A Simplified Ultrasonographic Score For The Prediction Of Cytologically Suspicious Thyroid Nodules

Purpose The major aim of ultrasound (US)-based risk stratification systems is to reduce unnecessary thyroid biopsies without losing the ability to recognize nodules with clinically significant malignancy. Each of the classic suspicious features of a thyroid nodule detected on US scan (hypoechoic pattern, microcalcifications, irregular margin, taller than wide shape, irregular vascularization) is significantly independently associated with the probability of malignancy, but none of them has good diagnostic accuracy. Thus, we evaluated the predictive value of a binary score simply based on the combination of these US features, regardless of the specific predictive value of each US feature, against the outcome of suspected malignancy at cytological diagnosis (TIR3 to TIR5 categories by SIAPEC-IAP [TIR+]).Materials and Methods 1009 thyroid nodules from 1081 patients were considered. The US features of suspicion of all nodules were categorized in 5 binary scores (U1 to U5), each including from 1 to 5 of those features.Results U2 ( at least 2 US suspicious features) was the most balanced predictor of TIR+ (PPV 0.48, NPV 0.93, LR+ 3.05 and LR- 0.24). Weighting the predictivity of the single features did not improve the estimate. Using U2 as the criterion to send nodules to FNAC would have reduced the number of biopsies by 60 % (604 patients) and the false negatives would have only accounted for 41 cases out of 237 TIR+ (17 %) with 39 cases of TIR3 and 2 cases of TIR4, including only 6 malignant nodules on histological examination. U2 performed much better than the ATA recommendations for detecting those nodules, resulting in TIR+ at cytology.Conclusion This simple and reproducible sonographic score based on 2 US features of suspicion of malignancy has quite a good performance with respect to identifying thyroid lesions categorized by cytology as medium-high risk of malignancy and could allow us to reduce cytology costs for low-risk nodules.