Several studies have evaluated the association of patient factors, such as sex and age, and disease outcomes with conflicting results (23,24,32,37-40). Certain studies found an association between younger age and the risk of persistent disease (23,38,40), whereas others have failed to find such an association (24,32,37,39). Males are more likely to have a poorer prognosis based on certain studies (24,39,40); however, other studies did not confirm these findings (23,32,37,38). The majority of studies showed no association between tumour size, histological type or extrathyroidal invasion, with a risk of persistent disease (23,24,37-40), but conflicting results have been reported on multifocality and tumour staging (24,32,37-40). Nevertheless, the majority of studies have shown an association between lymph node and distant metastasis with persistent disease (24,32,37,39). Of note, a study that evaluated prognostic factors in a population of 65 patients with DTC under the age of 20 years showed that lymph node and distant metastases were the only predictors for persistent disease (37). However, Mihailovic et al (38) observed different results in a population of 51 patients with DTC of the same age group. They found that diagnosis at a younger age, less radical primary surgery and tumour multifocality were also strong predictors for disease recurrence.

Different risk stratification systems combining several risk factors have been proposed to predict the outcome of patients with DTC. In general, these systems aim to estimate recurrence risk and mortality, guide follow-up and treatment, and ensure effective communication with patients and diverse professionals while permitting benchmarking (36,41). However, the current systems have limitations, particularly for paediatric patients. Factors affecting disease recurrence/persistence and survival prediction are distinct. Additionally, these tools have poor performance in predicting outcomes for patients in the early stages of disease, considered low risk (primarily stages I and II), which comprise the majority of patients with DTC (26,42). Moreover, they employ only information regarding disease presentation, but do not incorporate response to treatment and have not been validated for several populations, including paediatric patients (41). As illustrated in Table I and discussed below, there is conflicting evidence on the performance of these prognostic factors in children and adolescents (23,24,32,37-40).

The TNM/AJCC staging system is the most commonly used staging system, and is recommended by the ATA DTC paediatric guidelines (13,36). This system is focused on predicting mortality. It includes as variables the age of the patient at diagnosis (stratified around 55 years), the size of the tumour, and the presence of lymph node and distant metastases. Adult patients are classified into four stages, with a progressive decline in survival for stages I, II, III and IV. Due to the age at which patients are stratified, children and adolescents are classified only in stages I and II (with or without distant metastases, respectively), limiting the discriminatory factor in determining the prognosis for this population. Patients classified as TNM/AJCC I have a survival rate close to 100% (43).

The primary criticisms of this system are the lack of inclusion of variables known to influence the evolution and prognosis of patients such as, histological type/subtype and treatment-related data, and its inability to predict outcomes other than mortality (such as recurrences and persistent disease). The TNM/AJCC is updated periodically, and the 8th edition is the most recent version (Table II) (44).

Since DTC mortality rates in children and adolescents are very low, systems that capture the likelihood of relapse or persistent disease in the long-term follow-up are essential for defining the therapeutic strategies in this population. The ATA risk stratification incorporates a system that addresses the risk of persistent cervical disease and identifies which patients should undergo imaging to assess the presence of distant metastases (13). In this system, the patient is categorized into three levels of risk: Low, intermediate or high (Table III). However, its value is limited since it only considers histopathological data and does not consider the response to therapy.

The classification systems based on clinicopathological features use information from the patient's initial assessment for categorization of risk, without changes in this classification over time (13,44). The use of response to initial treatment has been advocated to estimate the risk of recurrence and death (45-50). This new modality risk stratification was termed DRS, based on the observation that a patient's risk may change over time, according to new data gathered during follow-ups (45). In this system, patients are classified into four categories: Excellent, biochemical incomplete, structural incomplete and indeterminate response (Table IV) (46,47).

The utility of DRS has been shown in several DTC cohorts (46-48). A study by Vaisman et al (48) showed that patients with an excellent response after the initial therapy had a risk of only 1.4% for persistent/recurrent disease (48). Conversely, amongst patients with persistent structural disease, only 9% were classified as excellent response, even after several additional therapies.

