This is Part 2 of a four-part series. See also Part 1, Part 3, and Part 4.
Read a PDF of the entire series.
11 October 2012. On September 27th, researchers, participants, and other stakeholders in the Dominantly Inherited Alzheimer Network (DIAN) gathered at the network’s hub at Washington University, St. Louis, Missouri. They spent a moment basking in pride that DIAN has surpassed its original aims, and that it is now poised to start therapeutic trials. Then they quickly moved on to exchange new research data DIAN is churning out, and to work on the numerous practical, scientific, and ethical challenges facing this rapidly expanding, global study for families with autosomal-dominant Alzheimer’s disease.
DIAN to date has enrolled 290 members of families with autosomal-dominant AD at 11 sites in Australia, the U.S. and the UK; 215 of the participants are asymptomatic. This number exceeds the original enrollment target of 240 in the DIAN grant from 2008, and is on track to meet an expanded target of 400. “All sites are operational and contribute participants. Assessment completion rates are upwards of 80 percent,” said John Morris of WashU, who is the principal investigator of DIAN. New sites are starting to operate in Germany and the U.S., and yet more are planning to do so. DIAN has begun collecting skin biopsies from participants to generate induced pluripotent stem cells and neuronal cell models. Most importantly to participants, the network’s trial unit has clinched an agreement with Eli Lilly and Company and Roche to run a first preclinical treatment trial of three drugs in 160 participants (see Part 1).
“This is the holy grail of DIAN,” Morris said.
In the face of such growth, the head of DIAN’s external advisory group, Thomas Bird of the University of Washington, Seattle, set himself up for a friendly ribbing when he confessed that at the project’s start in 2008 he doubted whether it would succeed, given its ambitious aims . “I am glad I was wrong,” he told Alzforum while an NBC television crew was milling around among scientists, DIAN families, staff and representatives of funding agencies. Bird co-discovered the Volga German presenilin 2 mutation, and several younger relatives of the family who worked with him for this discovery now participate in DIAN (see Part 1, Part 4).
DIAN is an arduous study for participants. Even so, they embrace it enthusiastically, said Nigel Cairns of WashU, who leads the network’s neuropathology core. They fly or drive to their nearest study site for at least three days of being poked, scanned and—most grueling for many—sitting for hours of cognitive testing. “The test days bring up all the fears you might otherwise manage to bottle up in daily life,” said a participant who requested anonymity. DIAN is also arduous for the sites, requiring much advance planning and coordination, as well as sophisticated data handling and development of new data analysis tools. Even so, recruitment is continuing apace and participants are returning for repeat visits.
Overall, the data DIAN has generated is resoundingly validating AD biomarkers, Morris believes. “DIAN shows that only the mutation carriers have the biomarker changes, and they all are destined to develop AD dementia. Their non-carrying siblings have neither the biomarker changes nor get the disease. This means these are markers of AD,” Morris said. Following the announcement of the first DIAN therapeutic trial (see Part 1 of this series), below is a summary on new scientific data from the observation study. (For tricky ethical issues DIAN is confronting at this point, see Part 3; to get a sense of what the participants are thinking, see Part 4).
CSF Tau Shoots Up When Symptoms Start
Last August, DIAN formally published its first dataset of cognitive, fluid, and imaging markers (ARF related news story on Bateman et al., 2012); more recent results were presented at the Alzheimer’s Association International Conference in Vancouver, Canada (ARF related news story). Since then, DIAN scientists, who ‘freeze’ data of this ongoing, multicenter, longitudinal study every three months for successive analyses, have done so once again. They presented the latest cut to their steering, external advisory, and pharma committees on 27 September.
The biomarker core offered a glimpse at the first set of longitudinal data available so far. Anne Fagan showed that over a period of one to three years, CSF tau levels shot up in those mutation carriers who were just beginning to develop symptoms, rising much faster than in fellow carriers who are still presymptomatic. In non-carriers, CSF tau was lower in absolute amount and stayed flat over time. This might imply that treatment trials right around the age of symptom onset might find a robust measure in CSF tau. By comparison, CSF Aβ42 change over the same time frame was more variable, partly due to fickleness of the currently available CSF Aβ42 assays. In the subcohort with longitudinal biomarker data available, absolute CSF Aβ42 values were lower at both time points in carriers than non-carriers, but more time points on more samples are needed to sort out the pattern, Fagan said. This is perhaps not surprising given that CSF Aβ42 starts dropping many years earlier but does so slowly, said David Holtzman of WashU.
