By Paul Nicolaus
November 15, 2018 | A year ago, as Microsoft co-founder Bill Gates announced his decision to invest in Alzheimer’s research for the first time, he pointed out that people are living longer than they used to thanks in large part to scientific advancements. Across the globe, fewer people are dying young from heart disease, cancer, and infectious diseases. This ought to be good news. “But what happens,” he asked, “when it’s not?”
As we live longer, our odds of developing a chronic condition that lessens our quality of life grows. Among the array of late-life illness possibilities, Alzheimer’s stands out as a particularly significant threat. After age 65, the risk of developing the disease doubles every five years, and in the U.S., it is the only top 10 cause of death that lacks meaningful treatments and becomes more prevalent each year. Beyond that, it is one of the most expensive conditions, requiring round-the-clock care for those in the grips of its late stages.
“I first became interested in Alzheimer’s because of its costs—both emotional and economic—to families and healthcare systems,” Gates wrote. The interest is also personal considering the men in his family, including his father, have suffered from the illness. “I know how awful it is to watch people you love struggle as the disease robs them of their mental capacity, and there is nothing you can do about it.”
Today’s diagnostic methods, which evaluate cognition using traditional pen and paper tests, are less than ideal. The results can be combined with physical health, medical history, and information from family members or caregivers. If a patient does not perform well on a cognitive test, there are other potential causes of memory loss that still need to be ruled out.
The most certain form of diagnosis still comes from an autopsy following death, but there are other reliable options. An amyloid PET scan can help physicians determine if someone has high levels of this hallmark protein of the disease in their brain. A spinal tap can help as well, but the former is costly and the latter is invasive, which means many do not wind up taking advantage of these tests.
There are efforts underway to make amyloid PET imaging more affordable for patients. The Alzheimer's Association has been leading a four-year, $100 million Imaging Dementia-Evidence for Amyloid Scanning (IDEAS) study that will provide access to imaging for over 18,000 individuals. The hope is that the study will yield enough data to assess whether amyloid imaging has a positive impact on patient outcomes and, in turn, determine whether it should be reimbursed by Medicare and other third-party payers.
While not yet complete, the study’s findings have already revealed that a significant number of those with dementia or mild cognitive impairment who are being treated for Alzheimer’s may not have the disease. “What that says to me is that paper and pencil tests—the traditional way of doing an assessment for a patient—isn’t really adequate to give an accurate diagnosis,” James Hendrix, director of global science initiatives at the Alzheimer's Association told Diagnostics World News. “And more biologically-based tools are needed to help physicians give a more accurate diagnosis.”
Regulatory and Scientific Developments
One of the biggest drawbacks of the current status quo is that patients aren’t being tested until they begin showing signs of cognitive decline. “Our brains are changing long before the onset of symptoms,” Hendrix said, noting that amyloid begins to appear in the brain 10 to 20 years before warning signs set in.
This delayed diagnostic approach is troubling for a number of reasons. Similar to cancer, the hypothesis is that prevention is easier than repair. Catching it early offers more promise than attempting to deal with it after the disease has had the opportunity to progress.
Delayed diagnosis also hinders the search for treatments. Pharmaceutical companies have poured billions into efforts to come up with a solution that could curb or cure the effects of Alzheimer's, but many believe drug trials have failed at least in part because treatments were tested on participants whose brains were already too far gone to benefit.
In other words, drugs need to be tested earlier on, and earlier diagnosis could help. “We need tools to find people who are early on in the disease,” Hendrix said. What’s needed are screening tools that are inexpensive, widely available, and effective. Once that becomes a reality, “we could funnel a whole lot more people into these clinical trials,” he added, “and that could actually lead us to better treatments.”
Recent developments are opening doors to new possibilities. To begin with, the very definition of Alzheimer’s disease is undergoing change. The U.S. Food and Drug Administration (FDA) issued guidance for industry that indicated the agency would consider drug trials based on biological markers, which could allow therapies to be tested much earlier in the disease process.
In terms of scientific progress, Hendrix pointed to one paper that he said has “electrified the field” this year. A study published in Nature (doi.org/10.1038/nature25456) that came out of a collaboration between Japanese and Australian researchers showed the promise of using a blood test to measure amyloid status with up to 90% accuracy.
Blood tests could be the tip of the iceberg. Some are pursuing the development of digital pens that can pick up subtle changes in the way a person writes while others are looking to harness the potential of everything from retinal imaging to speech patterns. “People are basically throwing the kitchen sink at this problem,” he said, and there are plenty of interesting advances on the horizon.
