Artificial intelligence defines protein-based classification of thyroid nodules

• Published in: Cell discovery Vol. 8; no. 1; p. 85
• Main Authors: Sun, Yaoting, Selvarajan, Sathiyamoorthy, Zang, Zelin, Liu, Wei, Zhu, Yi, Zhang, Hao, Chen, Wanyuan, Chen, Hao, Li, Lu, Cai, Xue, Gao, Huanhuan, Wu, Zhicheng, Zhao, Yongfu, Chen, Lirong, Teng, Xiaodong, Mantoo, Sangeeta, Lim, Tony Kiat-Hon, Hariraman, Bhuvaneswari, Yeow, Serene, Alkaff, Syed Muhammad Fahmy, Lee, Sze Sing, Ruan, Guan, Zhang, Qiushi, Zhu, Tiansheng, Hu, Yifan, Dong, Zhen, Ge, Weigang, Xiao, Qi, Wang, Weibin, Wang, Guangzhi, Xiao, Junhong, He, Yi, Wang, Zhihong, Sun, Wei, Qin, Yuan, Zhu, Jiang, Zheng, Xu, Wang, Linyan, Zheng, Xi, Xu, Kailun, Shao, Yingkuan, Zheng, Shu, Liu, Kexin, Aebersold, Ruedi, Guan, Haixia, Wu, Xiaohong, Luo, Dingcun, Tian, Wen, Li, Stan Ziqing, Kon, Oi Lian, Iyer, Narayanan Gopalakrishna, Guo, Tiannan
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 06.09.2022; Springer Nature B.V; Nature Publishing Group
• Subjects: 631/337/475; 631/67/70; Accuracy; Artificial intelligence; Biomarkers; Biomedical and Life Sciences; Biopsy; Cell Biology; Cell Culture; Cell Cycle Analysis; Cell Physiology; Life Sciences; Malignancy; Neural networks; Nodules; Paraffin; Proteins; Proteomes; Proteomics; Stem Cells; Thyroid; Thyroid gland
• ISSN: 2056-5968; 2056-5968
• DOI: 10.1038/s41421-022-00442-x

Determination of malignancy in thyroid nodules remains a major diagnostic challenge. Here we report the feasibility and clinical utility of developing an AI-defined protein-based biomarker panel for diagnostic classification of thyroid nodules: based initially on formalin-fixed paraffin-embedded (FFPE), and further refined for fine-needle aspiration (FNA) tissue specimens of minute amounts which pose technical challenges for other methods. We first developed a neural network model of 19 protein biomarkers based on the proteomes of 1724 FFPE thyroid tissue samples from a retrospective cohort. This classifier achieved over 91% accuracy in the discovery set for classifying malignant thyroid nodules. The classifier was externally validated by blinded analyses in a retrospective cohort of 288 nodules (89% accuracy; FFPE) and a prospective cohort of 294 FNA biopsies (85% accuracy) from twelve independent clinical centers. This study shows that integrating high-throughput proteomics and AI technology in multi-center retrospective and prospective clinical cohorts facilitates precise disease diagnosis which is otherwise difficult to achieve by other methods.