Introduction

Papillary thyroid carcinoma (PTC) is the most commonly encountered malignant endocrine tumor and accounts for up to 85% of all thyroid malignancies [1, 2]. Although PTC usually affects adults with a female predominance, small numbers of PTCs occur in children and youth [2–11]. In these pediatric PTCs, histopathologically and clinically distinct features are reported [3, 6]. For instance, children and youth have a predilection to be affected by morphologically unique variants of PTC, such as solid variant (SV), solid-follicular variant (SFV), and diffuse sclerosing variant (DSV) [8, 12–14]. Despite a higher frequency of lymph node (LN) and distant metastases, some reports indicated that the overall survival of pediatric PTCs is more favorable than adult PTCs [7, 11].

Molecular pathogenesis of adult PTCs has been clarified by previous studies [1, 15, 16], including a next-generation-sequencing analysis by the Cancer Genome Atlas [17]. Activating mutation of the BRAF gene is the most frequent driver alternation of PTC [1, 15, 18]. BRAF V600E is generally identified in 35 to 80% of adult PTCs [1, 15, 18]. Conversely, there have been limited studies on the BRAF mutation in pediatric PTCs [5, 6, 8–11, 19, 20]. In addition, the number of examined cases in those studies was relatively small, and the mutation frequency varied widely from 0 to 36.8%. Since mutated BRAF is a promising target for molecular therapy [21], it is important to discover the true incidence and significance of BRAF mutation in pediatric PTCs. Another important genetic alternation in thyroid cancer is TERT promoter mutation. Some earlier studies showed that TERT mutations are associated with aggressive clinicopathological features [22–24]. However, only few studies reported its incidence and significance in pediatric PTCs [20, 25].

Radiation exposure is a well-established risk factor for thyroid cancer. Indeed, after the nuclear meltdown accident at Chernobyl in 1986, the incidence of PTC had increased in the radiologically contaminated areas of Belarus, Russia, and Ukraine [26]. In Japan as well, the Fukushima Dai-ichi nuclear accident in 2011 has created great concern about the risk for pediatric thyroid cancer, especially in eastern Japan. Recently, Mitsutake et al. reported the genetic profile of thyroid carcinoma in the young population in Fukushima [27]. Their study enrolled 67 PTCs and one poorly differentiated carcinoma detected by the thyroid ultrasound screening for children in the Fukushima prefecture aged 0–18 years old at the time of the accident. Of note, quite different from post-Chernobyl PTCs, BRAF V600E was highly prevalent in the pediatric PTCs in Fukushima. Based on this finding, they suggested non-radiogenic etiology in the pediatric PTCs in Fukushima. The conclusion is well concordant with the lower estimated thyroid doses of radioiodine exposure in Fukushima [28]. However, their study had no comparison with sporadic non-radiation-associated pediatric PTCs in Japan. To evaluate the actual impact of the Fukushima accident, it is essential to review a large cohort of Japanese pediatric PTCs from before the disaster and to delineate their clinicopathological and genetic features.

The purpose of this study is, therefore, to evaluate the frequency of BRAF and TERT promoter mutations and their clinicopathological significance in a series of non-radiation-associated pediatric PTCs in Japan.

Materials and Methods

Patients, Tissue Samples, and Data Collection

In the present study, we defined PTC occurring up to and including age 20 years as pediatric PTC. We explored the pathological records of University of Yamanashi Hospital and Kuma Hospital from 1991 to 2013 and collected cases of 81 pediatric PTCs. As the control group, we collected 91 cases of adult PTCs from patients aged 21 years or over. We performed additional pathological analyses on the formalin-fixed and paraffin-embedded (FFPE) tissue blocks from these cases and obtained the following clinicopathological data from the medical records under an appropriate identification: age, sex, tumor size (cm), extrathyroidal invasion, LN metastasis, and distant metastasis. All patients studied lived in iodine-sufficient areas, and there were no individuals with previous history of radiation exposure. The study protocol was approved by the ethical review board of University of Yamanashi (approval code: 1332).

Histopathological Evaluation

Two pathologists (N.O. and R. K.) examined the histology of the most representative slide from each PTC without any prior clinicopathological or genetic information on the case. They used the histological classification of the WHO Classification Tumors of Endocrine Organs [29]. However, the WHO classification defines the SV as a PTC dominated by solid sheets of tumor cells, whereas our case studies rarely had a purely solid pattern. The solid architecture was usually intermingled with some follicular pattern. These PTCs were very similar to the SFV described by Harach and Williams [30]. Hence, in the current study, we combined these histological types in an SV/SFV group.

