A total of 144 pediatric patients with DTC who had undergone thyroid operations from August 1982 to December 2013 were retrospectively reviewed at Yonsei University Hospital (Seoul, Korea). Sixteen patients were excluded because they had inadequate follow-up data; thus, 128 patients were analyzed by complete review of medical charts and pathology reports. Among them, 40 (31.2%) had undergone bilateral total thyroidectomy (BTT), and 43 (33.6%) had undergone BTT with therapeutic modified radical neck dissection (mRND). These patients received subsequent radioactive iodine (RAI) ablation. The mean follow-up duration was 11.5 years (range, 2.1–29.8 years). This study was approved by the at Yonsei University Hospital Institutional Review Board (4-2015-1160).

The management protocol followed the ATA management guidelines [17]. Tg and TgAb concentrations were regularly measured, and physical examinations, neck ultrasonography and thyroid function tests were performed at 3 and 6 months and annually thereafter. Postoperative RAI treatment was performed 4 to 6 weeks after operation, and WBSs were performed 5 to 7 days after RAI ablation. All patients were prescribed a suppressive dose of L-thyroxine (thyroidstimulating hormone [TSH] < 0.05 mIU/L). Patients who had clinical suspicion of recurrence of disease or distant metastasis were evaluated by additional diagnostic imaging, including computed tomography, positron emission tomography/computed tomography, and/or RAI WBS. To confirm the recurrence of disease, ultrasound-guided fine needle aspiration biopsy was performed in addition to imaging modalities.

The 2015 ATA risk stratification system for pediatric patients was applied to the study subjects, dividing them into low, intermediate, and high-risk groups [17]. The low-risk group was defined as pediatric patients with DTC grossly confined to the thyroid (AJCC/UICC TNM staging system T1 to T3) with clinical N0/Nx disease or with pathologic N1a micrometastases (number of LNs ≤ 5 and largest dimension of metastatic foci < 0.2 cm). The intermediate-risk group was defined as pediatric patients with DTC accompanied with pathologic N1a metastases (number of LNs > 5 or largest dimension of metastatic foci ≥ 0.2 cm) or minimal N1b metastases (number of LNs ≤ 10 and largest dimension of metastatic foci < 3 cm). The high-risk group was defined as pediatric patients with DTC with pathologic N1b metastases (number of LNs > 10 or largest dimension of metastatic foci ≥ 3 cm), locally invasive disease (T4), or distant metastasis.

The DRS system classified pediatric patients into 4 groups, as shown in previous studies: excellent response, indeterminate response, biochemical incomplete response, and structural incomplete response [18192021]. The best response to the initial treatment was evaluated between 1 and 2 years after diagnosis of DTC with Tg level, TgAb level, and imaging findings. The definition of DRS classification according to the extent of surgery is shown in Table 1.

Patients with negative stimulated Tg, negative TgAb, and no evidence of structural disease at the final follow-up were defined as having no evidence of disease (NED). Biochemical-persistent disease in patients who underwent total thyroidectomy with RAI ablation was defined as negative imaging findings with suppressed Tg > 1 ng/mL or increasing TgAb level; in patients who underwent lobectomy, this was defined as negative imaging findings with nonstimulated Tg > 30 ng/mL or increasing Tg level with similar TSH level or increasing TgAb. Structural persistent/recurrent disease was defined as pathologically or cytologically verified malignancy and/or metastasis by imaging modalities regardless of Tg or TgAb level [20].

Continuous, quantitative variables are provided as mean with standard deviation. Categorical and qualitative variables are reported as number with percentage. Multivariate Cox regression analysis was performed to identify the independent risk factors of DFS. DFS was compared by Kaplan-Meier analysis with log-rank test between groups. The proportion of variance explained (PVE) value was calculated with the Cox regression model, using the following equation:

PVE = 1 − exp(-G²/n)

where G² is the maximum likelihood ratio from the Cox regression model, and n is the total number of patients [22]. Hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated, and 2-sided P-values of <0.05 were regarded as statistically significant. All statistical analyses were performed using R package version 3.1.3 (http://www.R-project.org).

