Posted 19 April 2010
8th Annual Symposium on Early Alzheimer’s Disease: Focus on Early and Preclinical Alzheimer’s Disease
12-13 March 2010
Miami Beach, Florida
View Powerpoint Slide Index (Links to presentations also appear in the summary below.)
Last year, ARF and the organizers created a virtual conference for the 7th Annual Miami Mild Cognitive Impairment Symposium, and with the help of the speakers and organizers, we are happy to reprise that format this year. Read our summary of the meeting below, and click on the links to view the speakers’ slides. We welcome your comments!
Biomarkers Weigh In at MCI Meeting
19 April 2010. This year, the Miami Beach symposium kicked off on a rainy Friday with a new name and a new focus on biomarkers. For seven years, the meeting has been called the Mild Cognitive Impairment Symposium, but that changed this year to the Symposium on Early Alzheimer’s Disease. The change comes as a nod to a recent and dramatic shift in how Alzheimer disease is perceived, and what researchers increasingly see as the most promising angle of attack for future therapeutics. Part of the name change comes from the appreciation that AD probably begins years, or even decades before any outward cognitive changes, even the mildest ones. Many researchers now believe that during that clinically silent time, amyloid builds up, brain metabolism declines, and neurons are at risk and dying. The duration of this clinically silent stage seems to be influenced by genetic and lifestyle factors, and by “cognitive reserve,” an expression of the resiliency of the brain. Only when significant brain damage has occurred do the telltale memory signs show up, followed by progression to dementia. Probing the earliest stages of AD requires biomarkers, which are being tested at a rapid pace as a wealth of data emerges from the large Alzheimer’s Disease Neuroimaging Initiative study (ADNI; see ARF related news story).
However, the Miami meeting has traditionally been a place for neuropsychologists to have their say, too. The main sense this reporter got from talks on cognitive testing is that the psychologists feel they have an important and complementary role to play alongside biomarker analyses. Several speakers bemoaned the fact that the cognitive measures used to define mild cognitive impairment in the ADNI and other studies are fairly crude, and in fact define a rather late stage of disease. Some called for bringing more sophisticated measures to the table to help understand and track cognitive changes that are clearly occurring before the MCI stage. Their efforts parallel those of researchers in ADNI, which, in its second stage, will include a group of very mildly impaired subjects whose clinical state falls somewhere between normal and MCI (see ARF related conference story).
The realization that AD starts decades before dementia has driven a paradigm shift in the field, from cure to prevention, said keynote speaker John Morris, Washington University, St. Louis, Missouri. He outlined the view of AD as a progressive brain disease with two major stages: a preclinical or presymptomatic stage, and a symptomatic stage including prodromal (incipient) dementia and dementia. Both stages are part of a continuum of brain damage that proceeds from Aβ deposition to neurodegeneration.
The new concept turns away from using clinical or functional thresholds to define demented/non-demented, which can only be “arbitrary, ambiguous, and variably applied,” Morris said. The continuum model eschews the clinical category of mild cognitive impairment, which is based on impaired performance (compared to population norms) on a memory test, without reference to underlying etiology. Because of that, MCI has proved a heterogeneous clinical grouping that, while enriched for people with AD, captures many non-Alzheimer-related causes of cognitive impairment, and does not incorporate the important dimension of decline over time. Morris also avoids the term conversion (“It’s not a religious experience,” he said), in favor of the idea of a progressive loss of function.
Instead of relying on memory performance, the new view depends on using biomarkers to unmask underlying changes in the brain that begin during this long period of normal cognition and eventually lead to end-stage dementia and death. Top biomarkers include cerebrospinal fluid Aβ42 and tau, cerebral amyloid imaged by PET scanning using Pittsburgh Compound B (PIB) and newer ligands, structural changes detected by MRI, and deficits in glucose uptake on FDG-PET scans. Advances in biomarkers, fueled by the work of Morris and colleagues, and by the ADNI project, among others, suggest a model where an orderly succession of biomarker and cognitive changes reveal an underlying pathological progression from abnormal amyloid deposition to neuronal dysfunction, injury, and neurodegeneration (for more on this, see the recent review by Clifford Jack, Mayo Clinic, Rochester, Minnesota, and the recent ARF Live Discussion).
Understanding the preclinical phase of AD is important, because researchers believe that is where preventive measures and therapies will be most effective. An area of current research at Washington University, said Morris, is to define that progression in additional longitudinal studies of people at high risk for AD. At Wash U, that includes the Adult Children Study, which is following people whose parents had AD, and studies of familial AD with the Dominantly Inherited Alzheimer Network. View Morris Presentation [.ppt] (819 KB).
