Integrating Ultrasound and Clinicopathologic Characteristics to Predict the Invasive Papillary Thyroid Carcinoma Among Indeterminate Thyroid Nodules

Introduction

Thyroid nodules are a common clinical finding, with detection rates having increased markedly over the past three decades.¹ Ultrasonography is the primary modality for the initial assessment of thyroid nodules.² Ultrasound-guided fine needle aspiration (FNA) cytology is a key method for further evaluating the benign or malignant nature of these nodules.³ FNA results are classified using The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC), which categorizes nodules into Bethesda classes I through VI. Specifically, Bethesda classes III, IV, and V are classified as indeterminate thyroid nodules (ITNs),⁴ with associated malignancy risks of 13–30%, 23–34%, and 67–83%, respectively.^4,5 Management options for these categories include repeat FNA, molecular testing, diagnostic lobectomy, total thyroidectomy, or surveillance, posing significant challenges for clinical decision-making.

When clinical evidence suggests a high suspicion of malignancy in ITNs, further intervention is warranted. Over 80% of malignant thyroid nodules are papillary thyroid carcinoma (PTC).⁶ Within PTC, a subset of low-risk variants may be managed with alternatives to lobectomy, such as active surveillance or minimally invasive ablation techniques.^7–10 These approaches aim to reduce surgical risks, avoid scarring, and mitigate overtreatment of indolent tumors.¹¹ In addition, patients with low-risk papillary thyroid microcarcinoma (PTMC) may experience decision regret after thyroidectomy,¹² which also provides important evidence for exploring more conservative management strategies. However, some PTCs exhibit aggressive features, including local or distant metastasis and extrathyroidal extension, which are associated with relatively poor prognoses.¹³ For such cases, lobectomy alone may constitute insufficient initial treatment, making more extensive total thyroidectomy the preferred option, often accompanied by central compartment lymph node dissection.^8,14

Current research on ITNs primarily focuses on assessing malignancy risk,^15–21 rather than stratifying surgical patients by risk. Existing risk stratification systems, such as the TI-RADS and ATA guidelines, are predominantly designed for preoperative malignancy risk assessment and have limited capacity to guide the extent of surgical intervention. Ultrasound remains the principal imaging modality for thyroid nodule diagnosis,²² and is a valuable examination tool,²³ with high-resolution ultrasound being standard for evaluating thyroid cancer.²

This study retrospectively analyzes patients with ITNs who underwent surgical resection and received a postoperative diagnosis of PTC. We aim to identify risk factors for invasive PTC and construct a predictive model based on available ultrasonic and clinicopathological characteristics. In contrast to existing systems that focus on malignancy detection, our model is specifically designed for preoperative risk stratification of invasiveness, aiming to complement current guidelines by assisting clinicians in determining the appropriate extent of surgery (eg, lobectomy vs. total thyroidectomy). This model is expected to offer precise individualized treatment plans, improving treatment efficacy while avoiding unnecessary surgery.

Materials and Methods

Study Design and Population

This retrospective study was approved by the Ethics Committee of Affiliated Hospital of Jiangsu University (Approval Number: SWYXLL20190225-2).

We retrospectively collected data from patients who underwent ultrasound-guided FNA at the Affiliated Hospital of Jiangsu University between March 2016 and October 2024. Inclusion required cytological results classified as Bethesda III, IV, or V, and a postoperative pathological diagnosis of PTC. Clinical, pathological, and preoperative thyroid ultrasound data were collected. Based on postoperative pathology as the gold standard, patients were categorized into invasive PTC and non-invasive PTC groups.

Invasive PTC was defined as exhibiting extrathyroidal extension (ETE) and/or local lymph node metastasis. ETE was defined according to the 8th edition of the American Joint Committee on Cancer (AJCC) Cancer Staging Manual as tumor invasion into surrounding structures, including strap muscles, larynx, trachea, esophagus, recurrent laryngeal nerve, prevertebral fascia, or encasement of the carotid artery or mediastinal vessels.²⁴

Inclusion Criteria

(a) Age ≥18 years;

(b) Preoperative routine thyroid ultrasound examination performed;

(c) US-FNA cytology classified as Bethesda III, IV, or V followed by surgical resection and postoperative pathological confirmation of PTC;

(d) Availability of complete preoperative clinical records.

Exclusion Criteria

(a) Family history of thyroid cancer;

(b) History of any malignancy;

(c) History of head and neck radiotherapy;

(d) Postoperative pathology indicating a benign lesion or a thyroid malignancy other than PTC;

(e) Incomplete thyroid function laboratory data.

After applying these criteria, 494 patients were included in the final analysis (Figure 1).

Figure 1 Patient Selection Flowchart.

Ultrasound (US) Evaluation

All thyroid nodules underwent ultrasound assessment. Two senior radiologists, each with over 10 years of experience in thyroid imaging, independently evaluated the US features. Any discrepancies between their assessments were resolved by re-evaluation and consensus with a third experienced radiologist.

