Categories: Pediatric Head and Neck Imaging, Pediatric, Thyroid Cancer, Thyroid Nodule, Ultrasound, K-TIRADS
Source: Journal of the Korean Society of Radiology
Pediatric thyroid cancer exhibits different clinical characteristics from those in adults, requiring a tailored approach for imaging evaluation and management. Although thyroid nodules are less common in children, they carry a higher risk of malignancy and often present with cervical lymph node or distant metastases, while long-term outcomes remain favorable. The revised 2021 Korean Thyroid Imaging Reporting and Data System (K-TIRADS) aims to reduce unnecessary biopsies by refining the ultrasound lexicons and biopsy criteria. In pediatric patients, biopsy thresholds for high-suspicion nodules have been lowered, and a risk factor-based strategy has been adopted for intermediate-suspicion nodules. This review summarizes the clinical and molecular features of pediatric thyroid cancer, outlines key updates in the 2021 K-TIRADS, and discusses important considerations for its application in children, with the aim of supporting radiologists in appropriately applying adult-based risk stratification systems in pediatric practice.
Pediatric thyroid cancer accounts for approximately 9.5% of all thyroid cancers (1), and in Korea, from 2004 to 2016, the age-standardized incidence rate was 0.92 per 100,000 people, indicating a relatively low incidence; however, the annual percentage change of 4.0% suggests a gradual increase over recent decades (2). Although thyroid nodules are less common in children than in adults, when detected, the likelihood of malignancy is higher (22%–26%), emphasizing the importance of radiological evaluation (34). Pediatric thyroid cancer also frequently presents with lymph node (LN) or distant metastases at diagnosis, demonstrating an aggressive initial presentation but an excellent long-term prognosis (5).
These clinical characteristics suggest limitations in evaluating pediatric thyroid nodules and cancers using the same criteria as those used for adults. The most commonly used US-based risk stratification systems (US RSS) have been developed for adult populations (36789). Direct application to pediatric patients may fail to adequately reflect the distinct natural history of pediatric thyroid cancer, the relatively smaller thyroid volume in children (10), and a familial predisposition, potentially leading to either excessive biopsies or underestimation and insufficient evaluation of lesions. In particular, the diffuse sclerosing subtype of papillary thyroid carcinoma, which is more common in children than in adults, may not present as a discrete nodular lesion and therefore requires careful US interpretation (11).
In this context, the revised 2021 Korean Thyroid Imaging Reporting and Data System (2021 K-TIRADS) introduced several important changes to overcome the limitations of the previous system (7). The revised system adjusted the US lexicons and biopsy criteria based on the nodule size and proposed a selective biopsy strategy considering pediatric characteristics, aiming to reduce unnecessary examinations and treatment while avoiding missed clinically significant lesions.
This review briefly summarizes the clinical characteristics of pediatric thyroid nodules and thyroid cancer and discusses the emerging molecular biological background. It further outlines key updates in the 2021 K-TIRADS and important considerations for pediatric applications from a radiological perspective. In addition, clinically relevant biopsy strategies, the role of CT as an adjunct imaging modality, and future directions for radiological evaluation are discussed, to provide practical guidance for radiologists less familiar with pediatric thyroid nodule evaluation.
As noted above, pediatric thyroid cancer demonstrates a clinical behavior distinct from that of adult thyroid cancer. Cervical LN metastasis is present in approximately 60%–80% of pediatric patients at diagnosis, compared to 20%–50% in adults. Distant metastasis is also more frequent in children (5%–25%) than in adults (2%–5%). In contrast, long-term prognosis is generally better in children. All-stage mortality is approximately 1%–2% in adults and <1% in children. Even in the presence of distant metastasis, mortality is 25%–40% in adults, compared to only 1%–7% in children (5). Thus, pediatric thyroid cancer can be understood as a distinctive disease that presents with aggressive clinical features at diagnosis but demonstrates a favorable long-term prognosis.
