Dementia refers to the loss of intellectual or thinking abilities and Alzheimer’s disease is its most common cause. Although we don’t understand the cause of Alzheimer’s disease, we know that at early stages of the disease a protein known as beta-amyloid becomes deposited in the brain in the form of plaques. Evidence suggests that this beta-amyloid is harmful to brain cells. Although we don’t know why amyloid is deposited, research in animal models suggests that the activity of the brain cells themselves may lead to its deposition.

We can measure brain activity in living people using a technique known as functional magnetic resonance imaging (fMRI), and we can also measure amyloid deposition using positron emission tomography (PET scanning). Leal et al. have now used these techniques to learn about how brain activity might influence amyloid deposition.

Volunteers initially performed a memory task while their brain activity was measured using fMRI. This gave a “baseline” level of brain activity. Over the course of several years, the volunteers returned for PET scans and further memory tests. Cognitively normal older adults with greater baseline levels of brain activity – particularly in the hippocampus, a brain region involved in the formation of new memories – showed more beta-amyloid accumulation over the next three to four years. Furthermore, people who accumulated more beta-amyloid also showed a more severe decline in memory.

To strengthen these results, a follow up study should be performed that examines how brain activity and amyloid deposition change together over time. In addition, it will be important to test whether methods that reduce brain activity could affect amyloid deposition, thus perhaps reducing the risk of Alzheimer’s disease.

Beta-amyloid (Aβ) is one of the core pathologies of Alzheimer’s disease (AD) and Aβ deposition is posited to be the earliest event in a long and complex pathological cascade leading to AD. The amyloid cascade hypothesis suggests that soluble forms of Aβ represent the start of AD, which is ultimately associated with damage in synaptic structure and function and the formation of Aβ plaques (Hardy, 2009). However, it remains unknown why Aβ deposition occurs in the first place.

A substantial body of preclinical research suggests that synaptic activity is linked to Aβ deposition (Nitsch et al., 1993; Kamenetz et al., 2003; Cirrito et al., 2005; Bero et al., 2011). Studies using in vivo microdialysis with concurrent field potential recordings have shown that Aβ levels are directly influenced by synaptic activity, in which increases in neuronal activity are sufficient to modulate Aβ levels and plaque growth in a local, region-specific manner (Cirrito et al., 2005; Bero et al., 2011). Furthermore, using an acute brain slice model, rapid effects of synaptic activity on Aβ levels have been shown to be primarily linked to synaptic vesicle exocytosis (Cirrito et al., 2005). There is evidence that hyperactivity in the CA3 subregion of the hippocampus in aged rodents may underlie memory deficits associated with aging (Wilson et al., 2005; El-Hayek et al., 2013; Haberman et al., 2013). This hyperactivity has been regarded as epileptiform in nature in rodents (Koh et al., 2010) and AD transgenic mouse models (Palop et al., 2007). A reversal of memory impairment has been shown with a reduction of hippocampal hyperactivity (via a low-dose of an anti-epileptic drug) in age-impaired rodents, AD rodent models, and in mild cognitive impairment (MCI) (Koh et al., 2010; Bakker et al., 2012, 2015; Sanchez et al., 2012). There are at least two proposed hypotheses regarding increased hippocampal activation. It has been suggested that hyperactivity within the hippocampus may occur over the course of normal aging, but hyperactivity may eventually lead to the secretion of Aβ more quickly than would otherwise occur. Alternatively, some have suggested that Aβ may lead to increased neuronal activity. These explanations are not mutually exclusive; both may be occurring leading to a vicious cycle of Aβ production.

Human studies have also shown that increased hippocampal activity is associated with memory function and Aβ deposition in aging and MCI (Bakker et al., 2012; Brody et al., 2008; Mormino et al., 2012; Yassa, 2011; Dickerson et al., 2005). High-resolution fMRI during memory task performance has localized hyperactivity to the dentate gyrus/CA3 subregions of the hippocampus, which has also been linked to memory deficits in aging and MCI (Bakker et al., 2012; Yassa, 2011). Cognitively normal older adults with high Aβ have increased hippocampal and default mode network (DMN) activation during episodic memory processing (Mormino et al., 2012; Sperling et al., 2009; Vannini et al., 2012). Subjects with MCI have shown heightened medial temporal lobe (MTL) activation during episodic memory processing that is associated with subsequent cognitive decline (Dickerson et al., 2005, 2004; O'Brien et al., 2010). A recent study found that individuals with MCI who were Aβ+ at baseline showed increased hippocampal activation at baseline and longitudinally in addition to longitudinal cognitive deficits compared to MCI subjects who were Aβ− (Huijbers et al., 2015). Apolipoprotein E4 (ApoE4) carriers have also shown increased MTL activation (Dennis et al., 2010; Bookheimer et al., 2000). Eventually, it appears that hippocampal activity decreases as AD pathology progresses (O'Brien et al., 2010; Celone et al., 2006).

In addition to MTL activity, human imaging studies have also shown increased cortical activity in regions associated with the DMN, but also in regions beyond the DMN (Mormino et al., 2012; Bookheimer et al., 2000; Elman et al., 2014; Jones et al., 2016; Myers et al., 2014). A recent study found that older adults with Aβ deposition had increased memory task-related neural activation in the parietal and occipital cortex that was linked to more detailed memory encoding (Elman et al., 2014), suggesting the activation increases were compensatory. Other studies have discovered links between Aβ and the DMN such that Aβ aggregates in areas of high intrinsic connectivity (Myers et al., 2014) and that there is high connectivity between the posterior DMN and hubs of high frontal lobe connectivity that are associated with Aβ accumulation (Jones et al., 2016). High Aβ deposition has been found in locations of cortical hubs, such as posterior cingulate, lateral temporal, lateral parietal, and medial prefrontal cortices (Buckner et al., 2009). It remains unclear as to whether local amyloid deposition is driving local hyperactivity, if region-specific hyperactivity is linked to global amyloid deposition, or both may be occurring. Previous studies have shown that hyperactive neurons were typically found near amyloid plaques in the cortex (Busche et al., 2008), however, increased hippocampal activity occurred before transgenic mice began to deposit amyloid plaques. Hippocampal hyperactivity can occur independent of plaque deposition, while increased activity in cortical neurons is more directly linked to the presence of amyloid plaques. This may be due to a difference in local concentrations of amyloid, different vulnerabilities of cortical versus hippocampal neurons, or both (Busche et al., 2012). Another possibility is that hippocampal hyperactivity may drive increased cortical activity in downstream regions that are more directly linked to amyloid deposition or that hippocampal hyperactivity is driving tau deposition in the MTL, which then leads to Aβ deposition.