The US evaluation was performed according to the 2023 Korean Thyroid Radiology Society (K-TRADS) consensus²⁵ and the 2017 American College of Radiology Thyroid Imaging Reporting and Data System (ACR TI-RADS) guidelines.²⁶ The following features were assessed: (a) echogenicity; (b) aspect ratio; (c) calcification; (d) location; (e) vascularity; (f) distance to capsule; (g) maximum nodule diameter.

Clinicopathologic Evaluation

Thyroid function test results were interpreted based on the reference ranges applicable at the time of testing. Values within the reference range were considered normal, while those outside the range were considered abnormal. Histopathological evaluation followed a two-step review process. Initial diagnosis was performed by a pathologist with at least five years of experience. All cases were then reviewed and the final pathological diagnosis was confirmed by a second, senior pathologist with over ten years of experience. The final diagnosis of ITNs, serving as the gold standard, was established based on these postoperative pathological findings.

Statistical Analysis

Potential risk factors for invasive PTC were first identified using univariate logistic regression analysis. Variables showing a significant association (P < 0.05) in univariate analysis were then included in a multivariate logistic regression model to identify independent predictors.

A predictive model for invasive PTC was developed based on the results of the multivariate analysis. The discriminative ability of the model was evaluated using receiver operating characteristic (ROC) curve analysis, with the area under the curve (AUC) calculated. The clinical utility of the model was assessed using decision curve analysis (DCA). A nomogram was constructed to visualize the final predictive model.

All statistical analyses were performed using SPSS software (version 25.0, IBM Corp., Armonk, NY, USA) and R software (version 4.4.2, R Foundation for Statistical Computing, Vienna, Austria). A p-value < 0.05 was considered statistically significant.

Results

Clinical and Pathological Data and Statistical Analysis

A total of 494 patients were included in the final analysis. The cohort comprised 366 females (74.1%) and 128 males (25.9%), with a mean age of 46.40 ± 11.96 years. Overall, 262 patients (53.0%) were >45 years old and 232 (47.0%) were ≤45 years old (Table 1).

Table 1 Clinical and Pathological Characteristics

Based on postoperative pathology, 353 patients (71.5%) were diagnosed with non-invasive PTC and 141 patients (28.5%) with invasive PTC.

Univariate analysis revealed that male, age ≤45 years, presence of microcalcifications, abundant vascularity, distance from the thyroid capsule ≤2 mm, larger maximum nodule diameter, FT4 levels, and TgAb levels were significantly associated with invasive PTC (all p<0.05). In contrast, nodule location, FT3 levels, TSH levels, and Tg levels showed no significant association (all p>0.05) (Table 2).

Table 2 Results of Univariate and Multivariate Analysis

Multivariate logistic regression analysis identified the following independent predictive risk factors for invasive PTC: male (OR = 2.91, 95% CI: 1.78–4.76, P < 0.001), age ≤45 years (OR = 1.93, 95% CI: 1.23–3.03, P = 0.04), abundant vascularity (OR = 4.60, 95% CI: 2.42–8.75, P < 0.001), distance from the capsule ≤2 mm (OR = 3.63, 95% CI: 2.15–6.14, P < 0.001), and abnormal TgAb levels (OR = 0.38, 95% CI: 0.18–0.77, P = 0.008) (Table 2).

Development and Validation of the Clinical Prediction Model

The 494 patients were randomly divided into a training set (80%) and a validation set (20%) using a computer-generated random sequence. The clinical prediction model for invasive PTC was developed in the training set based on the independent predictors identified by multivariate logistic regression analysis. The model’s performance was then evaluated in the validation set.

Receiver operating characteristic (ROC) curve analysis demonstrated moderate discriminative ability of the model. The area under the curve (AUC) was 0.776 (95% CI: 0.728–0.825) for the training set and 0.759 (95% CI: 0.643–0.874) for the validation set (Figure 2).

Figure 2 (a) The receiver operating characteristics (ROC) curve and area under the curve (AUC) in the training set; (b) The ROC curve and AUC in the test set.

Decision curve analysis (DCA) further assessed the clinical utility of the model across a range of threshold probabilities. The DCA curves for both the training and validation sets indicated that the model provides a good net benefit compared to treating all or no patients, supporting its potential clinical usefulness (Figure 3).

Figure 3 (a) Decision curve analysis in the training set; (b) Decision curve analysis in the test set.

Construction of the Nomogram for Invasive PTC

Multivariate logistic regression analysis confirmed that male, age ≤45 years, abundant vascularity, distance from the thyroid capsule ≤2 mm, and abnormal TgAb levels were independent predictors of invasive PTC in patients with ITNs. Based on these significant predictors, a static nomogram was constructed to provide a visual representation of the predictive model, facilitating the estimation of individual risk (Figure 4).

Figure 4 The nomogram for predicting invasive PTC in ITNs.

Additionally, to enhance accessibility and ease of use, an interactive online dynamic nomogram has been developed and is publicly accessible at: https://indeterminate-345.shinyapps.io/nomogram/.

Figure 5 illustrates a representative example of using the nomogram for risk prediction in a patient with indeterminate thyroid nodules, alongside the corresponding postoperative pathological findings confirming invasive PTC.