DNA Extraction from FFPE Samples

We extracted DNA from FFPE using RecoverAll total nucleic acid isolation kit (Applied Biosystems, Tokyo, Japan) according to the manufacturer’s protocol. To validate the quality and quantity of the extracted DNA, we performed PCR for phosphoglycerate kinase 1 gene (PGK1). The PCR amplification was carried out in a reaction volume of 50 μL with 10× PCR buffer (QIAGEN, Tokyo, Japan), 200 μM of each dNTP (QIAGEN), 500 nM of each primer (listed in Table S1), and 1.25 units of HotStarTaq DNA polymerase (QIAGEN). The thermal cycling condition was as follows: 15 min at 95 °C, followed by 40 cycles of 94 °C for 30 s, 58 °C for 30 s, and 72 °C for 30 s, with a final extension at 72 °C for 10 min. We used 3% agarose gel electrophoresis with Midori Green DNA Stain (NIPPON Genetics Co., Ltd., Tokyo, Japan) and light-emitting diode (LED) illuminator (NIPPON Genetics) to analyze the PCR products.

Allele-Specific PCR and Sanger Sequencing for BRAF Mutation

To detect BRAF mutations, we carried out two-step testing with initial allele-specific PCR (AS-PCR) and subsequent Sanger sequencing. We screened for BRAF V600E, the most common type of BRAF mutation, using AS-PCR, and subsequently performed Sanger sequencing for BRAF exon 15 in the AS-PCR-negative PTCs. AS-PCR is a concise method using the specific primer pair for BRAF V600E mutant allele (listed in Table S1), and we performed the AS-PCR as described previously [31, 32]. In Sanger sequencing for BRAF exon 15, 1 μL of DNA in a final volume of 50 μL with 10× PCR buffer, 200 μM of each dNTP, 500 nM of each primer (listed in Table S1), and 1.25 units of the polymerase was amplified. The thermal cycling condition was as follows: 15 min at 95 °C, followed by 40 cycles of 94 °C for 30 s, 58 °C for 30 s, and 72 °C for 30 s, with a final extension at 72 °C for 10 min. The PCR products were submitted to Sanger sequencing with the forward primer using BigDye Terminator v1.1 Cycle Sequencing Kit and ABI Prism 3130xl Genetic Analyzer (Applied Biosystems, CA, USA).

Sanger Sequencing for TERT Promoter Mutations

We examined TERT promoter mutations by standard Sanger sequencing. In brief, 1 μL of DNA in a final volume of 50 μL with 5× Q-solution, 10× PCR buffer, 200 μM of each dNTP, 500 nM of each primer (listed in Table S1), and 1.25 units of the polymerase was amplified. The thermal cycling condition was the same as described in BRAF exon 15. After electrophoresis, the PCR products were sequenced with the forward primer.

Sanger Sequencing for CTNNB1 and PIK3CA

Mutations of CTNNB1 and PIK3CA have been reported in cribriform-morular variant (CMV)-PTCs [33, 34]; we explored mutations at exon 3 of CTNNB1 and exon 9 and exon 20 of PIK3CA in one CMV-PTC using Sanger sequencing. Used primers are listed in Table S1. The protocol for PCR and Sanger sequencing was the same as that for BRAF exon 15.

Statistical Analysis

Chi2 test and the Fisher exact test were used for ordinal or nominal categorical variables and discrete numerical variables. Student’s t test and Mann-Whitney test were used for continuous numerical data appropriately, after Shapiro-Wilk test for normality. All statistical calculations were performed with the SPSS Statistics 22 software package (IBM Japan, Tokyo, Japan), and a p value of <0.05 was considered statistically significant.

