High Incidence of Distant Metastasis Is Associated With Histopathological Subtype of Pediatric Papillary Thyroid Cancer - a Retrospective Analysis Based on SEER

Objective Children with papillary thyroid cancer (PTC) have a higher invasive rate and distant metastasis rate, but the mortality rate is lower with unknown reasons. The majority of PTC cases comprise classical papillary thyroid carcinoma (CPTC) and follicular variant papillary thyroid carcinoma (FVPTC). This study aimed to determine the relationship between histopathological subtype and rate of distant metastasis and investigate factors influencing distant metastasis in pediatric PTC. Methods A total of 102,981 PTC patients were recruited from SEER registry, 2004-2015. Proportion of distant metastasis between children (≤18 years) and adults with different histopathological subtypes was compared by propensity score matching. The cut-off age for distant metastasis in children was calculated by receiver operating characteristic (ROC) curve, and the risk factors for distant metastasis in pediatric patients were analyzed by logistic regression models. Results Among the 1,484 children and 101,497 adults included in the study, the incidence of CPTC patients with distant metastasis in children was higher than that in adults (p<0.001). The ROC curve was calculated, which yielded a cut-off age for distant metastasis in CPTC children as 16 years old. In CPTC, the proportion of young children (2-16 years) with distant metastasis was higher than that of adolescents (17-18 years) and adults (>18 years) (both p<0.001). While there was no such trend in FVPTC. In young children (2-16 years), the incidence of CPTC with distant metastasis was higher than FVPTC (p=0.006). There was no difference between the proportion of CPTC and FVPTC with distant metastasis in adolescents (17-18 years) and adults. Logistic regression models revealed that extrathyroidal extension, lymph node metastasis and CPTC histopathological subtype were risk factors for distant metastasis in young children aged 2 -16 years. Conclusions In CPTC, the incidence of distant metastasis in young children (2-16 years) was significantly higher than that in adolescents (17-18 years) and adults (>18 years). In patients with distant metastasis aged 2-16 years, the proportion of CPTC was higer than that of FVPTC. Extrathyroidal extension, lymph node metastasis, and CPTC histopathological subtype were risk factors for distant metastasis in young children aged 2-16 years.


INTRODUCTION
Thyroid cancer is the most common endocrine cancer in the pediatric population (1). In a cross-sectional study based on the Surveillance, Epidemiology, and End Results (SEER) database including individuals younger than 20 years who had a diagnosis of thyroid cancer, the incidence of pediatric thyroid cancer increased by 1.1% per year from 1973 to 2006, with a significant increase of 9.5% per year from 2006 to 2013 (2). Compared to adult thyroid carcinoma, the prognosis of pediatric papillary thyroid cancer is generally fair (3). The reported mortality rate of pediatric papillary thyroid cancer (PTC) is very low with a higher rate of distant metastasis (DM) in most series despite more advanced disease at presentation and a higher risk of recurrence (4)(5)(6). A study involving patients aged 12 to 83 years showed that approximately half of patients with welldifferentiated thyroid cancer with DM die of disease within 5 years of initial diagnosis despite thyroid surgery and radioactive iodine (RAI) (7). Indeed, the presence of DM is reported to be an independent predictor for poor overall survival (8). Therefore in a disease with a low mortality rate such as PTC, it is critical to identify tumors at initial presentation that are at risk of developing DM to assist in clinical decision making. Despite incidence of DM of PTC in children (≤18 years) was higher than in adults (>18 years), the American Thyroid Association Guidelines for the Treatment of Thyroid Nodules and Differentiated Thyroid Carcinoma do not differentiate children from adults in diagnosis and treatment, thereby calling for further studies to guide treatment strategies. Pediatric PTC patients aged less than 18 years might benefit from tailored disease management by cut-off age and expect better prognosis.
Some published studies have suggested that the frequency of DM was different among the various histopathological subtypes (8), the majority of PTC cases comprise classical papillary thyroid carcinoma (CPTC) and follicular variant papillary thyroid carcinoma (FVPTC) histopathological subtypes (9), which account for 67-74% and 26-30% of cases, respectively (9,10). According to previous reports, extrathyroidal extension (ETE) and cervical lymph node metastases (LNM) are more common in CPTC than in FVPTC (11). The current consensus among thyroid academia is that there are only a few differences between CPTC and FVPTC. Moreover, the overall management of the two malignancies is similar, and patients with CPTC and FVPTC have similar long-term outcome (11). However, despite these similarities, minimal ETE and aggressive histopathological features, such as thyroid capsule infiltration, are significantly more common in CPTC than in FVPTC (11). Besides, patients with mETE showed significantly higher rates of lymph node metastases in the neck (12) and all levels of ETE, including microscopic ETE, were associated with increased risk for lymph node metastasis and DM (13). Thus, we propose the following hypothesis: The high DM rate in pediatric PTC aged less than 18 years old is associated with histopathological subtype.
Although there were some reports focusing on those young children had a greater degree of ETE and lymph node involvement than adolescents and were more prone to DM (4). Due to the lack of pediatric patients who are necessary to determine the explanation for the differences in clinicopathological outcomes observed in pediatric PTC patients, the focus on the reason why young children and adolescent patients show this difference has been rarely explored. More unfortunately, little attention was paid to differences in pathological subtypes among young children and adolescents. To assess our hypothesis and calculate the cut-off age for DM in children, as well as investigate factors influencing distant metastasis in pediatric PTC, we conducted a retrospective analysis at the first time using a population-based data set with a large sample.

