Scientists gathered on March 1 in Bethesda, Maryland, for the 2018 NIH Alzheimer’s Disease Research Summit, ready to absorb the 82 presentations scheduled over two days. Then a nor’easter forced closure of U.S. government buildings and put the kibosh on Day 2. Even so, researchers seemed impressed by the summit. Scientists interacted across disciplines with colleagues whom they might not otherwise meet, and the NIH took away 75 pages of funding recommendations for future research. “There were a lot of interesting ideas, from basic science to clinical studies,” said Marco Colonna, Washington University, St. Louis. “It was a bit overwhelming, but extremely valuable.”
This was the third AD research summit hosted by the NIH, as mandated by the National Alzheimer’s Project Act. Recommendations from the first two summits in 2012 and 2015 helped shape the NIA-funded research agenda. With the NIH budget for AD and related disorders almost tripling to $1.414 billion in the last three years, the third summit has the potential to make a major impact on the field. As NIH Director Francis Collins emphasized in his opening address, this is a critical time in Alzheimer’s research. “Any notion that a quick path to prevention or treatment was going to emerge from 2012 … has turned out to be naïve,” Collins said. But he noted that the field knows more than in 2012 and urged leaders to bring boldness, audacious optimism, and a readiness to consider dramatic new approaches to the problem.
The first day of the summit outlined some of those approaches. Organized by Suzana Petanceska and Laurie Ryan of the National Institute on Aging, the summit was to cover seven themes:
- Novel Mechanistic Insights into the Complex Biology and Heterogeneity of AD
- Enabling Precision Medicine for AD
- Translational Tools and Infrastructure for Predictive Drug Development
- Emerging Therapeutics
- Understanding the Impact of the Environment to Advance Disease Prevention
- Advances in Disease Monitoring, Assessment, and Care
- Building an Open Science Research Ecosystem to Accelerate AD Therapy Development.
Theme sessions featured three to four 15-minute talks followed by five-minute, one-slide summaries of related topics by up to nine selected panelists, leaving little time for Q&A or debate. The cancelled second day will be rescheduled soon, NIH staff said.
Big data and precision medicine dominated the day, building on plenary talks by Eric Schadt from the Icahn School of Medicine at Mount Sinai, Los Angeles, and Joni Rutter of the NIH. Rutter gave an update on the All of Us research program, formerly called the Precision Medicine Initiative cohort. It plans to gather data on 1 million Americans as a foundation to enable precision medicine, or medical treatment and care tailored to individual patients. So far, 16,500 participants have enrolled. They have provided medical histories and agreed to share medical records and give blood, urine, and saliva for future analysis. The program hopes to eventually incorporate behavior and lifestyle data, including information from wearable devices.
Schadt thinks the possibilities from mining such data are staggering. He emphasized that a broad array of upcoming technologies will generate vast amounts of information. Data from next-generation sequencing, electronic medical records, wearables, geocoding that catalogs a person’s “exposome,” i.e., exposure to environmental hazards such as air pollution or even getting stuck in traffic, and other sources will add up to yottabytes, Schadt said. For the mathematically minded, that is 1024 bytes, or 10 trillion gigabytes, enough to store 2,000 trillion songs on your smartphone. The challenge, said Schadt, is analyzing the data.
Enter machine learning. Algorithms can be trained to identify, in unbiased fashion, characteristics that distinguish healthy controls from people with disease. Schadt showed how such in silico crunching of data from the AMP-AD knowledge portal uncovered VEGF as a master regulator gene downregulated in AD. Overexpressing this gene in 5xFAD mice dramatically reduced Aβ pathology, said Schadt.
Joel Dudley, fromMount Sinai in New York, told the audience that, to his great surprise, machine learning picked up a greater abundance of viral RNA and DNA in various brain regions of AD patients. Dudley was mining open data from the AMP-AD portal for network states that distinguish normal controls from early and late AD, with a view to drug targeting. “As we looked at network transitions, we kept getting screaming signals for viral biology,” he said. Specifically, he found that herpes viruses 6 and 7, as well as herpes simplex viruses, are amplified in Alzheimer’s. The findings recapitulated across multiple datasets including those from Mount Sinai, Rush’s Religious Orders Study and Memory and Aging Project, and from the Mayo Clinic sample. Taking it a step further, Dudley found that the abundance of viruses correlated with clinical and neuropathological traits of AD, such as the CDR, plaque load, and Braak staging.
