Abstract
The entorhinal-hippocampal circuit is a strategic hub for memory but also the first site to be affected in the Alzheimer’s Disease (AD)-related pathology. We investigated MRI patterns of brain atrophy and functional connectivity in a study cohort obtained from the Alzheimer’s Disease Neuroimaging Initiative database including healthy control (HC), Mild Cognitive Impairment (MCI), and AD subjects. MCI individuals were clinically evaluated 24 months after the MRI scan, and the group further divided into a subset of subjects who either did (c-MCI) or did not (nc-MCI) convert to AD. Compared to HC subjects, AD patients exhibited a collapse of long-range connectivity from the hippocampus and entorhinal cortex, pronounced cortical/sub-cortical atrophy, and a dramatic decline in cognitive performances. c-MCI patients showed entorhinal and hippocampal hypo-connectivity, no signs of cortical thinning but evidence of right hippocampus atrophy. On the contrary, nc-MCI patients showed lack of brain atrophy, largely preserved cognitive functions, hippocampal and entorhinal hyper-connectivity with selected neocortical/sub-cortical regions mainly involved in memory processing and brain meta-stability. This hyper-connectivity can represent an early compensatory strategy to overcome the progression of cognitive impairment. This functional signature can also be employed for the diagnosis of c-MCI subjects.
Introduction
Brain aging and aging-related neurodegenerative disorders are a significant health challenge for contemporary societies. Brain aging represents a favorable background for the onset and development of neurodegeneration and dementia. Alzheimer’s Disease (AD) is a condition associated with the development of irreversible cognitive and behavioral deficits and preceded by a prodromal stage known as Mild Cognitive Impairment (MCI). MCI patients do not fulfill the diagnostic criteria for dementia but show significant cognitive deficits that mostly occur in mnemonic domains (Petersen et al. 2010). The MCI stage progresses to AD in 60-65% of cases (Busse et al. 2006) with a conversion rate that reaches 8.1% per year (Mitchell and Shiri-Feshki 2009). Thus, the early identification of the brain changes associated with MCI is critical to catch the disease at its initial stage, unravel the pathogenic mechanisms involved in AD and help the design of more effective therapeutic interventions.
Neuroimaging approaches have been extensively employed to detect the initial changes associated with the early stages of AD (Frisoni and Jessen 2018). Resting-state functional Magnetic Resonance Imaging (rs-fMRI) is a non-invasive tool that allows the investigation of the operational changes and network reconfigurations that occur in several neurological/neurodegenerative conditions including AD. In MCI patients, this technique has been successfully employed in the quest to detect abnormalities in the brain functional connectivity that occur before the appearance of patent signs of structural damage (Badhwar et al. 2017, Drzezga et al. 2011, Canuet et al. 2015).
The entorhinal-hippocampal circuit is a strategic region for the control of cognitive processes and the first site to be affected by the AD-related pathology (Braak et al. 2013, Gomez-Isla et al. 1996). In the AD brain, the early signs of synaptic degradation occur within the perforant path, the neurodegeneration then spreads to the layers II-III of the entorhinal cortex and the hippocampal CA3/DG regions, eventually reaches the subicular areas, and ultimately affects the whole hippocampus (Yassa et al. 2010). The entorhinal-hippocampal complex plays a critical role in the processing of long-term memory (Preston and Eichenbaum 2013). The region sustains the network brain stability and promotes the adaptive neuroplasticity that copes with the underlying pathological stressors that are triggering the structural damage (van den Heuvel and Sporns 2011, Hillary and Grafman 2017).
In this study, we investigated, in a cohort of one hundred thirty-five individuals, differences in structural MRI (sMRI) and rs-fMRI features that occurred within the cortico-hippocampal and cortico-entorhinal circuits. The study group included Healthy Control (HC) (n=40), MCI (n=67), and AD (n=28) subjects. The dataset also provided information on the demographic, neuropsychological/clinical, and APOE status as well as the CSF levels of AD-related pathogenic proteins like the amyloid β1–42 peptide (Aβ1–42), tau phosphorylated at threonine 181 (p-tau181), and the ratio of p-tau181/Aβ1–42.
sMRI data were employed to investigate differences in brain volume and cortical thickness among study participants. Rs-fMRI data were used to evaluate differences in the functional connectivity (FC) occurring in the circuits linking the hippocampus or the entorhinal cortex to the cortex. The progression or clinical stability of HC or MCI subjects was assessed by using clinical follow-up data obtained 24 months after the initial MRI session. With the help of these longitudinal data, the MCI group was therefore divided into two subsets: patients who converted (c-MCI) or did not convert (nc-MCI) to AD. Finally, the CSF data were plotted against the rs-fMRI results to explore correlations between the FC strength and levels of Aβ1–42 and p-tau181 as well as the p-tau181/ Aβ1–42 ratio.
