For consummate physician-researchers, what better way to mark an auspicious anniversary than with a scientific meeting? So it was that some 250 researchers gathered at Massachusetts General Hospital on September 19 to celebrate the 30th anniversary of the founding of the Massachusetts Alzheimer’s Disease Research Center in the first batch of what was to become a nationwide system of federally funded ADRCs. The crowd comprised the instigators of the centers system, former trainees-turned-research-leaders, as well as current faculty, residents, and students from the United States and abroad. After a quick look back in history, the day turned to where the field is today, with new and unpublished data on clinical and basic science.

The Massachusetts ADRC was one of five centers funded in 1984, after Congress mandated the National Institute on Aging to establish a nationwide Alzheimer’s centers program. (The others were in Baltimore under Don Price, in New York City with Ken Davis, in Los Angeles under Tuck Finch, and in San Diego under Bob Katzman.) The mandate followed Katzman’s seminal 1976 editorial, in which he equated the rare early onset disease then called Alzheimer’s to the common dementia of old age, and correctly predicted a tsunami of cases (Katzman, 1976). Leading the Physiology of Aging branch at the NIA at the time, Zaven Khatchaturian had pushed for the centers system to be established and, in December 1983 issued the first request for applications, on a crisp four pages.

John Growdon—already sporting his signature bow tie—in 1984 at the announcement of the Massachusetts Alzheimer’s Disease Research Center. Seated left to right are Zaven Khatchaturian, Secretary of Health and Human Services Margaret Heckler, and Dean Daniel Charles Tosteson of Harvard Medical School.

“Today, Alzheimer’s research is front and center, but in the ’70s at neurology meetings, there was no discussion of Alzheimer’s. Occasional papers by Katzman or Bob Terry attracted scant attention,” said John Growdon, who was the center’s founding director and led it until 2006.

Joseph Martin, who in 1984 headed MGH’s neurology department, recalled that a center for Huntington’s disease, established in 1980 following Nancy Wexler and Woody Guthrie’s advocacy, prompted him to pounce on the RFA for an ADRC. Growdon assembled a regional consortium with labs from across Harvard, MIT, and UMass Medical Center in Worcester. Its initial investigators were Flint Beal, Jim Gusella, Dennis Landis, Charles Marotta, Marsel Mesulam, Ralph Nixon, Joseph Rogers, Dennis Selkoe, Peter St. George-Hyslop, and Rudy Tanzi, among others. In 1989, Martin recruited Brad Hyman, who became director in 2006. 

Ties that bind.

At a symposium marking the founding of the Massachusetts ADRC, Brad Hyman flaunts a bow tie in honor of John Growdon, whose leadership style was as gentle as his looks were sharp.

The ADRCs started with a $3.5 million grant, but since 1984 have drawn well upwards of $1 billion, said Tony Phelps of the NIA, who oversees the program. At present there are 27 centers, down from a maximum of 30 some years ago. Some centers fail their five-year renewal process, others join, but the Massachusetts ADRC has been operating continually since 1984. Its future is bright, said Phelps. It trained seven winners of the Potamkin award, its junior faculty currently hold 40 Career Development (K) awards, and several of its senior faculty help chart the future of research as part of the NIA council. 

What has happened in these 30 years? Martin captured the state of neurology in 1984 with a twist on a management joke. Two men flying in a hot-air balloon get fogged in and drift aimlessly for hours. When the air clears they see a man in the street. They shout down asking where they are. “You are in a balloon!” comes the answer. “Just our luck. We get a neurologist,” says one. “How do you know he’s a neurologist?” says the other. The response: “The information is accurate but completely useless.”

“In 2014, that is not true for Alzheimer’s anymore,” Martin told the audience. “It has taken us longer than we thought, but I believe we finally are on the cusp of figuring out what it all means.”

In the mid-1980s, there were no Alzheimer’s genes or animal models. The neurochemical hypothesis of acetylcholine deficiency created optimism that replacing acetylcholine would improve Alzheimer’s as dramatically as L-Dopa treated Parkinson’s, but that did not happen. “This turned out to be scientifically solid but insufficient to the task,” recalled Growdon. In the ’90s, the discovery of APP and presenilin mutations and then ApoE opened up the molecular biology underlying Alzheimer’s for investigation, and subsequent cell-based screening systems and animal models have led to a second wave of therapeutics. 