However, whilst DRS has been validated in the adult DTC population, the assessment of its role in children and adolescent management is still limited (49,50). Indeed, the current ATA guidelines for children with DTC do not suggest the use of DRS for children (13). Lazar et al (49) evaluated DRS in a cohort of 54 patients with a median age at diagnosis of 13.9 years and a median follow-up of 8.8 years. They found that patients classified as having an excellent response after the initial treatment presented a favourable prognosis: 82.9% of them remained classified as excellent at follow-up. Conversely, all patients with an incomplete response after the initial therapy remained with persistent disease. Sung et al (50) recruited a cohort of 77 paediatric patients with DTC and demonstrated that DRS was useful in predicting disease outcome at follow-up. When compared to the group with an excellent response, the risk of persistent/recurrent disease was significantly higher in patients with an indeterminate or incomplete structural response. Recently, our group conducted a multicentre study involving four institutions to evaluate DRS in children and adolescents (32). A total of 66 patients with a diagnosis of DTC before 18 years of age were included. In this study, a multivariate analysis including tumour size, lymph node and distant metastasis, ATA paediatric risk stratification and DRS was performed. The results showed that DRS was the only predictor of persistent/recurrent disease, with odds ratios (confidence intervals) of 35.2 (3.7-762.5), 54.9 (2.5-3,933.1) and 13.9 (1.1-313.7) for indeterminate, biochemically persistent and structurally persistent disease, respectively.

For the majority of patients, the initial postoperative evaluation is performed ~3 months after surgery (13). This assessment aims to evaluate persistent locoregional disease and identify patients who may benefit from RAI dosing, such as those with known or suspected distant metastases (13). Low-risk patients should undergo thyroglobulin (Tg) measurement using levothyroxine (Tg-T4) and cervical ultrasound. In turn, in patients at intermediate and high risk, the addition of stimulated Tg (sTg) for improved risk stratification and determination of the need for RAI treatment is useful. Thus, a more individualized and conservative approach to treatment and postoperative staging can reduce unnecessary exposure to RAI in children with no evidence of disease, in whom the risks of routine therapy with RAI probably outweigh the benefits. Additionally, certain patients will require additional imaging techniques, such as neck and chest computed tomography (CT), especially those with detectable Tg-T4. The value of PET/CT has been poorly studied in this population and is not routinely recommended for children (13).

Serum Tg levels serve as a marker of recurrent disease, and ultrasensitive serum Tg assays are considered the most sensitive method for the detection of residual thyroid cancer (13,51). Measurement of serum Tg levels is critical for the management of paediatric patients with DTC, both at the initial postoperative staging and during long-term follow-up (13). Therefore, monitoring Tg under levothyroxine therapy (Tg-T4) is the ideal approach to evaluate disease recurrence or progression (13). Interestingly, the Tg levels may be higher in children than in adults with a similar extent of disease (52).

The role of sPOTg as a prognostic factor for DTC in the paediatric population has been recently addressed by several studies. The first study included 32 children and adolescents diagnosed with DTC <18 years old and found that the ideal cut-off value for the prediction of excellent response was 31.5 ng/ml, with a sensitivity and specificity of 100% (53). Similar results were observed in a larger sample of 66 young patients: A cut-off of 37.8 ng/ml showed 81% sensitivity and 100% specificity (32). More recently, a Chinese study with 118 paediatric patients (<20 years old) evaluated the prognostic factor of pre-ablation sPOTg and found that the ideal cut-off to predict disease-free status was 17.8 ng/ml, with a negative predictive value of 96.8% (54).

TgAc are present in ~25% of patients with DTC, and their positivity may determine laboratory interference with Tg measurement (55). As the concentrations of TgAc respond to changes in circulating Tg antigen levels and thus indirectly represent changes in thyroid tissue mass, TgAc levels may serve as a surrogate tumour marker for DTC (13,55). As a result, it is recommended to evaluate TgAc levels in all patients with DTC during their follow-up (56).

Most studies in adult populations have reported that recurrence, persistence, or a rising trend in postoperative TgAc concentrations are significant risk factors for persistent or recurrent disease (57). However, it is not known whether a positive TgAc value correlates with disease extension/invasiveness or prognosis (58). A decline in TgAc levels suggests a decreasing disease burden, considering an average of 3 years to eliminate TgAc after cure of DTC (59). A significant increase in TgAc may indicate disease progression, and this should be assessed in more detail. Similar to Tg measurements, the trend in TgAc concentrations is more relevant for disease detection than a single determination (58).

The proto-oncogene RET, located on chromosome 10q11.2, encodes a tyrosine kinase receptor (89,90). At least 12 types of RET/PTC rearrangements have been described, with types 1 and 3 being the most common (89,90). In the paediatric population, RET/PTC mutations are the most common type of mutations, ranging from 15-77% based on different studies (65, 67-70,72,73,76,78,80-83,88).

Several studies have examined the role of RET/PTC as a prognostic factor in paediatric DTC patients. Whilst the majority of studies have failed to demonstrate an association (67,68,72,76,78,80,82), a recent Brazilian study reported an association between RET/PTC3, larger tumour size and multifocality (83).

