By the time a person starts exhibiting the memory problems and other symptoms of Alzheimer’s disease, a catastrophic cascade of cellular events has been playing out inside their brain for years or even decades. Misfolded proteins and fragments of protein have been ever so slowly clumping together, forming microscopic plaques and tangles that interfere with the function of neurons. Eventually, these brain cells die and this neurodegeneration takes its toll on memory and cognition.
The slow-burning fuse of Alzheimer’s creates a terrible predicament for doctors, patients and the scientists working to develop therapies: Once symptoms appear, so much damage has already been done that it may be impossible to reverse. This might explain why clinical trials for experimental drugs intended to slow or reverse the cellular damage caused by Alzheimer’s disease have so far yielded one demoralizing failure after another. (Several drugs have been approved to treat the symptoms by boosting certain neurotransmitters, but none target the underlying cause of the disease.)
Many researchers now believe that those clinical trials may have failed because the experimental drugs were given too late in the course of the disease. Some of the same drugs — or new ones — might have a better chance if administered earlier, even before symptoms appear. It’s a hypothesis with logical appeal, but it’s still unproven. And it depends on the ability to detect signs of Alzheimer’s in people who aren’t yet experiencing symptoms.
In recent years, scientists have developed several methods for doing just that. The best-established biological markers are based on brain scans and tests of cerebrospinal fluid. More recently, researchers have finally made major breakthroughs in their decades-long pursuit of a simple blood test for Alzheimer’s-related proteins.
Biomarker tests have clarified how the disease progresses, and some are now widely used in Alzheimer's clinical trials. They’re not yet commonplace in clinical practice, but that’s likely to change in the next few years.
“There’s evidence that biomarkers start to show changes 20 years before onset of dementia,” says Andrew Saykin, director of the Alzheimer’s Disease Research Center at the Indiana University School of Medicine. That presents a huge window for early therapeutic intervention. Once effective therapies finally become available, many researchers think they’ll work best when given as early as possible. “It’s very tough to reverse neurodegeneration once it’s happened, but if we could prevent it from happening there’s really an opportunity to have a major impact on the disease,” Saykin says.
Visualizing the problem
Alzheimer’s disease is the most common cause of dementia, affecting an estimated 5.8 million Americans. Memory problems are often the first sign, but confusion and other cognitive difficulties appear as the disease progresses, along with changes in mood, behavior and personality. To make a diagnosis, doctors typically interview the patient and a family member, and conduct cognitive assessments. They may also order brain scans or other tests to rule out a stroke or another type of dementia.
A definitive diagnosis, however, can come only after a person has died, by examining slices of brain tissue under a microscope. The pathological signature of Alzheimer’s has two components: plaques, which are clumps of a protein fragment called beta-amyloid that can be seen in the spaces between cells, and tangles, twisted strands of a protein called tau that form inside of cells. This combination of abnormalities is what distinguishes Alzheimer’s disease from other types of dementia that are also caused by abnormal deposits of proteins, including Lewy body dementia and fronto-temporal dementia. Scientists still don’t understand exactly how amyloid and tau cause neurodegeneration, but most believe these two compounds are key links in the disease mechanism. (An increasingly vocal minority of researchers, however, argues that more investigation of alternative mechanisms is needed.)
In 2002, Alzheimer’s researchers made a huge breakthrough with a method that made it possible to visualize amyloid buildup in the living human brain for the first time. Up to that point, more than a decade of pathology studies, mouse experiments and other research had implicated amyloid as a culprit in the disease, but there was no way to detect it in the brains of living people. The solution, developed by a team led by researchers at the University of Pittsburgh, was a compound that incorporates a radioactive isotope of carbon (C11) and binds to amyloid, making it visible on a positron emission tomography (PET) brain scan.
“Amyloid PET scanning was really just transformative,” says William Jagust, a neuroscientist the University of California, Berkeley. The new scans revolutionized clinical trials for Alzheimer’s disease, he and other experts say. Many leading candidate therapies are drugs or antibodies designed to clear amyloid from the brain, or at least stop it from building up. As researchers began using amyloid-targeted PET, they realized that a significant proportion of the patients enrolled in these trials didn’t actually have elevated amyloid in their brains, despite having memory impairments or other cognitive symptoms.
