Evaluating modified ACR TI-RADS nodule size for thyroid nodules in pediatric population

Abstract

Purpose

In this study, we aimed to investigate the diagnostic performance of modified thyroid nodule size based on the American College of Radiology Thyroid Imaging Reporting and Data System (ACR TI-RADS) for the pediatric population.

Method

This retrespective study enrolled patients younger than 21 years who underwent thyroid nodule US examination between January 2021 and June 2024. Thyroid nodules were assessed according to the ACR TI-RADS and the modification of size criteria. The diagnostic performance of sensitivity, specificity, area under the receiver operating characteristic curve (AUC), unnecessary biopsy rates, and missed malignancy rates were calculated and compared.

Results

A total of 204 thyroid nodules were retrospectively included, of them 51 were confirmed malignancies. For ACR TI-RADS, the sensitivity, specificity, and AUC were 76.5% (95% CI 62.5%-87.2%), 75.2% (95% CI 67.5%-81.8%), and 0.825 (95% CI 0.756–0.894), demonstrating moderate diagnostic performance. When applied modified lower nodule size, the sensitivity was improved (82.4%, 95% CI 69.1%-91.6%, and 88.2%, 95% CI 76.1%-95.6%); however, which was at the cost of decreasing specificity (65.4%, 95% CI 57.3%-72.9% and 53.6%, 95% CI 45.4%-61.7%).

Conclusion

ACR TI-RADS demonstrated moderate diagnostic performance in the younger population. Lowering the thyroid nodule size criteria for FNAB leads to a higher detection rate of malignant, however, which at the cost of a significantly increased number of biopsies. Larger multicentric studies are required to customize the recommendation further before it can be accepted and used generally in the pediatric population.

Introduction

Thyroid cancer accounts for 3% of all cancers worldwide, with approximately 586,000 new cases every year [1]. In the past three decades, the incidence of thyroid cancer has risen significantly; however, the mortality rates of thyroid cancer are low and have remained stable or even decreased, indicating that the increase in incidence is primarily because of overdiagnosis [2]. Compared with only about 5% malignant thyroid nodules in adulthood, the malignancy rate increases up to 18–26% among children and adolescents [3]. Moreover, the pediatric population often has more advanced cancer when diagnosed, with greater lymph node extension, metastases, and a higher chance of recurrence [4]. These underscore the critical need for early detection and timely intervention. Fine-needle aspiration biopsy (FNAB) remains the most accurate tool currently available to evaluate thyroid nodules by extracting cellular material for cytological analysis. Although FNAB is generally safe and cost-effective, it can yield indeterminate or false-negative results and may lead to repeated procedures or unnecessary surgeries. Therefore, accurate risk stratification is essential to ensure that FNAB is reserved for nodules with a higher likelihood of malignancy, particularly in pediatric patients, where overdiagnosis and overtreatment remain key concerns [5].

In the past decade, several standardized risk stratification systems for thyroid cancer were proposed. Of them, the two most commonly used are the American Thyroid Association (ATA) guidance and the ACR TI-RADS [6, 7]. The primary difference between the above two guidelines is that ACR TI-RADS raised the nodule size threshold recommended to FNAB. In the ATA guidelines, FNAB is recommended for high- or intermediate-suspicion nodules ≥ 1 cm, for low-suspicion nodules ≥ 1.5 cm, and for very-low-suspicion nodules ≥ 2 cm. By comparison, according to ACR TI-RADS, FNAB is recommended for nodules of TR3 ≥ 2.5 cm, TR4 ≥ 1.5 cm, and TR5 ≥ 1 cm. While the ACR TI-RADS has been widely utilized in adult populations, its application in children is less understood, particularly regarding how nodule size impacts risk stratification and clinical decision-making. Nodule size plays a critical role in the ACR TI-RADS classification system, as larger nodules are associated with a higher risk of malignancy. However, pediatric thyroid nodules differ biologically from adult nodules, raising questions about the appropriateness of size thresholds and the system’s sensitivity to malignancy in younger patients [3]. In fact, no available dedicated risk stratification systems for children currently. In recent years, some studies tried to validate these adult scoring systems in the pediatric population. However, diagnostic performance among these studies varied widely [8]. This paper aims to evaluate the applicability of ACR TI-RADS in pediatric patients, with a specific focus on the role of nodule size in malignancy risk assessment, then provide insights into potential modifications or considerations for improving pediatric thyroid nodule evaluation using current adult ACR TI-RADS.

