The incidence of papillary thyroid cancer (PTC), the most common type of differentiated thyroid cancer (DTC), has increased over the last decades (1, 2). The use of high-resolution ultrasonography (US) and fine needle aspiration (FNA) biopsy techniques have led to an increase in the diagnosis of PTC (3). However, the sensitivity of US in assessing deep anatomical spaces of the central neck compartment is quite low (4).

In 2015 the American Thyroid Association Guidelines recommended against prophylactic central lymph node dissection (PCLND) for small (T1-T2), non-invasive, clinically node-negative (cN0) PTC, while PCLND could be considered for patients with T3-T4 or cN1b (5). However, the Chinese Thyroid Association recommends routine central lymph node dissection (CLND) for all PTC patients to stage disease and prevent recurrence (6). Thus, the routine CLND for T1-T2 patients without certain lateral lymph node metastases (LLNM), comparing to T3-T4 or cN1b patients, remains controversial. In the 8^th edition of American Joint Committee on Cancer (AJCC) TNM classification, PTC patients with minimal extrathyroidal extension (mETE) and tumor diameter less than 4 cm were classified into T1-T2 stage, and there was also controversy about the treatment of central lymph nodes of these patients (7, 8). According to relevant studies, over 50% of patients with PTC have central lymph node metastases (CLNM) (9), and the majority of these patients are in the T1-T2 disease subgroup. Even in patients with clinically negative cervical lymph nodes (cN0), the rate of CLNM can range between 15.9% and 53% (10). Except for that, several studies still advocate CLNM is a risk factor for recurrence (11, 12). Considering the complications of CLND, such as laryngeal nerve injury and hypoparathyroidism, decision-making for these patients should better depend on individual patients rather than single approach recommended by guidelines. Based on preoperative clinical characteristics, several nomograms have been built to predict central or cervical LNM in PTC patients (13–16). However, there is no available nomogram based on the SEER database to predict CLNM in T1-T2 PTC patients without LLNM.

To that end, the present study used the preoperative clinical characteristics of T1-T2 PTC patients without lateral neck metastasis retrieved from the SEER database from 2010 to 2015 to construct a nomogram and assist clinicians in choosing the optimal treatment strategy for this particular patient group.

The incidence of papillary thyroid cancer has increased yearly, mainly due to advances in diagnostic technology. Early-stage PTC tends to have a good prognosis, which dictates a conservative approach to diagnosis and treatment. According to the management guidelines of the ATA (2015), for adult patients with thyroid nodules and differentiated thyroid cancer, fine needle aspiration (FNA) was only recommended for patients whose maximum tumor diameter was larger than 10 mm combined with highly suspicious sonographic features (5). One retrospective multicenter study concluded that there was no significant difference in the risk of regional recurrence whether CLND was performed or not in cN0 PTC (17). In contrast, for some patients, incomplete lymph node dissection increases the rate of regional recurrence and repeated surgery (12). For T1-T2 PTC patients without LLNM in China, routine CLND is still recommended, which is not recommended for all such patients in guidelines of the ATA (2015) (5, 6). Additionally, for the change of the AJCC TNM classification, some T3 patients in the 7^th edition of AJCC TNM classification were reclassified into the T1-T2 stage, who were recommended a PCLND in the past (7, 8). In light of this, it is important to construct a visual predictive model to assist surgeons in deciding whether CLND is appropriate for these patients.

For low-grade, low-risk cN0 PTC patients, active surveillance (AS) and thermal ablation (AB) are also considered treatment modalities (18, 19). Considering the high incidence of CLNM in PTC, the long-term prognosis of these treatment options remains to be determined. According to Kuma Hospital, low-risk papillary thyroid microcarcinoma (PTMC) should be treated with AS instead of surgery, and patients with tumor diameters >1 cm do not always require immediate surgical intervention. Patients with older age, multifocality, and minimal invasion could also be targeted for AS (18). In a recent study by Sasaki et al., the need for conversion of AS to surgery for low-risk PTMC patients has decreased over time (20). However, consistent with previous studies and meta-analyses (21–23), our study found that age <44 years, multifocality, and mETE were risk factors for CLNM in small PTC (T1-T2) patients with no LLNM. Notably, there is a relatively high incidence of CLNM in cN0 patients (10), which led to some of these patients being treated with AS. To date, there are no studies on the long-term outcomes of patients treated with AS. Several studies (19, 24, 25) indicated that thermal ablation (TA) could be used as a treatment option for low-risk PTMC, but the longest follow-up time was 5 years. Considering that the incidence of CLNM is high, the impact on long-term prognosis is thus unclear. In our study, the CLNM rate of small PTC (T1-T2) with no LLNM was 32.1%-34.5%. Moreover, age, gender, race, tumor diameter, multifocality and mETE were considered risk factors for CLNM; therefore, AS or TA was probably unsuitable for high-risk patients in the prediction nomogram.