“About a third of the people who were enrolled in these clinical trials for ‘Alzheimer’s disease’ actually turned out not to have Alzheimer’s disease,” says Clifford Jack, a brain imaging researcher at the Mayo Clinic in Rochester, Minnesota. “That’s a really huge problem.”
Now virtually all clinical trials of anti-amyloid therapies use amyloid PET to screen patients for enrollment and to track whether the therapy being tested actually removes amyloid from the brain. Newer compounds based on a different radioactive isotope, Fluorine-18, have expanded their use in research even more. The original C11 compound has a half-life of about 20 minutes, which limits its use to hospitals with a cyclotron (a particle accelerator) and chemists on-site to manufacture their own supply, says Richard Carson, a PET imaging researcher at Yale University, and coauthor of a 2019 paper on the method in the Annual Review of Biomedical Engineering. “You have to be able to make the molecule very close to where you’re going to use it,” he says. In contrast, the new F18 compounds have about a two-hour half-life, which means they can be made in one place and shipped to hospitals that don’t have the capacity to make it themselves.
Amyloid PET has also made it possible to test experimental therapies in people who show signs of amyloid accumulation but aren’t yet experiencing memory problems. The largest such effort to date is called the A4 study, which has enrolled 1,169 people between the ages of 65 and 85 with abnormal amyloid PET scans but normal scores on cognitive tests (A4 stands for Anti-Amyloid Treatment in Asymptomatic Alzheimer’s). Participants were randomly assigned to receive either anti-amyloid therapy or a placebo, and researchers used cognitive tests and additional PET scans to track each person for four-and-a-half years.
The trial, which began enrolling patients in 2014, was nearing completion but has experienced delays due to the pandemic, says Reisa Sperling of Harvard Medical School, one of the study’s leaders. The trial will be an important test of the early-is-better treatment strategy. Sperling expects to have results by early 2023.
In addition to changing clinical trials, amyloid PET and other biomarkers have changed the way researchers conceptualize Alzheimer’s, says neurologist Michael Weiner of the University of California, San Francisco, who leads the Alzheimer’s Disease Neuroimaging Initiative. The project, funded by more than $190 million from the federal government and the pharmaceutical industry since its launch in 2004, focuses on developing Alzheimer’s disease biomarkers. “We don’t think of the diagnosis being made by symptoms [anymore], we think of identifying pathology using biomarkers,” Weiner says. “It’s a huge shift in our thinking and our diagnostic and therapeutic approach.”
That shift is spelled out in detail in research recommendations published in 2018 by the National Institute on Aging and the nonprofit Alzheimer’s Association that urge researchers — though not yet clinicians — to define Alzheimer’s using biomarkers, much as clinicians already do for other conditions, such as using blood sugar tests to diagnose diabetes and bone density tests to diagnose osteoporosis. Doing so, the researchers argue, would enable researchers to design studies based on subjects with similar biology rather than focusing on cognitive symptoms, which can have differing biological underpinnings. “In that paper we basically say that someone has Alzheimer’s disease if they have positive amyloid and positive tau biomarkers,” says Jagust, who was one of the authors. A patient’s symptoms don’t factor into this definition.
The ability to visualize tau in the living human brain has become possible only in recent years, with the development of new compounds that make tau visible on a PET scan. The emerging research on tau PET scans suggests that they may match up better with a person’s cognitive status than amyloid PET scans do, Jagust says. “If you have a whole lot of tau in your brain, it’s very improbable that you’re going to be cognitively normal, whereas you can have a whole lot of amyloid in your brain and still be cognitively normal.”
That fits with the idea that tau accumulation occurs at a later stage of the disease than amyloid accumulation. Indeed, the overall picture emerging from Alzheimer’s biomarker research is that different biomarkers track different stages of the disease process, much as Jack and colleagues proposed in a widely cited paper more than a decade ago (see graphic below).
The earliest indicator that trouble is brewing appears to be tests that detect amyloid fragments circulating in the cerebrospinal fluid (CSF) that bathes and cushions the brain and spinal cord. The type of amyloid implicated in Alzheimer’s disease is a fragment created when enzymes break down a much larger protein. What this protein and its fragments normally do in the brain, and why this particular fragment clumps together in the brains of Alzheimer’s patients, is poorly understood. Somewhat counterintuitively, researchers find that low levels of amyloid circulating in the CSF correlates with more amyloid building up in the brain.
