Development and validation of a nomogram based on clinicopathological characteristics and multimodal ultrasound parameters for predicting lateral lymph node metastasis in papillary thyroid carcinoma

Introduction

In the past decade, the detection rate of papillary thyroid carcinoma (PTC) increased year by year, while the total mortality rate had not increased significantly. The debate on whether PTC is overdiagnosed and overtreated has become increasingly heated (1). Previous studies showed that approximately 20% to 67% of PTC patients were diagnosed with lateral cervical lymph node metastasis (LLNM) at the time of initial diagnosis (2,3). Some PTC patients even presented with contralateral LLNM relative to the primary tumor site, which, if undetected, could result in incomplete surgical dissection (4). Lymph node metastasis (LNM), especially LLNM, is an important risk factor for PTC recurrence, distant metastasis and reduced survival rate (2-5). However, compared with central lymph node dissection (CLND), the incidence of postoperative complications after lateral cervical lymph node dissection (LLND) is much higher than that without dissection (6). The American Thyroid Association (ATA) recommends that surgeons should not prophylactically dissect lateral lymph nodes (LLNs) unless there is clear evidence of metastasis, such as preoperative imaging and fine needle aspiration biopsy (FNAB) confirming suspected LLNM (7-9). Nevertheless, the false negative rate of LLNM cannot be ignored, as the accuracy of preoperative detection for LLNM depends largely on the experience of ultrasound physicians, radiologists and pathologists. As the currently preferred imaging methods for preoperative assessment of cervical lymph node status, conventional ultrasound and computed tomography are of limited value in assessing LLNM, with high specificity (85.0–97.4%), but low sensitivity (36.7–61.0%) (10). Similarly, invasive diagnostic techniques such as FNAB and thyroglobulin measurements in washout of FNAB (FNAB-Tg) can be used to clarify the status of lymph nodes, but also demonstrated a clinically significant false negative rate in the diagnosis of LLNM (10-12). These diagnostic limitations may consequently contribute to rapid postoperative relapse in patients who have undergone thyroidectomy alone (13), which could affect the timing and quality of treatment (9). Therefore, improving the accuracy of preoperative diagnosis for lateral cervical lymph nodes is very important.

In recent years, superb micro-vascular imaging (SMI) and strain ultrasound elastography (SUE) are additional tools for conventional ultrasound that can be used to evaluate microvessels and tissue rigidity, which play an important role in the differential diagnosis of benign and malignant thyroid nodules (14,15). Nevertheless, despite their clinical potential, these emerging medical imaging technologies are rarely integrated into diagnostic models for predicting LLNM. Most of the prediction models extensively used in clinical practice are based on clinical and pathological characteristics. However, the diagnostic accuracy of these models in detecting LLNM still remains suboptimal. Therefore, in this study, we established a nomogram model according to the clinicopathological and ultrasonic parameters, including the perforator vessels shown by SMI and the tissue hardness shown by SUE, to explore the independent risk factors for LLNM in PTC patients. We present this article in accordance with the TRIPOD reporting checklist (available at https://gs.amegroups.com/article/view/10.21037/gs-2024-525/rc).

Methods

Study subjects

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Ruijin Hospital Luwan Branch, Shanghai Jiao Tong University School of Medicine (approval code: IRB-2022-006) and individual consent for this retrospective analysis was waived. Patients who underwent thyroid ultrasound examinations and surgery at Ruijin Hospital Luwan Branch, Shanghai Jiao Tong University School of Medicine from January 2019 to February 2024 were included. Inclusion criteria: (I) adult patients (age ≥18 years old) with complete preoperative ultrasound data, and postoperative pathology was confirmed as PTC; (II) all cases had no history of previous thyroid surgery, medical treatment, I-131 treatment or radiotherapy; (III) thyroid surgery was completed within a short period of time after receiving ultrasound examination. Exclusion criteria: (I) the maximum diameter of the target nodule was >4.0 cm; (II) cases with suboptimal ultrasound image quality; (III) combined with other types of malignant tumors, such as follicular thyroid carcinoma, medullary carcinoma, undifferentiated carcinoma, or thyroid metastasis from other organs. Finally, 703 PTC patients were included, and the screening process of the study subjects is shown in Figure 1. All patients were stratified into a training cohort (494 cases) and a validation cohort (209 cases) according to the chronological sequence of surgery with a ratio of 7:3. Patients of two cohorts were then stratified into LLNM(+) group (pN1) and LLNM(−) group (pN0), respectively. LLNM(+) was confirmed by postoperative pathology, while LLNM(−) comprised: (I) cases were diagnosed with non-LLNM by postoperative pathology; (II) cases were diagnosed with non-LLNM by FNAB, and LLND was not performed; (III) cases with no suspicious lymph nodes on ultrasound, and FNAB or LLND was not performed.

