Just as cash in a rainy day fund can help people weather future economic downturns, extra years of school seem to have a protective effect in Alzheimer disease. This theory—that education builds “cognitive reserve,” which guards against later dementia amid buildup of brain amyloid—now draws support from the largest cohort study to detect AD pathology in living people. Exactly how the more highly educated cope as their brains fill with plaques remains unclear. An intriguing clue comes from another study in which researchers use functional magnetic resonance imaging (fMRI) to reveal neurophysiological differences that might explain in part why some folks have better memory even as young adults. Whether genetics, education, or other factors influence these brain features is not yet known, but the findings could help researchers develop measures that portend future cognitive impairment.

Data linking formal education with enhanced ability to withstand the burden of AD pathology has come primarily from studies assessing brain plaques and tangles at autopsy (e.g., Bennett et al., 2003; Roe et al., 2008). Such investigations are limited by variable lag times between the neuropsychological evaluation during a person’s life and the pathological evaluation after death. This gap closed with the development of positron emission tomography (PET) radiotracers (e.g., Pittsburgh Compound B [PIB]) that detect amyloid in living tissue. Earlier this year, researchers at the University of Turku in Finland published the first, small study using live brain imaging to demonstrate the education effect in 25 patients with mild AD (Kemppainen et al., 2008). Led by Juha Rinne, these investigators found that the most educated among their group of equally cognitively impaired research participants had more amyloid pathology than the least educated. A new study, published in this month’s Archives of Neurology by John Morris and colleagues at Washington University School of Medicine, St. Louis, Missouri, reports a similar trend in 198 seniors (161 non-demented, 37 with mild AD). “We found that for people who did not have Alzheimer’s plaques in the brain, there was no association [between education and cognitive test performance],” first author Catherine Roe told ARF. “But for people who did have amyloid plaques, they showed fewer dementia symptoms the more years of education they had.”

Drawn from community volunteers in ongoing memory and aging studies at the university, participants underwent all clinical and cognitive tests over a period of several weeks, with PIB-PET scans being done on average within seven to eight months. Fifty-nine people came up PIB-positive, the remaining 139 PIB-negative. This study differs from the Finnish study in how it was done, though not in the overall finding. The Finnish work related PIB load to education in equally impaired people; the St. Louis study related performance to education in subjects with mild AD or no dementia. On three measures of global cognitive function (Clinical Dementia Rating Sum of Boxes, Mini-Mental State Examination, Short Blessed Test), the WashU team found no influence of formal schooling on test performance among PIB-negative individuals. But in the PIB-positive group, the education effect jumped out: those with more than 16 years of education had the highest mean cognitive scores, followed by the group with 13-16 years and then those with 12 or fewer years of school. Roe and colleagues saw this in all three outcome measures, and got similar results when PIB uptake was considered as a continuous variable (as opposed to the dichotomous PIB-positive and PIB-negative categories used in the main analysis).

“It's a very nice study,” said Bill Klunk of the University of Pittsburgh School of Medicine. “And it follows a very nice study by the Finnish group.” Because PET-PIB imaging is costly and technology-intensive, “no one group can go and study hundreds and hundreds of subjects,” noted Klunk, who co-developed PIB with radiochemist Chester Mathis, also at Pittsburgh. “Everybody is putting pieces of preliminary data into the puzzle, and we will see if it holds up across a bunch of studies.”

Interestingly, no education effect showed up in a small PIB-PET study that found significant amyloid deposition in about a fifth of cognitively normal seniors. Klunk and colleagues published this study (Aizenstein et al., 2008) in the same current issue of the Archives of Neurology after having presented some of its data this past April at the Human Amyloid Imaging conference in Chicago (see ARF related news story). Among the nine of 43 healthy volunteers who came up amyloid-positive in this group, education appeared not to affect performance on cognitive tests. In the Washington University investigation, Roe and colleagues picked up an education effect in a PIB-positive group of 59 subjects, more than half of whom presumably had mild AD. By pooling non-demented and AD patients, Klunk said, the data could not reveal whether the presence or absence of dementia contributed to the education phenomenon.