BRAF kinase, whose encoding gene is located on chromosome 7, is the most potent activator of the MAPK pathway (89,90). Over 40 mutations of the BRAF gene have been identified, with the T1799A mutation being the most common (89,90). This missense mutation, due to a somatic transversion of thymine to adenine at position 1,799 in exon 15, results in the substitution of a valine amino acid for glutamic acid at position 600 (BRAFV600E). In children and adolescents, this is the second most prevalent mutation, found in ~28% of cases, with prevalence ranging from 0-68% (68-88).

The association of the BRAFV600E mutation with disease outcome in paediatric patients is still controversial. Alzahrani et al (77) evaluated 55 children and adolescents with DTC and found that persistent/recurrent thyroid cancer was more prevalent in patients with the BRAFV600E mutation (66.7 vs. 34.1%) and more pronounced in patients with classic PTC (77.8% vs. 33.3%). Onder et al (79) observed that the classic architecture with multicentricity and local recurrence was correlated with BRAFV600E mutation (79). In contrast, several studies found no association between BRAFV600E mutation and disease prognosis (68,71,72,74-76,80,82,85-87).

The NTRK1 receptor gene, located on chromosome 1, encodes the high-affinity nerve growth factor receptor and is activated via the MAPK pathway (73). ETV6-NTRK3 is the result of an interchromosomal translocation (12; 15) (p13; q25) that juxtaposes exons 1-4 of ETV6 to exons 12-18 of NTRK3 (73). This gene has recently been studied and is gaining importance due to its high prevalence (~12%, ranging from 7-26%) being the third most common in the paediatric population. Moreover, studies have shown an association between NTRK fusions and worse clinical outcomes (73,76,81-83). Compared with BRAF mutations, the presence of fusion genes has been associated with larger tumours (2.2 vs. 1.5 cm), aggressive histology (84% vs. 0%), and lymph vascular invasion (92.3% vs. 46.1%) (76).

RAS genes encode highly related G proteins, which serve a central role in intracellular signal transduction by activating the MAPK and other signalling pathways, such as PI3K/AKT (89,90). Amino acid modifying mutations of RAS generally occur at codons 12, 13 or 61 of H-RAS, K-RAS or N-RAS proteins (89,90).

Mutations in the RAS gene were the first studied in the DTC paediatric population. RAS mutations are much less prevalent in paediatric patients than in the adult population, with an estimated rate of 2.7% (prevalence range, 0-16%) (66,68,69,72,73,76,78,80-83,87). No associations have been reported between RAS mutations and disease presentation in paediatric DTC (66,72,78,82,87).

Despite the excellent prognosis of DTC in paediatric patients, a small subset of this population may show progressive and RAI refractory disease (13-17). In such a case, systemic therapy should be considered (13). Identifying tumour molecular profiles may be critical in selecting the most appropriate therapy (Fig. 2) (91). Of note, the majority of the current knowledge of targeted kinase inhibitors in this population is based on case reports and anecdotal clinical experience (8,13). In addition, long-term side effects in the paediatric age group are unknown (91). Two drugs have been approved by the Food and Drug Administration (FDA) for DCT refractory disease, namely, sorafenib and levantinib, although several other drugs are in clinical trials (8,92,93). Anti-neoplastic therapy in children should be performed in centres experienced with the use of these therapeutic agents in paediatric patients (13).

Sorafenib therapy has been reported in three patients with lung metastatic and progressive disease (14-16). The first case was a 14-year-old girl who experienced a significant reduction in lung metastasis after 2 months of sorafenib therapy, although with side effects, such as cutaneous toxicity and neutropenia (14). The second patient was an 8-year-old boy with hypoxaemia and a need for mechanical ventilation (15). The patient was weaned off mechanical ventilation, and CT showed regression of the pulmonary metastasis after 2 months of therapy. The third case was an 11-year-old boy who showed stable disease after 24 months of sorafenib use (16).

Treatment with levantinib has been used in a small series of paediatric patients with extensive bilateral metastatic pulmonary disease, including one patient who previously used sorafenib (17). All three patients had respiratory distress requiring oxygen therapy. After a few weeks of treatment with levantinib, all patients were successfully weaned off oxygen. The drug was well tolerated, and proteinuria was the only major adverse effect. Two patients had stable disease at 11 and 23 months after the initiation of levantinib. The third patient switched treatment to a tumour-specific target.

More recently, larotrectinib, a highly selective inhibitor of tropomyosin receptor kinase, was approved by the FDA for patients with solid tumours harbouring NTRK fusions in adult and paediatric populations (94). The drug was tested in a phase I/II clinical trial, which included 24 children with solid tumours and two with PTC. These two patients showed stable disease for >7 months of follow-up (94,95). In another phase II study, 55 adolescent and adult patients were included, with five diagnosed with thyroid cancer. Of these, four patients presented with a partial response, and one showed complete response (94).