The overall aim of the study was to disclose the contributions of the hippocampus and entorhinal cortex in ongoing neurodegenerative processes and the transition from different steps of the AD-related spectrum.
Materials and methods
Experimental design
Data employed for this article were obtained from the ADNI-GO/2 database. ADNI was launched in 2003 as a public-private partnership led by Michael W. Weiner. The primary goal of ADNI is to employ serial MRI, PET, biological markers, and clinical and neuropsychological data to investigate the features of patients affected by the AD spectrum. For up-to-date information on the initiative, see www.adni-info.org.
Experiments fulfilled the ethical standards and the Declaration of Helsinki (1997) and subsequent revisions. Informed consent was obtained from study participants or authorized representatives. Study participants had a good general health status and no diseases that are expected to interfere with the study. Overall, the ADNI-GO/2 database included one hundred seventy participants who have completed the 3T-sMRI and 3T-rs-fMRI and with an age range between 57 and 88 years old.
Participants who did not complete a clinical follow-up performed 24 months after the first MRI session or those who showed technical issues related to their MRI raw-data (i.e., artifacts, dis-homogeneity in acquisition parameters, images deformed for missing information raw-file) were excluded from the study sample (Supplementary Fig. 1). Our final sample included one hundred thirty-five participants divided into forty HC subjects, sixty-seven MCI patients, and twenty-eight AD patients. Based on the clinical follow-up, the MCI group was further subdivided into a group of fifty-four nc-MCI patients and thirteen c-MCI patients.
Neuropsychological assessment
All subjects underwent clinical and cognitive evaluations at the time of the MRI scan. The ADNI neuropsychological dataset includes the Mini-Mental State Examination (MMSE) (Folstein et al. 1975) and the Montreal Cognitive Assessment (MoCA) (Nasreddine et al. 2005) to investigate global cognition; the Functional Activities Questionnaire (FAQ) for the assessment of daily living activities (Pfeffer et al. 1982); the Alzheimer’s Disease Assessment Scale-Cognitive subscales (ADAS - 11 items scores; ADAS - 13 items scores) to evaluate the severity of impairments of memory, learning, language (production and comprehension), praxis, and orientation (Mohs and Cohen 1988; Mohs et al. 1997); the Animal Fluency (Morris et al. 1989) and the 30-item Boston Naming Test (BNT) (Kaplan, et al. 1983) to investigate semantic memory and language abilities; the Trail Making Test (TMT), part A and B (time to completion) to assess attention/executive functions (Spreen 1998); the Rey Auditory Verbal Learning Test (RAVLT) to investigate recall and recognition (Rey, 1964).
HC subjects were free of memory complaints and without significant impairment as far as general cognitive functions or daily living activities. The inclusion criteria for HC subjects were: MMSE scores between 24 and 30, a global score of 0 on the Clinical Dementia Rating Scale (CDR-RS; Morris JC 1993), and a score above the cutoff level on the Logical Memory II, subscale of the Wechsler Memory Scale-Revised (WMS-R; Wechsler, 1987) (≥ for 0-7 years of education, ≥ for 8-15 years, and ≥ for 16 or more years).
The inclusion criteria for MCI patients were: MMSE scores between 24 and 30, memory impairments identified by the partner with or without complaints by the participant, a CDR score of 0.5, and memory deficits as indicated by scores below the cutoff level on the WMS-R Logical Memory II (0-7 for years of education 3-6, 5-9 for 8-15 years, and for >15 years ≤ Their general cognition status and functional performances were sufficiently preserved to exclude a diagnosis of AD.
AD patients fulfilled the criteria of probable AD, set by the National Institute of Neurologic and Communicative Disorders and Stroke (NINCDS) as well as the Alzheimer’s Disease and Related Disorders Association (ADRDA). AD patients had MMSE scores between 20 and 26 and CDR-RS scores between 0.5 and 1.0.
