28 April 2008. What is going on in older folks who seem to do fine with amyloid-peppered brains? Do their brains rewire themselves to compensate for the unwelcome protein deposits? To address such questions, studies using functional magnetic resonance imaging (fMRI) have zeroed in on what’s been dubbed the “default network”—brain areas that are most active during passive daydreaming periods and typically turn off when the person engages in a focused mental task (see Lustig et al., 2003; Greicius et al., 2004; ARF related news story). Impaired default activity—in other words, failure to brush off distractions and focus on the task at hand—might help explain why people with MCI or AD have a harder time remembering things (see Buckner et al., 2005 and ARF related news story).

According to more recent studies led by Reisa Sperling, director of clinical research at the Memory Disorders Unit at Brigham and Women’s Hospital, Boston, coordinated deactivation of medial parietal lobe activity and activation of the hippocampus are required for a person to learn and remember a face-name pair (see Miller et al., 2008 and ARF related news story). Another study by Sperling and colleagues (Pihlajamaki et al., 2008) found that mild AD patients undergoing memory recall tests had abnormal increased activity in the medial temporal lobe (MTL), a larger brain area that houses the hippocampus. These results raised the possibility that MTL hyperactivation could help older people compensate for cognitive problems associated with amyloid buildup at the earliest stages of AD. To make this claim, the researchers needed to look more closely at the low-performing subjects and determine whether amyloid was in fact piling up in the brain areas with the odd network activity.

At this year’s Human Amyloid Imaging meeting, held 11 April 2008 in Chicago, Sperling presented results of this new study, which examined the relationship between fibrillar amyloid deposition detected by PIB-PET and regional brain activity during the face-name memory test as assessed by fMRI. The work was done in collaboration with Keith Johnson and Randy Buckner at Massachusetts General Hospital. In their analysis of 31 non-demented seniors (mean age 77.7; 19 with Clinical Dementia Rating (CDR) 0, 12 with CDR 0.5 who did not yet meet MCI criteria as defined in Petersen, 2004), high levels of PIB retention were associated with abnormal increased activity in the PCC (precuneus/posterior cingulate cortex) and medial frontal areas. These are part of a default network and typically “turn off” during successful memory formation—in this case, predicting the ability to remember a face-name pair 30 minutes later. Furthermore, high PIB uptake in the PCC correlated with increased hippocampal activation only when subjects recalled the face-name pair correctly but not when their memory failed.

These findings suggest that PIB-positive individuals who maintain good performance in cognitive tests might be sending their hippocampus into overdrive to do so, Sperling said. She noted in the Q&A, however, that her data do establish any causal relationships. At present, the study leaves unclear whether amyloid buildup makes the default network go awry or whether neuronal problems within the default network feed back and lead to amyloid deposition. “I don't have arrows saying that amyloid causes this dysfunction, although my bias is that it does,” Sperling said. “All I can say here is that these findings go together and that the disruption of functional activity occurs even before you have any medical symptoms.”

In a related study, neuropsychologist Dorene Rentz and colleagues (including Sperling and Johnson) at Harvard Medical School showed on a poster that normal elderly with more amyloid in the precuneus at baseline did noticeably worse on memory tests one year later. Though 15 of the 31 non-demented subjects were classified as CDR 0.5, the researchers saw the amyloid-memory correlation even in the remaining CDR 0 individuals, after adjusting neuropsychological test scores for cognitive reserve by using measures such as education and IQ. That any decline was seen after just one year, Rentz said, suggests that amyloid in the brains of apparently normal people might not be so harmless, after all.

On the other hand, data presented by Ira Driscoll of the National Institute on Aging, Bethesda, Maryland, seem to indicate that healthy seniors can tolerate a fair amount of amyloid before brain atrophy is seen on an MRI scan. As part of the Baltimore Longitudinal Study of Aging, the research measured amyloid load in 56 well-functioning elderly (mean age 78.7, mean MMSE 28.9), with PIB-PET scans taken about a decade after the first of up to 10 MRI scans. MRI data revealed declines in whole brain volume and volumes of all regions tested (ventricular CSF; white matter (WM); gray matter (GM); hippocampus; and frontal, temporal, parietal, and occipital WM and GM); no significant association was found between PIB retention and brain atrophy rates in the clinically normal study participants. The findings are consistent with the threshold model of disease, which would predict that the brain tolerates a certain degree of amyloid pathology until a threshold is crossed. In other words, Driscoll said, amyloid load may not affect brain volume or performance within the clinically normal range of function.

