The earliest signs of Alzheimer disease are bound to be subtle, arising from perturbations in brain function at the level of individual synapses. Picking up these early signals is critical to a sharper diagnosis and eventually, the early treatment that most experts agree will be the key to staving off AD.

In advances in early detection via brain scans, two new papers, one from Jeffrey Petrella and colleagues at Duke University, Durham, North Carolina, and the other originating from Christian Sorg and colleagues at the Technische Universitaet Muenchen, Germany, both show functional MRI (fMRI) measures that detect something amiss during the pre-diagnosis phase of AD marked by mild cognitive impairment (MCI). In the Petrella study, changes in cortical activity prospectively predict which patients will progress to AD. A third paper, from Mony de Leon’s lab at New York University School of Medicine, reports an even earlier change, using fluorodeoxyglucose-PET (FDG-PET) scans to measure glucose utilization. Those researchers reveal lower glucose uptake in people who have a family history of late-onset AD and hence a greater risk of developing disease, but who are (still) cognitively normal. Interestingly, poor glucose uptake was limited to people whose mothers had AD, but not those with a paternal history. The results imply that an unidentified maternal genetic factor decreases brain energy uptake, and may increase the risk of AD. The finding could help pin down the elusive genetic determinants of late-onset AD.

With the wider availability of fMRI in recent years, researchers studying neuronal activity have discovered consistent and striking alterations in brain function in AD. Some of these are changes in the default mode network, an interconnected distributed network that includes the hippocampus and posterior cortical regions. This network is activated when the brain is not attending to any particular thought (see ARF related news story), and becomes deactivated (that is to say, the fMRI BOLD signal declines) when people are asked to engage in tasks (see ARF related news story). People with AD show a loss of the deactivation response to memory tasks early on in the disease, and also display changes in resting activity in the network (Celone et al., 2006; ARF related news story; and ARF news story).

But are those changes useful for the early diagnosis of imminent AD? They may be, according to the work from Petrella and colleagues. Writing in the October issue of PLoS ONE, these authors present data that loss of deactivation in the posteromedial cortex during a memory test in patients with mild cognitive impairment predicted further memory decline and the progression to dementia in the ensuing years.

Petrella zeroed in on the posteriomedial cortex (PMC) because it is the earliest region affected by AD (see ARF related news story /new/detail.asp?id=1247), and his previous work had established that deactivation in the region during memory tasks was impaired in AD and in MCI (Petrella et al., 2007). That work uncovered a continuum of impairment, from the normal response in cognitively normal people, to a decline in amnestic MCI (a preclinical phase that often, but not always, leads to AD), and a further worsening in AD.

To assess the prognostic value of these changes, in the current study the researchers scanned 75 subjects (34 MCI, 13 AD, and 28 normal) and then followed their clinical progress for an average of 2.5 years. Among the MCI group, 11 progressed to AD during this time. The patients with MCI who converted to AD had significantly more failures to deactivate. Only half the group showed normal PMC deactivation, compared to 73 percent of subjects with MCI who did not convert. In the normal control group, 79 percent of subjects displayed deactivation, while only 23 percent of AD patients did. Even with the small sample size, the differences were statistically significant. Thus, the approach looks promising for helping to identify the subgroup of MCI subjects at greatest risk for progressive cognitive decline, the authors write.

In a second report on network activity in MCI, published in this week’s PNAS online, Sorg and colleagues report that, among eight different distributed neuronal networks they detect by fMRI, the default mode network is most affected in AD. That work is notable in that Sorg searched for network activity throughout the brain in a non-biased way, by analyzing scans to identify regions in which activity on fMRI went up and down in a coordinated way in resting subjects. Doing this, they found eight different resting state networks. When they compared activity in these regions between healthy elderly subjects and 24 amnestic MCI patients, only the default mode network regions showed a significant difference, consistent with previous results. The researchers also detected atrophy of the hippocampus and a lack of functional connectivity between the hippocampus and the posterior cingulate region of the cortex in people with MCI. Their work fits with the idea that the default mode network structures are among the first to be affected in AD. The authors report that they are now carrying out a 2-year follow-up of their patients, assessing symptom progression to find out if analysis of functional connectivity might predict the conversion to dementia, in the same way that the deactivation signal of Petrella and colleagues does.

The work may have relevance to other neurodegenerative or developmental brain disorders, as well. Just as the researchers have identified a selective disruption of the default mode network in AD, they may find that other diseases have different disconnectivity or disruption profiles that may prove useful for diagnosis.