CSF and APOE genotyping
CSF data were available for 87.4% of the total sample. The set included information on levels of the amyloid β1–42 peptide (Aβ1–42), total tau (t-tau), and tau phosphorylated at threonine 181 (p-tau181). Highly standardized Aβ1–42, t-tau and p-tau181 levels were measured using the Roche automated immunoassay platform (Cobas e601) and immunoassay reagents. Details on the methods for the acquisition and measurement of CSF are reported at the ADNI website (http://www.adni-info.org). The apolipoprotein E (APOE) ε allele frequency was also investigated at the screening stage.
MRI acquisition protocol
MR data were acquired with a Philips 3T scanner (see details at http://adni.loni.usc.edu/wp-content/uploads/2010/05/ADNI2_MRI_Training_Manual_FINAL.pdf). T1-weighted images were obtained using 3D Turbo Field-Echo sequences (TFE, Slice Thickness=1.2 mm; TR/TE=6.8/3.1 ms). One run of Resting-state Blood Oxygen Level Dependent (BOLD) fMRI data was acquired using gradient-echo T2*-weighted echo-planar (EPI) sequence (in-plane voxel size=3.3125 mm x 3.3125 mm, slice thickness 3.3125 mm, and TR/TE=3000/30 ms. Subjects were instructed to lay still and keep their eyes open during acquisition.
MRI data analysis
FreeSurfer (version 6.0) was employed to perform sMRI and rs-fMRI data analysis. For each study participants, T1-weighted images were analyzed using the “recon-all -all” command line to obtain automated reconstruction and labeling of cortical and subcortical regions (Fischl et al. 2004). The pre-processing steps encompassed magnetic field inhomogeneity correction, affine-registration to Talairach Atlas, intensity normalization and skull-strip. The processing steps involved segmentation of the subcortical white matter (WM) and deep grey matter (GM) volumetric regions, tessellation of the GM and WM matter boundary, automated topology correction, surface deformation following intensity gradients to optimally position the GM and WM and GM/cerebrospinal fluid borders at the location where the greatest shift in intensity delineates the transition to the other tissue class. The total volume of the left and right hippocampi and the estimated total intracranial volume (eTIV) were calculated using the “asegstats2table”. The hippocampal volumes were normalized by eTIV. The left and right masks of the hippocampi and entorhinal cortices were obtained by the “recon-all -all” command lines and used as “seed regions” for FC analysis using FreeSurfer - Functional Analysis Stream (http://surfer.nmr.mgh.harvard.edu/fswiki/FsFastFunctionalConnectivityWalkthrough). The ‘‘preprocsess’’ command line was employed to perform motion and slice timing corrections, masking, registration to the structural image, sampling to the surface, and surface smoothing by 5 mm as well as sampling to the MNI305 with volume smoothing. Surface sampling of time-series data was carried out onto the surface of the left and right hemispheres of the “fsaverage” template of FreeSurfer. Nuisance regressors were obtained for each study participants by extracting the EPI average time courses within the ventricle mask and the white matter mask (taking into consideration the top 5 principal components). These regressors, the motion correction parameters, and a fifth order polynomial were eliminated from the EPI time series. Temporal band-pass filtering (0.01<Hz<0.1) was applied to analyze only rs-fMRI data within this frequency range. The first four rs-fMRI time points were discarded to allow T1-weighted equilibration of the MRI signal. The mean signal time course within each seed region was employed as “regressor” to assess FC. With the “selxavg3-sess” command line, we performed the first level analysis (single subject analysis) including the computation of the Pearson correlation coefficient (r-value) between the time series within the seed and the time series at each voxel. The obtained correlation maps were then converted to Z-score maps before entering the second level analysis (group analysis). The “isxconcat-sess” command line was employed to create a “stack” of maps from each subject.
The Desikan-Killiany’s Atlas (Desikan et al. 2006) was employed to identify the location of clusters displaying structural MRI differences. In addition, two functional atlases, focused on cortical (Yeo et al. 2011) and cerebellar (Buckner et al. 2011) networks, were used to integrate the information provided by the Desikan-Killiany’s Atlas and define the positioning of clusters showing between-group differences or within-group correlations.