In anticipation of more tools that detect and measure the biology associated with Alzheimer’s, an Alzheimer’s Association-led Workgroup that includes Hendrix has published appropriate use criteria for lumbar puncture and spinal fluid analysis (doi: 10.1016/j.jalz.2018.07.220).
Could AI Play a Greater Role?
Artificial intelligence (AI) is yet another technology to enter into the mix as a team of scientists has trained an algorithm to predict cognitive decline leading to Alzheimer’s disease. Led by Mallar Chakravarty, a computational neuroscientist at the Douglas Mental Health University Institute and an assistant professor of psychiatry at McGill University, the research group designed an algorithm that can detect patterns across MRI images, genetics, and clinical data up to five years before major symptoms appear.
Their study, published this fall in PLOS Computational Biology (doi.org/10.1371/journal.pcbi.1006376), used data from healthy seniors, those experiencing mild cognitive impairment, and those with Alzheimer’s. The results, which relied on data from the Alzheimer's Disease NeuroImaging Initiative, were replicated using a sample from the Australian Imaging, Biomarker & Lifestyle Flagship Study of Ageing.
“Depending on the methodologies we use, we see anywhere between 90 and 99% accuracy, which is somewhat unheard of in this field,” Chakravarty told Diagnostics World News. “One reason I think we get this accuracy is because a lot of this work doesn’t rely on the diagnostic labels that a clinician gives an individual.” From his perspective, this attempt to capture cognitive decline as an objective measure rather than a diagnostic category is one of the big advances he and his team are conceptually putting out there with this study.
One drawback, however, is that the findings are of limited value in a clinical setting, according to John Walker, a computer scientist and co-founder of uMETHOD Health—a company that uses advanced AI to fight Alzheimer's disease using a precision medicine approach. “First, the authors’ focus is on the quality with which someone’s cognitive decline can be predicted, given a limited set of inputs,” he explained, and they don’t discuss the range of multi-domain interventions that could impact disease progression.
Second, the input data is limited in a number of ways. The study mentions looking at the APOE gene, a key influencer in the development of Alzheimer’s disease, but this is just one of hundreds of additional genes that ought to be considered. In terms of demographics, the study takes age and sex into consideration, but Walker says ethnicity, menopause status, and medical history are all critical for the prediction of disease onset and progression.
Finally, while MRI scans are a useful data point, Walker argues that they are not the most important. “A well-known conundrum of Alzheimer’s disease is that not everyone with brain pathology, as seen on a scan, ends up with symptoms,” he explained. “This is because of the complexity of cognitive decline.”
One example of this complexity is the cognitive reserve theory, which attempts to understand why the more education someone has had throughout their life, the less likely they are to experience cognitive decline from this disease. “Learning builds neurons and re-wires the brain,” Walker added, “and is one of many protective things you can do throughout life.” Educational attainment is, in turn, a factor to include when using predictive analytics.
For Chakravarty’s team, the next step is to find larger datasets to train and validate on. Moving forward, he is also interested in determining whether it is possible to make accurate predictions in real-time as opposed to using data that’s previously been collected.
They have recruited and acquired data on a cohort of about 280 individuals and are interested in performing a second assessment, he explained, in order to find out whether the algorithm predicts what their cognitive decline state would be like in the three years that have passed since the data was first collected. “So that’s a really exciting project that’s coming up—trying to marry traditional data collections with these AI-driven recruitment strategies,” he added.
New Funding Backs Bold Ideas
We’ve entered a season of new funding for Alzheimer’s research. The Rainwater Charitable Foundation, one of the largest independent funders of neurodegenerative disease (ND) research, announced last week that it is launching the Rainwater Prize Program to encourage and reward scientific progress toward new treatments for neurodegenerative diseases related to the accumulation of tau protein in the brain, which includes Alzheimer’s. The foundation is offering up to $10 million for ground-breaking discoveries.
And this summer, Gates came to the table once again, this time as he and Alzheimer's Drug Discovery Foundation (ADDF) co-founder Leonard Lauder announced a new initiative called the Diagnostics Accelerator. With the support of other donors such as the Dolby family and the Charles and Helen Schwab Foundation, this project will look to spark development of novel biomarkers for earlier and better detection of Alzheimer's disease and related dementias.
Collectively, the coalition of philanthropists has committed over $30 million. Over the next three years, these dollars will be doled out to those who are working on the most promising ideas to diagnose the illness before the most devastating symptoms set in. The endeavor is expected to take more risks on bold ideas than traditional venture capital while placing a greater focus on developing real products for real people than basic research commonly backed by governments or charitable organizations.
Funding is open to scientists and clinicians in the U.S. and worldwide working in academic medical centers, universities, nonprofits, or biotechnology companies.