Figure 1 Flowchart showing the patient selection process in this study. PTC, papillary thyroid carcinoma.

Instruments and methods

Toshiba Aplio500 ultrasound machine (Tokyo, Japan) with frequency of 5–14 MHz probe, equipped with SMI and SUE software was used. Conventional ultrasound was used to assess the tumor size, echogenicity, margin, microcalcification, aspect ratio (defined as longitudinal diameter/transverse diameter >1), capsule invasion, Hashimoto’s thyroiditis (HT), and cervical lymph nodes. Conventional color Doppler flow imaging (CDFI) was used to assess the blood flow of primary thyroid nodule. SMI imaging mode was used to observe the presence of perforating vessels. SUE was activated to measure the hardness of the primary thyroid nodule and the strain ratio with the surrounding tissue.

In this study, the Chinese Thyroid Imaging Reporting and Data System (C-TIRADS) (16) was used to stratify the risk of target thyroid nodules. For multifocal nodules, the nodule with the highest C-TIRADS grade was selected. Capsule invasion was defined as the edge of the suspicious nodule in contact with the thyroid capsule by more than 25% on conventional ultrasound (17). The perforating vessels were defined as tiny blood vessels extending from the outside of the nodule to the inside on SMI images (18). The diagnosis of HT was assessed by ultrasound, and it was manifested as uneven echogenicity of the thyroid parenchyma, with a few or multiple lamellar hypoechoic areas showing grid-like changes (17). According to the 2015 ATA guidelines (8), lymph nodes were considered clinically suspicious for metastases when at least one of the five criteria was met: (I) focal or diffuse hyperechogenicity, (II) microcalcifications or macrocalcifications, (III) cystic changes, (IV) abnormal vascular pattern (chaotic or peripheral vascular pattern), or (V) rounded (length/transverse diameter ratio <1.5).

After the primary thyroid nodule was assessed on grayscale ultrasound, the SUE mode was initiated. The size of the sampling frame was adjusted to include as much as possible of both tumor and surrounding thyroid parenchyma (STP), and a circular region of interest (ROI) was placed inside the tumor, which was regarded as the target area (S-T1). For mixed cystic and solid nodules, the ROI was placed over the solid part of the lesion. The transducer was used to apply periodic external compression while SUE images of the tumor were acquired. Then, a similar ROI was placed in the STP at the same depth as the ROI in the tumor (S-R). Strain rate value (SRR, SRR = S-R/S-T1) of the tumor was automatically calculated. After the SMI function was enabled, the sample frame was adjusted to be slightly larger than the nodule to avoid pressurizing the nodule, and the blood flow gain was adjusted appropriately to just show the tiny blood vessels without overflow. The static and dynamic images were stored and imported into the hard disk for further analysis. The relevant features of thyroid nodules were independently assessed by two radiologists with more than 5 years of experience in thyroid ultrasound without knowledge of the patient’s pathological results. All datas were collected from the Medical Record System, including sex, age, results of thyroid function tests, HT, tumor size, and ultrasound characteristics such as aspect ratio, hypoechoic, multifocality, capsular invasion, microcalcification, perforator vessel, S-T1, S-R, SRR, the number of central lymph node metastasis (CLNM) and ratio of metastatic lymph nodes (LNR). LNR was calculated as the ratio of CLNM to the number of CLND (9,19).

Surgical method

In this study, patients diagnosed with PTC by FNAB or intraoperative rapid frozen section underwent unilateral lobectomy with or without isthmus resection and ipsilateral prophylactic CLND, while patients who underwent total thyroidectomy underwent bilateral CLND. The PTC patients clinically suspected of LLNM additionally received ipsilateral LLND. These patients comprised: (I) FNAB-confirmed LLNM cases; (II) cases were diagnosed with non-LLNM by FNAB and even without FNAB, but ipsilateral LLND was performed due to suspicious ultrasound findings, extreme anxiety or other individualized considerations.

