With the increasing focus on preclinical disease, scientists want to know what goes wrong first in AD brains, and where. In the April 9 Journal of Neuroscience, researchers led by Willem Huijbers and Reisa Sperling at Brigham and Women’s Hospital, Boston, report that amyloid deposits in the neocortex correlate with a failure to dial down the activity of the entorhinal cortex during memory tasks. These changes occur before any overt memory problems or hippocampal deficits, and are distinct from age-related decline, the authors found. The findings point toward the entorhinal cortex as a potential target for prevention studies, Huijbers suggested.

“This is a very nice paper that ties Aβ in the neocortex with dysfunction in the medial temporal lobe. In the past, it has been difficult to connect these two stories,” said William Jagust at the University of California, Berkeley. He was not involved in the work. The medial temporal lobe, which includes the entorhinal cortex, controls memory formation and contains mostly tau pathology early in AD.

Previous studies by Sperling’s group and others have shown that amyloid accumulates first in the default mode network (DMN), a set of connected regions largely in the neocortex that are active when the brain is at rest (see Feb 2009 news story). As this network becomes clogged with amyloid, its activity wanes (see Mar 2004 news story). Less is known, however, about how early amyloid deposits affect the functions of the hippocampus and entorhinal cortex. In some studies, these medial temporal lobe regions synchronize their activity with the DMN, suggesting a close connection.

Huijbers and colleagues wanted to investigate the relationship between neocortical amyloid and memory. Since memory also fades in normal aging, the authors first compared brain activity during a memory task in young and old volunteers. They recruited 48 participants around 75 years old, as well as 21 who were about 25, from the Harvard Aging Brain Study. The volunteers, all cognitively normal, first tried to memorize 150 faces paired with random first names. Then they were shown those pairings along with 150 new face-name pairs while their brains were scanned using functional MRI. Participants had to indicate whether they had seen each pair before. This paradigm allowed the researchers to see how brain activity changed when the volunteers recognized or failed to recognize a familiar pair, as well as when they memorized or failed to memorize a new pair. Brain activity was judged by changes in brain blood flow (blood-oxygen-level dependent, or BOLD, signal).

During both memorization and recognition tasks, participants activated several brain regions that included the hippocampus, while simultaneously quieting the DMN and entorhinal cortex. Older adults were less able to turn on the hippocampus and to squelch EC activity than were young adults, the researchers found (see image below). This difference in brain activity correlated with poorer performance on the task. The older participants recognized 57 percent of the known face-name pairs, significantly worse than the 63 percent that young volunteers got right.

The pattern of regional brain activation (yellow) and deactivation (blue) during a memory task becomes less precise with age. [Courtesy and with permission of Huijbers et al., The Journal of Neuroscience, 2014.]


How does the presence of amyloid change the picture? To determine this, the authors divided the older cohort into people with or without Aβ accumulation in their neocortex as seen by PIB-PET imaging. Those with amyloid scored just as well as those without on the memory task, but fMRI revealed differences in brain activity. While hippocampal activation was similar in both groups, those with amyloid had significantly more trouble turning off the EC during the memory task. This problem was worst when memorizing new face-name pairs, but also occurred when recognizing known pairs, and remained even after correcting for the effects of age, gender, brain atrophy, and ApoE genotype. The results suggest that Alzheimer’s strikes first in the EC, before it affects hippocampal function. This agrees with other studies (see Dec 2013 news story). 

“It’s been difficult to show behavioral effects in healthy people who have amyloid,” noted Denise Park at the University of Texas, Dallas. “This paper used a very clever design to separate the effects of healthy aging from amyloid pathology. It’s a good model for how we can use fMRI to understand what neural circuits are affected earlier and later by amyloid.” 

What factors might link amyloid in the neocortex to functional deficits in the entorhinal cortex? One possibility is that neurofibrillary tangles account for the EC’s sluggish response. The EC develops tau tangles during normal aging. This pathology accelerates in AD and likely precedes neocortical amyloid deposits (see Nelson et al., 2012). Thus, people with amyloid in the DMN should also have substantial tangles in EC. Huijbers plans to follow the participants for five more years using both amyloid and tau imaging to try to disentangle the effects of these two pathologies.