Pre-MCI: How Long Has This Been Going On?
The mother of all longitudinal studies, the Framingham Heart Study, began in 1948 with 5,209 subjects and is now in its third generation. In 1976, the researchers started neuropsychological testing of subjects as part of the epidemiology of dementia study, and have tracked participants since then for onset of dementia. Rhoda Au, Boston University School of Medicine, reported evidence of attenuated baseline performance on tests of memory, verbal learning and fluency, and abstract reasoning among people as much as 31 years before AD diagnosis. Those baseline performances raise questions about the length of the preclinical period of AD, and what is going on, disease-wise, in younger people who are many years away from AD, versus those older and perhaps closer to dementia. View Au Presentation [.ppt] (445 KB).
Separate from her talk, Au told ARF that the Framingham study is also starting to look at biomarkers. The researchers incorporated MRI measures into their study in 1999, and now have roughly 3,000 images, with several hundred participants who have had multiple MRIs. The heart disease arm of the study has been collecting blood data, too, including inflammatory markers that may be useful for AD studies, and Au said they are starting to take plasma Aβ measures as well.
Richard Caselli, Mayo Clinic Arizona, Scottsdale, studies cognitive changes in carriers of the ApoE4 allele, who are at high risk for AD. He previously demonstrated that this group, which on the surface appears to age normally, actually experiences accelerated memory loss beginning in their fifties (see ARF related news story on Caselli et al., 2009). Converging data on imaging biomarkers including MRI, FDG-PET, PIB-PET amyloid imaging, and white matter imaging, all suggest that this early memory decline represents a presymptomatic phase of AD. These asymptomatic but high-risk individuals make up a good target population for testing new therapies, Caselli argues. Others agree (see ARF related news story). View Caselli Presentation [.ppt] (4.4 MB).
In his talk, David Loewenstein, University of Miami and Mount Sinai Medical Center, Miami Beach, Florida, showed data on the identification of pre-MCI states in their Florida Alzheimer’s Disease Research Center population using a combination of neuropsychological and clinical data. He focused on a subset of patients whom either the psychologist or neurologist, but not both, classify as MCI. These patients do not meet the consensus criteria for MCI, but they had more apathy, more medial temporal atrophy, and greater deterioration in cognitive function than normal older adults. He found scores on specialized cognitive tests (most significantly, the Fuld Object Memory Evaluation), and measures of hippocampal atrophy correlated with the risk of progression to MCI within one to three years, even though the scores and measures all fell within age-adjusted norms.
The work highlights shortcomings in the current criteria for MCI, or “the elephants in the room,” as Loewenstein called them. One limitation is that the widely used brief measures of cognition, such as simple paragraph recall, may be affected by attentional problems, hearing difficulties, or anxiety. In addition, comparing test results to norms based on age and education levels and setting arbitrary cut-offs to define normal versus pathologic are problematic when the normative databases likely contain people with MCI or AD, and don’t account for quality of education, suggested Loewenstein. As he showed, moving to earlier pre-MCI states will require more sophisticated cognitive tests that are less susceptible to age and education effects. He called for more research on combining clinical measures and biomarker criteria. View Loewenstein Presentation [.ppt] (154 KB).
The question-and-answer session centered on the issue of whether biomarkers will make neuropsychological measures moot. For screening, what will be better, a cognitive measure, or a brain scan or spinal tap? Steven DeKosky, University of Virginia, Charlottesville, opined that the lumbar puncture necessary to collect cerebrospinal fluid for testing might not present the barrier some people think. The procedure, now viewed as highly invasive, was routine for many conditions 30 years ago, and is still common in cases of suspected multiple sclerosis. John Morris endorsed that view, saying, “People will undergo a lumbar puncture if they find an incentive to do so. If we have truly helpful medicine for AD, it will not be a barrier. Thousands are done daily in women undergoing delivery,” Morris said. He made sure to be clear that he values neuropsychology and its continuing role, but says for early AD diagnostic purposes it will need additional support, most likely from biomarkers.
