The Auditory Afferent Pathway as a Clinical Marker of Alzheimer’s Disease

INTRODUCTION

Dementia, in particular Alzheimer’s disease (AD), poses a substantial economic burden of direct and indirect costs in Australia which exceeded 14.6 billion in 2016 and are predicted to rise to 36.85 billion AUD by 2056 [1]. Globally we are facing a pandemic of AD with 9.2 million cases predicted for the year 2050, such that 1 in 85 people on our planet will be living with this disease [2]. According to the World Health Organization (WHO) there are nearly 10 million new cases of dementia every year, with AD contributing to 60 to 70% of those cases [3].

AD has traditionally been a clinical diagnosis with standardized memory tests performed contemporaneously with an initial clinical assessment. The diagnosis can be supported by neuropsychological testing and structural imaging modalities, the latter of which are performed to exclude alternative diagnoses as well as characterize regional abnormalities, for example, by performing cross sectional magnetic resonance imaging (MRI) with hippocampal volumetry. These modalities are limited in their ability to accurately determine the underlying etiology in subjects presenting with mild cognitive impairment (MCI) [4]. This may potentially influence the findings of clinical trials as was found in the solanezumab EXPEDITION and EXPEDITION2 phase III studies in which 25% of the study population did not have imaging evidence of amyloid at baseline [5]. Recent modalities, such as amyloid and tau positron emission tomography (PET) and cerebrospinal fluid analysis (CSF) of amyloid-β (Aβ) 42/40 ratio and CSF tau [6], are not readily available for the general community and relatively expensive. Peripheral blood tests that measure the ratio of the two isoforms of Aβ are a promising future alternative to these more complex measures [7].

More recently, it has been postulated that various brain stem neural tracts and nuclei may be disturbed prior to observable neuronal atrophy in AD [8–10]. It follows that a functional examination of such pathways, using non-invasive techniques, such as magnetic resonance spectroscopy (MRS) [11], brain stem auditory evoked potentials (BAEPs), and event-related potentials (ERPs) [2] may improve diagnostic accuracy, particularly if structural abnormalities not observable using traditional imaging modalities, can be detected by observable functional deficits. A hybridization of MRS, BAEPs, and ERPs may be of value in identifying patients who are at risk of cognitive decline. It does not require use of a radiopharmaceutical compound and is non-invasive when compared to PET and/or CSF studies.

Hearing in patients with dementia is a focus of growing clinical interest, with increasing evidence that hearing loss may predict or accelerate cognitive deterioration [12, 13] and alterations of hearing may manifest as complex cognitive and behavioral symptoms relevant to the differential diagnosis of dementias [14–19]. The proposition that neural tracts and nuclei reacting before any identifiable atrophy of the cortex is a rather new hypothesis which is supported by neuropathological studies of brain sections of patients with AD [9]. The inferior and superior colliculi and medial geniculate body have been found to exhibit senile plaques and neurofibrillary tangles, respectively, in the brains of patients with AD [9]. Morphological changes in these two regions including neuronal loss and synaptic alterations, prior to the appearance of senile plaques or neurofibrillary tangles, have also been reported in early cases of AD [10]. It seems appropriate to investigate the anatomical structure of the brain stem, which has a high content of neural tracts.

Neuropathological studies in AD have demonstrated involvement of brain stem nuclei. One of these nuclei, the Tegmentopontine reticular nucleus, is affected in early stages during the development of AD (preclinical AD) before the neuropathological changes affect the limbic system (composed of the thalamus, hypothalamus, hippocampus, and the amygdala) and cerebral cortex [19]. The solitary tract and nucleus as well as dorsal nucleus of the vagus nerve are also affected in the early stages of AD. Histologically, most of these affected nuclei exhibit neurofibrillary tangles or senile plaques in the brains of patients with AD [9].

MAGNETIC RESONANCE SPECTROSCOPY

Patients with amnestic mild cognitive impairment (MCI) have a higher risk of progression to AD compared with their cognitively normal peers, and proton magnetic resonance spectroscopy (¹H-MRS) is a non-invasive biomarker, in such conditions, that can be diagnostically useful through its ability to measure numerous metabolites in the human brain [20].