Statistical analysis
One-way analysis of variance (ANOVA) and Bonferroni post-hoc test were employed to evaluate the group differences regarding demographic/clinical data as well as the hippocampal and entorhinal morphometry. Chi-squared test was used to investigate the group differences on gender and the APOE ε4 carrier status. For analyses related to FC and cortical thickness, general linear models (https://surfer.nmr.mgh.harvard.edu/fswiki/FsgdFormat) were used. The analysis investigated the differences between groups (I comparison: HC subjects, MCI patients, AD patients; II comparison: HC subjects, nc-MCI patients, c-MCI patients, AD patients). Moreover, further general linear models were used, in the MCI patients, to assess, relationships between FC strength and other variables of interest like the subject age, the seed morphometric measures, the cortical thickness in each vertex, and the CSF biomarkers. The correlation analyses between the variables of interest and the FC of a given seed region FC were performed in a vertex-by-vertex computation by using the “pvr” option in “mri_glmfit”. Using the “mri_concat” command line, conjunction maps were created to highlight the sites of overlaps occurring between clusters expressing significant group difference (HC vs. MCI) and clusters that indicate significant correlations between FC strength and variables of interest. All the results are shown on statistical maps and adjusted by applying cluster-wise correction for multiple comparisons (Hagler et al. 2006).
Results
Demographic and clinical features of the study groups
Global cognition and episodic memory (recall and recognition) were affected in MCI and AD patients. When compared to MCI and HC subjects, AD patients showed significant impairment of semantic memory, verbal fluency, language ability, and executive functions. AD patients also showed significantly higher frequency of APOE ε4 when compared to MCI or HC subjects. Within the MCI subsets, the global cognition and episodic memory (recall) were found to be more compromised in the c-MCI group. Between the two MCI subsets, no differences were found when considering other neuropsychological and clinical features or the APOE ε4 frequency. When compared to HC, levels of Aβ1–42, t-tau and p-tau181 were found to be altered in AD patients. No statistically significant differences were found when comparing levels of CSF biomarkers in HC vs. nc-MCI or c-MCI vs AD. Higher levels of t-tau, p-tau181, Aβ1–42/t-tau, and p-tau181/ Aβ1–42 as well as lower concentrations of Aβ1–42 were found in c-MCI patients when compared to HC subjects, and in AD patients when compared to both the nc-MCI and HC subjects. No differences regarding age and educational levels were observed among the study groups (HC, MCI, and AD) or the MCI subsets (nc-MCI and c-MCI). Statistics on demographics and clinical features of the study groups are shown in Tables 1 and 2.
Morphometric variations in the study groups
AD patients showed hippocampal atrophy. No statistically significant differences in hippocampal volumes were found between the two MCI subgroups. However, when compared to HC subjects, c-MCI exhibited signs of atrophy in the right hippocampus. AD patients showed bilateral hippocampal atrophy when compared to HC or nc-MCI patients whereas no differences in hippocampal volumes were found when compared to c-MCI patients. No differences in the estimated total intracranial volume (eTIV) were seen in the study groups or the MCI subgroups. The statistical analysis of the structural data is shown in Tables 1 and 2.
Cortical thinning was found in AD patients when their cortical thickness was compared to HC (Supplementary Fig. 2A; Supplementary Fig. 3A) or MCI subjects (Supplementary Fig. 2B). The cortical atrophy, found in AD patients, was more prominent in the insula, the temporo-occipital areas, the supramarginal and angular, the temporo-parietal junction, the posterior cingulate cortex/precuneus, and the parahippocampal and entorhinal cortices. These regions actively participate in the connectivity patterns of the Default-Mode Network (DMN) and the Posterior Medial Network (PMN), thereby indicating the presence of disease-driven differences of critical functional value. No significant differences in cortical thinning were instead observed when comparing MCI vs. HC subjects.
In AD patients, the comparison with nc-MCI (Supplementary Fig. 3A) or HC (Supplementary Fig. 3C) subjects indicated greater thinning of brain regions that belong to the DMN and PMN. Significant thinning of the mesial temporal regions was instead found in AD patients when compared to c-MCI patients (Supplementary Fig. 3B). Finally, no statistically significant differences in cortical thickness were instead found when comparing the c-MCI with nc-MCI patients or HC subjects.