Study design

In order to avoid the influence of confounding factors, the above variables in the training cohort were filtered by the least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation. The variables with non-zero coefficients were integrated and included in the backward stepwise multivariate logistic regression analysis of the Akaike information criterion (AIC) to screen out independent predictors of LLNM and the prediction model was constructed. The receiver operating characteristic (ROC) curve of the training cohort and the validation cohort were drawn, and the area under the curve (AUC) was calculated to evaluate the discrimination of the model. The Delong test was used to compare the differences in AUC of different models. The accuracy of the optimal cutoff value was estimated by calculating the following parameters: sensitivity, specificity, predictive value, and likelihood ratio. Finally, the clinical utility of the nomogram was determined by quantifying the net benefit of combining different threshold probabilities in the training and validation cohorts using decision curve analysis (DCA) and clinical impact curve (CIC).

Sample size estimation

The estimation was performed using R software (version 4.3.2, https://www.r-project.org/). “Type” of this study was set as “binary”, abbreviated as “b”; “parameters” was calculated to be 20 in accordance with the number of our independent variables; “shrinkage” refers to the shrinkage rate of model coefficients, which is usually set to 0.9. The c-statistic was calculated to be 0.864, and the prevalence was calculated to be 0.218 in accordance with the study by Feng (19). The sample size was automatically estimated to be 528.

Statistical analysis

Statistical analysis was performed using R software (version 4.3.2, https://www.r-project.org/). The Shapiro-Wilk test was performed for the quantitative data. The data that met the normal distribution were expressed as mean (standard deviation), and those that did not meet the normal distribution were expressed as the median (interquartile range). The count data were expressed as frequency (n) and percentage (%), and the Chi-square test was used to compare the baseline characteristics between the training cohort and the validation cohort. A two-sided P value <0.05 was considered statistically significant.

Results

Patient characteristics

This study retrospectively analyzed 703 PTC patients, including 191 (27.2%) males and 512 (72.8%) females; the age ranged from 19 to 81 years, with a median age of 43.0 years. Among these PTC patients, 448 patients underwent lobectomy with or without isthmus resection, while the remaining 255 received total or near total thyroidectomy. Based on the preoperative imaging findings, all patients underwent either prophylactic or therapeutic CLND. Among them, 121 patients additionally received LLND, and a total of 98 patients (13.9%) were diagnosed with LLNM by postoperative pathology, including 58 (59.2%) females and 40 (40.8%) males (P=0.002). In the training cohort, there were 67 (13.6%) and 427 (86.4%) cases in the LLNM(+) group and the LLNM(−) group, respectively. In the validation cohort, there were 31 (14.8%) and 178 (85.2%) cases in the LLNM(+) group and the LLNM(−) group, respectively. There was no significant difference in LLNM between the two cohorts (P=0.75). In addition, except for elastic S-R, there were no significant differences in the rest of characteristics between the LLNM(+) group and the LLNM(−) group in the two cohorts, indicating that there was no grouping bias in this study. The clinical and ultrasound characteristics of the patients in the training cohort (494 cases) and the validation cohort (209 cases) are summarized in Table 1.

Development of prediction model

Among the 20 potential risk factors, 10 factors with non-zero coefficients were finally screened by LASSO regression analysis, including sex, age, tumor size, multifocality, capsular invasion, microcalcification, perforator vessel, number of CLNMs, LNR, and SRR (Figure 2). The 10 risk factors with non-zero coefficients were included in the logistic regression multivariate analysis, and 8 independent risk variables including sex, tumor size, multifocality, capsular invasion, microcalcification, perforator vessel, LNR and SRR were screened to establish a nomogram prediction model for LLNM (Table 2, Figure 3).

Figure 2 Selection of variables using the LASSO algorithm in the training set. (A) The LASSO coefficient profiles of the 20 features. The coefficient profiles were plotted as a function of the log (λ) sequence. (B) A vertical line indicates the λ value selected via 5-fold cross-validation, which resulted in 10 features with non-zero coefficients. The selection of risk factors for LLNM in patients with PTC using LASSO regression. The binomial deviance curve is plotted against the log (λ) sequence. LASSO, least absolute shrinkage and selection operator; LLNM, lateral cervical lymph node metastasis; PTC, papillary thyroid carcinoma.

Figure 3 A nomogram for preoperative prediction of LLNM risk in PTC. LLNM, lateral cervical lymph node metastasis; LNR, ratio of metastatic central lymph nodes; PTC, papillary thyroid carcinoma; SRR, strain rate ratio.