Clinical Trials of the Future
Future trials of early-stage AD will likely use patients who are at a high genetic risk of AD, and Robert Green, Boston University, talked about this strategy. In their REVEAL study, his group focused on the effects of gene testing and disclosure of ApoE4 status to healthy adults. Their studies reveal that for the most part, people experience little short-term distress upon hearing they are positive for ApoE4 and thus at higher risk for AD (see ARF related news story on Green et al., 2009). With longer follow-up, however, he did find that the information could change behavior. People who were informed they were E4-positive were five times more likely to buy long-term health insurance, and three times more likely to be taking nutritional supplements a year later. Adding more information seemed to magnify effects on lifestyle: Subjects told that E4 also increased their risk for cardiovascular disease were twice as likely to increase their level of exercise compared to those who only learned about AD risk.
Going forward, Green said they want to expand the study to look at the effect of ApoE information on people with MCI. They will offer subjects with a memory problem, and thus already at risk for AD, a more precise estimate of their AD risk in three years by adding in E4 status. In older age groups, the risk for E4-positive elders to progress to AD is 60 percent, compared to 40 percent for E4 non-carriers (Petersen et al., 2005). Given those odds, Green expressed surprise that Elan had done so well setting up two separate trials in E4 carriers versus non-carriers (see ARF related news story), but then the subjects already have a diagnosis of probable AD. Such trial stratifications may be more difficult in asymptomatic subjects, he said.
Another sticking point for trial enrichment based on E4 status may be numbers. Given that 25 percent of people are E4 positive, 50,000 volunteer genotypes have to be revealed to identify 1,000 homozygotes. In the REVEAL study, Green says they sent 1.2 million letters out to ultimately enroll 300 people. View Green Presentation [.ppt] (2.5 MB).
Joseph Quinn, Oregon Health and Science University, Portland, Oregon, presented the results of two clinical trials of dietary supplementation with the omega-three fatty acid docosahexaenoic acid (DHA). One trial, sponsored by the Alzheimer’s Disease Cooperative Study (ADCS), enrolled subjects with mild to moderate AD, while the other, sponsored by the supplement manufacturer, was done in healthy elderly with age-related memory decline, as defined by low scores on a logical memory test. (For more details, see ARF related news story.) In both cases, the treatment had no significant effect on the primary endpoints (ADAS-cog and CDR-SOB for AD, or two tests of memory in the healthy adults). There was a whiff of an effect of DHA in ApoE-negative AD patients, who showed a benefit on the ADAS-cog and MMSE but no other measures. In the healthy elderly, a post-hoc analysis suggested some improvement in changes in episodic memory. Based on these results, some investigators have called for further study of DHA in MCI and/or E4-negative subjects, but Quinn said the ADCS steering committee has decided that the data do not justify trials in MCI at this point. “If we are able to continue this line of study, it will be with biomarker-based investigation,” he said, involving smaller studies on defined targets with different doses and combinations of fatty acids. View Quinn Presentation [.ppt] (1.2 MB).
Eric Siemers, Eli Lilly and Company, Indianapolis, Indiana, weighed in with the corporate view of clinical trials. It is too early for primary prevention trials yet, he said, because of a lack of a good presymptomatic screen with a high positive predictive value for AD. However, the predictive value of a screen jumps when people move from cognitively normal to MCI, making what he calls “secondary prevention trials” (treating people with MCI) feasible now. For outcomes, he likes continuous measures (CDR-SOB is most sensitive) compared to categorical measures (i.e., conversions to dementia) because the former may be more clinically relevant and allow smaller sample sizes. In terms of which candidate treatments should move to Phase 3, Siemers said that the question of what types of Phase 2 data might predict Phase 3 success remains an unanswered but major issue for the field.
A treatment that could slow progression by 50 percent could have a dramatic impact on AD prevalence, Siemers said. He modeled screen-and-treat scenarios assuming a lifetime risk of AD of 10 percent for a 65-year-old, and a test that was 90 percent sensitive and specific. In that model, the prevalence of AD could be cut in half by screening every five years, but at the cost of treating nearly half of the population. That may be similar to statins, he said, where close to half of 65-year-olds take the drug. He also opined that if a person is ApoE4-positive, PIB-positive, and has indicative spinal fluid Aβ/tau markers levels, that combination would give a pretty accurate assessment of AD risk, as good as genetic testing for some other diseases.