Nevertheless, this PIB-PET study clearly supports the idea that more education helps people sustain higher brain Aβ loads before succumbing to AD. Similar support for the cognitive reserve theory came from a 21 October Neurology paper, in which Italian researchers assessed neurodegeneration not by amyloid imaging but by using fluorodeoxyglucose (FDG)-PET to measure brain glucose activity in 458 people (242 with probable AD, 72 with amnestic mild cognitive impairment (aMCI), 144 healthy controls). Among those with probable AD and those with amnestic MCI that progressed to AD during the study, Daniela Perani and colleagues at Vita Salute San Raffaele University in Milan, found that those with higher education and mentally demanding jobs had lower glucose metabolism in their posterior temporoparietal cortex and precuneus given a comparable cognitive impairment (Garibotto et al., 2008). In other words, the extra mental stimulation from school and vocation seemed to have helped delay onset of cognitive symptoms in people with AD-related changes in the brain.

These studies do not address what it is that might mediate the protective effect of education in seniors with amyloid and other tissue damage in their brain. (For a review of possible mechanisms of cognitive reserve, see Stern, 2006). “We don't know if there's a causal relationship between education and being able to withstand more AD pathology,” Roe said. “It could be some third variable that we're not measuring that causes you to stay in school longer and also causes you to do better on these cognitive tests.”

For instance, what if certain people’s brains are wired to remember things more efficiently? This possibility might not be so far-fetched in light of a paper published online on 10 November in PNAS Early Edition. Led by William Kelley of Dartmouth College, Hanover, New Hampshire, researchers used fMRI to measure brain activity in healthy young adults taking computerized tests that alternate between simple cognitive tasks and rest periods. They found that people with greater task-induced deactivation of blood-oxygen-level dependent (BOLD) activity in the medial temporal lobe (MTL) performed better on memory tests given during a separate visit.

AD researchers should perk up at the mention of the medial temporal lobe, as it houses several brain areas affected earliest in AD, such as the hippocampus and entorhinal cortex (see ARF related news story). The MTL forms a large part of the “default network”—brain areas that rev up when the mind is at rest and tone down during focused mental tasks. Resting-state activity is reduced in the MTL of AD patients (see ARF related news story) and of seniors with memory decline (Small et al., 2000; Small et al., 2002). Earlier this year, researchers reported that memory in older adults hinges upon their ability to turn down activity in the medial parietal area while firing up the hippocampus (see ARF related news story). This series of findings in older people led first author Gagan Wig to consider whether brain deactivation differences might correlate with memory ability at younger ages.

To address this question, Wig and colleagues recruited 45 healthy volunteers, mostly college students (mean age 20 years), from the Dartmouth area. Each participant underwent two fMRI scans coinciding with cognitive batteries involving different task-rest transitions. In the first test, subjects spent 30 seconds making odd/even judgments on a random set of numbers, followed by 30 seconds of rest, and so forth. Their memory performance correlated with resting-state BOLD activity in two MTL regions—one in the left hippocampus, another in the right hippocampus extending into the right parahippocampal gyrus. Those who did better on the memory tests were the people who deactivated their MTL most strongly during the task periods of the fMRI procedure.

To see if the results would hold across a modified experimental design, the subjects took a second test. This time, the task—pressing buttons to indicate when a flickering checkerboard comes on or off—was alternated between variable, shorter (0-10 sec.) rest periods. “The actual task was unimportant,” said Wig, now at Harvard University. “We just wanted to get that contrast between task and rest.” In the second test, resting activity in the right MTL once again correlated with memory scores. Effects in the left MTL did not replicate, but Wig and colleagues found correlations with other nearby brain areas, suggesting that the relationships on both sides of the MTL were meaningful, he told ARF.

As for why certain people had greater task-induced deactivation in the MTL, the authors could only speculate. “I don't think these people were doing anything intentionally different in the scanner. I don't think they were sitting there actively memorizing anything,” Wig said, noting studies showing that active encoding of recent experiences helps later recollection (e.g., Otten et al., 2006). One possibility, the authors suggest, is that the BOLD signal variability could reflect inherent differences in neuronal firing rates or metabolic activity that do not result from conscious information processing. Supporting the notion that such differences may be inborn, a recent investigation unveiled a set of genetic variants that boost memory and hippocampal activity in people (see ARF related news story).

