The French FRACTURE database: A way to improve knowledge on management of children with very rare tumors

About 2200 new cases of child and adolescent cancers are diagnosed every year in France. They are systematically registered in the French National Registry of Childhood Cancer (Registre National des Cancers de l'Enfant - RNCE).¹ Overall, 8%–10% of these malignancies are considered “very rare tumors” (VRTs), with an annual incidence of less than two per million inhabitants.² VRTs in children and adolescents represent a heterogeneous group of cancers. Due to their extremely low incidence, knowledge on these tumors is scant and they are the subject of only very few prospective studies; for example, a nasopharyngeal carcinoma study in Germany.³ Consequently, therapeutic recommendations are difficult to produce, and physicians have to treat VRTs on an individual basis with weak evidence-based medicine support.⁴ In order to face these difficulties, over the foregoing years, the European Pediatric Oncology Community has developed projects specifically dedicated to VRTs.⁵ The main types of tumors anticipated were undifferentiated nasopharyngeal carcinoma (UCNT), solid pseudopapillary neoplasms of the pancreas (SPPP), pleuropulmonary blastoma (PPB), carcinoid tumor, pheochromocytoma/paraganglioma, adrenocortical tumors (ACT), thyroid carcinoma, and cutaneous melanoma. Some VRTs such as PPB arise exclusively in pediatric patients, while other tumors develop generally during adulthood but are rarely seen in children and adolescents (i.e., colon cancer and malignant melanoma). These rare malignant tumors are less known to pediatricians and even pediatric oncologists. However, in recent years, the European EXPeRT group (European Cooperative Study Group for Pediatric Rare Tumors) published some specific recommendations to help clinicians in treating patients with these rare entities.^5-12 The PARTNER (PAediatric Rare Tumours Network - European Registry) project aims to extend knowledge of theses rare diseases and to build a single European registry that links the existing national registries dedicated to children and adolescents (0–18 years) with VRTs. European countries, where no VRT registry exists, should register their patients directly in this new European registry.⁵ In France, the Very Rare Tumor Committee (FRACTURE) of the Société Française de lutte contre les Cancers et les leucémies de l'Enfant et de l'adolescent (French Society of Pediatric Oncology - SFCE) was founded in 2007 with several objectives.⁵ Its main aim was to record as many childhood and adolescent cases of VRTs as possible on a national scale, using the FRACTURE database, a national collection database open since 2012. The purpose of the database was to facilitate analysis of these rare diseases, mainly in order to develop harmonized dedicated guidelines for diagnosis, treatment, and follow-up.^13-16 The number and heterogeneity of these pathologies, the variety of their age of onset, and the various sites of care (surgery, oncology, dermatology, ophthalmology, endocrinology, etc.) make it difficult to systematically register these patients. To date, very few countries have reported data from their national database or experience with rare pediatric tumors (Germany, Italy, Brazil). Here, we present the French experience, which represents one of the oldest and largest databases.

The main objective of this article is to describe the cases of VRTs registered in the FRACTURE database and to perform a detailed analysis for the most common entities, describing their treatments and outcomes. The secondary objective is to compare these data with those registered in the French RNCE to analyze the rate of effective registration in the dedicated VRT database and to explore the reasons of discrepancies.