Biomarkers Give Early Insight
The next session dealt with biomarkers, namely FDG-PET (Norman Foster, University of Utah, Salt Lake City), fMRI, and amyloid imaging (Reisa Sperling, Harvard University). Foster stressed that glucose uptake measured by FDG-PET likely reflects synaptic integrity and function, and as such will read out abnormalities seen early in AD. Foster praised ADNI, which he said has been very successful at unifying data acquisition and quality control at multiple centers, and generating lots of high-quality FDG-PET data. ADNI is the largest ever FDG-PET imaging study ever done in aging and dementia, and the first multicenter longitudinal study to look at normal controls and MCI subjects. The results of scans on 200 subjects with MCI are being analyzed by investigators at three institutions. Data from William Jagust at the University of California, Berkeley, showed that decreased glucose uptake, as measured by FDG-PET, is a better indicator of disease status, because it correlates well with scores on the MMSE, unlike PIB-PET and CSF Aβ42. Foster’s group and Eric Reiman’s group at the Banner Alzheimer’s Institute in Phoenix, Arizona, have identified similar metabolic declines in MCI as in AD. The analyses they developed suggest FDG-PET will be useful for looking at individual patients for disease classification and possibly selection for clinical trials, and for determining prognosis. Foster reminded the audience that the original goal of ADNI was to determine if biomarkers could be used as measures of disease progression and surrogate endpoints of outcome in clinical trials, not to validate biomarkers for diagnostic or prognostic measures. However, he said with FDG-PET that aim has changed to seeing if scans can predict individual outcomes and be used to select subjects for clinical trials.
These FDG-PET studies will continue in the proposed ADNI2, where they will be extended to very mild MCI patients. In addition, subjects will be imaged for amyloid with the new agent 18F-AV45 (Choi et al., 2009). Foster also said that “pragmatic” trials of FDG-PET funded by Medicare are underway to determine if the procedure will be useful in diagnosing and managing elders with progressive cognitive impairment before they have reached the stage of dementia. View Foster Presentation [.ppt] (11.8 MB).
Sperling studies another early imaging marker for AD—the loss of network connectivity in the brain measured by fMRI. In particular, she has been interested in the 30 percent of elders who have normal cognitive function but show extensive amyloid deposition on PIB scans. By using fMRI during a face-naming task, Sperling has identified defects in functional connectivity in a distributed memory network in these subjects. She likens the memory task to a “stress test” for the brain, which can show up early evidence of functional compromise even before outward cognitive symptoms. Her group recently found similar functional problems in the default mode network that correlate with amyloid load measured by PET scan (see ARF related news story on Sperling et al., 2009).
Interestingly, that study showed no correlation between the extent of amyloid deposition and how people actually performed on the memory test. Could the disconnect between amyloid load and memory performance reflect cognitive reserve? In a separate study, Sperling and colleagues looked at the relationship between amyloid load, performance on a battery of neuropsychological tests, and cognitive reserve, as proxied by educational level and intelligence measures. In the study, individuals deemed to have high cognitive reserve did not show significant correspondence between amyloid load and cognitive performance compared to subjects with less education or lower IQ (Rentz et al., 2010). Sperling thinks this may present a systematic problem in academic research centers, where subjects tend to be people of high socioeconomic status and educational attainment. She is starting a new project to recruit working class or lower socioeconomic status subjects for studies of amyloid load and cognitive performance. In addition, her lab is beginning to look at vascular lesions using diffusion tensor imaging and white matter imaging to see if the presence of other factors might act in concert with amyloid to affect performance. Finally, she showed some evidence that using a more difficult memory test could unmask defects in cognitively normal people with high levels of amyloid, in effect overcoming their cognitive reserve. This suggests that researchers and clinicians need more challenging tests to discriminate between amyloid-positive and -negative subjects within the normal cognition group. View Sperling Presentation [.ppt] (15.0 MB).
In the final biomarker talk, Lisa Mosconi, New York University School of Medicine, presented new data on the FDG-PET in an at-risk population, namely the offspring of mothers with late-onset AD. Epidemiology studies indicate that maternal inheritance may account for 20 percent of late-onset AD, and Mosconi previously showed that cognitively normal adult children of mothers with AD have lower baseline FDG-PET and more rapid decline with age (see ARF related news story on Mosconi et al., 2007). Mosconi also recently reported that these same subjects have more amyloid than do controls (see ARF related news story on Mosconi et al., 2010). Such changes were not observed in brains of matched subjects whose fathers had AD. View Mosconi Presentation [.pdf] (1.1 MB).
