Part 2 of a five-part report.

With therapy trials of the dominantly inherited Alzheimer’s network gearing up (DIAN-TU, see Part 1 of this story), what is the status of the network’s observational study (DIAN-Obs)? Its size fluctuates somewhat. From a high point of 465 participants in 2013, 89 transferred into the treatment trial but 111 have discontinued. Some people died, others had to leave because their disease had advanced too far for them to travel, take tests, and put up with the needles and scanner. In fact, some participants declined so rapidly that, on the current DIAN-Obs protocol of visits every other year in the symptomatic phase, they were able to return only twice after their onset, leaving the DIAN progression model with a dearth of data for rates of change in this late phase of the disease. To fill this gap, DIAN will from now on invite symptomatic patients back for annual visits. Yet other participants, however, missed visits because scheduling was tricky, or the tests were burdensome, or for unknown reasons. At an internal DIAN meeting, John Morris of Washington University, St. Louis, asked all site leaders to redouble their efforts to engage and support participants in their concerns, so that they will choose to stick with the study and contribute longitudinal data.

DIAN is replenishing its ranks as additional family members come forward. Moreover, sites in Germany, Argentina, and Japan have begun enrolling, with 55 new participants between them thus far. A site in Seoul, South Korea, is about to begin enrollment, and sites in Mexico, Israel, and Taiwan have expressed interest in joining. The University of Antioquia in Medellin, where Francisco Lopera’s team already conducts an Alzheimer Prevention Initiative therapy trial in the world’s largest ADAD cohort of families carrying the E280A “Paisa” presenilin-1 mutation also is considering becoming a DIAN site. After all, Colombia is home to families with other autosomal-dominant AD mutations, as well, Lopera told Alzforum. Morris said that going forward, he aims to offset attrition or departure into trials recruitment in order to maintain a DIAN-Obs cohort of about 300 people.

In the observational study, three-quarters of participants are asymptomatic at entry; of those, three-quarters carry the family mutation. They are in their 20s, 30s, and 40s, with an average 14.7 years of education. Roughly 30 percent are ApoE4 positive. This risk gene exerts a far smaller influence on age of onset in ADAD than in LOAD, possibly because it is a surrogate for amyloid deposition, which is already driven far more strongly by the APP or presenilin mutation, said Randy Bateman of WashU.

DIAN participants handle the study’s intense regimen of tests well, though some sites are more successful than others at obtaining CSF and PET scans, Morris said. Lumbar puncture headaches have annoyed quite a few among these young-adult and middle-aged people. At AAIC, Morris showed results from a recent survey suggesting that once participants had decided to undergo spinal fluid sampling, they were more likely to do it on follow-up visits, regardless of whether they had suffered a headache the previous time. Recent research on lumbar punctures suggests that using particular atraumatic needles and letting the CSF drip rather than pulling it into a syringe reduces the risk of this unpleasant consequence of donating CSF for research. “This data has busted some myths about CSF collection,” Morris said. (Moulder et al., submitted; Duits et al., 2016.)

Throughout AAIC, scientists presented data from the observational cohort study. Jason Hassenstab of WashU described how cognitive performance in the long preclinical period changes over time. Traditionally in AD research, cognitive decline has been measured with less-sensitive tests against group norms. In staging diagrams, cognition has been pegged to start changing only a few years before clinical diagnosis, certainly after CSF tau goes up. In many instances, cognitive decline was conceptualized as mild cognitive impairment and applied to a heterogeneous mix of people who had different underlying diseases affecting their cognition. In contrast, when measuring cognition repeatedly in people who are known with certainty to harbor the pathogenic process of AD, it turns out that changes can be well-characterized early in the disease process, suggesting that cognition changes early, at least in the purer form of AD seen in DIAN participants, Hassenstab showed.

Hassenstab used results of 16 cognitive tests that tap episodic memory, attention control, working memory, processing speed, and semantic memory, plus a DIAN cognitive composite of four tests. So far, 432 participants have taken these tests at least once, 235 at least twice, 76 at least three times, and 35 people sat them at least four times. Some of these people became clinically symptomatic while in DIAN. Hassenstab compared rates of cognitive decline between asymptomatic and symptomatic mutation carriers, and non-carriers, and he modeled the rates in carriers relative to their expected year of onset (EYO). Overall, cognitive decline in otherwise asymptomatic carriers was detectable as far back as 15 years prior to EYO, Hassenstab said at AAIC. The first test to flag change was on ability to remember word lists; other tests of episodic memory and attention control revealed a signal nearly as early. Rates of cognitive decline accelerated as people edged closer to their EYO.

Scientists are now drilling down on whether different mutations in PS-1, PS-2, or APP affect progression differently. This research is one way of trying to explain the variance seen within the DIAN data set; however, thus far rates of cognitive decline look similar for the three genes in question, Hassenstab reported. Overall, the longitudinal data mirror the published cross-sectional data, he said.

CSF analysis is generating longitudinal data, as well. At AAIC, Anne Fagan updated her surprising discovery that the known increase in CSF tau prior to EYO does not continue, but rather levels off and may even reverse once people have been symptomatic for some years. This was published based on 75 longitudinal samples from 37 DIAN participants (Fagan et al., 2014). “I hoped and prayed it would hold up in a larger number of people,” Fagan quipped to the DIAN steering committee. It did. As of now, 329 samples from 134 DIAN participants confirm the same finding. CSF Aβ42 goes down over time and stays down after EYO; but both phospho-tau and, two years later, total tau increase as people approach EYO, then both markers plateau and decrease again.