For example, in the 1980s David Drachman of UMass Medical Center, who was associate director of the ADRC until 2004, oversaw the characterization of an extended family that led to the identification of the first autosomal-dominant presenilin 1 mutation by St. George-Hyslop, Tanzi, and others. This story is told in the book Hannah’s Heirs by Daniel Pollen, 1993, Oxford University Press. Thus far, drugs based on presenilin have failed in the clinic, but at the symposium, St. George-Hyslop showed how this historic line of research continues in his lab today in search of better therapeutics. A current focus is on using high-resolution electron microscopy for a detailed structural understanding of how substrates bind to the γ-secretase enzyme of which presenilin is part, and how allosteric inhibitors work. This work is making progress, though scientists remain unable to generate atomic-level models of this large protein complex.

In the meantime, neuroimaging and fluid biomarkers evolved. They now enable scientists to identify patients and track them in natural history studies with means beyond the clinical descriptions of yore. With those markers, a new wave of clinical trials is attempting to stem Alzheimer’s disease at pre-dementia stages.

At MGH, Cliff Jack of the Mayo Clinic in Rochester, Minnesota, noted that a new three-stage classification system for preclinical Alzheimer’s disease that was proposed in 2011 is currently being validated in longitudinal cohorts internationally. In those cohorts, a quarter of the people tend to fall out; they have forms of neurodegeneration without amyloid, i.e., conditions other than Alzheimer’s (Jack et al., 2012). The proposed staging appears to hold up for the other three-quarters, in that people who have biomarker evidence of both brain amyloid and neurodegeneration as evidenced by CSF tau or structural MRI—but perhaps not people who have only brain amyloid—decline cognitively in the three-year time frame that puts them within reach of secondary prevention trials (e.g., Sep 2014 news story; Vos et al., 2013). 

A range of studies support the notion that some biomarker evidence of neuronal degeneration is important to choose the right patients for intervention trials that use cognition as a measure of success. For example, evidence of neurodegeneration appears to be absent from the brains of “mismatches.” That is what Teresa Gomez-Isla, who co-directs the Massachusetts ADRC’s clinical core, calls people who lived to old age and died cognitively intact despite having enough plaque and tangle pathology to meet an Alzheimer’s diagnosis. Many autopsy studies have described such people. “Would they all have developed AD symptoms, or do some have features in their brain that render them less vulnerable?” Gomez-Isla asked. At the symposium, she presented ongoing studies to find out what is different about their brains. Her group noticed that the neuronal anatomy of mismatches looked intact compared to that of people with similar pathology and dementia. Mismatches had more neurons, their cortex was thicker and more amply stocked with synaptic markers. Mismatches had fewer tangles outside neurons and more inside neurons than people who had had dementia, suggesting that their neurons survived with tangles. Their plaques were as numerous but smaller than those of people with dementia, and their neurites were plump and axons straight, not dystrophic and curvy as seen in dementia. Finally, their glial cells were not activated, as they are in dementia. 

This suggests that some people’s brains tolerate abundant plaques and tangles without developing neurodegeneration and dementia, Gomez-Isla said. Why this difference? One hypothesis holds that the concentration of oligomeric assemblies of Aβ or even tau might explain it; alas, Gomez-Isla’s attempts to measure these species in the available brain samples have thus far proven inconclusive. “Trying to measure Aβ oligomers in tissue is challenging,” Gomez-Isla told the audience.

Of course, the opposite is also true. Several large Phase 3 trials, the Alzheimer’s Disease Neuroimaging Initiative ADNI, and a recent analysis of the ADRC’s collective pathology dataset all showed that a significant fraction of people who present clinical symptoms of AD lack its defining pathology; they probably have a different disease (e.g., Serrano-Pozo, 2014). Mismatches and people whose symptoms masquerade as AD may be common enough to obscure a drug effect in a trial; hence researchers broadly agree that selecting presymptomatic people accurately with more sensitive cognitive tests and biomarkers—and measuring their progression with new tools—is critical for the next wave of therapeutic trials. 

At the symposium, Reisa Sperling of Brigham and Women’s Hospital and MGH summarized current work to prepare earlier-stage tools for use in secondary prevention studies that are ramping up across the country and abroad. For one, these trials eschew the ADAS-cog in favor of more sensitive tests that are being validated. The A4 study, for example, uses a preclinical composite (Jun 2014 news story), as well as a cognitive composite delivered on an iPad. A new cognitive function index developed by Rebecca Amariglio of the Massachusetts ADRC is in press; it lets patients themselves report standardized outcomes by way of a questionnaire and correlates their self-assessment with cognitive testing.

