Xenogen Corp.) according to instruction manual from images collected at week 15, with 2.5-s exposure, and at emission wavelengths of 620 and 660 nm. 3D reconstruction program was designed for use with bioluminescence data and does not account for attenuation of fluorescence excitation signal by absorption and scatter. However, because emission wavelengths used (620 and 660 nm) are similar to those used during luciferase-based BLI, and photon densities and source intensities are similar to those observed with BLI, we have been advised that these results likely represent good first estimate of tissue penetration (communications with technical support at Xenogen Corp.).

FIGURE 13

Optical imaging of metastatic potential. Prostate tumors LAPC-4 and LAPC-9 are implanted and resected on reaching 1 cm (day 0). BLI is performed on day 20 then again on day 40. No metastases are evident from more indolent LAPC-9 tumors. LAPC-9 tumors would generate metastases after transfection with VEGF-C (data not shown). (Images courtesy of Dr. Lily Wu, UCLA.)