One proposed model suggests that life-long patterns of neural activity may lead to Aβ deposition (Jagust and Mormino, 2011). Thus far, most studies have examined the link between Aβ and brain activation using cross-sectional data. Longitudinal studies are essential for testing hypotheses about brain activation leading to Aβ accumulation over time and to determine if the rate of Aβ deposition is different in people who show more or less brain activation.

In the present study, our goal was to examine how hippocampal neural activity influenced the longitudinal accumulation of global Aβ and if these factors were related to changes in cognition. We hypothesized that cognitively normal older individuals with greater hippocampal activity at baseline would show increased longitudinal Aβ accumulation globally and that this would be associated with memory decline. To test this hypothesis, we performed a study with a baseline assessment of memory during acquisition of functional magnetic resonance imaging (fMRI) to measure task-related neural activity, followed by longitudinal [¹¹C] Pittsburgh Compound-B positron emission tomography (PIB-PET) imaging to measure global Aβ deposition across time in addition to repeated assessments of memory. We focused on the California Verbal Learning Test Long-Delay Free Recall (CVLT) to measure long-term memory changes over time.

Forty-five cognitively normal older adults participated in the fMRI experiment, the results of which have been previously published (Mormino et al., 2012). For the current study, we included all subjects who had: (1) participated in the original fMRI experiment, (2) at least two PIB scans (first near the time of fMRI), and (3) at least three neuropsychological testing sessions. Fifteen of the 45 subjects did not return for follow-up PIB-PET imaging after their initial PIB scan. Three additional participants were excluded due to technical factors. This left us with a sample of twenty-seven subjects (N = 27) who met the criteria for the study (see Table 1).

Participants performed a memory task inside the MRI scanner that involved being shown two hundred images of natural outdoor scenes with instructions to indicate whether water was present in each image. After the MRI scan (15 min later), a surprise recognition task was conducted that included all stimuli presented during encoding as well as one-hundred novel stimuli. For each image, subjects were asked if they had seen the image before and were allowed to respond with one of four responses: (1) high-confidence yes, (2) low-confidence yes, (3) high-confidence no, and (4) low-confidence no. The recognition task was self-paced and subjects were encouraged to be as accurate as possible.

We examined three key measures: (1) task-related activity in the right hippocampus and several non-hippocampal regions that showed increased activation in PIB+ compared to PIB- individuals during high-confidence hits versus misses (Mormino et al., 2012), (2) PIB distribution volume ratios (DVR) to measure global Aβ deposition across frontal, parietal, temporal, and cingulate cortices, and (3) performance on the CVLT Long-Delay Free Recall to measure changes in long-term memory. The right hippocampus was chosen for analysis because this hippocampal region demonstrated the strongest effect of Aβ on activation in the original study, perhaps because the stimuli were visually presented scenes (Mormino et al., 2012).

To examine relationships between hippocampal activation at baseline, longitudinal Aβ accumulation, and longitudinal memory performance, we conducted a series of linear mixed models. All models included age, sex, and education as covariates, as well as a random intercept to account for individual variability in initial PIB DVR or CVLT scores and random slopes to account for individual variability in PIB DVR or CVLT slopes. Age was mean-centered at 76.5 years, education was mean-centered at 17.4 years, and females were the reference group, so that resulting parameter estimates could be interpreted as representing an example 76.5-year-old female with 17.4 years of education.

Figure 1 shows the slopes of PIB DVR and CVLT over time for each subject. PIB DVR slopes across time (relative to baseline MRI) for each subject are shown in Figure 1a. CVLT slopes across time (relative to baseline MRI) are shown in Figure 1b. The average time from the baseline MRI to last PIB scan was 3.4 years (range 2.8–6 years) and from MRI to last neuropsychological session was 2.7 years (range 1.5–6.1 years). Participants came in for PIB scans and neuropsychological testing sessions on separate days. The average time between MRI and first PIB scan was about one week and the average time between MRI and first neuropsychological session was about eight weeks. At the baseline exam, six subjects were classified as PIB+ (four were ApoE4 carriers) and by the end of the study eight subjects were PIB+ (both who converted to PIB+ were ApoE4 carriers).

Figure 1

Download asset Open asset

Figure 1—source data 1

This file contains the source data for Figures 1–4.

https://doi.org/10.7554/eLife.22978.005

Download elife-22978-fig1-data1-v2.xlsx

Hippocampal activation at baseline associated with longitudinal Aβ accumulation

To determine whether hippocampal activation at baseline predicted longitudinal accumulation of global Aβ, we examined associations between our independent variables of interest (hippocampal activation and time of PIB) and the dependent variable (PIB DVR) using a linear mixed model (with age, sex, and education as covariates). Time of PIB was determined as relative to baseline MRI. The interaction between hippocampal activation and time of PIB (hippocampal activation × time) was the primary effect of interest because it represented the relationship between hippocampal activation and longitudinal PIB DVR change. Results revealed that hippocampal activation at baseline was positively associated with longitudinal amyloid accumulation such that PIB DVR increased at a rate of 0.001 units per year for every one hippocampal activation unit [interaction, p=0.004, Table 2]. To visualize this, we plotted the relationship between PIB slope from the model and hippocampal activation contrast values at baseline (Figure 2a), which showed a positive correlation between PIB slope and hippocampal activation.