The precise mechanisms underlying these characteristics have not yet been clearly elucidated; however, differences in the molecular biological features between pediatric and adult thyroid cancers have been suggested to contribute in part. The BRAF mutation, which is the most common genetic alteration in adults, has been identified in only approximately 25% of pediatric patients (5) and is observed more frequently in patients aged 15–24 years than in those younger than 15 years. In contrast, oncogenic fusions involving RET, NTRK, and ALK are commonly observed in pediatric patients (1213). Indeed, thyroid cancers harboring gene fusions have been reported to demonstrate more aggressive clinical behaviors. In a study on pediatric differentiated thyroid cancer, Franco et al. (14) reported that tumors with RET/NTRK fusions showed a higher frequency of lateral cervical LN metastasis (69.8%) than those with the BRAF mutation (15.4%). Furthermore, distant metastasis was observed in 39.5% of the patients in the RET/NTRK fusion group, whereas it was not observed in the BRAF mutation group.
Understanding the clinical and molecular characteristics of pediatric thyroid cancer provides an important basis for refining the US risk assessment, interpreting imaging findings within the appropriate clinical context, and guiding decisions regarding biopsy, molecular testing, and treatment strategies.
US risk stratification of pediatric thyroid nodules using the 2021 K-TIRADS is summarized in Table 1. Compared to the 2016 K-TIRADS, the 2021 K-TIRADS introduced changes in US lexicons, nodule classification criteria, and management and treatment decision-making based on the US diagnosis. The most important change was the adjustment of biopsy criteria according to the nodule size in cases with suspicious findings of high-risk thyroid cancer, with the aim of reducing unnecessary biopsies while maintaining the appropriate sensitivity for high-risk thyroid cancer (715).
Several important changes have been made to the US lexicons for thyroid nodules (15). The composition remained largely unchanged compared with that of the 2016 K-TIRADS. For echogenicity, the definition of markedly hypoechoic was adjusted from “lower than the strap muscles” to “similar to or lower than the strap muscles,” and mildly hypoechoic was clearly defined as lower than the normal thyroid parenchyma but higher than the strap muscles. In cases of coexisting diffuse thyroid disease, the 2016 K-TIRADS determined the echogenicity relative to the adjacent thyroid parenchyma, whereas the 2021 K-TIRADS recommended assessment based on the typical echogenicity of normal thyroid tissue. For orientation, the assessment of a nonparallel orientation was modified to be evaluated only in the transverse plane, improving the diagnostic agreement and ease of application. Spiculated or microlobulated margins were unified and simplified under the category of irregular margins. Echogenic foci were previously described as calcifications in the 2016 K-TIRADS. However, because echogenic foci may also represent dense colloids rather than true calcifications, the terminology was revised to echogenic foci.
The key changes to consider when evaluating pediatric thyroid nodules using the 2021 K-TIRADS are summarized as First, the biopsy threshold for K-TIRADS 5 nodules was reduced from >1.0 cm in adults to >5 mm in children. This reflects the difficulty of applying active surveillance, which may be considered for nodules ≤1 cm in adults, to pediatric patients, as well as the fact that surgery is the standard treatment once thyroid cancer is diagnosed in children (1516). According to a study by Kim et al. (17), the malignancy risk of pediatric K-TIRADS 5 nodules is very high at 94%, and performing biopsy for K-TIRADS 5 nodules ≤1 cm demonstrates a higher diagnostic sensitivity and accuracy than applying the adult criterion (171819). However, nodules ≤5 mm are generally not candidates for surgical treatment and are often too small for practical biopsy; therefore, follow-up observation is recommended. Second, for K-TIRADS 4 nodules, when thyroid cancer risk factors were present, the biopsy threshold was reduced from >1.5 cm to >1.0 cm. Thyroid cancer risk factors include increased ^18^F-fluorodeoxyglucose uptake on PET, a family history of thyroid cancer or hereditary cancer syndromes, and clinically concerning symptoms such as dysphonia. In addition, the nodule location, detailed US features, and patient factors, such as age, comorbidities, and preferences regarding evaluation and treatment, should be considered. Representative hereditary cancer syndromes include DICER1 syndrome (20), PTEN hamartoma tumor syndrome (21), Carney complex (22), familial adenomatous polyposis (23), Werner syndrome (24), multiple endocrine neoplasia type 2 (2526), and Bannayan-Riley-Ruvalcaba syndrome (327). In addition, other factors that can be considered risk factors for thyroid cancer in children include a history of head and neck radiation therapy (2829) and Hashimoto thyroiditis (30). Third, diffuse infiltrative lesions were incorporated into K-TIRADS 4. Consequently, conditions such as the diffuse sclerosing subtype of papillary thyroid carcinoma (Fig. 1), in which punctate echogenic foci can be observed throughout the thyroid gland, as well as lymphoma and Langerhans cell histiocytosis (which can present as diffuse lesions), can be classified as K-TIRADS 4 without confusion (Fig. 2).
