**3. The default mode network (DMN)**

The distribution of amyloid has been noted to overlap with the neural system related to the default mode of brain function [5]. The observation has prompted hypotheses about the relationship of neural activity, the default mode network, and AD that continue to evolve.

PET studies have shown a broad region of relative deactivation in resting states compared to active states with greatest deactivation within the ventromedial prefrontal cortex (VMPFC) with peak minimum at Brodmann area 10 (BA10; **Figure 1**) [40]. Additional regions showing deactivations localized bilaterally to the inferior and superior frontal cortex, PCC/precuneus, prefrontal cortex, inferior parietal cortex, and several temporal areas.

The deactivation in the VMPFC during rest was shown not to reflect a relative activation as the OEF did not change significantly from whole-brain (e.g., **Figure 1**;

#### **Figure 1.**

*Transverse sections (z = −6) of stereotactically normalized parametric data showing convergence across studies between CBF change and FC as well as relative stability of OEF. Left: Region of common relative CBF deactivation (VMPFC) in mega image contrast between active scans and passive scans (peak Z-score − 7.7; modified from [40]); middle: Default mode oxygen extraction fraction during rest with eyes closed; uniform OEF; no OEF increase in VMPFC [41]. Right: Resting state FC of the VMPFC with the DMN [49]. Middle* & *left: Copyright (2003) National Academy of Sciences, U.S.A.*

Raichle et al. [41]). If the VMPFC were activated, it should show increased OEF (as some regions in the visual cortices, see red color). The activated network during passive tasks was hypothesized instead to reflect a return to baseline or default mode network (DMN) that was interrupted by active tasks. The regional specificity of the DMN suggested some ill-defined brain function.

The DMN can be detected even in the presence of deep anesthesia suggesting a degree of invariance with respect to consciousness possibly reflecting intrinsic brain organization such as anatomical connectivity [42]. The VMPFC becomes more active not only at rest relative to other active task states (e.g., attentional) but also during a variety of other conditions where attention is directed away from the external environment. Such states include introspection, "mind wandering" [43], self-appraisal or introspection [44], stimulus independent thought [45], episodic future simulation, trait emotional self-awareness [46], and interoceptive tasks (e.g., recall of visceral information) [47]. As this book notes, these are reflections of the "autobiographical self."

Previous work in the mid to late 1990s observed physiological fluctuations occurring during active states and even during rest in the MR blood oxygen level dependent (BOLD) signal in humans and in CBF in rodents that were correlated at low frequency (0.1 Hz) across regions known to have anatomical and functional relatedness [48]. However, it remained unclear how these temporal signals related to the DMN as defined from PET studies.

Remarkably, when the VMPFC region in the DMN (showing relative deactivation during active tasks using PET) was used as a seed region to correlate BOLD signals throughout the brain, a network surface based on interregional temporal coherence of the BOLD MR signal that was visually superimposable on the DMN. Use of a spiral MR pulse sequence avoiding signal dropout in ventral prefrontal regions enabled good signal recovery within VMPFC (**Figure 1** [49]). Subsequent fMRI studies examining resting state networks displayed a more variable pattern in the medial prefrontal regions frequently showing more dorsal localization [50–52]. The significance of this disparity is unclear but may relate to low signal recovery with fMRI in ventral brain regions or other subnetworks (see below).

The analysis of BOLD data from the resting state using independent component analysis of FC identified numerous subnetworks showing both anatomical (known afferent and efferent anatomical projections) and functional (coactivation during tasks) architecture converging with other datasets (**Figure 2**) [53]. The four principal RSNs are the dorsal attention network (DAN), DMN (as above), salience network (SAL; also termed cingulo-opercular), and bilateral frontoparietal control network (FPC). The latter is sometimes segmented to dorsal and ventral systems [54].

The precise components of these RSNs need further refinement. For example, how does the "anterior medial prefrontal cortex" relate to VMPFC, pgACC, dACC, and medial superior frontal gyrus? What elementary cognitive operations do these different areas serve? Similarly, how does the "PCC/precuneus" relate to the various medial parietal subregions, and what functions do they serve? In this regard, preliminary dissection divides the PCC into dorsal and ventral regions, each with two additional subregions that in turn connect with other differing cortical projection regions [55].

These four major networks based on data averaged across individuals only hint at the complexity on a more fine-grained analysis. ICA can produce many more networks, and the Human Connectome Project points to 180 parcels per hemisphere [56]. When resting BOLD is collected over many hours for one individual, much more complex, interdigitated, parallel, distributed networks become apparent without the blurring caused by inter-subject averaging [57].

**19**

below) [60, 66].

