Print
facebook
twitter
Linkedin
Introduction
Thyroid nodules are detected in up to 68% of adults undergoing high-resolution ultrasound (US) examination [1]. In euthyroid patients, further evaluation typically involves cytological assessment using fine-needle aspiration (FNA) or histological evaluation using core needle biopsy (CNB), depending on the level of suspicion. Even when biopsy results are benign, continued ultrasonographic surveillance is recommended, and nodules demonstrating significant growth are often re-biopsied to exclude missed malignancy.
However, consensus regarding optimal follow-up strategies for thyroid nodules remains limited. The American Thyroid Association provides a weak recommendation, supported by low-quality evidence, to repeat US after 12–24 months for nodules with low- to intermediate-suspicion patterns. Repeat FNA should be considered if there is a ≥20% increase in at least two dimensions (minimum increase ≥2 mm) or a ≥50% increase in volume [2]. Similar recommendations have been proposed by the American Association of Clinical Endocrinologists [3]. In addition, the Korean Society of Thyroid Radiology recommends US follow-up at 1, 3, and 5 years for low-suspicion nodules (Korean Thyroid Imaging Reporting and Data System [K-TIRADS] category 3) that do not meet biopsy criteria [4]. For nodules confirmed as benign on FNA, follow-up US is recommended at 2 years.
Although nodule growth during surveillance has been proposed as a predictor of malignancy [5,6], its clinical utility remains uncertain. Some benign nodules exhibit slow growth, whereas others grow more rapidly [7–9]. Consequently, whether growth of low-suspicion (K-TIRADS 3) nodules on US predicts malignancy or neoplasm remains unclear. In the absence of definitive evidence, many clinicians continue to recommend annual follow-up, an approach that may be overly conservative and cost-inefficient.
This study evaluated nodule growth using the longest diameter alone in low-suspicion nodules and compared discrepancies between diameter- and volume-based growth assessments. It further assessed whether significant nodule growth (SNG) can distinguish malignant from benign nodules or thyroid neoplasms from nodular hyperplasia within the low-suspicion US category.
Materials and Methods
Compliance with Ethical Standards
This retrospective study was approved by the Institutional Review Board of the authors’ affiliated institution (IRB No. 2024-05-006), and the requirement for informed consent was waived owing to its retrospective design.
Study Patients
A retrospective review was conducted of 309 patients with 334 nodules classified as K-TIRADS 3 who underwent at least three thyroid US examinations over a follow-up period exceeding 1 year, with final diagnoses established between January 2020 and December 2024 at the authors’ institution. The median follow-up duration was 41 months (range, 16 to 239 months). During follow-up, nodules that developed suspicious US features and were reclassified as K-TIRADS 4 were excluded. Cytologic or histologic biopsy, including repeat biopsy, was performed based on the following indications: nodule size >2 cm, interval growth, or the presence of contralateral or coexisting malignancy. Surgical resection was subsequently performed according to biopsy results. A final diagnosis (benign or malignant) was assigned when one of the following criteria was met: (1) surgical pathology; (2) conclusive results from FNA or CNB on at least two occasions, reported according to the Bethesda System for FNA and the Korean Endocrine Pathology Thyroid Core Needle Biopsy Study Group classification for CNB [10,11] (conclusive categories defined as Bethesda II [benign], V [suspicious for malignancy], or VI [malignant]); or (3) an initial benign biopsy result with stable or decreased size on follow-up US performed more than 2 years later (Fig. 1).
Patient age, sex, cytopathologic findings, and final histopathological diagnoses were obtained from medical records. Because definitive differentiation between nodular hyperplasia and thyroid neoplasms requires surgical histopathological confirmation, only surgically resected nodules were included in this subgroup analysis.
US Examinations and Follow-up Protocol
All thyroid US examinations were performed using a 5–12 MHz linear-array transducer with either a Logiq 700 scanner (General Electric Healthcare, Milwaukee, WI, USA) or an IU22 scanner (Philips Medical Systems, Bothell, WA, USA). Examinations were conducted by one of four radiologists with 3–19 years of experience in thyroid imaging. Two radiologists (J.H.S. and M.K.K., with 20 and 4 years of experience, respectively) retrospectively reviewed the images in consensus. Thyroid nodules were evaluated retrospectively in three dimensions using US images to measure diameters and calculate nodule volume on both transverse and longitudinal planes. Surveillance US examinations were performed at intervals ranging from 5 to 24 months. Follow-up duration was defined as the interval between the initial US examination and the examination at which FNA or CNB was performed, or the preoperative US examination in surgically treated nodules.
