The impact of incomplete registration on survival rate of children with very rare tumors

Pediatric very rare tumors (VRTs) represent a heterogeneous subset of childhood cancers, with reliable survival estimates depending dramatically on each (un)registered case. The current study aimed to evaluate the number of VRTs among Lithuanian children, to assess the impact of the registration status on survival rates and to track changes in treatment outcomes over the 16-year study period. We performed a population-based retrospective study across children below 18 years old diagnosed with VRTs in Lithuania between the years 2000 and 2015. The identified cases were cross-checked with the Lithuanian Cancer Registry—a population-based epidemiology cancer registry—for the fact of registration and survival status. The overall survival was calculated in relation to the registration status and treatment period. Thirty-seven children with VRTs were identified within the defined time frame. Six of them (16.2%) were not reported to the Lithuanian Cancer Registry at diagnosis. The probability of overall survival at 5 years (OS5y) differed significantly between the registered (n = 31) and unregistered (n = 6) cohorts: 51.6% versus 100%, respectively (p = 0.049). A 5-year survival estimate for children diagnosed with a VRT at the age of 0–14 years differed by 10 percentage points according to the registration completeness: 52.1% calculated for the entire cohort versus 42.1% for registered patients only. The OS5y has not improved over the analyzed period: 61.1% in 2000–2007 versus 57.9% in 2008–2015 (p = 0.805). The survival continued to decline beyond 5 years post-diagnosis due to late cancer-related adverse events: 59.5% of patients were alive at 5 years as compared to 44.3% at 10 years. The OS5y of children affected by VRT was lower than in more common childhood cancers. The survival rate of the unregistered patients may lead to misinterpretation of treatment outcomes. Meticulous registration of VRTs is crucial for correct evaluation of treatment outcomes, especially across small countries with few cases.


Introduction
Malignant tumors in children are very rare: a recently calculated incidence of pediatric cancer in childhood was 155.8 cases per million children between the ages of 0 to 19 years (1). Although infrequent, the vast majority of pediatric cancers can be treated in international clinical trials or following treatment guidelines developed by expert groups. Despite the well-elaborated management strategies for the most common childhood malignancies, childhood cancer remains the second leading cause of illness-related mortality among children in developed countries (2).
Pediatric very rare tumors (VRTs) represent a particular subset of childhood cancers, comprising approximately 9 to 11% of all malignancies occurring in children below the age of 20, 75% of them diagnosed between the ages of 15 to 19 years old (3,4). The European Cooperative Study Group for Pediatric Rare Tumors (EXPeRT) defines a VRT as any solid malignancy or borderline tumor that has an incidence rate of ≤ 2 per million per year and/or is not considered for treatment in clinical trials (5). As a result, VRTs pose a real diagnostic and therapeutic challenge due to their rarity and lack of international treatment guidelines or ongoing clinical trials. Based on this observation, we initiated the current population-based study that aimed to evaluate the number of VRTs in Lithuanian children and to verify their registration status in the LCR as a potential predictive factor of the treatment outcome.

Study population and design
We performed a retrospective population-based study of children diagnosed and treated for VRTs in Lithuania between 2000 and 2015. A VRT was defined as described by the International EXPeRT Cooperative Group (see above) (5). Pediatric age was considered as less than 18 years of age at the time of the diagnosis.

The institutional databases of both Lithuanian Pediatric Oncology Centers at Vilnius
University Santaros Klinikos (VUHSK) and Hospital of Lithuanian University of Health Sciences Kaunas Clinics (LUHSKK) were reviewed to identify VRTs as defined previously. Thereafter, the retrieved cases were compared with the LCR database to verify patients' registration and survival status and calculate survival rates.
Baseline characteristics were collected either electronically through institutional databases, or manually through paper records. The primary end point to assess treatment outcome was defined as an overall survival at five years after diagnosis (OS 5y ). In order to evaluate changes in the survival rates over the 16- Table 1 for baseline characteristics).
The median age at the time of their diagnosis was 12 (range 0-17) years with a slight female predominance (n = 27; 61.4%). The distribution of histologic types amongst the study subjects is depicted in Fig. 1. Overall, 23 histologic tumor types were identified: adrenocortical carcinoma was the most frequent one (n = 7; 15.9%), followed by hemangioendothelioma (n = 4; 9.1%), renal/thyroid carcinomas and rhabdoid tumors (n = 3, 6.8% each). There were two cases of pheochromocytoma, gastric adenocarcinoma, salivary gland carcinomas, hemangioblastoma, inflammatory myofibroblastic tumors, and ovarian cancers (4.5%) and 12 single cases of different tumor types (Fig. 1). Most patients (n = 39; 88.6%) presented with localized tumors at the time of diagnosis, half of them remained in complete remission at the time of evaluation. Tumor progression compromised the cure in 47.7% of cases, one patient (2.3%) deceased due to toxic complications.

Registration at the Lithuanian Cancer Registry
The data verification with the LCR revealed that nine of 44 patients (20.5%) were not registered at the time of diagnosis (Fig. 1). Eight of nine missing patients (88.9%) were diagnosed at LUHSKK and comprised 44.4% (8/18) of all identified pediatric VRTs treated at the center as compared to 3.8% (1/26) of patients treated at VUHSK. The most common types of non-reported cases were salivary gland carcinoma, inflammatory myofibroblastic tumor and thyroid carcinoma (2 cases each), followed by single cases of adrenocortical carcinoma, hemangioblastoma and carcinoma of the uterus (Fig. 1).

