With the development of high-frequency ultrasonic devices, more thyroid nodules have been diagnosed. Although most thyroid nodules are benign, a diagnosis differentiating malignant from benign nodules is still of vital importance for determining the appropriate therapy. A differential diagnosis is still a difficult problem for clinicians. Elastography has emerged as a valuable technology to assist in the diagnosis of thyroid nodules due to its ability to reflect the stiffness of tissue that is related to malignant lesions. A large number of studies have shown the excellent performance of elastography in the differentiation of malignant and benign nodules[15-17]. Nevertheless, several studies have revealed the opposite result[11, 12]. The present study was the first multicenter study in China with a large sample size to determine the value of real-time strain elastography in the discrimination of malignant from benign thyroid nodules with high-suspicion characteristics based on the 2015 ATA guidelines. The results showed that the combination of USE with the ES had good efficacy for the discrimination of malignant from benign high-suspicion thyroid nodules, with a high sensitivity (92.4%), accuracy (79.0%) and NPV (85.5%), especially in nodules ≥1 cm (95.0%, 85.9%, and 87.8%, respectively). Additionally, logistic regression analysis showed that the ES and SR were independent predictors of malignancy.
US is the preferred technology for thyroid nodules due to its convenience, maneuverability and low cost. Conventional US malignant characteristics, including microcalcifications, hypoechogenicity, irregular margins, and taller than wide were significantly different between malignant and benign nodules in this study. Logistic regression analysis showed that microcalcifications, taller than wide, irregular margins and ETE were independent predictors of malignancy. However, previous studies have reported largely varied sensitivities and specificities[5-7]. The diagnostic efficacy of conventional US parameters in this study was also unsatisfactory. Although irregular margins and ETE yielded good sensitivity, their specificity was rather poor. Similarly, despite their high specificity, the sensitivity of taller than wide and disrupted rim calcifications with a small extrusive hypoechoic soft tissue component was unacceptable. The accuracy of conventional US parameters was also limited.
Elastography takes advantage of the change in elasticity of soft tissues resulting from specific pathological or physiological processes, and the low elasticity observed on USE is highly correlated with malignancy. Since the emergence of elastography, it has shown an outstanding sensitivity, specificity, PPV and NPV[16,18- 20], with the specificity and PPV reaching up to 100%. Elasticity, which is absent in conventional US, is the single feature with the best diagnostic performance, as well as a potent predictor of malignant thyroid nodules[21, 22]. With an increase in the ES, the percentage of malignancy increased significantly in this research, confirming that the ES is correlated with malignancy. In this research, we used the 5 scoring parameters proposed by Itoh, who defined an ES >3.5 as the cut-off value; the sensitivity, specificity, accuracy, PPV and NPV were 92.4%, 60.7%, 79.0%, 76.3% and 85.5%, respectively, with an AUC of 0.828. However, contrary opinions have also been presented. A multicenter study by Moon et al reported that elastography showed an inferior performance in the differentiation of malignant and benign thyroid nodules relative to grayscale US features. They investigated all solid nodules, including hyperechogenic, isoechogenic, hypoechogenic and marked hypoechogenic nodules, in their research. In this study, we recruited solid hypoechoic nodules according only to the recommendations of the 2015 ATA guidelines. Another study with three scoring methods showed that USE had a limited sensitivity and PPV in detecting malignant thyroid nodules and was not superior to conventional US. A retrospective study involving 197 thyroid nodules utilized the iU22 system for elastography and classified the color mapping as blue versus not blue, showing that the ES and SR had a limited ability in differentiating benign from malignant thyroid nodules. However, Magri et al showed that the strain index was significantly higher in malignant thyroid nodules than in benign thyroid nodules and displayed a good diagnostic performance. It can be speculated that the instruments and scoring method used are among the factors that affect the diagnostic efficacy of elastography.
Elasticity is a qualitative method; thus, it is inevitably affected by the practitioner. The SR, which is an objective and semiquantitative technology, was introduced to address this disadvantage. However, whether the ES or SR is more accurate is a long-standing controversy[20, 26-28]. Furthermore, the cut-off value of SR varied largely in different studies[28, 29]. In this research, using an SR>2.99 as the cut-off value, the sensitivity, specificity, accuracy, PPV and NPV were 81.1%, 50.1%, 68.9%, 65.9% and 67.9%, respectively, with an AUC of 0.738, showing no advantage over the ES. Though the SR is a semiquantitative method, it is defined as the ratio obtained by dividing the mean strain within the lesion by the mean strain of the surrounding normal tissue; thus, thyroid echogenicity surrounding thyroid nodules will inevitably affect the SR. As a result, it might be inferred that no additional information is provided by the SR, which is more time consuming than the ES, in the discrimination of malignant from benign thyroid nodules. Although shear wave elastography (SWE) has drawn more attention in recent years and has been reported to be more accurate than USE in some studies[30-32], Tian et al’s and Hu et al’s meta-analysis showed that USE had a better diagnostic performance than SWE, with comparable specificity between methods[33, 34]. Even more recently, a prospective study comprising 243 nodules revealed that USE yielded the highest performance compared with the TI-RADS score and SWE. The aforementioned data confirm that USE is still a promising diagnostic tool for discriminating malignant from benign thyroid nodules.
Another controversial topic is whether the combined application of US and ES may provide better results for thyroid nodule characterization. The results of the present research demonstrated that the sensitivity and NPV were dramatically higher with the combination than with the TI-RADS and ES alone, and the accuracy was higher with the combination than with the TI-RADS alone but not with the ES alone, consistent with the study by Cantisani. However, the specificity decreased after the two were combined.
According to the 2015 ATA guidelines, nodules ≥1 cm in the largest dimension with a high-suspicion sonographic pattern are recommended for FNA. We compared the diagnostic efficacy of elastography in nodules ≥1 cm and <1 cm, and the results showed that the AUC of the ES for nodules ≥1 cm was 0.875, with excellent sensitivity, accuracy, PPV, and NPV and good specificity. The aforementioned parameters were all superior to those of nodules <1 cm. The SR demonstrated a similar trend, except it had a lower sensitivity and NPV in nodules ≥1 cm than in nodules <1 cm. Wang’s research also showed a similar result: elastography yielded higher sensitivity for nodules larger than 1 cm. It might be concluded that elasticity is a helpful tool for discriminating benign and malignant thyroid nodules. FNA is not popular in many developing countries because of the lack of skillful cytologists and the high dependence on the skill and experience of the operator and cytologist. Furthermore, FNA is an invasive technique that is time consuming and costly. Thus, the use of FNA is limited in certain areas.
This study has certain limitations. Cytology without lesion growth was also accepted as a reference method for benign lesions, as suggested by international guidelines, which may have resulted in selection bias. However, including only histology would increase the number of malignant lesions. Furthermore, the number of follicular carcinomas and medullary carcinomas, which was determined by morbidity, was low, consistent with previous studies[16, 38]. No reproducibility tests were performed in this study because all US diagnoses were performed by two physicians.