Results

Clinical and Histopathological Features

Table 1 summarizes clinicopathological features of both pediatric and adult PTCs. In the pediatric PTC group, patients’ ages were distributed from 6 to 20 years. Children 15 years or under accounted for 22% (18/81) of the cases. Females accounted for 91% (74/81) of pediatric PTCs. Although female predominance was also apparent in adult PTCs (62% (56/91)), the disparity between the sexes was more significant in the pediatric PTCs (p < 0.001). Figure 1 shows the representative histology of pediatric PTCs. There was a higher prevalence of non-classical PTCs including SV/SFV and DSV in the pediatric group than in the adult control group (19% (15/81) vs 4% (4/91), p = 0.003). Furthermore, pediatric PTCs exhibited significantly advanced feature in tumor size and LN metastasis. The mean tumor size was significantly larger in pediatric PTCs than in adult PTCs (3.0 ± 1.5 vs 1.3 ± 0.9 cm, p < 0.001). Microcarcinoma, defined as carcinoma with a diameter 1.0 cm or smaller, was identified in only 5% (4/75) of pediatric PTCs, while 49% (42/86) of adult PTCs were microcarcinoma (p < 0.001). The incidence of LN metastasis was significantly higher in pediatric PTCs than adult PTCs (90% (72/80) vs 66% (58/88), p < 0.001). Furthermore, extensive LN metastasis to central and lateral nodes (pN1b) was identified in 68% (54/80) of pediatric PTCs and in only 16% (14/88) of adult PTCs (p < 0.001). However, there was no statistically significant difference in distant metastasis between pediatric and adult PTCs (4% (3/80) vs 1% (1/88), p = 0.27). Table 2 indicates the clinicopathological difference of each histology in pediatric PTCs. Classical PTCs occurred in significantly older pediatric patients (17.8 ± 2.6 in classical, 15.3 ± 2.5 in SV/SFV, 16.0 ± 2.9 in DSV, p = 0.034). Of note, distant metastasis was exclusively observed in 29% (2/7) of SV/SFVs and 25% (1/4) of DSV (p = 0.002). Clinicopathological feature of the pediatric PTCs with distant metastasis is summarized in Table S2. All the three pediatric PTCs with distant metastasis exhibited multiple lung metastases. In addition, the DSV-PTC showed multiple bone metastases. While the patient with DSV-PTC died of the pulmonary fibrosis after the radioactive iodine therapy, the other two patients were alive at the last follow-up (18 and 9 years after the initial operation, respectively). Other clinicopathological factors including tumor size, extrathyroidal invasion, and LN metastasis showed no association with the histology in pediatric PTCs.

Table 1 Clinicopathological features of pediatric and adult PTCs
Full size table
Fig. 1
figure 1

Representative histology in pediatric papillary thyroid carcinoma (PTC). a Classical PTC. The predominant papillary configuration is present. b Solid variant (SV). Tumor cells exhibit predominantly solid sheet-like architecture. c Solid-follicular variant (SFV). Intermingled follicular pattern is identified in the solid component. d Diffuse sclerosing variant (DSV). Neoplastic cells accompanied by psammoma bodies and squamous metaplasia present in a lymphatic sinus. Note the marked infiltration of lymphocytes

Full size image
Table 2 Histology and clinicopathological features of pediatric PTCs
Full size table

BRAF Mutations

We identified BRAF V600E mutation in 54% (44/81) of pediatric PTCs and 85% (77/91) of adult PTCs (Table 3); the BRAF mutation frequency was significantly lower in pediatric PTCs than in adult PTCs (p < 0.001). No BRAF mutations other than BRAF V600E were identified in pediatric or adult PTCs. Table 4 and Table S3 show the association between BRAF V600E status and clinicopathological features in pediatric and adult PTCs, respectively. In pediatric PTCs, BRAF V600E was associated with an older patient’s age (18.2 ± 2.1 vs 16.5 ± 3.0, p = 0.009). The association between BRAF V600E and patient’s age was also significant, when the pediatric PTC patients were subdivided into two age groups: 28% (5/18) in 15-year-old or younger vs 62% in 16- to 20-year-old (39/63) (p = 0.011). However, the older group of pediatric PTCs still showed significantly lower frequency of BRAF V600E than the adult PTCs: 62% (39/63) vs 85% (77/91), p = 0.001.

Table 3 BRAF and TERT mutation status in pediatric and adult PTCs
Full size table
Table 4 BRAF V600E mutation status and clinicopathological features in pediatric PTCs
Full size table

From a morphological aspect, the BRAF mutation was positively associated with classical histology. Indeed, all of 44 BRAF-mutated pediatric PTCs were classical PTC; conversely, 67% (44/66) of pediatric classical PTCs were positive for BRAF mutation. On the other hand, non-classical pediatric PTCs exhibited no BRAF mutation (p < 0.001). We also confirmed this correlation between BRAF mutation and histological appearance in adult PTCs (p = 0.011) (Table S3). However, regarding PTCs with classical histology, the frequency of BRAF mutation was still lower in pediatric classical PTCs than in adult classical PTCs (67% (44/66) vs 87% (76/87), p = 0.003).

