Concordance Study on Molecular Detection of Thyroid Nodules by Ultrasound Guided Fine Needle Aspiration
Su Meng, Jianxiong Sun, Changliang Zhang, Zhixue Yang, Jinrong Wei, Si Chen, Yanan Li, Xiaoxiao Dai, Guoqin JiangABSTRACT
Background
Fine needle aspiration (FNA) constitutes the primary preoperative evaluation for thyroid nodules. Targeted Next‐generation sequencing (NGS) of FNA samples increasingly informs clinical assessment. Molecular concordance between FNA and surgical tissues may be modulated by lesion diameter and heterogeneity.
Methods
Among 91 prospectively enrolled patients with suspected thyroid nodules, FNA–fresh surgical tissue matching yielded 76 patients per cohort: 85 matched pairs for DNA NGS, 82 pairs from 72 patients for RNA NGS after four pairs failed library construction. Concordance was assessed by Cohen's kappa, positive (PPA) and negative (NPA) percent agreement; generalized estimating equations and patient‐level cluster bootstrap addressed within‐patient clustering.
Results
DNA NGS concordance was 88.24% ( κ = 0.60, cluster‐adjusted 95% CI: 0.37–0.79; PPA 89.04%, NPA 83.33%); RNA NGS concordance was 98.78% ( κ = 0.88, 95% CI: 0.54–1.00). DNA yield was higher in surgical tissue than FNA ( p < 0.0001); library yield did not differ ( p = 0.376). Lesion diameter was comparable between concordant and discordant pairs ( p = 0.840; minimum detectable effect d ≈0.95). Concordance did not differ between solitary‐ and multiple‐nodule specimens ( p = 0.418) or between malignant and benign lesions on both platforms, although patient‐level discordance was numerically higher in multinodular disease (37.5% vs. 10.3%).
Conclusion
Inter‐specimen concordance is driven mainly by tumor heterogeneity; lesion diameter is unlikely to be a major determinant. Tumor cell proportion, though unquantified, remains a plausible contributor. Surgical tissue yields richer profiles but is not a perfect reference. Standardized FNA sampling and separate puncture of each suspicious nodule may improve concordance.