This is Part 1 or a two-part series. See also Part 2.
22 November 2011. Regular visitors to Alzforum will be familiar with the use of positron emission tomography, magnetic resonance imaging, and cerebrospinal fluid measurements as markers for Alzheimer’s disease (AD). But these three are not the only AD biomarkers. There is also the humble event-related potential (aka ERP). Determined by electroencephalography (EEG), this lesser-known electrophysiological measurement has been a relative wallflower, but is now gaining some popularity among AD trialists. At the 4th International Conference on Clinical Trials on Alzheimer's Disease (CTAD) held 3-5 November 2011 in San Diego, California, ERPs had a symposium and some posters devoted to them. “As a tool for assessing cognitive function, ERPs have been around for a long time, but mostly in the realm of scientists doing basic studies of memory and other processes rather than in the realm of trialists,” Andrew Budson of Boston University, told ARF. Budson jointly organized the ERP symposium at CTAD with Lon Schneider of the University of Southern California, Los Angeles.
ERPs have remained below the radar in clinical studies, Schneider told Alzforum. “AD researchers have been relying on biomarkers like CSF proteins and brain imaging. They have not had the motivation to use EEG,” he said. The lack of new therapies for AD and drug failures in late-stage clinical trials may now be providing that motivation, for example, to try to develop a cheaper, more immediate neurophysiological indicator of drug effect, he added.
What are ERPs, anyway? They represent EEG recordings elicited in response to particular stimuli. One common example is the “oddball” paradigm, in which researchers ask people to distinguish an infrequent, “oddball” tone (say, a high-pitched 2,000 Hz sound) among more frequent 1,000 Hz tones. This paradigm generates EEGs that, when averaged to cancel out random variation, form a reproducible wave pattern of neuronal activity, or the ERP. Its components—each being a discrete peak or trough of few microvolts in magnitude—represent summed inhibitory or excitatory postsynaptic potentials. They are designated as N or P, depending on whether they are negative or positive, followed by a number that indicates their timing (or latency).
Each ERP component results from one or more mental operations occurring at a specific time after the stimulus, with a set amplitude and distribution across a person’s scalp. Some components have standing names and are well known to scientists. For example, the auditory oddball paradigm elicits early components, such as the N100 (or N1) and P200 (P2), which represent activity in the first cortical areas to receive and evaluate the sensory input. Later components of the same ERP, such as the intensely studied P300 (P3) component, represent the processing of information at more advanced cognitive levels. Scientists think that the P300 elicited by oddball stimuli reflects the working memory updating mental representations. Still later components of the ERP (i.e., N400 or P600) involve cognitive functions that have to do with the processing of language and syntax, as well as memory.
There is a substantial literature of some 370 papers showing that various ERP components are altered in either amplitude or latency in AD patients, and that ERPs could thus be used as diagnostic markers of the disease. The first paper from Floyd Bloom’s group at The Scripps Research Institute in La Jolla, California, showed that the P300 component elicited by an oddball auditory stimulus was smaller in amplitude and had a longer peak latency in AD patients than in normal controls (Polich et al., 1990).
But the value of ERPs in AD research was only recently appreciated with the recognition that AD is a primarily synaptic disorder. In 2002, Dennis Selkoe at Brigham and Women’s Hospital in Boston, Massachusetts, drew attention to prior findings by others on synapse loss early in AD, and proposed that amyloid-β oligomers cause synaptic dysfunction before the deposition of amyloid plaques in the brain (Selkoe, 2002). Nine years later, researchers updated the model of biomarker changes associated with AD development (see ARF related Webinar) to include changes in synaptic function as one of the earliest markers in the preclinical stage of AD, occurring before tau-mediated neuronal injury or brain structure changes (Sperling et al., 2011).
Because ERPs provide a sensitive measure of synaptic function, they can provide clues into the early steps of AD, argued John Olichney, University of California, Davis, in his presentation at CTAD. ERPs constitute a millisecond-by-millisecond record of neural information processing; this stands in contrast to a temporal resolution of seconds to tens of seconds with functional MRI or PET. On the other hand, “ERPs are an older technique with limited spatial resolution compared to current neuroimaging methods,” wrote Olichney in an e-mail to Alzforum. That might be one reason they have been underutilized in clinical studies. Another is that “clinical validation of these sensitive measures has been slow and difficult work,” Olichney added.