Previous studies have established several predictive models to predict CLNM in PTC patients. These had several limitations, such as a small sample size, the lack of external validation, or low racial heterogeneity. For instance, Gao et al. (13) collected information on clinicopathological characteristics and preoperative ultrasound images to predict CLNM in cN0 PTC patients. However, only 296 patients in a single center were included, and no external validation was performed. In a nomogram for predicting CLNM in PTC patients developed by Yang et al. (14), the internal and external validation C-indexes reached 0.857 and 0.825, and the model was significantly better compared to ultrasound; however, only 1,731 patients from China were included. One nomogram developed by Kim et al. (15) had a population size of 7,535 patients, but they were all of Korean descent. Furthermore, a recent analysis by Zhu et al. (26) in T1-T2, non-invasive, and cN0 PTC patients revealed that gender, tumor size and location, multifocality, age, and Delphi node status were independent predictors of CLNM. Moreover, by comparing the predictive ability of different types of models, a predictive model with an AUC of 0.70-0.75 was obtained. This was a single-center study that also lacked external validation. Feng et al. (16) previously used the SEER database to construct a nomogram for predicting cervical LNM in PTC patients and obtained excellent C-indexes of 0.716-0.733. There were also some studies, such as Min et al. (27) and Wei et al. (28), which have constructed excellent nomograms for predicting CLNM in different subgroup of PTC. And for T1-T2 with no LLNM, there were few studies. To our knowledge, we are the first to construct a nomogram based on the SEER program with internal and external validation cohorts for predicting CLNM. Moreover, we focused on T1-T2 patients with no LLNM, which was not studied in previous studies. To some extent, our nomogram may better predict CLNM in small PTC (T1-T2) patients, which is pretty common in clinical practice. As the restaging of T stage of the 8^th AJCC, some patients with mETE were identified as T1-T2. We also concluded that mETE was the risk factor of CLNM, and the HR was 2.568 (2.206, 2.989), which was not low. Additionally, we comprehensively validated our findings using ROC, calibration and DCA curves. Therefore, this nomogram could possess a higher predictive value in clinical practice. To make personalized decisions for patients, surgeons may use this nomogram independently of imaging examinations such as ultrasounds and CT scans.

Nevertheless, our study has some limitations. Firstly, due to the SEER database’s limitations, we could not include some variables, such as Hashimoto’s thyroiditis (HT), BRAFV600E mutation, several laboratory variables, and preoperative imaging features, which tend to associate with CLNM (13, 14, 29). Secondly, the external validation cohort is from a single center, and all patients were Chinese. Therefore, further external validation of this nomogram using patients from different countries and medical centers is warranted. Thirdly, this model was only designed for classic PTC patients, and was not applicable to other pathology types of PTC.

Male gender, age <44 years, non- black population, tumor size larger than 1 cm, multifocality, and mETE were identified as risk factors for CLNM in T1-T2 PTC patients without LLNM. Although most T1-T2 PTC patients achieve long-term favorable prognoses, surgeons should reconsider whether neck lymph node dissection, TA and AS interventions are less suitable for patients at high risk of CLNM. Herein, using clinicopathological characteristics obtained from the SEER database, a nomogram was developed and validated for the prediction of CLNM in T1-T2 PTC patients without LLNM. Overall, this nomogram demonstrated robust effectiveness and could be used by clinicians to make personalized clinical decisions regarding T1-T2 patients.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

YS contributed to the conception of the study, performed the data analyses and wrote the manuscript. WS and JX helped perform the analysis with constructive discussions. HZ contributed to the conception and supervision of the manuscript. All authors contributed to the article and approved the submitted version.