Method and materials

Patient population

This retrospective single-center study was approved by the Institutional Review Board (IRB) of Suqian People’s Hospital, and the requirement for written informed consent was waived by the IRB of Suqian People’s Hospital. This study was conducted in accordance with the Declaration of Helsinki, and all data were collected in compliance with the Health Insurance Portability and Accountability Act (HIPAA). Between January 2021 and June 2024, patients met following criteria were included: (1) younger than 21 years when underwent US examination; (2) nodules demonstrated on thyroid US; and (3) with FNAB or surgical histopathological results as the reference standard. Patients who met any following criteria would be excluded: 1)cases with cytology classified as atypia of undetermined significance or follicular lesion of undetermined significance, non-diagnostic results, or lacking subsequent histopathologic examination; 2) diagnosed as suspicious for follicular neoplasm on biopsy but without final histopathologic report from the surgical specimen; and 3) diagnosed with benign initially, but showing increased nodule size on follow-up US and without histopathologic confirmation.

US examination and nodule classification

All US examinations were performed with 10–12-MHz linear probes and real-time US systems (EPIQ 5 and EPIQ 7, Philips Healthcare). Nodule images were evaluated using the gray-scale image and Doppler US, with transverse and longitudinal planes. Thyroid nodule features used for evaluation are as follows: composition (cystic, spongiform, mixed, or solid), echogenicity (anechoic, hyperechoic/isoechoic, hypoechoic, or very hypoechoic), echogenic foci (none/large comet tail artifacts, macrocalcifications, peripheral calcifications, punctate echogenic foci), margins (smooth, ill-defined, lobulated/irregular, extra-thyroidal extension), and shape (wider-than-tall, or taller-than-wide). Lymph nodes were deemed abnormal if they exhibited calcifications, had a rounded shape, lacked a visible echogenic hilum, or displayed abnormal echotexture. Nodule size was determined based on the maximum dimensions, and in patients with multiple lesions, only the most suspicious nodule was analyzed. Initial US examinations were performed by 4 radiologists with more than 5 years of experience. In this study, US images were interpreted by two experienced US radiologists (with 6 and 17 years of experience) who were blinded to pathological results. Each thyroid nodule was categorized according to the ACR TI-RADS, in which US features were used to evaluate and assign individual lesions. As thyroid nodule size in the pediatric population is smaller than that in adulthood, the nodule biopsy threshold for adults may not be appropriate in children and adolescents. Therefore, we assessed the impact of nodule size cutoff as follows: (1) for TI-RADS category TR1 (benign) and TR2 (not suspicious), the original ACR guidelines recommend no FNAB, while Strategy 2 introduces follow-up for TR1 and TR2 nodules between 1.0 and 4.0 cm and allows FNAB for nodules larger than 4.0 cm; (2) in the TR3 category (mildly suspicious), the ACR recommends FNAB only if the nodule is ≥ 2.5 cm, while both revised strategies lower this threshold to ≥ 1.5 cm. Additionally, Strategy 2 recommends follow-up for nodules of 1.0–1.5 cm; (3) for TR4 nodules (moderately suspicious), the ACR recommends FNAB at ≥ 1.5 cm, while both revised strategies reduce this to ≥ 1.0 cm, with Strategy 2 also suggesting follow-up for nodules as small as 0.5 cm; (4) in the TR5 category (highly suspicious), all three strategies support FNAB at ≥ 1.0 cm. Strategy 2 is the most aggressive, recommending consideration of FNAB even for nodules between 0.5 and 0.9 cm, along with follow-up for all nodules ≥ 0.5 cm [9, 10]. The modified nodule size criteria are demonstrated in Table 1.

Table 1.

ACR TI-RADS and revised thyroid nodule size criteria

Category	ACR TI-RADS	Strategy 1 [9]	Strategy 2 [10]
TR1 (benign)	No FNAB	No FNAB	> 4.0 cm: FNAB 1.0–4.0 cm: Follow-up
TR2 (not suspicious)	No FNAB	No FNAB	> 4.0 cm: FNAB 1.0–4.0 cm: Follow-up
TR3 (mildly suspicious)	≥ 2.5 cm: FNAB ≥ 1.5 cm: Follow-up	≥ 1.5 cm: FNAB	≥ 1.5 cm: FNAB 1.0–1.5 cm: Follow-up
TR4 (moderately suspicious)	≥ 1.5 cm: FNAB ≥ 1.0 cm: Follow-up	≥ 1.0 cm: FNAB	≥ 1.0 cm: FNAB ≥ 0.5 cm Follow-up
TR5 (highly suspicious)	≥ 1.0 cm: FNAB ≥ 0.5 cm: Follow-up	≥ 1.0 cm: FNAB ≥ 0.5 cm: Follow-up	≥ 1.0 cm: FNAB 0.5–0.9 cm: Consider FNAB ≥ 0.5 cm Follow-up