The final report in the trio skips from fMRI to another measure of neuronal activity—the uptake of glucose as measured by FDG-PET scans. Decreases in glucose uptake in brain are measurable years before disease onset in people with familial AD mutations, and in people with MCI. Reduction occurs in the same regions that are picked out by fMRI, including the posterior cingulate, which is part of the posteromedial cortex. The new work, also published in this week’s PNAS online, indicates that a family history of AD, specifically a maternal history, is associated with a decrease in glucose uptake in cognitively normal elderly people.

To ask whether glucose uptake might be linked to risk of AD, first author Lisa Mosconi studied 49 cognitively normal elderly, 24 with a family history of dementia and 25 without. FDG-PET scans revealed significant glucose uptake reductions in the group with a family history, compared with those without. Upon closer inspection, however, Mosconi and colleagues found that the decrease was entirely accounted for by people with a maternal history, and that those whose fathers had AD showed normal uptake, similar to people with no family history. The authors suggest that the mechanism of transmission of lower glucose metabolism and/or heightened AD risk could be either X-linked or driven by mitochondrial DNA. Taken together with other data, including the observation that siblings of either sex incur similar risk from a parent with AD, they concluded that the most likely scenario is that the risk is passed along in mitochondrial DNA. This could easily account for changes in the capacity for glucose oxidation. They point out, however, that their study was small and the results need to be replicated in larger groups to determine the clinical applicability, if any.

These researchers, too, are following up with their patients with an eye to determining which ones go on to develop AD. “If these metabolic abnormalities predispose individuals to develop AD, FDG-PET studies of normal FHm [family history on the maternal side] individuals could provide a homogenous group to direct investigation of potential susceptibility genes for AD, to examine brain changes predisposing to AD, and to select participants for prevention studies.” The same could be said for any of the methods presented here, which provide a spot of hope for future diagnostic tests for early or incipient AD.—Pat McCaffrey.

Petrella JR, Prince SE, Wang L, Hellegers C, Doraiswamy PM. Prognostic value of posteromedial cortex deactivation in mild cognitive impairment. PLoS ONE. 2007 Oct 31;2(10):e1104. Abstract

Sorg C, Riedl C, Muhlau M, Calhoun VD, Eichele T, Laer L, Drzezga A, Forstl H, Kurz A, Zimmer C, Wohlschlager AM. Selective changes of resting-state networks in individuals at risk for Alzheimer's disease. PNAS Early Edition, week of Nov 5. Abstract

Mosconi L, Brys M, Switalski R, Mistur R, Sobanska L, Pirraglia E, Tsui W, De Santi S, de Leon M. Maternal family history of Alzheimer's disease predisposes to reduced brain glucose metabolism. PNAS Early Edition. 2007 November 5. Abstract


  1. This study is an elegant demonstration of alterations in the "cruise control" networks involved in default brain states. The strength of the authors’ methodology is that it is "model-free," that is, there are no assumptions about how individual events influence the brain. Even so, their conclusion supports our conclusion, which we reach by modeling our data, that is, posterior regions of the default mode network will be important areas for assessing clinical status of brain disorders, including dementia. These regions strongly influence how attention and memory are regulated and therefore are affected by Alzheimer's pathology and, more importantly, the prodromal stage of the disease, amnestic MCI.

    View all comments by Jeffrey Petrella
  2. Analyses of resting-state connectivity constitute a rapidly growing subfield of functional brain imaging. This technique is particularly appealing when studying neuropsychiatric patients, any of whom have difficulty performing tasks in the scanner. Depending on how thinly one slices resting-state fMRI data, there appear to be between six and 10 spatially distinct resting-state networks (RSNs) corresponding to canonical sensory and cognitive domains such as vision, sensory-motor function, executive function, and salience processing (1,2). The default-mode network (DMN) is perhaps the most heavily studied RSN. It has been implicated in episodic memory processing and self-referential thought (3-5). It has also been linked, by our group and several others, to Alzheimer disease (AD) (4-8). We had previously used independent component analysis (ICA) to demonstrate disrupted DMN connectivity in patients with mild AD compared to healthy elderly controls (4).

    The elegant study by Sorg and colleagues in Munich applies this same RSN approach to an earlier segment of the AD pathology spectrum, demonstrating reduced resting-state DMN connectivity in patients with amnestic MCI. This paper includes some particularly informative additional analyses. First, a voxel-based morphometry (VBM) analysis was used to demonstrate that reduced connectivity in the DMN was not driven by atrophy. Second, the authors supplemented their ICA with a region-of-interest (ROI)-based connectivity analysis to demonstrate reduced hippocampal-to-posterior cingulate connectivity in the MCI group (which was missed with ICA). Finally, by examining several RSNs, they demonstrate that the disrupted connectivity in MCI is relatively specific to the DMN.

