The diagnostic value of ultrasound on different-sized thyroid nodules based on ACR TI-RADS

The thyroid nodule is one of the most common endocrine system diseases. Risk classification models based on ultrasonic features have been created by multiple professional societies, including the American College of Radiology (ACR), which published the Thyroid Imaging Reporting and Data System (TI-RADS) in 2017. The effect of the size in the diagnostic value of ultrasound remains not well defined. The purposes of our study aims to explore diagnostic value of the ACR TI-RADS on different-sized thyroid nodules. A total of 1183 thyroid nodules were selected from 952 patients with thyroid nodules confirmed by surgical pathology from January 2021 to October 2022. Based on the maximum diameters of the nodules, they were stratified into groups A ( ≤ 10 mm), B ( > 10 mm, < 20 mm) and C ( ≥ 20 mm). The ultrasonic features of the thyroid nodules in each group were evaluated and scored based on ACR TI-RADS, and the receiver operating characteristic curve (ROC) was plotted to determine the optimal cut-off value for the ACR TI-RADS scores and categories in each group. Finally, the diagnostic efficacy of ACR TI-RADS on different-sized thyroid nodules was analyzed. Among the 1183 thyroid nodules, 340 were benign, 10 were low-risk and 833 were malignant. For the convenience of statistical analysis, low-risk thyroid nodules were classified as malignant in this study. The ACR TI-RADS scores and categorical levels of malignant thyroid nodules in each group were higher than those of benign ones (p < 0.05). The areas under the ROCs (AUCs) plotted based on scores were 0.741, 0.907, and 0.904 respectively in the three groups, and the corresponding optimal cut-off values were > 6 points, > 5 points and > 4 points respectively. While the AUCs of the ACR TI-RADS categories were 0.668, 0.855, and 0.887 respectively in each group, with the optimal cut-off values were all > TR4. Besides, for thyroid nodules of larger sizes, ACR TI-RADS exhibited weaker sensitivity with lower positive prediction value (PPV), but the specificity and negative prediction value (NPV) were both higher, presenting with statistically significant differences (p < 0.05). For thyroid nodules of different sizes, the diagnostic efficacy of ACR TI-RADS varies as well. The system shows better diagnostic efficacy on thyroid nodules of > 10 mm than on those ≤ 10 mm. Considering the favorable prognosis of thyroid microcarcinoma and the low diagnostic efficacy of ACR TI-RADS on it, the scoring and classification of thyroid micro-nodules can be left out in appropriate cases, so as to avoid the over-diagnosis and over-treatment of thyroid microcarcinoma to a certain extent.


Introduction
The thyroid nodule is one of the most common endocrine system diseases.Studies have shown that about 70% of the people worldwide are suffering from thyroid nodules, most of which are benign while 5-10% are malignant [1][2][3].Although the diagnosis and treatment methods of thyroid microcarcinoma are still controversial, the early detection of thyroid cancer and appropriate diagnosis and treatment measures are still associated with the improved prognosis of patients with thyroid tumors.Therefore, early diagnosis can substantially affect the morbidity and mortality of the disease.
Ultrasound is easy and quick to perform with high resolution.As its value in the detection and evaluation of thyroid nodules has gradually become clinically recognized, it is now a preferred screening tool of thyroid nodules.However, ultrasonographers usually interpret the images subjectively, leading to the constant inconsistencies between reported terminology and treatment recommendations.On this account, all varieties of thyroid imaging reporting and data systems (TI-RADSs) have been created, the most influential of which is the one published by the American College of Radiology (ACR) in 2007.In this system, the thyroid nodules are classified into TR1-5 based on the composition, echogenicity, morphology, margin, hyperechoic foci of the nodules.The higher categorical level is associated with the greater risk of being malignant [4].It has been demonstrated that ACR TI-RADS boasts high diagnostic efficacy and is widely applied worldwide [5].
The ultrasonic features of different-sized thyroid nodules also vary.The suspicious ultrasonic features of benign and malignant thyroid nodules are more likely to overlap among smaller nodules, while for larger ones, the features of benign and malignant nodules are more distinguishable [6].Therefore, the size of thyroid nodules can also affect the diagnostic efficacy of ultrasound in a certain degree, thus influencing the clinical decisions made by physicians.
The current study aims to analyze the ultrasonic features of different-sized thyroid nodules, and to investigate the diagnostic value of ACR TI-RADS on the thyroid nodules of various sizes.

