This is Part 1 of a four-part series. See Part 2, Part 3, and Part 4.
7 May 2009. Slowly but surely, Alzheimer disease researchers are coming to grips with the possibility that some experimental therapies could be failing because they have been tested in people whose disease is too advanced. The field’s focus is therefore shifting toward earlier diagnosis, even prevention. Reflecting the new emphasis, this year’s Human Amyloid Imaging (HAI) meeting drew some 150 researchers to Seattle on 24 April to share and discuss the latest in brain imaging, which will be crucial for identifying at-risk individuals and helping them resist impending AD.
A growing literature documents elevated brain amyloid in a substantial proportion of seniors who appear cognitively normal. This finding has stirred up new questions—not the least of which is whether this amyloid foretells future AD. That issue remains to be clearly resolved. However, ask any number of researchers at the HAI meeting, and chances are they’ll reckon that having a head full of amyloid is worrisome. “There was consensus that among normal people, amyloid is associated with changes in the brain,” said Reisa Sperling of Brigham and Women's Hospital in Boston, Massachusetts, in a conversation with this reporter during the poster session. “That might be very valuable in identifying individuals who will get preventive treatment. If we can identify people who are going to get AD a decade later, we have a window to treat people before they get symptoms,” Sperling noted.
To address that “if,” HAI’s opening session explored the relationship between amyloid deposition and functional changes in the brain. Each can be measured by positron emission tomography (PET)—the former with radiolabeled amyloid tracers, the latter using fluorodeoxyglucose (FDG) metabolism. Previous live brain imaging with the PET tracer Pittsburgh Compound-B (PIB) has revealed elevated Aβ in 10 to 30 percent of cognitively normal elderly, and has shown that amyloid load tracks longitudinally with whole brain atrophy. Elizabeth Mormino, a Ph.D. student in Bill Jagust’s lab at the University of California, Berkeley, addressed whether high amyloid deposition also coincides with reduced glucose metabolism, i.e., portends a loss of brain function. In her study, normal seniors and AD patients received magnetic resonance imaging (MRI) to track brain atrophy, FDG-PET to measure glucose metabolism, and PIB-PET to detect amyloid load.
Based on a published method for defining cut-offs (Aizenstein et al., 2008), all AD patients in Mormino’s group were classified as “high PIB,” as were 11 of 40 normal participants. Among the remaining healthy seniors, 21 came up as having “low PIB,” and the rest fell into an intermediate zone. The subjects were also grouped according to FDG metabolism. Within the cognitively normal group, high PIB was associated with reduced glucose metabolism, albeit with FDG-PET patterns less pronounced than in demented populations. Among those with high PIB, lower glucose metabolism also correlated with worse episodic memory. This trend did not hold for the people with low PIB. All told, the data suggest that elevated PIB uptake in normal seniors may suggest preclinical AD, Mormino said.
Ann Cohen, a postdoctoral fellow in Bill Klunk’s group at the University of Pittsburgh, Pennsylvania, also reported a link between amyloid load (PIB-PET) and cerebral metabolism (FDG-PET). In her study of 51 healthy seniors, 38 had high PIB uptake and 13 fell into the low-PIB group. Cohen used software that establishes cut-offs for “abnormal” and “normal” metabolism through automated iterative analysis of FDG-PET scans. By this method, Cohen further subdivided PIB groups according to glucose metabolism status. On her poster, Cohen reported that PIB-positive elderly were six times more likely to have abnormal metabolic patterns than were PIB-negative people. On the flip side, those with disrupted glucose metabolism were four times more likely to have elevated brain amyloid. However, a handful of participants had discordant PIB and FDG profiles, i.e., high PIB with normal FDG-PET, or low PIB with abnormal FDG-PET. This suggests that the correlation between the two techniques is not airtight. “People are realizing more and more that you need a multi-modal approach,” Cohen said.
Inconvenient and expensive as it is, this conclusion also emerged in a talk by Gil Rabinovici of the University of California, San Francisco. Rabinovici compared the value of PIB- and FDG-PET in distinguishing clinically diagnosed AD and frontotemporal lobar degeneration (FTLD). He chose FTLD because it is a non-Aβ dementia whose patients tend to be younger, which makes age-related amyloid less of a concern. Using either PIB or FDG correctly diagnosed about 80 percent of the patients (40 AD, 36 FTLD), but each method misclassified a different set of people. Hence, though the techniques showed high diagnostic sensitivity and specificity, they agreed with each other moderately at best, suggesting to Rabinovici that they are “far from redundant and are providing us with complementary information.”
