Ultrasound characteristics of thyroid nodules facilitate interpretation of the malignant risk of Bethesda system III/IV thyroid nodules and inform therapeutic schedule

Abstract Background This study was designed to explore whether ultrasound of thyroid nodules facilitates the interpretation of the malignant risk of Bethesda III/IV thyroid nodules to inform further therapies. Methods We reviewed patient records in which the results of ultrasound‐guided fine‐needle aspiration (US‐FNA) were classified by the Bethesda III/IV in our institution between January 2016 and June 2018. Studies were retrieved from PubMed, Cochrane Central Register of Controlled Trials, ISI Web of Science, Science Direct, Wiley Online Library, EMBASE, China National Knowledge Infrastructure, WanFang, and Chinese VIP. The odds ratio (OR) was used to measure associations between risk factors and thyroid nodule malignancy. Results Fifty‐nine cases of Bethesda III/IV with corresponding surgeries were included, and the malignancy risk was 54.2%. Meta‐analysis revealed irregular borders, solitary nodules, hypoechogenicity, microcalcifications, and being taller than wide, all of which increased the malignancy risk of thyroid nodules. Combined ORs for these factors were 4.08 (95% CI: 2.34‐7.14, P < .001), 2.18 (95% CI: 1.39‐3.42, P = .001), 2.02 (95% CI: 1.35‐3.01, P = .001), 3.21 (95% CI: 2.26‐4.56, P < .001), and 4.35 (95% CI: 3.07‐6.15, P < .001), respectively. Conclusion As the risk of malignancy for papillary thyroid carcinoma (PTC) is high, when any one of the five ultrasound features of malignancy were confirmed, repeated FNA is recommended to confirm PTC‐type malignancy, even though nodules were Bethesda III/IV classification. However, repeated FNA should be avoided when none of these ultrasound features are identified because repeated FNA does not contribute to identifying non‐PTC type malignancies, such as follicular thyroid carcinoma and poorly differentiated thyroid carcinoma.


| INTRODUCTION
Thyroid nodules are clinically common; however, the occurrence of malignancy in thyroid nodules remains relatively low, ranging between 5% and 7%. 1,2 Ultrasound-guided fine-needle aspiration (US-FNA) is a useful test for the evaluation of thyroid nodules that has been widely applied as the primary diagnostic procedure. 3  unclear. The risk of malignancy for Bethesda III is estimated at 5%-15%, 2 and 10%-40% of Bethesda IV nodules are malignant. 3 Due to the large range of malignancy risk, repeated FNA or surgery is recommended, and more than half have a determined diagnostic result. [5][6][7] Approximately 15.6%-48.6% of nodules remain AUS/FLUS upon repeat FNA, 5,6,8 indicating a need to assess the risk of malignancy in Bethesda system III/IV nodules.
Ultrasound is the preferred method to detect thyroid nodules, and ultrasound factors, such as hypoechogenicity, irregular shape, being taller than wide and microcalcifications, are considered to be features of malignant nodules. Some studies have demonstrated the use of ultrasound to predict the risk of malignancy in Bethesda III and IV thyroid nodules during the initial US-FNA. [9][10][11][12] However, one study demonstrated that the thyroid ultrasound risk stratification system has poor specificity and accuracy for classifying indeterminate lesions. 13 Another study reported that age and gender, but not ultrasound characteristics, influence the decision to perform surgery in AUS/FLUS patients. 14  Chinese VIP from inception to August 2018 using the keywords: "thyroid," "Bethesda system," "The Bethesda System for Reporting Thyroid Cytology "TBSRTC," "fine-needle aspiration," "FNA," and "FNAC." Inclusion criteria were studies using ultrasound to estimate the malignancy of Bethesda III/IV nodules in the initial US-FNA.
Reviews and references were also evaluated. Reviews, abstracts, letters, and duplicate data were removed. No language restrictions were applied.
Two authors (Fu Li and Denghua Pan) independently performed the literature search and data screening. When there was controversy, the disagreement was resolved by a third reviewer (Yun He).

| Data extraction
Two investigators (Fu Li and Denghua Pan) independently extracted data from eligible studies. The following characteristics were extracted: first author's name, year of publication, country, and number of patients. To evaluate risk factors associated with thyroid cancer, the following ultrasound parameters were extracted: irregular borders, solitary nodule, hypoechogenicity, microcalcifications, and being taller than wide.