According to the K-TIRADS classification [4], low-suspicion nodules (category 3) were defined as partially cystic or iso- to hyperechoic nodules without any of the following suspicious features: punctate echogenic foci, nonparallel orientation, or irregular margins.
Evaluation of Nodule Growth
SNG or shrinkage was evaluated using two established approaches [12,13]. The same three orthogonal diameters measured on ultrasonography (length, width, and height) were used for both the volume- and diameter-based assessments. Under the volume-based criterion (VBC), a change was considered significant when nodule volume increased or decreased by at least 50%. Nodule volume was estimated using the ellipsoid formula (V=π/6×length×width×height), and changes of <50% were classified as stable. Under the diameter-based criterion (DBC), a significant change was defined as a ≥20% increase or decrease in at least two diameters, provided that the absolute difference was ≥2 mm [14]. Nodules that did not meet either criterion were classified as stable. In addition, for low-suspicion nodules, we assessed interval change using the longest diameter alone and defined a significant change as a ≥20% increase or decrease with a minimum absolute change of 2 mm.
Data Analysis and Statistics
The nodules were dichotomized into two outcome categories: (1) malignant versus benign and (2) thyroid neoplasm versus nodular hyperplasia. Continuous variables were summarized as medians and interquartile ranges (IQRs). The Mann-Whitney U test or Kruskal-Wallis test was used to compare continuous variables, and the chi-square test or Fisher exact test was used to compare categorical variables. Associations between nodule growth status and malignancy or neoplasm were evaluated using logistic regression.
SNG was assessed using the three growth criteria, and agreement among the results was evaluated using the kappa statistic. A kappa value >0.8 was interpreted as excellent agreement, whereas a value of 0.60–0.80 was considered substantial agreement.
Statistical significance was set at P < 0.05. All statistical analyses were performed using SPSS software (version 27, IBM Corp., Armonk, NY, USA).
Results
Baseline Clinicopathologic Characteristics of Patients
The baseline characteristics are summarized in Table 1. Final histopathological analysis identified 267 benign and 67 malignant lesions. In addition, surgically resected subgroups comprising nodular hyperplasia (n=69) and thyroid neoplasms (n=155) were analyzed. Malignant thyroid tumors included 12 follicular thyroid carcinomas (FTCs), 26 follicular variant papillary thyroid carcinomas, one differentiated high-grade thyroid carcinoma, six oncocytic carcinomas, six papillary thyroid carcinomas, 12 noninvasive follicular thyroid neoplasms with papillary-like nuclear features, and four tumors of uncertain malignant potential. Benign thyroid lesions included 173 nodular hyperplasias, 80 follicular adenomas, and 14 oncocytic adenomas.
The median patient age was 56 years (IQR, 45 to 64), and most patients were female (77.2%). The median overall follow-up duration, defined as the interval between the first and last US examinations, was 4.08 years (IQR, 2.39 to 8.41). There were no significant differences in age or sex between benign and malignant nodules or between nodular hyperplasia and thyroid neoplasms. The median initial nodule size was 2.1 cm (IQR, 1.4 to 3.1) in the overall cohort. Initial nodule size did not differ significantly between benign and malignant nodules (2.1 cm [IQR, 1.4 to 3.1] vs. 2.3 cm [IQR, 1.6 to 3.4], P=0.231) or between nodular hyperplasia and thyroid neoplasms (2.2 cm [IQR, 1.4 to 3.6] vs. 2.3 cm [IQR, 1.6 to 3.2], P=0.873). The median initial nodule volume was 2.0 mL (IQR, 0.6 to 5.8) in the overall cohort. Likewise, initial nodule volume did not differ significantly between benign and malignant nodules or between nodular hyperplasia and thyroid neoplasms. Nodule composition was significantly associated with the distinction between nodular hyperplasia and thyroid neoplasm (P<0.001), with solid nodules being more common in thyroid neoplasms (65.2%) than in nodular hyperplasia (33.3%).
The median growth rate of thyroid nodules was 1.02 mm/year (IQR, 0.43 to 1.98) overall. Malignant nodules showed a significantly higher growth rate than benign nodules (1.57 mm/y vs. 0.98 mm/y, P=0.010). Thyroid neoplasms likewise showed a higher growth rate than nodular hyperplasia (1.48 mm/y vs. 0.99 mm/y), although this difference was not statistically significant (P=0.274). Final diagnoses were established through several diagnostic pathways. Surgical excision was performed in 88.6% of nodules, with a malignancy rate of 21.9%. Among nodules managed with repeat biopsy or imaging follow-up, the malignancy rate ranged from 0% to 10%. Nodules managed without surgery had substantially longer follow-up durations than surgically excised nodules (Table 2).