    It seems, therefore, that resting-state fMRI is a sensitive tool for detecting group-level effects of early AD pathology—this despite the fact that the current study only acquired 4 minutes of resting-state data (scanning one brain volume every 3 seconds) on a 1.5T scanner. We typically acquire twice as many time points (6 or 8 minutes of rest scanning one brain volume every 2 seconds) and have found, anecdotally, that 3T is far superior to 1.5T in detecting resting-state functional connectivity. As such I suspect that this approach will perform even better in MCI when applied at higher field with the acquisition of longer time series. Obvious but important next steps include seeing whether DMN connectivity predicts conversion from MCI to AD and whether DMN connectivity can provide an early objective marker of treatment efficacy in clinical trials (granting the weighty presumption that novel therapeutics are near at hand).


    . Dissociable intrinsic connectivity networks for salience processing and executive control. J Neurosci. 2007 Feb 28;27(9):2349-56. PubMed.

    . Investigations into resting-state connectivity using independent component analysis. Philos Trans R Soc Lond B Biol Sci. 2005 May 29;360(1457):1001-13. PubMed.

    . Effect of Alzheimer disease risk on brain function during self-appraisal in healthy middle-aged adults. Arch Gen Psychiatry. 2007 Oct;64(10):1163-71. PubMed.

    . Default-mode network activity distinguishes Alzheimer's disease from healthy aging: evidence from functional MRI. Proc Natl Acad Sci U S A. 2004 Mar 30;101(13):4637-42. PubMed.

    . Molecular, structural, and functional characterization of Alzheimer's disease: evidence for a relationship between default activity, amyloid, and memory. J Neurosci. 2005 Aug 24;25(34):7709-17. PubMed.

    . Altered resting state networks in mild cognitive impairment and mild Alzheimer's disease: an fMRI study. Hum Brain Mapp. 2005 Dec;26(4):231-9. PubMed.

    . Functional deactivations: change with age and dementia of the Alzheimer type. Proc Natl Acad Sci U S A. 2003 Nov 25;100(24):14504-9. PubMed.

    . Alterations in memory networks in mild cognitive impairment and Alzheimer's disease: an independent component analysis. J Neurosci. 2006 Oct 4;26(40):10222-31. PubMed.

    View all comments by Michael Greicius
  3. This is a clever and potentially important study. As the authors discuss, there is a stronger maternal inheritance factor for Alzheimer disease (AD). It may relate to mitochondrial DNA, which comes exclusively from the maternal side, in addition to genomic DNA that can, of course, come from either parent. By showing a relative AD pattern of CMRgl reduction in maternal vs. paternal or no family history individuals, the authors show that individuals whose mother had AD have a metabolic pattern suggesting either early-stage disease or, at the very least, topographically relevant cortical vulnerability to AD. Replication of these findings will be important for establishing the validity of this observation, especially considering the small number of people in the study as the authors readily point out. But if confirmed, it may provide further impetus to the study of maternal factors that influence AD inheritance.

    View all comments by Richard Caselli

Make a Comment

To make a comment you must login or register.


News Citations

  1. Daydreams Traced to Default Network Activity
  2. Network Diagnostics: "Default-Mode" Brain Areas Identify Early AD
  3. Boston: Resting State MRI Shows Loss of Network Connectivity Early in AD

Paper Citations

  1. . Alterations in memory networks in mild cognitive impairment and Alzheimer's disease: an independent component analysis. J Neurosci. 2006 Oct 4;26(40):10222-31. PubMed.
  2. . Cortical deactivation in mild cognitive impairment: high-field-strength functional MR imaging. Radiology. 2007 Oct;245(1):224-35. PubMed.
  3. . Prognostic value of posteromedial cortex deactivation in mild cognitive impairment. PLoS One. 2007;2(10):e1104. PubMed.
  4. . Selective changes of resting-state networks in individuals at risk for Alzheimer's disease. Proc Natl Acad Sci U S A. 2007 Nov 20;104(47):18760-5. PubMed.
  5. . Maternal family history of Alzheimer's disease predisposes to reduced brain glucose metabolism. Proc Natl Acad Sci U S A. 2007 Nov 27;104(48):19067-72. PubMed.

Further Reading

Primary Papers

  1. . Prognostic value of posteromedial cortex deactivation in mild cognitive impairment. PLoS One. 2007;2(10):e1104. PubMed.
  2. . Maternal family history of Alzheimer's disease predisposes to reduced brain glucose metabolism. Proc Natl Acad Sci U S A. 2007 Nov 27;104(48):19067-72. PubMed.
  3. . Selective changes of resting-state networks in individuals at risk for Alzheimer's disease. Proc Natl Acad Sci U S A. 2007 Nov 20;104(47):18760-5. PubMed.