MATERIALS AND METHODS Data Source
Data for this study were extracted from the National Cancer Institute's SEER data (Surveillance, Epidemiology, and End Results), which is one of the most representative large oncology registry databases in North America, covering 34.6% of the U.S. population and collecting case information from 18 population-based cancer registries (14). Age was categorized as ≤18 years and >18 years. The patients aged ≤ 18 years were regarded as children and patients aged >18 years were regarded as adults. Ethnicity was categorized according to the record in the SEER database as Black, White and other. Histopathological subtype was based on ICD-O-3. Tumor size less than 15cm are preserved. ETE was based on "CS extension (2004-2015)" codes. DM and LNM were based on AJCC (Eighth Edition). Foci was based on "CS site-specific factor 1" codes.

Statistical Analysis
The database materials were obtained by SEER*Stat 8.3.9 software, processed by WPS 2.7.1 software and statistically analyzed by SPSS 26.0 (IBM) software. Patient information, including demographic data and cancer-related information, was compared with histopathological subtypes. In our data, only two histopathological subtypes, CPTC and FVPTC, were present in patients ≤ 18 years old with DM. Therefore, this study focused on CPTC and FVPTC. Categorical variables were reported as frequency and proportion. We used the chi-square test or Fisher's exact test to compare these variables. To minimize selection bias, propensity score matching (PSM) was performed on ETE and LNM when comparing the proportion of DM in children and adults with different histopathological subtypes. The 1:1 matching scheme was used for matching, and a caliper of 0.05 SD for the probit value. The age cut-points at which CPTC and FVPTC developed DM in children were calculated by receiver operating characteristic (ROC) curve. Logistic regression analysis was used to calculate the risk factors for DM in children and regression coefficient (B), odds ratio (OR), the 95% confidence interval (CI) was used for reporting. Bilateral p value <0.05 was considered statistically significant difference. Variables with p value <0.05 in univariate analysis were included in multivariate analysis.