BRAF-mutated pediatric PTCs showed significant association with the presence of LN metastasis (98% (43/44) vs 81% (29/36), p = 0.014). However, when focusing on the pediatric PTCs with LN metastasis, extensive LN metastasis (pN1b) was paradoxically less frequent in BRAF-mutant PTCs than BRAF-wild-type PTCs: 65% (28/43) vs 90% (26/29), p = 0.016. Furthermore, all the three pediatric PTCs with distant metastasis were BRAF-wild-type (Table S2). Other three factors of sex, tumor size, and extrathyroidal invasion were independent from BRAF mutation status in both pediatric and adult PTCs.

TERT Promoter Mutations

Table 3 indicates TERT promoter statuses in both pediatric and adult PTCs. All of the pediatric PTCs and all but six adult PTCs were informative for TERT promoter mutation status. None of the 81 pediatric PTCs exhibited TERT promoter mutations, while 13% (11/85) of adult PTCs harbored the mutations (p = 0.001). Among the 11 TERT-mutated adult PTCs, C250T and C228T accounted for 27% (3/11) and 73% (8/11), respectively. All the TERT promoter mutations in adult PTCs were heterozygous. The age of the youngest and oldest patient with TERT-mutated PTC was 61 and 96 years, respectively. Furthermore, all these TERT-mutated PTCs (11/11) harbored BRAF mutation (Fig. 2). As indicated in Table S4, the adult PTCs with TERT promoter mutation were associated with an older age at operation (73.3 ± 9.8 vs 55.7 ± 10.3, p < 0.001) and UICC stage IV (46% (5/11) vs 14% (10/74), p = 0.019).

Fig. 2
figure 2

A representative adult PTC with BRAF and TERT promoter mutations is shown. a Sanger sequencing shows heterozygous T1799A (V600E) mutation in BRAF (arrow). b Heterozygous cytidine-to-thymidine transition at chr5, 1,295,228 (TERT promoter mutation C228T) is identified (arrow)

Full size image

Mutation Analysis in a CMV-PTC

In the pediatric PTC group, there was one CMV-PTC. The patient was an 18-year-old female without history of familial adenomatous polyposis. In genotyping, no BRAF, CTNNB1, PIK3CA, or TERT promoter mutation was evident in the CMV-PTC.

Discussion

We reviewed and investigated 81 PTCs from patients aged 20 years or less and compared them to 91 adult PTCs. We found that BRAF V600E-mutated tumors accounted for 54% of overall pediatric PTCs. Although reported proportions of BRAF V600E in pediatric PTCs have varied widely from 0 to 36.8% [11], our current study identified relatively higher frequency of BRAF mutation in Japanese pediatric PTCs. There may be several potential causes for this observation. First, the inclusion criteria for pediatric PTC can affect the BRAF mutation frequency. Our definition of pediatric PTC was PTC in a patient of 20-year-old or younger, while some other studies set the cutoff age of 15 or 16 years. Supporting this hypothesis, as shown in the present study, an older patient’s age was significantly associated with higher BRAF mutation frequency even in pediatric PTCs. Second, racial and/or geographical factors may be another cause. Comparing with the reported data in USA and Europa, the frequency of BRAF mutation in adult PTCs is quite higher in East Asia, particularly in South Korea [35]. In fact, the frequent BRAF mutation of 85% in our adult PTCs is well concordant with previous data from East Asia. Hence, as well as adult PTC, the relatively higher BRAF mutation in Japanese pediatric PTCs may stem the racial and/or geographical factors.

To evaluate the impact of internal exposure after the Fukushima accident, it is important to compare our findings with the recently published study of Japanese pediatric PTCs detected by ultrasound screening program in Fukushima [27]. The Fukushima study reported slightly higher BRAF V600E prevalence in pediatric PTCs: 64% (43/67) in Fukushima vs 54% (44/81) in our study, although the difference is not statistically significant. This minor discordance in BRAF V600E frequency can be largely attributed to the difference of how to detect pediatric PTCs. The Fukushima study investigated ultrasonographically screened pediatric PTCs, while our study enrolled clinically symptomatic pediatric PTCs. This may bias the proportion of histological type in pediatric PTCs. Indeed, the Fukushima study had no SV/SFV- and DSV-PTCs that are clinically aggressive and negative for BRAF mutation, while our study enrolled eight and four cases of SV/SFV and DSV, respectively. When comparing with pediatric classical PTCs of Fukushima and our study, the difference of the BRAF V600E frequencies becomes much smaller: 70% (43/61) in Fukushima vs 67% (44/66) in our study. Collectively, the BRAF mutation status is quite similar between our non-radiation-associated Japanese pediatric PTCs and pediatric PTCs in Fukushima. This finding suggests no alternations in driver point mutation after the Fukushima accident and implies non-radiogenic etiology of PTCs in Fukushima.