Validation of the LLNM prediction model

The prediction model was internally validated by the Bootstrap resampling method to test the discrimination and draw the ROC curve (Figure 4A,4B). In the training cohort, the AUC of the model was 0.895 [95% confidence interval (CI): 0.854–0.936], while the AUC of the control group model without perforator vessel and SRR was 0.860, and the difference was statistically significant (P=0.003). In the validation cohort, the AUC of the model was 0.866 (95% CI: 0.816–0.917), while the AUC of the control group model without perforator vessel and SRR was 0.787, and the difference was statistically significant (P=0.04). The Hosmer-Lemeshow test was performed and the calibration curve was drawn. The calibration curves were close to the ideal curve in both the training cohort and the validation cohort (Figure 4C,4D). The optimal cutoff value (0.149) of the prediction model was determined by ROC curve analysis, demonstrating a sensitivity of 0.806, a specificity of 0.845, a positive predictive value (PPV) of 0.450, and a negative predictive value (NPV) of 0.965 in the training cohort. DCA curves (Figure 5A,5B) and CIC (Figure 5C,5D) were generated for both the training cohort and validation cohort to evaluate the clinical utility of the nomogram. In the training cohort, the DCA curve indicated a net benefit when the threshold probability exceeded 0.07 (Figure 5A), while in the validation cohort, the net benefit was observed at the threshold probability above 0.10 (Figure 5B).

Figure 4 ROC and calibration curves of the prediction models of the training cohort and validation cohort. (A) Training cohort and (B) validation cohort display the ROC curves. (C) Training cohort and (D) validation cohort present the calibration curves. AUC, area under the curve; ROC, receiver operating characteristic.

Figure 5 DCA and CIC of the prediction models of the training cohort and validation cohort. (A,B) The DCA of the prediction models for the training cohort and validation cohort, respectively. The curve of “All” means that all patients are assumed to have lymph node metastasis. The blue Fit1 and model1 represent the models constructed in this study with perforator vessels and SRR included. The red fit2 and model2 are comparison models without perforator vessels and SRR. The curve of “None” means that no patient is assumed to have CLNM. (C,D) The CIC of prediction model in training cohort and validation cohort, respectively. The graph depicts the number of patients (out of 1,000) identified as LLNM positive at varying threshold probabilities, with the red curve representing the assumed positive cases, the blue curve indicating the true positive cases and the boundary lines curve representing 95% CI. CI, confidence interval; CIC, clinical impact curve; CLNM, central cervical lymph node metastasis; DCA, decision curve analysis; LLNM, lateral cervical lymph node metastasis; SRR, strain rate ratio.

Discussion

LLNM is the most important risk factor for local recurrence and distant metastasis of PTC, and its associated mortality rate is three times higher than that of LLNM-negative patients (15). Undoubtedly, LLNM should be treated surgically as soon as possible. The ATA guidelines (8) recommended FNAB and FNAB-Tg for suspicious lymph nodes shown by imaging examinations, and LLND for patients with biopsy-proven LLNM (5). Unfortunately, the accuracy of preoperative ultrasound for pathological cervical lymph node is not ideal (15). A previous study showed that the incidence of occult LLNM was 18.6% to 64% (20). Although more advanced diagnostic tools are now available, LLNM is still difficult to predict (5). In clinical practice, which patients with PTC should undergo LLND remains a controversial topic. Therefore, it is necessary to further explore more effective methods to predict the incidence of LLNM in PTC patients (10). To the best of our knowledge, there is currently no unified imaging standard and consensus for predicting the risk of PTC LLNM. Accordingly, this study analyzed the risk factors for LLNM in PTC by combining conventional ultrasound with emerging modalities such as SMI and SUE and then constructed a prediction model to assist clinicians in making treatment decisions.

Previous studies have found that women are more likely to develop PTC (21,22), but male PTC patients are more likely to develop LLNM than female patients (10,23), which is consistent with the results in our study. This may be related to sex hormones and the higher basal metabolic rate of men for promoting cell division (23,24). In addition, the results of this study showed that multifocality and capsular invasion were also risk factors for LLNM, which is consistent with previous studies (17,25). Another independent risk factor for LLNM in the study was tumor size. Patients with larger tumors are more likely to develop LLNM. As the tumor size increased, the cancer cell multiplication rate accelerated, and the metastasis rate increased accordingly (21,24). Previous studies showed that the combination of HT was negatively correlated with LLNM (26); however, it was not confirmed in this study. The reason may be that potential factors such as thyroid drug treatment and immune status of different patients could affect the results.