Subjects
All procedures performed in this study involving human participants were in accordance with the Declaration of Helsinki (as revised in 2013).The study was approved by the Ethics Committee of Affiliated Hospital of Jiangnan University and written informed consent was obtained from the patients.A total of 1183 thyroid nodules were selected from 952 patients (male, n = 231; female, n = 741; age, 15-80 years, mean age, 49.07 ± 12.20 years) with thyroid nodules confirmed by surgical pathology in Affiliated Hospital of Jiangnan University and Zhejiang Cancer Hospital from January 2021 to October 2022.The maximum diameters of the nodules ranged from 2 mm to 86 mm, with the mean diameter being 11.65 ± 11.12 mm.Principle surgical indications: (1) the nodules were confirmed or suspected to be thyroid cancer through fine-needle thyroid aspiration; (2) no definite pathological diagnosis was made, but the nodules were > 10 mm and accompanied by bleeding or highly suspicious on ultrasound; (3) the nodules were ≤ 10 mm and the fine-needle aspiration yielded inconclusive results, but the categorical levels of them were ≥ TR4 under ACR TI-RADS, and the patients were too anxious to accept follow-up observation, requiring surgical treatment; (4) the nodules are large in size, compressing the surrounding tissues.Inclusion criteria: (1) patients with initially detected thyroid nodules; (2) patients whose thyroid nodules were confirmed by surgical pathology; (3) patients with complete and reliable ultrasonic and clinical data before aspiration/surgery.Exclusion criteria: (1) patients who underwent aspiration for definitive diagnosis or treatment in other hospitals, which might accept the ultrasonic features of the thyroid nodules; (2) no pathological confirmation or only fine-needle aspiration confirmation; (3) patients with diffuse lesions but with no obvious nodules on ultrasound.

Instruments and methods
The mindray R7 EXP color doppler ultrasound with a linear array probe (frequency, 7.8-15 Mhz) was adopted.During the examination, the patients were in a supine position with both sides of the neck fully exposed.Both the cross and longitudinal sections of the thyroid were scanned, and the nodules were observed from multiple sections and angles.The nodules were then evaluated with the classification methods set by ACR TI-RADS, which took into account the factors including the site, location, margin, acoustic halo, structure, echogenicity, hyperechoic foci, posterior echo, and size.Based on the maximum diameters, the thyroid nodules were divided into groups A ( ≤ 10 mm), B ( > 10 mm, < 20 mm), and C ( ≥ 20 mm).By evaluating the ultrasonic features of the thyroid nodules from each group with ACR TI-RADS, the nodules were finally scored and classified.All the cases were scanned and evaluated by 2 attending or higher-seniority ultrasonographers with more than 10 years of work experience.The results were reviewed by a chief physician in case of inconsistent interpretations.

Statistical analysis
The statistical software of SPSS26.0 and MedCalc19.3.1 were adopted.The measurement data confirming to normal distribution were expressed as mean ± standard deviation, with the independent sample t-test applied for inter-group comparisons.In contrast, the non-normally distributed measurement data were represented by median (P25, P75), with the Mann-Whitney U test utilized for inter-group comparisons.The enumeration data were exhibited in frequency numbers and percentages, with the χ 2 test and Fisher exact probability method adopted for inter-group comparisons.The receiver operating characteristic curves (ROCs) was plotted based on the scores and categories of the thyroid nodules in each group, and the areas under the curves (AUCs) as well as the Youden indices were calculated.The AUC of 0.85-0.95indicated excellent diagnostic efficacy, 0.70-0.85indicated moderate diagnostic efficacy, and 0.50-0.70indicated low diagnostic efficacy.The score corresponding to the maximal Youden index was the optimal cut-off value for each group, with p < 0.05 were considered statistically significant.

