Diagnostic Imaging
November 2000

Neurological Disease:
MR and PET vie in diagnosis of Alzheimer’s disease

Functional and structural studies make detection of disease possible before obvious symptoms appear

By Jane Lowers

Among imaging researchers who study Alzheimer’s disease, 25 years of work have produced an intriguing hypothesis. All things being equal, patients who live to old age without succumbing to heart disease, cancer, or some other illness will probably develop a degenerative brain disease. Alzheimer’s is a predictable side effect of aging that, somewhat surprisingly, may begin quite early in life.

Yet agreement is not universal. Some patients could live to be 200 without showing marked cognitive decline, while others show decreases in brain activity in their 20s. Some patients have memory loss in their later years but never progress to actual Alzheimer’s. The factors that distinguish who develops the disease, how, when, and why make the field one of the most dynamic in brain imaging.

"People at 85 or 90 are still developing Alzheimer’s. Is it inevitable? That’s an open question," said Neil Buckholtz, Ph.D., chief of dementias of the aging at the National Institute on Aging. "If we can get rid of other health problems, are we just opening ourselves up for Alzheimer’s?"

The ability to determine whether an individual patient will develop Alzheimer’s is the goal, said Dr. Clifford Jack, a professor of radiology at the Mayo Clinic and author of a number of papers based on structural MR and MR spectroscopy. A conclusive diagnosis at the first sign of cognitive decline could give patients access to treatments to slow or halt brain damage, particularly once current clinical trials show results. The technology to make that diagnosis already exists in the form of PET or MR imaging. The real question is which modality will end up being the one that primary-care physicians, Medicare, and others look to for the answers.

No single modality seems likely to outstrip the rest in the field. Structural MR offers widespread accessibility on standard 1.5-tesla scanners but requires extensive postprocessing. Functional MR, which is still in early trials, shows great detail but so far requires a high-field magnet. PET offers both early detection and monitoring but is limited to test centers within an hour’s reach of a cyclotron. MR spectroscopy may one day open up plaque visualization but needs work before that happens. At least one researcher thinks coregistering MR and PET images may push the field ahead.

Diagnosis of Inclusion

PET researchers are more ready than those who study MR to say that their technology can determine conclusively whether or not a patient has Alzheimer’s.

"Before, an Alzheimer’s diagnosis was a workup of exclusion: You ruled out other things, but you needed an autopsy to get a definitive answer," said Dr. Daniel Silverman, head of neuronuclear imaging at the University of California, Los Angeles. "Now, we’re showing sensitivity of PET to be 96% compared with autopsy-based diagnoses."

By the time the first symptoms appear in the form of memory loss, for example, a scan with fluorine-18 fluorodeoxyglucose can show significant decreases in metabolic activity in the posterior cortex.

"If you can catch patients early in dementia, before they could even be diagnosed with Alzheimer’s, that gives you the best chance of intervening before substantial neurodegeneration has occurred," Silverman said. "By the time the first symptoms appear, it’s easy to see the loss of brain tissue. No other modality is as well documented for showing early problems."

PET can also indicate that a perceived memory problem is not related to Alzheimer’s. The changes associated with Alzheimer’s occur specifically in the posterior cingulate, parietal, and temporal regions early on, followed by parts of the frontal cortex. Other forms of dementias, such as Huntington’s, have their own distinct patterns.

Other researchers, meanwhile, say structural MR can pin down changes in the size of the hippocampus or entorhinal cortex that can be used as markers for early diagnosis.

"The advantage of looking at the entorhinal cortex is that it helps track the transition from mild cognitive impairment (MCI) to Alzheimer’s," said Mony de Leon, D.Ed., a professor of psychiatry at New York University. "We’re at the juncture of evaluating how to measure it and what changes in its size represent an increased risk for Alzheimer’s."

MCI is a stage of cognitive decline that precedes, but does not always predict, Alzheimer’s. While de Leon views the cortex as a sort of early warning sign-it appears to shrink before the hippocampus does-significant shrinking in the hippocampus has been tracked by several institutions as a distinct marker of cognitive decline. In MCI, it may shrink by 15%, and patients with Alzheimer’s may show decreases of 25% to 70% compared with healthy patients.

One problem with structural MR readings is the daunting amount of postprocessing needed to make such measurements: Although Jack’s Mayo team uses three-dimensional imaging on 1.5-tesla scanners with a minimum of home-grown modifications, the volume measurements are still made with manual tracing. Well-trained staff can produce variations in volume well below 1%, Jack said, but the time involved may make structural MR unwieldy for clinical use.

