Hürthle cell-predominant thyroid fine needle aspiration cytology: A four risk-factor model highly accurate in excluding malignancy and predicting neoplasm.

BACKGROUND:Interpretation of Hürthle cell-predominant cytologies (HCP) is very challenging as a majority is diagnosed as indeterminate. Prior studies have reported various cytologic features to help distinguish non-neoplastic (NN) from neoplastic and malignant lesions but had contradicting results. Our aim was to identify risk factors predictive of neoplasm and/or malignancy by correlating cytologic features with clinical and ultrasound findings. METHODS:Sixty-nine HCP cases with surgical follow-up were identified, including 35 NN, 20 adenomas, and 14 carcinomas. Ultrasound data were recorded utilizing Thyroid Imaging Reporting and Data System (TI-RADS) and American Thyroid Association (ATA) scoring systems. Sixteen cytologic criteria were evaluated and semi-quantitatively scored. Data were assessed by univariable, multivariable and stepwise logistic regression analysis; and statistical significance achieved at P-value <0.05. RESULTS:On univariable analysis, significant predictors of neoplasm were high cellularity, isolated single cells, absent colloid, non-uniform HC population (anisonucleosis), larger nodule size, and higher ATA score. Large-cell dysplasia and transgressing blood vessels were not found to be significant factors. Multivariable analysis identified a combination of four risk factors (high cellularity, anisonucleosis, absent colloid, and size ≥2.9 cm) that was associated with neoplasm in 10/11 patients. None of 15 patients with zero or 1 out of 4 risk factors had malignancy or neoplasm on follow-up. This model also significantly outperformed ATA and TI-RADS scoring systems. CONCLUSION:In the absence of four or three risk factors, the model excluded malignancy and neoplasm in all patients. The presence of all four factors predicted neoplasm and malignancy in 91% and 46% of cases, respectively.