**Figure 2.**

**4. Cerebral energetics**

*Fact, Fiction, or Evolution: Mechanism Hypothesis of Alzheimer's Disease*

Data on the effects of aging and amyloid/tau deposition on these subnetworks continue to accrue, but there appears differential vulnerability of different subnetworks. For example, amyloid decreased the FC of the DMN subnetworks relevant to episodic memory (PCC, angular gyrus, VMPFC) while increasing FC in dorsolateral and anterior medial prefrontal cortices as well as lateral temporal regions [58]. The latter regions were interpreted as reflecting compensatory responses to the amyloidrelated dysfunction in the memory subnetworks. Furthermore, the mere presence of biomarkers such as amyloid in preclinical AD can confound FC findings within and across RSNs. Cognitively normal subjects without vs. with amyloid positivity show small vs. large age-related changes in RSN functional, respectively [59].

*Four major RSNs revealed by ICA: DAN, dorsal attention network; DMN, default mode network; SN, salience network (also called cingulo-opercular network); right FPC, frontoparietal control network. Modified from [54].*

The DMN shares considerable connectivity with the hippocampus [60, 61]. The DMN couples with different sectors of the hippocampus [62] through the parahippocampal gyrus depending on task context during rest [63]; spontaneous, unconstrained thought (e.g., thinking about one's past or future [64]); episodic memory retrieval [65]; and associative episodic memory encoding [63]. This network becomes disrupted early occurring both in preclinical and early AD (see Section 6,

The energetic balance sheet indicates a large part of oxidative metabolism maintains the resting state [67–69]. The classic work of Seymour Kety showed oxygen consumption in the brain differed little across a wide variety of abnormal mental states such as in psychosis, whether in schizophrenic decompensation or acute drug intoxication [70]. In response to external stimulation, the brain only increases oxygen consumption by 5% [71]. Similarly, few differences in oxygen consumption occur between sleep vs. wakefulness [72]. Although the bulk of brain work at rest and on activation derives from oxidative phosphorylation [71], the metabolism hypothesis of AD focuses on a specific metabolic pathway: AG (i.e., glycolysis in the presence of adequate levels of oxygen; i.e., nonoxidative metabolism of glucose [73]). Regions high in oxidative phosphorylation do not necessarily have high rates of AG. As an example, the visual cortex has very high glucose metabolism; high cerebral blood flow (CBF); high oxidative metabolism (cerebral metabolic rate for oxygen) with high levels of cytochrome oxidase; but low AG [73]. The focus on AG follows from the visual cortex having relative resistance to amyloid deposition and being one of the regions showing little decrease in metabolism with aging. In contrast, the PCC has high flow, oxidative metabolism, glucose metabolism, and high

*DOI: http://dx.doi.org/10.5772/intechopen.83824*

*Fact, Fiction, or Evolution: Mechanism Hypothesis of Alzheimer's Disease DOI: http://dx.doi.org/10.5772/intechopen.83824*

**Figure 2.**

*Redirecting Alzheimer Strategy - Tracing Memory Loss to Self Pathology*

of the DMN suggested some ill-defined brain function.

"autobiographical self."

ventral systems [54].

tion regions [55].

to the DMN as defined from PET studies.

Raichle et al. [41]). If the VMPFC were activated, it should show increased OEF (as some regions in the visual cortices, see red color). The activated network during passive tasks was hypothesized instead to reflect a return to baseline or default mode network (DMN) that was interrupted by active tasks. The regional specificity

The DMN can be detected even in the presence of deep anesthesia suggesting a degree of invariance with respect to consciousness possibly reflecting intrinsic brain organization such as anatomical connectivity [42]. The VMPFC becomes more active not only at rest relative to other active task states (e.g., attentional) but also during a variety of other conditions where attention is directed away from the external environment. Such states include introspection, "mind wandering" [43], self-appraisal or introspection [44], stimulus independent thought [45], episodic future simulation, trait emotional self-awareness [46], and interoceptive tasks (e.g., recall of visceral information) [47]. As this book notes, these are reflections of the

Previous work in the mid to late 1990s observed physiological fluctuations occurring during active states and even during rest in the MR blood oxygen level dependent (BOLD) signal in humans and in CBF in rodents that were correlated at low frequency (0.1 Hz) across regions known to have anatomical and functional relatedness [48]. However, it remained unclear how these temporal signals related

Remarkably, when the VMPFC region in the DMN (showing relative deactivation during active tasks using PET) was used as a seed region to correlate BOLD signals throughout the brain, a network surface based on interregional temporal coherence of the BOLD MR signal that was visually superimposable on the

DMN. Use of a spiral MR pulse sequence avoiding signal dropout in ventral prefrontal regions enabled good signal recovery within VMPFC (**Figure 1** [49]). Subsequent fMRI studies examining resting state networks displayed a more variable pattern in the medial prefrontal regions frequently showing more dorsal localization [50–52]. The significance of this disparity is unclear but may relate to low signal recovery

