Abstract B56: Autofluorescence visualization to screen for premalignant and malignant lesions of oral cavity and oropharynx

Abstract

Abstract B56 Objective To determine if the addition of autofluorescence (AF) visualization to conventional white light exam (WL) improves the efficiency in detecting pre-malignant and malignant lesions in the oral cavity of high risk patients, compared to WL exam alone. Methods A total of 47 patients underwent the screening procedure at the department of Dentistry and Head and Neck Surgery at Roswell Park Cancer Institute, between October 2006 and June 2007, either to investigate suspicious oral lesions or to identify any co-existing lesions in patients with a recently diagnosed untreated oral pre-malignant lesions or cancer of oral cavity. Four patients had a second exam at least 6 months after the initial visit, making up a total of 51 screening visits. The patients underwent a comprehensive oral exam with an autofluorescence device which was equipped with WL source, an AF source (blue excitation light at 405 nm and imaging at 530-550 & 630-650 nm) and a video camera. Any area that looked fluorescent green was classified as normal and any area that showed loss of fluorescence (grey or black) was considered to be suspicious on AF. All oral exams were performed by one dentist and the biopsies were reviewed by the same pathologist. A total of 170 lesions were identified on either WL or AF and biopsied, and 1 contra-lateral control biopsy from a normal looking area was taken per visit. For convenience of analysis, the pathologic diagnoses were grouped into four categories: (1) ‘benign lesions’(BL) - normal/ unremarkable samples with benign conditions and parakeratosis without atypia (2) ‘low grade lesions’(LGL) - parakeratosis with atypia and mild dysplasia (3) ‘high grade lesions’(HGL) - moderate dysplasia (MD), severe dysplasia (SD) and CIS (4) ‘Carcinoma’(CA). These groups contributed 48, 129, 26 and 15 biopsies respectively. Results On lesion-by-lesion basis, while WL exam did not identify 8 HGLs (4 MD and 4 CIS), AF missed only one HGL (1 MD). Similarly, WL exam did not identify 4 CAs, whereas AF exam missed only 1 cancerous lesion. On combined sequential screening by WL followed by AF (WL+AF), none of the HGLs or CAs was missed. For the LGLs, the sensitivity improved from 44% with WL to 63% with AF and 76% with WL+AF exam. Similarly, sensitivity for detecting HGLs increased from 69% with WL to 96 and 100% with AF and WL+AF, respectively. For CAs, the sensitivity improved from 73% to 93 and 100% with AF and WL+AF respectively. The negative predictive value of WL for HGLs and CAs was 96 and 97% respectively, which increased to 100% with AF alone. The specificity for HGLs and CAs was superior with WL at 75% compared to 52% with AF alone. The addition of AF to WL increased the relative sensitivity by 83%, 47% and 45% in identifying LGLs, HGLs and CAs respectively. One per-person basis, we examined the efficiency of AF in identifying the highest grade lesion in a person. While, WL missed the highest grade lesions in 2 patients with MD, 2 with CIS and 4 with CAs; AF did not miss the worst lesion in any patient with a HGL or CA. The efficacy of AF in identifying the highest grade lesion in each individual was 100%. Conclusion Our results demonstrate that sequential exam by AFL visualization following a WL exam is very efficient in identifying pre-malignant and malignant oral lesions not detected with WL alone. This technique has the potential as a non-invasive and effective tool for oral cancer screening. Citation Information: Cancer Prev Res 2008;1(7 Suppl):B56.