Meanwhile, intrathyroidal ectopic thymus requires caution, as it can present as a hypoechoic solid nodule and may occasionally exhibit punctate echogenic foci (Fig. 3) (3132). Consequently, it may be classified as K-TIRADS 4 or 5, leading to unnecessary biopsies. Although not mentioned as a separate entity in the current K-TIRADS, it should be considered in the differential diagnosis as a relatively common benign lesion encountered during pediatric thyroid nodule evaluation. Generally, intrathyroidal ectopic thymus can be differentiated from malignant thyroid nodules based on the 1) US features similar to those of the normal mediastinal thymus, 2) it is typically identified in infants and young children <4 years of age, a demographic in which thyroid cancer is extremely rare, and 3 it tends to remain stable or decrease in size during follow-up, with some cases showing increasingly indistinct margins or a transition to higher echogenicity. Therefore, when a nodule with punctate echogenic foci is detected in a child <4 years of age, it is advisable to consider intrathyroidal ectopic thymus as a key differential diagnosis and prioritize periodic follow-up observation over immediate biopsy.
The size criteria for biopsy determination according to the K-TIRADS-based risk classification in children are summarized in Table 1. In the pediatric population, biopsy is recommended for K-TIRADS 5 nodules >0.5 cm. For K-TIRADS 4 nodules, selective biopsy is suggested for the nodules >1.0–1.5 cm, depending on the presence of risk factors. Although biopsy is performed for K-TIRADS 3 nodules >2.0 cm, K-TIRADS 2 nodules are generally not considered indications for biopsy.
US-guided fine-needle aspiration (FNA) is a widely used primary diagnostic tool in adults. Core needle biopsy (CNB) may be used as a supplementary or alternative test in cases of repeated nondiagnostic results or when differentiation of follicular lesions is required.
Indeed, according to a meta-analysis of 4,235 thyroid nodules initially presenting as nondiagnostic or atypia of undetermined significance/follicular lesion of undetermined significance (AUS/FLUS), a definitive diagnosis was obtained in 87.9% of cases with CNB compared to 68.4% with repeat FNA, with a statistically significant difference. The area under the curve for CNB was notably high at 0.981 (33). Furthermore, CNB demonstrated a higher diagnostic yield than did FNA for follicular-patterned neoplasms (prematching: 9.0% vs. 0.5%, p < 0.001; 9.0% vs. 0.6%, p < 0.001) (34). Studies have also reported high diagnostic performance when using CNB as the primary biopsy method (35363738). However, complications such as hematoma, hemorrhage, pain, infection, transient hemoptysis, and nerve injury may occur. Serious complications such as arteriovenous fistulas have rarely been reported (39). Additionally, CNB can be technically challenging when the nodule is located posteriorly or adjacent to vital structures, such as the carotid artery or trachea, which may increase the risk depending on the practitioner’s experience (40). Another concern is the lack of a standardized pathological classification system or established management guidelines for CNB specimens. Therefore, K-TIRADS states that CNB can serve as an alternative to FNA in selected cases when performed by experienced practitioners (741).
