Abstract
The epithelial-connective tissue interface (ECTI) plays an integral role in epithelial neoplasia, including oral squamous cell carcinoma (OSCC). This interface undergoes significant alterations due to hyperproliferating epithelium that supports the transformation of normal epithelium to precancers and cancer. We present a method based on nonlinear optical microscopy to directly assess the ECTI and quantify dysplastic alterations using a hamster model for oral carcinogenesis. Neoplastic and non-neoplastic normal mucosa were imaged in-vivo by both multiphoton autofluorescence microscopy (MPAM) and second harmonic generation microscopy (SHGM) to obtain cross-sectional reconstructions of the oral epithelium and lamina propria. Imaged sites were biopsied and processed for histopathological grading and measurement of ECTI parameters. An ECTI shape parameter was calculated based on deviation from the linear geometry (ΔLinearity) seen in normal mucosa was measured using MPAM-SHGM and histology. The ECTI was readily visible in MPAM-SHGM and quantitative shape analysis showed ECTI deformation in dysplasia but not in normal mucosa. ΔLinearity was significantly (p < 0.01) higher in dysplasia (0.41±0.24) than normal (0.11±0.04) as measured in MPAM-SHGM and results were confirmed in histology which showed similar trends in ΔLinearity. Increase in ΔLinearity was also statistically significant for different grades of dysplasia. In-vivo ΔLinearity measurement alone from microscopy discriminated dysplasia from normal tissue with 87.9% sensitivity and 97.6% specificity, while calculations from histology provided 96.4% sensitivity and 85.7% specificity. Among other quantifiable architectural changes, a progressive statistically significant increase in epithelial thickness was seen with increasing grade of dysplasia. MPAM-SHGM provides new noninvasive ways for direct characterization of ECTI which may be used in preclinical studies to investigate the role of this interface in early transformation. Further development of the method may also lead to new diagnostic approaches to differentiate non-neoplastic tissue from precancers and neoplasia, possibly with other cellular and layer based indicators of abnormality.