Demographic and Clinicopathological Characteristics
There were 102,981 patients diagnosed with PTC between 2004 and 2015, who met the inclusion criteria. Table 1  The ROC curve was calculated, which yielded a cut-off age of CPTC children with DM as 16 years old, with the area under the curve of 0.707, sensitivity of 0.750 and specificity of 0.606 ( Figure 1A). Therefore, we divided patients into young children (2-16 years), adolescents (17-18 years), and adults (>18 years). In CPTC, DM was observed in 31 (5.16%) young children, 5 (1.04%) adolescents and 617 (0.93%) adults (p<0.001) (

DISCUSSION
Thyroid cancer is the most rapidly increasing cancer in the United States (16). The incidence rates of pediatric thyroid cancer (patients younger than 20 years) increased more rapidly from 2006 to 2013 than from 1973 to 2006 (2). The incidence of DM is significantly more frequent in children (<19 years) than in adults, but the prognosis is generally good (3) with unclear reasons. By analyzing the demographic information and clinicopathological characteristics of a large number of PTC patients, this study confirmed that the incidence of DM in pediatric PTCs (≤18 years) was higher than that in adults, and the incidence of pediatric CPTCs (≤18 years) was also higher than that in adults. We compared the proportion of children (≤18 years) and adults with DM in CPTC and FVPTC respectively by propensity score matching. The results showed that in CPTCs, children had a higher rate of DM than adults. However, in FVPTCs, there was no difference in the incidence of DM between children and adults. We also used ROC curve to calculate the cut-off age for DM in CPTC children, and compared the difference in the proportion of histopathological subtypes above and below the cut-off age. According to our calculations, CPTCs had a higher rate of DM than FVPTCs in both children (≤18 years) and young children (2-16 years). There was no difference in DM rate between CPTC and FVPTC in adolescents (17-18 years) and adults. These findings further validated the correlation between CPTC and DM in young children. Finally, logistic regression models revealed that ETE, LNM, and CPTC histopathological subtype were risk factors for DM in young children aged 2-16 years. However, CPTC histopathological subtype was not a risk factor for DM in adolescents aged 17-18 years and in adults aged older than 18 years. PTC is more common in female than in male in a research including patients aged 0-24 years and the incidence of thyroid cancer increased with age (17). For pediatric population, difference in gender starts just above age 10, with increasing distinction above age 15 (18). There had been previous studies dividing thyroid cancer patients into pediatrics (<13 years) and adolescents to compare the metastasis and disease progression (19). Within the pediatric group under 18 years of age, special attention should be paid to male patients under 15 years old, as they are associated with a more advanced disease at diagnosis (5).  Tumors were generally infiltrative in patients younger than 15 years (20). Therefore, we believe that it is necessary to take the age into consideration to divide young children separately from adolescents during the development and disease progression of PTC. We calculated the cut-off age for DM in children as 16  A previous study had shown that CPTC and FVPTC were different in driving somatic genetic alterations and cell signal transduction, which leaded to the poor differentiation and strong invasiveness of CPTC (21). CPTC was a strong predictor of high    recurrence risk and high cancer-specific mortality, with a worse prognosis than FVPTC (9). In a study including 163 patients aged less than 18 years diagnosed as PTC, BRAF mutations and RET and NTRK fusions were detected mainly in CPTCs (20). The presence of a BRAFV600E mutation was reported to be correlated significantly with the need for a second treatment during the follow-up in patients under 18 years of age, and BRAF mutations might be associated with more aggressive clinical features and a higher risk of recurrence or persistence of disease in the pediatric population (5). Furthermore, the fusion-driven tumors, in general, displayed a lower thyroid differentiation score than mutation-driven samples (≤18 years), suggesting that gene fusion-positive pediatric PTCs are less differentiated (22). Besides, fusion gene-positive pediatric PTC cases (6-20 years) had more aggressive disease with more frequent extrathyroidal extension and lymph node and distant metastases than patients without fusion genes (23). This finding prompted us to speculate that the poor outcome in classical PTCs compared to FVPTCs was largely attributable to higher proportion of BRAF V600E mutations and RET and NTRK fusions in the former group. Clinically, the prevalence of highrisk parameters was significantly different among the two subtypes. Risk factors including ETE, LNM, stages III/IV, disease recurrence, radioiodine treatment, as well as mortality