Detection accuracy for epithelial dysplasia using an objective autofluorescence visualization method based on the luminance ratio

Abstract

The autofluorescence visualization method (AVM) uses blue excitation light to assist in the diagnosis of epithelial dysplasia. It detects epithelial dysplasia as a black area, which is known as fluorescence visualization loss (FVL). In this study, we evaluated the detection accuracy for epithelial dysplasia of the tongue using the objective AVM and assessed its possible clinical utility. Seventy-nine tongue specimens clinically suspected to have leukoplakia or squamous cell carcinoma (SCC) were analyzed. First, the AVM was subjectively performed using the Visually Enhanced Lesion scope (VELscope), and the iodine-staining method was then performed. After biopsy, the histopathological results and the luminance ratio between the lesion and healthy tissue were compared, and a receiver operating characteristic curve was created. The cutoff value for the objective AVM was determined; the lesion was considered FVL-positive or -negative when the luminance ratio was higher or lower than the cutoff value, respectively. The histopathological diagnoses among the 79 specimens were SCC (n=30), leukoplakia with dysplasia (n=34), and leukoplakia without dysplasia (n=15). The cutoff value of the luminance ratio was 1.62, resulting in 66 FVL-positive and 13 FVL-negative specimens. The luminance ratio was significantly higher in the epithelial dysplasia-positive than -negative group (P<0.000 1). The objective AVM showed much higher consistency between histopathological results than did the two methods (kappa statistic=0.656). In conclusion, objective autofluorescence visualization has a potential as an auxiliary method for diagnosis of epithelial dysplasia.

Figure 1

Mechanism of autofluorescence visualization method using the VELscope. Blue excitation light from the VELscope is emitted to the oral mucosa, and the fluorescence energy from the endogenous autofluorescence substances in the healthy mucosa emits green light. Conversely, low levels of endogenous autofluorescence substances in the tissue affected by epithelial dysplasia cannot emit green light. Therefore, epithelial dysplasia can be detected as fuorescence visualization loss (FVL), which is shown as a dark area.

Figure 2

Schema of definition of cutoff value. First, each lesion-to-normal tissue uptake ratio was entered into EZR statistics software, and a receiver operating characteristic curve was created. The area under the curve was 0.84, indicating moderate accuracy. Next, the maximum point of addition of the sensitivity and specificity (1.620) was defined as the cutoff value of lesion-to-normal tissue uptake ratio.

Figure 3

Typical lesion well detected by both autofluorescence visualization method and iodine-staining method. (a) Macroscopically, there was an outwardly growing lesion on the lateral edge of the tongue. The lesion showed (b) fluorescence visualization loss with fluorescent light emitted from the VELscope and (c) an iodine-unstained area by the iodine-staining method. (d) Histopathologically, a multi-layered epithelium and destroyed structure of the epithelial basal membrane were observed leading to diagnosis of squamous cell carcinoma. Bar=500 μm.

Figure 4

Distribution of the lesion-to-normal tissue uptake ratio in the epithelial dysplasia-positive and -negative groups. The lesion-to-normal tissue uptake ratio of the dysplasia-positive group was 2.71, which was significantly higher than that of the dysplasia-negative group (1.68) (P<0.000 1).

Figure 5

False-negative specimen by the objective autofluorescence visualization method. (a) There was an outwardly growing lesion on the lateral edge of the tongue. (b) The lesion did not show apparent fluorescence visualization loss with fluorescent light emitted from the VELscope leading to low luminance ratio. One of the reasons could be that the lesion is located on the lateral edge of the tongue. Another reason could be possible genetic alteration of the specimen. (c) However, the lesion was iodine-unstained. (d) The histopathological diagnosis was squamous cell carcinoma (Specimen No. 2). Bar=500 μm.

Figure 6

False-negative specimen by the iodine-staining method. (a) There was also an outwardly growing lesion on the lateral edge of the tongue. (b) The lesion showed relatively apparent fluorescence visualization loss with fluorescent light emitted from the VELscope. (c) The same lesion was iodine-unstained. (d) The histopathological diagnosis was squamous cell carcinoma (Specimen No. 4). Bar=500 μm.