In vivo molecular mapping of the tumor microenvironment in an azoxymethane-treated mouse model of colon carcinogenesis

Abstract

Background and objective: Development of miniaturized imaging systems with molecular probes enables examination of molecular changes leading to initiation and progression of colorectal cancer in an azoxymethane (AOM)-induced mouse model of the disease. Through improved and novel studies of animal disease models, more effective diagnostic and treatment strategies may be developed for clinical translation. We introduce use of a miniaturized multimodal endoscope with lavage-delivered fluorescent probes to examine dynamic microenvironment changes in an AOM-treated mouse model.

Study design/materials and methods: The endoscope is equipped with optical coherence tomography (OCT) and laser induced fluorescence (LIF) imaging modalities. It is used with Cy5.5-conjugated antibodies to create time-resolved molecular maps of colon carcinogenesis. We monitored in vivo changes in molecular expression over a five month period for four biomarkers: epithelial growth factor receptor (EGFR), transferrin receptor (TfR), transforming growth factor beta 1 (TGFβ1), and chemokine (C-X-C motif) receptor 2 (CXCR2). In vivo OCT and LIF images were compared over multiple time points to correlate increases in biomarker expression with adenoma development.

Results: This system is uniquely capable of tracking in vivo changes in molecular expression over time. Increased expression of the biomarker panel corresponded to sites of disease and offered predictive utility in highlighting sites of disease prior to detectable structural changes. Biomarker expression also tended to increase with higher tumor burden and growth rate in the colon.

Conclusion: We can use miniaturized dual modality endoscopes with fluorescent probes to study the tumor microenvironment in developmental animal models of cancer and supplement findings from biopsy and tissue harvesting.

Keywords: endoscopy; fluorescence; microenvironment; molecular imaging; optical coherence tomography.

Figure 1

Data analysis and graphical summary for tumor (OCT) and fluorescence emission intensity (LIF) map time point comparisons. Tumor and fluorescence emission intensity maps were each discretized to 5x8 matrices. Sensitivity and specificity of the fluorescent marker for tumor areas were evaluated over a range of intensity values using tumor maps as truth and comparing to fluorescence emission intensity maps collected at the same or earlier time points. The areas under the resulting ROC curves (AUC) were then calculated; the numerical values converted to a grey scale, and used to fill in elements of a grid displaying correlation between tumor and fluorescence emission intensity maps over time.

Figure 2

Images from a single AOM-treated mouse colon with lavage-delivered EGFR-targeted fluorescent antibodies, obtained at the final time point. (a) Gross tissue image of the excised colon, sliced longitudinally and opened with the lumen side up. The locations of adenoma are highlighted with black boxes. (b) Fluorescence microscope image of the same excised colon using Cy5.5 emission and excitation filters. White boxes correspond to adenoma locations shown in panel a. Scale bar on panel b applies to panels a and b and represents 5 mm. The distal colon is sampled in vivo with 30 mm longitudinal scans collected at 8 evenly spaced rotations; the 8 rotations are vertically combined to provide tumor location (c) and contrast agent distribution (d) from the perspective of the colon lumen. The horizontal axis is lateral location and the vertical axis represents in vivo rotation, with the ventral rotation at the bottom. Proximal to distal colon is oriented from left to right. (c) Representative tumor map of the colon derived from OCT images collected in vivo. Tumor locations are indicated in red. Space between consecutive vertical lines represents 5 mm increments. (d) Molecular map of EGFR expression derived from LIF spectra obtained in vivo. Intensity map represents Cy5.5 fluorescence intensity on an arbitrary units scale from 0 to 1000 (blue to red). Panels c and d are proportionately scaled.

Figure 3

Cy5.5 fluorescence of targeted agents delivered in vivo (left) and immunohistochemical staining using DAB (right) for each of four biomarkers: EGFR(a), TfR(b), TGFβ1(c), and CXCR2(d). Fluorescence and IHC images for each marker were collected from similar regions of the same colon. Scale bar applies to all images and represents 100 um.

Figure 4

Time series of OCT and LIF derived tumor and fluorescence emission intensity maps of the distal colon for AOM-treated mice administered TfR-targeted (left) or nonspecific (right) fluorescent antibodies. OCT and LIF data are collected simultaneously. The most distal 5 mm have been removed from the images as this portion was not used for subsequent fluorescence or ROC analysis. Colons were sampled at 16, 20, 24, and 28 weeks of age. Black indicates adenoma location in OCT-derived tumor maps. The intensity scale applies to Cy5.5 fluorescence emission intensity in LIF-derived maps and has an arbitrary units scale from 0 to 1000.

Figure 5

Plots of average Cy5.5 fluorescence emission intensity of the distal mouse colon as a function of tumor burden (left column) or tumor growth (right column) for EGFR (a, b), TfR (c, d), TGFβ1 (e, f), or CXCR2 (g, h) –targeted or nonspecific (i, j) fluorescent antibodies. Tumor burden is approximated using OCT images of the colon. For plots of fluorescence vs. tumor burden, points are collected at 16 (+), 20 (●), 24 (▼), and 28 (□) weeks of age. For plots of fluorescence vs. tumor growth, points are derived from collected data between 16 and 20 (●), 20 and 24 (▼) and 24 and 28 (□) weeks of age. Linear regression fits account for all mice and time points of a treatment group and are represented as dotted lines. R-squared values are presented in the bottom right corner of each graph.

Figure 6

Location correlation between tumor and fluorescence emission intensity maps for EGFR- (a), TfR- (b), TGFβ1- (c), or CXCR2- (d) targeted, or nonspecific (e) fluorescent antibodies. Tumor maps were computationally correlated to fluorescence emission intensity maps collected at the same time point as the tumor map (positive diagonal) and each previous time point (lower right-hand elements). Grey scale represents the area under the ROC curve comparing fluorescence emission intensity vs. presence or absence of disease over a range of fluorescence intensities for each agent. The grey scale applied to the area under the curve for each tumor map/accumulation map comparison is provided in the figure, with OCT time point on the horizontal axis and LIF time point on the vertical axis. The intensity scale applies to all figures and ranges from 0 to 1. Black represents a null value.