“At this point, it looks like the CSF biomarkers become abnormal a little bit before the imaging biomarkers,” says Jack. That makes sense, he says, because the CSF tests probably indicate that something has gone wrong with amyloid production or turnover that’s causing it to build up in the brain. But it has to amass for a while — possibly years — before it can be seen with PET. “It takes a pretty substantial amount of amyloid to be deposited before you can actually detect it out of the noise with amyloid PET,” he says.
Tau appears to begin accumulating later, overlapping considerably with neurodegeneration and memory problems but preceding the severe cognitive decline that robs people of their independence in later stages of the disease. By that time, atrophy of certain brain regions can often be seen on MRI scans, which show brain anatomy in greater detail than PET does.
The distribution of tau in the brain differs from that of amyloid, says Jagust. “Amyloid seems to arise in multiple areas of the brain much at the same time, almost like it’s like blossoming everywhere,” he says. Tau, in contrast, appears to pile up first in the medial temporal lobe, a region critical for memory, and then spread to other areas, PET studies have shown. In a 2017 paper, Jagust and colleagues suggested that focusing on tau accumulation in specific brain regions instead of just detecting its presence or absence might provide a better indicator of neurodegeneration and cognitive symptoms.
While many researchers remain focused on refining biomarkers for amyloid and tau, others are investigating additional markers that could help push detection of the disease even earlier. These include genetic tests to assess an individual’s risk (see sidebar), tests for olfactory deficits or other sensory abnormalities that may predict Alzheimer’s disease, and digital diagnostics that use computers or other devices to assess cognition without a visit to the doctor’s office. Other approaches include investigating whether machine-learning algorithms can help neurologists pick up signs of trouble in brain scans, and new PET compounds that can detect the loss of synapses, the tiny gaps between neurons that are essential for communication from one cell to another.
Even the more established biomarkers for Alzheimer’s disease aren’t yet widely used in clinical practice. Medicare, the government health insurance program that covers most older Americans, does not cover PET amyloid scans for most patients, and few opt to pay out of pocket for a test that costs several thousand dollars and likely wouldn’t change doctors’ treatment plan, given the absence of disease-altering drugs. Medicare makes exceptions, however, for certain cases in which a negative PET amyloid scan would aid diagnosis by ruling out Alzheimer’s disease, and for clinical research. (The FDA just approved the first tau PET test at the end of May.)
One recent study examined the medical records of 11,409 Medicare patients who had received an amyloid PET scan under one of these exemptions. The researchers, led by Gil Rabinovici at UCSF, reported last year in the Journal of the American Medical Association that amyloid PET scan results altered physicians’ treatment plans. Doctors were more likely to prescribe certain drugs for people with a positive PET scan, even if they hadn’t been diagnosed with Alzheimer’s disease. These drugs alter neurotransmitter function and have been shown to improve cognition in some people with Alzheimer’s disease, although they do not prevent neurodegeneration. Whether the patients benefitted in the long run is the focus of a follow-up study now underway.
How patients themselves react to amyloid PET scan results was the focus of an even more recent study. Emily Largent and Jason Karlawish at the University of Pennsylvania and colleagues interviewed 80 cognitively unimpaired participants of the A4 study after their amyloid PET scan. Elevated amyloid, the researchers explained, means a higher but uncertain risk of eventually developing Alzheimer’s-related memory loss. Those who tested positive saw the result as “more serious and sensitive” than other types of medical test results, the researchers reported in February in the journal PLOS ONE. “A colonoscopy isn’t going to change who I am ... this is my brain involved,” one participant told the researchers. Many of those who tested positive also feared stigmatization or social isolation as a result and reported spending more time contemplating and planning for their future.
Karlawish says the findings underscore the need for doctors to help patients prepare for amyloid PET scans and discuss the results with them afterwards. “Before you order a test, particularly one that has consequences like this, you need to sit down and make sure the person understands what they’re walking into,” he says.
