Detailed analysis of NANOG expression in head and neck precancerous and cancerous lesions: an approach using QuPath digital image analysis

초록

Background & objectives: The grading criteria for head and neck (H&N) precancerous lesions is obscure, leading to low observer reproducibility. In this study, we performed NANOG immunohistochemical (IHC) staining on H&N dysplastic lesions and quantitatively evaluated their expression using QuPath digital image analysis. Methods: Tissue from 68 patients diagnosed with oral/laryngeal dysplasia or invasive carcinoma was classified into four groups through a central review: normal (NL), low-grade dysplasia (LG), high-grade dysplasia (HG), and invasive carcinoma (CA). After performing NANOG IHC, the scanned IHC images were annotated by group. Subsequently, the intensity of NANOG expression for each group was quantitatively evaluated using QuPath. Results: NANOG immunostaining revealed predominantly cytoplasmic staining patterns in both dysplastic and invasive carcinoma areas, but it also stained normal submucous glands and ciliated cells. Using the ‘cell detection’ tool in QuPath, the NANOG staining intensity (measured as mean optical density per cytoplasm; mOD) of cells within each group was evaluated. The results showed a gradual increase in mOD from NL through LG to HG, followed by a decrease in CA compared to HG (NL: 0.06±0.02, LG: 0.09±0.03, HG: 0.18±0.08, CA: 0.16±0.07). Significant differences in mOD were observed in the pair-wise comparisons of NL- LG, LG-HG, and HG-CA (P <0.001, all). Conclusion: Currently, there is no established biomarker to aid in the grading H&N dysplastic lesions. In this study, to minimize tissue heterogeneity, cases diagnosed with dysplasia or cancer were annotated into NL, LG, HG, and CA groups through central review. Subsequently, we quantitatively confirmed differences in NANOG expression across these groups. Notably, the differences in NANOG intensity between the groups were also visually discernible. This suggests that NANOG IHC could enhance the accuracy and improve diagnostic reproducibility of H&N dysplastic lesions.