In addition, both the A4 and DIAN trials are adding tau PET to their docket. At present, this new form of brain imaging is generating the most excitement in clinical AD research. Keith Johnson, who co-directs the neuroimaging core of the Massachusetts ADRC, recalled how researchers used to chuckle at the idea that there would ever be a ligand that entered neurons and labels tau with sufficient concentration. They appear to have gotten exactly that in T807. “As a PET tracer, it is terrific. It has low background binding and good kinetics,” said Johnson. Research at MGH is ramping up, with follow-up scans in longitudinal studies beginning this month and a tau PET study in young adults about to start even as researchers are still trying to define exactly which molecular forms of tau T807 bind in the brain. They know that T807 binds tau pathology in AD and several forms of frontotemporal dementia. It does so in brain areas that correspond to the patient’s functional deficits and disease stage; e.g., in AD it increases with decreasing MMSE.

Ongoing studies hint that tau PET may recapitulate the progressive anatomical spread of tau pathology that Heiko Braak described in postmortem studies. What’s more, by comparing how amyloid and tau PET relate in people at different stages of AD, Johnson is asking whether accelerating tau binding in the medial temporal lobe could represent a tipping point at which a person who has tolerated brain amyloid for some time becomes symptomatic. This is the point in the long run-up to Alzheimer’s dementia that secondary prevention trials are trying to target. A prime suspect appears to be a time, around Braak stages 3 or 4, when T807 binding fans out from the hippocampus and engulfs the neocortex. “Can we find a measure in the brain that can tell us whether this agent can indicate impending impairment? ” Johnson asked. He hopes that tau PET will enable scientists to test an old hypothesis. It holds that some tau pathology in the medial temporal lobe can simmer in the brain for many years as part of aging, but intensifies and branches out once the brain has accumulated high levels of amyloid pathology, as well.

Tau PET is being added to A4 and some other trials as a baseline screen. Once longitudinal data are in hand, researchers can calculate a slope for how quickly tau binding rises over time, and use it as a basis for deploying tau PET as an outcome measure. Sperling is eyeing tau PET binding in the inferior temporal lobe as a marker of longitudinal memory decline, but much more data is needed.

While tau PET is grabbing headlines, a different way of imaging the brain in preclinical Alzheimer’s is quietly coming along in its shadow. Resting-state functional connectivity MRI has begun to flourish in the past decade. It is also starting to be used in A4 and other trials in an exploratory way. This technique uses the BOLD signal to probe the integrity of anatomically distributed networks that are tied to certain behaviors. Compared to PET, connectivity fMRI is cheaper, requires no radiation, and can be added to safety scans and repeated. It started to draw serious attention in AD research when PiB PET patterns showed that amyloid deposition occurred in cortical networks that were “incredibly complex and distributed,” said Randy Buckner, who came to the Massachusetts ADRC 10 years ago. So instead of parsing specific circuits right away, Buckner first tried to understand the general characteristics of the affected regions. By studying young people, he discovered that the regions most strongly affected by amyloid deposition in a person’s 60s and 70s are also the “hubs” that connect many distributed networks in the brain and are marked by the highest activity levels in adult life.

At the MGH symposium, Buckner offered an evolutionary perspective. Those hubs in the frontal, temporal, and parietal association cortices that receive projections from the hippocampus and have the most amyloid are the parts of the brain that expanded massively relative to apes starting some 2.5 million years ago. Back then, the size of the human brain began shooting up exponentially as the genus Homo arose and started building tools, languages, social groups, and then civilizations. The brain of modern humans is three times the size of the brain of modern great apes, Buckner said. If one were to lay out the cortical mantle of a human, it would be the size of a large pizza, 1,000 times larger than a mouse’s. The expanded regions are those of densest connectivity. They develop later than sensory or motor cortex in embryonic development, and their metabolic activity is highest throughout adulthood (Buckner and Krienen, 2013).

Alzheimer’s researchers are now dissecting these networks in search of a preclinical biomarker. At the MGH symposium, Jasmeer Chhatwal of Brigham and Women’s Hospital in Boston noted that earlier work had reported the default mode network (DMN) as being a place of amyloid deposition and functional degradation in presymptomatic Alzheimer’s disease (Chhatwal et al., 2013). That is still true but since then, it has become clear that there are overlapping changes in the DMN in both aging and AD, and researchers have gone out in search of other networks that would allow them to more cleanly distinguish between normal aging and preclinical AD.

For that, Chhatwal is using three comparisons. In the DIAN cohort, he compares asymptomatic and symptomatic mutation carriers. In cognitively normal people, he compares people in their 20s to old people who are free of brain amyloid. Finally, he compares the young adults to cognitively normal old people who do have amyloid. These successive comparisons are beginning to crystallize a signature of presymptomatic AD. Full data are in press, but the upshot is that cognitive networks such as the dorsal attention network are more degraded than motor or sensory networks in both autosomal-dominant and sporadic presymptomatic AD than in healthy aging. When integrated into a composite score, changes across this broader range of networks could yield an fMRI measure for secondary prevention trials, Chhatwal said. This jibes with a similar comparison between the DIAN cohort and a longitudinal cohort at Washington University, St. Louis (Aug 2014 news story). 