Besides this longitudinal finding, Fagan described how the case-control data derived from data freeze 4 had added power to the data freeze 1 results published in the NEJM. As of 31 August, DIAN’s biomarker core has analyzed a total of 262 CSF and 302 plasma samples. This larger set confirmed the findings reported in the paper. Notably, both total tau and phospho-tau is low in mutation non-carriers of all ages whereas it rises with age in carriers. Aβ42 is the same across age in non-carriers whereas in carriers it is high in young adults in their 20s and successively lower in people in their thirties and forties as they approach their estimated age of dementia onset. DIAN scientists are currently drilling deeper into exactly what happens in those few years when a carrier’s CSF Aβ42 concentration temporarily looks like that of a non-carrying sibling because it is passing through the normal range on its way down. This finding has implications for presymptomatic treatment trials if CSF Aβ42 were to be an inclusion criterion. A solution may lie in using the tau/Aβ42 ratio, Fagan said, as this measure predicts cognitive decline with high accuracy at all ages.
In preparation for the clinical trial, Fagan is currently transitioning her laboratory over to adopting Good Laboratory Practice (GLP) standards. For the trial, choosing the right assay will be critical, she said. Also in view of supporting the clinical trial, she would like to add measurement of the protein α-synuclein in the observational DIAN CSF samples. “Maybe if someone does not respond to a therapeutic, it could be because other pathologies are also present, such as Lewy bodies,” Fagan said. “It is important to capture all pathology in CSF to study individual response to treatment.”
DIAN’s neuropathology results already suggest there is reason to measure α-synuclein in this population. Nigel Cairns of WashU, who leads DIAN’s pathology core, told the audience that all autopsies of affected members of DIAN families so far have shown the β amyloidosis and tauopathy characteristic for Alzheimer’s. Beyond that, however, Cairns also saw sufficient α-synuclein pathology in DIAN participants and family members to make a combined diagnosis of AD and dementia with Lewy bodies (DLB), with advanced Braak Lewy body staging, in two of six cases autopsied to date. Cairns uses the new NIA-Alzheimer’s Association criteria for pathologic diagnosis of AD (Montine et al., 2012). TDP-43 or significant vascular pathology have not, so far, made an appearance. Of the DIAN participants themselves, four have passed away since 2008. Unfortunately, autopsy was possible only in two cases, as in one case the family declined and one was a suicide, Cairns said. DIAN checks in with participants for suicidality every three months, as populations at risk for autosomal-dominant neurodegenerative diseases are thought to be at somewhat elevated risk (see Alzforum essay).
Cairns further noted that DIAN autopsies show amyloid-β deposition in the cerebellum. This brain region is widely used as a reference region for amyloid imaging, but is seen as unsuitable for longitudinal studies because it does acquire some amyloid deposition as the disease progresses. For this reason, DIAN normalizes its amyloid imaging data against the brainstem.
ARIA Are Part of Autosomal-Dominant Alzheimer’s
On brain imaging, Tammie Benzinger of WashU, who heads the network’s imaging core, focused her talk on new data analyzed after the DIAN imaging presentations at AAIC (see ARF related news story). In Vancouver, the buzz had been about the basic timeline of biomarker change that DIAN had worked out. Perhaps for this reason, a separate result Benzinger also presented—about unexpected white matter atrophy in mutation carriers—got less attention. At this latest DIAN conference in St. Louis, however, white matter findings came to the fore. In particular, emerging data on amyloid-related imaging abnormality, or ARIA, not only may change scientists’ view of where ARIA fits into AD, but even call into question current safety rules for AD clinical trials.
Benzinger showed slides of DIAN participants who had numerous white matter abnormalities on their MRI scans that expanded between baseline and the person’s second visit two years later. A fuller analysis of one type of those abnormalities—tiny bleeds called microhemorrhage—showed that among 141 mutation carriers whose mean age was 39, 6 percent of the presymptomatic carriers had such bleeds in their brain. So did 25 percent of mildly symptomatic carriers, sixteen percent having one to four of these microhemorrhages, nine percent having more than five. “This is markedly abnormal for 40-year old people,” said Benzinger. Similar bleeds have been reported in older people with sporadic AD.