| Statistical analysis
Two independent t-tests and chi-square test were applied to analyze differences between benign and malignant groups using SPSS22.0 (SPSS Inc., Chicago, Illinois). For meta-analysis, the odds ratio (OR) was used to explore associations between risk factors and thyroid nodule malignancy. OR > 1 indicates that risk factors were more likely to correlate with the malignancy of thyroid nodules. For heterogeneity analysis, chi-squared test was used to assess heterogeneity among studies. P < .05 was considered to represent significant heterogeneity, and then a fixed effect model was used. Otherwise, a random effects model was used. Publication bias was measured using funnel plots. Meta-analysis was conducted in STATA12.0 (STATA Corp., College, Texas). P < .05 was considered statistically significant. No significant difference was observed with respect to age between benign and malignant groups (P > .05). In the benign group, average nodule size was 2.51 cm, and in the malignant group, mean nodule size was 1.44 cm, which were significantly different in two groups (P = .019). Additionally, there was a significant difference between benign and malignant groups with respect to patient sex (P = .013) in thyroid nodules. In the malignant groups of thyroid nodules, the parameters of irregular borders, solitary nodules, hypoechogenicity, microcalcifications and being taller than wide were more common (all P < .05). We also compared clinical ultrasound and ultrasonographic characteristics between PTC and benign thyroid nodules ( Table 2). The mean age of PTC cases was 41.27 years old, and the average age of the benign group was 46.41 years old, with no significant difference between the two group (P = .119). Average size in the benign group was larger than in the PTC group (P = .023). In addition, there were significant differences in sex, irregular borders, solitary nodules, hypoechogenicity, microcalcifications, and being taller than wide between PTC and benign groups (all P < .05). Figure 1 shows the procedures of this study, which screened 368 studies. After reading full texts of the studies, 16 publications were included in the meta-analysis. 9,[15][16][17][18][19][20][21][22][23][24][25][26][27][28][29] Table 3 shows the characteristics included studies. All studies had sufficient data to calculate the associations between risk factors and nodules. This meta-analysis observed significant associations between irregular borders, solitary nodules, hypoechogenicity, and microcalcifications and nodules that are being taller than wide or malignant. Combined ORs were 4.08 (95% CI: 2.34-7.14, P < .001; Figure 2A Table 4).

| Publication bias
Begg's test was performed to measure publication bias of this metaanalysis. Publication bias was found in groups with irregular borders  hypoechogenicity ( Figure 3D, P = .055), and those that are being taller than wide ( Figure 3E, P = .173).

| Discussion
Ultrasound is the preferred method to detect thyroid nodules, and  Figure 4A shows that a solid thyroid nodule was located in the left lobe of the thyroid. The nodule was a hypoechoic nodule with irregular border and microcalcifications (TI-RADS 4c). FNA was conducted, and results revealed cytological atypia: focal nuclear changes with extensive but mild nuclear changes ( Figure 4B, Figure 4C). Therefore, we diagnosed this thyroid nodule as AUS/FLUS.
Finally, the patient decided to undergo surgery to resect this nodule, and the histopathology was PTC ( Figure 4D). Figure 5 was a case of an FN/SFN thyroid nodule. Figure 5A shows a solid thyroid nodule located in the right lobe of the thyroid. The nodule was hypoechoic with irregular border and macrocalcifications (TI-RADS 4b).
FNA results showed mild nuclear changes (increased nuclear size, nuclear contour irregularity) without true papillae and intranuclear pseudoinclusions ( Figure 5B,C). We diagnosed this thyroid nodule as FN/SFN and recommended surgery for this patient. Histopathology of this thyroid nodule was FTC ( Figure 5D). In this study, we unveiled significant associations between solitary nodules, hypoechogenicity, irregular borders, being taller than wide, microcalcifications, and malignant nodules.  F I G U R E 4 A case with AUS/FLU thyroid nodule. A, Imaging of the thyroid nodule. B,C, Results of FNA. D, Histopathology results. Figure 4A shows a solid thyroid nodule located in the left lobe of the thyroid. The nodule was hypoechoic with irregular border and microcalcifications (TI-RADS 4c). Figure 4B,C show cytological atypia: focal nuclear changes and extensive but mild nuclear changes (×400). Hence, we diagnosed this thyroid nodule as atypia of undetermined significance or follicular lesion of undetermined significance (AUS/FLUS). Figure 4D shows histopathology of PTC for this thyroid nodule (×400  Figure 5A shows a solid thyroid nodule located in the right lobe of the thyroid. The nodule was hypoechoic with irregular border and macrocalcifications (TI-RADS 4b). Figure 5B,C show mild nuclear changes (increased nuclear size, nuclear contour irregularity) without true papillae and intranuclear pseudoinclusions (×400). Hence, we diagnosed this thyroid nodule as follicular neoplasm or suspicious for a follicular neoplasm (FN/SFN). Figure 5D shows the histopathology was FTC for this thyroid nodule (×40) [Color figure can be viewed at wileyonlinelibrary.com]