Source: http://cebp.aacrjournals.org
Author: Calum MacAulay, Martial Guillaud, Lewei Zhang, Catherine Poh, and Miriam Rosin
 

Abstract: Oral cancer like many epithelial cancers which are readily accessible are much more treatable if caught in their early pre-invasive/minimally invasive stages. The oral cavity is easily examined and sampled. In British Columbia we have established an Oral Cancer Prevention Program in which we are evaluating and investigating several technologies and their interactions for the screening and follow-up of oral cancer to be implemented into a population based screening program. These sensitive “field of view” image based screening technologies are generally sensitive for the detection of suspect OPLs (oral premalignant lesions), but can highlight areas who’s actual characteristics may be masked by inflammation/ulceration and other conditions. As part of a comprehensive management program we present our pilot data on the use of oral cytological samples collected by targeted brushing and analyzed by a fully automated high resolution image cytometry device (cyto-savant). For this study we collected 196 cytological samples using targeted brushing of select areas in the oral cavity from individuals with squamous cell carcinoma (SCC), carcinoma in situ (CIS), severe dysplasia, no areas of abnormality and subjects with areas of inflammation/infection etc. All of these samples were spun down onto slides and the DNA quantitatively labeled with a modified Feulgen-Thionin stain and the slides automatically scanned by the cyto-savant. For each object (cell/debris) imaged ~120 features were calculated and used by a cell recognition decision tree (originally trained for cervical cell recognition) to differentiate cells from debris. The 108 samples from known normal areas and 60 samples from (SCC, CIS and severe dysplasia) were used to determine the appropriate thresholds for the frequency of cells displaying characteristics previously known to be associated with cancer detection (in cervix and lung). Using these thresholds the system correctly identified 86% of the abnormal cases and 86% of the normal cases. When tested on the 28 inflammation/infection cases the system correctly identified 92% of the evaluable samples as non OPLs. These pilot results indicate that image cytometry may have a roll in oral cancer screening. Supported by grant RO1-DE017013-01, NIDCR

 

*This news story was resourced by the Oral Cancer Foundation, and vetted for appropriateness and accuracy.