Last week’s PNAS online describes some new gross anatomical and electrophysiological changes that accompany Alzheimer disease (AD) and frontotemporal dementia (FTD). For one, researchers led by Klaus Linkenkaer-Hansen at VU University, Amsterdam, the Netherlands, report that α and ϑ brain waves in mild to moderate AD patients exhibit subtle changes in timing. In the authors’ view, the changes support the idea that default networks are compromised in AD, and they could also serve as a marker to monitor disease progression or treatment. For another, Maria Gorno-Tempini and colleagues in the U.S. and Italy used MRI scans to examine the relationship between ApoE status and brain atrophy in Alzheimer’s and behavioral variant FTD (bvFTD). While there is an established link between the ε4 variant of ApoE and brain atrophy in AD, the researchers now extend this link to bvFTD as well, supporting the idea that ApoE4 may be a risk factor for multiple neurodegenerative diseases.

The Dutch brain wave study employed magnetoencephalography (MEG) rather than the more common electroencephalography (EEG). “Electric fields can give rise to a potential on the scalp that makes it difficult to determine where the signals come from,” Linkenkaer-Hansen told ARF. With MEG, the source of the signal can be more accurately pinpointed, indeed closely enough to allow this technology to guide neurosurgical treatment of intractable epilepsy and brain tumors. MEG is also relatively unobtrusive and fast. It does not require a multitude of electrodes taped to the scalp; instead, the magnetic field is measured by a helmet-type device that the participant can slip on and off in seconds. A typical MEG recording takes three minutes compared to the 15 minutes required for an EEG. “From the time the patient walks into the room to the time they leave is about 10 minutes,” said Linkenkaer-Hansen. Speed is clearly an advantage when dealing with AD patients, who can get agitated during testing conditions.

The researchers took MEG recordings of 19 people with mild to moderate AD and 16 age-matched controls, and assessed temporal changes in different brain waves. It is established that amplitudes of different brain waves, including α, γ, and ϑ bands, are different in AD patients compared to controls (see recent review by Jackson and Snyder, 2008). What joint first authors Teresa Montez, Simon-Schlomo Poil, and colleagues investigated was not simply the amplitude in such bands, but how they are modulated temporally. “Normal EEG and MEG analysis throw out all temporal information in the signal,” explained Linkenkaer-Hansen. By contrast, he showed that normal volunteers generate oscillatory activity that has refractal character in time. This essentially means the oscillation can somehow carry a memory of its own dynamic for several tens of seconds (see Linkenkaer-Hansen et al., 2001). “I suspected that it is important for memory that you have this coordination of brain activity over time,” he said.

In this study, the researchers show that this temporal dynamic breaks down in AD patients. For example, α band (in this case 6-13 Hz) oscillations in the parietal region normally bias activity in the same region several tens of seconds later in normal people, but this bias is significantly reduced in AD. The authors suggest that the “temporal structure is at least as important as the magnitude of the oscillations as a marker of pathophysiology and, possibly, mnemonic operations.” In contrast, the authors found that temporal bias was significantly and greatly strengthened among ϑ frequency (4-5 Hz) oscillations emanating from the medial prefrontal cortex in AD patients compared to controls. “Our interpretation is that this might reflect some sort of compensatory mechanism,” said Linkenkaer-Hansen. There is growing evidence for compensatory mechanisms in Alzheimer’s, including hippocampal hyperactivation (see ARF related news story) and increased frontal activity (see ARF related news story).

Finally, the authors suggest that these MEG recordings support the concept that default networks—a baseline neural activity that occurs while people do not focus on a given mental task—is gone awry in people with dementia (see ARF related news story). ϑ oscillations are part and parcel of this default network activity (see Gusnard and Raichle, 2001).

Previous EEG studies have found differences in brain wave amplitudes between people with dementia and controls (see, for example, Moretti et al., 2008), and even correlated those differences with hippocampal atrophy (see Babiloni et al., 2009) and ApoE genotype (see Kramer et al., 2008). Several magnetic resonance imaging studies, as well, have linked ApoE status with rates of brain atrophy in AD patients (see, for example, van de Pol et al., 2007 and Hämäläinen et al., 2008); however, only one small study has examined the relationship between ApoE status and atrophy in FTD (see Boccardi et al., 2004). Now, researchers led by Maria Gorno-Tempini at the University of California at San Francisco report analysis of a larger FTD imaging study focusing on patients with behavioral variant FTD (bvFTD), also called Pick’s disease. This is the most common form of FTD and is characterized by altered social behavior, emotions, and self-awareness. The study indicates that bvFTD patients who also carry the ApoE4 allele have greater brain atrophy in select regions of the brain than bvFTD patients who do not carry that particular isoform of the apolipoprotein. ApoE4 is already a known risk factor for this type of FTD (see Engelborghs et al., 2006).

