An Alzheimer's Fingerprint?

For decades, researchers have been trying to devise a reliable diagnostic test for Alzheimer's disease. But the goal has proven elusive. Today, even with the best techniques available, patients are technically classified as having "possible" or "probable" Alzheimer's, with a definitive conclusion becoming possible only upon death, when the brain can be autopsied. That's why a study appearing today in the online version of the journal Annals of Neurology could be so significant. In the new study, Kelvin Lee and Erin Finehout, chemical and biochemical engineers at Cornell University, joined forces with neurologist Norman Relkin at Weill Cornell Medical College to develop and test a new approach to diagnosing Alzheimer's, based on an analysis of proteins in patients' spinal fluid. The approach won't translate into a commercially available test for several years. But when and if it does, it will be a major breakthrough.

It can't come soon enough. Already some 4.5 million Americans have Alzheimer's disease. By 2050, as increasing numbers of seniors reach the age of 85 and older, the tally could swell to as much as 16 million. Researchers are racing to develop drugs that would slow the progression of the disease—but in order for these treatments to be useful, one needs diagnostic tests to determine who can benefit. So how does the possible new testing method work? NEWSWEEK's Anne Underwood spoke with Dr. Relkin. Excerpts:

NEWSWEEK: What is the basis for this potential new test?

Relkin: It's a test for specific proteins whose levels are different in people with Alzheimer's than in other people—either normal people or patients with other forms of dementia. We found 23 proteins with altered levels in the Alzheimer's patients. Together they form a kind of fingerprint of the disease.

What are the proteins you pinpointed?

Some were involved in binding A-beta [the toxic protein in Alzheimer's]. Others play a role in inflammation, and others, in synaptic function—the interaction between brain cells. These are all processes associated with Alzheimer's, so it makes sense.

Had these proteins been identified before as potentially useful in diagnosing Alzheimer's?

Some of these proteins had been reported to relate to Alzheimer's in past studies, and others had not. However, none had stood out on their own as a diagnostic. It was only when considered together as a group that their power to identify Alzheimer's patients became evident.

What about the usual suspects— like A-beta 42 [the most toxic form of A-beta] and tau [a marker of neuronal death]? Did those turn up in your mix?

Those were not among the markers that came out in our study. I don't want to suggest they're not valid [targets for testing]. But because of the nature of our technology, they didn't fall into this pattern. In an ideal world, you might want to include those, but in a sense, you don't need them as long as you can get a characteristic group of proteins that will reliably show who has the disease and who doesn't.

How reliable was the test?

We actually report on two studies in this paper. In the first one, we had an initial group of 34 patients and 34 age-matched controls with other forms of dementia. The importance of this first group was to determine what the markers are. Autopsy studies showed the test was 94 percent [accurate]. We then performed a small validation study in a second group of patients with probable Alzheimer's to see if the results held.

But the validation test had a somewhat lower accuracy rate.

These were living patients, so we couldn't compare against autopsy results. Even in expert hands, clinicians only identify about 90 percent of cases correctly. [The lower accuracy rate] is not necessarily a weakness in our test, but could be a limitation in clinical diagnosis.

But you're still quite a long way from having a commercial test, aren't you?

I would estimate that it will be at least three to five years before a commercial test is available. I base that on the length of time it will take to carry out larger validation studies and to put the test into a form in which it is practical for commercial or clinical use. Right now, it is a laborious process to detect these markers. But the technology exists to measure them in other ways.

The major drawback I see here is that you have to put people through a spinal tap to do this test.

That is a drawback, I agree. It would certainly be wonderful to have a noninvasive test or a less invasive test, such as a blood test. But as a matter of practical consideration, spinal fluid is the only bodily component we can assess that's in direct contact with the brain. Blood is one or two steps removed, and it contains a more complex mix of proteins influenced by functions of all organs of the body. That's not an insoluble problem, but it does make it harder.

Would this test also work for early detection of Alzheimer's?

We don't know. There is evidence from other studies that some biological markers remain constant from the early to late stages, while others change as a function of the disease. Unless we test this combination of markers in people with early stages, we won't know for sure.

What tests exist now?

Companies like Athena Diagnostics and Innogenetics make commercial tests for A-beta 42, total tau and hyperphosphorylated tau. They are not used in routine evaluations, but as an adjunct diagnostic tool by some clinicians.

Early last year, news reports trumpeted another possible test— a test for ADDL's [small clumps of A-beta strands that show up in spinal fluid early in the disease process.]

I'm excited about those, too. But the technology for quantifying levels is still being developed. I've worked with three groups on detection methods, but all of them still need work. We can't detect ADDL's with our current proteomic test because their concentration is very low, they're constantly being converted to other forms, and the gels we use denature them. ADDL's are a very hot area from the standpoint of potential diagnosis and treatment. But our proteomic test is much further along than this. All we need is further validation studies.

What about brain imaging? There are ways of scanning for plaques in the brain.

In my program, we're collaborating with the University of Pittsburgh to study imaging. It's a breakthrough technique, but all such techniques have their limits. We need to develop many approaches simultaneously. No test will be 100 percent accurate. In the same way that we need multiple treatments, we need a full armamentarium of diagnostics that can be applied in cost-effective ways.

An Alzheimer's Fingerprint? | News