The prevalence of thyroid cancer is increasing worldwide (1). Notably, papillary thyroid carcinoma accounts for 85% of differentiated thyroid cancers with a high 10-year survival rate (2–4). CVPTC is the most prevalent variant and is believed to be a less aggressive histological subtype that has a lower risk for death and metastatic disease (5). Prophylactic central lymph node dissection remains controversial according to different guidelines (6–8). However, up to 50% of papillary thyroid carcinoma (PTC) patients develop lymph node metastasis (LNM) (9). Large-volume or high-volume LNM (HVLNM) was defined as >5 metastatic lymph nodes (8). Increasing evidence has shown that patients with HVLNM have poorer outcomes than those with small-volume LNM, which includes higher recurrence rates and lower disease-free survival (10, 11). Consequently, the latest American Thyroid Association (ATA) guidelines have determined clinical N1 disease or >5 pathological lymph nodes (less than 3 cm) as characteristics of patients with an intermediate risk for postoperative risk stratification (8).

Ultrasound (US) is the primary choice for routine thyroid examination and preoperative staging of thyroid cancer. Several associations have published reporting systems for the assessment of thyroid nodules based on ultrasonographic patterns that are helpful in diagnosing nodules (8, 12–14). However, due to the complex anatomical structure of the neck and the physical limitations of US, the diagnostic performance of preoperative US evaluation in detecting lymph node involvement in the cervical region, especially in the central compartment, is not particularly effective (15). Another reason that the detection rates of positive lymph nodes are low is that the procedure relies heavily on the proficiency of operators. Therefore, during the preoperative examination, the potential risks of LNM and HVLNM may be overlooked, further misleading the management of vulnerable patients.

Many studies have focused on the association between imaging patterns of thyroid nodules and LNM. A few studies have concentrated on the risk factors for HVLNM (16–20). Furthermore, in these studies, multifocal lesions and postoperative diagnosis were analyzed in most cases. And the association between preoperative ultrasonographic features and HVLNM, such as capsule morphology, has never been investigated thoroughly. Clinical decisions may be altered if feasible preoperative prediction models can be established, and more aggressive treatment modalities may be found suitable for some CVPTC patients.

Thus, this study aimed to construct a preoperative nomogram to predict HVLNM in CVPTC patients with a solitary lesion based on demographic and ultrasonographic features. Besides, the predictive value of the nomogram was assessed using a validation set consisting of the patients from the last 6 months of the cohort. The developed nomogram may help clinicians select a follow-up approach for the entire diagnosis and treatment process.

Evidence has suggested that the prognosis of PTC was linked to the number and size of involved lymph nodes. Randolph et al. revealed a marked difference in the median risk of recurrence in pathological N1 patients between <5 positive nodes (4%, range 3%–8%) and >5 nodes (19%, range 7%–21%) (10). Moreover, an analysis based on data from the National Cancer Data Base and SEER database, showed that a higher number of metastatic lymph nodes (up to six metastatic) was associated with lower overall survival (HR 1.12, 95% CI 1.01–1.25). In contrast, no additional mortality risk was found with more positive nodes (HR 0.99, 95% CI, 0.99–1.05) (11). All these findings allude a more concerning message about HVLNM. Thus, the 2015 ATA guidelines have modified the clinical and pathological nodal status as a characteristic that can stratify the risk of recurrence in PTC patients. The status of more than five metastatic lymph nodes is classified as intermediate risk, indicating a >20% risk of recurrence (8).

Precise identification of patients with HVLNM preoperatively may facilitate the selection of rigorous screening and treatment protocols. However, the low diagnostic performance of lymph nodes in preoperative assessments is unfavorable for patients and physicians. It was reported that the diagnostic efficacy was unsatisfactory, with a pooled sensitivity between 0.31 and 0.35 for detecting involved lymph nodes in the central neck (22). Another meta-analysis reported that US + Computed Tomography and FNA cytology + FNA-thyroglobulin (FNA-Tg) showed good diagnostic performance (23). The limitations are that these methods are radiative or invasive. For these reasons, it would be helpful to establish a prediction model for PTC patients combined with demographic features and nodule characteristics using US.

In a previous study, preoperatively sonographic characteristics combined with serum Tg antibodies levels were found to help predict central LNM (CLNM) (24). Differently, our current study developed a preoperative nomogram based on preoperative sonographic patterns to visualize the prediction model for HVLNM in solitary CVPTC patients. The AUC was 0.842 in the training set and 0.875 in the validation set. Moreover, the HVLNM rate was found in 9.1% (46/503 cases) in the training set and 8.1% (10/123 cases) in the validation set, which was lower than that reported by Liu et al. (12.3%, 254/2,073 cases) (16). Different designs of the single lesion in this study, the inclusion criteria, or the sample capacity might have caused the divergence. When comparing the two groups in the training set, the data revealed that male sex, tumor size >10 mm, ETE on neck US, and capsular contact >50% were independent risk factors for HVLNM. Contrastingly, middle age and old age were significant protective factors. It is suggested that increased vigilance would be required for potential HVLNM patients.

