From Mild Cognitive Impairment to Alzheimer’s Disease: A New Perspective in the “Land” of Human Brain Reactivity and Connectivity

The study “Promising role of neuromodulation in predicting the MCI progression to dementia” by Naro and colleagues which is published in the present issue of the Journal of Alzheimer’s Disease [1] is dealing with a hot topic of modern neurosciences, namely how subjects with mild cognitive impairment (MCI) who will progress to Alzheimer’s disease (AD) can be intercepted with high sensitivity/specificity. This is because the MCI population is considered as a prodromal AD state since the condition carries a high risk of developing dementia with an annual rate of conversion which is about 20 times higher than the general elderly non-MCI population [2]. Moreover, a bulk of clinical and experimental evidence is prompting the idea that mechanisms of neurodegeneration in AD (i.e., a progressive aggression starting with synaptic disruption, followed by amyloid-β plaques aggregation and tau protein deposition leading to neuronal death and loss of neural circuits) act for many years before symptoms appear. Meanwhile, it is known that appropriate lifestyle, currently available drugs, and cognitive rehabilitation as well as future treatments are most effective if initiated well before the clinical symptoms of AD fully develop. Thus, there is a need of identifying patients that are at risk of progression. Currently available methods either provide a relatively late warning (i.e., when amyloid aggregation is already widespread) or are expensive and not diffusely available (PET scan with amyloid-β radioligands, volumetric MRI) or invasive (lumbar puncture). There is a growing interest, at present, concerning deteriorating brain electroencephalographic (EEG) oscillations as a predictor of MCI-to-AD conversion [3]. For this reason, the authors experimentally modulated the γ-band oscillations (GBO)-gamma EEG rhythms are mainly involved in cognitive functions [4]-in a sample of MCI and AD patients and an age-matched healthy elderly group, using a transcranial alternating current stimulation (tACS) protocol. Stimulation was applied for 10 min at γ-band frequency (i.e., continuously and randomly ranging from 40 to 120 Hz, with the same number of discrete steps delivered at 20 Hz), with zero-degree phase-lag and at a very low (barely distinguishable) intensity (peak-to-peak current intensity of 1 mA). This approach was chosen because it covers the whole spectrum of GBO that can be recruited by tACS. Interestingly enough, only those MCI subjects who lacked significant neuropsychological and electrophysiological after-effects to tACS, similar to AD individuals and differently from healthy controls, converted to AD in a 2-year follow-up. The authors conclude that this type of approach might be of support for differential diagnosis of MCI and AD and help to intercept early in the disease course those MCI subjects who are in a pre-symptomatic stage of an already established AD. This prediction index may help the clinician to adopt a better prevention/follow-up strategy.

The effects of transcranial direct current stimulation (tDCS) on brain functions (including rhythmic EEG oscillations) and the underlying molecular mechanisms are still largely unknown. Synchronization and phase coherence of EEG rhythms affects communication between neuronal groups [5]. In particular gamma-band (30–90 Hz) synchronization modulates excitation rapidly enough that it escapes the following inhibition and activates postsynaptic neurons effectively. Synchronization also ensures that a presynaptic activation pattern arrives at postsynaptic neurons in a temporally coordinated manner. Multiple presynaptic groups, representing different stimuli, thus converge toward postsynaptic neurons, but if a stimulus is selected depending on attention, its neuronal representation shows stronger and higher-frequency gamma-band synchronization and the attended stimulus representation selectively entrains postsynaptic neurons. This entrainment creates sequences of short excitation and longer inhibition that are coordinated between pre- and postsynaptic groups to transmit the attended representation and shade the network from competing inputs. The predominantly bottom-up-directed gamma-band influences are controlled by predominantly top-down-directed alpha-beta-band (8–20 Hz) influences. Subject attention itself samples stimuli at a 7-8 Hz theta rhythm. Thus, several oscillating EEG rhythms and their interplay render neuronal assemblies communication effective, precise, and selectively tuned on the transmission of the relevant information.

Recent reports on mice subjected to tDCS exhibited increases in hippocampal LTP, learning and memory changes lasting up to 1 week [6]. These effects were associated with an enhancement of: 1) acetylation of brain-derived neurotrophic factor (BDNF) promoter I; 2) expression of I - IX BDNF exons; and 3) BDNF protein levels. Altogether, experimental models suggest that tDCS increase hippocampal LTP and memory via chromatin remodeling of BDNF regulatory sequences leading to increased expression of this gene, and prompts the idea that it might particularly modulate those functions first affected by impaired neuroplasticity.

The evaluation of the sensitivity of brain oscillations, especially GBO, to external perturbations, apart from offering a possible and reliable index of cognitive reserve, could be currently considered the best methods to study and understand brain connectivity phenomena. The growing interest in the “human connectome” fits strongly in a long-standing debate in neuroscience, between localizationist and holistic views. The former highlight the specificity and modularity of brain organization, whereas the latter stress global functions and mass action. This controversy mirrors two contrasting properties that coexist in the brains of higher vertebrates: the functional segregation of different brain regions and their integration in perception and behavior. Evidence that the brain is functionally segregated at multiple levels of organization is overwhelming. Developmental events and activity-dependent selection result in the formation of neuronal assemblies of strongly interconnected cells sharing inputs, outputs, and response properties. Each group tends to be simultaneously, but transiently, connected to several and specific subsets of other neuronal aggregates. Further evidence for functional segregation in a variety of systems is provided by the analysis of the specific deficits produced by localized cortical lesions. In contrast to such local specialization, brain activity is globally integrated at many levels ranging from the individual neuron to inter-area interactions and to overall behavioral output. The interplay of segregation and integration in brain networks generates the huge variety of functions both strongly diversified and integrated, thus creating patterns of high complexity. This is also the basis of the individualistic behavior, which typifies the human species.