Modeled as rates of change, Fagan said, the data reveal a linear decrease for Aβ42. For phospho-tau, they reveal a kink—a change point in biostatistics lingo—at six years prior to EYO. This finding is puzzling because tau PET indicates increased tracer uptake well after that; tau PET data thus far is largely cross-sectional. Many researchers at AAIC emphasized that hypotheses of how a given marker might change over time that are drawn from cross-sectional comparisons must be tested with in-person, serial measurements. Cross-sectional data compare different people to each other, with the individual variability this brings.

Emerging CSF biomarkers are starting to round out the picture. Nelly Joseph-Mathurin of WashU showed new data on the neuronal calcium sensor protein VILIP-1. “This calcium sensor protein could be a preclinical marker of neuronal injury,” Joseph-Maturin said. VILIP-1 is thought to change similarly to tau, but much less is known about it.

In the DIAN cohort, most VILIP-1 samples analyzed thus far, from 202 participants, are cross-sectional. Levels are highest in preclinical mutation carriers who already have brain amyloid deposition, Joseph-Mathurin showed. Intriguingly, in the first batch of repeat measurements analyzed thus far, from 41 participants, VILIP-1 appears indeed to behave like tau. It rises while a person approaches his or her EYO, but then drops again during the symptomatic phase of disease. It is almost as though the most acute neuronal death occurred before overt clinical illness, followed by a relative slowing, said Fagan.

CSF VILIP-1 appears correlated to CSF total and phospho-tau in LOAD, as well. In the first few participants who received tau PET scans within DIAN, VILIP-1 again was highest in asymptomatic carriers who were positive for both amyloid and tau tracer uptake, but much more data is needed on these emerging studies.

In LOAD, additional markers correlate with VILIP-1, Fagan said. They include neurogranin, YKL-40, and the synaptic protein SNAP-25. “There are many interesting comparisons to be made in the DIAN dataset, but this was underfunded,” Fagan noted. Doing those types of analyses on the DIAN samples could help scientists slot a rapidly growing panel of injury and inflammation markers downstream from Aβ42 into the disease’s evolving staging diagram. This would flesh out the field’s understanding of what happens during the crucial years before and around clinical onset of AD.

For example, at AAIC, Christian Haass, Ludwig-Maximilian University in Munich, used DIAN samples to advance the story of sTREM2. This is the shedded, soluble stub of the microglial transmembrane receptor that has been genetically linked to a range of neurodegenerative diseases. Haass reported in Toronto that CSF sTREM2 levels in AD appear to be most dynamic during the five years before and five years after clinical symptoms begin. sTREM2 creeps up somewhat as healthy people get older, reflecting a mild immune stimulation with age. In LOAD, a much steeper increase is seen in preclinical stage 2 of the NIA-AA preclinical research diagnostic criteria, i.e., in people who have both amyloid and tau pathology. sTREM2 is not particularly elevated in the amyloid-only preclinical stage 1, Haass reported. In this study, sTREM2 is also up in SNAP, i.e., people with an as-yet vaguely defined brain atrophy that is not triggered by amyloid.

Haass’ lab obtained aliquots of DIAN cross-sectional CSF samples, and in Toronto, he showed results. Mutation non-carriers had the same steady, slight elevation with age seen previously in other people. Carriers have a much steeper increase during the decade of minus five to plus five years from their EYO. “That is the window of time for TREM2,” Haass said. CSF sTREM2 correlates with tau, but starts its rise a few years after tau does.

But is sTREM2 truly a marker of microglial activation? The answer might be yes, Haass believes, at least according to an unpublished mouse triple PET study visualizing the new microglial tracer TSPO together with amyloid and tau pathology in presenilin/APP transgenic lines. Based on these data, it would appear that microglial activation, and sTREM2 as an indicator of it, fits into the amyloid cascade after tau. “You must have cellular debris from neurodegeneration to fully activate microglial phagocytosis,” Haass told the AAIC audience on the last morning of the conference.

Like CSF research, brain imaging is also grappling with the confluence of neurodegeneration and inflammation in early AD. In one presentation at AAIC, Juan Domingo Gispert of Barcelonaβeta Brain Research Center in Barcelona, Spain, reported that CSF sTREM2 levels were up at the same time that MRI indicated brain swelling in AD areas as judged by increased gray-matter volume and decreased water diffusivity. Gispert’s new data came from people at the amnestic MCI stage of LOAD, but a similar finding was reported last year at AAIC by Philip Weston and Natalie Ryan at University College London in presymptomatic people with autosomal-dominant AD, as well (see Aug 2015 conference news). For more data on imaging in dominantly inherited AD, see Part 3.—Gabrielle Strobel

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. At Age 8, DIAN Is Churning Out Data and Growing into a Movement
  2. New Data on Autosomal-Dominant Alzheimer’s Point to Early Fissures in the Brain’s Microarchitecture
  3. Brain Imaging in DIAN: Atrophy Rates—Check. Tau PET—Not Yet.

Mutations Citations

  1. PSEN1 E280A (Paisa)

Biomarker Meta Analysis Citations

Paper Citations

  1. . Performance and complications of lumbar puncture in memory clinics: Results of the multicenter lumbar puncture feasibility study. Alzheimers Dement. 2016 Feb;12(2):154-63. Epub 2015 Sep 11 PubMed.
  2. . Longitudinal change in CSF biomarkers in autosomal-dominant Alzheimer's disease. Sci Transl Med. 2014 Mar 5;6(226):226ra30. PubMed.

Further Reading

No Available Further Reading