LNM of PTC usually follows a specific pattern. According to the lymphatic drainage pathway, it first spreads to the central lymph nodes and then involves the lateral areas. CLNM was selected as an important risk factor in the LLNM prediction model in previous studies. Previous studies showed that the number of CLNM and LNR may affect the results of skip metastasis and LLNM (5,13,19,20,27). An increase in the number of CLNM tended to promote regional recurrence and was associated with the median and high stratification of recurrence risk (20). Therefore, this study selected the number of CLNM and LNR rather than just the presence of metastases. We found that the number of CLNM and the LNR in the prediction model had a significant impact on the dependent variable, which was consistent with the results of previous studies (12,13,19,20).

Jiang et al. reported that extracellular matrix cross-linking was an important component of cancer cell biology and associated with tissue fibrosis, affecting tumor progression by regulating soluble factors that can trigger inflammation and angiogenesis and induce cell growth and invasion (15). Previous studies showed a correlation between the hardness of PTC and its invasiveness; the higher SRR indicated a harder PTC, and higher hardness was also associated with a greater probability of CLNM (21,28). However, to the best of our knowledge, the relationship between thyroid cancer hardness and LLNM has not been well documented in the literature. In this study, the SRR in the LLNM(+) group was significantly higher than that in the LLNM(−) group, and multivariate logistic regression model showed that it was an independent predictor of LLNM. Our results were similar to those of Liu et al. (21), who reported that the SUE elasticity score was independently associated with LLNM. They used semiquantitative parameters, whereas quantitative parameter SRR was used in our study, which was more objective. Xu et al. (28) reported that after adjusting for other factors through multivariate logistic analysis, conventional tumor hardness grading ST1 and SR were not helpful in predicting LNM, which may be related to the fact that ST1 and SR only revealed the hardness of the tumor itself, while SRR reflected the relative hardness of the target tumor relative to the adjacent thyroid parenchyma (28). Compared with SRR, other SUE parameters were relatively susceptible to factors such as tumor size, depth, pre-compression, neck morphology, and fibrosis caused by previous cervical surgery (21).

The importance of neovascularization in the progression and metastasis of malignant tumors has been confirmed in several studies (18,29,30). Compared with normal tissues, cancer cells required more oxygen and nutrients due to their abnormally active metabolism, resulting in relative hypoxia and nutrient deficiency, which in turn stimulated the formation of more microvessels extending from normal tissues into the tumor to maintain the metabolic homeostasis of the tumor, that were, perforator vessels (18). Stabenow et al. (31) found that metastatic thyroid cancer had a higher immunohistochemical expression of vascular endothelial growth factor. Cancer cells stimulated the formation of irregular and outward-extending microvessels inside the tumor by secreting a variety of vasoactive substances such as vascular endothelial growth factor. These microvessels often grew in a disordered and uncontrollable manner, which could explain the higher invasiveness of tumors with perforating vessels inside (31). Although a large number of literature has confirmed that tumor progression and prognosis were closely related to new microvessels, there was still controversy about using the internal vascular conditions of the tumor as a basis for judging the prognosis of PTC. The low sensitivity of blood flow detection technology may be one of the reasons for this disagreement (Figure 6). Conventional CDFI can hardly show microvessels with a diameter less than 0.1 mm or low-speed blood flow with a velocity less than 10 mm/s (Figure 6A). Contrast-enhanced ultrasound (CEUS) can clearly show the microcirculation of nodules through contrast agent microbubbles, however, injections of additional contrast agents were needed. SMI technology effectively separated blood flow signals and motion artifacts through adaptive algorithms and wall filters, and could show microvessels and low-speed blood flow without relying on contrast agents (Figure 6B,6C). Zhao et al. discovered that the effectiveness of SMI in showing micro-perforating blood vessels in tumors was highly consistent with that of CEUS (30). There have been many studies on the value of SMI and SUE in distinguishing benign and malignant tumors such as breast and thyroid, but the value in predicting LLNM in PTC patients remains to be verified.

