I’m giving an overview of brain tumour diagnostics and classification, looking back at how the World Health Organisation has done this over the past approximately fifty years, talking about the recent 2016 WHO classification of brain tumours, then talking about the response that the community has had to the 2016 classification. Finally, for most of the talk, I’ll be discussing the future of diagnostics as an integrated specialty.
How are these tumours classified?
Until recently, until the 2016 WHO classification, they were classified nearly exclusively on the basis of light microscopic techniques such as standard light microscopy as well as immunohistochemistry. It became clear over the 1990s and the first decade of this century that molecular genetics could play a role in the classification of these tumours in addition to light microscopy. The 2016 classification introduced molecular parameters in addition to microscopic parameters as the basis for brain tumour classification.
Can you tell us some more about the 2016 WHO classification?
For many of the major tumour types, and by major tumour types I mean the most common ones such as the diffuse gliomas which are the most common brain tumour types of adults as well as the most common malignant tumours of childhood, the medulloblastomas, we introduced genetic parameters as a way of classifying the tumours over and above the histological or microscopic parameters for classification. So it affected a good number of the brain tumours, whether you had to include genetic analyses in their classification.
What are the specific genetic parameters?
For the diffuse gliomas, which are the most common types in adults, the most important genetic parameters are mutations in the IDH, or isocitrate dehydrogenase, genes, those are IDH1 and IDH2, and those are used to classify all of the types of diffuse glioma such as glioblastomas, astrocytomas and oligodendrogliomas. The other key parameter is the copy number status of chromosomal arms 1p and 19q and those changes are essential in the classification of oligodendroglial tumours. For medulloblastomas there’s a broader array of alterations that are used in their classification and how one does it in medulloblastomas, whether one uses RNA-based approaches or DNA-based approaches or even protein-based approaches, is still being debated by the field. Nonetheless, by identifying specific pathways such as the Wnt pathway and the sonic hedgehog pathway you can divide medulloblastomas as well.
How has molecular diagnostics helped treatment in a practical way?
At the present time it has mostly helped assign tumours into much more specific categories that then can be treated in a much more uniform manner. It has also helped assign tumours into categories for clinical trials in which one can use molecularly targeted therapies against the aberrant genetic abnormalities. These, however, have mostly been in the setting of clinical trials.
There is also the practical step in the tumour known as glioblastoma, which is the most common brain tumour, of testing for MGMT promoter methylation, that’s methylguanine methyltransferase. This is a gene, when its promoter is methylated it has prognostic and what we call predictive importance, predictive meaning that it predicts the response to therapies. It’s clear that if a glioblastoma has its MGMT promoter methylated it will show better responses to the standard therapeutic approaches at the present time.
What is the future for diagnostics?
When you look over how technology has impacted diagnostics in the last hundred years you note that each time a new-fangled idea comes in, a hundred years ago it was light microscopy, in the 1960s it was electron microscopy, in the 1980s and ‘90s it was immunohistochemistry, in the 1990s and then 2000s it was PCR and then DNA sequencing and then RNA expression profiling. Each time one of these comes in it’s a big impact on the field, people get very excited about it and you spend a while figuring out what is the exact situation that that new technology will be used in in a practical manner. The most recent technology to make a big splash has been methylone profiling, a very important technology, and again that will make an impact in specific situations.
The question for the future, however, is not necessarily will there be new technologies that make an impact, the answer to that is yes, but how does one take multiple different methodologies and integrate them together to produce a diagnostic estimate that’s more accurate, more predictive, more prognostically accurate than any individual one. So what we’re looking at now is the use of larger datasets, using disparate types of data input, not only the type of data that one can find from looking through a microscope or doing a methylone profile, but also the kinds of information that can come from a patient’s blood samples or a patient’s background DNA, from their genetics, potentially in the future even their microbiome. But there are multiple different data sources and so what we’re exploring at the present time is the use of computational methods to integrate those many different types of data sources to produce a more accurate diagnosis.
Is there much international collaboration in this field?
Sure, in brain tumour research there is greater international collaboration. This has been the case for a couple of decades in the paediatric realm in which paediatric brain tumours are relatively uncommon. So in order to make progress the paediatric brain tumour community has done a very good job of collaborating internationally to move the field forward. It’s less common in the more common brain tumours such as the adult brain tumours, but what we have seen over the years, most recent years, is a move towards greater collaboration, greater national and international efforts, one example being the Cancer Genome Atlas pushed forward by the National Cancer Institute in the USA.
The diagnosis of brain tumours has gotten much better over the last couple of decades; it’s gotten more accurate in terms of prognosis and prediction. It has a long way to go and probably through integrating larger datasets we’ll make further progress.