1 MATERIALS AND METHODS

This retrospective study analyzed all data recorded in the FRACTURE database between 2012 and 2018 and compared them to data available in the RNCE during a 4-year period (2012–2015). The FRACTURE database is a prospective voluntary repository that since 2012 has collected clinical, biological, and therapeutic information on patients from 0 to 18 years old treated in one of the 32 SFCE centers for very rare malignant and borderline tumors. The inclusion criterion was a very rare malignant tumor or a very rare tumor of intermediate malignancy occurring in a person younger than 18 years old. VRTs were described as tumors with an annual incidence of less than two cases per million inhabitants, as defined by Ferrari et al.² Central nervous system tumors and hematological malignancies were excluded, as were other tumors within the field of another SFCE working group or registered in a specific clinical trial: rhabdoid tumors, hepatoblastoma, retinoblastoma, malignant germ cell tumors, and rare histotypes of soft part sarcomas. The purpose of the FRACTURE database is not to provide extensive epidemiological analysis, as the voluntary nature of the report was not expected to ensure its completeness. The informed consent of the family and/or the patient is collected in each center by the clinician. Data are anonymously collected in a secured electronic database (EnnovClinical software) housed in the Angers University Hospital. Authorization from the National Commission for Data Protection and Liberties (CNIL-France) was obtained in October 2011. At regular points, a clinical update is performed to collect the last follow-up information. The validation of the data is carried out by the clinical research staff of the University Hospital of Angers. For this study, data were extracted on January 31, 2019, with an update on December 31, 2021.

All data were analyzed using descriptive statistical methods. Tumor failure was defined as relapsed disease, progressive disease (PD), or death. Locoregional relapse was defined as recurrence at the primary thoracic site or in directly adjacent thoracic tissues, and metastatic relapse was defined as appearance of tumors at distant sites. Survival time was calculated from the date of diagnosis (initial biopsy/surgery) to the time of last follow-up or event. Overall survival was calculated by the Kaplan–Meyer method. R software (version 3.5.1) and R packages wordcloud2 (version 0.2.1) and ggplot2 (version 3.0.0) were used to create the figures.

The RNCE is a population-based registry that ensures an exhaustive retrospective collection of all cancer cases in children under 18 years of age in mainland France and four overseas departments.^{17, 18} Based on the previous French RNCE data, the initial annual expected number of VRT cases was between 150 and 200. The main source of collection is active research by trained investigators of pediatric hematology and oncology departments in France. Active registration is supplemented by an annual request for claim data from the Information System Medicalisation Programme (PMSI) of University Hospitals (CHU) and Cancer Centers, which lists all children aged less than 15 years who have been admitted to any department with a diagnosis of cancer or a brain tumor, including a few borderline diagnoses. This source of information enables identification of children with cancer treated in departments other than pediatric oncology departments. Systematically recorded data include child characteristics, information on cancer diagnosis and patients’ pathway from diagnosis to treatment, initial treatment, vital status, and last contact date. The RNCE produces incidence and survival estimates on a national scale.¹⁹ The RNCE is evaluated by the Registry Evaluation Committee, which certifies the quality and completeness of the collected data. At the time of the analysis, RNCE data were exhaustive up to 2015, so the selected period for the comparison with the FRACTURE database was 2012–2015. This comparison was made on August 31, 2019. Moreover, as some VRTs are actually nonmalignant tumors, we widened the selection of the RNCE to include borderline tumors that are also recorded in the registry but not included in the incidence figures. Databases were combined using patients’ initials, date of birth, date of diagnosis, and type of tumor. Pediatric VRTs represent a very heterogeneous group of tumors, with difficult classifications for some entities. To simplify the comparative analysis complicated by this great heterogeneity, we decided to consider for analysis only tumors for which more than 15 cases were reported (RNCE alone, FRACTURE alone, or in both) over the period 2012–2015.

For the classification of patient treatment sites, the term “specialized pediatric oncology unit” corresponds to pediatric oncology units in university hospitals and pediatric oncology units in cancer centers.

2 RESULTS

2.1 Description of the FRACTURE database

Overall, 477 patients fulfilling the inclusion criteria were included in the FRACTURE database, accounting for 97 different tumor histological types (Figure 1, Table 1) over a period of 7 years (January 1, 2012 to December 31, 2018). Total 258 patients were female (58%) and 185 were male (42%) (unknown: 34 cases). The median age of patients at diagnosis was 12 years (range: 0.0–18.9). Many different entities were included, regardless of the age of the patient. Two peaks of occurrence are present: before the age of 3, and after the age of 10 years (Figure 1).