First author Federica Agosta and colleagues used voxel-based morphometric (VBM) analysis to measure brain atrophy in 51 people with AD, 31 with bvFTD, and 51 controls. Consistent with previous imaging studies, they found greater atrophy in hippocampus and parietal cortex in ApoE4-positive AD patients. In the bvFTD group, Agosta and colleagues found that ApoE4-positive patients had greater atrophy in regions of the brain that typically degenerate in this disease, most notably both sides of the anterior cingulate cortex, and a broad region of only the right frontal cortex. Other small areas that appeared to atrophy more in ApoE4 carriers than non-carriers include the right caudate, superior temporal gyri, and left frontal gyri. The results suggest that ApoE4 somehow influences the underlying pathology of FTD. According to the authors, that might explain some curious observations. For example, one concerned a pathological dichotomy between two first-degree relatives who both had FTD—the brother who was homozygous for ApoE4 had more profound behavioral and cognitive problems than his sibling, who carried two copies of ApoE3.

Whether ApoE affects the progression of bvFTD is unknown. In AD, it is generally accepted that E4 brings on the disease earlier and speeds it up. The authors had not set out to address this question in this study, but while they saw no correlation between cognition and ApoE status, they do write that “Certainly, the atrophy patterns in our VBM study suggest that the FTD ε4 carriers may be at higher risk for rapid clinical decline.” This question requires a longitudinal study.—Tom Fagan


  1. Since its introduction, the electroencephalogram (EEG) was viewed with great enthusiasm as the only methodology allowing a direct and online view of the “brain at work.” In the last decades, the introduction of structural and metabolic brain imaging methods (i.e., PET, MRI, and fMRI) have relegated neurophysiological techniques to such a relatively low level of interest for diagnosis and research in Alzheimer disease and related forms of dementias that clinical guidelines do not include EEG recordings as a primary step for correct diagnosis. However, despite the fact that modern methods of functional brain imaging combine extremely precise anatomical details to inform on brain metabolism and function (very high spatial resolution), the question has been more frequently raised as to how could the temporal resolution (that is, the ability to follow information processing in brain circuitries with a time sampling not of minutes or seconds, but at a millisecond level, which is the proper brain timing) be improved. Moreover, it is becoming evident that indirect information stemming from neurovascular coupling (that is, the linkage between the firing characteristics of neurons attending a given task and their blood supply and oxygen-glucose consumption) is only partly known, with a number of exceptions to the general rules, which makes large-scale clinical applications of these imaging modalities (MRI, PET, fMRI) still premature. Finally, PET and fMRI are expensive devices, available only in highly specialized centers and—at least in the case of PET—the use of ionizing radioisotopes is required; both reasons inhibit short-term follow-up exams in large groups of subjects.

    For all the above reasons, in recent years the “old” EEG has been revisited by using a number of modern approaches that can analyze and localize sources of EEG rhythms and signals in 3D, as well as track neural wiring and connectivity that characterize the hierarchy of the electromagnetic brain activity sustaining a given function. The approach is gradually regaining the interest of the scientific community as a tool theoretically able to discern—with a time resolution that follows the “brain time,” that is, in the order of milliseconds or even fractions of milliseconds—the sequential recruitment of relays within networks sustaining the investigated task, according to its “natural” hierarchy. This ability is not really surprising, if one considers that the same neuronal circuits responsible for behavior, mood, emotions, memory, movements, sensations, language, and attention produce electromagnetic transient or rhythmic signals that vary in time in parallel with the evolving cognitive process. It is, therefore, one of the most fascinating challenges of modern neuroscience to disentangle cerebral electromagnetic activities causally linked to a specific brain function from the bulk of spontaneous transient and oscillating brain waves.

    This study by Teresa Montez and her colleagues in Amsterdam approaches the analysis of electromagnetic brain signals in a relatively new way, namely by evaluating time-varying characteristics of the main rhythms of oscillation, looking at their interactions across different brain areas within a given instant (cross-correlation analysis) as well as within the same brain area in a relatively extended time window and examines auto-correlation properties (that is the coordination, if any, of rhythmic brain activity in time within a given cortical district), which is linked to serial processing. In a group of 19 patients affected by early-stage Alzheimer disease, the authors found a significant reduction of time fluctuations of α rhythms (as depicted by sequential representation of high- and low-amplitude bursts of such rhythm) over temporo-parietal brain regions, combined with weaker than normal autocorrelations on long-time scales (time window between 1 and 25 seconds). Meanwhile, the same parameters for ϑ rhythms were significantly increased on medial, prefrontal brain regions. They interpret α changes as secondary to damage to cholinergic, encoding/retrieval memory mechanisms, while ϑ modifications in the prefrontal cortex are considered compensatory mechanisms pointing toward functional maintenance. Follow-up studies will confirm or disconfirm this interesting hypothesis.

    Within this vein of reasoning we should finally consider that spatio-temporal characteristics of electromagnetic brain signals contain relevant information on pathologic processes underlying different types of dementias. It is worth remembering that the integrated analysis of EEG power and coherence provided reliable predictions of MCI to AD progression within a relatively short time-frame (about one and a half years), demonstrating that low temporal δ source and low γ band coherence along the fronto-parietal midline correctly predict around 10 percent of annual rate of conversion to AD. The progression to AD conversion was predicted to be faster across about one year in individual MCI subjects with δ sources and fronto-parietal coherence higher than cut-off points obtained from control population. Moreover, methods producing “clusters” (Supervised Artificial Neuronal Networks; see Rossini et al., 2008) are able to classify subjects on an individual basis, i.e., whether they fall within or outside a normative cluster, as is required for clinical purposes, with a sensitivity and a specificity around 90 percent.