Abbreviations: ACR, American College of Radiology; FNAB, fine needle aspiration biopsy; TI-RADS, Thyroid Imaging Reporting and Data System

Fine needle aspiration biopsy

Biopsy was conducted by a radiologist at our institution, with a free-hand technique under US guidance. The final pathological diagnosis of nodules was decided by cytopathologic results, according to the Bethesda System for Reporting Thyroid Cytopathology [11]. In cases where a child with concerning cytology underwent thyroidectomy, histology findings replaced the cytology results. For those whose cytology did not indicate the need for thyroidectomy, appropriate follow-up, including either repeat FNAB or US after at least one year, was conducted to confirm the benign status of nodules. Generally, the indication of FNAB of thyroid nodules depended on clinicians, including US imaging features, nodule size, patient age, and underlying condition.

Statistical analysis

The mean ± standard deviation (SD) or median (interquartile range, IQR) is used to present descriptive statistics. For continuous variables, comparisons between two groups were evaluated using Mann-Whitney U test for non-normally distributed variables and independent t-test for normally distributed variables. The Shapiro-Wilks test was used to test the normal distribution of data. The diagnostic performance for the detection of malignancy was evaluated in terms of sensitivity, specificity, and AUC, which was compared with the DeLong test [12]. Additionally, the number of unnecessary biopsy rates and the number of malignant nodules missed were calculated. Statistical analyses were performed using R software (version 3.6.3), with P < 0.05 indicated as statistically significant.

Results

Patient characteristics

A total of 204 young patients were included in the final study population. The median age was 14.6 (range 6–20, IQR 13.7–18.1), with the majority of patients being female (142/204, 69.6%), the characteristics of included thyroid nodules are described in Table 2. Of all 204 thyroid nodules, 51 were diagnosed as malignant and all were papillary thyroid carcinoma, including classic variant papillary thyroid carcinoma (n = 35, 68.6%), follicular variant of papillary thyroid carcinoma (n = 14, 27.5%), and the diffuse sclerosing variant of papillary thyroid carcinoma (n = 2, 3.9%).

Table 2.

Characteristics of patients

Variable	Benign(N = 153)	Malignant (N = 51)	Total (N = 204)
Age (Years, Median, IQR)	14.6/13.7–18.1	15.3/14.6–18.5	14.6/13.7–18.1
Gender (Female/Male)	114/39	40/11	142/62
Nodule Size (Median, IQR)	2.1/1.4–3.5	2.2/1.5–3.6	2.1/1.4–3.6
Size (cm)
< 1.0	86	9	95
1.0 ~ 1.5	38	16	54
≥ 1.5	29	26	55

Abbreviations: IQR, interquartile range; SD, standard deviation

Diagnostic performance in predicting malignant thyroid nodules

Most nodules were classified as TR3 category (n = 82, 40.2%), followed by TR4 category (n = 50, 24.5%), TR2 category (n = 39, 19.1%), and TR5 category (n = 27, 13.2%). Category TR1 accounted for the least number of thyroid (n = 6, 2.9%). The nodules within the TR5 were associated with the highest rate of malignancy (n = 21/27, 77.8%), then followed by the TR4 risk category with a 36.0% cancer rate (n = 18/50). Most malignant nodules were classified into these two categories (n = 39/77, 50.6%). However, there are 11 nodules were classified in the intermediate TR3 category (n = 11/82, 13.4%), and 1 malignant nodule classified in the TR2 category. No malignant nodule was categorized as TR1.

Figure 1 shows the ROC analysis of each scorings. While using the cutoff ≥ 4, the sensitivity and specificity were 76.5% (95% CI 62.5%-87.2%) and 75.2% (95% CI 67.5%-81.8%), respectively, with AUC of 0.825 (95% CI 0.756–0.894). With strategy 1, the sensitivity and specificity were 82.4% (95% CI 69.1%-91.6%) and 65.4% (95% CI 57.3%-72.9%), respectively, with AUC of 0.831 (95% CI 0.768–0.895). By lowering the nodule size cutoff for TR3 and TR4, 3 more malignancies were recommended FNAB. Nevertheless, 15 benign nodules were recommended FNAB. When using nodule size ≥ 4.0 cm for TR1 and TR2, 3 more malignancies were identified. The sensitivity increased to 88.2% (95% CI 76.1%-95.6%), whereas the specificity decreased to 53.6% (95% CI 45.4%-61.7%), with AUC of 0.799 (95% CI 0.737–0.860). Details on the diagnostic performance and number of FNAB are presented in Table 3; Fig. 2. A representative case of 1.0 cm ACR TI-RADS 4 nodule is presented in Fig. 3.

