29 May 2007. Functional imaging (fMRI) of brain activity during cognitive testing has shown that, in the downtime between tests, the resting brain is not really at rest. Instead, the scans detect coordinated activity in a distributed network of cortical regions when the brain is not engaged in a specific task. The network, commonly called the default-mode network, includes the hippocampus and other regions that are known to be impaired early in Alzheimer disease, and to be targets for early amyloid deposition (see Forum Interview with Randy Buckner). Michael Greicius and colleagues at Stanford University in Palo Alto, California, found that people with early AD displayed decreased default-mode network activity during breaks between simple motor tasks, suggesting that changes in default-mode network activity could be an early diagnostic marker (see ARF related news story). Another study found decreased default-mode network activity in patients with MCI challenged with visual or memory tasks (Rombouts et al., 2005).
Functional MRI is a complex technique that’s not widely available outside academic brain imaging centers. It requires asking patients to solve puzzles while undergoing MRI and having the expertise to analyze their responses. But diagnosing default network disconnection may be simpler than that, according to new data from Greicius. Speaking at the American Academy of Neurology meeting in Boston on 3 May, he showed that default-mode network activity was detectable by fMRI without requiring a task or stimulus. Measured in this way, the resting-state connectivity between different brain regions was weaker in people with mild cognitive impairment (MCI) than in healthy controls.
In the study, the researchers analyzed spontaneous BOLD (blood-oxygenation-level dependent) signals during 8-minute, task-free MRI scans on nine patients with mild cognitive impairment and nine healthy controls. Charles DeCarli and coworkers at the University of California at Davis performed the scans. Then, Greicius and coworkers performed an automated analysis to identify regions of correlated activity. Next, they compared how well the subjects’ network activity pattern matched a healthy pattern taken from reference scans. The scientists found significant decreases in network connectivity in the bilateral hippocampus and precuneus of the MCI group compared to the healthy controls.
The results suggest that fMRI-based detection of resting default-mode network activity could form a simple functional test for the earliest stages of AD. Because the subjects did not have to complete a task during the imaging, the measure would be easier on patients and independent of learning effects. The test could even be done on people in advanced stages of dementia. An advantage to using resting-state measures in MCI or AD is that issues of poor blood flow at baseline or regions of atrophy should not confound results, as the analysis is not looking at the magnitude of the BOLD signal or its change with a task, but instead measures the correlation between regions. The sensitivity of the test was comparable to structural measures that gauge changes in hippocampus size, Greicius said.
Greicius pointed out a recent study that reported the detection of default-mode network activity in anesthetized monkeys (Vincent et al., 2007). This data supports the idea that the default-mode network is a critical basal state of brain activity. Thus, its disruption may reveal a fundamental underlying pathology early in the course of AD.
“This work provides more evidence that important default network function is affected early in the disease process,” said Reisa Sperling from Harvard Medical School in Boston, who heard Greicius’s talk. “These functional changes will predict structural changes,” she added, but questioned whether resting-state methodology is the best way to measure network breakdown in the early stages. “You can get similar network data from task-related scans, and it’s possible that stressing the network with a task will reveal even earlier changes, like the difference between doing a cardiac stress test versus taking a resting EKG.”
While it is possible to measure network connectivity from data collected during the short rest periods in task-dependent fMRI scans, collecting a solid block of 8-minute activity gives better results, Greicius said. “I hope this data will convert some people to want to tack a resting-state fMRI to their protocols,” he said. “It would be nice if the Alzheimer Disease Neuroimaging Initiative would do it.”—Pat McCaffrey.