Results
In group A, a total of 766 nodules from 566 patients (male, n = 128; female, n = 438) aged from 18 to 76 years (mean age, 48.25 ± 11.42 years) were included.The maximum diameters of the nodules were 2-10 mm, with the mean diameter being 5.74 ± 2.10 mm.Among these nodules, 357 were located in the left lobe, 397 in the right lobe, and 12 in the isthmus.In group B, a total of 229 nodules from 203 patients (male, n = 43; female, n = 160) aged from 16 to 79 years (mean age, 50.41 ± 12.73 years) were included in group B. The maximum diameters of the nodules were 11-19 mm, with the mean diameter being 14.02 ± 2.66 mm.Among these nodules, 110 were located in the left lobe, 119 in the right lobe, 4 in the isthmus, and 1 in the pyramidal lobe.And in group C, a total of 188 nodules from 183 patients (male, n = 50; female, n = 133) aged from 16 to 79 years (mean age, 50.41 ± 12.73 years) were included.The maximum diameters of the nodules were 20-86 mm, with the mean diameter being 34.05 ± 11.06 mm.Among these nodules, 81 were located in the left lobe, 105 in the right lobe, and 1 in the isthmus.
Table 1 shows the pathological results of the thyroid nodules in each group based on the 5th edition of WHO Classification of Endocrine Organ Tumors (thyroid) published in 2022 [7,8].In group A, there were 99 benign nodules with an incidence rate of 12.92%, and 667 malignant nodules with an incidence rate of 87.08%.In group B, there were 111 benign nodules with an incidence rate of 48.47%, 4 low-risk nodules with an incidence rate of 1.75%, and 114 malignant nodules with an incidence rate of 49.78%.In the group C, there were 130 benign nodules with an incidence rate of 69.15%, 6 low-risk nodules with an incidence rate of 3.19%, and 52 malignant nodules with an incidence rate of 27.66%.For the convenience of statistical analysis, low-risk thyroid nodules were classified as malignant in our study.
Table 2 shows the ultrasonic features of thyroid nodules in each group.The results of χ 2 test shows that there were significant statistical differences in composition, echo, Table 3 shows the ACR TI-RADS scores and categories of thyroid nodules in each groups.The ACR TI-RADS scores were 0-14 for all the included thyroid nodules, while the mean scores were 7.92 ± 2.64 for the 766 nodules in group A, 6.82 ± 3.80 for the 229 nodules in group B and 4.51 ± 3.49 for the 188 nodules in group C. The results of independent sample t-test shows that ACR TI-RADS scores of the benign nodules in each group were lower than the scores for malignant ones, and the differences were statistically significant (p < 0.05).
The ROCs were plotted based on the benignity and malignancy of the nodules as well as their scores and categories (Fig. 8).It was shown that the AUCs of group A were evidently smaller than those of groups B and C (p < 0.005).Besides, in each group, the AUCs of the ACR TI-RADS scores were larger than those of the ACR TI-RADS categories (Z values were 4.355, 15.008 and 15.426 respectively, with p values < 0.001).Table 4 shows the AUC, optimal cutoff values and diagnostic efficacy indicators of each groups.
The results suggested that the sensitivity and positive predictive value of the ACR TI-RADS scoring system for thyroid malignant nodules gradually decreased as thyroid nodules increased, the differences were statistically significant (χ 2 = 9.448, 16.093, p < 0.001).The specificity and negative prediction gradually increase, the difference were statistically significant (χ 2 = 68.442,34.918, p < 0.001).As thyroid nodules increase, the sensitivity and positive predictive value of the ACR TI-RADS classification system for thyroid malignant nodules gradually decrease, and the differences were statistically significant(χ 2 values were 21.118 and 6.934, p < 0.001), the specificity and negative prediction gradually increased as well, and the difference were statistically significant (χ 2 = 86.377,29.055, p < 0.001).