There is no consensus regarding the impact of age on LNM, and age classification varies in different studies. Ito et al. pointed out that tumor progression was the greatest in young patients and the poorest in older patients with papillary thyroid microcarcinoma (PTMC), indicating that tumors presented increased aggressiveness in younger patients compared to older patients (25). A meta-analysis, including 9,369 PTC patients, reported that age <45 years increased the risk of lymph node metastasis (pooled OR 1.52, 95% CI 1.14–2.01, P < 0.00001) (26). In the study, age classification was consistent with that of two recent studies by Oh et al. and Shen et al. (19, 20). According to our data in the training set, people of middle age and older people groups had less than half the incidence of HVLNM compared to the young age group. Multivariate analysis revealed that middle-aged and older patients had lower rates of HVLNM than younger patients. In addition, studies have investigated age's influence on HVLNM in patients with PTMC and derived some approximations. For instance, Zhang et al. found that the risk of HVLNM in PTMC was significantly lower in patients of middle age (40–59 years) (OR 0.313, 95% CI 0.191–0.515) and older (≥60 years) patients (OR 0.085, 95% CI 0.012–0.633) (18). Another finding worth mentioning is that Liu et al. revealed that age ≥40 years was an independent protective factor (17). Similarly, elderly age (≥55 years) was also found to be a protective factor of high-volume CLNM within PTMC in Wei’s study (27). However, age was not found to be significantly different among participants in a study of 2,073 patients with PTC using the cutoff of 55 years (16). Overall, grouping by age varied in some studies; thus, detailed and well-recognized stratification of age groups may provide a more accurate risk assessment. Apart from age, our results demonstrated that the risk of HVLNM in males was 3.396 times higher than that in females (95% CI 1.579–7.304), indicating a similar adverse effect of male sex on LNM in previous studies (26).

Importantly, primary tumor size is the most intuitive parameter in the preoperative US that has long been analyzed in previous studies and included in several scoring schemes (AGES, AMES, and MACIS) (28–30). To distinguish PTMC and conventional papillary thyroid cancer, 10 mm is used as a benchmark. It has been proposed that tumors with a diameter larger than 10 mm have a higher incidence of invasion, LNM, and CLNM, thus requiring more radical tactics to improve outcomes (26, 31–33). Likewise, in our study, the HVLNM portion of patients with tumor size ≤10 mm was 4.2%, while in the >10 mm group was 18.8% in the training set. Multivariate analysis also demonstrated that a tumor diameter >10 mm significantly increased the risk of HVLNM.

Numerous studies have shown that ETE was recognized as a predictor of metastatic diseases, such as locoregional LNM and distant metastasis (34, 35). In addition, Feng et al. used the SEER database and single-center data to evaluate the correlation between demographic and clinicopathologic characteristics and CLNM in CVPTC patients (33). ETE on pathology was significantly different between the two groups (33). On the contrary, Tao et al. did not find significant associations of ETE with CLNM and lateral LNM (LLNM) in PTMC (36). However, the definition of ETE differed in some of these studies, and most were based on postoperative pathological results, which limited their utility in the presurgical setting. Lamartina et al. reported that by combining US signs of minimal or gross ETE and taking the presence of microscopic or gross ETE as a reference on histology, preoperative US achieved an accuracy of 81.5% (37). In addition, gross ETE is believed to have a higher incidence of tumor recurrence than microscopic ETE (38, 39). Accordingly, preoperative US signs of ETE are of great diagnostic importance. In our current study, the incidences of HVLNM were significantly different in terms of ETE. Nevertheless, further studies are still needed to determine the diagnostic criteria for ETE using US.

Few studies have estimated the degree of capsular contact and analyzed its impact on LNM in PTC (40–42). Animal and human studies suggested that lymphatic networks and vessels appeared denser at the periphery of the gland (43, 44). A study in Japan revealed that increased lymphatic density was also correlated with vascular endothelial growth factor-D expression and LNM in PTC patients (45). These findings could explain that the state of contact with the glandular capsule has facilitated the spread of the tumor to regional lymph nodes. Research in clinical practice concurred with these findings. Ye et al. found that capsular extension >50% was associated with LLNM in PTC (40). Kwak et al. found >25% contact with the adjacent capsule was a risk factor for LLNM in PTMC (46). Contrastingly. Lin et al. and Zeng et al. found no significant association between capsular contact and LLNM after multivariate logistic regression (47, 48). Different from previous studies, our study highlighted the association between capsular contact and HVLNM and found that capsular contact >50% was the most common in the HVLNM group in the training set, contributing the most to the prediction model. Since the all-around measurement is needed during the examination, real-time observation of US or dynamic images after the examination could benefit the analysis after the examination.

Despite the promising results, limitations remain in the current study. First, this was a retrospective case-control study at a single center, and selection bias was inevitable. Second, only a small number of patients had HVLNM, and comparisons were only made between the HVLNM and non-HVLNM groups, while factors related to other nodal statuses were not investigated. Moreover, our nomogram included only five factors, and potential variables might need to be analyzed and validated. Thus, a large-sample cohort study involving external validation from a multicenter study is required.

In conclusion, HVLNM is relatively uncommon in CVPTC patients with a solitary lesion. Male sex, larger tumor size (>10 mm), ETE on US, and capsular contact >50% increased the risk of HVLNM, whereas middle and old age were significant protective factors. These findings may be essential for implementing more vigilant and aggressive preoperative examinations and treatment strategies for CVPTC patients with a high risk of HVLNM based on the nomogram. Additionally, real-time US plays a vital role in preoperative assessment.