1.1 Background

Recent technological and methodological advances have improved the availability of biomarkers allowing an early identification of the specific pathology underlying Alzheimer's disease (AD).1 Standard magnetic resonance imaging (MRI) can detect AD-typical patterns of morphological changes of the brain, such as hippocampal atrophy or microbleeds.2 Further, positron emission tomography (PET) in combination with specific radiotracers provides early information about regionally reduced glucose metabolism as well as amyloid and tau deposits.3-5 So far, however, many diagnostic markers have lacked predictive value for cognitive decline. Additionally, even if such imaging markers can support diagnosis, the understanding of pathological processes underlying the development and progression of dementia is still insufficient, hindering the development of successful therapeutic concepts.

Recently, non-invasive measurements of the brain tissue sodium (Na+) concentration (TSC) from 23Na-MRI have become increasingly available due to technical and methodological improvements and have already been successfully applied to investigate sodium alterations in different diseases such as tumors and neurodegenerative or chronic inflammatory diseases.6-10 In contrast to classical 1H-MRI, 23Na-MRI sets its focus on sodium changes as an early hallmark of neurodegeneration.

Sodium depicts a potentially interesting marker of cellular metabolic state: its transmembrane ion gradient defines the neuronal resting potential maintained by the cellular Na+/K+-ATPase (NKA), which is highly dependent on energy in the form of adenosine triphosphate (ATP). This resting potential is crucial for cell homeostasis and successful neurotransmission. Disturbances in this overall cascade of energy synthesis on mitochondrial level, NKA, and ion channels for stabilizing cell homeostasis can lead to increased intracellular sodium concentration with a breakdown of the resting potential, which finally results in impaired neurotransmission and ultimately cell death.6, 10-12

In the context of AD, baseline studies have indeed shown a close interplay between the pathological hallmarks of AD, that is, amyloid beta (Aβ) deposits and tau tangles with the NKA in disease models.13-16 Correspondingly, sodium and potassium concentration imbalances have been detected in post mortem brain tissue, and increased sodium concentrations have also been found in blood plasma of AD patients and in cerebrospinal fluid (CSF) of those at risk.17-20

These findings point to the hypothesis that local sodium concentration changes detected in vivo by 23Na-MRI might be useful as an early marker of neuronal dysfunction in AD, which could be associated with cognitive decline, and to investigate the crucial interplay between metabolic deficiency as a potential consequence of energy deficits and AD-related amyloid and tau pathologies. This type of imaging goes beyond the characteristic volumetric alterations of brain regions in AD pathology and could point to a more substantial hypothesis of failed neurotransmission as the basis of clinical implications. It might even be that the vital energy breakdown at the cellular level could be considered a potential starting point due to an impairment of mitochondrial function,21, 22 possibly interacting with amyloid and tau deposits. In particular, there are findings that the ATP-synthase, an essential pump for ATP production as an energy source, might be affected in AD.23

Only two preliminary 23Na-MRI studies have been carried out so far in AD, one on a small cohort of five mild AD patients,24 and one recent study at 3 Tesla25 pointing to increased intensities from 23Na-MRI in brain regions typically affected by AD, such as the hippocampus. However, there have not been any previous 23Na-MRI studies so far in AD under technically advanced conditions with the improved signal-to-noise ratio (SNR), improved resolution of 7 Tesla magnetic fields, and with a quantification of TSC complemented by high-resolution 1H-MRI, tau and amyloid PET neuroimaging as well as clinical assessment.

In the current study, we examined 17 patients with AD via 23Na-MRI and classical 1H-MRI under ultra-high-field MR conditions at 7 Tesla. AD subjects also received [11C]-PIB and [18F]-flortaucipir PET for quantification of regional amyloid and tau load. A matched control sample of 22 cognitively healthy control subjects was also included. All subjects also received neuropsychological assessment. At the whole-brain level, we indeed found increased TSC in several brain regions of AD patients, such as regions of the temporal and parietal lobes, as typically affected by neurodegeneration. These regions did not always overlap with atrophy clusters that were computed in parallel. Apart from TSC increases in our patients’ cohort, we further investigated which factors influence regional sodium increases. We found that TSC is indeed predictive for cognitive state, evaluated via a cognitive screening test, the Montreal Cognitive Assessment (MoCA), and can be further influenced by local brain volume, age, and sex. TSC in the fusiform gyrus, a brain region in the temporal lobe, has furthermore higher sensitivity in discriminating between mildly affected patients and healthy controls compared to grade of atrophy.