Figure 6 A 9-mm papillary thyroid carcinoma (the green arrow) in a 40-year-old man. (A) CDFI demonstrates a few dot-like flow signals inside and around the nodule. (B,C) SMI demonstrates perforator vessels extending from the outside of the nodule to the inside. (D) Ultrasound elastography demonstrates the hardness distribution of tumor tissue (S-T1) and surrounding thyroid parenchyma (S-R) in a sampling box. CDFI, color Doppler flow imaging; S-T1, strain of tumor tissue; S-R, strain of surrounding tissue; SMI, superb microvascular imaging.

A nomogram graphically depicts the relationship between a specific disease and its risk factors quantitatively to predict the probability of clinical events. In recent years, it has been extensively used in oncology research. Previous studies have analyzed various risk parameters and constructed nomogram models to predict the risk of LLNM in PTC patients. Liu et al. developed a dual-center nomogram prediction model for CLNM and LLNM based on clinical pathological characteristics and screened the risk factors for cervical LNM (10). However, the AUC value of their prediction model for LLNM was 0.706 (sensitivity: 0.711, specificity: 0.630), while it was 0.895 (sensitivity: 0.806, specificity: 0.845) in our study. This difference may be attributed to the fact that their study failed to include ultrasound-related imaging parameters, which have been confirmed to be crucial for oncology research. Jiang et al. (23) predicted cervical LNM in PTC patients based on the CEUS-based nomogram and achieved a certain degree of discrimination effect. However, the sample size was relatively small. The construction of the prediction model required adequate samples as training cohorts to improve accuracy and reduce errors. Therefore, its practical value has yet to be verified and cannot be used as high-quality evidence for evidence-based medical research. The prediction model in this study was constructed based on the clinical information and above-mentioned multimodal ultrasound parameters, which included anatomical and biological information of different aspects of the tumor.

We applied LASSO regression to filter risk factors, and risk factors with non-zero regression coefficients were integrated into multivariate logistic regression analysis. The LASSO algorithm is a variable selection method for fitting high-dimensional generalized linear models based on constructing a penalty coefficient (λ) and compressing a variable set (15). Finally, eight independent risk factors were screened out to establish a nomogram prediction model. The ROC curves were drawn, and the AUC values of the model were calculated to evaluate the discrimination of the model. This study used the Hosmer-Lemeshow goodness of fit test to test the accuracy of the model and drew calibration curves. The results demonstrated that the AUC values of the prediction model were 0.895 and 0.866 in the training cohort and validation cohort, respectively, indicating that the model had a high predictive efficiency, while in the comparison group, the AUC values of the prediction model without perforator vessels and SRR were 0.860 and 0.787, respectively. The differences in the AUC values of the two models in the two cohorts were statistically significant (P<0.05 for all), suggesting that additional SMI and SUE indicators improved the diagnostic performance of the model. The calibration curves and the ideal curves fit well in both the training cohort and the validation cohort, which indicated that the prediction results of the model were consistent with the actual incidence of LLNM. DCA revealed that net benefit would be achieved when the threshold probability of the prediction model is above 7% (training cohort) and 10% (validation cohort), which suggested that over 90% of PTC patients will benefit from the prediction model. Compared with the prediction model containing only independent clinical risk factors, the utilization of SUE and SMI significantly improved the predictive capability of the LLNM prediction model. As shown in the CIC, excellent prediction performance and significant clinical net benefit were obtained.

There were also some limitations in this study. First, only subjects with suspected LLNM were subjected to LLND by clinical or imaging examinations, and patients without LLND were considered negative; patients with occult LLNM were neglected. Second, for multifocal tumors, only the nodule with the highest C-TIRADS grade was analyzed, and the characteristics of other tumors were not analyzed. It was not verified which suspicious nodule caused LLNM. Third, on SUE images, the hardness of PTC was not uniform, especially for larger PTCs. We selected the hardest area on the SUE image for measurement, and the location of the ROI may affect the results. Finally, no follow-up information such as surgical complications and LLNM recurrence rate was provided.