This issue is still open, though. In a study combining PIB-PET with fine-tuned volumetry, Keith Johnson reported that amyloid plaques do accumulate in brain areas marked by reduced cortical thickness. The researchers examined 55 people (32 classified as CDR 0; 15 as CDR 0.5; and 8 as CDR 1) and found that, at the vertex level, both amyloid buildup and cortical thinning were significantly greater in CDR 0.5/1 individuals than in the CDR 0 group. Local amyloid deposition was correlated with cortical thinning primarily in posterior cingulate/precuneus and temporal cortices—brain regions that commonly sustain neuronal damage in AD. During the Q&A after his talk, Johnson said he and colleagues are discovering that the brain areas for which the amyloid-cortical thinning relationship holds are quite different from areas where this correlation is not seen. Future studies will take a closer look at whether and how these regions relate to memory function. For a separate, recently published study comparing PIB-PET to MRI, see Jack et al., 2008. In Chicago, Cliff Jack of the Mayo Clinic, Rochester, Minnesota, presented follow-up work to this paper but still found no consistent relationship between PIB binding and memory performance in normal controls, unlike the Melbourne group, which did (see Part 2 of this series).

Regional differences also showed up in the work of Ansgar Furst, a postdoctoral fellow in William Jagust’s lab at the University of California, Berkeley. Using structural MRI, [11C]PIB and [18F]FDG PET, Furst examined the relationship between amyloid burden and glucose metabolism in 13 patients with probable AD (mean age 63.9, mean MMSE 20.0) and 11 healthy controls (mean age 72.6, mean MMSE 29.4). Voxel-wise, atrophy-corrected analyses showed that increased amyloid load in temporo-parietal regions (including the precuneus) and the anterior and posterior cingulate was associated with lower glucose use in these areas, whereas even whopping amounts of amyloid in the frontal lobes, striatum, and thalamus were not coupled with similar metabolic drops. These data support the idea that amyloid plaques exert non-uniform effects across different brain areas, and they invoke again the old question of what molecular mechanisms are behind the selective vulnerabilities of different brain areas in AD.

Get Your ZZZ’s: A Connection Between Aβ and Sleep?
Why does amyloid seem to preferentially deposit in specific brain areas in the first place? This question was addressed by HAI keynote speaker David Holtzman of Washington University, St. Louis. Holtzman first reviewed published data that make the case for CSF Aβ42 and tau as antecedent biomarkers for AD. In particular, the combination of amyloid imaging and CSF markers offers strong predictors for future AD (Fagan et al., 2006). He then described work first author John Cirrito and colleagues reported this month in Neuron showing that synaptic activity and Aβ levels are linked via the endocytic pathway (Cirrito et al., 2008 and ARF related news story). Holtzman concluded with intriguing new data on a connection between sleep and Aβ levels.

Why sleep? Having established that synaptic activity regulates Aβ levels in the brain’s interstitial fluid (ISF), the researchers wondered whether ISF Aβ is dynamically modulated over the course of a day, and whether drugs that modulate synaptic activity and Aβ might therefore have therapeutic potential. Using EEG electrodes to track brain activity in mice over several days, Jae Eun Kang, a graduate student in the Holtzman lab, observed that the animals’ Aβ levels are higher at night (when rodents tend to be more awake) and lower during the day (when rodents sleep more). The amount of wakefulness (minutes awake per hour) was significantly correlated with ISF Aβ levels, she found. Holtzman noted that Randy Bateman, also at Washington University, is assessing human CSF Aβ levels over time and has data to suggest that similar trends may be occurring in people, as well.

When the mice were sleep-deprived, their Aβ levels remained higher than they otherwise would have been during a period where sleep is more common, Holtzman said, suggesting that something associated with wakefulness is boosting Aβ. In rodents who were put on the sleep-enhancing drug diazepam, baseline Aβ levels dropped by about 30-40 percent. When given modafinil, a wake-promoting agent, the mice had increased Aβ levels. These findings support the idea that the Aβ fluctuations during sleep-wake cycles might stem from changes in synaptic activity, Holtzman said, but this connection has not been proven. Asked whether the new data suggest that certain anticonvulsants could be candidates for AD prevention, Holtzman mentioned a possible scenario in which drugs that regulate synaptic activity in specific ways without affecting cognition might be able to modulate Aβ levels, which ultimately determine whether Aβ aggregation occurs. The drug effects would likely be brain region-dependent, he said. His group is starting to do regional microdialysis experiments to look more closely at specific brain regions—for instance, those involved in the default network—to see whether Aβ levels correlate with localized synaptic activity.—Esther Landhuis.

This is Part 3 of a four-part series. See also Part 1, Part 2.