In parallel to new tools being fashioned for trials, a new wave of experimental therapeutics is wending its way through the clinical pipeline. In 2014 all eyes are on BACE inhibitors, from Merck’s leading Phase 3 compound on down to others that are close behind. As more of those drugs enter larger trials, a chorus of researchers can be heard suggesting that they be evaluated closely for potential side effects (Dec 2013 conference series). 

At the MGH symposium, Christian Haass of the Munich site of German Center for Neurodegenerative Diseases, or DZNE—Germany’s correlate of the U.S. ADRCs—spoke about published and unpublished biological effects of BACE inhibition. Moving beyond the 34 brain-specific substrates for BACE1, and the enzyme’s known role in forming adult muscle spindles, Haass presented new data indicating that pharmacological BACE inhibition increases the generation of an additional APP fragment that Haass called Aeta. Barely studied to this day, this fragment blocks LTP, Haass said. He noted that he has seen this fragment since he got his start in AD research in Dennis Selkoe’s lab at the Massachusetts ADRC in the early 1990s, and showed bands of it in a paper from those days (Haass et al., 1992). His group only recently began to characterize this band in depth, and in doing so realized that it represented the product of a heretofore-unappreciated APP cleavage pathway. “This is a physiological processing pathway, and it is abundant,” Haass said.

At the symposium, Selkoe, who is studying a similar synaptotoxic fragment he calls “N-terminally extended,” agreed that it points to a new APP processing pathway that deserves attention (Welzel et al., 2014). 

Pharma researchers have expressed confidence in the extensive safety data of the compounds they have moved into the clinic. On September 16, AstraZeneca and Lilly announced a $500M collaboration to jointly develop AstraZeneca’s BACE inhibitor AZD3293. At the MGH symposium, Sperling said that she is going ahead with planning a BACE inhibitor trial on the A4 platform, but that safety monitoring will take into account new research findings emerging on BACE1.

These groups are working toward a vision of personalized medicine, where a panel of molecular diagnostics would tell an aging person which types of misfolded protein are present in their brain—Aβ, tau, α-synuclein, TDP-43, SOD1—and doctors would then treat those pathologies in a secondary prevention mode to stave off symptoms. Neurologists of the future, listen up: The day ended with the announcement of an endowed Growdon Fellowship to start in July 2015. “This was named in honor of the teaching and gentleness John has shown us over the years,” said Hyman into a standing ovation for Growdon, who remains active in research and clinical care.—Gabrielle Strobel


Comments on this content

No Available Comments

Make a Comment

To make a comment you must login or register.


News Citations

  1. Together, Aβ and Neurodegeneration Spell Cognitive Decline in Three Years
  2. Test Battery Picks Up Cognitive Decline in Normal Populations
  3. Cloistered Retreat Takes the Pulse of BACE Research

Therapeutics Citations

  1. Verubecestat
  2. AZD3293

Paper Citations

  1. . Editorial: The prevalence and malignancy of Alzheimer disease. A major killer. Arch Neurol. 1976 Apr;33(4):217-8. PubMed.
  2. . An operational approach to National Institute on Aging-Alzheimer's Association criteria for preclinical Alzheimer disease. Ann Neurol. 2012 Jun;71(6):765-75. PubMed.
  3. . Preclinical Alzheimer's disease and its outcome: a longitudinal cohort study. Lancet Neurol. 2013 Oct;12(10):957-65. PubMed.
  4. . Mild to moderate Alzheimer dementia with insufficient neuropathological changes. Ann Neurol. 2014 Apr;75(4):597-601. Epub 2014 Apr 8 PubMed.
  5. . The evolution of distributed association networks in the human brain. Trends Cogn Sci. 2013 Dec;17(12):648-65. Epub 2013 Nov 7 PubMed.
  6. . Impaired default network functional connectivity in autosomal dominant Alzheimer disease. Neurology. 2013 Aug 20;81(8):736-744. PubMed.
  7. . Amyloid beta-peptide is produced by cultured cells during normal metabolism. Nature. 1992 Sep 24;359(6393):322-5. PubMed.
  8. . Secreted amyloid β-proteins in a cell culture model include N-terminally extended peptides that impair synaptic plasticity. Biochemistry. 2014 Jun 24;53(24):3908-21. PubMed.

External Citations

  1. Hannah’s Heirs
  2. A4 study

Further Reading

No Available Further Reading