Discussion
TI-RADS is a thyroid classification system established on the basis of thyroid image reporting and management Fig. 1 The incidence of benign and malignant nodules for different ACR TI-RADS categories in each group.Panel (A) shows the distribution of benign and malignant thyroid nodules ≤10 mm; panel (B) shows the distribution of benign and malignant thyroid nodules >10 mm but <20 mm; and panel (C) shows the distribution of benign and malignant thyroid nodules ≥20 mm Fig. 2 The left thyroid nodule with the maximum diameter of 6 mm in a 60-year-old female patient, postoperative pathology: nodular goiter Endocrine (2023) 82:569-579 Fig. 4 The right thyroid nodule with the maximum diameter of 11 mm in a 64-year-old male patient, postoperative pathology: papillary thyroid carcinoma Fig. 5 The right thyroid nodule with the maximum diameter of 19 mm in a 45-year-old female patient, postoperative pathology: follicular adenoma Fig. 3 The right thyroid nodule with the maximum diameter of 6 mm in a 43-year-old female patient, postoperative pathology: papillary thyroid microcarcinoma Fig. 6 The left thyroid nodule with the maximum diameter of 35 mm in a 74-year-old male patient, postoperative pathology: medullary thyroid carcinoma Fig. 7 The right thyroid nodule with the maximum diameter of 30 mm in a 53-year-old female patient, postoperative pathology: nodular goiter Fig. 8 ROCs of each group.Panel (A) shows diagnostic efficiency of ACR TI-RADS score and classification system for thyroid nodules ≤10 mm; panel (B) shows diagnostic efficiency of ACR TI-RADS score and classification system for thyroid nodules >10 mm but <20 mm; and panel (C) shows diagnostic efficiency of ACR TI-RADS score and classification system for thyroid nodules ≥20 mm system.With this system, a standardized imaging dictionary has formed and the malignant risk of thyroid nodules has been stratified in a standardized method, which can diminish the impact brought by ultrasonographer's subjectivity to a certain extent, thus facilitating the wide application of the system.Currently, there are multiple TI-RADSs worldwide, the most influential one of which is the ACR TI-RADS published in 2017.Research has indicated that compared with other risk stratification systems [9].In the present study, with the surgical pathology as the gold standard, we conducted analyses on 1183 different-sized thyroid nodules and found that the ultrasonic features of thyroid nodules of different sizes varied to a certain extent.Besides, the diagnostic efficacy of ACR TI-RADS on different-sized thyroid nodules also differed, which was obviously better on thyroid nodules of > 10 mm than on the nodules of ≤ 10 mm.
A total of 1183 thyroid nodules were included in the study, including 766 thyroid nodules with maximum diameters of ≤ 10 mm.It was revealed that the malignant nodules (ACR C-TIRADS score, 8.45 ± 2.36) were primarily the PTMCs, and 4 malignant ones were follicular adenocarcinomas.While among the benign nodules (ACR C-TIRADS score, 5.33 ± 2.41), the pathological types included nodular goiter, chronic lymphocytic thyroiditis, follicular adenoma and granulomatous thyroiditis.Although the scores of malignant micro-nodules were visibly higher than those of benign ones, the score spans of benign and malignant nodules were both large, resulting in a great overlap in the scores of these two types of nodules.Similarly, the ultrasonic features of benign and malignant thyroid micro-nodules and thyroid microcarcinomas were also largely overlapped [10].On ultrasound, the typical PTMC is solid, hypoechoic/extremely hypoechoic and in vertical position with ill-defined margins and microcalcifications [11][12][13].As research continues to advance, the diagnostic accuracy of ultrasound has gradually improved, but for the benign nodules in group A, the chronic lymphocytic thyroiditis and granulomatous thyroiditis share obviously overlapped ultrasonic features with PTMC [14][15][16].Chronic lymphocytic thyroiditis, can be called Ha shimoto's thyroiditis as well, is the most common autoimmune thyroiditis.It is considered a risk factor of papillary thyroid carcinoma with an incidence