TABLE 1. Number of cases included in the very rare tumor FRACTURE database for each tumor type between 2012 and 2018

Histotypes	Number of cases
Neuroendocrine tumors: Appendiceal Lung Pancreas Liver	66 54 9 2 1
Thyroid carcinoma Papillary thyroid carcinoma Follicular thyroid carcinoma Medullary thyroid carcinoma	52 39 9 4
Nasopharyngeal carcinoma UCNT Nasopharyngeal papillary carcinoma	44 43 1
Pancreatic tumors Solid pseudopapillary neoplasm Acinar cell carcinoma Pancreatoblastoma Undifferentiated carcinoma	38 35 1 1 1
Pulmonary and pleural tumors Pleuropulmonary blastoma Mucoepidermoid carcinoma NUT carcinoma Adenocarcinoma Epidermoid carcinoma Mesothelioma Pulmonary chondroid hamartoma	35 25 3 3 1 1 1 1
Pheochromocytoma/paraganglioma Pheochromocytoma Paraganglioma	34 21 13
Cutaneous tumors Melanoma Basal cell carcinoma Solitary cutaneous mastocytoma	28 26 1 1
Adrenocortical tumor	27
Salivary glands tumor Mucoepidermoid carcinoma Acinar cell carcinomas Undifferentiated carcinoma Mammary analog secretory carcinoma Salivary gland analog tumor Adenoid cystic carcinoma Sialoblastoma	27 14 7 2 1 1 1 1
Hemangioendothelioma	14
Ovarian non-germ cell tumor Mucinous cystadenoma Carcinoma Adenocarcinoma Serous cystadenoma Sclerosing stromal tumor Steroid cells tumor	13 5 2 2 2 1 1
Gastrointestinal carcinoma	12
Kidney tumors	10
Giant cells tumors	9
Thymic tumors	8
Gastrointestinal stromal tumor	5
Others rare tumors	55
Total	477

• Abbreviations: FRACTURE, French Very Rare Tumors Committee; UCNT, undifferentiated nasopharyngeal carcinoma.

3 DISCUSSION

In recent years, a number of countries have set up their own registries for very rare pediatric tumors in order to better understand these diseases.^20-23 The Italian TREP project (Project on Rare Pediatric Tumors) was the first of these when it was launched in 2000.²¹ Overall, 297 patients were registered in this database between 2000 and 2005. Our French experience shows a similar distribution of the different tumors with a median age among those patients registered of about 12 years. The German Pediatric Rare Tumor Group (STEP) was founded in 2006.²³ Between 2008 and 2018, 623 patients with VRTs were registered in STEP registry.²² The inclusion criteria and the type of tumors registered are very similar to those of the FRACTURE database.

Registration in the FRACTURE and the RNCE databases is mainly carried out in pediatric oncology departments. Even though both databases register a significant number of cases each year, there is a significant lack of comprehensiveness in the FRACTURE database (53 cases per year) when compared to the RNCE (193 cases per year), with 74% of patients exclusively registered in RNCE, 7% in FRACTURE, and 19% in both databases. There are several possible explanations for this difference, including the fact that registration in the FRACTURE database is self-reported by a physician with no specific funding for these inclusions, whereas the RNCE provides funding for time spent in the centers. A dedicated team of clinical research associates proactively collect and cross-check data from many sources, such as medical procedure coding systems and pathology laboratories. Finally, registering patients and providing regular follow-up is a time-consuming process for clinicians receiving no financial support for this extra effort. However, and despite all these difficulties, nearly 500 patients were registered in a 7-year period with a large amount of medical data, thus enabling retrospective studies to improve medical knowledge on such rare entities. However, the objectives of these two databases are different. The RNCE is more exhaustive, as it is a systematic publicly funded collection of data, the completeness of which is checked by using the Information System Medicalization Programme.¹⁷ It provides essentially epidemiological data. In contrast, the FRACTURE database provides clinical data available for clinical studies and patient treatment. Several retrospective studies have already been performed with these medical data, improving knowledge of some rare entities.^{11, 13, 24-26} A number of these retrospective studies were performed with the collaboration of other European VRT groups that registered their own patients and within the framework of EXPeRT group.^{9, 27} Finally, all of these studies allowed the EXPeRT group to build and publish care recommendations for the treatment of such rare diseases.^{5, 8}