The analysis of BOLD data from the resting state using independent component analysis of FC identified numerous subnetworks showing both anatomical (known afferent and efferent anatomical projections) and functional (coactivation during tasks) architecture converging with other datasets (**Figure 2**) [53]. The four principal RSNs are the dorsal attention network (DAN), DMN (as above), salience network (SAL; also termed cingulo-opercular), and bilateral frontoparietal control network (FPC). The latter is sometimes segmented to dorsal and

The precise components of these RSNs need further refinement. For example, how does the "anterior medial prefrontal cortex" relate to VMPFC, pgACC, dACC, and medial superior frontal gyrus? What elementary cognitive operations do these different areas serve? Similarly, how does the "PCC/precuneus" relate to the various medial parietal subregions, and what functions do they serve? In this regard, preliminary dissection divides the PCC into dorsal and ventral regions, each with two additional subregions that in turn connect with other differing cortical projec-

These four major networks based on data averaged across individuals only hint at the complexity on a more fine-grained analysis. ICA can produce many more networks, and the Human Connectome Project points to 180 parcels per hemisphere [56]. When resting BOLD is collected over many hours for one individual, much more complex, interdigitated, parallel, distributed networks become apparent

without the blurring caused by inter-subject averaging [57].

with fMRI in ventral brain regions or other subnetworks (see below).

**18**

*Four major RSNs revealed by ICA: DAN, dorsal attention network; DMN, default mode network; SN, salience network (also called cingulo-opercular network); right FPC, frontoparietal control network. Modified from [54].*

Data on the effects of aging and amyloid/tau deposition on these subnetworks continue to accrue, but there appears differential vulnerability of different subnetworks. For example, amyloid decreased the FC of the DMN subnetworks relevant to episodic memory (PCC, angular gyrus, VMPFC) while increasing FC in dorsolateral and anterior medial prefrontal cortices as well as lateral temporal regions [58]. The latter regions were interpreted as reflecting compensatory responses to the amyloidrelated dysfunction in the memory subnetworks. Furthermore, the mere presence of biomarkers such as amyloid in preclinical AD can confound FC findings within and across RSNs. Cognitively normal subjects without vs. with amyloid positivity show small vs. large age-related changes in RSN functional, respectively [59].

The DMN shares considerable connectivity with the hippocampus [60, 61]. The DMN couples with different sectors of the hippocampus [62] through the parahippocampal gyrus depending on task context during rest [63]; spontaneous, unconstrained thought (e.g., thinking about one's past or future [64]); episodic memory retrieval [65]; and associative episodic memory encoding [63]. This network becomes disrupted early occurring both in preclinical and early AD (see Section 6, below) [60, 66].

### **4. Cerebral energetics**

The energetic balance sheet indicates a large part of oxidative metabolism maintains the resting state [67–69]. The classic work of Seymour Kety showed oxygen consumption in the brain differed little across a wide variety of abnormal mental states such as in psychosis, whether in schizophrenic decompensation or acute drug intoxication [70]. In response to external stimulation, the brain only increases oxygen consumption by 5% [71]. Similarly, few differences in oxygen consumption occur between sleep vs. wakefulness [72]. Although the bulk of brain work at rest and on activation derives from oxidative phosphorylation [71], the metabolism hypothesis of AD focuses on a specific metabolic pathway: AG (i.e., glycolysis in the presence of adequate levels of oxygen; i.e., nonoxidative metabolism of glucose [73]).

Regions high in oxidative phosphorylation do not necessarily have high rates of AG. As an example, the visual cortex has very high glucose metabolism; high cerebral blood flow (CBF); high oxidative metabolism (cerebral metabolic rate for oxygen) with high levels of cytochrome oxidase; but low AG [73]. The focus on AG follows from the visual cortex having relative resistance to amyloid deposition and being one of the regions showing little decrease in metabolism with aging. In contrast, the PCC has high flow, oxidative metabolism, glucose metabolism, and high

AG [73]. As summarized above, the PCC is very susceptible to amyloid deposition and is among the earliest dysfunctional regions in AD. These observations further refined the metabolism hypothesis of AD.

In a group of mostly cognitively intact elders, those globally without amyloid did not have tau accumulation in areas prone toward tau deposits (precuneus, amygdala, entorhinal, inferior temporal, inferior and superior parietal, fusiform, and lateral occipital cortices) and did not have decreased CMRO or AG [69]. They showed a positive correlation between AG and CMRglu; no correlations surfaced between CMRO, CMRglu, or tau deposition. In contrast, those who were amyloid positive globally showed an inverse relationship between tau and AG but not between tau and CMRO or CMRglu. These data suggest the loss of AG in tau-prone regions with tau accumulation leads to decreased plasticity and decreased neuroprotection (i.e., decreased redox buffering) leading to accelerated tauopathy.