Treatment outcome
The OS 5y of the entire cohort was 55.8%. Surprisingly, the cure rate did not improve over the analyzed time periods -54.2% in 2000-2007 vs 49.4% in 2008-2015 (Fig. 2a). The median follow-up time of the compared patients' groups was 11.0 and 4.6 years, respectively (Table 1). In line with our expectations the OS 5y rate of the unregistered patients was significantly higher than that of the registered cohort -100% vs 45.1% (p = 0.016) (Fig. 2b) with similar median follow-up time of 8.4 and 9.9 years, respectively ( Table 1). The difference in OS rates remained detectable 10 years after the diagnosis accounting for 80% in the unregistered cohort as compared to 45.1% in the registered one (Fig. 2b).

Discussion
Our study aimed to address epidemiological data of pediatric VRTs in Lithuania -a year. The main drawback of our population-based study was its retrospective nature that did not allow us to verify relevant parameters (e.g. details on adverse events, treatment etc.) due to limited data source availability. Nevertheless, we were able to demonstrate that the survival of VRTs in children was inferior (55.8% in total) as compared to the average 80% cure rate of pediatric malignancies. Similar results were reported in adults by population-based analysis of the Surveillance of Rare Cancers in Europe (EUROCARE) project (10) -with rare cancers displaying lower survival rate (47%) than the common cancers (65%). The scarce expertise due to rarity of cases and absence of clinical trials were main contributors to the inferior survival.
The lack of improvement in cure rate over time (54.2% in 2000-2007 vs 49.4% in 2008-2015) was rather unexpected. National population-based studies in leukemia (11,12) and single-center reports in solid tumors (13)(14)(15) (6). Thus, expanding national, regional and international collaboration with a special focus on VRTs is crucial to overcome incurable cancers.
As expected, the survival rate of unregistered patients was higher than the one of the registered patients (100% vs 45.1% at 5 years). The difference remained significant for at least 10 years after diagnosis. Given the extreme rarity of VRTs, accurate reporting to cancer registries is crucial for reliable calculation of treatment outcome. Insufficient registration of pediatric cancers was confirmed in more common childhood cancers, e. g. for CNS tumors -the study that analyzed survival of European children based on the national population-based cancer registry data highlighted incomplete registration of non-malignant entities in many countries and, as a consequence, a lower overall survival (7). The recent survey focused on the rate of pleuropulmonary blastoma in Europe also demonstrated lower than estimated number of reported cases in Eastern / Central European countries (16).
A high rate of unregistered tumors in our study (20.5%) could be partially underpinned by an inconsistency in the national regulatory requirements: there was a formal obligation for health care providers to report every new cancer case to the cancer registry, however the legal status registry was not accordingly formalized.
This resulted in different interpretation of the reporting obligation and restriction data flow. The majority of the unregistered cases (88.9%) came from one of two pediatric oncology centers and reflected institutional policy with regard to the obligation to report new cancer cases to the LCR. All the missing data identified during this study were entered retrospectively and included death certificate only.
An additional contributing factor to the high proportion of unreported cases could be insufficient awareness of surgeons, who used to be the first to encounter a VRT in children and adolescents, about the importance of meticulous registration of every pediatric cancer case. In our study most of the tumors (88.6%) were localized at the time of diagnosis and did not required adjuvant chemotherapy subject complete tumor resection. Therefore, presumably pediatric oncologists were not involved in the patient care of at the initial stage. Several studies have shown that multidisciplinary teamwork affects the diagnosis, management and quality of care in cancer patients (17,18), thus regular tumor boards, including virtual tumor boards as well as international collaborations, should be regarded as a standard of care in the management of childhood cancers (19,20). Improvement in multidisciplinary collaboration between pediatric oncologist, surgeons, and other specialists could increase the registration rate.
Our results clearly demonstrated that timely (not retrospective) and complete registration must be ensured for accurate statistical analysis and data evaluation. In addition to mandatory national reporting regulations, an ongoing European PARTNER (Pediatric Rare Tumor Network -European Registry) project supported by the European Reference Network for Paediatric Cancer aims to create a pan-European system that should enhance international communications between members of the European Union by combining national registries focused on VRTs and creating registries for countries that do not have one, as well as linking these registries with virtual consultation systems (https://webgate.ec.europa.eu). The undertaken action could strengthen registration at the national level. It could be further improved by seeking an agreement on the entities to be registered, especially those of low-grade histology, as an absence of uniform guidelines defining registration of benign CNS tumors resulted in inequalities of survival rate across European countries (7). In our study we identified two cases of non-malignant inflammatory myofibroblastic tumor that is currently not defined as a VRT and not reported to the cancer registries (21).
However, both affected patients underwent multiple surgical interventions resulting in almost mutilating late sequelae. Therefore, a revision of the registration criteria for some tumor entities should be considered.

Conclusions
Incomplete registration of VRTs in cancer registry is an important issue and can significantly affect epidemiologic and outcome data. Regular verification of the pediatric cancer cases could ensure data quality and completeness of registration.
Based on our results we would strongly advocate for an active collaboration between pediatric oncology centers and national cancer registries to prevent dramatic deviation in statistical analysis and calculation of survival data.  Comparison of five-year overall survival according to: a diagnosis and treatment period; b re