Conclusions

In conclusion, SMI and SUE demonstrated favorable predictive performance and clinical utility for LLNM in PTC patients, particularly when combined with conventional ultrasound parameters. This nomogram incorporating the aforementioned ultrasound parameters might be used to make accurate preoperative risk stratification of LLNM, which could help clinicians in formulating individualized treatment strategies.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://gs.amegroups.com/article/view/10.21037/gs-2024-525/rc

Data Sharing Statement: Available at https://gs.amegroups.com/article/view/10.21037/gs-2024-525/dss

Peer Review File: Available at https://gs.amegroups.com/article/view/10.21037/gs-2024-525/prf

Funding: This work was supported by grants from Putian University Research Foundation (Grant No. 2023096); Huangpu District Medical Specialty Construction Project in Shanghai (Grant No. 2023ZDZK02); Huangpu District Health Commission Foundation in Shanghai (Grant No. HLM202403); Shanghai Municipal Health Commission (202340290); and Shanghai Huangpu District Top-notch Medical Talent Program (2023BJ03).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://gs.amegroups.com/article/view/10.21037/gs-2024-525/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Ruijin Hospital Luwan Branch, Shanghai Jiao Tong University School of Medicine (approval code: IRB-2022-006) and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Boucai L, Zafereo M, Cabanillas ME. Thyroid Cancer: A Review. JAMA 2024;331:425-45. [Crossref] [PubMed]
2. Wei B, Yao J, Peng C, et al. Clinical features and imaging examination assessment of cervical lymph nodes for thyroid carcinoma. BMC Cancer 2023;23:1225. [Crossref] [PubMed]
3. Zhuo X, Yu J, Chen Z, et al. Dynamic Nomogram for Predicting Lateral Cervical Lymph Node Metastasis in Papillary Thyroid Carcinoma. Otolaryngol Head Neck Surg 2022;166:444-53. [Crossref] [PubMed]
4. Zhao Y, Li W, Tao L, et al. Risk factors and nomogram to predict skip metastasis in papillary thyroid carcinoma. Gland Surg 2024;13:178-88. [Crossref] [PubMed]
5. Zhao H, Huang T, Li H. Risk factors for skip metastasis and lateral lymph node metastasis of papillary thyroid cancer. Surgery 2019;166:55-60. [Crossref] [PubMed]
6. Liu SQ, Feng JW, Yan ZT, et al. Constructing a nomogram based on the distribution of thyroid nodules and suspicious lateral cervical lymph nodes in fine-needle aspiration biopsies to predict metastasis in papillary thyroid carcinoma. Front Endocrinol (Lausanne) 2023;14:1242061. [Crossref] [PubMed]
7. Shao L, Wang Z, Dong W, et al. Risk factors associated with preferential lateral lymph node metastasis in papillary thyroid carcinoma. Cancer Med 2023;12:20670-6. [Crossref] [PubMed]
8. Haugen BR, Alexander EK, Bible KC, et al. 2015 American Thyroid Association Management Guidelines for Adult Patients with Thyroid Nodules and Differentiated Thyroid Cancer: The American Thyroid Association Guidelines Task Force on Thyroid Nodules and Differentiated Thyroid Cancer. Thyroid 2016;26:1-133. [Crossref] [PubMed]
9. Lai SW, Fan YL, Zhu YH, et al. Machine learning-based dynamic prediction of lateral lymph node metastasis in patients with papillary thyroid cancer. Front Endocrinol (Lausanne) 2022;13:1019037. [Crossref] [PubMed]
10. Liu W, Zhang D, Jiang H, et al. Prediction model of cervical lymph node metastasis based on clinicopathological characteristics of papillary thyroid carcinoma: a dual-center retrospective study. Front Endocrinol (Lausanne) 2023;14:1233929. [Crossref] [PubMed]
11. Zhao L, Zhou T, Zhang W, et al. Blood immune indexes can predict lateral lymph node metastasis of thyroid papillary carcinoma. Front Endocrinol (Lausanne) 2022;13:995630. [Crossref] [PubMed]
12. Gao L, Li X, Xia Y, et al. Large-Volume Lateral Lymph Node Metastasis Predicts Worse Prognosis in Papillary Thyroid Carcinoma Patients With N1b. Front Endocrinol (Lausanne) 2021;12:815207. [Crossref] [PubMed]
13. Huang Y, Yin Y, Zhou W. Risk Factors for Central and Lateral Lymph Node Metastases in Patients With Papillary Thyroid Micro-Carcinoma: Retrospective Analysis on 484 Cases. Front Endocrinol (Lausanne) 2021;12:640565. [Crossref] [PubMed]
