Glycan-Based Near-infrared Fluorescent (NIRF) Imaging of Gastrointestinal Tumors: a Preclinical Proof-of-Concept In Vivo Study

Abstract

Purpose: Aberrantly expressed glycans in cancer are of particular interest for tumor targeting. This proof-of-concept in vivo study aims to validate the use of aberrant Lewis glycans as target for antibody-based, real-time imaging of gastrointestinal cancers.

Procedures: Immunohistochemical (IHC) staining with monoclonal antibody FG88.2, targeting Lewis^a/c/x, was performed on gastrointestinal tumors and their healthy counterparts. Then, FG88.2 and its chimeric human/mouse variant CH88.2 were conjugated with near-infrared fluorescent (NIRF) IRDye 800CW for real-time imaging. Specific binding was evaluated in vitro on human gastrointestinal cancer cell lines with cell-based plate assays, flow cytometry, and immune-fluorescence microscopy. Subsequently, mice bearing human colon and pancreatic subcutaneous tumors were imaged in vivo after intravenous administration of 1 nmol (150 μg) CH88.2-800CW with the clinical Artemis NIRF imaging system using the Pearl Trilogy small animal imager as reference. One week post-injection of the tracer, tumors and organs were resected and tracer uptake was analyzed ex vivo.

Results: IHC analysis showed strong FG88.2 staining on colonic, gastric, and pancreatic tumors, while staining on their normal tissue counterparts was limited. Next, human cancer cell lines HT-29 (colon) and BxPC-3 and PANC-1 (both pancreatic) were identified as respectively high, moderate, and low Lewis^a/c/x-expressing. Using the clinical NIRF camera system for tumor-bearing mice, a mean tumor-to-background ratio (TBR) of 2.2 ± 0.3 (Pearl: 3.1 ± 0.8) was observed in the HT-29 tumors and a TBR of 1.8 ± 0.3 (Pearl: 1.9 ± 0.5) was achieved in the moderate expression BxPC-3 model. In both models, tumors could be adequately localized and delineated by NIRF for up to 1 week. Ex vivo analysis confirmed full tumor penetration of the tracer and low fluorescence signals in other organs.

Conclusions: Using a novel chimeric Lewis^a/c/x-targeting tracer in combination with a clinical NIRF imager, we demonstrate the potential of targeting Lewis glycans for fluorescence-guided surgery of gastrointestinal tumors.

Keywords: Aberrant glycosylation; Carbohydrates; Fluorescence-guided surgery; Lewis glycans; Monoclonal antibody.

Fig. 1.

a Immunohistochemical FG88.2 staining in a colon tumor and in normal colonic crypts. b FG88.2 staining in a gastric tumor and in normal gastric glands. c, d FG88.2 staining in pancreatic tumor tissue and pancreatitis tissue (and normal pancreatic tissue derived from the same patient (III)). Red-dotted lines represent the tumor (c) or pancreatitis-normal pancreatic tissue border (d). Overview images are taken at × 50 magnification and inserts at × 200 magnification. Scale bars represent 500 μm and 100 μm, respectively. Scale bars represent 500 μm and 100 μm for overview and insert images, respectively.

Fig. 2.

a Cell-based plate assay of FG88.2-800CW at 1.25, 2.5, 5, and 10 μg/ml dilutions on gastrointestinal cell lines. b Flow cytometry of FG88.2 on HT-29, BxPC-3, and PANC-1. Red-dotted lines represent conjugate controls and blue lines represent FG88.2 fluorescence signals. c Immunofluorescence analysis of CH88.2-800CW binding on HT-29, BxPC-3, and PANC-1 cells. AF488 signals and 800CW signals are represented in green and red, respectively. DAPI was used to stain nuclei (blue channel).

Fig. 3.

a Average tumor and background MFIs over time in HT-29 colon cancer-bearing mice injected with CH88.2-800CW using the Pearl preclinical imager. b Mean TBRs over time. c Representative black-and-white, NIRF, and merged images of HT-29 tumor-bearing mice at 72 h and 96 h post-injection.

Fig. 4.

a Average tumor and background MFIs over time in BxPC-3 pancreatic cancer-bearing mice injected with CH88.2-800CW using the Pearl preclinical imager. b Mean TBR over time. c Representative black-and-white, NIRF, and merged images of BxPC-3 tumor-bearing mice at 72 h and 96 h post-injection.

Fig. 5.

a Representative color, NIRF, and merged images of CH88.2-800CW binding specificity in a HT-29 tumor-bearing mouse model using the clinical Artemis NIR imaging system at 150-ms exposure. Regions of interest were selected in similar fashion to the Pearl as shown by the red and blue shapes, corresponding to the tumor and background area, respectively (only displayed in the left figure). To allow better visualization of the field of interest, the tumor-bearing skin was manually mobilized to the center of the camera’s optical field as is displayed by left and right back images. b Representative images of CH88.2-800CW binding specificity in a BxPC-3 tumor-bearing mouse model.

Fig. 6.

a Representative examples of ex vivo hematoxylin-eosin (HE) staining, NIR fluorescence heatmap (800 nm), and FG88.2 staining on HT-29 and BxPC-3 tumor tissue sections. Overview images are taken at × 25 magnification and inserts at × 100 magnification. Scale bars represent 500 μm and 100 μm for overview and insert images, respectively. b Average tumor-to-liver, tumor-to-colon, and tumor-to-pancreas ratios in HT-29 and BxPC-3 tumor-bearing mice at 168 h/1 week post-injection. c Biodistribution of CH88.2-800CW at 168h/1week post-injection expressed as tumor or organ MFI. d  Ex vivo fluorescent images of resected tumors and organs. Sk, skin; Hrt, heart; Lu, lungs; Li, liver; St, stomach; Sp, spleen; Pa, pancreas; Du, duodenum; Co, colon; Ki, kidneys; Mu, muscle; Tu, tumors (under brackets); Br, brain.