Clues in the blood
It’s a scenario that could soon become more common. One of the fastest-developing areas of Alzheimer’s biomarker research is on tests that can be done with an ordinary blood sample. It’s a goal that researchers have pursued for decades but begun to achieve only in the last few years.
One reason it’s been so difficult is that the blood-brain barrier, a cellular defense wall that protects the brain from any toxins circulating through the rest of the body, prevents potential biomarker molecules from getting into the bloodstream in significant amounts. Another reason is that, compared with CSF, blood contains many additional proteins that can foul up the analysis. “You have something like 50,000 times less signal to noise in the blood,” says Randall Bateman, a neurologist at Washington University in St. Louis. Bateman’s research group is one of several around the world that have recently begun to overcome these obstacles with new, super-sensitive assays that can detect minuscule amounts of amyloid or tau in blood samples.
One of the most promising methods involves a combination of techniques. The first is immunoprecipitation, which uses immune system proteins called antibodies to seek out a particular protein in a sample and concentrate it. The concentrated protein can then be fed into a mass spectrometer, a machine used by chemists to identify and quantify compounds by their molecular weight.
In 2017, Bateman and colleagues reported that a blood test based on this approach matched amyloid PET scan results in 41 adults age 60 or older. All study subjects who had positive PET scans had amyloid blood levels below a certain threshold, indicative of a higher risk of Alzheimer’s. However, several subjects with negative PET scans also had amyloid blood levels below this threshold, suggesting that improvements may be needed to avoid false positives.
Bateman’s group and others have been working to refine the tests. In 2018, researchers in Japan and Australia published details of a similar blood test that matched amyloid PET results with about 90 percent accuracy in a larger sample of 254 older adults. Preliminary findings reported earlier this year suggest that these blood tests have promise for predicting the onset of Alzheimer’s disease in people who don’t yet have symptoms or are experiencing only mild cognitive impairment.
Researchers have also made recent progress on a blood test for tau. In a paper published July 28 in the Journal of the American Medical Association, an international team of researchers reported that a blood test for a specific subtype of tau accurately distinguished Alzheimer’s disease from other neurodegenerative diseases in 1,402 study subjects from Colombia, Sweden and the United States. The blood test was as accurate as CSF tests and PET scans for tau. In one arm of the study, the blood test picked up signs of tau accumulation about 20 years before the onset of symptoms in people with a gene mutation that causes a rare, early-onset form of Alzheimer’s.
If blood tests for amyloid and tau prove to be reliable for diagnosing and predicting the course of Alzheimer’s disease, the impacts could be huge. Unlike PET scans, which are expensive and require injecting a radioactive tracer into the patient, and CSF tests, which involve inserting a needle into the spinal column, a blood test could be used far more widely. “By having a blood test, the numbers of people we can screen for Alzheimer’s and enroll in clinical trials grows by orders of magnitude,” says Bateman. It’s also much faster and cheaper than the alternatives, he says. “At the end of the day, what that really means is we are accelerating the time to find more effective treatments.”
A blood test could also help remove some of the uncertainty around the clinical diagnosis of Alzheimer’s disease in routine practice, Bateman says. Memory problems have many potential causes, so doctors can’t always confidently diagnose patients. Not knowing can be stressful for patients and their families, Bateman says. “It helps a lot when people know what they’re dealing with and what they can expect, and we come up with a plan for what to do about it.”
Some researchers see a day when doctors will use blood tests to gauge a patient’s risk of Alzheimer’s disease much as they now use blood tests for cholesterol to gauge the risk of heart disease — by plugging the test results into a formula that also considers their age, family history and other risk factors. “I can imagine that within five years from now, we will have that for Alzheimer’s disease,” says Sperling. “But I don’t think we have it yet.”
For the time being, Sperling sees no reason for people to seek out biomarker tests for Alzheimer’s disease unless their doctor has advised it. “Right now, I would tell people to do exactly the same things … whether they have a genetic risk factor or a clear biomarker or not,” says Sperling. That includes the usual advice to be physically active, stay mentally and socially engaged, and eat a heart-healthy diet. Biomarkers will make a real difference in helping patients, she says, only once effective therapies finally become available. “When we have a drug that alters amyloid or tau, everything will change.”
Editor's note: This story was updated on August 21, 2020, to clarify that only some biomarkers are now widely used in clinical research, not all.