How Much Risk?

While most of the emphasis in Alzheimer’s imaging has been at the point of MCI or advancing disease, PET imaging may help ferret out high-risk patients before they’re old enough to vote. At the Society of Nuclear Medicine meeting in July, Silverman and colleagues presented a study in which they calculated the levels of brain metabolic activity in middle-aged and elderly patients with low and high educational levels and extrapolated what their metabolic activity would have looked like at age 20. They found differences in the posterior cingulate cortex sufficient to indicate that cerebral metabolic decline could begin early enough in life to affect how much education patients obtain.

Silverman was careful to point out that, although associations between cerebral metabolism, educational level, and Alzheimer’s disease have been identified, they must currently be regarded as correlative, and much more work with young and pre-adults will be needed before cause-and-effect relationships could be established.

"What is clear at this stage is that significant differences in posterior cingulate metabolism between high-education and low-education groups are detected by PET, even decades before Alzheimer’s disease is typically diagnosed," he said.

Differences between patients with high and low educational levels were more pronounced than differences between patients who carried the apolipoprotein E-e4 allele, which has been linked to a higher incidence of Alzheimer’s. Duke University researchers found that patients with two copies of apo E-e4 had a risk of developing the disease 12 times greater than normal. Although only 3% of the population carries two copies, 23% carry one copy and face a threefold greater risk than normal. Most of the population carries two copies of the apo E-e3 allele, but a small minority carries one or two copies of apo E-e2, which is thought to convey some protective benefit.

Functional MR is also studying genetic differences, focusing on the hippocampus. In a study published in the Aug. 17 New England Journal of Medicine, UCLA researchers tracked brain activation during memory tasks for patients with the apo E-e4 and E-e3 alleles and found that the brains of patients with apo E-e4 have to work harder to accomplish the same tasks as patients with apo E-e3. Lead author Susan Bookheimer, Ph.D., and colleagues used 3-tesla scanners with echo-planar imaging and found that the hippocampal regions produced higher intensity signals in patients with apo E-e4 alleles during memory activation. In a subset of patients measured two years later, the amount of activation in the left hemisphere correlated with decline in verbal recall ability.

Despite the increased risks, some patients who carry the apo E-e4 allele and some patients who develop MCI never progress to Alzheimer’s. A study published in the Sept. 12 issue of Neurology suggests that an as-yet-unidentified recessive gene may be involved: A survey of Arab patients in Israel, who often have consanguineous marriages, found that 20% of patients older than 65 had the disease, more than double the global average. Prevalence of apo E-e4 was much lower than in other populations, however.

Monitoring Progression

With cognitive tests, genetic markers, and imaging in hand, a primary-care physician can do a reasonably good job of diagnosing Alzheimer’s, but the clinical changes that mark the disease’s progression are more obscure. Each new finding, from metabolite levels to physical shrinking or amylotic plaques, opens up a new avenue to discover the disease’s mechanisms and potential vulnerabilities.

Jack and colleagues tout the prospects for MR spectroscopy, particularly for the presence of N-acetylaspartate (NAA) and myo-inositol. Both, he said, may serve as proxies for the presence of plaque. Myo-inositol levels appear to rise first, as plaques develop and inflame surrounding tissue. Once the plaques or the inflammation begins killing off neurons, NAA levels drop accordingly. MR spectroscopy may gauge the success of anti-inflammatory drugs now being considered as a possible treatment for the disease.

"If you’ve got a group of patients with MCI, you can’t necessarily tell which ones are likely to decline," Jack said. "This might be a way to tease out who has a high or low probability, and that’s a very valuable use of imaging."

The elusive master indication for Alzheimer’s imaging, one that would provide both certain diagnosis and a way to monitor treatments, is amyloid plaque. Given that a large plaque may stretch across only 100 mm, imaging has been difficult to achieve in vivo: PET, MR spectroscopy, and MR microscopy all require improvements in resolution to achieve reliable results.

While PET and MR vie for the leading role in Alzheimer’s imaging, few studies have compared them outright. After spending some time with both, de Leon suggests that a combination of the two, in the form of coregistered images, which can find and sample small areas and also relate them to the larger background, might provide the refinement the field seeks.

"PET lags on measuring small volumes, but it’s exquisitely sensitive at identifying vulnerable regions of the brain," he said. "Everyone tries to push the bottom line of what’s going to be most useful for a drug trial. In the best world, I think it would be a combination.