The standard of care for premalignant lesion risk assessment is dysplasia grading by histopathology. With significant overlap between dysplasia grades and high inter- and intraobserver variations, histopathology dysplasia grading alone is not a useful prognostic tool. Our aim is to investigate whether a method for quantitatively assessing S100A7, a prognostic biomarker, using image analysis can better predict clinical outcome in cases with oral dysplasia.Study Design.
Using the Visiopharm image analysis system, we analyzed a cohort of 150 oral biopsy samples to build and test Straticyte, a model for individualized assessment of the 5-year risk of progression of oral precancerous lesions to invasive squamous cell carcinomas.Results.
Straticyte classified lesions more accurately than histopathological dysplasia grading for risk to progression to cancer over the following 5 years. The sensitivity of low-risk versus intermediate- and high-risk Straticyte groups was 95% compared to 75% for mild versus moderate and severe dysplasia. Furthermore, the negative predictive value for low-risk versus intermediate- and high-risk Straticyte groups was 78% compared to 59% for mild versus moderate and severe dysplasia.Conclusion.
By quantitatively assessing S100A7, Straticyte better defines the risk for developing oral squamous cell carcinoma than histopathological dysplasia grading alone.