Table 2 provides the baseline clinicopathologic characteristics of the 128 pediatric patients with DTC. The mean age of study patients was 16.5 years (range, 4–19 years); among them, 9 (7.0%) were younger than 10 years. A total of 114 patients (89.1%) were female, and the mean tumor size was 2.1 cm (range, 0.2–6.0 cm). A total of 103 patients (80.5%) were diagnosed with conventional papillary thyroid carcinoma (PTC), and 9 patients (7.0%) were diagnosed with follicular thyroid carcinoma (FTC). The remaining patients (12.5%) were diagnosed with other types of PTC. Multiplicity and bilaterality of carcinoma were found in 36 (28.1%) and 26 patients (20.3%), respectively. Seventy-six patients (59.4%) were identified with ETE. Forty patients (31.2%) underwent BTT, and 43 patients (33.6%) underwent BTT with therapeutic mRND due to N1b nodes. The numbers of patients with T1, T2, T3, and T4a were 29 (22.7%), 19 (14.8%), 74 (57.8%), and 6 (4.7%), respectively. Forty-six patients (35.9%) had cervical LN metastasis, and 43 patients (33.6%) had lateral LN metastasis. Only 6 patients (4.7%) were diagnosed as having distal metastasis of the lung at the same time as diagnosis of thyroid carcinoma. Loco-regional recurrence was detected in 18 patients (14.1 %), and recurrence with distant metastasis to the lung was detected in 5 patients (3.9%). The mean follow-up duration was 11.5 years (range, 2.1–29.8 years).

Using the ATA risk stratification system, 62 patients (48.4%) were classified into the low-risk group, 38 patients (29.7%) were classified into the intermediate-risk group, and 28 patients (21.9%) were classified into the high-risk group. Table 3 shows the clinical outcomes according to the ATA risk stratification system. Most of the patients (87.1%) in the low-risk group had NED, and only 1 patient (1.6%) was diagnosed with structural persistent/recurrent disease. On the other hand, 8 patients (28.6%) in the high-risk group were diagnosed with structural persistent/recurrent disease. The DFS of the ATA risk stratification system were significantly different (P = 0.041) (Fig. 1).

Using the DRS system, the patients were reclassified into 4 groups: excellent response, 63 (49.2%); indeterminate response, 37 (28.9%); biochemical incomplete response, 18 (14.1%); and structural incomplete response, 10 (7.8%). Table 4 presents the clinical outcomes according to the DRS system. Sixty-one patients (96.8%) in the excellent response group had NED, and only 2 patients (3.2%) had biochemical persistent disease. The majority of patients (78.4%) in the indeterminate group had NED, while 7 patients (18.9%) and 1 patient (2.7%) were verified as having biochemical persistent and structural persistent/recurrent disease, respectively. On the other hand, slightly more than half of the patients (55.6%) in the biochemical incomplete group were diagnosed with biochemical persistent disease, and 4 patients (22.2 %) were diagnosed with structural persistent/recurrent disease. Eight patients (80%) in the structural incomplete response group were diagnosed with structural persistent/recurrent disease, and only 1 patient (10%) had NED. The Kaplan-Meier survival analysis showed statistically significant differences among the response groups with the DRS system (P < 0.001) (Fig. 2).

The correlation between the ATA risk stratification system and the DRS system is shown in Table 5. Of the patients who had excellent responses at initial evaluation after initial therapy, 42 (66.6%) were in the low-risk group, while 7 (11.1%) were in the high-risk group. Twelve patients (66.6%) with biochemical incomplete response were in the intermediate-risk group, and all patients with structural incomplete response were in the high-risk group.

To compare the predictive accuracy of prognosis between the ATA risk stratification system and the DRS system, the PVE was calculated. The PVE value of the DRS system was 54.9%, which was higher than that of the ATA risk stratification system, 29.8%.

To determine the independent risk factors associated with DFS, a multivariate analysis was performed (Table 6). Tumor size over 1 cm (HR, 15.130; 95% CI, 0.997–197.19; P = 0.016) and N1b stage (HR, 7.914; 95% CI, 5.192–9.683; P = 0.001) were confirmed as significant predictors for recurrence. In the ATA risk stratification system, the risk of recurrence was significantly higher in the high-risk group (HR, 9.371; 95% CI, 0.283–450.487; P = 0.045) than in the low-risk group. The risk of recurrence in the DRS system was significantly higher in all other groups than in the excellent response group: indeterminate group (HR, 6.104; 95% CI, 1.258–29.613; P = 0.025), biochemical incomplete group (HR, 7.562; 95% CI, 1.131–50.539; P = 0.037), and structural incomplete group (HR, 44.487; 95% CI, 3.343–592.01; P = 0.004).