Alzheimer’s is among aging’s most enigmatic diseases. Despite decades of research, there is no cure, no preventative, and little in the way of treatment. Moreover, the disease is still difficult to diagnose before deteriorating cognitive abilities make it obvious too late for treatments that could add meaningful years to a patient’s life if given at an early stage.
“Every single Alzheimer’s agent for the past 25 years has failed egregiously,” said Derek Lowe, a medicinal chemist at Vertex Pharmaceuticals in Boston and long-time drug discovery blogger. “If we’re going to have any hope at all, we’ve got to get in as early as possible.”
A handful of recently announced trials that are doing just that. Last week (July 15), for example, Novartis announced its partnership with the Banner Alzheimer’s Institute in Phoenix to test two drugs known to target the disease’s characteristic amyloid plaques in the brain. The trial aims to recruit more than 1,300 healthy individuals, ages 60 to 75, carrying two copies of the APOE4 gene (who are therefore at high risk of developing Alzheimer’s), to see if these drugs are effective at delaying or preventing the onset of dementia if taken before symptoms appear.
Thus, one highly sought-after goal of Alzheimer’s research is to identify blood-borne biomarkers that can predict the eventual development of dementia. Unfortunately, efforts to track down reliable blood biomarkers for neurodegeneration have yielded precious few leads. Likely due in part to degradation by proteases and clearance by the liver and kidneys, blood levels of classic Alzheimer’s players such as amyloid beta (Aβ) and tau proteins are not predictive of future cognitive decline. “What we measure in the blood does not reflect what’s happening in the brain,” said clinical neurochemist Henrik Zetterberg of the Sahlgrenska Academy at the University of Gothenburg in Sweden, and the University College London Institute of Neurology.
Researchers have also turned to screening diverse molecules in groups of healthy patients to see if collective panels might be indicative of future dementia. This year, two groups announced successes using this strategy to develop screens that were reportedly 90 percent and 87 percent accurate at predicting, respectively, the onset of dementia in healthy people and the progression from mild cognitive impairment (MCI) to Alzheimer’s disease.
In March, Howard Federoff of Georgetown University Medical Center in Washington, DC, and his colleagues published in Nature Medicine on 10 lipids—identified through screens of many thousands of metabolites—that could distinguish those who would develop cognitive impairment from those whose minds would remain sharp with a 90 percent sensitivity (meaning it will miss, or give a false negative to, every one in 10 people who do have the disease) and a 90 percent specificity (it will yield a false positive diagnosis for every one in 10 people who don’t have the disease). “There was a fair amount of excitement” about those results, said neurologist Douglas Galasko of the University of California, San Diego, School of Medicine.
Then, in a study appearing this month (July 8) in Alzheimer’s & Dementia, Simon Lovestone of King’s College London and University of Oxford and his colleagues published on a panel of 10 proteins that could predict progression from mild cognitive impairment (MCI) to full-blown Alzheimer’s with a reported accuracy of 87 percent (calculated as the total number of true results, both positives and negatives, over the total number of tests given).
Both sets of findings were widely covered, including by The Scientist, and many touted the studies as promising steps in a slow-moving field. At the same time, discussions erupted in the blogosphere on the meaning of “accuracy,” with some accusing the media, institutional press offices, and even the researchers themselves of misrepresenting the significance of the blood tests.
So what do these two high-profile papers really mean for Alzheimer’s research? It depends on context—these tests are not intended to be used to screen everyone over the age of 60—and on the reliability of these screens in future studies. But if the results hold up to further testing, they may well help the field continue to shift its focus to early diagnoses.
“It’s possible that [one day] we may actually find therapeutics that delay, among those who test positive, the emergence of symptoms,” said Federoff. “And if we’re successful, it could change the face of how we ultimately prepare for people who would test positive.”
Not for everyone
When Lovestone began proteomics work to look for blood-based biomarkers of Alzheimer’s disease, he didn’t expect to succeed. “I didn’t really think there would be a signal at that point,” he admitted. But over the course of 10 years of scouring blood samples from Alzheimer’s patients, he and his colleagues did find a signal, in the form of more than 20 proteins—different levels of which correlated with the development of dementia. In this month’s paper, the team describes a panel of 10 proteins, identified from a short list of 26 previously identified and validated markers, that could predict the progression from MCI to Alzheimer’s with 85 percent sensitivity and 88 percent specificity.
“It looks like, from Howard’s work and our work and others’, that we might actually find that thing that I thought was impossible that is, a signal in blood,” Lovestone said.
But just like Federoff’s publication a few months earlier, the report was met with mixed reviews. In response to each study, University College London biophysicist David Colquhoun and others pointed out that, with a relatively rare disease like Alzheimer’s, which afflicts some 5 million Americans, a widely used screen of 90 percent sensitivity and 90 percent specificity would yield more false positives than real ones. “You’d be telling people they were in for a terrible, lingering death, and they’re not,” Colquhoun told The Scientist. “It would be madness.”
Lowe was also outspoken about the problem of prevalence. “[Even] 99 percent is not very good for a diagnostic test,” he said. “This is a constant problem in all areas of diagnostics. If you’re going after any disease with a reasonably low prevalence rate, it just means your accuracy has to be so much higher.”
Of course, the researchers aren’t suggesting that these tests be used in the clinic without further refinement and testing. Since Federoff’s group published on its Alzheimer’s blood test four months ago, the researchers have added additional metabolites to the panel for an accuracy of more than 99 percent, he said, adding that his group is now in discussions with potential collaborators to look at the stored blood of Alzheimer’s patients to see how the test performs on a much larger population. “We want to look at many thousands of people,” said Federoff.
Moreover, Lovestone makes the point that most Alzheimer’s tests would likely never be administered to the broader population, but rather only those patients who come to their doctors complaining of cognitive problems. “The clinical purpose [our test] might have some use for—in the future, if it replicates—is in a memory clinic, where the prevalence of conversion [from MCI to dementia] is 20 percent or more,” he said, in which case the statistics are much more favorable. “If the test replicates at the same figures, then it has real utility, especially for clinical trials.”
Federoff is similarly undeterred. “Like a lot of new types of investigation, people wind up having a full range of enthusiasm and skepticism,” he said. “And I think we have to be responsive to the skeptics and continue to move the work forward.”