Currently, clear evidence and standardized guidelines for the biopsy of pediatric thyroid nodules have not been established. Therefore, it is advisable to follow the same principles as in adults. However, there are several important considerations for pediatric patients. First, the thyroid gland is smaller, and adjacent vital structures are more closely situated in children than in adults, requiring greater caution during the biopsy. In particular, meticulous evaluation is required not only for structures that are relatively visible on US, such as the common carotid artery, internal jugular vein, trachea, and esophagus, but also for structures that can be easily overlooked, including the vagus nerve, recurrent laryngeal nerve, cervical sympathetic ganglion, superior and inferior thyroidal arteries, and anterior jugular vein (42). Second, sedation may be required for biopsies of pediatric patients. Children have a low tolerance for anxiety and fear and may struggle to control spontaneous movements during the procedure. Performing a biopsy without adequate sedation can increase the risk of procedural failure, nondiagnostic results, or complications. In addition, because completely suppressing subtle movements is difficult, even with sedation, ensuring an appropriate depth of sedation while establishing a safe biopsy route and procedural plan is essential. As sedation can be associated with adverse effects such as respiratory depression, airway obstruction, and hypoxia, continuous monitoring of vital signs and respiration by skilled personnel is mandatory during and after the procedure (43). Third, because obtaining sufficient specimens and performing repeat biopsies are more challenging in children than in adults (3), the decision to proceed with a biopsy should be carefully considered based on the US findings and risks. If a decision is made to proceed with a biopsy, every effort must be made to ensure that a sufficient sample is acquired, to minimize the need for repeated procedures.
The 2021 K-TIRADS does not explicitly provide separate management strategies for pediatric thyroid nodules based on biopsy results. When applying the guidelines directly, the optimal timing for a repeat biopsy of nondiagnostic or AUS/FLUS nodules, typically within 6–12 months, is determined by the nodule size, US features, clinical factors, and presence of poor prognostic factors, such as suspected nodal metastasis or gross extrathyroidal extension. US surveillance may also be considered for AUS/FLUS cases. Even with benign cytopathological results, a repeat biopsy within 12 months is recommended for high-suspicion nodules. For intermediate- or low-suspicion nodules, follow-up US is performed after 24 months. Diagnostic lobectomy is generally recommended for follicular neoplasm/suspicious for a follicular neoplasm nodules, although follow-up US or molecular testing may be considered in selected cases. Surgical treatment is primarily recommended for nodules that are malignant or suspicious for malignancy.
The 2023 revised Bethesda System (The Bethesda System for Reporting Thyroid Cytopathology) explicitly includes independent discussions on pediatric thyroid nodules and provides the risk of malignancy (ROM) for each category along with potential management strategies (44). The European Thyroid Association (ETA) also provides management strategies based on pediatric biopsy results and recommends repeat US and FNA at a minimum interval of 3 months for Bethesda I, follow-up at 6–12-month intervals for 5 years for Bethesda II, repeat FNA after 6 months for Bethesda III or IV, and surgery for Bethesda V or VI (Table 2) (45). The 2015 American Thyroid Association guidelines recommended surgery even for the indeterminate category; however, these recommendations were published based on limited data during a period when pediatric-specific evidence was relatively scarce (3).
In the 2021 K-TIRADS, cervical LN US is considered an essential component of thyroid nodule evaluation. These guidelines recommend prioritizing morphological abnormalities over the size when determining the likelihood of metastasis. Given that pediatric patients exhibit a higher incidence of cervical LN metastasis than adults at the time of diagnosis, a thorough US evaluation of the cervical LNs is critical, alongside primary nodule assessment.
The US features suggestive of metastasis include cystic changes, echogenic foci indicative of calcification, focal or diffuse cortical hyperechogenicity, and abnormal peripheral or diffuse vascularity. These findings are recognized as highly useful predictors of LN metastasis; the presence of even a single feature classifies the node as suspicious, with a reported ROM of 73%–88%. Conversely, probably benign LNs are defined by typical benign characteristics, such as an echogenic hilum or radiating hilar vascularity, in the absence of suspicious findings, and carry an ROM of <3%. Indeterminate LNs do not exhibit features of either the suspicious or probably benign categories, with an estimated ROM of approximately 20%. The 2021 K-TIRADS recommends FNA for suspicious LNs with a short-axis diameter >3–5 mm and for indeterminate LNs >5 mm. During FNA, measuring thyroglobulin levels in the needle washout (Tg-washout) is recommended to improve the diagnostic sensitivity (46).
