A retrospective study correlating sonographic features of thyroid nodules with fine-needle aspiration cytology in a South African setting.

BACKGROUND

Thyroid nodules are prevalent worldwide. Detection rates are increasing because of the use of ultrasonography. Ultrasound has become the first-choice imaging modality in evaluating nodules. The decision to perform an US-guided fine-needle aspiration (FNA) is based on a nodule's sonographic features. Thus, it is essential to accurately risk stratify thyroid nodules so that they are appropriately referred for FNA.

OBJECTIVES

The aim of this study was to correlate the ultrasound imaging features of thyroid nodules with FNA cytology and surgical histopathology results, and to risk stratify patients using the American Thyroid Association (ATA) classification for each imaging characteristic with the likelihood of the nodule being malignant.

METHOD

Retrospective analysis of a thyroid ultrasound database at Chris Hani Baragwanath Academic Hospital, over the period 2015-2017. Frequencies and percentages were used to summarise the data. Univariate logistic regression analyses were used to assess the accuracy of sonographic features in predicting the histologically determined diagnosis for thyroid tumours.

RESULTS

A total of 113 nodules underwent FNA, of which 104 were diagnostic. The best three ultrasound features that pose a higher risk for malignancy are absent halo, presence of microcalcifications and hypoechoic appearance. No single nodule feature is an absolute indicator for malignancy. There is a high agreement between ATA classification and cytopathology or histology when nodule features are grouped into clusters. Agreement between the ATA classification and cytopathology/histology was 86.7% with a kappa of 0.714. The agreement between the cytopathology FNA results and lobectomy histopathology was 98.8% with a kappa of 0.973.

CONCLUSION

This study contributes to the paucity of data available for sub-Saharan Africa and provides reassurance that our results are consistent with international studies. The study confirms that the usage of a thyroid nodule classification system improves characterisation and increases accuracy in detecting thyroid malignancies, thus sparing many patients the morbidity of unnecessary thyroid surgery.