were lower in FVPTC (24). After studying a subgroup of FVPTC with an intact tumor envelope and very good prognosis, it is referred to as "non-invasive follicular thyroid tumor with papillary features" (NIFTP) and is classified as a noncancerous tumor (25). In conclusion, the biological behaviors and disease prognosis of CPTC and FVPTC differ significantly, and they should be distinguished in diagnosis and treatment. It has been reported that CPTCs were less represented in patients aged less than 15 years than in patients aged 15-18 years, while FVPTCs occurred more frequently in the former group (20). However, our results showed that the proportion of CPTCs was higher than that of FVPTCs, both in children aged ≤18 years and in young children aged 2-16 years regardless of with or without DM. The reason for this difference may be that our study is based on a large sample size, but the previous study included only 163 samples. In the other hand, our logistic regression analysis showed that tumor diameter > 1 cm was not a risk factor for DM in patients aged 2-16 years (p>0.05). This indicates that although children, when compared to adults, had larger primary tumors (26) and the tumor diameter is related to the poor prognosis of FVPTC (9), but tumor diameter is not a risk factor for DM in patients aged 2-16 years, which may be one of the reasons why the proportion of FVPTCs in patients aged 2-16 years with DM is lower than that of CPTCs.
The molecular biological characteristics of the pathogenesis of thyroid cancer in children and adults may explain the differences in clinical manifestations and prognosis (27). Despite this, the clinical assessment and treatments used in pediatric thyroid cancer are the same as those implemented for adults (18). Histopathological subtypes have recently been shown to play an important role in determining the persistence and/or recurrence of disease (28). Our data show that the incidence of DM in young children (2-16 years) with CPTC is significantly higher than that in adolescents (17-18 years) and adults, and that the same histopathological subtype presents different clinical and histopathological features in different age groups. ETE, LNM, and CPTC subtype are risk factors for DM in pediatric patients aged 2-16 years. We suggest that the histopathological subtypes of CPTC and FVPTC should be classified and managed separately in patients aged 2-16 years to cope with the persistent or recurrent risk of disease, but further studies are needed to expand our findings, which may guide therapeutic strategies.
This study has certain limitations. First, it was a retrospective analysis based on the SEER database. There was an inherent selection bias. To control for selection bias, we adopted a rigorous scientific study design, clarified the inclusion criteria and exclusion criteria of subjects, and unified the disease diagnosis. Second, not all data are available from the SEER database, such as patient recurrence information, not allowing analysis of the subsequent DM occurrence in different histopathological subtypes. Third, the use of stratified analysis in this study would cause statistical deviation due to the small number (41 cases) of pediatric patients with DM. Finally, data in this study were extracted from the SEER database from 2004 to 2015, but the 2017 WHO classification introduced the NIFPT terminology for encapsulated FVPTCs. As a result, a minority of the "FVPTCs" reported in the SEER database may in fact be NIFPTs. It might affect the differences in outcome between bona fide FVPTCs vs CPTCs within an acceptable range.
However, our study is the first with a large sample to investigate the role of common histopathological subtypes in determining DM of pediatric PTC and the reasons of the high DM incidence in children. We propose that pediatric PTC patients should be divided into patients aged 2-16 years and patients aged 17-18 years. CPTC patients aged 2-16 years might be treated more aggressively. Further studies in the future may be helpful to guide the treatment strategies for PTC in pediatric patients.
In conclusion, this study showed that in CPTC, the incidence of distant metastasis in young children (2-16 years) was significantly higher than that in adolescents (17-18 years) and adults (>18 years). There was no such difference among patients with FVPTC. In patients with distant metastasis aged 2-16 years, CPTC patients had a higher rate of DM than FVPTC patients. Extrathyroidal extension, lymph node metastasis, and CPTC histopathological subtype were risk factors for distant metastasis in young children aged 2-16 years.

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

AUTHOR CONTRIBUTIONS
XZ, ZW, ZG, JX, MC, WS, LH, WD, JH, DZ, CL, TZ, LS, PZ, and HZ contributed to this study. ZW contributed to the conception and design of this study. XZ collected data. XZ and ZG performed the statistical analysis. XZ and ZW drafted and wrote the manuscript. All authors contributed to the article and approved the submitted version.