of 0.5-30% in Hashimoto's thyroiditis patients which is higher that in general population [17].Despite a higher incidence of papillary thyroid carcinoma in patients with Hashimoto's thyroiditis, it's more difficult to distinguish between benign and malignant nodules in these patients because they often present a coarse and heterogeneous thyroid parenchyma caused by the repetitive damage of chronic inflammation [18].Granulomatous thyroiditis, can be called subacute thyroiditis as well, is caused by the infiltration of inflammatory cells.In most cases,subacute thyroiditis has typical clinical and laboratory features, with a focal or multifocal blurred margin and an irregularly shaped hypo-echoic lesion in the thyroid lobes on ultrasonography.The classic sonographic features of subacute thyroiditis are thyroid enlargement, blurred margins and irregularly shaped hypoechoic thyroid lesions, usually without microcalcifications and with poor color flow in the hypo-echoic area, In most of these atypical cases, which are termed clinically atypical subacute thyroiditis, the primary presentation is the presence of painless thyroid nodules found by physical examination and by ultrasonography [19,20].Thus, when the clinical, sonographic and even laboratory features of subacute thyroiditis are non-characteristic, which are termed clinically atypical subacute thyroiditis (CAST), the CAST lesions are frequently misdiagnosed as malignant lesions, which often results in an unnecessary thyroid lobectomy [14].Thyroid follicular tumors include thyroid follicular adenoma and follicular carcinoma, which have overlapping clinical manifestations, ultrasound features, and molecular biology.It is also difficult to distinguish between follicular adenoma and follicular carcinoma before surgery through ultrasound, FNA, and immunohistochemistry [21].Meantime, most nodular goiters are both cystic and solid with clear margins and a transverse diameter that is larger than the longitudinal diameter on ultrasound [22].But some nodular goiters in our study are initially discovered with small sizes and similar features to PTMC.These nodules are mostly benign degenerative thyroid nodules, primarily induced by bleeding, degeneration, infarction, and fibrosis.
As the nodules gradually shrink in the course of hematoma absorption, the thyroid degenerative nodules become increasingly irregular, presenting with malignant sonographic features [23,24].Additionally, if the nodular goiters are accompanied by interstitial collagenization, calcification and crystal formation, hyperechoic dots may appear with time in some of the goiters, but it is hard to accurately discriminate solid thyroid micro-nodules from psammoma bodies with ultrasound [25].Besides, the small sizes of nodules are associated with the low probability of cystic degeneration, making it more difficult to differentiate the nodules from thyroid carcinomas.That's why many benign nodules in the current study were nodular goiters with the ACR TI-RADS categories of ≥ TR4.Among nodules with larger sizes, the rate of surgically confirmed malignant nodules is lower.We propose this phenomenon may be partially related to the more evident distinction between the ultrasonic features of benign and malignant nodules as well as the better diagnostic accuracy of the ultrasound.For the thyroid nodules with larger sizes in group C, the malignant thyroid nodules of some pathological types are more clearly characterized, including the undifferentiated and medullary carcinomas that are solid and large in sizes [26,27].While most of the benign thyroid nodules are nodular goiters and follicular adenomas, the larger sizes of which are associated with the higher probability of cystic degeneration [28,29], leading to the lower scores of benign thyroid nodules.In the current study, the optimal cut-off value for both benign and malignant thyroid nodules of ≥ 20 mm was > 4 points, lower than the values of nodules in other groups.Despite the low cut-off value, the AUC reached 0.904, indicative of high diagnostic efficacy.Therefore, we believe that the features of benign and malignant thyroid nodules of larger sizes are less likely to overlap ultrasonically, which promotes the further evaluation of nodules by ultrasound and also explains why the diagnostic efficacy of ultrasound for larger thyroid nodules is higher.Meanwhile, the thyroid nodules of group B lie between those of groups A and C in size, and combine the features of the nodules in the remaining two groups, with the repetition rate of the ultrasonic features of benign and malignant nodules also falling between the rates of the other two groups.