The entorhinal and hippocampal FC shows a distinct pattern in MCI subjects compared to AD patients
The analysis of rs-fMRI data revealed the presence of increased hippocampal FC in the MCI patients while the FC was decreased in AD patients. The investigation of the regional distribution of FC differences indicated that, compared to HC subjects (Supplementary Fig. 4A), MCI individuals exhibited enhanced connectivity of the hippocampus with the right medial prefrontal cortex (mPFC), cerebellar regions that are part of the DMN, right hypothalamus and left caudate, and hypo-connectivity between the hippocampus and of cerebellar regions that are part of the SN/CON. On the other hand, compared to MCI (Supplementary Fig. 4B; Fig. 3B) or HC (Supplementary Fig. 4C) individuals, AD subjects showed a decreased FC that took place in the DMN/PMN, striatum, and brainstem, and cerebellum regions that are part of the sensorimotor network. The entorhinal FC was found to be unaffected in MCI subjects and reduced in AD patients. Compared to HC subjects, AD patients displayed hypo-connectivity with DMN-related regions as well as with brainstem, striatum and cerebellar areas that are functionally linked to attentional and sensorimotor networks (Supplementary Fig. 5B, for the I model; Fig. 4C, Supplementary Table 1, for the II model). In addition, compared to MCI subjects, AD patients showed hypo-connectivity with cerebellar areas as well as with brainstem and thalamus (Supplementary Fig. 5A).
Divergent patterns of functional connectivity within the MCI subgroups
The analysis of the hippocampal FC of the two MCI groups identified significant differences. Nc-MCI patients were characterized by enhanced FC of the hippocampus with the mPFC as well as with cerebellar regions that are part of DMN and fronto-parietal network (FPN) and with subcortical regions like the thalamus, hypothalamus, striatum (ventral and dorsal portions), and superior colliculus (Fig. 1A, Supplementary Table 3). Compared to HC subjects c-MCI patients showed no differences in the hippocampal FC with the rest if the brain (Fig. 1B). Finally, nc-MCI patients displayed extensive hyper-connectivity between the hippocampus and cerebellar areas that are functionally associated to DMN, FPN, and SN/CON (Fig. 1C, Supplementary Table 4).
As far as the entorhinal cortex, nc-MCI patients showed hyper-connectivity in cerebellar areas that are part of the SN/CON, limbic and sensorimotor networks (Fig. 2A, Supplementary Table 5). Conversely, compared to HC (Fig. 2B, Supplementary Table 6) or nc-MCI subjects (Fig. 2C, Supplementary Table 7), c-MCI patients showed reduced connectivity with the lateral-occipital cortex, cortical and cerebellar regions that are part of attentive networks, brainstem, striatum, thalamus, and hypothalamus.
Compared to nc-MCI subjects, AD patients showed diffuse patterns of hippocampal hypo-connectivity with cortical and cerebellar regions that are mainly involved in the process belong to SN, DMN/PMN and sensorimotor network (Fig. 3A, Supplementary Table 8). Moreover, AD patients displayed entorhinal hypo-connectivity with brainstem, cortical regions of the DMN and SN/CON, and cerebellar areas that are part of the sensorimotor and attentional networks (Fig. 4A, Supplementary Table 9). In contrast, compared to c-MCI patients, AD individuals did not show differences in hippocampal FC but displayed entorhinal hypo-connectivity with the amygdala and brainstem (Fig. 4B, Supplementary Table 10).
Relationship of rs-fMRI data with age, structures and CSF biomarkers of MCI patients
In the MCI group, by using a whole brain correlation analysis, we found that the features of hippocampal FC were associated with altered CSF levels of p-tau and amyloid while no relationships were found with age and alterations in the structural integrity of the hippocampi or cortices. Increased FC occurring between the hippocampus and DMN regions was negatively associated with levels of p-tau181 and p-tau181/Aβ1–42, and positively associated with Aβ1–42 levels (Fig. 5, Supplementary Table 11).
The reduction of the strength of the right entorhinal FC was associated with cortical thinning of the right lingual gyrus and cuneus. No further associations were found between the entorhinal FC and another variable of interest (Supplementary Fig. 6).