Alternatively, sputtering connections in the DMN may isolate the EC and weaken its ability to switch off in response to activity elsewhere. Huijbers and colleagues found that EC activity synchronized closely with DMN activity during the face-name task, while the hippocampus did not. Moreover, the EC meshed less well with the DMN in people with brain amyloid than in those without, although the trend did not reach statistical significance. The data hint that an overall loss of brain connectivity might underlie the earliest cognitive deficits in people on the path to AD. This fits with other data showing that people with Alzheimer’s experience a general breakdown of brain networks and lose the ability to switch smoothly between them (see Jul 2012 news storyMar 2014 news story). 

The findings may focus more attention on EC connectivity, said Howard Aizenstein at the University of Pittsburgh. “The dissociation between the entorhinal cortex and hippocampus proper is interesting, and this paper is the first to show that functionally,” he told Alzforum. Aizenstein wondered if EC dysfunction might eventually serve as a biomarker for clinical research, identifying those people with amyloid who will progress most quickly. Researchers agreed, however, that complex fMRI measures like this are unlikely to make good biomarkers for clinical practice.—Madolyn Bowman Rogers


  1. The authors made an elegant study in which they assessed the profile of activations and deactivations in young adults and cognitively normal elderly, and then performed additional analyses from two regions of interest (ROI): one preferentially located in the hippocampus and showing activation in both groups, and the other located mainly in the entorhinal cortex and showing deactivation (i.e., lower activity during the task than during the reference—cross fixation). They showed that age educed both hippocampal activation and entorhinal deactivation, and that amyloid deposition was associated with less deactivation in the entorhinal cortex.

    This study is interesting in that it emphasizes the complex relationships among amyloid deposition and brain activity and connectivity, hence reviving the debate on whether Aβ causes abnormal activity or whether aberrant activity promotes amyloid deposition. Longitudinal studies would probably help answer this question. The fact that this Aβ-related activity change was unrelated to memory performance further complicates the picture and our understanding of the effects of amyloid deposition and Aβ-related brain changes on cognition and clinical outcome.

    An intriguing finding is the different networks highlighted in assessing the functional connectivity of each ROI, with the default mode network (DMN) highlighted with the entorhinal ROI, while the hippocampal ROI revealed a more restricted network, including posterior ventro-medial areas. Whether the hippocampus is or is not part of the DMN may also in part reflect the antero-posterior location of the ROIs (when using seed-based connectivity analyses) because important intrinsic connectivity differences have been highlighted over the antero-posterior axis of the medial temporal lobe (see, for instance, La Joie et al., 2014; and Ranganath and Ritchey, 2012). Similarly, the inclusion of the hippocampus in the deactivation network may depend on the task, and is less likely to occur when using episodic memory tasks. This is another interesting question to explore: How much are the results, and especially the effect of amyloid deposition on the age-related decrease of entorhinal cortex deactivation, related to the nature of the activation task used to measure deactivation?


    . Intrinsic connectivity identifies the hippocampus as a main crossroad between Alzheimer's and semantic dementia-targeted networks. Neuron. 2014 Mar 19;81(6):1417-28. PubMed.

    . Two cortical systems for memory-guided behaviour. Nat Rev Neurosci. 2012 Oct;13(10):713-26. PubMed.

    View all comments by Gael Chetelat

Make a Comment

To make a comment you must login or register.


News Citations

  1. Cortical Hubs Found Capped With Amyloid
  2. Network Diagnostics: "Default-Mode" Brain Areas Identify Early AD
  3. New Look into Early AD Reveals Lateral Entorhinal Cortex Vulnerability
  4. Communication Breakdown: Multiple Networks Decline in AD Brains
  5. Cognitive Deficits Arise From Network, Not Regional, Dysfunction

Paper Citations

  1. . Correlation of Alzheimer disease neuropathologic changes with cognitive status: a review of the literature. J Neuropathol Exp Neurol. 2012 May;71(5):362-81. PubMed.

Further Reading

Primary Papers

  1. . Amyloid deposition is linked to aberrant entorhinal activity among cognitively normal older adults. J Neurosci. 2014 Apr 9;34(15):5200-10. PubMed.