Functional connectivity and neuronal networks in brain tumor patients with epilepsy

Approximately 80% of patients with low-grade and 45% of patients with
high-grade glioma suffer from epileptic seizures. Among these patients, there
is a large inter- and intra-individual variability in seizure severity and
frequency, the cause of which has not been elucidated in earlier research.

In order to understand more of the complex mechanisms involved in epileptic
seizures in brain tumor patients, the neural network architecture of the brain
can be investigated. It is widely acknowledged that the paroxysmal phenomenon
of epilepsy is related to hyperexcitation of neurons, leading to changes in the
synchronization of large neuronal networks during the seizure. Spatial features
of neural networks determine to what extent the network facilitates such
synchronization. Moreover, networks with more long distance connections, a
so-called *random* configuration, are thought to have a lower threshold for
synchronization. A possibly influential feature of functional networks is the
small-world characteristic, referring to the high degree of interconnectedness
between single units in a network.

Previous studies using computational models have suggested a correlation
between changes in network structure and epilepsy. Netoff et al. simulated a
network model of excitatory neurons in hippocampal slices and changed the
parameters of the network (e.g. proportion of local versus long-distance
connections). They found that the start of the bursting phase corresponds with
a more random architecture. Percha et al. showed a potential mechanism
underlying seizure generation, as properties of phase synchronisation changed
radically depending on the structure of the network, which may play an
important role in the emergence of seizures. Dyhrfjeld-Johnsen et al. recently
demonstrated by means of a computational model that in the rat dentate gyrus
neuronal loss, similar to the loss observed in hippocampal sclerosis, is
accompanied by a more random network configuration.

On the basis of these studies, it is suggested that the architecture of the
underlying neuronal network determines whether a patient is more or less prone
for the development of seizures. The more random a network, the more
susceptible it may be to whole system synchronization. In contrast, networks
with a less random and more small-world configuration, while still having a low
threshold for synchronization, are expected to be more stable [16, 17]. We
think it highly interesting to investigate whether network characteristics and
epilepsy features are correlated in glioma patients.

In our previous research, low-grade glioma (LGG) patients displayed cognitive
deficits. These deficits were strongly related to epilepsy burden (assessed by
the Engel Scale), and the use of antiepileptic drugs (AEDs). We have recently
shown that a correlation exists between cognitive functioning and functional
connectivity in LGG patients. The relation between (1) network structure, (2)
epilepsy, and (3) cognition remains to be clarified.

Language-study:
Surgical treatment of brain tumours has been advocated by numerous authors in
the past decade (Duffau, 2008). The goal is to obtain maximum resection of the
tumour, as survival is improved with this (Duffau, 2009, Sanai, 2008). Although
survival is improved, there is a chance that there is an occurrence of
functional loss after surgical removal of the tumour. In order to prevent
functional loss, intraoperative function mapping is applied during awake
surgery.
One of the important functions that can be affected during resective brain
surgery is language. Intraoperative language mapping is considered the gold
standard to identify cortical and subcortical language structures. A picture
naming task is administered, while the neurosurgeon applies direct cortical
stimulation on various (sub)cortical sites. If a naming error occurs during
stimulation this indicates that the stimulated (sub)cortical site is essential
for language function. This site will then be preserved during tumour
resection, to minimize language deterioration postoperatively. Currently,
picture naming and counting are the most commonly used tasks during
interaoperative language mapping to identify language sites (Duffau, 2008,
Sanai, 2008, Ilmberger, 2008).
Most studies use a picture naming task to test the language abilities in the
postoperative period. Research shows that in the immediate postoperative period
there is a decrease in language performance on this task compared to the
performance on the task before surgery. Three months after tumour resection
most studies indicate that patients returned to their baseline score or even
had an improved score compared to score before surgery. These results suggest
that there is no longer a reduced language performance in these patients.
However, in a subset of patients language performance is reduced. As the
picture naming task is relatively simple, it is presumed that the task may not
be sufficient enough to detect subtle but important language deficits.
Therefore, an extensive language task is needed to detect language deficits.
Only a few studies have used an extensive language task to examine language
performance in patients treated with resective surgery. One study investigated
writing, reading, repeating, naming and counting, and showed that six months
after surgery a subset of patients had a persisting language deficit (Sanai,
2008). Two studies used parts of the AAT to investigate language performance.
One study found that 3 months after surgery 42 patients out of 128 patients who
had no disturbance pre-operatively showed a language disturbance (Ilmberger,
2008). It was found in another study that three months after surgery 15 out of
29 patients had a reduction in spontaneous speech and mean length of
utterances. (Visch-Brink, 2010) Another study found that although positive
language sites were preserved, some patients were unable to write after removal
of the tumour (Duffau, 2009).
The present study will therefore examine language performance of the treated
patients with an extensive language test to characterize language disorders
resulting from resective brain surgery. In addition, a language related
questionnaire will also be obtained. This will provide insight in how patients
experience their own language performance.

The current study aims to investigate whether neural network characteristics of
the brain can predict the frequency and severity of epileptic seizures in brain
tumor patients. Our secondary objective is to determine the correlation between
seizures, network architecture and cognition in this patient group.

Language-study:
The primary aim of the research is to establish the characteristics of language
disorders in patients that have had resective surgery of a tumour in the
dominant hemisphere.

Longitudinal case-controlled observational study

Language-study:
This is a prospective controlled study and will be conducted at the VU
University Medical Centre in Amsterdam.
Brain tumour patients who received intraoperative language mapping and
resective brain surgery in their dominant hemisphere will be asked to
participate in the present study. In order to assess their language outcome
after surgery an extensive language test and a language related questionnaire
will be conducted. The inclusion of patients and data obtainment will start in
January 2011 and will end in April 2011.
We will investigate whether language performance is affected in patients
treated with resective brain surgery, despite the preservation of eloquent
language areas indicated with intraoperative language mapping. An extensive
language test will be used to objectively measure the language performance in
this patient group. A naming task will be included as well, to compare pre and
post operative data. In addition, a language related questionnaire will be
utilized to investigate how patients perceive their language performance.

For patients, the burden associated with participation consists of a number of
visits to the outpatients* clinic for MEG measurements and neuropsychological
screening. Furthermore, they will be asked to keep a diary regarding the
frequency and severity of epileptic seizures during the course of their
disease, which is routine practice in patients with epilepsy. For controls, the
burden of participation consists of a one time only visit to the outpatients*
clinic for MEG registration. No health-related risks are involved in this
study. In our view, the burden associated with participation is proportionate
to the potential value of the research for (glioma) patients suffering from
epilepsy.