Figure 2

Download asset Open asset

Table 2

Parameter	Estimate (SE)	df	t	p
Intercept	1.00 (0.07)	22	14.58	<0.001
Age (at MRI, mc)	−0.001 (0.01)	22	−0.14	0.887
Sex (F)	0.10 (0.08)	22	1.27	0.216
Education (years, mc)	−0.02 (0.02)	22	−0.89	0.385
Hippocampal activation (baseline)	0.004 (0.002)	22	1.64	0.115
Time (PIB DVR)	0.003 (0.003)	12	0.87	0.399
Hippocampal activation * Time (PIB DVR)	0.001 (0.0002)	12	3.58	0.004

1. Dependent Variable: PIB DVR.

2. Key: SE = Standard Error, mc = mean-centered, PIB DVR= Pittsburgh Compound B Distributed Volume Ratio.

We also performed the same linear mixed model described above, but with non-hippocampal activation as an independent variable to determine if brain activation predicting longitudinal Aβ accumulation was specific to the hippocampus or existed for cortical activity as well. We chose the non-hippocampal regions of bilateral occipital cortex and bilateral inferior frontal gyrus, which similarly to the hippocampus, showed increased activity in PIB+ compared to PIB- subjects in the original study (Mormino et al., 2012). However, we did not find an interaction between non-hippocampal cortical activation and longitudinal Aβ accumulation [interaction, t(21) = 1.02, p=0.319; Figure 2b], suggesting increased brain activation at baseline associated with increased global Aβ accumulation may be MTL specific.

To determine whether hippocampal activation at baseline was associated with change in memory performance over time, we examined associations between our independent variables of interest (hippocampal activation and time of CVLT) and the dependent variable (CVLT) using a linear mixed model. As with time of PIB, time of CVLT was determined as relative to baseline MRI. The interaction between hippocampal activation and time of CVLT (hippocampal activation × time) was not significant [interaction, t(14) = 0.786, p=0.445] indicating no relationship between hippocampal activation at baseline and the subsequent rate of memory decline.

Increased Aβ accumulation associated with a longitudinal decline in memory performance

Next, we examined the relationship between longitudinal Aβ accumulation and memory performance. We examined associations between the independent variables of interest (PIB slope and time of CVLT) and the dependent variable (CVLT) using a linear mixed model. The interaction between PIB slope and time of CVLT (PIB slope × time) was the primary effect of interest because it represented the relationship between Aβ accumulation and memory change over time. Results revealed that increased longitudinal Aβ accumulation was associated with a decline in CVLT score [interaction, p=0.023, Table 3]. To visualize the effect, we plotted the relationship between CVLT slope from the model and PIB slope (Figure 3), which showed a negative correlation between CVLT slope and PIB slope. Sex [p = 0.034] and time of CVLT [p = 0.023] also predicted the average CVLT score (Table 3). The effect of time of CVLT can be seen in Figure 1b, where the average CVLT slope increases over time, by 0.41 CVLT units/year. This suggests that while many participants retain their memory abilities over time, longitudinal Aβ accumulation predicts memory decline over time.

Figure 3

Download asset Open asset

Table 3

Parameter	Estimate (SE)	df	t	p
Intercept	10.07 (0.89)	18	11.27	<0.001
Age (at MRI, mc)	−0.06 (0.09)	16	−0.69	0.499
Sex (F)	2.54 (1.10)	16	2.31	0.034
Education (years, mc)	−0.17 (0.28)	17	0.59	0.562
Slope (PIB DVR)	15.6 (28.32)	21	0.55	0.587
Time (CVLT LD FR)	0.41 (0.17)	23	2.44	0.023
Slope (PIB DVR) * Time (CVLT LD FR)	−20.06 (8.26)	25	−2.43	0.023

1. Dependent Variable: CVLT LD FR.

2. Key: SE = Standard Error, mc = mean-centered, PIB DVR= Pittsburgh Compound B Distributed Volume Ratio, CVLT LD FR = California Verbal Learning Test - Long Delay Free Recall.

Longitudinal amyloid accumulation mediates the influence of hippocampal activation on memory decline

Based on the observation that hippocampal activation at baseline was related to increasing Aβ deposition over time, we conjectured that the relationships between the three variables in this study were complex. Thus, we examined whether the increased longitudinal Aβ deposition associated with baseline hippocampal activation was linked to memory decline. We tested this relationship using a mediation model, where PIB slope (mediator, M) may mediate the relationship between hippocampal activation at baseline (predictor, X) and CVLT slope (outcome, Y), with age, sex, and education as covariates. Step 1 tested the effect of X on Y, not including M. Step 2 tested X predicting M, and Steps 3 and 4 tested M|X significant predictor of Y and X|M not a significant predictor of Y. In Step 1 of the model, the regression of hippocampal activation on memory performance, ignoring PIB slope, was not significant [β = 0.005, t = 0.473, p=0.641]. In Step 2, the regression of hippocampal activation on PIB slope was significant [β = 0.0005, t = 2.26, p=0.034]. In Step 3, the regression of PIB slope, controlling for hippocampal activation, on memory performance was significant [β = −25.75, t = −2.91, p=0.008]. In Step 4, when controlling for PIB slope, hippocampal activation was not a significant predictor of memory performance [β = 0.018, t = 1.76, p=0.094]. There were no significant effects of age, sex, or education. The bootstrapped unstandardized indirect effect was −0.016 and significantly differed from zero, as revealed by a 95% bootstrap confidence interval that was entirely below zero (confidence interval: −0.031 to −0.002). Thus, longitudinal amyloid accumulation mediated the relationship between increased baseline hippocampal activation and greater memory decline over time (Figure 4).