In a similar vein, Shizuo Hatashita of Shonan-Atsugi Hospital and Clinic in Kanagawa, Japan, presented data suggesting that while brain amyloid load may indeed predict future development of AD, it seems to mark an earlier stage than does FDG-PET. In his study, all 56 AD patients, and 28 of 58 MCI patients, had robust PIB binding in cortical areas, reflecting a typical AD pattern. But reduced glucose metabolism accompanied this high PIB uptake in only 29 AD and in merely two MCI patients. Among the 91 healthy seniors in this study, 17 had elevated brain amyloid; their PIB patterns had lower intensity but otherwise resembled those of the PIB-positive MCI and AD participants. None of these 17 healthy seniors with high PIB had reduced glucose metabolism. Across all groups, higher PIB load correlated with worse cognitive test performance (Mini-Mental State Examination and CDR sum of boxes). Taken together, Hatashita’s data suggest that brain amyloid may track poorly with glucose hypometabolism, but nonetheless holds promise as a means for diagnosing preclinical AD, Hatashita said. In a post-meeting conversation with this reporter, he proposed a model of AD whereby amyloid deposition strikes early and would require another “hit” (e.g., mitochondrial abnormalities, free radical damage, calcium dysregulation) to produce neuronal dysfunction detectable by FDG-PET.
In an effort to bring some closure to what was becoming an increasingly complex story regarding PIB-PET and FDG-PET, Jagust asked during a Q&A if these two measures are correlated, anti-correlated, or have no relationship. “It depends on where in the brain you look, and when you look,” Dawn Matthews offered in reply. Matthews heads Abiant, Inc. in Deerfield, Illinois, a company that conducts imaging studies to help pharmaceutical companies assess their drugs. She and collaborators at New York University identified 50 participants of the Alzheimer’s Disease Neuroimaging Initiative (ADNI) who had both FDG and PIB-PET scans, preferably at least two PIB reads. In her talk, Matthews reported that her team saw progressively lower glucose metabolism levels in particular brain areas when comparing baseline scans of participant groups with increasing degrees of clinical decline. That is, normal controls had the highest metabolism levels; then came normal controls who converted to MCI; then MCI who stayed MCI; then MCI who converted to AD; and AD, whose metabolism was lowest. “Higher amyloid load and lower glucose metabolism seem to be correlated in regions known to be affected first in each modality, even in normal subjects,” she said. To explain, Matthews said via e-mail: “We did not find correlations between increased amyloid…and decreased glucose metabolism within medial frontal gyrus or prefrontal cortex, two of the primary regions in which amyloid has been found to accumulate. However, there was a correlation between amyloid in those regions and decreased glucose metabolism in the hippocampus (a brain structure involved in memory that is among the regions of earliest and most substantial atrophy in AD), as well as other regions where glucose metabolism declines in AD.”
Klunk parried Jagust’s question in the group Q&A with a question of his own. “Could increased basal metabolism to begin with accelerate amyloid deposition?” he asked, noting work by Washington University researchers that links higher synaptic activity to Aβ release into the interstitial fluid (see ARF related news story and Cirrito et al., 2005). Klunk then followed with this proposal: If people start at a high level of metabolism and can maintain that, perhaps it takes more of a second insult, e.g., Aβ, for them to progress from the MCI group into AD. Earlier in the day, Mormino offered a similar temporal argument for hypometabolism. “Maybe if your metabolism is slow and then you get hit by amyloid,” she said, you become more susceptible to decline. Then again, she said, “it could be that the amyloid induces the FDG-PET changes down the line.” Not to get lost in uncertainties, it is worth noting that this notion crystallized for the HAI conference participants: Amyloid deposition correlates with functional brain changes. See Part 2 of this series for more.—Esther Landhuis.
This is Part 1 of a four-part series. See Part 2, Part 3, and Part 4.
All attendees are invited to send additions and corrections to firstname.lastname@example.org.