Alzheimer’s Disease (AD), the most prevalent form of dementia among the elderly (Mckhann et al., 1984), entails irreversible neurodegeneration. Progressive decline in cognitive abilities and memory deterioration are typically associated with AD spectrum disorders, with memory being the primary affected function, followed by decline in other cognitive domains (Dubois et al., 2016). Presently, more than 6.7 million elderly individuals in the United States are afflicted by dementia, a figure projected to double by 2060 (Alzheimer's Disease Facts and Figures, 2023), posing significant healthcare challenges. AD is a continuum of biological and clinical manifestations, with amnestic mild cognitive impairment (aMCI) as its precursor, characterized by memory impairment and declining cognitive abilities, increasing the risk of AD dementia (Sperling et al., 2011). The transition from aMCI to clinically diagnosed AD remains inadequately understood, impeding the development of effective therapies (Sun et al., 2018). Investigating the progression of AD and aMCI is crucial for unveiling underlying mechanisms and advancing early interventions, potentially enhancing patient outcomes and alleviating societal burdens.

Resting-state functional magnetic resonance imaging (rs-fMRI) is a state-of-the-art, non-invasive method for assessing spontaneous fluctuations in blood oxygen level-dependent (BOLD) signals, which serves as a proxy for neural activity. Rs-fMRI techniques such as amplitude of low-frequency fluctuations (ALFF) (Xi et al., 2013), regional homogeneity (ReHo) (Zhang et al., 2012), and seed-based functional connectivity (FC) (Wang et al., 2006) have been extensively employed to investigate functional alterations in individuals with AD and aMCI. Unlike measures that focus on local areas like ALFF, ReHo, and subnetwork-scale seed-based FC, degree centrality (DC) offers an evaluation of the overall connectivity strength of the entire human brain on a global connectome scale. By identifying nodes with the highest number of direct connections, DC determines the relative importance of nodes within the network (Zuo et al., 2012). It has been utilized in the pathological and physiological exploration of AD and aMCI. A study (Xiong et al., 2021) discovered that, compared to healthy controls (HC), patients with MCI exhibit diminished DC values in the left inferior temporal gyrus. Additionally, these DC values exhibited positive correlations with Mini-Mental State Examination (MMSE) scores and negative correlations with disease progression. Another investigation (Li et al., 2018) revealed that individuals with subjective cognitive decline (SMC) demonstrate increased DC in both hippocampi (HP) and the left fusiform gyrus, along with decreased DC in the inferior parietal gyrus. This implies that DC can elucidate the inherent connectivity disruption patterns in the whole-brain functional network of AD patients at the voxel level.

However, in addition to intra-hemispheric connections, the functional connections between cerebral hemispheres are equally critical. Previous studies have indicated that AD patients may experience deficiencies in interhemispheric information integration (Lakmache et al., 1998). Therefore, it is advisable to employ voxel-mirrored homotopic connectivity (VMHC), an analytical approach detecting alterations in functional connectivity between homotopic regions across hemispheres (Zuo et al., 2010b), to delve further into the connectivity of homotopic regions and illuminate the underlying mechanisms of AD and aMCI. A study (Liao et al., 2018) discovered that the AD group exhibits lower VMHC values in the anterior cingulate gyrus and medial prefrontal cortex compared to the aMCI group and HC group. Utilizing rs-fMRI analysis grounded in both DC and VMHC can assess both whole-brain functional connectivity and interhemispheric connectivity. Examining the interplay between these two metrics may offer a more profound comprehension of the pathological mechanisms of AD.

Moreover, most investigations focusing on brain function in AD spectrum patients have centered on the classical frequency range (0.01–0.08 Hz) of rs-fMRI. However, research has indicated that rs-fMRI signals within the cortical regions of the brain manifest distinct characteristics and sensitivities to specific brain activity patterns across diverse frequency ranges (Buzsáki and Draguhn, 2004; Zuo et al., 2010a). This may relate to different neural representations, and previous study (Yuen et al., 2019) have already identified high reproducibility of these frequency clusters, regardless of the sampling rate of rs-fMRI data. Numerous studies (Han et al., 2011; Liu et al., 2014) have disclosed that in AD and aMCI patients, ALFF within the slow-5 frequency band (0.01–0.027 Hz) exhibits greater sensitivity to alterations in the default mode network (DMN) in comparison to the slow-4 frequency band (0.027–0.073 Hz). Given this, we previously investigated the discriminative capacity of the ALFF index in different frequency bands for aMCI patients and discovered that the slow-5 frequency band may offer insights into the pathogenesis of AD and its distinct stages (Wang et al., 2021). Nevertheless, this study did not delve deeper into whether individuals with AD spectrum disorders also display frequency-dependent characteristics in whole-brain functional connectivity. Hence, it is imperative to differentiate between frequency bands when examining alterations in DC and VMHC results in aMCI and AD patients.