Figure 4

Download asset Open asset

Since prior studies have noted increased Aβ accumulation over time in those with elevated baseline Aβ (Villemagne et al., 2011; Sojkova et al., 2011) and because we previously found a relationship between baseline Aβ and activation (Mormino et al., 2012), we analyzed the relationship between baseline PIB, hippocampal activation, and PIB slope to ensure that our effects on longitudinal Aβ accumulation were not being driven solely by elevated baseline Aβ. To determine whether increased baseline amyloid levels led to increased baseline hippocampal activation, which led to greater longitudinal Aβ accumulation, we conducted a mediation analysis, with age, sex, and education as covariates. We found no evidence of mediation, as revealed by a 95% bootstrap confidence interval that included zero (confidence interval: −0.004 to 0.044). Thus, baseline amyloid levels did not lead to greater baseline hippocampal activation to influence longitudinal Aβ accumulation, suggesting the effect of hippocampal activation on Aβ accumulation does not solely rely on baseline amyloid levels. Furthermore, by including random intercepts into our linear mixed models, we accounted for initial differences in amyloid. We then conducted a moderation analysis using hierarchical linear regression to determine if baseline amyloid levels influenced Aβ accumulation, but depended on the level of hippocampal activation. The two predictors (hippocampal activation and baseline PIB) were first entered into the regression analysis to determine each predictor’s effect on PIB slope and then the interaction term was added. Results indicated that baseline hippocampal activation [b = −0.004, p=0.034] but not baseline PIB [b = 2.01, p=0.608] was associated with PIB slope and accounted for a significant amount of variance in PIB slope [R² = 0.524, F(6,20) = 3.67, p=0.013]. The interaction between hippocampal activation and baseline PIB explained a significant increase in variance in PIB slope [ΔR² = 0.15, F(1,20) = 6.32, p=0.021]. Thus, the influence of baseline amyloid levels on longitudinal Aβ accumulation depends on hippocampal activation at baseline. To further characterize the nature of this interaction, we used the Johnson-Neyman technique to identify at which point in the range of hippocampal activation contrast values where the effect of the predictor on the outcome transitions from being statistically significant to non-significant (Hayes, 2013). We found that when hippocampal activation was greater than or equal to a contrast value of 12.18, higher baseline amyloid led to greater Aβ accumulation. This suggests that while baseline amyloid levels do not directly lead to greater levels of hippocampal activation, baseline amyloid’s relationship with Aβ accumulation may depend on higher levels of hippocampal activation at baseline.

In this study, we used fMRI with PIB-PET imaging and neuropsychological testing to investigate the influence of memory-related hippocampal activation at baseline on longitudinal Aβ accumulation and cognitive decline in normal older adults. While the amyloid hypothesis places Aβ deposition as the earliest event in a pathological cascade, why Aβ accumulates in the first place is not known. Although Aβ can induce structural and functional alterations to synapses and synaptic plasticity (Selkoe, 2002; Almeida et al., 2005; Coleman and Yao, 2003), the reverse relationship also exists such that early synaptic activity can influence Aβ (Nitsch et al., 1993; Kamenetz et al., 2003; Cirrito et al., 2005; Bero et al., 2011; Brody et al., 2008). A recent mouse model of AD found that hyperactivity of hippocampal neurons precedes Aβ plaque formation, suggesting that hippocampal hyperactivity is one of the earliest dysfunctions in the pathophysiological cascade of AD (Busche and Konnerth, 2016). This indicates that a major target of Aβ itself, synaptic activation, can control Aβ levels.

These results support the hypothesis that increased hippocampal activation at baseline is associated with increased longitudinal Aβ accumulation across frontal, parietal, temporal, and cingulate cortices and that longitudinal Aβ accumulation mediates the effect of baseline hippocampal activation on memory decline. Although previous studies have shown that baseline PIB is associated with the rate of longitudinal amyloid accumulation (Villemagne et al., 2011; Sojkova et al., 2011), our results are not simply dependent on baseline PIB for several reasons. First, we included random slopes and random intercepts in our linear mixed models to account for baseline PIB and different PIB slopes for each participant. Second, we performed a mediation analysis to determine if baseline PIB acted through baseline hippocampal activation to lead to increased longitudinal Aβ accumulation, which was revealed not to be the case. Finally, a moderation analysis showed that baseline PIB predicted increased longitudinal Aβ accumulation, but only when baseline hippocampal activation was high. These data suggest that while baseline amyloid may be linked to increased Aβ accumulation, this did not drive the relationships between baseline hippocampal activity and longitudinal Aβ accumulation.

Previous studies have described increased brain activity in those at risk for AD in different regions: the dentate gyrus/CA3 region of the hippocampus (Bakker et al., 2012; Yassa, 2011), the entire hippocampus (Mormino et al., 2012; Dickerson et al., 2005), the MTL more broadly (Dickerson et al., 2004), or neocortex (Mormino et al., 2012; Elman et al., 2014; Jones et al., 2016; Myers et al., 2014). These studies suggest that increased activation during memory performance is linked to risk for AD in older adults with memory impairment, MCI, ApoE4+, or those who are Aβ+. Our data support a role for task-related brain activation in the hippocampus that is linked to global Aβ deposition. Indeed, we found that only hippocampal activation at baseline and not non-hippocampal cortical activation was linked to increased longitudinal Aβ accumulation, suggesting regional specificity in these relationships. The regional pattern of relationships between cortical activation, hippocampal activation, and AD-related pathology may appear perplexing. Cross-sectional neuropathological data suggest that Aβ is first deposited in the cortex and then spreads to the MTL (Thal et al., 2002), while tau deposition in the form of neurofibrillary tangles deposit in the MTL initially with later spread to neocortex (Braak and Braak, 1995). We find that activation in the hippocampus predicts cortical Aβ while cortical activation does not. There are a few possible explanations for this. Hippocampal hyperactivity may be more indicative of pathological effects in those at risk for AD, while cortical activation may reflect other processes. For example, we previously reported that increased cortical activity, but not hippocampal activity appeared to be compensatory during memory encoding (Elman et al., 2014). This is consistent with data associating increased hippocampal activation with memory deficits (Yassa et al., 2011), and with the findings that reversing hippocampal hyperactivity rescues memory deficits (Bakker et al., 2012, 2015). Another possibility is that increased hippocampal activation is driving neuronal activity elsewhere, in cortical regions such as retrosplenial cortex and precuneus with strong hippocampal connectivity and early Aβ deposition. Another possibility is that hippocampal activation is driving tau deposition in the MTL, which then leads to Aβ deposition. In fact, preclinical data suggest a crucial role for tau accumulation in producing aberrant neural hyperactivity (Roberson et al., 2011). These possibilities are not mutually exclusive and will be important to examine in the future.

