Validation of a five-tier cytodiagnostic system for thyroid fine needle aspiration biopsies using cytohistologic correlation

Objective: Approximately 70-80% of thyroid fine needle aspiration biopsies (FNAB) can distinguish benign from malignant thyroid nodules. However, much interpretive diagnostic difficulty arises with the remaining 20-30% of cases. These problematic thyroid aspirations have been placed in various diagnostic categories, which collectively have led to confusion and a negative impact on the clinical management of patients with thyroid nodules. We present our experience using a five-tier system, including the diagnostic terminology: benign, indeterminate, suspicious, malignant, and unsatisfactory. Design: Thyroid FNABs were diagnosed using a five-tier system from 200 consecutive patients and the subsequent surgical excisions were correlated. Main outcome: Overall, there was an excellent association between the five diagnostic categories and predicting benign versus neoplastic thyroid nodules (LR = 96.06, chi(2) = 76.49, and phi = 0.618, df = 4, p < 0.0001). A negative cytologic diagnosis carries a negative predictive value of 92%, while an indeterminate, suspicious, and malignant cytologic diagnosis carries a positive predictive value of 50, 71, and 100% respectively. The estimated sensitivity for an indeterminate, suspicious, or malignant cytologic diagnosis varied from 70 to 89%, while the specificity increased from 57 to 92 to 100%, respectively. Conclusions: The data presented shows that the five diagnostic categories of thyroid FNAB are excellent at distinguishing benign from neoplastic thyroid nodules. Both the indeterminate and suspicious categories, while not statistically different from each other in predicting benign from neoplasia, are statistically different from obviously benign and obviously malignant categories. These results support the need for an indeterminate and/or suspicious category.