Discussion
In the present study, we investigated patterns of hippocampal and entorhinal FC in a cohort of HC, MCI and AD subjects. The rs-fMRI data were also evaluated in relation to the clinical progression of the study participants. Overall, the analysis indicates that AD patients showed a synergic and parallel process of hypo-connectivity localized in the hippocampus and entorhinal cortex. The analysis of the MCI subsets shows that, while the c-MCI patients are characterized by hypo-connectivity, nc-MCI patients exhibited hyper-connectivity of both entorhinal and hippocampal regions.
Patterns of functional connectivity, cognitive status and structural damage in AD patients
Our AD subjects were characterized by the collapse of hippocampal and entorhinal connectivity, the decline in memory and executive skills, and the presence of marked signs of cortical and subcortical atrophy. These findings confirm the notion that macro-structural damage severely impairs global communication efficiency, prevents the adaptive functional reorganization of the brain networks, and ultimately sets the stage for the disease progression (Hillary and Grafman 2017). Of note, we observed a hypo-connectivity in the angular gyrus and retrosplenial/posterior cingulate cortex, two areas that are strictly involved in memory retrieval and intimately connected to the hippocampus and entorhinal cortex (Eichenbaum 2017; Sestieri et al. 2017). It is therefore possible that, in AD subjects, the reduced connectivity of the hippocampus and entorhinal cortex represents the functional correlate of the defective episodic memory retrieval that is typically found in the disease.
Patterns of functional connectivity, cognitive status and structural features in nc-MCI patients
Our study shows that nc-MCI patients exhibited hippocampal and entorhinal hyper-connectivity, relative preservation of cognitive functions and brain structures, and non-pathological levels of the AD-related CSF biomarkers. The findings are in line with previous studies showing patterns of increased hippocampal FC occurring in healthy, but at-risk for AD, individuals as well as in MCI patients (Bookheimer et al. 2000; Bondi et al. 2005; Hamalainen et al. 2007; Kircher et al. 2007; Das et al. 2013; Putcha et al. 2011). The hippocampal hyperactivity exhibited by MCI patients is controversial in value. While some authors have proposed that the process plays a compensatory role and helps to maintain cognitive performances (Sperling et al. 2009; Mormino et al. 2012; Oh and Jagust, 2013; Huijbers et al. 2015), others have considered the phenomenon disadvantageous and set to promote cognitive impairment (Das et al. 2013; Pasquini et al. 2015).
Our results that indicate the presence of enhanced FC between the hippocampus, thalamus, striatum, and mPFC lend support to the “compensatory hypothesis”. The thalamus is a structural and functional hub of the communication occurring between the hippocampus and mPFC, thereby supporting strategic cognitive functions, including memory consolidation (Ferraris et al. 2018) (Eichenbaum 2017). The striatum, along with the PFC, is also implicated in the modulation of memory retrieval (Scimeca JM and D Badre 2012). The mPFC is part of an integrated system (DMN) that sustains the global communication and meta-stability of the brain (Hellyer et al. 2014) and, ultimately, modulates a wide-range of high-order cognitive functions as well as the resilience against neurodegenerative processes (Hillary and Grafman 2017). The hyper-connectivity with the mPFC, is in line with different modelizations of brain ageing-related dynamics (i.e., HERA, HAROLD, PASA, CRUNCH, STAC, GOLDEN Aging) that postulate an increased engagement of the prefrontal brain regions to compensate for the functional decline of the posterior regions (Tulving et al. 1994; Cabeza et al. 1997; Davis et al. 2008) (Schneider-Garces et al. 2010; Park and Reuter-Lorenz 2009; Reuter-Lorenz and Park 2014; Fabiani M 2012). The compensatory hypothesis fits with evidence indicating that the mPFC and the hippocampus, the two areas where we observed increased FC, are tightly interconnected by bidirectional projections that are structurally and functionally integrated. The oscillatory synchronic activity between these two regions supports the organization and processing of the episodic memory (Eichenbaum 2017). The mPFC is strategic for memory as the area receives information on contextual cues from the anterior hippocampus and, in turn, indirectly sends the information, via thalamus and perirhinal/entorhinal cortices, to the posterior hippocampus (Fig. 6). In this context, the hippocampus acts as a key region set to control the memory organization and encoding, whereas the mPFC is implicated in the retrieval of context-appropriate memory engrams, the suppression of distractors or interference and the switching or selection of episodic memories according to contextual rules (Eichenbaum 2017). Furthermore, the presence of altered connectivity between the mPFC and the hippocampus impairs the object-place and temporal-order memory and leads to severe impairment in conditional visual discrimination as well as to learning and memory deficits related to defective suppression of irrelevant memory engrams (Eichenbaum 2017; Barker et al. 2007).