Although there is no apparent difference between the AUCs of groups B and C, the optimal cut-off value of benign and malignant nodules of group B is > 5 points.The findings further suggest that the ultrasonic feature repetition rate of benign and malignant thyroid nodules in this group is low.
Moreover, the current study also demonstrated that the sensitivity corresponding to the optimal cut-off value of ACR TI-RADS was lower among larger thyroid nodules, with statistical differences observed only between groups A and C. The findings indicate that the cut-off value of ACR TI-RADS owns significantly stronger capacity in accurately identifying benign thyroid nodules of > 10 mm than those of ≤ 10 mm, which is partially because of the more suspicious features and higher scores of some smaller malignant thyroid nodules.In contrast, the specificity corresponding to the optimal cut-off value of ACR TI-RADS was higher among larger nodules, implying the higher incidence of mis-diagnoses among benign thyroid nodules of smaller sizes.By evaluating the different-sized thyroid nodules with ACR TI-RADS, the current study found that compared with the thyroid nodules of > 10 mm, the ACR TI-RADS only exhibited moderate diagnostic efficacy on nodules of ≤ 10 mm with a cut-off value of > 6 points and low specificity, indicating that ACR TI-RADS have suboptimal performance in managing micro-thyroid nodules, and the diagnostic efficacy of ultrasound was overestimated.This is consistent with the results of ACR TI-RADS in managing pediatric patients with thyroid nodules [30].Given that, if the prognosis of thyroid microcarcinoma is favorable and the diagnostic efficacy of ACR TI-RADS on it is moderate, follow-up observation should be conducted on the thyroid micro-nodules so as to avoid the over-diagnoses and overtreatment of thyroid micro-nodules to a certain extent.At the same time, there is still some controversy regarding the principles for handling suspicious thyroid nodules, while most TI-RADS do not recommend FNA for thyroid nodules ≤ 1 cm without high-risk factors [4,31,32].In response to the low specificity and negative predictive value of ACR TI-RADS for thyroid nodules in this study, the authors believe that it may be appropriate not to score and classify ACR TI-RADS for thyroid nodules ≤ 10 mm without highrisk factors.And it is recommended to only conduct regular follow-up observation based on the ACR TI-RADS management principles, When the nodule is greater than 10 mm, scoring and classification can also help reduce the incidence of over diagnosis and treatment of thyroid nodules.
Our works have an limitation.In our present study, the diagnostic efficacy of ACR TI-RADS on thyroid nodules was assessed based on the cut-off value corresponding to the maximum AUC, not on the need for further invasive examinations.In this context, whether the optimal cut-off values obtained in the study can be further applied to clinical practices still requires more exploration.
In conclusion, the ultrasonic features of different-sized thyroid nodules vary in a certain degree, impacting the diagnostic efficacy of ultrasound on thyroid cancer.The ACR TI-RADS showed obviously better diagnostic efficacy with lower cut-off values on larger thyroid nodules than on smaller ones.Therefore, ACR TI-RADS may be more suitable to evaluate the thyroid nodules of > 1 cm.

Table 1
Pathological results of thyroid nodules in each group n (%)

Table 2
Ultrasonic features of the thyroid nodules in each group n (%)

Table 3
Ultrasound scores of the thyroid nodules in each group based on ACR TI-RADS

Table 4
Optimal cut-off values and diagnostic values in each group