Glioblastoma (GBM), the brain cancer better known for having killed Senator Ted Kennedy a few months ago, is the most aggressive and lethal of all brain tumours. But a collaboration between researchers from the University of Minho in Portugal, and the University of California-San Francisco, has found new reasons to be optimist. Their research, published in the journal Cancer Research(1), reveals that activation of a gene called HOXA9 is linked to GBM severity – the more activated HOXA9 is, the more aggressive the tumour turns out to be. The good news is that this activation can be reverted. This is possible because it is, not the result of a mutation, but an epigenetic change. Epigenetic mechanisms change the activity of a gene without altering its DNA sequence and, instead, a molecule sits, literally, on the top of the DNA activating or inhibiting specific genes. The work has two major implications: first reveals the therapeutic potential of drugs able to revert the HOXA9 aberrant activation – after all, GBM patients with less HOXA9 are the ones living longer - and, second, uncovers HOXA9 as a new prognostic tool to allow doctors and patients more educated choices on how to deal with a terrible disease. Also interesting is the fact that, yet again, epigenetic mechanism are behind cancer, highlighting the need to pursue more studies – and new methods – looking for epigenetic changes, especially in those cancers difficult to explain by mutations.

Glioblastoma is the most common and most aggressive type of brain cancer. Without treatment patients last no more than 3 months and, even when treated with optimal therapy - including surgery, radiation therapy, and chemotherapy – they only go to survive an average of 12 months, with less than a quarter of all cases surviving up to 2 years and fewer than 10% up to 5 years. The disease, like the name indicates, is a cancer of the brain glial cells, which are non-neural cells that provide support and protection, as well as oxygen and nutrients, to the brain neurons.<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" />

To develop treatments against any disease it helps if we understand its cause(s), but GBM, so far, has been extremely difficult to link to any of the usual cancer mutations. Genome-wide studies - that compare the patients’ genome with healthy subjects’ looking for changes associated with the disease - have found a couple of altered genes but nothing is very clear. The best prognosis marker so far is the MGMT promoter methylation, which is associated with milder disease, but even this not always work. Methylation is the addition of a methyl group and if done to a gene is an epigenetic change that can activate or repress the gene. As MGMT is a DNA repair gene that also protects cells from chemotherapy, and its methylation leads to gene silencing, MGMT promoter methylation is a positive prognostic marker because it makes cancer cells more sensitive to treatment.

It has also been know for some time that aberrant activation of several Homeobox (HOX) genes occur in the brain of GBM patients (in normal brains, HOX genes are not active). The HOX family is known by being responsible for the correct body plan during embryo development and the role of this abnormal activation in GBM, or why it occurs is not clear. 

The study just published – that resulted from a collaboration between Bruno Costa and Rui Reis at the Life and Health Sciences Research Institute (ICVS) from the University of Minho, Portugal and Justin Smith and Joseph F. Costello from the Department of Neurological Surgery from the University of California-San Francisco in the US - is an attempt to understand better the causes of GBM, as well as investigating the potential role of HOX activation in diseased brains.

For this the researchers tested cells from dozens of GBM patients with known disease history so they could link molecular observations with disease levels. Costa, Smith and colleagues start by trying to confirm that HOX activation was in fact aberrant in GBM but found that while this was truth, much more interesting was the fact that the activation of HOXA9 – a member of the HOX family – actually correlated with disease severity  - patients with higher HOXA9 activation had the worst disease history.

PI3K  is a protein linked to the transmission of signals between cells and also to cell growth that has been found in abnormally high quantities in several cancers. In GBM, a known regulator of PI3K is altered but until now the significance of this was not known. Linking these two facts Costa and Smith hypothesised that maybe HOX9 over-activation could be linked to PI3K.

To test this possibility they added a PI3K inhibitor to GBM cells growing in laboratory, and found that this reduced their HOXA9 activation as well as the quantity of HOXA9 protein produced, a promising result as it suggested that, if HOXA9 activation was behind the cancer  it could be treated. Also interesting was the fact the PI3K inhibition changed the methylation in the chromosome around the HOXA9 gene indicating that PI3K acted by an epigenetic mechanism.

The next step was to confirm a link between the disease and HOXA9 activation, and for that the researchers tested glioblastomas and non-cancerous brain cells, both with and without HOXA9 production. It was found that cells with higher HOXA9 showed higher cell division and reduced cell death, what explained their association with worst GBM prognoses. In fact, cancer cells are characterised by abnormal and uncontrolled cell division together with a resistance to death, and both characteristics are behind the cancer uncontrolled growth that takes over and eventually kills normal tissues and organs.  When the PI3K inhibitor – that reduces HOXA9 activation - was added to HOXA9-high GBM cells their death rate was substantially increased showing the therapeutic potential of such inhibitor.

To further confirm the link between disease severity and HOXA9 levels, Costa Smith and colleagues traced the disease history of the patients from whom the tested cells have been obtained, and found that the worst overall survival and treatment response was found on those individuals with higher HOXA9 activation.

Costa, Smith and colleagues’ work show that in GBM HOXA9 activation correlates with disease severity what is explained by the fact that GBM cells with higher HOXA9 divide more and die less typical oncogenic characteristics. HOXA9 aberrant brain activation in these patients is regulated by PI3K but, most importantly, this activation is epigenetic and can be reversed leading to a reduction in the cancer cells’ aggressiveness.

MGMT promoter methylation is a powerful prognosis tool that marks GBM patients with milder disease and is currently being proposed to be used in clinic. The problem is that some patients positive for this marker have, nevertheless, severe disease what makes its use problematic. After their results Costa and Smith hypothesised that maybe HOXA9 could be used in conjunction with MGMT promoter methylation to help in this problem. And in fact, several of the patients with MGMT promoter methylation showing aggressive disease were discovered to be positive for HOXA9 expression, while those positive for MGMT promoter methylation, and with reduced or no HOXA9 activation, were shown to suffer, by large, milder disease.

The study now published has several major implications: by revealing the molecular mechanisms that lead to increased GBM aggressiveness, and by showing that HOXA9 activation is reversible, Costa and Smith launch the basis for the development of new therapies against the disease. Then, by showing how HOXA9 levels correlate with patients’ survival, they provide a new prognosis tool that, when used together with MGMT promoter methylation, not only improves the efficiency of the prognosis but also extends its use to many more cases improving patients and doctors’ decision-making when dealing with this devastating illness.

Also interesting is the fact this work shows that the (one of them at least) “error” behind the disease is linked, not to a mutation in genes linked to the disease,  but to an epigenetic mechanism. In fact, and especially after the human genome project , most studies of cancers that do not have family history are done by looking at the DNA sequences of key tumour-suppressor genes and oncogenes searching for mutations. GBM is not an exception and just recently (end of 2008) two papers – one in the journal Science and one in the journal Nature – looked at DNA sequences to try to find the cause(s) of the disease with some,  but not huge, success. But as more and more of the so called “junk” DNA – which do not produce proteins – is being revealed as having important epigenetic properties these mechanisms need investigation in cancer, especially in those cases where mutations are nt enough tto explain the disease.  The results now obtained by Costa and Smith are a good example of how important this type of studies can be in the study of cancer

Cancer Research Cancer Research 70, 453, January 15, 2010. doi: 10.1158/0008-5472.CAN-09-2189