Interestingly, nc-MCI individuals were characterized by hippocampal hyper-connectivity with the hypothalamus, superior colliculus, and cerebellar areas that are functionally associated to the DMN and the FPN/DAN. These subjects also displayed hyper-connectivity between the entorhinal cortex and cerebellar regions that are part of the somatosensory network. Thus, our findings support a close functional connection between the hippocampus and cerebellum. The interplay between these regions, through circuits that involve the entorhinal cortex, hypothalamus, superior colliculus, and thalamus (including the cerebello-thalamo-cortical and cortico-ponto-cerebellar pathways), is strategic for the modulation of cognitively relevant prefrontal and parietal activities (Yu W and E Krook-Magnuson 2015). Moreover, the involvement of the cerebellum is intriguing as recent evidence indicates that the region acts as a critical hub for the control of a wide range of cognitive processes encompassing language, visual-spatial, executive, and working memory processes (Stoodley 2012). Thus, the hyper-connectivity between the hippocampus and cerebellum is functionally relevant and potentially associated with the relative preservation of the high-order cognitive functions of nc-MCI individuals.
Thus, from a theoretical standpoint, the increased FC that we observed in nc-MCI subjects may help to transiently cope with, and counteract, the undergoing neurodegenerative process and related cognitive impairment.
Patterns of functional connectivity, cognitive status and structural alterations in c-MCI patients
In contrast to nc-MCI subjects, c-MCI patients did not show signs of hippocampal hyper-connectivity with mPFC. This lack of hyper-connectivity may result in the reduced compensatory engagement of prefrontal areas and a more severe cognitive decline. In line with previous MRI studies (Grundman et al. 2002; Jack CR Jr. et al. 2004; Apostolova et al. 2006; Henneman et al. 2009), c-MCI patients showed hippocampal atrophy. These patients also showed entorhinal hypo-connectivity with cortical and cerebellar regions that take part in the modulation of long-term memory and attentional systems. Entorhinal hypo-connectivity, in particular, should be considered in relation to the role played by the superficial layers, II-III, of this region (Fig. 6). These layers act, in fact, as relay stations that carry, through the perirhinal or parahippocampal cortices, unimodal/multimodal cortical information from cortical associative areas to the hippocampus (Canto et al. 2008; Ranganath and Ritchey 2012). Moreover, the deep layers, V-VI, of the lateral entorhinal cortex send projections from the posterior hippocampus, via the cingulum, to the parahippocampus and the cortical areas involved in attentional networks and the DMN/PMN bundle (Kahn et al. 2008; Lacy and Stark 2012; Libby et al. 2012). These circuits promote the integration of spatial information as well as the representation of retrieved events (Preston and Eichenbaum 2013; Vann et al. 2009). It is therefore conceivable that the c-MCI reduced FC within the DMN/PMN and attentional/associative networks represents a functional marker of underlying alterations that occur before the onset and development of AD.
Overall, these data are in agreement with neuropathological evidence indicating that the entorhinal cortex and the hippocampus are the first brain regions to display tau-pathology and neurodegeneration in the course of AD (Braak et al. 1994). In line with this notion, our c-MCI patients showed decreased levels of Aβ1–42 and increased levels of t-tau, p-tau181, t-tau/Aβ1–42, and p-tau181/Aβ1–42, CSF alterations that went along with the presence of more profound memory deficits.
