Accuracy of thyroid fine-needle aspiration using receiver operator characteristic curves

Area of Science:

• Endocrinology
• Pathology
• Oncology

Background:

• Thyroid fine-needle aspiration (FNA) is a key diagnostic tool for thyroid nodules.
• Previous accuracy assessments of thyroid FNA have been limited by suboptimal statistical methods.
• Standardized methods are needed to accurately evaluate thyroid FNA performance.

Purpose of the Study:

• To determine the accuracy of thyroid fine-needle aspiration (FNA) using receiver operator characteristic (ROC) curves.
• To evaluate the impact of repeat aspirations on diagnostic accuracy.
• To assess the utility of subcategorizing nondiagnostic and malignant thyroid FNA cases.

Main Methods:

• Correlation of 1,085 thyroid fine-needle aspirates with histologic/cytologic follow-up over 4 years.
• Application of receiver operator characteristic (ROC) curves to calculate accuracy, including nondiagnostic and indeterminate cases.
• Analysis of accuracy changes with repeat aspirations and subcategorization of specific diagnoses.

Main Results:

• Overall accuracy for a single thyroid FNA session was 0.90 +/- 0.02.
• Nondiagnostic aspirates carried a 16% risk of malignancy.
• Repeat aspiration improved overall accuracy, with 70% of patients achieving adequate negative results; subcategorization did not significantly improve accuracy.

Conclusions:

• Receiver operator characteristic (ROC) curves provide a robust method for assessing thyroid FNA accuracy.
• Repeat aspiration of nondiagnostic thyroid nodules significantly enhances diagnostic accuracy.
• Subcategorization of thyroid FNA results, including papillary carcinoma subtypes, does not improve overall test accuracy.