ERPs as Diagnostic and Prognostic Markers
Two talks at the CTAD symposium showcased the clinical value of ERPs in predicting risk of AD development. The first, by Olichney, focused on changes in the N400 and P600 responses. Olichney’s group has previously developed a common paradigm for testing them. It begins with giving patients several phrases, each describing a category, for example, “a breakfast food” or “a continent.” Subsequently the patient hears a word that is either congruous (i.e., “waffle” following the breakfast food phrase) or incongruous (i.e., “table” following continent) with its preceding category. In healthy people, new congruous words elicit large P600 and N400 amplitudes, compared to incongruous words. Once the congruous words are repeated, i.e., no longer new, the N400 and P600 amplitudes shrink relative to their initial values; this shrinking is referred to as the “repetition effect.” Olichney and others have reported that the repetition effect is less apparent in mild AD patients than in healthy people (Olichney et al., 2006).
In a study on 32 patients with mild cognitive impairment (MCI), Olichney showed that abnormal ERP word repetition effects provide a marker for high risk of progressing to AD. MCI patients who had abnormal repetition effects (either N400 or P600) at baseline had an 87 to 88 percent likelihood of developing dementia three years later (Olichney et al., 2008). “The hazard ratio of the risk prognosis compares favorably with MRI and FDG-PET in predicting AD progression from MCI as established by ADNI [the Alzheimer’s Disease Neuroimaging Initiative],” said Olichney at CTAD (see ARF related news story).
For his part, Karim Bennys at Montpellier University Hospital in France described comparable data focusing on the P300 ERP component. This component can be broken into P3a, which measures focal attention originating from the prefrontal cortex, and P3b, which measures working memory. Bennys examined 71 patients of mean age of 71 years diagnosed as having MCI on the basis of the Petersen criteria (Petersen, 2001), compared to 31 cognitively normal controls with similar age and education level. These people were examined at baseline and again one year later using an auditory oddball paradigm.
One year later, 41 people in the MCI group were defined as progressors because their executive function had declined significantly compared to the 30 non-progressors. At baseline, the progressors showed increased N200 and P3b latency and lower N200 and P3b amplitudes than non-progressors. In the progressors, the gradient with which the P3b appears from the front to the back of the head was inverted, with a significant decrease of amplitude in the parietal cortex compared to non-progressors. By this point in time, 17 of the progressors met clinical criteria for AD. “The modification of N200 and P3b amplitude and latency appear very early in MCI patients at high risk of progressing to AD,” Bennys said at CTAD. He added that P3b amplitude alone is a robust marker for the conversion from MCI to AD, with 80 to 90 percent sensitivity and 70 to 80 percent specificity.
As the AD field is moving toward conducting trials in mildly or even presymptomatic people (see ARF related news story and ARF related Webinar), a recent paper has shown the value of ERPs in this group. A study by the Alzheimer's Prevention Initiative (see ARF conference series) used high-density ERPs to examine brain physiology in 10 presymptomatic carriers (average age of 34 years) of the E280A mutation in the presenilin-1 gene (Quiroz et al., 2011). These people performed a visual recognition memory test during which ERPs were recorded, and the recordings were then compared to those taken from 11 siblings without the mutation. The results, published this year in Neurology by first author Yakeel Quiroz at Boston University and the University of Antioquia in Medellin, Colombia, showed that despite identical test performance, mutation carriers had smaller amplitude from 200 to 400 ms in the frontal brain regions and a higher amplitude in the occipital regions than did controls (Quiroz et al., 2011). “I hear from our Colombian colleagues that one advantage of ERP biomarkers is that not everyone can afford to do a PET scan, but everyone has an EEG recorder,” said Budson, who is senior author on the paper.
Budson, Olichney, and others are in the process of looking at MCI and AD patients to determine which paradigms and ERP components provide the best measures for clinical diagnosis. “It may be that we will develop diagnostic markers made up of one or two different paradigms, or that some paradigms will be more or less sensitive for different stages of disease,” Budson told ARF.—Laura Bonetta.
This is Part 1 or a two-part series. See also Part 2.