Looking for quantitative trait loci that change with viral load (vQTLs), HHV6/7 and HSV vQTLs correlated with AD traits. With these vQTLs, Dudley built viral-host molecular networks in an effort to take the data beyond correlations and look for causality. This turned up vQTLs that elicited changes in the host, and vice versa. These networks were enriched in genetic loci that had turned up in previous AD genome-wide association studies.
Dudley is not sure what this data means. He told Alzforum that others at the summit have similar findings. He does not think viruses cause AD. Rather, because they integrate into the host genome, they may become activated when neurons come under stress, as during disease. Then they may start to co-opt the cell’s transcriptional machinery for viral reproduction, tipping the whole system toward a more pathological state. “This type of work highlights the benefit and complexity of a data-driven approach,” said Dudley, who believes big data has the power to shake up the field. “Studying AD necessitates a systems view of how information flows from the immune system to neurons to metabolism and back. By understanding these relationships, we hope we can put together a much larger picture of the disease that will reveal non-traditional [drug] targets,” he said.
Other massive datasets are coming online that will help in that analysis. Matthias Arnold from the Helmholtz Center Munich, Nick Seyfried from Emory University, Atlanta, Rachel Whitmer at the University of California, San Francisco, and Cornelia van Duijn from Erasmus University Medical Center, Rotterdam, the Netherlands, showed, respectively, how metabolomics, proteomics, epidemiology, and genetics can be woven into a big data fabric.
Other speakers presented tools and techniques that expand our understanding of the brain. Sean Bendall from Stanford University co-leads MIRIAD, a.k.a. multiplexed imaging of resilience in Alzheimer’s disease. Bendall and colleagues use a technique called multiplexed ion beam imaging (MIBI) to simultaneously measure expression of up to 40 molecules at a time in a tissue section. MIBI uses antibodies labelled not with fluorescent dyes but with specific isotopes that are liberated by the ion beam and measured by mass spectrometry. The technique can resolve down to the level of individual synapses. Bendall uses MIBI to search for expression patterns that emerge more in people who are resilient or prone to AD, again using machine learning to tease different patterns out from vast amounts of data.
Other panelists emphasized how epidemiology, genetics, and comorbidities affect resilience. Ben Readhead from Arizona State University, Phoenix, described a single-cell RNAseq atlas of brain transcription. It could catalog the degree of transcriptomic heterogeneity in the brain, uncover new subtypes of cells, and identify shifts in gene expression that occur in the face of emerging pathology, Readhead said.
Other speakers addressed the role in AD pathobiology of proteostasis, nitrosylation, microglia and the immune system, stress granules, autophagy, and the chaperome—that is, the suite of molecules that help proteins fold correctly. Greg Carter from the Jackson Laboratory, Bar Harbor, Maine, reviewed progress in the MODEL-AD project, which has released ApoE and TREM2 mouse models, including R47H and Y38C, and plans to issue eight to 10 new lines per year, including one that will have the complete tau locus humanized.
Ryan gave an overview of the new ACTC (Dec 2017 news), while Cynthia Musante from Pfizer in Cambridge, Massachusetts, championed the power of quantitative systems pharmacology as a means to better predict how modulation of drug targets translates into clinical endpoints. In the emerging therapeutics session, Frank Longo from Stanford, Roberta Brinton from the University of Arizona, and Mark Gurney from Tetra Discovery Partners in Grand Rapids, Michigan, shared data from early stage trials of p75 receptor antagonists to stem neurodegeneration, allopregnanolone to promote regeneration, and a phosphodiesterase 4D allosteric inhibitor to improve memory, respectively (Dec 2017 conference news; May 2017 conference news).
Research Models Citations
- Clinical Trials Consortium Succeeds ADCS, Focuses on Prevention
- In the Running: Trial Results from CTAD Conference
- Non-Amyloid Treatments: Inflammation, Epigenetics, Regeneration