Usefulness of combined use of contrast-enhanced ultrasound and TI-RADS classification for the differentiation of benign from malignant lesions of thyroid nodules

Early diagnosis, correct identification and diagnosis of the benignity and malignancy of thyroid nodules has important significance for clinical treatment method selections and prognosis predictions. TI-RADS classification identifies the benignity and malignancy of nodules through a comprehensive analysis of nodule morphological characteristics, and this classification provides a standard for the diagnosis of thyroid nodules; however, improvements can be made in its application for the identification of benignity and malignancy among thyroid nodules [2‐4, 11]. CEUS is a recently developed ultrasound imaging technology, and studies have shown that monitoring microcirculation perfusions after the injection of microbubble contrast agents can be used to identify the benignity and malignancy of thyroid nodules [5‐10]. However, to our knowledge, studies presenting TI-RADS classification criteria that also consider CEUS diagnoses have not been reported. Thus, the primary goal of the present study was to develop a new and effective TI-RADS classification standard by investigating the value of diagnosing thyroid nodules using CEUS combined with TI-RADS and determining whether improvements were made to the diagnostic accuracy.

Currently, a number of scholars have begun applying TI-RADS classifications to thyroid nodules [2‐4]; however, a generalised standard has not been developed. In the present study, we have referenced BI-RADS classifications and the classifications by Park et al. [2], Horvath et al. [3] and Kwak et al. [4] to classify thyroid nodules as score 1-5, with score 4 divided into the subscores 4a and 4b. Clinical practice has confirmed that this classification is simple and practical. To develop an objective classification, the American scholar Horvath et al. [3] first proposed a classification using ten US sonographic characteristics, whereas Park et al. [2] proposed using 12 US sonographic characteristics. Recently, Korean scholars Kwak et al. [4] proposed using five ultrasound sonographic characteristics for the classification, and these characteristics are simple and generalised, and highly practical in clinical practice. Therefore, in the present study, we used these five US signs for the classification of each thyroid nodule, although certain flaws were noted [4]. In the classification, malignant indicators are not available for scores 2 and 3; therefore, the difference between these two scores is vague. In the present study, we separated score 2 from score 3 and classified nodules with one malignant sign as score 3. This classification facilitates the standardisation of scores 2 and 3, and reduces the subjectivity of their classification. In addition, the standard presented here classifies nodules with one malignant sign as score 4a and nodules with two malignant signs as score 4b; thus, there may be higher false-positive rates.

To verify the accuracy and effectiveness of the classification standard in the present study, we performed a comparative analysis with the classification method by Kwak et al. [4]. The results showed that our classification method has a higher accuracy than the classification method by Kwak et al. (90.3 % vs 73.8 %). Moreover, because the criteria presented here classify nodules with two malignant signs as score 4a and nodules with three malignant signs as score 4b, the false-positive rate is significantly reduced, which may prevent unnecessary surgical treatment. In the present study, we misdiagnosed 31 nodules based on TI-RADS standards, which accounted for 9.7 % of the diagnoses. The misdiagnosed nodules were mainly in scores 4a and 4b, indicating that these two scores have a degree of overlap.

As an important technique in US medicine, CEUS has played an important role in the diagnosis of thyroid lesions over the last decade [5‐10, 12]. Studies from numerous researchers suggest that most of the contrasts indicating low enhancement and heterogeneous enhancement are malignant thyroid nodules [5, 9], whereas most contrast modes indicating high enhancement and circular enhancement are benign nodules [7]. CEUS has been proposed to evaluate the thyroid nodules and the sensitivity, specificity and accuracy for the diagnosis of malignant thyroid nodules were reported to be 76.9–88.2 %, 84.8–95.1 % and 82.6–90.4 % [5‐7], respectively. The results presented here are consistent with the results of the above-mentioned studies, which indicates that CEUS has clinical value in the identification and diagnosis of benignity and malignancy of thyroid nodules. In this study, 31 nodules were misdiagnosed based only on TI-RADS standards, and 32 nodules were misdiagnosed based only on CEUS standards. A statistically significant difference was not observed in the diagnostic accuracy of CEUS and TI-RADS. Nineteen misdiagnosed nodules based only on TI-RADS standards were correctly diagnosed by CEUS, and 20 misdiagnosed nodules based only on CEUS standards were correctly diagnosed by TI-RADS. The results indicated that the two methods may have complementary effects in the differential diagnosis of thyroid nodules.

In our study, there were eight thyroid nodules coexisting with Hashimotoʼs thyroiditis in our study, which were all accurately diagnosed by CEUS and partly misdiagnosed by TI-RADS. The result was consistent with that of Zhao et al. [13]. The result indicated that CEUS is also effective in the diagnosis of malignant thyroid nodules coexisting with Hashimotoʼs thyroiditis and can improve the diagnostic accuracy. Of five inflammatory lesions, three nodules showed low-enhancement mode and were misdiagnosed. It indicated that CEUS may not be able to provide more valuable information for inflammatory lesions, and further studies with greater sample size are required to validate the study results.

At logistic regression analysis of 4a and 4b thyroid nodules, the following features showed an association with malignancy: solid component, marked hypoechogenicity, irregular margins, microcalcifications, taller than wide shape and low enhancement. These results were consistent with those of Hyobin et al. [14]. It further confirmed that the above features are effective in the diagnosis of malignant thyroid nodules. To use TI-RADS and CEUS on a regular basis in clinical routine it is important that it is highly reliable. The inter-examiner measurement agreement for TI-RADS and CEUS had a high positive correlation even between experienced and inexperienced examiners (k = 0.885, k = 0.990, respectively). It indicated that TI-RADS descriptors and CEUS mode are simple, objective and are less influenced by US physicians’ experience.

Although a combination of both methods presents significant improvements in diagnostic accuracy, 15 nodules were still misdiagnosed, which may have been caused by the following: (1) certain benign nodules may have fibroplasias during hyperplasia and form calcified fibrous septa and calcified nodule walls, thereby affecting the interpretation of CEUS images; (2) the contrast enhancement modes of certain benign and malignant nodules in scores 4a and 4b could have overlapped; (3) two malignant nodules that were diagnosed as benign nodules were the nodular goitres with complications from local papillary carcinoma and from local follicular cancer; this local carcinoma increased the difficulty of providing a correct diagnosis; (4) the pathological results of the two misdiagnosed TI-RADS score-5 nodules were inflammatory lesions, and conventional US and CEUS results for such lesions are similar to malignant nodules, thereby increasing the difficulty of providing a correct diagnosis.

The present study included the following limitations. (1) Most of the malignant nodules were papillary carcinomas, whereas most of the benign nodules were nodular goitres, and fewer additional pathological types were analysed. Therefore, the present study mainly confirms the diagnostic value of TI-RADS and CEUS for papillary carcinoma and nodular goitre, whereas the diagnostic value of these methods for other benign and malignant thyroid pathological types requires further investigation. (2) The present study represents a single thyroid clinic’s work; thus, results have to be confirmed with multi-centre studies and a large sample size.

In summary, CEUS and TI-RADS showed similar accuracy in the identification of benign and malignant thyroid nodules. The improved TI-RADS combined with CEUS can significantly increase the diagnostic accuracy for the qualitative identification of thyroid nodules, particularly for TI-RADS score 4a and 4b lesions, and if these findings can be replicated, then it may be possible to reduce the number of patients subjected to FNAB or short-term follow-up for benign-appearing solid thyroid nodules.