Anatomic Pathology / INTEROBSERVER VARIABILITY IN THYROID FNA INTERPRETATION Interobserver Variability in Thyroid Fine-Needle Aspiration Interpretation of Lesions Showing Predominantly Colloid and Follicular Groups

We studied interobserver variability in the assessment of thyroid fine-needle aspiration (FNA). We limited our cases to those showing predominantly colloid and follicular cell groups. Twenty cases of thyroid FNA diagnosed by 1 experienced cytopathologist were reviewed by 4 other cytopathologists who made their own diagnoses while unaware of the original diagnoses. Two cytopathologists then assessed the cytologic features of the 20 cases. Interobserver variability was calculated for noncollapsed and collapsed diagnoses. Diagnoses and observer agreement were compared with cytologic features. There was little correlation among observers regarding the diagnosis of follicular “lesion” vs “neoplasm.” Interobserver variability was somewhat poor before data were collapsed (κ = 0.35) but was relatively good when data were collapsed to treatment recommendations (κ = 0.65). Cellularity, cyst change, and amount of colloid correlated with treatment recommendations; no specific features correlated with poor performance. Thyroid FNA shows good interobserver agreement in the diagnoses of lesions showing predominantly colloid or follicular cells (when collapsed). We speculate that interobserver variability is poor in some cases owing to difficulty assessing thin colloid, some lack of agreement regarding criteria for adequacy, and a possible “gray zone” that might exist with lesions showing colloid and abundant follicular cells. Fine-needle aspiration (FNA) has been shown to be an effective method for the preliminary interpretation and triaging of thyroid lesions.1-9 Recently, it has become clear that even the “gold standard” for diagnosing follicular-patterned thyroid lesions, surgical pathology, shows considerable interobserver variability.10-12 Also, it has been shown that thyroid pathology contributes significantly to second-opinion discrepancies for surgical pathology and cytopathology specimens.13-16 Because of the “gray zone” that exists in the interpretation of cytologic samples of thyroid lesions composed predominantly of colloid and follicular groups, we speculated that considerable interobserver variability might exist for the interpretation of these cases.17-21 Colloid nodule (CN) is the most common diagnosis made with thyroid FNA, and it represents, by far, the most common thyroid nodule.1,6,7,9,22,23 The differential diagnosis for CN most often includes other follicular-patterned thyroid lesions, generally diagnosed as follicular “lesions” (FLs) or follicular “neoplasms” (FNs) by FNA. Because thyroid pathology is vast, we limited our study to this common diagnostic dilemma. Therefore, we reviewed interobserver variability with thyroid FNA of specimens that showed predominantly colloid and follicular groups. Materials and Methods The cytology files of 1 cytopathologist who reviews more than 600 thyroid FNA specimens per year were reviewed. All slides from 10 consecutively diagnosed CNs and 10 consecutively diagnosed FLs or FNs were retrieved. Each case was placed into a separate folder and was accompanied with a