Pre-operative diagnosis of suspicious thyroid nodules ‒ an integrated multidisciplinary approach

Abstract

Introduction and aim. Thyroid nodules are common, though few are malignant. In lower-middle-income countries such as Sri Lanka, limited diagnostic resources make accurate preoperative assessment essential to avoid unnecessary surgery. While the Thyroid Imaging reporting and Data System (TIRADS) and Bethesda systems are widely used, both have limitations, particularly in differentiating follicular lesions. This study aimed to evaluate the potential value of immunocytochemistry (ICC) as an adjunct to preoperative diagnosis in suspicious thyroid nodules.

Material and methods. A prospective observational diagnostic accuracy study was conducted on 106 patients with TIRADS ≥3 nodules. Fine-needle aspiration cytology was classified using the Bethesda system, and cell blocks from 34 cases underwent ICC with HBME-1, CK-19, and CD-56. Final diagnoses were established using histopathology where available or a composite reference standard based on stable imaging follow-up over two years. Diagnostic performance was assessed using logistic regression, ROC curves, and confusion matrices.

Results. Bethesda showed sensitivity of 80.8% and specificity of 60.6% (AUC=0.73), while TIRADS showed sensitivity of 89.0% and specificity of 30.3% (AUC=0.65). Among the ICC-tested cases, specificity was 100%; sensitivities were 94.1% for HBME-1, 88.2% for CD-56, and 41.2% for CK-19. The HBME-1⁺/CD-56⁻ pattern was strongly associated with malignancy. The combined Bethesda+ICC model demonstrated the best predictive performance (lowest AIC). Papillary carcinoma was the most frequent malignancy.

Conclusion. In this cohort, ICC using HBME-1, CK-19, and CD-56 showed potential as an adjunct to the Bethesda classification for the preoperative assessment of thyroid nodules. These findings suggest a possible role for ICC in selected cases, although further studies with larger histologically confirmed cohorts are needed.