← Back

From molecule to behavior: using network theory to link cellular biology and cognitive functioning in glioma patients

Networks form an important fiber in current day society: we all participate in a large number of social contexts, (try to) traverse road and railway grids efficiently, and receive much of our daily newsfeed through the World Wide Web.

Text by Linda Douw

Since the seventies, mathematical analysis of how these networks are organized has led to wonderful and particularly useful insights into their functioning. What if we can even use these easy-to-access networks to understand how the brain works? After all, the brain is ‘just’ a network, like a computer, even though it is the most complex one known to man. Indeed, both microscopic and macroscopic brain networks abide by the rules of network theory, showing local specialization, global integration and modular organization. However, bridging the gap between these two scales of measurements has been difficult: for instance, translating a particular proteomic tissue profile to a corresponding network correlate has rarely been done.

“What if we can use easy-to-access networks to understand how the brain works?”

In December 2013, I took on the daunting but exciting task of using network theory to unite microscopic neuroscience with neuroimaging thanks to Veni and Branco Weiss Fellowships, and most recently a Neuroscience Campus Amsterdam (NCA) Team Leader grant. I do so in a very special patient population: glioma patients. Although there are many types of primary brain tumors that are reasonably benign, patients with the type of glioma I am particularly interested in generally have dismal prognoses, ranging from 7 years in low-grade glioma to only 14 months in glioblastoma multiforme (GBM). Adding to this, patients often suffer from epileptic seizures and may experience cognitive deficits.

So, what makes glioma patients such an interesting population to study? First off, major recent advances in our understanding and particularly treatment of glioma can be counted on the fingers of two hands. Clinically most relevant in the past decade has been the discovery of a mere 2-month survival benefit of combined irradiation and temozolomide chemotherapy over radiotherapy alone in GBM patients. Moreover, it has become evident that glioma subtypes with distinct growth and progression patterns can be distinguished based on the molecular and genetic make-up of the tumor. For example, gliomas with loss of heterozygosity at chromosomal arms 1p and 19q have a better prognosis than those without and respond differently to chemotherapy. However, the plethora of genetic markers and molecular pathways we now know of has not led to much control over tumor progression, let alone curative treatment options. Therefore, research is also aimed at early identification of glioma subtypes in order to speed up (tailored) treatment. Importantly, traditional neuroimaging has had limited power in this setting, leaving room for innovative non-invasive approaches of determining glioma subtypes and predicting treatment response.

Secondly, almost all glioma patients undergo surgical resection of (peri)tumor areas, allowing for extensive analysis of fresh brain material with microscopy, morphometry and genetics. In combination with the fact that neuroimaging and neuropsychological assessments are performed at regular intervals before and after neurosurgical intervention, this population offers a unique window into the associations between molecular brain tissue features, network disturbances and cognitive correlates of glioma.

At this time, a pipeline for tissue and neuroimaging data collection in glioma patients has been established, both at the VUmc and at Massachusetts General Hospital (MGH) in Boston (MA, USA). In Amsterdam, the main focus is on magnetoencephalography (MEG), cognitive profiling and investigation of both oncological and more general plasticity-related markers in brain tissue. In Boston, routine oncological analysis of resected tissue is combined with state-of-the-art neuroimaging, including resting-state functional magnetic resonance imaging (rs-fMRI) and diffusion weighted imaging (DWI).


We have already been able to bridge the gap between human whole-brain functional networks and cellular dysfunction with respect to seizure symptomatology in glioma patients. Resected tissue protein expression (particularly synaptic vesicle protein 2A, which is important for neuronal excitability) is associated with network characteristics of the tumor area. Seizure frequency correlates with both protein expression and network features, suggesting that network topology may even be used as an intermediate between molecular properties and behavior.


I am very excited to take this project to the next level, together with the fantastic people working with me both in Amsterdam and Boston. In a couple of years, I hope we have elucidated some of the links between brain functioning at the microscopic and connectomic levels, while I also aspire to establish network-based biomarkers of glioma genotypes and phenotypes. More to come!