Functional MRI May Be Useful for Monitoring Cognitive Decline in the Elderly


by Jim Schnabel

September 8, 2010

Alzheimer’s researchers have long wanted to find better ways not only to diagnose the disease but also to monitor its progression from the earliest stages. A highly sensitive measure of Alzheimer’s-affected brain functions, for example, could enable clinical trial researchers to determine more easily whether a drug is working, thus making such trials faster and cheaper, as well as more accurate. A new study suggests that functional magnetic resonance imaging (fMRI), a technique currently used mainly for neuroscience research or to guide brain surgery, could be useful in this clinical role.

“It does seem to have potential utility in monitoring changes in brain function over time, which no one has ever shown for fMRI before,” says Brad Dickerson, a neuroimaging researcher at Massachussetts General Hospital and Harvard Medical School, and a co-author of the study.

The study appears in the June issue of Neurology and was led by senior author Reisa Sperling, of the Athinoula A. Martinos Center for Neuroimaging at Brigham and Woman’s Hospital, also part of the Harvard Medical School system. Sperling is also a member of the Dana Alliance for Brain Initiatives.

At the outset of the experiment, Sperling and Dickerson and their colleagues assembled two groups of elderly individuals: 21 showed strong cognitive functioning, while the remaining 30 scored poorly enough on a standard, interview-based dementia rating system to suggest that an early progression towards dementia had begun.

Having defined these groups, Sperling and her colleagues gave each person additional learning tests designed to measure cognitive abilities, and also used functional magnetic resonance imaging (fMRI) to record neural activation patterns in the hippocampus of each subject while he or she performed a standard memory test. The fMRI-monitored memory test required the proper association of names and faces—a function that requires the proper working of the hippocampus, an important memory region that is hit early and hard in Alzheimer’s disease. Two years later, Sperling and her colleagues performed the same set of tests on the 51 subjects, including fMRI scans. (An fMRI scan is sensitive to the flow of oxygen-carrying blood in brain regions; more flow suggests higher brain cell activity.)

The subjects whose scores declined the most on basic clinical tests showed the greatest loss of fMRI-measured activation in the right hippocampus, which typically is more active than the left when faces and other visual memories are stored and recalled. But there was another, more striking result:  These fastest-declining subjects also had the highest fMRI-measured right-hippocampal activation levels at the start of the test. The researchers aren’t sure why this would be. Among other possibilities, they speculate, the hippocampus might be compensating with greater activity in the earliest stages of the dementia process. “Hints of this inverted-U-shaped curve of hippocampal activation were seen in some previous studies,” Dickerson says.

Researchers have been relatively enthusiastic about the more common clinical type of MRI, known as “structural” MRI, to predict Alzheimer’s and monitor its early progression [see story, “Advances in Neuroimaging Help Refine Search for Biomarkers for Alzheimer’s Disease”]. But because structural MRI in this role aims to detect changes in the volume of the hippocampus—in other words, large-scale losses of neurons—in principle it should be less sensitive than fMRI. Hippocampal neurons should die only after their activity has gone into a long decline. Sperling, Dickerson and their colleagues scanned the hippocampi of their subjects with structural MRI and found, as expected, no significant changes in volume during the two-year period, despite clear evidence of activity changes on fMRI.

Dickerson notes that the results of the study were reassuring for another reason:  Functional MRI measurements generally require sophisticated analyses and have acquired the reputation among some researchers of being too variable, or “noisy.” But the subjects who showed no significant changes in their cognitive scores also showed no changes in their fMRI measures of hippocampal activation. “This is really the first time anyone has shown that fMRI is fairly stable over two years in people who start out normal and remain normal,” says Dickerson.

Functional MRI is expensive, requires specialist training, and is still rarely used in a clinical setting. But Dickerson thinks that it might be possible to combine it with a structural MRI scan in a way that makes it competitive with another functional brain-imaging technique, a positron-emission tomography (PET) scan, which typically costs several thousand dollars.

“We don’t really know, because for the most part nobody’s really tried to do that yet. But I do think it’s a direction we should be moving in,” Dickerson says.

Ansgar Furst, a neuroimaging specialist at a Veterans Affairs health center in Palo Alto, Calif., and co-author of a commentary on the Sperling group’s paper in Neurology, calls it an elegant and thought-provoking study.  He notes that more studies are needed to establish that a characteristic curve of fMRI-measured hippocampal activity can usefully predict Alzheimer’s-type dementia or track its course. If fMRI can indeed live up to its promise, further work would be needed to integrate this technique into an ordinary clinical setting. “But altogether I’m cautiously optimistic,” says Furst.