One specific aim of the national registry is to improve knowledge about cancer epidemiology. Our analysis confirms that VRTs represent a total of nearly 200 new cases per year in France, that is, about 10% of all new cases of tumors registered each year in children and adolescents. This constitutes a significant proportion of cases, therefore justifying the existence of specialized working groups and committees to treat and follow-up these patients as best as possible.

To increase the number of inclusions in the FRACTURE database, more complete information should be provided to pediatric oncologists and surgeons should be better informed. Multidisciplinary consultation meetings also serve as a reminder of the existence of the database and raise awareness of the inclusions.²⁸ However, our study confirms that patient registration in the FRACTURE database is mostly performed by pediatric oncologists, probably because this database was initially developed within the framework of the SFCE. It appears more difficult to regularly remind all physicians treating these rare tumors, particularly in units that take care of types of cancer typically seen in adolescents or adults (like thyroid carcinoma or melanoma), especially in non-university hospitals.

This study also highlighted that many pediatric patients treated for VRTs are not directly managed by pediatric hematologists/oncologists. Therefore, patients with pediatric thyroid cancers are generally treated by dedicated endocrinologists, patients with cutaneous melanomas by adult dermatologists, those with NETs by surgeons, and so forth. In these situations, pediatricians may not be involved in the treatment of such patients, which is a matter of concern for several reasons. First, in some cases, the same pathology may require different managements between adults and children; for example, hemicolectomies are not indicated in the treatment of NETs of the appendix in childhood,²⁹ or radiotherapy dosages might sometimes be reduced for teenage and young adult patients with UCNT after a favorable tumor response to initial chemotherapy.^{8, 14} Second, pathologies can differ in their biology (e.g., childhood melanomas^{25, 30}) or etiology (e.g., thyroid carcinoma and DICER1 syndrome^{31, 32}) in pediatric cases and therefore need to be considered specifically in the light of the patient's age. Pediatric oncologists seem to be solicited mostly to manage patients with the most complex cases, as suggested by the proportion of metastatic forms of melanoma, which is three times higher in the FRACTURE database than in the literature (15% vs. 5%).³⁰ Previous studies have shown that children and adolescents with metastatic melanomas are more often managed in pediatric oncology departments than other patients with more localized melanomas.³³

4 CONCLUSIONS

This study confirmed the great diversity of VRTs in children, which nevertheless represents 10% of all pediatric cancers diagnosed each year. After 10 years of operation, the main goal of the FRACTURE database has been achieved. Enough patients have been registered to enable the database to contribute to all collaborative studies undertaken with other European countries and for researchers to perform collaborative retrospective studies. As a result, several collaborative national and international studies have already been finalized using these data, and have even led to recommendations on the diagnosis and treatment of some VRTs, such as pancreatoblastoma or ACC.^{7-9, 27, 30, 34} The sustainability of this collection is therefore important, and a regular systematic collaboration between the FRACTURE database and the RNCE register helps to provide a more exhaustive picture of these VRTs and allows research completeness for specific groups of patients.

Overall, 39.1% of children and adolescents with these VRTs were not treated in or linked to a specialized pediatric oncology unit. Communication on these rare pathologies and their specificities according to the age of onset must be a major objective of the working groups on VRTs in children and adolescents.

CONFLICT OF INTEREST

The authors declare they have no conflicts of interest.

FUNDING INFORMATIONS

French FRACTURE database supported by “Enfant, cancers et santé”