This study has several limitations. Not all subjects had the same number of PIB scans (each subject had two or three scans), which may influence the accuracy of PIB slope, however, linear mixed models are equipped to handle missing data. We found that there was no significant difference in PIB slope between participants with two or three scans or a relationship between years followed and PIB slope (p’s > 0.05). Sample size was relatively small and we did not have a large enough number of subjects to fully examine sex or ApoE4 carrier differences adequately. Absence of association may be due to lack of power rather than absence of a relationship, such as the absence of ApoE4 effect and absence of mediation through baseline Aβ. However, to determine if the effects of hippocampal activity on amyloid accumulation was driven by ApoE4 carriers, we removed them from the sample and re-ran the linear mixed model only in ApoE4 non-carriers. We found the interaction between hippocampal activation and amyloid accumulation remained significant [p = 0.018]. Thus, it appears the effect of hippocampal activation on increased amyloid accumulation is not specific to those who are ApoE4 carriers, but is also occurring in ApoE4 non-carriers that are also accumulating amyloid.

Another limitation of the study ignores the role of tau in producing these effects. As tau PET imaging is still relatively novel, future studies can incorporate tau into these models to determine if hippocampal activation at baseline influences longitudinal Aβ accumulation alone or if it influences how Aβ and tau interact during the progression of AD. It will also be important to include longitudinal fMRI data in addition to longitudinal PIB-PET data to measure how longitudinal change in hippocampal activation tracks longitudinal Aβ accumulation. One would expect that as hippocampal activation decreases over time (Celone et al., 2006; O'Brien et al., 2004), increases in Aβ accumulation will taper off, as this would indicate later stages of AD. In fact, data suggest that late in disease stage, Aβ deposition slows or even reverses, consistent with the idea that declining brain activity affects late-stage Aβ deposition (Villemagne et al., 2011; Jack et al., 2013). While we find that baseline amyloid levels did not drive our findings of hippocampal activation in predicting Aβ accumulation, it will be important for future studies to track those with little baseline amyloid deposition to determine if they eventually display a pattern of increased hippocampal activity followed by increased amyloid deposition.

1. 1. Almeida CG
   2. Tampellini D
   3. Takahashi RH
   4. Greengard P
   5. Lin MT
   6. Snyder EM
   7. Gouras GK

   (2005) Beta-amyloid accumulation in APP mutant neurons reduces PSD-95 and GluR1 in synapses

   Neurobiology of Disease 20:187–198.

   https://doi.org/10.1016/j.nbd.2005.02.008

   • PubMed
   • Google Scholar
2. 1. Bakker A
   2. Albert MS
   3. Krauss G
   4. Speck CL
   5. Gallagher M

   (2015) Response of the medial temporal lobe network in amnestic mild cognitive impairment to therapeutic intervention assessed by fMRI and memory task performance

   NeuroImage: Clinical 7:688–698.

   https://doi.org/10.1016/j.nicl.2015.02.009

   • PubMed
   • Google Scholar
3. 1. Bakker A
   2. Krauss GL
   3. Albert MS
   4. Speck CL
   5. Jones LR
   6. Stark CE
   7. Yassa MA
   8. Bassett SS
   9. Shelton AL
   10. Gallagher M

   (2012) Reduction of hippocampal hyperactivity improves cognition in amnestic mild cognitive impairment

   Neuron 74:467–474.

   https://doi.org/10.1016/j.neuron.2012.03.023

   • PubMed
   • Google Scholar
4. 1. Bero AW
   2. Yan P
   3. Roh JH
   4. Cirrito JR
   5. Stewart FR
   6. Raichle ME
   7. Lee JM
   8. Holtzman DM

   (2011) Neuronal activity regulates the regional vulnerability to amyloid-β deposition

   Nature Neuroscience 14:750–756.

   https://doi.org/10.1038/nn.2801

   • PubMed
   • Google Scholar
5. 1. Bookheimer SY
   2. Strojwas MH
   3. Cohen MS
   4. Saunders AM
   5. Pericak-Vance MA
   6. Mazziotta JC
   7. Small GW

   (2000) Patterns of brain activation in people at risk for alzheimer's disease

   New England Journal of Medicine 343:450–456.

   https://doi.org/10.1056/NEJM200008173430701

   • PubMed
   • Google Scholar
6. 1. Braak H
   2. Braak E

   (1995) Staging of Alzheimer's disease-related neurofibrillary changes

   Neurobiology of Aging 16:271–278.

   https://doi.org/10.1016/0197-4580(95)00021-6

   • PubMed
   • Google Scholar
7. 1. Brody DL
   2. Magnoni S
   3. Schwetye KE
   4. Spinner ML
   5. Esparza TJ
   6. Stocchetti N
   7. Zipfel GJ
   8. Holtzman DM

   (2008) Amyloid-beta dynamics correlate with neurological status in the injured human brain

   Science 321:1221–1224.

   https://doi.org/10.1126/science.1161591

   • PubMed
   • Google Scholar
8. 1. Buckner RL
   2. Sepulcre J
   3. Talukdar T
   4. Krienen FM
   5. Liu H
   6. Hedden T
   7. Andrews-Hanna JR
   8. Sperling RA
   9. Johnson KA

   (2009) Cortical hubs revealed by intrinsic functional connectivity: mapping, assessment of stability, and relation to Alzheimer's disease

   Journal of Neuroscience 29:1860–1873.

   https://doi.org/10.1523/JNEUROSCI.5062-08.2009

   • PubMed
   • Google Scholar
9. 1. Busche MA
   2. Chen X
   3. Henning HA
   4. Reichwald J
   5. Staufenbiel M
   6. Sakmann B
   7. Konnerth A

