Ultrasound characterization for thyroid nodules with indeterminate cytology: inter-observer agreement and impact of combining pattern-based and scoring-based classifications in risk stratification.

ABSTRACT

BACKGROUND: The American Thyroid Association (ATA) sonographic patterns stratify the risk of malignancy of cytologically indeterminate thyroid nodules (ITNs). This study aimed to (1) assess inter-observer agreement for sonographic features and patterns; (2) identify potential sources of disagreement; and (3) evaluate whether the number of suspicious features risk-stratifies non-ATA and high-suspicion patterns. METHODS: Three observers independently reviewed the ultrasound images of 463 ITNs with histological follow-up consecutively evaluated between October 2008 and June 2015 at an academic cancer center. Each observer evaluated individual sonographic features. ATA sonographic patterns were derived from the interpretation of sonographic features. Nodules not fitting into any of the proposed patterns were clustered into a non-ATA pattern. RESULTS: The inter-observer agreement for ATA sonographic patterns and echogenicity was fair, moderate for margins, good for composition and echogenic foci, and very good for extrathyroidal extension and lymph node metastasis. The interpretation of each sonographic feature was significantly different between observers, and there was complete disagreement in at least one of the features in 104 (22%) nodules. A total of 169 nodules (37%) were classified into the non-ATA pattern. The number of suspicious features allowed risk stratifying nodules with non-ATA and high-suspicion sonographic patterns. Most Non-invasive Follicular Thyroid Neoplasms with Papillary-like Nuclear Features had 0-1 suspicious features and none had >2. CONCLUSIONS: Echogenicity interpretation was the greatest source of disagreement. The number of suspicious features risk-stratifies ITNs with non-ATA or high-suspicion patterns. Future studies attempting to objectivize the interpretation of echogenicity and heterogeneity are needed.