Fig. 1.

ROC analysis. ACR, American College of Radiology; AUC, area under the receiver operating characteristic curve; FNAB, fine needle aspiration biopsy; TI-RADS, Thyroid Imaging Reporting and Data System

Table 3.

Diagnostic performance for ACR TI-RADS and revisions of thyroid nodule size criteria

Indicator	Sensitivity (95% CI)	Specificity (95% CI)	AUC (95% CI)	P	Unnecessary Biopsy	Missed Malignancy
ACR TI-RADS	76.5% (62.5-87.2%)	75.2% (67.5-81.8%)	0.825 (0.756–0.894)	/	38	12
Strategy 1	82.4% (69.1-91.6%)	65.4% (57.3-72.9%)	0.831 (0.768–0.895)	0.36*	53	9
Strategy 2	88.2% (76.1-95.6%)	53.6% (45.4-61.7%)	0.799 (0.737–0.860)	0.86*	71	6

Abbreviations: ACR, American College of Radiology; AUC, area under the receiver operating characteristic curve; CI, confidence interval; TI-RADS, Thyroid Imaging Reporting and Data System

* Compared with the ACR TI-RADS

Fig. 2.

Number of FNAB for each scorings. FNAB, fine needle aspiration biopsy

Fig. 3.

Longitudinal gray-scale US image in 18-year-old male patient with papillary thyroid cancer shows 1.0 cm thyroid nodule with hypoechogenicity, smooth margins, and no echogenic foci. According to the American College of Radiology Thyroid Imaging Reporting and Data System, this nodule received two points for composition, two points for echogenicity, 0 points for shape, 0 points for margin, and 0 points for echogenic foci, classifying as TR 4 and suggesting follow-up. However, according to Strategy 1 and Strategy 2, this nodule is recommended fine needle aspiration biopsy

Discussion

In this study, we investigated the ACR TI-RADS and its modification for the pediatric population. Our analysis demonstrated that the adult scoring systems performed moderately in pediatric populations, with an AUC of 0.825. The optimal threshold for differentiating malignancies from benign was TI-RADS ≥ 4, where the sensitivity and specificity were 76.5% and 75.2%. When lowered nodule size cutoff values, the sensitivity increased to 82.4% and 88.2%; however, the specificity decreased to 65.4% and 53.6%, respectively. In a recent meta-analysis, Xing et al. summarized the diagnostic accuracy from 8 studies applying TI-RADS for children and adolescents, in which the pooled sensitivity and specificity were 84% and 61%, with a pooled AUC of 0.85 [13]. Since ACR TI-RADS has been released, several studies have applied the revised TI-RADS to the pediatric population, in which the biopsy cutoff values for nodule categorized as TR3, TR4, and TR5 was lowered [10, 14, 15]. Nevertheless, the diagnostic accuracy among these studies varied widely. In Ahmed et al. a nodule size of 1.0 cm for TR4 and 1.5 cm for TR3 yielded higher sensitivity (84.6%) but lower specificity (52.4%) [9]. Another study also demonstrated that lower the size threshold criteria of the adult TI-RADS to guide management decisions for pediatric thyroid nodules may lead to early detection of malignant thyroid nodules in children (94.4%); however, this was at the cost of the increased number of unnecessary biopsies (63.3%) [15]. Huang et al. used a nodule size ≥ 3.5 cm as the cutoff value for FNAB in TR3 thyroid lesion, their study demonstrated that this could improve the diagnostic performance, reducing unnecessary biopsy rates and missed malignancy rates in patients < 19 years [16]. Nonetheless, considering that most studies suggested lowering the cutoff value of nodule size for FNAB, in the current study we did not validate larger cutoff values, as the nodule size is smaller in the pediatric population than in adults.