The correlation between altered CSF features and mPFC-related modifications is in line with studies showing that the decreased FC between central hubs of the DMN correlates with enhanced Aβ deposition (Buckner RL et al. 2005) (Elman et al. 2016; Foster et al. 2018; Grothe et al. 2016; Koch et al. 2010; Mutlu et al. 2017; Mormino et al. 2011; Palmqvist et al. 2017). The link between the presence of hyper-connectivity and enhanced signs of tau-pathology is less explored. However, fMRI/PET studies have recently shown that increased levels of tau-related pathology lead to a progressive decline of the brain FC (Cope et al. 2018; Hoenig et al. 2018; Jones et al. 2016; Schultz et al. 2017; Sepulcre et al. 2017) and the activation of the DMN in particular (Hoenig et al. 2018). These findings go along with evidence indicating that the loss of hippocampal GABAergic inter-neurons is closely associated with the appearance of enhanced signs of tau-pathology (Levenga et al. 2013). These processes may accelerate an ongoing pattern of hippocampal hyperactivity, micro/ macro-structural damage, atrophy, and the progression of cognitive and behavioral disorders (Gilani et al. 2014; Jones et al. 2016; Schmitz et al. 2017; Schobel et al. 2013).
We, therefore, propose a “work in progress” model by which a pattern of altered hippocampal FC may identify the degenerative processes that are driving MCI subjects to become AD patients. It is conceivable that the functional/dysfunctional value of the process varies upon different stages of the AD-related spectrum. At the MCI stage, the down-regulation of GABAergic neurotransmission may unleash a glutamatergic overdrive that promotes, a transiently beneficial, enhancement of the hippocampal activity that improves the functional coupling of the region with the cortex. This up-regulation is, for a while, advantageous and leads to increased communication between the hippocampus and cortical areas, like the mPFC, that are critically involved in the brain meta-stability and protection of cognitive functioning and, when deprived of activity, became more susceptible to AD and amyloidosis. This hypothesis is supported by the preservation of cognitive functions that we find in our nc-MCI patients. In the long run, the hippocampal increased FC may, however, set the stage for an enhanced, activity-dependent, damage of the region that is likely to be carried out by a combination of increased tau-related pathology [as we indirectly observed in the CSF of our c-MCI patients (Table 2)] and excitotoxicity, two processes ultimately leading to decreased FC and clinical conversion to AD (Fig. 7).
In conclusion, the study identifies the functional correlates of alterations that occur in patients at the early stages of the AD-related spectrum. Our study has some limitations. For instance, it should be underlined that all the study subjects are highly educated individuals who likely possess significant levels of cognitive reserve. Furthermore, the neuropsychological tests employed in the ADNI database are skewed toward the investigation of mnemonic domains and do not allow a detailed analysis of visuospatial and attentional domains. Our working hypothesis warrants future longitudinal investigation. For instance, the model should be tested and further validated by investigating changes in functional and structural connectivity in relation to ongoing processes of amyloid and tau deposition as assessed by PET imaging.
Acknowledgments
This work was supported by research grants from the Italian Department of Education (PRIN 2011; 2010M2JARJ_005) and the Italian Department of Health (RF-2013–02358785 and NET-2011-02346784-1).
Data collection and sharing for this project was funded by the ADNI (National Institutes of Health Grant U01 AG024904). ADNI is funded by the National Institute on Aging, the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: AbbVie, Alzheimer’s Association; Alzheimer’s Drug Discovery Foundation; Araclon Biotech; BioClinica, Inc.; Biogen; Bristol-Myers Squibb Company; CereSpir, Inc.; Cogstate; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. Hoffmann-La Roche Ltd and its affiliated company Genentech, Inc.; Fujirebio; GE Healthcare; IXICO Ltd.;Janssen Alzheimer Immunotherapy Research & Development, LLC.; Johnson & Johnson Pharmaceutical Research & Development LLC.; Lumosity; Lundbeck; Merck & Co., Inc.;Meso Scale Diagnostics, LLC.; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Company; and Transition Therapeutics. The Canadian Institutes of Health Research is providing funds to support ADNI clinical sites in Canada. Private sector contributions are facilitated by the Foundation for the National Institutes of Health (www.fnih.org). The grantee organization is the Northern California Institute for Research and Education, and the study is coordinated by the Alzheimer’s Therapeutic Research Institute at the University of Southern California. ADNI data are disseminated by the Laboratory for Neuro Imaging at the University of Southern California.
The authors are grateful to all the members of the Molecular Neurology Unit for helpful discussions and to Dr. Domenico Ciavardelli for helping with the statistical analysis.
Footnotes
↵* Data used in preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.edu). As such, the investigators within the ADNI contributed to the designed and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. A complete listing of ADNI investigators can be found at: http://adni.loni.usc.edu/wp-content/uploads/how_to_apply/ADNI_Acknowledgement_List.pdf