To date, the evidence evaluating the diagnostic performance of individual US features for cervical LN metastasis, specifically in pediatric patients, remains insufficient. In a study of 52 pediatric and adolescent patients with differentiated thyroid cancer, Navallas et al. (47) reported a sensitivity of 79%, specificity of 84%, and positive predictive value of 90% for US in the evaluation of cervical LN metastasis. Within this study, significant predictors of metastasis included changes in cortical echogenicity (p ≤ 0.0001), abnormal vascularity (p ≤ 0.0001), and a round shape, defined by a long-to-short axis ratio of ≤2 (p = 0.0002). As noted in the ETA guidelines, although pediatric-specific evidence remains limited, it is currently recommended to adapt and apply adult cervical LN US classification systems to children (45).
Although the 2021 K-TIRADS designates US as the primary imaging modality for thyroid cancer evaluation, it also acknowledges that CT plays a significant complementary role (7). The acquisition of both non-contrast and contrast-enhanced phases is recommended. Non-contrast CT is particularly useful for identifying calcifications, evaluating ectopic thyroid tissue, and differentiating tumor recurrence from normal remnant thyroid tissue after thyroidectomy. Contrast-enhanced CT is instrumental for evaluating cervical LN metastasis. LNs are classified as suspicious on CT if they exhibit any of the following cystic change, calcification, strong (focal or diffuse) enhancement, and heterogeneous enhancement. Nodes lacking these suspicious features but exhibiting a loss of hilar fat or normal vascular enhancement are classified as indeterminate, whereas those maintaining hilar fat or normal vascular enhancement are considered probably benign. According to one study, among the CT findings for cervical LN metastasis, cystic change (<25 Hounsfield unit) demonstrates a sensitivity of 24.9% and specificity of 99.0%, while calcification shows a sensitivity of 5.7% and specificity of 100%. Strong enhancement (greater than that of the adjacent muscle during the arterial phase) demonstrates a sensitivity of 92.4% and specificity of 70.1%, while the loss of hilar fat demonstrates a sensitivity of 97.8% and specificity of 34.0% (48).
Furthermore, CT offers distinct advantages in evaluating areas where US may be limited, such as the retropharyngeal space, lower neck, and upper mediastinum, and in defining the anatomical relationship with the major vessels, trachea, and esophagus. It is useful for both preoperative staging and postoperative follow-up to assess the extent of recurrence and establish plans for surgery or additional treatment.
However, as most evidence regarding the diagnostic performance of CT for detecting LN metastasis is based on studies in adult patients, caution is required when directly applying these criteria to the pediatric population. Given that children exhibit higher radiation sensitivity and a relatively greater long-term risk of carcinogenesis (49), the indications for CT must be carefully selected. Therefore, in pediatric patients with thyroid cancer, it is advisable to perform CT selectively, primarily in cases where US evaluation is restricted or when the scan can provide clinically critical information for preoperative staging. In addition, when a CT scan is performed, concerted efforts must be made to minimize radiation exposure by implementing low-dose protocols and avoiding unnecessary repeat imaging. The adoption of dual-energy CT or low-voltage CT protocols can significantly reduce both radiation and contrast medium doses, while maintaining the diagnostic image quality (5051). These advanced techniques can be considered useful alternatives for reducing radiation exposure in pediatric patients with thyroid cancer, particularly when both non-contrast and contrast-enhanced CT imaging are required.
Pediatric thyroid cancer exhibits clinical and molecular characteristics distinct from those in adults, and these differences influence the radiologic evaluation and management strategies for pediatric thyroid nodules. The 2021 K-TIRADS has improved diagnostic efficacy by refining US lexicons and biopsy criteria. Specifically, by providing a dedicated selective biopsy strategy for children, it aims to reduce unnecessary biopsies while effectively screening clinically significant lesions.
Although CT serves as a valuable supplementary tool, it is essential to establish appropriate indications and implement low-dose techniques to minimize radiation exposure in the pediatric population. Ultimately, the radiologic evaluation of pediatric thyroid nodules requires a customized approach that integrates the specific clinical context of children with the underlying intent of the 2021 K-TIRADS, rather than a simple application of adult standards.