   (2012) Critical role of soluble amyloid-β for early hippocampal hyperactivity in a mouse model of Alzheimer’s disease

   PNAS 109:8740–8745.

   https://doi.org/10.1073/pnas.1206171109

   • PubMed
   • Google Scholar
10. 1. Busche MA
    2. Eichhoff G
    3. Adelsberger H
    4. Abramowski D
    5. Wiederhold KH
    6. Haass C
    7. Staufenbiel M
    8. Konnerth A
    9. Garaschuk O

    (2008) Clusters of hyperactive neurons near amyloid plaques in a mouse model of Alzheimer’s disease

    Science 321:1686–1689.

    https://doi.org/10.1126/science.1162844

    • PubMed
    • Google Scholar
11. 1. Busche MA
    2. Konnerth A

    (2016) Impairments of neural circuit function in Alzheimer's disease

    Philosophical Transactions of the Royal Society B: Biological Sciences 371:20150429.

    https://doi.org/10.1098/rstb.2015.0429

    • Google Scholar
12. 1. Celone KA
    2. Calhoun VD
    3. Dickerson BC
    4. Atri A
    5. Chua EF
    6. Miller SL
    7. DePeau K
    8. Rentz DM
    9. Selkoe DJ
    10. Blacker D
    11. Albert MS
    12. Sperling RA

    (2006) Alterations in memory networks in mild cognitive impairment and Alzheimer's disease: an independent component analysis

    Journal of Neuroscience 26:10222–10231.

    https://doi.org/10.1523/JNEUROSCI.2250-06.2006

    • PubMed
    • Google Scholar
13. 1. Cirrito JR
    2. Yamada KA
    3. Finn MB
    4. Sloviter RS
    5. Bales KR
    6. May PC
    7. Schoepp DD
    8. Paul SM
    9. Mennerick S
    10. Holtzman DM

    (2005) Synaptic activity regulates interstitial fluid amyloid-beta levels in vivo

    Neuron 48:913–922.

    https://doi.org/10.1016/j.neuron.2005.10.028

    • PubMed
    • Google Scholar
14. 1. Coleman PD
    2. Yao PJ

    (2003) Synaptic slaughter in Alzheimer's disease

    Neurobiology of Aging 24:1023–1027.

    https://doi.org/10.1016/j.neurobiolaging.2003.09.001

    • PubMed
    • Google Scholar
15. 1. Delis DC
    2. Massman PJ
    3. Butters N
    4. Salmon DP

    (1991) Profiles of demented and amnesic patients on the California verbal learning test: implications for the assessment of memory disorders

    Psychological Assessment 3:19–26.

    https://doi.org/10.1037/1040-3590.3.1.19

    • Google Scholar
16. 1. Dennis NA
    2. Browndyke JN
    3. Stokes J
    4. Need A
    5. Burke JR
    6. Welsh-Bohmer KA
    7. Cabeza R

    (2010) Temporal lobe functional activity and connectivity in young adult APOE varepsilon4 carriers

    Alzheimer's and Dementia 6:303–311.

    https://doi.org/10.1016/j.jalz.2009.07.003

    • PubMed
    • Google Scholar
17. 1. Dickerson BC
    2. Salat DH
    3. Bates JF
    4. Atiya M
    5. Killiany RJ
    6. Greve DN
    7. Dale AM
    8. Stern CE
    9. Blacker D
    10. Albert MS
    11. Sperling RA

    (2004) Medial temporal lobe function and structure in mild cognitive impairment

    Annals of Neurology 56:27–35.

    https://doi.org/10.1002/ana.20163

    • PubMed
    • Google Scholar
18. 1. Dickerson BC
    2. Salat DH
    3. Greve DN
    4. Chua EF
    5. Rand-Giovannetti E
    6. Rentz DM
    7. Bertram L
    8. Mullin K
    9. Tanzi RE
    10. Blacker D
    11. Albert MS
    12. Sperling RA

    (2005) Increased hippocampal activation in mild cognitive impairment compared to normal aging and AD

    Neurology 65:404–411.

    https://doi.org/10.1212/01.wnl.0000171450.97464.49

    • PubMed
    • Google Scholar
19. 1. El-Hayek YH
    2. Wu C
    3. Ye H
    4. Wang J
    5. Carlen PL
    6. Zhang L

    (2013) Hippocampal excitability is increased in aged mice

    Experimental Neurology 247:710–719.

    https://doi.org/10.1016/j.expneurol.2013.03.012

    • PubMed
    • Google Scholar
20. 1. Elman JA
    2. Oh H
    3. Madison CM
    4. Baker SL
    5. Vogel JW
    6. Marks SM
    7. Crowley S
    8. O'Neil JP
    9. Jagust WJ

    (2014) Neural compensation in older people with brain amyloid-β deposition

    Nature Neuroscience 17:1316–1318.

    https://doi.org/10.1038/nn.3806

    • PubMed
    • Google Scholar
21. 1. Haberman RP
    2. Colantuoni C
    3. Koh MT
    4. Gallagher M

    (2013) Behaviorally activated mRNA expression profiles produce signatures of learning and enhanced inhibition in aged rats with preserved memory

    PLoS One 8:e83674.

    https://doi.org/10.1371/journal.pone.0083674

    • PubMed
    • Google Scholar
22. 1. Hardy J

    (2009) The amyloid hypothesis for Alzheimer’s disease: a critical reappraisal

    Journal of Neurochemistry 110:1129–1134.

    https://doi.org/10.1111/j.1471-4159.2009.06181.x

    • PubMed
    • Google Scholar
23. Book

    1. Hayes A

    (2013)

    Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach

    Guilford Press.