Putting his work into context with the cognitive reserve study, Wig said, “The Morris paper is demonstrating how environmental factors might influence one's ability to sustain pathological burden, whereas we might be coming at it from the other end of the spectrum. We're saying it might be the case that certain individuals have tools that allow them to engage brain regions that are important for memory at a higher level.” Given that neurophysiological differences showed up in his study at such an early age, Wig said, “If one were to go ahead and follow these individuals into aging, you might potentially have a metric by which to predict who will suffer from mnemonic deficits.” In future studies, Wig and colleagues hope to test whether the correlation between memory and MTL activity holds up in older people (age 65 and up) lacking overt memory problems.—Esther Landhuis


  1. This work adds yet another piece of evidence for the so-called cognitive reserve hypothesis, meaning that persons with greater cognitive reserve are able to withstand more Alzheimer-type pathology without becoming demented. The authors found that the uptake of [11C]PIB interacted with years of education in predicting scores in several measures of cognition. In other words, persons who were "[11C]PIB positive" and had the highest education were clinically rated as less impaired in cognitive performance than individuals with less education.

    The findings are in accordance with earlier epidemiological, clinical, pathological, and functional imaging (cerebral metabolic rate and amyloid imaging) studies supporting the cognitive reserve hypothesis and delayed clinical expression of the disease in high-educated patients with marked Alzheimer pathology.

    One strength of the study by Morris et al. is a large number (161 non-demented individuals and 37 patients with Alzheimer disease) of study participants allowing adequate statistical evaluation. Longitudinal follow-up studies in cohorts like the one in this study are important to verify whether the predictions derived from cross-sectional studies on the relationship between the degree of amyloid pathology and cognition and the possible modifying effect of education will hold true.

    View all comments by Rinne Juha
  2. This report also indirectly illustrates the complexities of using individual putative biomarkers in clinical trials, either to characterize disease severity or as a surrogate outcome measure, without interpreting the marker within the context of the demographic and clinical characteristics of the patients.

    For example, a drug may–hypothetically –markedly lower PIB uptake and may not have a perceptible clinical effect in, say, a group of higher-educated patients with high PIB uptake at baseline. Yet the same drug might be seen to not lower PIB uptake but improve cognitive function in a subgroup of lower-educated patients (who may or may not have lower PIB uptake and poorer cognitive function at baseline than the higher-educated group).

    Under this hypothetical, an otherwise respectable effect drug might not be recognized. In brief, it may be time to model "cognitive reserve" into clinical trial designs.

    View all comments by Lon S. Schneider
  3. I don't see much that is fundamentally new about these findings. My and other groups have demonstrated similar relationships in the past using cerebral blood flow as a proxy for pathology. In that instance, we were able to show that when controlling for clinical severity, people with higher education had lower CBF, suggesting that they had more advanced pathology. David Bennett did this with postmortem studies as well. In his study, too, at any level of cognition pre-death, those with higher education had more pathology. These new data reinforce that.

    Also, previous analyses were done on patients with AD. Here the analyses focus on people who are PIB-positive, whether or not they have AD, although no information is provided about the overlap between those who are PIB-positive and those who have an AD diagnosis. A lot more work will have to be done to understand the implications of being PIB-positive and nondemented.

    The major news here is the use of PIB as an indication of pathology in living people. This in theory is better than the CBF that we had used, because it is a direct measure of pathology, while CBF is an indirect measure. Unfortunately, for PIB the main thing that you get is absence or presence of pathology (rather than a continuous measure of pathology severity). Therefore, the analyses had to compare the relationship of education to cognition in people who were and were not PIB-positive.

    Our group is actively investigating the neural implementation of cognitive reserve. The epidemiology suggests that some people can cope with or compensate for AD pathology better than others, but how does this work in the brain? We have two main ideas.

    1. Neural reserve: there is inter-individual variability in the networks that subserve cognitive processes in healthy individuals such that these networks are more efficient or have higher capacity in some people than others. Those people with higher cognitive reserve might be able to tolerate pathology better because their existing networks are more efficient, have higher capacity, or are more resilient in some other way. We perform imaging studies that investigate this idea.

    2. Neural compensation: when pathology damages the neural networks that normally serve cognition, people might "compensate" by recruiting other networks not normally used. Perhaps people who can recruit these networks more readily can compensate better. Alternately, some people might be able to continue to use the existing networks longer before they have to compensate. We are also pursuing this idea with imaging studies.
    Finally, we have been pursuing the idea that there may be some generic cognitive reserve network or cognitive process that can aid performance across a wide range of tasks. The ability to invoke this process would allow people to cope better with pathology.