The last session on the last day, and hence the last word, was given over to neuropsychologists who peered into the early stages of AD, MCI, and pre-MCI through their cognitive testing lens. Keynote speaker Mark Bondi, University of California, San Diego, critically examined the concept of MCI and its various subtypes. He raised some of the same issues as Loewenstein had earlier, pointing out that the gate to the MCI category involves a cursory screening with the MMSE and one memory test of delayed recall, in addition to a subjective memory complaint with no functional impairment. Each of these criteria has been debated and has evolved over the years. To determine the validity of the construct of MCI (and its subtypes) requires more comprehensive testing with a thorough sampling of different cognitive domains, suggested Bondi. When more comprehensive criteria are adopted to define MCI, correlations appear with changes in hippocampal volume, he showed (Jak et al., 2009). Also, deficits in either learning or retention of information in people with MCI correlate with different patterns of gray matter loss (Chang et al., 2010). His message was that despite increasing sophistication in genetics, imaging, and biomarkers, concomitant advances in profiling cognition in MCI and AD are lacking. “We as neuropsychologists can do a better job of characterizing this clinical state,” he said. View Bondi Presentation [.pdf] (2 MB).
Catherine Roe, Washington University, St. Louis, Missouri, revisited the concept of cognitive reserve, and showed that in her study group, subjects with more education were able to withstand more amyloid accumulation (detected by PIB-PET) before becoming demented. Her results support the idea that cognitive reserve is driving the disconnect between biomarkers and symptoms, and that researchers will need to learn how to factor in cognitive reserve to increase the accuracy of biomarkers. View Roe Presentation [.ppt] (562 KB).
In her talk, Katherine Nutter-Upham, City University of New York, buttressed Bondi’s argument with results of verbal fluency performance in older adults with MCI, cognitive complaints, or no impairment. Difficulty in finding words during conversation is the single most common complaint in older adults, and by using multiple tests of word production, Nutter-Upham showed that the different tests and their distinct underlying cognitive processes discriminated healthy controls and subjects with MCI, but to different degrees. Her results argue that multiple tests that tap specific and distinct cognitive processes can give a better view of underlying defects. View Nutter-Upham Presentation [.ppt] (1.7 MB).
Amy Jak, University of California, San Diego, picked up the theme of defining objective cognitive impairment. By applying several different criteria proposed to define MCI to the same group of normal elderly subjects, she found that MCI prevalence rates and prediction of progression to AD varied widely (Jak et al., 2009). The criteria also affected individuals, with some ending up in different groups depending on the operational definition used. She showed that subjects who present with a single affected cognitive domain are the most likely to revert to normal (i.e., be falsely positive for early AD), and that diagnoses that rely on more than one test in each cognitive domain are better at predicting atrophy in the medial temporal lobe. Jak concluded that assessments should include measures in domains other than memory, and that memory assessments should cover both verbal and non-verbal memory. Most interesting, her analysis suggests that the cut-off scores on the current memory test used to define MCI for the ADNI study (single paragraph recall) may be capturing only the most severe cases. That supports the idea that MCI is quite far along the continuum to AD, and may, in fact, not represent the best target group for therapies. View Jak Presentation [.ppt] (600 KB).
Jennifer Manly, Taub Institute and Columbia University Medical Center, New York, talked about the problem of assessing cognitive impairment in minority populations. Her work in an immigrant community in upper Manhattan indicates that existing methods have poor specificity in predicting MCI and progression to AD (Manly et al., 2005). One thing not to do, Manly said, is try to get around language issues with a non-verbal test. On the other hand, comparing subjects to normative groups of similar race or ethnicity does reduce misdiagnosis. Manly showed that racial and ethnic differences in the rates of age-related cognitive decline and AD in the U.S. can be attributed to regional differences in the quality and quantity of schooling. Rather than years of education, it was the state where subjects were raised and their reading levels that made the best predictors of future cognitive decline. As people are evaluated earlier and earlier in disease, researchers will have to look carefully at these kinds of factors, Manly concluded. View Manly Presentation [.ppt] (1.3 MB).
As the meeting wrapped up (and the sun appeared), attendees could hit the beach, satisfied to have heard two sides of the early AD story—the biological and the psychological. At the outset, conference organizer Ranjan Duara, Mount Sinai Medical Center, Miami, said that part of the reason he changed the name of the meeting was mercenary. With the bad economy, support for meetings like this one has waned. Duara noted that the FDA has not approved any medications for the treatment of MCI, so he hoped the name change to early AD might attract more support from pharma for the meeting. As yet, that ploy has been unsuccessful, but Duara holds out hope that someone will step up to keep the annual event going for another year. View Duara Presentation [.ppt] (800 KB).—Pat McCaffrey.