    • Google Scholar
24. 1. Huijbers W
    2. Mormino EC
    3. Schultz AP
    4. Wigman S
    5. Ward AM
    6. Larvie M
    7. Amariglio RE
    8. Marshall GA
    9. Rentz DM
    10. Johnson KA
    11. Sperling RA

    (2015) Amyloid-β deposition in mild cognitive impairment is associated with increased hippocampal activity, atrophy and clinical progression

    Brain 138:1023–1035.

    https://doi.org/10.1093/brain/awv007

    • PubMed
    • Google Scholar
25. 1. Jack CR
    2. Wiste HJ
    3. Lesnick TG
    4. Weigand SD
    5. Knopman DS
    6. Vemuri P
    7. Pankratz VS
    8. Senjem ML
    9. Gunter JL
    10. Mielke MM
    11. Lowe VJ
    12. Boeve BF
    13. Petersen RC

    (2013) Brain β-amyloid load approaches a plateau

    Neurology 80:890–896.

    https://doi.org/10.1212/WNL.0b013e3182840bbe

    • PubMed
    • Google Scholar
26. 1. Jagust WJ
    2. Mormino EC

    (2011) Lifespan brain activity, β-amyloid, and Alzheimer's disease

    Trends in Cognitive Sciences 15:520–526.

    https://doi.org/10.1016/j.tics.2011.09.004

    • PubMed
    • Google Scholar
27. 1. Jones DT
    2. Knopman DS
    3. Gunter JL
    4. Graff-Radford J
    5. Vemuri P
    6. Boeve BF
    7. Petersen RC
    8. Weiner MW
    9. Jack CR
    10. Alzheimer’s Disease Neuroimaging Initiative

    (2016) Cascading network failure across the Alzheimer’s disease spectrum

    Brain 139:547–562.

    https://doi.org/10.1093/brain/awv338

    • PubMed
    • Google Scholar
28. 1. Kamenetz F
    2. Tomita T
    3. Hsieh H
    4. Seabrook G
    5. Borchelt D
    6. Iwatsubo T
    7. Sisodia S
    8. Malinow R

    (2003) APP processing and synaptic function

    Neuron 37:925–937.

    https://doi.org/10.1016/S0896-6273(03)00124-7

    • PubMed
    • Google Scholar
29. 1. Koh MT
    2. Haberman RP
    3. Foti S
    4. McCown TJ
    5. Gallagher M

    (2010) Treatment strategies targeting excess hippocampal activity benefit aged rats with cognitive impairment

    Neuropsychopharmacology 35:1016–1025.

    https://doi.org/10.1038/npp.2009.207

    • PubMed
    • Google Scholar
30. 1. Lange KL
    2. Bondi MW
    3. Salmon DP
    4. Galasko D
    5. Delis DC
    6. Thomas RG
    7. Thal LJ

    (2002) Decline in verbal memory during preclinical Alzheimer's disease: examination of the effect of APOE genotype

    Journal of the International Neuropsychological Society 8:943–955.

    https://doi.org/10.1017/S1355617702870096

    • PubMed
    • Google Scholar
31. 1. Mathis CA
    2. Wang Y
    3. Holt DP
    4. Huang GF
    5. Debnath ML
    6. Klunk WE

    (2003) Synthesis and evaluation of 11C-labeled 6-substituted 2-arylbenzothiazoles as amyloid imaging agents

    Journal of Medicinal Chemistry 46:2740–2754.

    https://doi.org/10.1021/jm030026b

    • PubMed
    • Google Scholar
32. 1. Mormino EC
    2. Brandel MG
    3. Madison CM
    4. Marks S
    5. Baker SL
    6. Jagust WJ

    (2012) Aβ deposition in aging is associated with increases in brain activation during successful memory encoding

    Cerebral Cortex 22:1813–1823.

    https://doi.org/10.1093/cercor/bhr255

    • PubMed
    • Google Scholar
33. 1. Myers N
    2. Pasquini L
    3. Göttler J
    4. Grimmer T
    5. Koch K
    6. Ortner M
    7. Neitzel J
    8. Mühlau M
    9. Förster S
    10. Kurz A
    11. Förstl H
    12. Zimmer C
    13. Wohlschläger AM
    14. Riedl V
    15. Drzezga A
    16. Sorg C

    (2014) Within-patient correspondence of amyloid-β and intrinsic network connectivity in Alzheimer's disease

    Brain 137:2052–2064.

    https://doi.org/10.1093/brain/awu103

    • PubMed
    • Google Scholar
34. 1. Nitsch RM
    2. Farber SA
    3. Growdon JH
    4. Wurtman RJ

    (1993) Release of amyloid beta-protein precursor derivatives by electrical depolarization of rat hippocampal slices

    PNAS 90:5191–5193.

    https://doi.org/10.1073/pnas.90.11.5191

    • PubMed
    • Google Scholar
35. 1. O'Brien JL
    2. O'Keefe KM
    3. LaViolette PS
    4. DeLuca AN
    5. Blacker D
    6. Dickerson BC
    7. Sperling RA

    (2010) Longitudinal fMRI in elderly reveals loss of hippocampal activation with clinical decline

    Neurology 74:1969–1976.

    https://doi.org/10.1212/WNL.0b013e3181e3966e

    • PubMed
    • Google Scholar
36. 1. O'Brien JT
    2. Lloyd A
    3. McKeith I
    4. Gholkar A
    5. Ferrier N

    (2004) A longitudinal study of hippocampal volume, cortisol levels, and cognition in older depressed subjects

    American Journal of Psychiatry 161:2081–2090.

    https://doi.org/10.1176/appi.ajp.161.11.2081

    • PubMed
    • Google Scholar
37. 1. Palop JJ
    2. Chin J
    3. Roberson ED
    4. Wang J
    5. Thwin MT
    6. Bien-Ly N
    7. Yoo J
    8. Ho KO
    9. Yu GQ
    10. Kreitzer A
    11. Finkbeiner S
    12. Noebels JL
    13. Mucke L

    (2007) Aberrant excitatory neuronal activity and compensatory remodeling of inhibitory hippocampal circuits in mouse models of Alzheimer's disease

    Neuron 55:697–711.

