Reply to the Letter to the Editor: “Ultrasound malignancy risk stratification of thyroid nodules based on the degree of hypoechogenicity and echotexture”
by Ji Ye Lee, Dong Gyu Na (email@example.com)Ultrasound malignancy risk stratification of thyroid nodules based on the degree of hypoechogenicity and echotexture
We would like to thank Dr. Rueda et al. for their interest in our recent study. Our study validated the feasibility of using adjusted definitions for echogenicity and mixed echogenicity which are differently used in various risk stratification systems. We agree with the comments suggesting that malignancy risk is not determined by a single ultrasound (US) predictor and should be assessed by a combination of US features. Among them, hypoechogenicity is a well-known independent predictor of malignancy by multivariate analysis [1, 2]. Based on this evidence, we performed a subgroup analysis to investigate whether the malignancy risk of US hypoechogenicity differs according to the various US patterns.
Currently, the definition of hypoechogenicity has not been standardized for the risk stratification of thyroid nodules. The American Association of Clinical Endocrinologists/ American College of Endocrinology/ Associazione Medici Endocrinologi guidelines, as well as the American College of Radiology (ACR) guidelines classify nodule hypoechogenicity as marked vs. mild for risk stratification. Nodules that show an echogenicity that is similar to strap muscles are considered as “mild” according to these guidelines but were considered “moderate” hypoechogenicity in our study [3, 4]. We found that classifying nodules with moderate hypoechogenicity together with marked hypoechoic nodules (rather than classifying them as “mild) was more appropriate. Moreover, according to the American Thyroid Association (ATA) and the Korean Society of Thyroid Radiology (KSThR) guidelines, marked/moderate hypoechogenicity was not classified separately from mild hypoechogenicity for risk stratification [5, 6]. Our study demonstrated that classifying nodule hypoechogenicity as marked/moderate vs. mild is optimal for malignancy risk stratification.
Regarding nodules with heterogeneous echogenicity, the European Thyroid Association (ETA) guidelines,  the KSThR guidelines, and the ACR guidelines [4, 6] provide conflicting definitions. Our study showed that when classifying mixed echogenicity nodules for risk stratification, classification should be based on the predominant echogenicity of the nodule.
In the current US lexicon, strap muscles are used as reference structures for determining marked hypoechogenicity. We used both strap and sternocleidomastoid (SCM) muscles as reference structures for evaluating hypoechogenicity and validated that this strategy could enhance the reliability of the hypoechogenicity lexicon. Considering the important role of the US lexicon for estimating the malignancy risk of thyroid nodules, the lexicon should include definitions that optimize the predictability of malignancy. This proposed nodule echogenicity lexicon will be useful for risk stratification and management of thyroid nodules and could provide a basis for future guidelines. An effort is underway to develop an international risk stratification system.
We agree with the opinion that having only two readers may have limited the assessment of inter-observer agreement when evaluating nodule echogenicity. However, we would like to stress that the two observers included in our study were independent radiologists who work at different institutions with different levels of expertise, which might reflect the heterogeneity of practitioners in real-world practice.
In the last point of the letter, Dr. Rueda and colleagues suggested that we provide the cytologic results of the surgically resected nodules. Of the 463 surgically resected nodules, 449 nodules underwent fine needle aspiration (FNA). When we diagnosed thyroid malignancy by using the criteria of Bethesda categories 4, 5, and 6 as test-positive on FNA, we obtained a sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of 70.4%, 95.7%, 93.8%, 77.7%, and 83.5%, respectively. When we diagnosed cancer using the criteria of Bethesda categories 5 and 6 as test-positive on FNA, we obtained a sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of 69.9%, 99.6%, 99.3%, 78.1%, and 85.3%, respectively.