231 Diagnostic utility of molecular and imaging biomarkers 2 Ultrasound assessment of Bethesda IV nodules Sixteen studies reported on 1,845 Bethesda IV nodules [232, 238, 240, 241, 246, 249, 250, 261, 262, 272, 280, 306, 308, 356-358]. The average malignancy rate in this group was 26.9% (496/1,845), including 196 PTC (39.5%), 123 FVPTC (24.8%), 134 FTC (27.0%), 26 FTC-OV (5.2%), four MTC (0.8%), 13 other thyroid malignancies (2.6%) and no unreported subtypes. In Bethesda IV nodules, absence of a halo is the only moderately sensitive feature (79%, 95% CI: 69%-90%). Nonetheless, an excellent specificity is found for a taller-than-wide shape (95.1%, 95% CI: 81.9%-98.8%), irregular margin (95.0%, 95% CI: 87.4%-98.1%) and macrocalcifications (95%, 95% CI: 92%-99%), followed by very good specificities for an increased intranodular vascularization (90.0% (95% CI: 73.9%-96.6%)) and microcalcifications (90.4% (95% CI: 79.5%-95.8%)) (Figure 66, Figure 67 and Table 53). A downside to these highly specific characteristics is that they occurred significantly less frequently in the Bethesda IV than Bethesda III population: 14.0% and 25.3% had taller-than-wide shape (Pearson chi-squared2, p<0.001), 11.3% and 39.5% had irregular margins (p<0.001), 21.4% and 26.9% had increased intranodular vascularization (p=0.02) and 16.8% and 22.5% had microcalcifications (p<0.001), respectively. The difference was nonsignificant for macrocalcifications, 18.0% versus 24.0% (p=0.08). TIRADS Diagnostic ultrasound scores or step-by-step algorithms aid the classification of US patterns and consequent risk of malignancy [238, 266, 273]. Best-known is the Thyroid Imaging Reporting and Data System (TIRADS), a classification to risk-stratify thyroid nodules, designed by Horvath et al. and modified by Kwak et al. following the example of the similar BIRADS classification for breast lesions [274, 275]. The TIRADS assigns nodules to a risk category based on five suspicious US features: solid appearance, (marked) hypoechogenicity, irregular margins, microcalcifications and a taller-than-wide shape. Nodules without any of these features are likely benign and categorized as TIRADS 3. Their risk of malignancy is ~1.7% in an unselected population. TIRADS 4 includes suspicious nodules, which are further classified according to an increasing malignancy risk into 4a (one suspicious US feature), 4b (two suspicious features) and 4c (three or four suspicious features). Nodules with all five suspicious US features are classified as TIRADS 5 and are associated with a high 87.5% risk of cancer in an unselected population [274]. Six studies validated the TIRADS in indeterminate thyroid nodules [240, 273, 276-279]. We performed a meta-analysis for the performance of the TIRADS, including 582 indeterminate thyroid nodules from five of these studies: 314 Bethesda III (54%) and 268 Bethesda IV nodules (46%). One study provided no summarized patient data and was disregarded [273]. Only 27 of the 582 indeterminate lesions included in all studies were TIRADS 5, 21 of which were malignant (78%). If lesions with TIRADS 4a or higher were considered suspicious for malignancy, sensitivity and specificity ranged from 61%-100% and 0%-65%, respectively. The estimated pooled sensitivity and specificity are 93% (95% CI: 68%-99%) and 13% (95% CI: 2%-52%), respectively. The AUC is 0.66 (95% CI: 0.62-0.70). Sensitivity is lower in Bethesda III than in Bethesda IV nodules: 81% (95% CI: 57%-93%) versus 95% (95% CI: 85%-100%) (Figure 69-74).
RkJQdWJsaXNoZXIy MTk4NDMw