Agreement between Growth Assessment Criteria
There was perfect agreement between SNG defined by VBC (>50% increase) and DBC (≥20% increase in at least two diameters with a minimum increase of ≥2 mm). All 193 nodules classified as showing SNG by volume also met the diameter-based criterion, and all 141 nodules classified as non-growing were stable by both methods (κ=1.00; 95% confidence interval [CI], 1.00 to 1.00). Agreement between VBC or DBC and growth defined by the longest diameter alone was substantial, with a kappa value of 0.681 (95% CI, 0.603 to 0.759). Among the 193 nodules classified as showing SNG by volume or diameter, 156 (80.8%) also showed growth based on the longest diameter, whereas 37 (19.2%) did not. Of the 141 nodules classified as stable by volume or diameter, 125 (88.7%) were also classified as stable based on the longest diameter.
Nodule Growth and Association with Malignancy in Low-Suspicion Thyroid Nodules
According to VBC, 193 nodules (57.8%) showed SNG and 141 (42.2%) were stable or showed shrinkage. Using DBC, the same distribution was observed: 193 nodules (57.8%) showed SNG and 141 (42.2%) were stable or showed shrinkage. When growth was assessed using the longest diameter alone, 172 nodules (51.5%) showed SNG and 162 (48.5%) were stable or showed shrinkage (Fig. 2).
For SNG defined by VBC or DBC and by the longest diameter alone, the median time to significant growth was 2.9 years (IQR, 2.0 to 4.8) and 2.6 years (IQR, 2.0 to 4.9), respectively (Table 3). Using VBC or DBC, the median time to SNG was significantly shorter in malignant nodules than in benign nodules (2.1 years [IQR, 1.6 to 4.4] vs. 3.1 years [IQR, 2.0 to 5.3], P=0.023). In contrast, when growth was assessed using the longest diameter alone, the median time to SNG did not differ significantly between benign and malignant nodules (3.0 years [IQR, 1.9 to 5.3] vs. 2.4 years [IQR, 2.0 to 4.5], P=0.358).
In low-suspicion nodules, SNG defined by VBC or DBC and by change in the longest diameter alone was not significantly associated with malignancy (odds ratio [OR], 1.19 and 1.21; 95% CI, 0.69 to 2.06 and 0.63 to 2.33, respectively) or neoplastic outcomes (OR, 1.05 and 1.15; 95% CI, 0.59 to 1.88 and 0.65 to 2.03, respectively) (Table 4). Similarly, among low-suspicion nodules measuring ≥2 cm, SNG defined by VBC or DBC and by change in the longest diameter alone was not significantly associated with malignancy (OR, 1.19 and 1.22; 95% CI, 0.64 to 2.18 and 0.67 to 2.21, respectively) or neoplastic outcomes (OR, 0.98 and 0.99; 95% CI, 0.51 to 1.90 and 0.52 to 1.89, respectively) (Table 5).
Discussion
In this study of low-suspicion thyroid nodules with long-term US follow-up, SNG, whether defined by volume, diameter, or longest diameter alone, was not associated with malignancy or thyroid neoplasm. Although more than half of the nodules exhibited significant growth, most were ultimately benign. In addition, the median time to significant growth among growing nodules was approximately 2.6 years, suggesting that rapid progression is uncommon. These findings support consideration of longer surveillance intervals and may help inform more evidence-based follow-up strategies for low-suspicion nodules.
Among low-suspicion nodules, VBC and DBC showed excellent concordance, supporting their interchangeable use in US surveillance. By contrast, agreement between these criteria and the longest-diameter criterion was substantial rather than perfect. Notably, 19.2% (37/193) of the nodules classified as showing significant growth by volume or diameter did not meet the growth threshold based on the longest diameter alone. For low-suspicion nodules, which are typically oval and isoechoic on ultrasonography and are usually managed conservatively because of their low baseline malignancy risk, assessment based solely on the longest diameter may be clinically practical, especially because volume- and multidiameter-based criteria are often cumbersome to apply in routine practice.
