The study evaluated population-based registry data on cases diagnosed with ependymal tumours defined as histology codes 9383, 9391–9394 in the brain or the central nervous system (CNS) primary sites C70.0–C72.9, C75.1–C75.3 (International Classification of Diseases for Oncology Third Edition (ICD-O-3)) (Fritz and World Health Organization, 2000). The data source for incidence was the Central Brain Tumor Registry of the United States (CBTRUS) analytic file, 1995–2009. Survival estimates were made using the SEER 13 registries research file for 1992–2009 (SEER, 2011a).

The CBTRUS analytic file consists of population-based incidence data on all primary brain and CNS tumours collected by 49 central cancer registries (Dolecek et al, 2012). Analyses were conducted using 2004–2009 data allowing the evaluation of both malignant and non-malignant ependymal tumours, given that the Benign Brain Tumor Cancer Registries Amendment Act (Public Law 107–260) was implemented in 2004. Long-term trends were assessed on a subset with primary malignant ependymal tumours for diagnosis years 1995–2009. The CBTRUS 1995–2009 analytic file captures incidence occurring over 97% of the US population.

For the estimation of survival rates, all primary malignant ependymal tumours in brain or CNS primary sites from the SEER 13 registries research file, April 2012, were evaluated for cases diagnosed from 1992 through 2009 (SEER, 2011b). The SEER 13 registries represent ∼14% of the US population.

Ependymal tumour groups were defined in accordance with the WHO Classification of Tumours of the Central Nervous Systems (Louis, 2007). Categorical classifications by ICD-O-3 histology/behaviour codes include: ependymoma (9391/3 cellular ependymoma, clear cell ependymoma and tanycytic ependymoma, and 9393/3 papillary ependymoma); anaplastic ependymomas (9392/3); myxopapillary (9394/1) and subependymomas (9383/1).

Demographic characteristics evaluated were gender; race (white; black; American Indian/Alaska Native; Asian; or Pacific Islander), ethnicity (Hispanic; non-Hispanic) and selected age groups. Age groups were based on accepted groupings, for example, 0–19 for children and adolescents, and arbitrary within adults to facilitate analysis. Tumour histology, primary site and behaviour (malignant and borderline malignant) patterns were assessed.

Frequencies, incidence rates, rate trends and relative survival rates were estimated using SEER*Stat 8.0.2 software (2011). Institutional Review Board approval was obtained under expedited review for research using the CBTRUS analytic file.

Population-based incidence statistics for all primary ependymal tumours are presented in Table 1 and Supplementary Tables 1–3. A total of 7303 ependymal tumours (4683 malignant and 2620 borderline malignant) were identified with an average annual age-adjusted rate (AAR) of 0.41 per 100 000 (0.27, malignant and 0.15, borderline malignant). Rates were higher in males than in females, whites than in other race groups and non-Hispanics compared with Hispanics. Among age groups, the highest rates were observed in age group 45–64 years. Most notable is the incidence rate ratio (IRR) of more than five times malignant to borderline malignant tumours for child/adolescent ages 0–19 years compared with an observed IRR of about 1.5 in adult age groups.

Ependymoma had the highest count and AAR followed by myxopapillary ependymoma, subependymoma and anaplastic ependymoma. Myxopapillary ependymoma represented approximately two-thirds and subependymoma the remaining third of borderline malignant tumours, while about 85% and 15% of malignant tumours were ependymoma and anaplastic ependymoma, respectively.

About half of all ependymal tumours were diagnosed in the spinal cord/cauda equina and the remaining half in the brain and other nervous system sites. However, the proportional primary site distribution for borderline malignant ependymal tumours was 64% spinal cord/cauda equina and 35% combined brain sites (Supplementary Table 1).

Site and histology patterns differed by age at diagnosis as shown in Figure 1. Whereas 50%–60% of ependymal tumours in adult ages were located in the spinal cord/cauda equine, only about 20% were found in the spinal cord/cauda equine primary site for children/adolescents. As shown in Supplementary Figure 1, the proportions of ependymal tumours diagnosed in the brain sites for the age group 0–19 years substantially exceeded those observed for adult age groups with the exception of ventricle. Among adult age groups, ventricle and cerebellum were the brain sites where proportional increases were observed with increasing age. The brain lobes site showed progressively lower proportions of cases with advancing age group.

The distribution by histology also differed by children/adolescent and adult age groups. While ependymoma was the most frequent histology in all age groups, anaplastic ependymoma was about six times more frequent in children and adolescents than adults, while the reverse pattern was apparent for subependymomas and myxopapillary ependymomas. For adult age groups, the proportional distribution pattern for subependymoma increased with advancing age, whereas decreases with age were observed for myxopapillary ependymoma.

Ependymal tumours exhibit a bimodal age distribution showing a peak in the 0–4 year age group declining through the early adolescent years and then rising to a second peak in the 55–59 year age group with the oldest age groups. Supplementary Figure 2 graphically shows the bimodal age distribution.

Supplementary Table 3 shows findings from the analysis of malignant ependymal tumour incidence for CBTRUS, 1995–2009.Trend analyses showed primary malignant ependymal tumour incidence rates to increase and in most instances significantly for most categories in gender, race, age, histology and primary site groups.

Relative survival estimates were made for a total of 1522 malignant ependymal tumour cases reported to the 13 SEER registries during 1992–2009 (Table 2). Overall, relative survival was ∼93%, 85%, 80% and 75% at 1, 3, 5 and 10 years post diagnosis, respectively. Poorer relative survival was observed for the youngest and oldest age groups, cases diagnosed with anaplastic ependymoma and/or tumours located in the brain sites. Additional relative survival estimate comparisons are graphically presented in Supplementary Figures 3–5.

Our comprehensive evaluation reaffirms findings from previous studies (Rodriguez et al, 2009; McGuire et al, 2009a, 2009b; Amirian et al, 2012; Bishop et al, 2012; Crocetti et al, 2012; Dolecek et al, 2012). Outcomes were generally favourable except for age groups 0–19 years and 65+ years, anaplastic ependymomas and tumour location in the brain site.

Children are more often diagnosed with anaplastic ependymomas and with tumours located in the brain sites. Furthermore, when the diagnosis is WHO grade II ependymoma, children fare worse than adults in the 20–44 and 45–64 year age groups. This is uncommon as most cancers of the same type and grade have improved outcomes in children compared with adults (Dolecek et al, 2012). Potential explanations include: (1) paediatric and adult cases having different biologically based tumours with gain of chromosome 1q and the absence of chromosomal imbalances as a marker for more aggressive disease (Korshunov et al, 2010; Wani et al, 2012), (2) although morphologically similar systematic bias by pathologists to report a better tumour grade/histology in paediatric cases, which has been demonstrated for classification of oligodendrogliomas (McCarthy et al, 2008); (3) differences in treatment, such as less radiation use in children (Koshy et al, 2011); and (4) differences in tumour biology based on location–genetic testing demonstrate subsets of ependymomas exhibit distinct patterns of chromosomal mutations and gene expression based on anatomical site (Spassky et al, 2005; Taylor et al, 2005).

The poorer relative survival for elderly cases may, in part, be explained by the fact that proportionately more diagnoses occurred in the brain sites among elderly than younger adult age groups (Supplementary Tables 2b, c and d). Our study has limitations, including the lack of detailed clinical data that may impact survival which, if available, could also shed light on this observation.

Insights from population-based incidence and survival analyses afford an opportunity to better understand the impact of site and histology patterns on outcomes associated with ependymal tumours and inform future research.