Differences in colonic crypt morphology of spontaneous and colitis-associated murine models via second harmonic generation imaging to quantify colon cancer development

Colorectal cancer remains the second leading cause of cancer death in the United States, and increased risk in patients with ulcerative colitis (a subset of inflammatory bowel disease) has motivated studies into early markers of dysplasia. The development of clinically translatable multiphoton imaging systems has allowed for the potential of in vivo label-free imaging of epithelial crypt structures via autofluorescence and/or second harmonic generation (SHG). SHG has been used to investigate collagen structures in various types of cancer, though the changes that colorectal epithelial collagen structures undergo during tumor development, specifically colitis-associated tumors, have not been fully investigated.

The results provide insight into the ability of SHG imaging to yield relevant data about the crypt microstructure in colorectal epithelium, specifically the potential to distinguish between spontaneous and colitis-associated murine models using quantification of crypt shape and distribution, informing future design of translational multiphoton imaging systems and protocols.

This study used two murine models, using A/J mice, one for spontaneous carcinoma and one for colitis-associated carcinoma, to investigate and quantify SHG image features that could potentially inform future study designs of endoscopic multiphoton imaging systems. The spontaneous tumor model comprised a series of six weekly injections of azoxymethane (AOM model). The colitis-associated tumor model comprised a single injection of AOM, followed by cycles of drinking water with dissolved dextran sodium sulfate salt (AOM-DSS model). SHG images of freshly resected murine colon were acquired with a multiphoton imaging system, and image features, such as crypt size, shape and distribution, were quantified using an automated algorithm.

The comparison of quantified features of crypt morphology demonstrated the ability of our quantitative image feature algorithms to detect differences between spontaneous (AOM model) and colitis-associated (AOM-DSS model) murine colorectal tissue specimens. There were statistically significant differences in the mean and standard deviation of nearest neighbor (distance between crypts) and circularity between the Control cohort, AOM and AOM-DSS cohorts. We also saw significance between AOM and AOM-DSS cohorts when calculating nearest neighbor in images acquired at fixed depths.