Over the past decade, the study of networks has rapidly expanded allowing the quantification of broad networks characteristics and revealing that the large-scale connectivity patterns of the cerebral cortex merge structural aspects of two extremes (‘regular’ versus ‘random’) and share some characteristics with the ‘small-world’ networks, including high levels of clustering coefficients and short characteristic path length [7–9].

In general terms, connectivity can be studied on many spatial and temporal scales as well as at several levels of complexity [10, 11]. Mainly three different types of connectivity can be recognized: anatomical, functional, and effective. Anatomical connectivity refers to a network of synaptic connections linking sets of neurons or neuronal elements, as well as their associated structural biophysical attributes condensed in parameters such as synaptic strength or effectiveness [12, 13]. Comprehensive data of anatomical connectivity are still scarce and difficult to obtain. Nowadays, achieving a detailed map of all the neurons and their interconnections in a mammalian brain is simply out of technological reach and only invasive tracing studies are capable of demonstrating direct axonal connections.

While functional connectivity only refers to statistical dependencies between spatially separated neuronal events, effective connectivity refers to a mechanistic model of how the data were caused [14, 15] and requires a causal or non-causal model, in which regions and connections of interest are specified and often constrained by a combination of neuroanatomical, neuropsychological, and functional neuroimaging data. Unlike structural connectivity, functional connectivity is highly time-dependent. Statistical patterns between neuronal elements fluctuate on multiple time scales, some as short as tens or hundreds of milliseconds. The accessibility of measurements of connectivity is driven by the measured parameters (electrophysiological, metabolic or structural imaging) and analysis methods [16].

Physiological recordings from multiple brain sites using techniques as EEG, magnetoencephalography (MEG) (Rossini & Ridding, unpublished) and very recently Hd-EEG in co-registration with transcranial magnetic stimulation has shown that remote neuronal assemblies can synchronize their firing. Neuroimaging studies, using fMRI, have reported the co-activation of distant brain regions under different experimental conditions, setting the foundations for novel approaches to understand the brain “at work” [17, 18].

Brain connectivity may be studied using “Graph Theory”, a branch of mathematics with diverse applications ranging from urban planning and traffic control to epidemiology and the analysis of complex biological systems. A graphs network can be described by several parameters, particularly by clustering coefficient, as a local connectedness measure, and characteristic path length (a global average of all distances), as an overall connectedness indicator. In this view, brain graphs provide a relatively simple way to model the human brain in a comprehensive map of neural connections: the connectome [19, 20]. Graph analysis of structural/anatomical (diffusion MRI, cortical thickness correlation) and functional (fMRI BOLD signals, EEG/MEG recordings) data have demonstrated a “small-world” configuration-a network with many local connections and a few random long-distance connections-in the healthy human brain possibly responsible for high efficiency of information processing related to cognitive and sensorimotor performance [21]. In the same line, alteration of “small-world” properties has been reported in brain diseases such as stroke and dementia [22–27]. With the aim of finding a post-hoc neurophysiological pattern prognosticating clinical outcome and characterizing effective rehabilitation, graph theory is helpful.

Modulation of EEG rhythms via transcranial stimulation is considered an emerging and potentially valuable method to study cortical reactivity, cortico-cortical connectivity, and brain functional organization. Two types of approach have been described and theorized [28]: 1) the “inductive” approach, based on the study of the spatio-temporal distribution of EEG wavelets induced transcranial stimulation; 2) the “rhythmic” approach (parallel to the methodology applied in the study by Naro et al. [1]) based on the principle that trains of rhythmic pulses can trigger or enhance brain oscillations, when these trains are frequency-tuned to the underlying EEG oscillations of the target cortical area (entrainment). Taking in account these features, the combination of transcranial stimulation and EEG allows to probe directly, with high spatio-temporal discrimination, whether and how the stimulation of a cortical cell assembly evokes synchronized neuronal activity in connected areas and its disruption/alteration in neurological diseases [22–27, 30].

In conclusion, over the last few years, a big effort has been made in order to find a marker able to predict reliably the progression from MCI to AD. Different techniques were explored in blood samples, cerebrospinal fluid, neuroimaging, and neurophysiological fields. Neuroimaging has the advantage of an excellent spatial resolution, but a relatively low time discrimination, and is time consuming and expensive. Neurophysiology can compensate the low spatial resolution with a good temporal resolution and its low cost. Moreover, some evidence suggest that functional alterations precede the structural ones and, at least theoretically, there should be a prolonged time window during which the brain anatomy is normal, while synaptic dysfunction is already affecting dynamic connectivity. Within this scenario neurophysiological techniques might be the ones to show early alterations. According to this view, in the early phases of neurodegeneration the recruitment of neural reserve networks help to maintain normal brain functions. However, in this condition brain reactivity is significantly reduced as demonstrated by the absence of tACS induced GBO changes in the Naro and colleagues study [1]. In this view, perturbation of EEG signals induced by non-invasive brain stimulation can be considered as a “stress test” able to verify the residual brain reactivity.

One of the major goals of modern research in the field of dementia is to find a marker that could be used in order to build up a method of screening for the diagnosis of AD or to identify patients with MCI who will convert in AD. A screening should be a method widely accessible, of relative low cost, and with a high specificity and sensitivity. At the moment, EEG and neurophysiological methods fulfill such characteristics. However, despite high specificity and sensitivity found in many studies, the efficacy of these methods in differentiating MCI converting and non-converting in AD should still be defined when applied at subject level. Further efforts should be made in order to test this or similar approaches in a larger sample of patients in order to validate a marker for early diagnosis as well as to track disease progression for clinical trials focused on new treatments.