A recent publication by Girardi et al.[1] provides a global overview of the histology (type of tissue) of brain tumors. However, there is a large variation in the distribution of brain tumors by histological type. In addition, there appear to be major international differences in the registration of brain tumors with non-malignant behavior, according to researchers Otto Visser (IKNL) and Henrike Karim-Kos (Princess Maxima Center for Pediatric Oncology and IKNL) and their colleagues. This can have a major impact on the estimates of incidence and survival.
The researchers also found these differences within Europe, despite the fact that the European Network of Cancer Registries (ENCR) already made recommendations in 1998 to include all tumors in the central nervous system (CNS, including brain tumors) in cancer registries, independent of their behavior. (whether the tumor is malignant or not).
Incomplete Sources and Internationally Different Coding Practices
Underlying causes of the variation are incomplete registration of non-malignant brain tumors due to incomplete signaling (eg, dependence on a source register that does not cover the entire CNS tumor population). Partly due to unclear sources, the correct code of conduct cannot always be registered. Out of a total of 22,147 European children, the researchers found that nearly 10% of all tumors were not confirmed microscopically. The coding practices of non-microscopically confirmed tumors differ widely from country to country and also contribute significantly to the differences found in the proportion of non-malignant tumors.
Analysis of effect of misclassification of tumor behavior
Studies using cancer registries often report brain tumors according to their code of conduct (malignant or not) and where they are located. The way of grouping could potentially lead to biased estimates of incidence and survival, the researchers found. To give an indication of the confounding effects that different proportions of non-malignant and malignant tumors can have, they analyzed data from the Dutch Cancer Registry (NKR) for the period 2000-2017 for children and adolescents <18 years diagnosed with a brain tumor (N=2705). In line with Ostrom et al who looked at the effect of misclassification of pilocytic astrocytomas, they reclassified a random sample of non-malignant tumors to malignant tumors in different proportions (i.e. 10%/90%, 20%/80% , 30%/70% and 40%/60%), or vice versa for the 50%/50% distribution (actual distribution 49%/51%). The researchers also estimated the effect on the age-standardized incidence rates and the overall 5-year survival. Incidence rates were calculated as the mean number of cases per million person-years using the annual mid-year population size, as obtained from the Central Bureau of Statistics (CBS). For standardization of incidence for the age group 0-17, the Segi world standard population method was used. 5-year survival was estimated using the Kaplan-Meier method.
Effect of misclassification on survival rates
The age-standardized incidence rates were equally and consistently affected for malignant and non-malignant tumors. Depending on the behavioral group, the incidence changed by ~0.4 per million person-years for every 10% shift. No effect was found on 5-year survival for non-malignant tumors; survival rates remained stable for all distributions made. However, large differences in survival rates were found for malignant tumors. Misclassification of 10% of non-malignant tumors as malignant tumors showed a change in 5-year survival of malignant tumors of up to +7 percentage points. A misclassification of 20% resulted in a further increase of +12 percentage points.
International alignment of registration guidelines
These results indicate that differences in the proportion of non-malignant tumors may directly affect estimates of incidence and survival. An important contributor to the differences in the proportion of non-malignant tumors are the non-microscopically verified tumors. Recently, the European Network of Cancer Registries (ENCR) established a working group to align registration guidelines for clinical diagnoses without microscopic verification (OV, face-to-face communication). This is the next step to improve the quality of European population-based brain tumor data, leading to fewer misclassifications and more accurate cross-country comparisons.
[1] Girardi F, Rous B, Stiller CA, et al. The histology of brain tumors for 67 331 children and 671 085 adults diagnosed in 60 countries during 2000- 2014: a global, population-based study (CONCORD-3). Neuro Oncol. 2021;23(10):1765–1776