BRAF V600E showed significant predilection to occur in classical PTCs. More recently, some studies have reported that classical morphology of PTC such as papillary configuration, typical nuclear features, and infiltrative growth is morphological hallmarks of BRAF mutation in adults [36–38]. Our findings confirmed the genotype-phenotype correlation even in pediatric PTCs. However, it should be noted that the frequency of BRAF mutation in pediatric classical PTCs was still lower than adult classical PTCs. This finding suggests that it is difficult to attribute the lower BRAF mutation frequency of pediatric PTCs to the lower fraction of classical PTCs in pediatric PTCs and implies that pediatric classical PTC is a distinct entity from classical PTC of the adult.

BRAF V600E mutation status was associated with the presence of LN metastasis in pediatric PTCs. However, paradoxically, the extensive LN metastasis was observed more frequently in pediatric PTCs without BRAF mutation. Furthermore, although not statistically significant, all the pediatric PTCs with distant metastasis were negative for BRAF mutation. In adult PTCs, there was no association between BRAF mutational status and aggressive risk factors. Collectively, in our case series, BRAF mutation may have less predictive value for clinical aggressiveness in both pediatric and adult PTCs. The clinicopathological significance of BRAF mutation and its prognostic impact have been widely discussed. A number of studies found BRAF mutation to be associated with worse prognostic factors including larger tumor size, extrathyroidal invasion, LN and distant metastases, and recurrence-free survival [39–41]. However, some reports failed to confirm its clinicopathological significance [42–44]. For instance, Cheng et al. reported that BRAF mutational status did not provide independent risk estimates within morphologically divided subgroups such as classical PTC and FV-PTC [45]. Li et al. also demonstrated that BRAF mutation is associated with aggressive clinicopathological factors not in classical PTCs but in FV, TCV, and in PTCs overall [44]. Based on this information, we can attribute the lack of clinicopathological significance of BRAF mutation in our adult PTCs to the abundance of classical PTC in this group. Because classical PTCs accounted for more than 90% of our adult subset of PTCs, the prognostic impact of BRAF mutation could be diminished by this histological bias. Focusing on pediatric PTCs, the absence of clinicopathological significance of BRAF mutation was also reported in several smaller studies in the United States, Europe, and Japan [8–11, 19, 20]. We confirmed these observations in our study with the largest patient series so far and demonstrated that the BRAF mutation is not a predictor of aggressiveness and adverse prognosis of pediatric PTCs.

The lack of TERT promoter mutations in pediatric PTCs is one of the pivotal findings of the current study: none of our pediatric PTCs exhibited TERT promoter mutations, whereas 13% of adult PTCs showed the mutations. Furthermore, the TERT-mutated adult PTCs were identified in significantly elder patients. Notably, Liu et al. demonstrated the close association between TERT promoter mutation and shorter telomere length of PTC cells, a biomarker of cell aging [23]. Taken together, our study with pediatric PTCs enhanced the evidence of the age dependence of TERT promoter mutations and suggested that TERT promoter mutation is a result of the aging tumor cells. The absence of TERT promoter mutations can have some implications on the prognosis of pediatric PTCs. As described above, our study showed that pediatric PTC is usually detected in advanced presentation. However, the prognosis of pediatric PTC has been reported to be generally favorable, and a critical case is quite rare [3, 7]. Since TERT promoter mutations are an independent indicator of a worse prognosis, the absence of TERT mutation rather than the presence or absence of the BRAF mutation status in pediatric PTCs may explain the paradox between clinical aggressiveness and overall survival in pediatric PTCs. Nonetheless, there were three pediatric PTCs with distant metastasis (one DSV- and two SFV-PTCs), and all they were negative for BRAF or TERT promoter mutation. This observation indicates that genetic biomarkers for clinical aggressiveness may be different between pediatric and adult PTCs. In pediatric PTCs, therefore, further exploration for a genetic biomarker should be done using advanced methods such as whole-exome-sequencing.

In conclusion, we examined a large series of Japanese pediatric PTCs and defined their clinicopathological, histopathological, and genetic features. In comparison to adults, pediatric PTC is characterized by more advanced clinicopathological features, a lower frequency of BRAF mutation, and complete absence of the TERT promoter mutations. BRAF V600E was significantly associated with classical PTC histology but did not affect the clinical aggressiveness. The lack of TERT promoter mutation may explain the favorable prognosis of pediatric PTCs. These findings outline the nature of radiation-independent pediatric PTCs in Japan.