This DIAN finding suggests that microhemorrhages and other white matter changes may be part of the Alzheimer’s process itself, not necessarily a response to immunotherapy. After all, none of the DIAN participants with white matter abnormalities have yet taken such drugs. This is important because, at present, four microhemorrhages 1 millimeter in size, or one larger one, are grounds for exclusion from an anti-amyloid therapeutic trial or for stopping dosing. This would affect some mildly symptomatic DIAN participants who might otherwise be eligible to join the trial, said WashU’s Joy Snider. Snider will lead the WashU site in DIAN’s upcoming therapeutic trial.
“There is going to be a certain amount of this pathology in the population as a whole. In the DIAN population, this finding becomes prevalent as people become symptomatic,” Benzinger said, adding, “It would be a shame if we removed someone from study drug just because their underlying illness was still present. If the drug can slow progression we would not want to stop it prematurely.” Spurred by findings of ARIA in transgenic mice and in bapineuzumab and other clinical trials, the FDA has mandated MRI monitoring of the white matter every 3 months in anti-amyloid trials. Radiologists have long recognized white matter changes as part of AD, but have not communicated this effectively to colleagues involved in trial design, Benzinger said.
Among older people, several causes are known for microhemorrhage or the two other forms of white matter abnormality, hyperintensity and atrophy. An older person could have hypertension, diabetes, trauma, and radiation therapy, all of which restrict blood supply to small vessels in the brain. Because DIAN participants are so young, it is now clear that AD pathology can also cause these white matter changes. If the DIAN population had the previously known causes, age-matched siblings and cousins would have white matter findings at a similar rate regardless of their mutation status; however, they do not. “We see white matter atrophy, hyperintensities, and microhemorrhages almost exclusively in carriers,” Benzinger said.
Brain β amyloid pathology increases over the course of 20 years before a person develops dementia. Courtesy of Tammie Benzinger,WashU.
Overall, How Do FAD and LOAD Match Up?
Taking a step back, scientists asked how DIAN’s biomarker results match up to similar data from WashU’s ongoing longitudinal studies of sporadic AD? Overall, both forms of the disease develop similarly, said Benzinger. Among symptomatic patients, the amyloid load is similar in cortical regions, though DIAN participants have more in the caudate and some other regions. Among preclinical cases, DIAN is finding that amyloid deposition starts in the precuneus about 15 years and hypometabolism some 10 years before expected onset. Based on data from people who later developed late-onset AD, Benzinger estimates the same time course and brain areas, again with the exception that caudate and striatum are hit harder in the DIAN population.
Colin Masters, who leads the DIAN site at the University of Melbourne, Australia, said this overall impression holds up when comparing DIAN mutation carriers to participants in the Australian Imaging, Biomarker & Lifestyle Flagship Study of Ageing (AIBL) who develop amyloid. The Melbourne group has been gathering longitudinal data to calculate rates of progression. “The rate of amyloid deposition in familial AD is very close to that in sporadic AD. We suspected but did not know this before,” Masters said.
AIBL now allows the Australian researchers to calculate how quickly amyloid accumulates, because this large study has collected three data points over the course of at least five years. Rendering this data as a best-fit curve, the scientists can for the first time tell how long it takes the average person to go from the cutoff for amyloid positivity, which in this case is defined as SUVR 1.5, to a mean Alzheimer’s load, defined as SUVR 2.3. “It takes almost 20 years to go from crossing the amyloid threshold to Alzheimer’s dementia,” Masters said.
Thinking past DIAN and AIBL, knowing a typical rate of progression represents a step toward offering people a prognosis based on their amyloid load in a scan taken while they are cognitively normal. The best-fit curve to the AIBL data is linear at 3 percent increase per year between the threshold of positivity and a load corresponding to symptomatic Alzheimer’s at CDR 1. “In future, we will be able to tell someone who has a given SUVR that, statistically speaking, they are likely to develop symptoms in a set number of years,” Masters said.
Quite a bit of work still separates reality from this future vision, Masters added. For one, cognitive reserve and modifying genes likely speed up or slow down the rate of progression in a given person. ApoE, it is known, does not do that. It merely moves the curve to a younger start age. But other genes do, and those still require a lot of research before genetic information can be incorporated into a prognosis.—Gabrielle Strobel.
This is Part 2 of a four-part series. See also Part 1, Part 3, and Part 4.
Read a PDF of the entire series.