Speaker: Andreas von Deimling
Department: Josephine Nefkens Institute for Oncology
Subject: Methylation analysis and NGS as basis for classification and grading of brain tumors
Location: Erasmus MC
Author: Renée van der Winden
Today Andreas von Deimling came in to talk to us about his team’s research in the diagnosis and grading of brain tumors. They have developed a method to classify the methylation patterns of tumors into distinct tumor subtypes. It was already known that methylation could distinguish four subgroups in medulloblastoma and it now turns out this also works for a lot of other types of brain tumors. During his talk Deimling often compared the result of methylation classification with that of WHO. Currently, WHO can distinguish 90 subtypes of brain tumors and methylation classification can distinguish 82 of those 90. What is very useful about methylation classification, is that methylation patterns in tumors turn out to be very stable. So the pattern in an early tumor is roughly the same as that in a tumor in a later stage.
To perform the classification a patients methylation pattern is run through approximately 10,000 decision trees which all point to a certain tumor subtype. To find which type is the right one, you simply have to pick the type on which most decision trees landed. As a supplement to this method, Deimling and his team also look at copy number variations in the tumor cells. A diagnosis like this takes around ten days to complete. It is then integrated with the diagnosis from WHO to give as accurate a diagnosis as possible. About 75% of the diagnoses overlap, which shows the methylation classification is reliable. In about 12% of the cases the methylation method yields a change in the diagnosis compared to WHO, these are of course the cases that are most interesting and show the best use of the methylation classification. Then there is 12% which does not give a match and less than 1% in which methylation classification is misleading.
As an example, Deimling talked about meningioma’s. Right now, mitosis is counted to distinguish between grade I, II and III tumors, which vary in aggressiveness. This is very arbitrary. Methylation classification can help to distinguish between these grades which leads to better treatment of the patients.
Clearly, this new method is useful for the classification of brain tumors, however, because methylation is stable in tumors, it is not very useful for grading the tumors and more research is needed on that topic.
Lastly, Deimling showed us a website on which doctors can easily access the methylation patterns of tumor subtypes. Moreover, since two weeks there is a prototype for a sarcoma classifier, thereby widening the usage of this method.
I thought this seminar was quite easy to follow and I liked the topic. I find it very inspiring that people are working to defeat a disease like cancer, which is lethal in many cases, and succeeding. I might want to join them one day. I loved the fact that Deimling and his team made a website for other doctors to access. It showed me that he really cares about his research and that he wants to help others with it.