Lisanne de Koster

50 chapter 2 with an incremental risk of malignancy [17]. Despite the obvious importance of both ultrasound and cytology, the ATA guidelines do not provide recommendations regarding (re-)interpretation of US characteristics after FNAC has resulted in indeterminate cytology. Follicular-type malignancies typically have a different US appearance. More often FTC may be iso- to hyperechoic, with a spherical shape, smooth regular margins and no calcifications [231, 232]. FVPTC may also show FTC-like or benign features rather than the classic suspicious features, although microcalcifications may be distinctive [232-234]. In the past years, Brito et al. and Remonti et al. performed meta-analyses on US assessment of unselected thyroid nodules. Both also briefly discussed its diagnostic value in indeterminate nodules, including a mere limited number of studies and also including cytology suspicious for malignancy. Increased central vascularization was most predictive of malignancy with reported 96% specificity [230]. Yet, in general US seemed less accurate in indeterminate nodules than in unselected thyroid nodules [229, 230]. In the dozens of available original ultrasound studies, individual US features generally demonstrated limited sensitivity in indeterminate thyroid nodules. Only the appearance of a solid thyroid nodule – as opposed to varying degrees of cystic content – had high sensitivity. Ranging between 46% and 100%, multiple studies demonstrated sensitivity above 90% [57, 58, 235-239]. A number of classic suspicious US characteristics, such as a taller-than-wide shape, presence of irregular margins and presence of microcalcifications, demonstrated valid specificity in indeterminate thyroid nodules. Specificities for each of these characteristics ranged from 72% to 99% [239-242], 65% to 100% [240, 243, 244] and 36% to 100%, respectively [245, 246]. Despite the wide range, presence of microcalcifications was more than 90% specific in many studies [235, 236, 238, 240-242, 244, 245, 247-249]. Large nodule size (defined as a diameter larger than 4 cm) was only investigated in a limited number of studies. Reported specificities ranged between 69% and 94% [245, 250]. Other features, such as a solitary nodule, hypoechogeneity and absence of a hypoechoic halo were associated with thyroid malignancy, but less accurately differentiated between benign and malignant indeterminate thyroid nodules [236, 239, 244, 250-255]. Additionally, opposing the results from one of the mentioned meta-analyses, central vascularization also does not seem very accurate in indeterminate thyroid nodules. Specificity ranged from 0% to 100%, although multiple studies demonstrated extremely poor specificity [57, 240, 254-258]. Results regarding two US features are remarkably contradicting. First, the absence of a hypoechoic halo is typically considered suspicious for malignancy, but showed overall poor and very heterogeneous diagnostic potential in indeterminate thyroid nodules [229]. Sensitivity and specificity ranged from 17% to 99% and 0% to 93%, respectively [238, 239, 243, 259, 260]. Presence of a hypoechoic halo is typically considered a benign feature, but has also been associated with follicular types of thyroid carcinoma [261]. Dogan et al. reported 88% specificity for presence of a halo in AUS/FLUS nodules and 78% in FN/SFN nodules [262]. Second, the ultrasonographic nodule shape seems ambiguous. Similar to the unselected population, a typically suspicious taller-than-

RkJQdWJsaXNoZXIy MTk4NDMw