Evaluation of characteristics of thyroid nodules on contrast-enhanced ultrasonography: a retrospective analysis of 252 cases

Abstract

Aim: This study investigated the enhancement patterns observed on contrast-enhanced ultrasound (CEUS) images for differentiating thyroid nodules.Material and methods: A retrospective review was conducted of CEUS cine loops of 252 nodules by two independent readers. Seven categories of enhancement patterns were identified: concentric hypoenhancement; heterogeneous hypoenhancement; hypoenhancement with sharp margin; homogeneous hyper/isoenhancement; hyper/isoenhancement with ring-like vascularity; island-like enhancement; and no perfusion. Associations between these patterns and the confirmed pathological/cytological outcomes (178 malignant, 74 benign) were analyzed and the sensitivity, specificity and positive predictive values (PPVs) determined. The agreement of the readers’ assessments was evaluated by Kappa value.Results: For malignant nodules, the predominant 3 patterns were: concentric hypoenhancement, heterogeneous hypoenhancement and homogeneous hyper/isoenhancement. For each of these, the diagnostic specificity was above 87% and the PPV more than 85%. Combining these patterns for malignancy the rates of sensitivity, specificity and PPV for reader 1 (reader 2) were 96.1% (98.9%), 71.6% (71.6%), and 89.1% (89.3%), respectively. For benign nodules, the predominant 4 patterns were: hypoenhancement with sharp margin; hyper/isoenhancement with ring-like vascularity; island-like enhancement; and no perfusion. The specificity for each was above 98% and the PPV more than 70%. Combining these patterns for benignity, the rates of sensitivity, specificity and PPV for reader 1 (reader 2) were 71.6% (71.6%), 96.1% (98.9%) and 88.3% (96.3%), respectively. The inter-reviewers agreement for classifying enhancement patterns was excellent (κ = 0.84, 95% CI: 0.79-0.89).Conclusions: Enhancement patterns of thyroid nodules on CEUS investigation, enable differentiation between malignant and benign lesions with good diagnostic sensitivity, specificity and PPV.