    Actually, a bulk of recent evidence points toward the idea that loss of synaptic contacts and neuronal connectivity begins longer before onset of clinical deficits in AD patients, and that they are compensated by neuroplastic phenomena, mainly based on recruitment of vicarious networks and enhanced network excitability. In a recent editorial in Annals of Neurology, commenting upon promising neuropsychological techniques for detection of prodromal (namely preclinical) signs of AD, Mortimer and Petersen (2009) speculated that if disease-modifying therapies will be found, they should be most effective when administered well in advance of symptom onset in at-risk individuals. This would urge us to find reliable biomarkers to detect such high-risk subjects many years before AD onset. Since the number of potentially at-risk individuals is extremely elevated, and widely distributed across the globe, such biomarkers should be low-cost, non-invasive, and widely available (otherwise, no health system could afford large-scale screening of general population); moreover, because of the ethical and sociological implications of such preclinical diagnosis, biomarkers should be extremely sensitive and specific, with an acceptable level of false positive and negative identifications. Needless to say, no biomarkers with such characteristics are currently available. However, neurophysiological techniques (namely EEG) do have many of the required properties since they are harmless, low-cost, and widely available on a worldwide basis. By using modern methods of signal analysis, EEG analysis is also becoming more and more sensitive and specific (see Rossini et al., 2007 for a review). The study by Montez et al. represents a further step toward this goal and even if they are using MEG (an expensive and relatively little diffused device), their type of analysis can be easily translated to EEG recordings. It is concluded that experts and professionals in the field of health organizations as well as drug companies and patients’ associations should look with careful and growing interest (and eventually invest some more resources) to the recent developments reached by neurophysiological methods within this field.


    . Altered temporal correlations in parietal alpha and prefrontal theta oscillations in early-stage Alzheimer disease. Proc Natl Acad Sci U S A. 2009 Feb 3;106(5):1614-9. PubMed.

    . Detection of prodromal Alzheimer's disease. Ann Neurol. 2008 Nov;64(5):479-80. PubMed.

    . Clinical neurophysiology of aging brain: from normal aging to neurodegeneration. Prog Neurobiol. 2007 Dec;83(6):375-400. PubMed.

    . Is it possible to automatically distinguish resting EEG data of normal elderly vs. mild cognitive impairment subjects with high degree of accuracy?. Clin Neurophysiol. 2008 Jul;119(7):1534-45. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Deactivation Flaws Predict Memory Troubles
  2. Neuron Refreshes Our Memory
  3. Network Diagnostics: "Default-Mode" Brain Areas Identify Early AD

Paper Citations

  1. . Electroencephalography and event-related potentials as biomarkers of mild cognitive impairment and mild Alzheimer's disease. Alzheimers Dement. 2008 Jan;4(1 Suppl 1):S137-43. PubMed.
  2. . Long-range temporal correlations and scaling behavior in human brain oscillations. J Neurosci. 2001 Feb 15;21(4):1370-7. PubMed.
  3. . Searching for a baseline: functional imaging and the resting human brain. Nat Rev Neurosci. 2001 Oct;2(10):685-94. PubMed.
  4. . Increase of theta/gamma ratio is associated with memory impairment. Clin Neurophysiol. 2009 Feb;120(2):295-303. PubMed.
  5. . Hippocampal volume and cortical sources of EEG alpha rhythms in mild cognitive impairment and Alzheimer disease. Neuroimage. 2009 Jan 1;44(1):123-35. PubMed.
  6. . EEG functional connectivity and ApoE genotype in Alzheimer's disease and controls. Clin Neurophysiol. 2008 Dec;119(12):2727-32. PubMed.
  7. . Baseline predictors of rates of hippocampal atrophy in mild cognitive impairment. Neurology. 2007 Oct 9;69(15):1491-7. PubMed.
  8. . Apolipoprotein E epsilon 4 allele is associated with increased atrophy in progressive mild cognitive impairment: a voxel-based morphometric study. Neurodegener Dis. 2008;5(3-4):186-9. PubMed.
  9. . APOE and modulation of Alzheimer's and frontotemporal dementia. Neurosci Lett. 2004 Feb 19;356(3):167-70. PubMed.
  10. . Dose dependent effect of APOE epsilon4 on behavioral symptoms in frontal lobe dementia. Neurobiol Aging. 2006 Feb;27(2):285-92. PubMed.

Further Reading

Primary Papers

  1. . Altered temporal correlations in parietal alpha and prefrontal theta oscillations in early-stage Alzheimer disease. Proc Natl Acad Sci U S A. 2009 Feb 3;106(5):1614-9. PubMed.
  2. . Apolipoprotein E epsilon4 is associated with disease-specific effects on brain atrophy in Alzheimer's disease and frontotemporal dementia. Proc Natl Acad Sci U S A. 2009 Feb 10;106(6):2018-22. PubMed.