Multi-modality diagnosis of oral premalignancy and malignancy

Objectives: Despite significant advances in treatment, oral cancer still results in 10,000 U.S. deaths annually. Early detection of cancer and its curable precursors remains the best way to ensure patient survival and quality of life. Specific aim of this study was to use a unique combination of innovative non-invasive optical in vivo technologies to test a multi-modality approach to non-invasive diagnostics of oral premalignancy and malignancy. Methods: In the hamster cheek pouch model (120 hamsters), in vivo two-dimensional and tomographic optical coherence tomography (OCT) and optical Doppler tomography (ODT) mapped epithelial, subepithelial and vascular change throughout carcinogenesis in specific, marked sites. In vivo multi-wavelength multi-photon (MPM) and second harmonic generated (SHG) fluorescence techniques provided parallel data on surface and subsurface tissue structure, specifically collagen presence and structure, cellular presence and vasculature. Images were diagnosed individually, on 2 separate occasions, by 2 blinded, pre-standardized investigators using a standardized scale of pathology from 0-6 for all modalities. After sacrifice, routine histopathological sections were prepared and pathology evaluated for each location imaged as described above on a scale of 0-6. ANOVA techniques compared imaging diagnostics with histopathology. 95% confidence limits of the sensitivity and specificity were established for the diagnostic capability of OCT/ODT+ MPM/SHG using ROC curves and kappa statistics. Results: Imaging data were reproducibly obtained with good accuracy. Carcinogenesis-related structural and vascular changes were clearly visible to tissue depths of 2mm. Sensitivity (OCT/ODT alone: 71-88%; OCT+MPM/SHG: 79-91%) and specificity (OCT alone: 62-83%;OCT+MPM/SHG: 67-90%) compared well with conventional techniques. Conclusions: OCT/ODT and MPM/SHG are promising non-invasive in vivo diagnostic modalities for oral dysplasia and malignancy. Supported by CRFA 27722, CRFA 30003, CCRP 00-01391V-20235, NIH RO21 CA8752701, TRDRP 71T-0192, NIH (LAMMP) RR01192, DOE DE903-91ER 61227, NIH EB-00293 CA91717, NSF BES-86924, TRDRP 9RT-0094.