*3.1. ACR vs. EU-TIRADS*

The comparison between the ACR and EU systems in classifying the 480 nodules is shown in Table 1.


**Table 1.** Comparison between the ACR and EU-TIRADS systems in the indication to perform FNA.

Considering the nodules requiring FNA as per the ACR (*n* = 223, 46.5%) and EU-TIRADS (*n* = 249, 51.9%) criteria, 86 (38.6%) and 90 (36.1%) were deemed to be ≥TIR3A/III after cytological assessment. Overall, a good agreement between the two systems was noted, reaching an accordance for FNA indication in 87.1% (418/480) of the nodules. The major discrepancies on the execution of cytology were observed in the setting of EU-TIRADS class 3, with 26 cases that would not be submitted to biopsy as per ACR, either for different US scores (*n* = 22) or size cut-off (*n* = 4). Similarly, 12 cases that received FNA following the ACR-TIRADS criteria did not reach the indication in the EU-TIRADS system, either for different US scores (*n* = 1) or size cut-offs (*n* = 11). Finally, 11.1% (6/54) of the cases labeled as class 5 in both the systems would be biopsied only following ACR due to the presence of a size "grey-zone" of exactly 1 cm in the classifications [3,5]. The assessment of the performances of these systems in the final FNA indication showed a significantly higher specificity for the ACR as compared to the EU-TIRADS (59.0% vs. 52.4%, *p* = 0.0012), with a similar sensitivity (58.9% vs. 61.6%, *p* = 0.3173), PPV (38.6% vs. 36.1%, *p* = 0.1116), and NPV (76.7% vs. 75.8%, *p* = 0.5288) (Tables 2 and 3). The difference noted in terms of the specificity was mainly due to the discordances of the two systems in terms of the size cutoff and the US feature "echogenicity".

**Table 2.** ACR-TIRADS vs. SIAPEC/Bethesda system cytological classifications. False negative cases are in bold. False positive cases are in italics.


**Table 3.** EU-TIRADS vs. SIAPEC/Bethesda system cytological classifications. False negative cases are in bold. False positive cases are in italics.


Histological evaluation was performed on 10.2% (49/480) of the nodules, 81.6% (40/49) of which had an FNA result ≥ TIR3B/IV. A total of 34 out of the 49 (69.3%) nodules

showed malignancy (27 PTC, 3 EFVPTC, 1 Hurthle cell carcinoma, 1 medullary carcinoma, 1 anaplastic carcinoma, and 1 metastasis of the melanoma). The ACR-TIRADS confirmed a significantly higher specificity (57.2% vs. 51.2%, *p* = 0.0019), with a similar sensitivity (67.6% vs. 70.6%, *p* = 0.6547), PPV (12.8% vs. 11.8%, *p* = 0.4030), and NPV (95% vs. 94.9%, *p* = 0.9435) in the diagnosis of malignancy (Table 4). As reported for the analysis of FNA indication, the difference in terms of the specificity between the two systems was mainly driven by the size cutoff and US feature "echogenicity".

**Table 4.** Diagnostic performances of TIRADS systems, cytology, and combination of US features and cytology in the diagnosis of malignant nodules (*n* = 401).


The best performances are in bold, 95% confidence intervals in brackets. TP: true positive; FP: false negative; TN: true negative; FN: false negative; PPV: positive predictive value; NPV: negative predictive value; EF/M: echogenic foci/margins.
