Notification of abstract acceptance
announced via E-mail on July 31st.
Abstract view



Oct. 18 (Thu)  Grand Ballroom 101 Room - SS01 TH
09:40-09:50 [SS 01 TH-01] 
The US-FNA system for malignancy risk in thyroid nodules: 5-Tier classification
   
Speaker Hyobin Seo (Seoul national university hospital Healthcare system Gangnam center)
Authors Hyobin Seo,Dong Gyu Na2,Ji Hoon Kim3
Affiliation Seoul national university hospital Healthcare system Gangnam center1,Human Medical Imaging and Intervention Center2,Seoul National University Hospital3
PDF View  SS 01 TH-01.pdf

PDF File View :
PURPOSE:
To evaluate diagnostic performance of the US-FNA (UF) system for thyroid malignancy by comparing with the Bethesda (BE) system of thyroid cytopathology.

MATERIALS AND METHODS:
US-guided FNA was performed at 2049 thyroid nodules of 1210 consecutive patients. Among these nodules, a total of 704 nodules with final diagnoses (214 malignant and 490 benign nodules) were included for this study. The malignancy risk of thyroid nodules were classified into 4-tiers by US features. The UF system was stratified into 5-tiers of malignancy risk by combination of US-based 4-tier classification and 6-tier FNA results of BE system. The 5-tier malignancy risk of UF system was categorized as probably benign (BE category 1 with US-probable benign, low or intermediate risk nodule; all nodules with BE category 2), low risk (BE category 3 with US-probably benign or low risk nodules and category 4 with US-probably benign), intermediate risk (BE category 1 with US-high risk, BE category 3 with US-intermediate risk nodules, and BE category 4 with low risk nodules), high risk (BE category 3 with US-high risk nodules and BE category 4 with US-intermediate or high risk nodules), and probably malignancy (all nodules with BE category 5 or 6). The diagnostic performance of each system for predicting malignancy was calculated with receiver operating characteristics (ROC) analysis.

RESULTS:
The thyroid nodules were classified into 4 tiers by US-based stratification of malignancy risk; probably benign (risk<3%, n=101), low risk (≥3% and <15%, n=149), intermediate risk (≥15 and <50%, n=211), and high risk (≥50%, n=243). The FNA results were BE category 1 (n=35), category 2 (n=404), category 3 (n=89), category 4 (n=9), category 5 (n=92), and category 6 (n=75). In nodules with BE category 1, malignancy were found only in US-high risk nodules, and malignancy risk gradually increased according to US-based malignancy risk in nodules with BE category 3. The malignancy risk of nodules was classified by the UF system as probably benign (n=418, 1.4%), low risk (n=32, 12.5%), intermediate risk (n=52, 40.4%), high risk (n=35, 48.6%), and probably malignancy (n=167, 99.4%). In overall nodules, the area under the ROC curve of UF system (Az=0.972) was greater than that of the Bethesda system (Az=0.948) (P=0.08).

CONCLUSION:
The UF system may be more accurate and effective for predicting malignancy compared with the FNA BE system alone.


ⓒ Korean Society of Radiology. All Rights Reserved.