    View all comments by Yaakov Stern
  4. This paper corroborates neuropathological findings: amyloid deposition can be found in non-demented patients. Correlation with cognitive deficits comes with intensification of amyloidosis and, most importantly, the extension of tau pathology.

    Question: why are Abeta oligomers, protofibrils and amyloid deposits observed in non-demented patients not toxic?
    Several hypotheses: neuronal reservoir, compensation, not-detectable MCI, or simply that they are not toxic in the human brain, but toxic in the experimental models constructed to demonstrate a toxicity.

    View all comments by Andre Delacourte

Make a Comment

To make a comment you must login or register.


News Citations

  1. HAI Chicago: PIB in Healthy People
  2. Tracing Alzheimer Disease Back to Source
  3. Network Diagnostics: "Default-Mode" Brain Areas Identify Early AD
  4. Deactivation Flaws Predict Memory Troubles
  5. Total Recall—Age, Genetic Variation, and Memory

Paper Citations

  1. . Education modifies the relation of AD pathology to level of cognitive function in older persons. Neurology. 2003 Jun 24;60(12):1909-15. PubMed.
  2. . Interaction of neuritic plaques and education predicts dementia. Alzheimer Dis Assoc Disord. 2008 Apr-Jun;22(2):188-93. PubMed.
  3. . Cognitive reserve hypothesis: Pittsburgh Compound B and fluorodeoxyglucose positron emission tomography in relation to education in mild Alzheimer's disease. Ann Neurol. 2008 Jan;63(1):112-8. PubMed.
  4. . Frequent amyloid deposition without significant cognitive impairment among the elderly. Arch Neurol. 2008 Nov;65(11):1509-17. PubMed.
  5. . Education and occupation as proxies for reserve in aMCI converters and AD: FDG-PET evidence. Neurology. 2008 Oct 21;71(17):1342-9. PubMed.
  6. . Cognitive reserve and Alzheimer disease. Alzheimer Dis Assoc Disord. 2006 Apr-Jun;20(2):112-7. PubMed.
  7. . Imaging physiologic dysfunction of individual hippocampal subregions in humans and genetically modified mice. Neuron. 2000 Dec;28(3):653-64. PubMed.
  8. . Imaging hippocampal function across the human life span: is memory decline normal or not?. Ann Neurol. 2002 Mar;51(3):290-5. PubMed.
  9. . Brain activity before an event predicts later recollection. Nat Neurosci. 2006 Apr;9(4):489-91. PubMed.

Further Reading


  1. . Cognitive reserve and Alzheimer disease. Alzheimer Dis Assoc Disord. 2006 Apr-Jun;20(2):112-7. PubMed.
  2. . Interaction of neuritic plaques and education predicts dementia. Alzheimer Dis Assoc Disord. 2008 Apr-Jun;22(2):188-93. PubMed.
  3. . Impaired medial temporal repetition suppression is related to failure of parietal deactivation in Alzheimer disease. Am J Geriatr Psychiatry. 2008 Apr;16(4):283-92. PubMed.

Primary Papers

  1. . Alzheimer disease and cognitive reserve: variation of education effect with carbon 11-labeled Pittsburgh Compound B uptake. Arch Neurol. 2008 Nov;65(11):1467-71. PubMed.
  2. . Medial temporal lobe BOLD activity at rest predicts individual differences in memory ability in healthy young adults. Proc Natl Acad Sci U S A. 2008 Nov 25;105(47):18555-60. PubMed.
  3. . Frequent amyloid deposition without significant cognitive impairment among the elderly. Arch Neurol. 2008 Nov;65(11):1509-17. PubMed.
  4. . Education and occupation as proxies for reserve in aMCI converters and AD: FDG-PET evidence. Neurology. 2008 Oct 21;71(17):1342-9. PubMed.
  5. . Cognitive reserve hypothesis: Pittsburgh Compound B and fluorodeoxyglucose positron emission tomography in relation to education in mild Alzheimer's disease. Ann Neurol. 2008 Jan;63(1):112-8. PubMed.