Global surveillance of survival from intrinsic brain tumours diagnosed during 2000-2014: trends by age and histology

This thesis provides a comprehensive examination of world-wide variation in population-based survival from brain tumours in children and adults, by histology. It comprises five chapters in the form of research papers. Population-based cancer survival estimates are key to assessing the overall effectiveness of a health care system in managing cancer. The third cycle of the CONCORD programme for the global surveillance of cancer survival (CONCORD-3), which included data for more than 37.5 million cancer patients diagnosed during 2000-2014, highlighted substantial world-wide disparities in survival for all brain tumours combined, in both children and adults. Survival comparisons for all brain tumours combined, however, are potentially confounded by international variation in histology distribution and the clinical heterogeneity of the various tumour subtypes. Robust comparisons of brain tumour survival require clinically relevant data by histology because histology is the main outcome predictor. Granular survival estimates by histology are therefore warranted. The first and second chapters, two systematic reviews, appraise the geographical coverage and methodology of studies of brain tumour survival by histology. The first chapter focusses on children, while the second centres on adolescents and young adults. Adolescents and young adults are a transitional age group for which the strategy for presenting epidemiological data by histology is still debated. These reviews show that very little is known about brain tumour survival by histology beyond Europe and North America, and that differences in study design may indeed hamper robust comparisons. The two systematic reviews inform the scope of the third chapter, in which distinct histology groupings for children and adults are defined. The groupings for children represent an enhanced version of the International Classification of Childhood Cancer (third edition), while those for adults are designed to produce clinically meaningful categories. These groupings schemes form the basis for an examination of the global variation in the histology distribution of brain tumours and for an account of some of the key quality indicators for cancer registration. 7 The histology distribution of brain tumours varied widely world-wide (2000- 2014). In children, the proportion of low-grade astrocytic tumours ranged from less than 10% to more than 30%. In adults, the proportion of glioblastoma varied between 9% and 69%. International comparisons were made difficult by wide differences in the proportion of tumours of unspecified histology, which accounted for up to 52% of diagnoses in children and up to 65% in adults. These histology groupings are then used to produce up-to-date analyses of global trends in brain tumour survival by histology, for children (chapter 4) and adults (chapter 5), using individual records from CONCORD-3 for 610,710 patients diagnosed with a brain tumour during 2000-2014 in 60 countries. Many of these countries have never previously been included in studies of brain tumour survival by histology. The studies deploy the same protocol for data collection, quality control and statistical methods for all the data sets. Brain tumour survival in children and adults were examined in two separate analyses. Five-year net survival for low-grade astrocytoma in children was in the range 84-100%, while it varied between 47% and 86% for medulloblastoma. In adults, survival from glioblastoma improved substantially after 2005 in most countries, mainly in the short term (up to two years from diagnosis). The increase in two-year survival between 2000-2004 and 2010-2014 was more remarkable for adults aged 40-70 years than for younger patients (15-39 years). Trends for 2000-2014 were upward for the 40-70 age group, while they were somewhat flat for the 15-39 age group. To our knowledge, this is the largest study of population-based survival from brain tumours by histology. The geographical breadth, the methodology for data collection and analysis, and the proposed histology groupings are expected to set a benchmark for future global comparisons.