In this study, we amassed rs-fMRI data to scrutinize disparities in DC and VMHC among the AD, aMCI, and HC groups. We also explored the association between variations in DC and VMHC in patients and neurocognitive assessments. More importantly, our study aims to ascertain whether DC and VMHC exhibit frequency band dependence in functional connectivity alterations among AD spectrum patients. This is the first study to apply DC and VMHC to evaluate multi-frequency band functional connectivity changes in patients with the AD spectrum. Building upon previous research, our hypothesis posits that aMCI and AD patients will manifest substantial abnormalities in DC and VMHC values relative to HC, and that these variations may correlate with disease severity. Additionally, we postulate that DC and VMHC within the slow-5 frequency band will demonstrate greater sensitivity to changes in functional connectivity in AD spectrum patients, in comparison to the slow-4 frequency band and the classical frequency range.

To the best of our knowledge, this study represents the first application of DC and VMHC to assess FC changes in AD and aMCI patients across multiple frequency bands. This approach addresses a limitation observed in most studies in this field, which typically focus solely on the classical frequency bands. We have identified significant differences in DC/VMHC across various brain regions, including the right CN (slow-4), left CN (classical, slow-4), right mSFG (slow-5), left mSFG (classical), left PCu (slow-5), right Lobule VI (classical), right Lobule VIII (slow-5), left Lobule VIII (slow-5, slow-4), right Lobules IV, V (slow-5), PCG (classical, slow-5), and the Crus II (slow-5, slow-4) when comparing AD and aMCI patients to the control group. Furthermore, alterations in DC/VMHC in the bilateral CN, left PCu, right Lobule VI, left Lobule VIII, right Lobules IV, V, PCG, and the Crus II were found to be closely linked to the clinical characteristics of AD and aMCI patients. Most notably, we observed that the intergroup differences in the slow-5 frequency band were more pronounced than those in the classical and slow-4 frequency bands. These findings suggest that patients within the AD continuum exhibit abnormal FC changes in intrinsic brain activity, influenced by specific frequency bands.

DC, as a graph-based measure, reflects the number of direct connections of a given voxel within the entire brain connectivity matrix, thereby representing its centrality in the whole brain network. This method has been employed in research on various disorders, including AD (Wang et al., 2019), schizophrenia (Wheeler et al., 2015), Parkinson’s disease (Baggio et al., 2014), among others.

In this study, we observed a progressive decrease in the DC values of the bilateral CN and the left mSFG as the disease advanced from NC to aMCI and ultimately to AD. This trend indicates a continuous reduction in the direct connectivity of these regions within the entire brain as the disease progresses. The CN plays a crucial role in cognitive processes such as attention, planning, and the execution of complex goal-directed behavior (Cummings, 1995). It integrates primary inputs from the dorsolateral prefrontal cortex and the orbitofrontal cortex. Lesions in the CN can lead to deficits in executive control and cognitive processing speed. Previous research has linked progressive atrophy of the CN (Rombouts et al., 2000), as well as the deposition of tau and amyloid-beta protein (Braak and Braak, 1990), to AD. Our study provides whole-brain functional connectivity evidence to corroborate the significant role of the CN in the underlying mechanisms of AD. We speculate that as tau proteins and β-amyloid proteins continue to accumulate in the CN, the connectivity between the CN and the dorsolateral prefrontal cortex and orbitofrontal cortex decreases, leading to cognitive impairment symptoms associated with AD. The mSFG, located in the frontal lobe, is closely associated with cognitive functioning, information encoding and retrieval, emotional thinking, and perception (Jobson et al., 2021). It has strong connections with the anterior part of the prefrontal cortex and plays a critical role in human memory processing. Multiple studies (Yue et al., 2015; Cai et al., 2017) have consistently reported decreased FC in the mPFC of individuals with aMCI, particularly in connections with the hippocampus. As the disease progresses to AD, this reduction in functional connectivity becomes more pronounced (Gili et al., 2011). These prior studies lend support to our current research findings. The decrease in DC value in the mSFG indicates a reduced direct connectivity between this brain region and the anterior part of the prefrontal cortex and the hippocampus, which is believed to be associated with the weakened cognitive control, decreased intelligence, and lowered cognitive abilities in AD patients. Therefore, we speculate that the decreased neural activity in the frontal lobe region of AD patients may be related to the deterioration of memory function. Furthermore, when compared to the HC and aMCI groups, the AD group exhibited significantly decreased DC values in the left PCu and right mSFG. The PCu, situated on the medial surface of the parietal lobe, is associated with various higher-level cognitive functions, including episodic memory, self-referential information processing, and aspects of consciousness. In patients with AD, the PCu is recognized as the region with the most prominent tau pathology deposition and neuroinflammation (Veitch et al., 2019). FDG-PET imaging studies (Strom et al., 2022) have revealed early regional hypometabolism in the PCu of AD patients. Dynamic FC studies (Zhao et al., 2022) have shown significantly reduced connectivity strength in the PCu in patients with aMCI and AD compared to the normal control group. Our research findings align with these previously reported results. We believe that the PCu may be one of the earliest regions to exhibit tau pathology deposition, leading to reduced perfusion and metabolism in this area. Consequently, cognitive processes in both resting state and cognitive activities are impacted in AD patients. Additionally, we found that in patients with AD and aMCI, the DC of the left PCu and bilateral CN positively correlated with MMSE and MoCA scores, while the DC of the left mSFG positively correlated with MoCA scores. These correlations suggest that the DC values in these brain regions are associated with patients’ cognitive function and can serve as predictors of disease progression.

