Early diagnosis of thyroid-associated ophthalmopathy using label-free Raman spectroscopy and multivariate analysis.

Thyroid-associated ophthalmopathy (TAO) is the most common orbital disease in adults, with complex clinical manifestations and significant impacts on the life quality of patients. The current diagnosis of TAO lacks reliable biomarkers for early and non-invasive screening and detection, easily leading to poor prognosis. Therefore, it is essential to explore new methods for accurately predicting TAO development in its early stage. In this study, Raman spectroscopy, with non-destructive, label-free, and high-sensitivity characteristics, was used to analyze the differences in biochemical components of orbital adipocyte and tear samples between TAO and control groups. Furthermore, a multivariate analysis method (i.e., Principal Component Analysis-Linear Discriminant Analysis (PCA-LDA)) was applied for data processing and analysis. Compared with controls, PCA-LDA yielded TAO diagnostic accuracies of 72.7% and 75.0% using orbital adipocytes and tears, respectively. Our proof-of-concept results suggest that Raman spectroscopy holds potential for exploring the underlying pathogenesis of TAO, and its potential application in early screening of other thyroid-associated diseases can be further expanded.