RESEARCH IN CONTEXT

Systematic Review: The authors performed a review of the literature using common sources such as PubMed. Post mortem studies in Alzheimer's disease (AD) have indicated sodium increases and possible interactions between amyloid and tau with the energy-dependent Na+/K+-ATPase. In vivo studies in line with these findings are still lacking.

Interpretation: Our findings support the hypothesis of a link between brain tissue sodium increases and the progression of AD-specific pathophysiology and cognitive decline.

Future Directions: The article proposes a framework for the continuing exploration of this hypothesis via the conduction of additional, longitudinal neuroimaging studies. Future directions include the evaluation of 23Na-MRI (magnetic resonance imaging) as a stage-dependent marker of AD, and exploration of the association with local hypometabolism and AD pathology. As a possible metabolic imaging marker, 23Na-MRI could support the understanding of reduced neurotransmission and monitor disease conversion and progression.

1.2 Study conclusions and implications

Our study therefore points at a TSC imbalance accompanying the neurodegenerative process in AD in vivo. This finding confirms observations from post mortem studies18, 26 and the recent detection of sodium increases in AD patients and subjects at risk.19, 20 Interestingly, TSC can be highly discriminative between control subjects and only mildly affected patients in temporal regions such as the fusiform gyrus and is predictive for cognitive state. TSC might therefore serve as a non-invasive and powerful marker of brain dysfunction even in early stages of the disease and could therefore be valuable, for example, as an outcome parameter for intervention trials.

What could be the neuropathological basis of the observed association of increased TSC with cognitive impairment? Because sodium is crucial for neurotransmission, it is likely that impairment in interneuronal transmission due to sodium imbalance would constitute a turning point for cognitive decline, possibly driven by local hypometabolism.

In this context, we further show that both amyloid and tau load are associated with TSC changes in AD. Notably, we found stronger associations of TSC with tau load compared to amyloid, especially for the fusiform gyrus as well as structures of the parietal lobe. This finding matches with previous post mortem studies in which the Braak stage, defined by the load of neurofibrillary tangles and representing the disease state, was associated with increased [Na+] in the parietal lobe in AD patients.18 This would also match the finding that tangle pathology seems to correlate more strongly with memory impairment than amyloid load and that tau seems to spread across neurons through neuronal connections,27 possibly mediated via amyloid.28, 29 Even if we found widespread TSC/tau and volume/tau associative patterns, the exact nature of the interplay between sodium imbalance, or glucose hypometabolism as reflected by fluorodeoxyglucose (FDG)-PET (e.g., if TSC precedes hypometabolism) and the proteinopathies in AD remains to be elucidated.32, 33

Altogether, sodium increases and therefore disturbances in the cell homeostasis give an important view on AD. These observations endorse the concepts that AD research should set a focus on managing these energy deficits, which are occurring early in the disease and go beyond the simple consequence of cell loss, which are seen ultimately as atrophy in our conventional anatomical MRI. Considering that almost all energy synthesis in the brain is oxidative and occurs in mitochondria, chronic and progressive mitochondrial dysfunction could be the starting point of this pathological process. The hampering of the electron transport chain by the accumulation of pathological amyloid and tau aggregates would then lead to a direct ionic channel downregulation and a therefore self-reinforcing cascade with a destabilized neurotransmission and the development of cognitive symptoms. A comparable hypothesis has been modeled for the case of motor neuron degeneration in amyotrophic lateral sclerosis.33 In these simulations, the reduced ATP supply led as projected to failed K+/Na+ homeostasis and an increasingly costly action potential for the neuron and a subsequent vulnerability. Because we also observed sodium increases in Huntington's disease9 and recently in Friedreich's ataxia, in which mitochondrial malfunction is hypothesized to play a major role, this should lay a basis for testing in vitro and in vivo the hypothesis of a possible common initial pathway in several neurological diseases.

23Na-MRI gives us a great opportunity to focus attention on a fundamental component of the disease, namely aberrations in signal transduction and alterations of cellular energy supply. Our study might be a motivation to deepen this metabolic approach, due to its strong link with cognitive decline. The exact relationship between increased TSC and synaptic/neuronal dysfunction or glucose metabolism on the one hand, and proteinopathies on the other hand, needs to be further studied in larger longitudinal cohorts combining 23Na-MRI with other imaging modalities and blood/CSF biological markers as well as cellular models, both in AD and in other neurodegenerative diseases. TSC might therefore have the potential of a non-invasive and powerful marker of brain pathology even in early stages and can therefore be valuable in translational approaches, for example, as an outcome parameter for intervention trials34 and/or for monitoring of treatments.