Background: Colorectal neoplasia is the second leading cause of cancer related deaths in the United States, resulting in approximately 149,000 new cases and about 55,000 deaths each year. This disease persists despite the wide practice of routine screening with colonoscopy. In addition to polyps, cancer may arise from sporadic flat lesions or from foci of chronic inflammatory bowel disease, both of which are endoscopically invisible on conventional white light endoscopy. An optical molecular imaging strategy that uses peptides that preferentially bind to pre-malignant mucosa and can be detected by endoscopic and confocal fluorescence imaging represents a new approach for the prevention and early management of this disease. Peptides offer many advantages over that of conventional probe development techniques, such as antibody or small molecule, that include low immunogenicity, short plasma half life, high complexity, good mucosal penetration and ease of use. Phage display is an unbiased, combinatorial approach for developing selective binding peptides. Methods: The selection of specific binding peptides is performed by biopanning the phage library, which contains a large number and diversity of peptides expressed by M13 bacteriophage, against target tissues. First, non-specific peptides are cleared from the library by biopanning against cells in culture (CRL 7869) that have intestinal morphology. Cultured cells are used to provide the large surface area needed to remove over 99% of the original library. Next, the remaining phage from the supernatant are biopanned first against biopsy specimens of normal colonic mucosa to remove additional non-specific binding phage, and then against pinch biopsies of colonic adenomas (dysplasia). The phage that binds to dysplastic mucosa are then eluted with acid, and screened for optimal binding affinity and avidity against 10 micron frozen sections of excised adenomas. Several candidate peptides associated with phage that exhibit the highest target-to-background ratio were sequenced, synthesized, purified by HPLC, and conjugated to FITC. After informed consent, 10 patients scheduled for routine endoscopy were sent for serum CBC, chemistries, and liver function tests. The candidate peptide (100 μM in 5 ml normal saline) was then topically administered by spraying onto adenomas found on routine colonoscopy of a patient undergoing routine screening and those with known flat lesions. After about a 1 minute of incubation period, the lesion is first imaged remotely by wide area fluorescence to demonstrate the ability to localize the lesion and then by a flexible, fiber-optic confocal microendoscope inserted through the instrument channel of the endoscope to observe sub-surface binding in the adenoma and surrounding normal colonic mucosa. A pinch biopsy was then taken from each site imaged, and compared with the appropriate confocal image. Follow up blood tests are taken 24 hours to assess for potential toxicity from the peptide reagents. Results: Endoscopic fluorescence images were collected from regions of colonic mucosa in vivo that varied in surface area from approximately 3 to 75 cm2, and revealed increased fluorescence intensity at the site of the adenomas compared to that of the surrounding normal mucosa. At a depth of 50 microns, the confocal fluorescence microscopy images showed increased fluorescence intensity in a pattern corresponding to the colonocytes surrounding the dysplastic crypts compared to that of the normal crypts. The target-to-background ratio between dysplastic and normal colonocytes measured on confocal images collected at the border of the adenoma, was ~7 with an average SNR of ~5. No differences were observed in fluorescence intensity of dysplastic and normal at the border with administration of a peptide that had the same amino acids but with the sequence scrambled. Furthermore, no differences were seen in the fluorescence intensity at the border of hyperplastic polyps. These results were confirmed on conventional confocal microscopy images collected ex vivo from the pinch biopsy. Finally, no toxicity associated with peptide administration was found. Conclusions: In vivo endoscopic images can detect fluorescence peptides that are topically administered onto colonic adenomas during routine colonoscopy. Furthermore, preferential peptide binding to dysplastic foci can be observed by confocal microendoscopy. This integrated optical imaging strategy provides a targeted approach to localize and identify pre-malignant mucosa in the colon and has potential to improve our ability to prevent the occurrence and to manage the consequences of this disease.

[Fifth AACR International Conference on Frontiers in Cancer Prevention Research, Nov 12-15, 2006]