14. Liu L, Jia C, Li G, et al. Can pre-operative ultrasound elastography predict aggressive features of solitary papillary thyroid carcinoma? Br J Radiol 2023;96:20220820. [Crossref] [PubMed]
15. Jiang M, Li C, Tang S, et al. Nomogram Based on Shear-Wave Elastography Radiomics Can Improve Preoperative Cervical Lymph Node Staging for Papillary Thyroid Carcinoma. Thyroid 2020;30:885-97. [Crossref] [PubMed]
16. Zhou J, Yin L, Wei X, et al. 2020 Chinese guidelines for ultrasound malignancy risk stratification of thyroid nodules: the C-TIRADS. Endocrine 2020;70:256-79. [Crossref] [PubMed]
17. Zhu J, Chang L, Li D, et al. Nomogram for preoperative estimation risk of lateral cervical lymph node metastasis in papillary thyroid carcinoma: a multicenter study. Cancer Imaging 2023;23:55. [Crossref] [PubMed]
18. Chen L, Zhan J, Diao XH, et al. Additional Value of Superb Microvascular Imaging for Thyroid Nodule Classification with the Thyroid Imaging Reporting and Data System. Ultrasound Med Biol 2019;45:2040-8. [Crossref] [PubMed]
19. Feng JW, Wu WX, Qi GF, et al. Nomograms based on sonographic and clinicopathological characteristics to predict lateral lymph node metastasis in classic papillary thyroid carcinoma. J Endocrinol Invest 2022;45:2043-57. [Crossref] [PubMed]
20. Wang Y, Deng C, Shu X, et al. Risk Factors and a Prediction Model of Lateral Lymph Node Metastasis in CN0 Papillary Thyroid Carcinoma Patients With 1-2 Central Lymph Node Metastases. Front Endocrinol (Lausanne) 2021;12:716728. [Crossref] [PubMed]
21. Liu L, Li G, Jia C, et al. Preoperative strain ultrasound elastography can predict occult central cervical lymph node metastasis in papillary thyroid cancer: a single-center retrospective study. Front Oncol 2023;13:1141855. [Crossref] [PubMed]
22. Wang D, Hu J, Deng C, et al. Predictive nomogram for central lymph node metastasis in papillary thyroid microcarcinoma based on pathological and ultrasound features. Front Endocrinol (Lausanne) 2023;14:1108125. [Crossref] [PubMed]
23. Jiang L, Zhang Z, Guo S, et al. Clinical-Radiomics Nomogram Based on Contrast-Enhanced Ultrasound for Preoperative Prediction of Cervical Lymph Node Metastasis in Papillary Thyroid Carcinoma. Cancers (Basel) 2023;15:1613. [Crossref] [PubMed]
24. Li J, Sun P, Huang T, et al. Preoperative prediction of central lymph node metastasis in cN0T1/T2 papillary thyroid carcinoma: A nomogram based on clinical and ultrasound characteristics. Eur J Surg Oncol 2022;48:1272-9. [Crossref] [PubMed]
25. Li P, Jiang W, Ding Y, et al. The impact of multifocality on lateral lymph node metastasis in papillary thyroid carcinoma. Eur J Surg Oncol 2023;49:106944. [Crossref] [PubMed]
26. Wang Y, Zheng J, Hu X, et al. A retrospective study of papillary thyroid carcinoma: Hashimoto's thyroiditis as a protective biomarker for lymph node metastasis. Eur J Surg Oncol 2023;49:560-7. [Crossref] [PubMed]
27. Ma Y, Li Y, Zheng L, et al. Prospective application of a prediction model for lateral lymph node metastasis in papillary thyroid cancer patients with central lymph node metastasis. Front Endocrinol (Lausanne) 2023;14:1283409. [Crossref] [PubMed]
28. Xu JM, Xu XH, Xu HX, et al. Prediction of cervical lymph node metastasis in patients with papillary thyroid cancer using combined conventional ultrasound, strain elastography, and acoustic radiation force impulse (ARFI) elastography. Eur Radiol 2016;26:2611-22. [Crossref] [PubMed]
29. Li W, Gao L, Du Y, et al. Ultrasound microflow patterns help in distinguishing malignant from benign thyroid nodules. Cancer Imaging 2024;24:18. [Crossref] [PubMed]
30. Zhao Y, Zhou P, Peng H, et al. Superb microvascular imaging compared with contrast-enhanced ultrasound to assess microvessels in thyroid nodules. J Med Ultrason (2001) 2020;47:287-97. [Crossref] [PubMed]
31. Stabenow E, Tavares MR, Ab'Saber AM, et al. Angiogenesis as an indicator of metastatic potential in papillary thyroid carcinoma. Clinics (Sao Paulo) 2005;60:233-40. [Crossref] [PubMed]