The Value of Negative Diagnosis in Thyroid Fine-Needle Aspiration: a Retrospective Study with Histologic Follow-Up.

The Bethesda System for reporting thyroid cytopathology (BSRTC) predicts an incidence of malignancy of less than 5% in thyroid nodules with a benign diagnosis on fine-needle aspiration (FNA). However, recent series have suggested that the true rate of malignancy might be significantly higher in this category of patients. We reviewed our experience by performing a retrospective analysis of patients with benign thyroid FNA results who underwent thyroidectomy between 2008 and 2013 at a large academic center. Information including demographics, ultrasound features, FNA diagnosis, and surgical follow-up information were recorded. Slides were reviewed on cytology-histology discrepant cases, and it was determined whether the discrepancy was due to sampling or interpretation error. A total of 802 FNA cases with a benign diagnosis and surgical follow-up were identified. FNA diagnoses included 738 cases of benign goiter and 64 cases of lymphocytic thyroiditis. On subsequent surgical resection, 144 cases were found to be neoplastic, including 117 malignant cases. False negative, defined as interpretation error and inadequate biopsy of the nodule harboring malignancy, was 6%. When cases of noninvasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP) were excluded from the analysis, false-negative rate was 5%. When microPTC cases were excluded, false-negative rate was 3% and was slightly less than 3% when both microPTC and NIFTP cases were excluded from the analysis. Retrospective review of neoplastic cases showed that 57% were due to sampling error and 43% were due to interpretation error. Interpretation error was more likely to occur in follicular patterned neoplasms (75%), while sampling error was more common in non-follicular variants of papillary thyroid carcinoma (non-FVPTC) (61%). With the exclusion of microPTC, interpretation errors were still more likely to occur in follicular neoplasms (79%) but there was no significant difference in sampling error between non-FVPTC (37%) and follicular patterned neoplasms (42%). Tumor size was larger in cases with interpretation error (mean = 2.3 cm) compared to cases with sampling error (mean = 1.4 cm). This study shows that the false-negative rate of thyroid FNA at our institution is not significantly above the rate suggested by the BSRTC. Interpretation errors were more likely to occur in follicular patterned neoplasms, while non-FVPTC was more frequently found in false negative cases due to inadequate sampling.