Figure 5 Dynamic nomogram prediction example for an indeterminate thyroid nodule. (a) Ultrasound image showing the thyroid nodule (0.70 cm × 0.62 cm) located≤ 2mm from the capsule. (b) Ultrasound image indicating abundant vascularity in the thyroid nodule. (c) The web-based dynamic nomogram predicts a risk score of 0.838 for invasive PTC in this patient. (d) Pathological findings confirm the nodule as PTC, with cancerous tissue invading vascular walls and surrounding thyroid soft tissue.

Discussion

The indeterminate cytological diagnosis of thyroid nodules by FNA presents a significant clinical dilemma in therapeutic decision-making. While conservative management may be appropriate for non-invasive papillary thyroid carcinoma (PTC) in indeterminate thyroid nodules (ITNs),^27,28 invasive PTC necessitates more extensive surgical resection. This study demonstrates that readily accessible ultrasonographic features combined with clinicopathological characteristics can effectively predict invasive PTC risk in ITNs, enabling clinicians to formulate individualized treatment strategies.

Consistent with extensive evidence,^29–33 our study confirms younger age (≤45 years) and male as independent predictors of invasive PTC, demonstrating 1.93-fold and 2.91-fold elevated risks respectively. This aligns with a meta-analysis of 27,741 patients across 41 studies.³⁴ Notably, while females exhibit higher overall PTC incidence, males face disproportionate invasive disease risks.^35–37 Potential mechanisms for this sex disparity include: androgenic promotion of tumor invasiveness,³⁷ increased environmental carcinogen exposure (eg, radiation/occupational hazards), higher prevalence of smoking/alcohol consumption, and potential diagnostic delays due to reduced healthcare engagement. For younger patients (≤45 years), accelerated tumor progression may stem from heightened metabolic activity promoting cancer cell proliferation, coupled with less effective immune surveillance against microscopic invasion.

Our analysis further establishes abundant vascularity and capsule proximity (≤2 mm) as novel independent predictors of invasive PTC in indeterminate thyroid nodules. This finding corroborates existing evidence of an inverse relationship between capsular distance and invasion risk.³⁸ Mechanistically, the thyroid capsule serves as a critical anatomical barrier against tumor spread. Nodules within ≤2 mm proximity promote extracapsular extension through both direct breaching of the capsular boundary and compression-induced invasive growth in confined microenvironments.

Concurrently, tumor neovascularization indicated by abundant vascularity promotes invasiveness via dual mechanisms: enhanced oxygen/nutrient delivery accelerating local proliferation, and providing hematogenous dissemination routes to cervical lymph nodes and distant sites (eg, lungs/bones). These vascular features signify biologically aggressive phenotypes.

The observed protective association of abnormal thyroglobulin antibody (TgAb) levels against invasive PTC in ITNs presents a novel finding. While elevated TgAb typically correlates with autoimmune thyroid disorders and has been linked to increased invasion risk in conventional PTC,^39,40 our data paradoxically demonstrate reduced aggressiveness in ITN contexts. This may reflect enhanced immune surveillance wherein TgAb activates antitumor responses that eliminate early malignant cells before invasion develops. Additionally, patients with TgAb abnormalities typically undergo more intensive thyroid monitoring, potentially enabling earlier intervention during pre-invasive stages. The differential expression patterns and immunological mechanisms of TgAb in ITNs warrant prospective investigation.

Regarding nodule size, while prior studies associate larger nodules (≥1 cm) and microcalcifications with invasiveness,^41–44 our multivariate analysis failed to establish these as independent predictors. The persistent association in univariate analysis suggests these features may serve as secondary indicators meriting consideration in comprehensive risk assessment.

This study has several limitations. First, its single-center retrospective design may introduce selection bias, and the moderate sample size warrants validation in larger prospective cohorts. Second, while internal validation demonstrated model efficacy, external validation across diverse populations is needed to confirm generalizability. Future multi-center studies should address this gap. Third, molecular testing was not systematically incorporated; while this reflects real-world scenarios where such testing may be cost-prohibitive or clinically unnecessary prior to surgery,⁴⁵ integrating molecular markers could enhance future models. Fourth, we acknowledge that the extended study period (2016–2024) may introduce potential temporal biases, and we recommend that future prospective studies with standardized protocols be conducted to validate our findings. Finally, the ultrasound protocol utilized conventional B-mode and Doppler imaging without advanced techniques like contrast-enhanced or elastographic ultrasound, which may provide additional predictive information.

Conclusion

We established and validated an integrated clinical-ultrasonic model incorporating five key predictors of invasive PTC in indeterminate thyroid nodules: male, age ≤45 years, abundant vascularity, capsule proximity (≤2 mm), and elevated thyroglobulin antibodies. This readily accessible tool provides individualized risk stratification to guide clinical decision-making, and is intended for preoperative risk assessment to assist in determining the extent of surgery, rather than as a substitute for pathological diagnosis. This model has the potential to reduce unnecessary aggressive surgery in low-risk cases while ensuring appropriate management of high-risk patients. The online nomogram implementation facilitates point-of-care application (https://indeterminate-345.shinyapps.io/nomogram/).