Several studies have reported that nodule growth does not reliably distinguish benign from malignant lesions [5,6,15], and the present findings are consistent with that literature. In this cohort, SNG was not significantly associated with malignancy or thyroid neoplasm, regardless of whether nodules were ≥2 cm. Furthermore, a recent study [16] of cytologically benign nodules found that nodules that remained stable during the first 3 years did not subsequently grow during later follow-up. Importantly, neither the presence nor the timing of growth was significantly associated with malignancy. In this cohort, the overall median growth rate was 1.02 mm/y, and malignant nodules showed a significantly higher median growth rate than benign nodules (1.57 mm/y). However, because a 2-mm increase is commonly used as the threshold for significant growth, and because US measurements are subject to interobserver variability and intrinsic measurement error, such small annual growth rates limit the clinical usefulness of growth-based assessment at annual follow-up examinations. In addition, when VBC or DBC was used, malignant nodules reached the significant-growth threshold earlier than benign nodules; however, even malignant nodules required more than 2 years to meet this criterion. Therefore, for nodules with benign imaging features, extending the follow-up interval to at least 2 years may be a practical approach. This observation may help optimize US surveillance protocols while reducing unnecessary imaging in patients with low-risk thyroid nodules.
The present study focused on low-suspicion nodules defined by US features rather than by cytological findings. Because of their sonographic features, low-suspicion nodules may include a relatively higher proportion of follicular neoplasms [17]. Nonetheless, Park et al. [18] reported that 47% of surgically resected FTCs were classified as K-TIRADS 3, the low-suspicion category used in the present study. In addition, Han et al. [19] reported a 53.2% malignancy rate on surgical pathology in circumscribed solid nodules lacking typical malignant sonographic features. Thus, focusing on surgically confirmed low-suspicion nodules is reasonable. Because growth did not differ significantly between benign and malignant nodules, surgical intervention could be considered for nodules exhibiting SNG within the first 2 years of follow-up, while extending the follow-up interval for stable nodules may help avoid unnecessary biopsies.
In the present study, low-suspicion nodules followed several diagnostic pathways. Nodules measuring ≥2 cm generally underwent FNA or CNB, and surgical excision was subsequently performed according to the biopsy findings. Accordingly, the higher malignancy rate observed in surgically treated nodules likely reflects differences in diagnostic pathways, as nodules diagnosed as malignant on repeat biopsy are typically referred for surgery. The relatively high surgical rate among low-suspicion nodules in the present cohort may likewise be attributable to biopsy-driven surgical decision-making, which may limit the generalizability of this study’s findings to the broader population of low-suspicion thyroid nodules. Prospective studies are needed to further validate these findings.
This study has several limitations. First, its retrospective design may have introduced selection bias. In addition, follow-up intervals were heterogeneous and were determined by routine clinical practice, and the median follow-up duration of 41 months may not fully reflect uniformly long-term observation. Nevertheless, inclusion of monitored low-suspicion nodules enabled longitudinal assessment of growth and definitive outcome determination. Second, even when US examinations are performed using the same equipment, intraobserver variability remains a concern, particularly when growth is assessed on the basis of changes as small as 1 mm or less. Furthermore, interobserver variability in US measurements of thyroid nodule diameter and volume has been reported to be approximately 13% and 7%, respectively [20]. These inherent measurement limitations may have affected the assessment of interval change. Third, because all measurements were performed retrospectively using saved US images, precise diameter assessment may have been limited. Nevertheless, in routine practice, radiologists are instructed to obtain transverse and longitudinal images that capture the maximal dimensions (length, height, and width) of each nodule, which may have reduced some of this variability. Fourth, nodules classified as benign solely on the basis of imaging stability without surgical confirmation may have included false-negative malignancies, potentially leading to underestimation of the true cancer rate in the nonsurgical group. Nevertheless, to the authors’ knowledge, this is the first study to evaluate the natural course of low-suspicion nodules and to compare growth between benign and malignant thyroid nodules.
Future studies are needed to validate these findings in broader and less selected populations. In particular, prospective studies using standardized follow-up protocols and uniform biopsy indications would help reduce selection bias and methodological heterogeneity. In addition, evaluation across a wider range of nodule sizes and inclusion of nodules without pathological confirmation may provide a more comprehensive understanding of growth dynamics and their clinical implications in low-suspicion nodules.
In conclusion, SNG over time, regardless of the measurement method used, was not associated with malignancy or thyroid neoplasm in low-suspicion thyroid nodules with cytopathologic confirmation. Given that the median time to SNG exceeded 2.6 years, the US follow-up interval for low-suspicion thyroid nodules may warrant cautious reconsideration.
Download Citation
PDF Links
PubReader
XML Download
Full text via DOI