Compared to the HC and aMCI groups, the AD group exhibited significant increases in DC values in the right lobule VI, left lobule VIII, and right lobules IV, V. This indicates an elevated status and role of the cerebellum in the whole-brain network. In addition to its role in coordinating fine motor functions, the cerebellum is involved in regulating cognitive functions such as attention, language, executive control, and emotion (Yang et al., 2012). Multiple studies have highlighted the crucial role of the cerebellum in the progression and pathogenesis of AD. From normal cognition to MCI and AD, there is a gradual decrease in cerebellar gray matter volume. A study on MCI converters revealed a sustained decline in DC values in the cerebellum compared to HC (Hu et al., 2021). The hypothesis is that the reduction in cerebellar DC may contribute to the improvement of cognitive and motor functions in MCI patients. Our research findings align with this perspective. In this study, we observed a significant increase in DC values in the right cerebellar lobule VI, left cerebellar lobule VIII, and right cerebellar lobules IV and V in patients with AD. Furthermore, in both AD and aMCI patients, the DC values in these cerebellar regions were negatively correlated with MMSE and MoCA scores. The increased DC in the cerebellum may indicate a worsening of cognitive function in AD patients. We speculate that the elevated status and role of the cerebellum in the whole-brain network would inevitably consume more neural resources, leading to a reduction in direct connections with other brain regions that are more beneficial for maintaining cognitive functions, such as the CN and mSFG mentioned earlier. Interestingly, our study found that compared to the AD and HC groups, the aMCI group exhibited a significant decrease in DC values in the right lobule VIII. This could possibly be a compensatory mechanism for maintaining cognitive function in patients with aMCI.

Interhemispheric functional connectivity is a common characteristic of the brain’s intrinsic functional networks (Wang et al., 2006). VMHC measures the integration of information between the two hemispheres by calculating the connectivity between each voxel in one hemisphere and its mirrored counterpart in the other hemisphere. Significant patterns of disrupted VMHC have been observed in various disorders, including depression (Wang et al., 2013), schizophrenia (Hoptman et al., 2012), and autism spectrum disorder (Anderson et al., 2011).

In our study, the AD group exhibited significantly reduced VMHC values in the PCG compared to the HC and aMCI groups. This indicates impaired functional connectivity within the hemisphere in the mentioned region for AD patients. The PCG is a key node in the structural and functional networks of the human brain. It is involved in processing episodic memory, working memory, and short-term memory, playing an important role in the progression of AD. Multiple studies have reported reductions in ALFF, ReHo, and FC in the PCG of AD patients (Zhang et al., 2021; Liao et al., 2022). Our study further highlights impaired functional connectivity within the PCG between brain hemispheres, and this impairment is positively correlated with disease progression. Subsequent correlation analysis confirmed this finding, demonstrating a positive correlation between the VMHC of the PCG and MMSE scores in the patient group. Along with the previously mentioned mSFG and PCu, the PCG constitutes core brain regions within the default mode network (DMN). The DMN is the most stable resting-state network, and its activity during periods of rest is crucial for memory consolidation in humans. Therefore, it is believed to have potential relevance to the development of AD (Dennis and Thompson, 2014). Both the reduced direct functional connectivity of mSFG and PCu across the whole brain and the disrupted functional connectivity between the two cerebral hemispheres of PCG can lead to the disruption of the DMN. The aberrant resting-state brain function of the DMN is a primary characteristic of AD patients.

Additionally, in comparison to the HC and AD groups, the aMCI group exhibited an increase in VMHC values in the Crus II, and these VMHC values in the patient group were positively correlated with MMSE scores. Some studies (Zhang et al., 2020) suggest that Crus II is associated with the DMN and the fronto-parietal network (FPN), among other cognitive networks, and it plays a crucial role in cognitive representations. Therefore, the increase in the VMHC value of Crus II is considered a compensatory phenomenon in aMCI patients to maintain their cognitive function within the cerebellar region. Before developing into AD, aMCI participants may be able to recruit additional neural resources in the Crus II region that is less affected or unaffected by the disease to compensate for the disrupted inter-hemispheric connectivity in the PCG region. Another study (McLaren et al., 2012) also found increased cerebellar activity in mild AD patients, which was positively correlated with improved memory.