The applicability of adult-based scoring systems, such as ACR TI-RADS, in predicting and managing pediatric thyroid nodules remains uncertain. While some studies have reported satisfactory performance, others have highlighted limitations in diagnostic accuracy, suggesting the need for potential modifications to better suit the pediatric population [10, 17]. Besides the ACR TI-RADS, other standardized guidelines for the risk stratification of thyroid nodules also have been tested on the pediatric population, of them the most reported was the ATA guidelines [18–20]. Despite the ATA proposed guidelines for pediatric thyroid nodules, this is mostly based on the stratification system for the adult population with additional emphasis on US characteristics (e.g., hypoechogenicity, irregular margin, and blood flow) and clinical context rather than size alone [18]. However, the ATA guideline recommends thyroid nodules ≥ 1 cm for biopsy, which may improve sensitivity but potentially reduce specificity. In a recent meta-analysis including 6 studies, the pooled sensitivity and specificity of ATA were 84% and 55%, with an AUC of 0.82, showing moderate diagnostic performance [13]. Another guideline that has been commonly reported is the K-TIRADS, which is also prone to generate high sensitivity but low specificity [21, 22]. Kim et al. evaluated the of K-TIRADS 2021 in pediatric patients, which demonstrated significant improvement in diagnostic accuracy as compared to 2016 K-TIRADS, particularly with a biopsy nodule size of 0.5 cm for K-TIRADS 5 and 1.0–1.5 cm for K-TIRADS 4 [23]. For the pediatric population, it advises to biopsy a lesion with size of 0.5 to 1.0 cm with strong suspicion. Additionally, raising the biopsy cut-off range for K-TIRADS 4 and 3 nodules improved diagnostic accuracy.

Due to the different biological behavior of thyroid cancer in children and adults, there is a need to modify the way protocols are used, especially for nodule size [24]. One of the primary advantages of ACR TI-RADS compared with other scoring systems was its accessibility, applicability, and general reputation and acceptance in clinical practice. Therefore, modifications to the ACR TI-RADS risk level for pediatric nodules might improve the detection of pediatric thyroid cancers. It has been demonstrated in the literature that using the size criterion of the adult ACR TI-RADS to direct the treatment of children’s thyroid nodules is not reliable. In the study of Richman et al., 404 pediatric thyroid nodules were evaluated with the adult ACR TI-RADS, they found that 17 out of 77 malignancies were missed at the initial assessment. In their study population, nine of these were suggested to be followed, while the remaining eight were neither recommended for FNAB nor follow-up [10]. Our study demonstrated that lowering the size threshold for thyroid nodules increased the detection of malignant lesions; however, it also significantly raised the number of benign nodules undergoing FNAB. Because malignancy is more accurately predicted by US features and clinical variables than by size alone, all US suspected nodules in children should receive FNAB, regardless of size. For pediatric thyroid nodules, ACR TI-RADS guidelines revealed a significant rate of undetected malignancy (21.7%) and unnecessary biopsy (62.7%), according to a previous meta-analysis analysis [8]. Nonetheless, it appeared that the rates of unneeded biopsies in children were greater than those in the adult population. This could be explained by the fact that ectopic thymus tissue in children’s thyroid glands can mimic a thyroid nodule, resulting in needless biopsy and overtreatment. Given that children had a higher overall malignancy rate of nodules (22–26%) than adults (5–10%), there may be a greater concern about cancer in children, which could account for the higher rate of needless biopsies in pediatric patients. Therefore, treating thyroid nodules in individual for children and adolescents may not be appropriate for adult recommendations. Increased intra-nodular vascularity and suspicious cervical lymphadenopathy both were concerning ultrasound features for malignancy; however, these characteristics were not included in the ACR TI-RADS.

Our study has several limitations. First, this is a single-center, retrospective study. Because the US is a dynamic imaging modality, selection bias is unavoidable when static images or limited sweeps are reviewed retrospectively. Furthermore, a retrospective study of such images is somewhat limited because US examination is very user-dependent and subjective to the images taken, with human specialists not always consistent in how findings are interpreted. Second, the small sample size of our study may affect the generalization of the conclusion in the clinical application; however, we overcame this constraint by using more reliable statistical analysis. Lastly, because the ACR TI-RADS does not include clinical history, lymph node morphology, or the existence of enhanced intra-nodular vascularity, these characteristics were not used to evaluate thyroid nodules in this study.

Conclusions

In the younger group, ACR TI-RADS showed a moderate diagnostic performance. However, lowering the thyroid nodule size criteria for FNAB may result in a higher rate of malignancy identification at the cost of a noticeable increase in the number of biopsies. Before the suggestion is widely accepted and applied in the pediatric population, larger multicentric studies are needed to further tailor it.

Author contributions

Guarantor of the article: Z.M. Conception and design: M.J. and Z.L.L. Collection and assembly of data: Z.L.L. and S.Y. Data analysis and interpretation: S.Y. All authors contributed to the article and approved the submitted version.

Funding

None.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Written informed consent was waived by the Institutional Review Board of Suqian People’s Hospital.

Consent for publication

Patients’ data were anonymized. Not applicable.

Competing interests

The authors declare no competing interests.