    https://doi.org/10.1016/j.neuron.2007.07.025

    • PubMed
    • Google Scholar
38. 1. Roberson ED
    2. Halabisky B
    3. Yoo JW
    4. Yao J
    5. Chin J
    6. Yan F
    7. Wu T
    8. Hamto P
    9. Devidze N
    10. Yu GQ
    11. Palop JJ
    12. Noebels JL
    13. Mucke L

    (2011) Amyloid-β/Fyn-induced synaptic, network, and cognitive impairments depend on tau levels in multiple mouse models of Alzheimer's disease

    Journal of Neuroscience 31:700–711.

    https://doi.org/10.1523/JNEUROSCI.4152-10.2011

    • PubMed
    • Google Scholar
39. 1. Sanchez PE
    2. Zhu L
    3. Verret L
    4. Vossel KA
    5. Orr AG
    6. Cirrito JR
    7. Devidze N
    8. Ho K
    9. Yu GQ
    10. Palop JJ
    11. Mucke L

    (2012) Levetiracetam suppresses neuronal network dysfunction and reverses synaptic and cognitive deficits in an Alzheimer's disease model

    PNAS 109:E2895–E2903.

    https://doi.org/10.1073/pnas.1121081109

    • PubMed
    • Google Scholar
40. 1. Selkoe DJ

    (2002) Alzheimer's disease is a synaptic failure

    Science 298:789–791.

    https://doi.org/10.1126/science.1074069

    • PubMed
    • Google Scholar
41. 1. Sojkova J
    2. Zhou Y
    3. An Y
    4. Kraut MA
    5. Ferrucci L
    6. Wong DF
    7. Resnick SM

    (2011) Longitudinal patterns of β-amyloid deposition in nondemented older adults

    Archives of Neurology 68:644–649.

    https://doi.org/10.1001/archneurol.2011.77

    • PubMed
    • Google Scholar
42. 1. Sperling RA
    2. Laviolette PS
    3. O'Keefe K
    4. O'Brien J
    5. Rentz DM
    6. Pihlajamaki M
    7. Marshall G
    8. Hyman BT
    9. Selkoe DJ
    10. Hedden T
    11. Buckner RL
    12. Becker JA
    13. Johnson KA

    (2009) Amyloid deposition is associated with impaired default network function in older persons without dementia

    Neuron 63:178–188.

    https://doi.org/10.1016/j.neuron.2009.07.003

    • PubMed
    • Google Scholar
43. 1. Thal DR
    2. Rüb U
    3. Orantes M
    4. Braak H

    (2002) Phases of A beta-deposition in the human brain and its relevance for the development of AD

    Neurology 58:1791–1800.

    https://doi.org/10.1212/WNL.58.12.1791

    • PubMed
    • Google Scholar
44. 1. Vannini P
    2. Hedden T
    3. Becker JA
    4. Sullivan C
    5. Putcha D
    6. Rentz D
    7. Johnson KA
    8. Sperling RA

    (2012) Age and amyloid-related alterations in default network habituation to stimulus repetition

    Neurobiology of Aging 33:1237–1252.

    https://doi.org/10.1016/j.neurobiolaging.2011.01.003

    • PubMed
    • Google Scholar
45. 1. Villemagne VL
    2. Pike KE
    3. Chételat G
    4. Ellis KA
    5. Mulligan RS
    6. Bourgeat P
    7. Ackermann U
    8. Jones G
    9. Szoeke C
    10. Salvado O
    11. Martins R
    12. O'Keefe G
    13. Mathis CA
    14. Klunk WE
    15. Ames D
    16. Masters CL
    17. Rowe CC

    (2011) Longitudinal assessment of Aβ and cognition in aging and Alzheimer disease

    Annals of Neurology 69:181–192.

    https://doi.org/10.1002/ana.22248

    • PubMed
    • Google Scholar
46. 1. Wilson IA
    2. Ikonen S
    3. Gallagher M
    4. Eichenbaum H
    5. Tanila H

    (2005) Age-associated alterations of hippocampal place cells are subregion specific

    Journal of Neuroscience 25:6877–6886.

    https://doi.org/10.1523/JNEUROSCI.1744-05.2005

    • PubMed
    • Google Scholar
47. 1. Yassa MA
    2. Lacy JW
    3. Stark SM
    4. Albert MS
    5. Gallagher M
    6. Stark CE

    (2011) Pattern separation deficits associated with increased hippocampal CA3 and dentate gyrus activity in nondemented older adults

    Hippocampus 21:968–979.

    https://doi.org/10.1002/hipo.20808

    • PubMed
    • Google Scholar

Author details

1. Stephanie L Leal

   Helen Wills Neuroscience Institute, University of California, Berkeley, United States

   Contribution

   SLL, Conceptualization, Formal analysis, Writing—original draft, Writing—review and editing

   For correspondence

   stephanieleal@berkeley.edu

   Competing interests

   The authors declare that no competing interests exist.

   "This ORCID iD identifies the author of this article:" 0000-0002-8082-8291

2. Susan M Landau

   1. Helen Wills Neuroscience Institute, University of California, Berkeley, United States
   2. Molecular Biophysics and Integrated Bioimaging, Lawrence Berkeley National Laboratory, Berkeley, United States

   Contribution

   SML, Formal analysis, Writing—review and editing

   Competing interests

   The authors declare that no competing interests exist.

3. Rachel K Bell

   Helen Wills Neuroscience Institute, University of California, Berkeley, United States

   Contribution

   RKB, Data curation, Formal analysis, Writing—review and editing

   Competing interests

   The authors declare that no competing interests exist.

4. William J Jagust

   1. Helen Wills Neuroscience Institute, University of California, Berkeley, United States
   2. Molecular Biophysics and Integrated Bioimaging, Lawrence Berkeley National Laboratory, Berkeley, United States

   Contribution

   WJJ, Conceptualization, Formal analysis, Funding acquisition, Writing—review and editing

   For correspondence

   jagust@berkeley.edu

   Competing interests

   The authors declare that no competing interests exist.

• 3,496

  views

• 666

  downloads

• 89

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.