Physiological Markers for the Prognosis of Memory Decline

What two fields of research can learn from each other: Memory Decline in Temporal Lobe Epilepsy and Mild Cognitive Impairment Against our scientific background, we found an analogy in one of their most disabling symptoms between the two different clinical groups that form the focus of our research interest: Epilepsies, especially temporal lobe epilepsies (TLE) and degenerative dementia in its earliest stage, mild cognitive impairment (MCI). While it is obvious that memory problems are the main concern of people with MCI, they are also a major predictor of impaired quality of life and social disability in TLE. Moreover, memory impairments are not only a link between MCI and TLE, but may also act as a starting point from which researchers in each field may add innovative aspects to the corresponding research area. Specifically, the pathogenesis of memory disturbance in both conditions is unclear when patients of each clinical group perform normally on standard neuropsychological tests of memory. This is the case in early stages of MCI as well as in TLE. However, correlates of the subjective observation of memory problems would be detectable with neuroimaging and neurophysiology. Thus, the overall aims of our study are: To identify eventual analogies in the pattern of memory impairments in TLE and MCI and to better understand the mechanisms of memory problems in both conditions. This objective will be addressed by comparing these groups by several measures. To increase the validity for prognosis of memory decline by implementing multimodal examination on a single subject base. In this study, MCI patients will be divided into a subgroup of evidenced memory impairment, as assessed by standardized tests, and in a subgroup of patients with subjective cognitive complaint, but without objectively measurable abnormality. TLE patients will be divided in a group of early TLE and in a group of pharmacoresistant TLE. By entering the study, each patient undergoes several neuropsychological tests on memory performance, document confounding variables, event-related electroencephalography and magnetic resonance imaging. In order to find individual abnormalities, we will use innovative data-processing techniques and single-subject non- parametric statistics. The extracted features of the clinical groups will be compared with those of a sample of healthy controls in order to determine abnormalities. Specifically, we will assess features which were shown to be of diagnostic or predictive value in one of the two assessed disorders (MCI or TLE). After 1.5 years, a second session of neuropsychological testing will reveal the degree of memory decline. To determine which features from neuroimaging and/or neurophysiology perform best for prognosis, being used alone or in some combination, support vector machines will be applied. Moreover, several automatic classifiers will be compared in order to identify the best suited machine-learning algorithm for prognosis of memory decline.

Combining several computer-based techniques will be increasingly incorporated into clinical diagnosis and prognosis of patients with neurological diseases such as temporal lobe epilepsy or mild cognitive impairment. Patients with temporal lobe epilepsy experience epileptic seizures arising from the temporal lobe of the brain. Deep within the temporal lobe there is the hippocampus. This region has the shape of a sea horse (latinum: hippocampus). It plays a core role for memory formation. Therefore it is not too surprising that patients with temporal lobe epilepsy complain about memory nuisances. Subjective cognitive complaints are often an early stage of Alzheimers dementia. The main symptoms are memory problems, just as it is in temporal lobe epilepsy. However, in early stages others may not notice these problems. Neuropsychological tests can detect the disorder only when the disease progresses to a clinically relevant level, such as mild cognitive impairment. Indeed, it would be important to recognize the disorder as early as possible, in order to take adequate countermeasures, once a disease modifying treatment becomes available. In both of these disorders, we can detect changes in size and shape of the hippocampus even at the beginning of the condition. In addition, both disorders show changes in the electrical activity of the brain. From the electrical brain activity we can create models of brain-networks. Both diseases heavily influence these networks. In temporal lobe epilepsy, the networks are altered either on the left or right, depending on the site of seizure onset. In patients with mild cognitive impairment the networks change continuously over the whole brain, by reducing the speed of these networks. These physiologic measures of size and shape of the hippocampus, of network properties and network speed could support clinical practice. With modern computerized algorithms, diagnostics in the clinics could be done more accurately and maybe even at an earlier stage of the disease. However, it is rather difficult to measure the size and shape of the hippocampus exactly. Automated measurements vary strongly, depending on which algorithm we choose. Manual measurement is also unreliable when different experts argue about the boarders of the hippocampus. The most challenging cases are those patients, where the hippocampus is strongly deformed because of pathology. Indeed, these are the patients in whom we would desire an accurate measurement. In the present project we found that combining different algorithms can solve this problem. Similarly, network characteristics are not straightforward to be applied in clinical practice. They are not constant over time. Most importantly, we could show that this variation over time is different between patient populations. Nevertheless, the stability of the networks itself could be a new clinical measure that could be indicative for epilepsy and early stages of Alzheimers dementia. In sum, our project could show that the combination of several physiological measures, e.g. the structure, perfusion, and electrical activity is highly promising and should be further developed and eventually incorporated into clinical practice.