Brain oscillations encompass a wide range of frequencies. Zuo et al. (2010a) divided the low-frequency oscillations (LFO) frequency band into slow-5 (0.01–0.027 Hz), slow-4 (0.027–0.073 Hz), slow-3 (0.073–0.198 Hz), and slow-2 (0.198 Hz-0.25 Hz). Slow-4 and slow-5 oscillations, primarily detected in gray matter, fall within the classical frequency range (0.01–0.08 Hz) and reflect spontaneous brain activity. The slow-2 and slow-3 frequency bands are mainly confined to white matter and are associated with respiratory and cardiac signal contamination (Cordes et al., 2001). Consequently, most brain function studies are conducted in the classical frequency range. However, an increasing number of studies have found that intrinsic activity varies across different frequency bands and exhibits frequency-dependent alterations in various brain disorders, such as major depression (Wang et al., 2018), alcohol dependence (Guo et al., 2019) childhood epilepsy (Jiang et al., 2020), and trigeminal neuralgia (Ge et al., 2024).

Our study reveals that different brain regions exhibit both overlapping and distinct changes in the same index across different frequency bands, and to some extent, they complement each other. This is particularly evident in VMHC, where the slow-5 frequency band precisely encompasses all the brain regions in the classical and slow-4 frequency bands. However, for DC, each frequency band has its specific set of brain regions. For example, right Lobule VI and left mSFG were only detected in the classical frequency band. The specific brain regions associated with the slow-5 frequency band include the right Lobule VIII, right Lobule IV, V, right mSFG, and left PCu. The right CN was only detected in the slow-4 frequency band. Left Lobule VIII overlaps in the slow-5 and slow-4 frequency bands, while left CN overlaps in the classical and slow-4 frequency bands. These overlapping brain regions may play a more stable and important role in the pathogenesis of AD. In future research, we should focus on analyzing these brain regions. The differences in brain regions included in the slow-5 frequency band were the greatest. These findings are consistent with our previous research (Wang et al., 2021) and suggest that the brain functions exhibit specific frequency and engagement patterns. The intrinsic brain activity in different frequency bands may have specific pathological significance. The slow-4 and slow-5 signals are distinct entities with different roles, with the slow-5 frequency band potentially providing better exploration of the pathological mechanisms underlying AD.

Several limitations must be addressed in our study. First of all, the absence of a biological diagnosis of AD based on CSF/PET biomarkers is the greatest limitation. In the following research, we should enhance the biological marker data to achieve better diagnosis of patients within the AD continuum. Secondly, our investigation aims to understand the transition from NC to aMCI and eventually to AD. However, our study is cross-sectional, so the next step should involve longitudinal investigations to examine the changes in DC and VMHC during disease progression. Thirdly, noise from respiration and heartbeat during the scanning process can affect the low-frequency amplitudes in the resting state. Currently, we can only partially mitigate the confounding effects of these underlying factors. In future research, we can consider employing synchronized gating techniques to synchronize the MRI acquisition with specific phases of the cardiac or respiratory cycle. Finally, this study did not integrate functional metrics with structural metrics to further analyze and explore their correlations. Nevertheless, functional changes often have underlying structural bases. For instance, alterations in the volume of the corpus callosum mediate changes in interhemispheric connectivity. Therefore, future studies should integrate structural investigations in participants to gain a better understanding of the relationship between brain structure and functional changes.

This study has demonstrated that the simultaneous utilization of DC and VMHC metrics allows for a more comprehensive insight into the functional abnormalities in the underlying brain networks of AD spectrum patients. Furthermore, we observed that brain regions exhibiting significant variations in various parameters were most pronounced within the slow-5 frequency band. Therefore, we hypothesize that brain function demonstrates a certain degree of frequency specificity, and the slow-5 frequency band holds the potential to provide deeper insights into the pathological and physiological mechanisms of AD.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding authors.

This prospective study received approval from the local Ethics Committee of Zhejiang Provincial People's Hospital (Ethics Approval No. 2012KY002) and was conducted in accordance with the principles outlined in the Declaration of Helsinki. All participants were right-handed and provided written informed consent before participating in the study.

HH: Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. LW: Methodology, Validation, Writing – original draft, Writing – review & editing. SA: Methodology, Software, Writing – original draft. XT: Methodology, Validation, Writing – original draft. QF: Software, Visualization, Writing – original draft. YM: Software, Writing – original draft. XG: Methodology, Writing – original draft. ZL: Conceptualization, Validation, Writing – review & editing. ZD: Conceptualization, Validation, Writing – review & editing.