There is evidence of progressive chemical changes that involve multiple brain regions during the progression of AD such as decreased neuronal integrity marker; N acetyl aspartate (NAA); or NAA/creatine ratio in the parietal and occipital cortex, gray matter, hippocampus, and posterior cingulate and increased glial marker myoinositol in the parietal and occipital cortex, gray matter, and posterior cingulate [21]. Elevated myoinositol/creatine and choline/creatine ratios and reduced NAA/creatine ratios in MCI and pre-symptomatic AD suggests that ¹H-MRS is valuable in predicting future development of dementia and monitoring early disease progression for preventive therapies [22].

No neuroimaging study has yet dealt with the anatomical structure of the brain stem nuclei concerning pathological changes caused by AD. We evaluated the ¹H-MRS of inferior colliculus findings in 54 males (27 healthy controls, 27 patients with MCI; aged 50–70 years). All patients underwent neuropsychological testing using Mini-Mental State Examination (MMSE) [23] and Clinical Dementia Rating (CDR) [24]. ¹H-MRS of the inferior colliculus was performed by the Point Resolved Spectroscopy (PRESS) system for sequencing, using both multi and single voxel with different voxel sizes (from 7–15) to obtain reproducible results with minimum noise. The study confirmed the feasibility of using ¹H-MRS in MCI for tracking neuronal changes in the brain stem (Figs. 1 and 2).

OPEN IN VIEWER

Fig. 1

The ¹H-MRS imaging plane used for the inferior colliculus.

OPEN IN VIEWER

Fig. 2

A normal MRI image from a mild congitive impairment patient, and a ¹H-MRS image from the same patient showing abnormal biochemical changes.

THE ASCENDING AUDITORY PATHWAY

Alteration of hearing may predict future onset of dementia such as AD. A relevant starting point in the examination of auditory neural tract disturbances would be the primary afferent pathways, more specifically the associated brain stem and midbrain nuclei as cortex activity is informed and governed by these structures including cochlear nuclei, inferior colliculi, and the medial geniculate nuclei.

We have selected to focus on the auditory pathway as it is the most external of the sensory pathways and it is also generally one of the last sensory pathways to be clinically detected as being disturbed in AD. It could be inferred that if damage is found in the auditory pathway, that other brain stem pathways, for example the nucleus tractus solitarius [25] and locus coeruleus [26], may already be damaged more severely.

Processing of sound begins in the ascending auditory pathway extending from the cochlea to the primary auditory cortex in Heschl’s gyrus (transverse temporal gyrus) and involves complex signal processing at each of the different levels of the pathway [27]. A meta-analysis concluded that hearing loss impacts on multiple domains of cognition and that there is a correlation between cognition and hearing impairment [28] that is not simply attributable to hearing loss confounding speech-based cognitive tasks [29] and has been observed in those with and without dementia [30]. Histopathological involvement of auditory cortices has been described in major neurodegenerative dementias [31] and deficits of auditory cognition are early features of these diseases. Limited histopathological data is available for the role of auditory system in common dementias, indicating that the major auditory relay nuclei are involved in the pathology of AD [9, 31] and abnormalities of auditory cortical evoked potentials precede the clinical symptoms in young carriers of pathogenic AD mutations [32]. Relatively few studies, on the role of hearing in dementia, have addressed cortical auditory processing specifically, perhaps due to the wide variation in the reported frequency of hearing impairment in AD [13].

Patients with dementia may have reduced perception of sound, disproportionate to any damage involving the cochlea or the ascending auditory pathways, which may manifest as cortical deafness syndrome [33] with peri-Sylvian degeneration [34]. Patients with clinically typical AD commonly report difficulty following conversations and other sounds against background noise which may contribute to avoidance of social situations and a general dislike of busy auditory environments [35]. Such symptoms are attributed to a nonspecific memory or attentional deficit that may signify AD-associated impairments of auditory scene with disintegration of a parieto-temporal network [14, 18] in an unfamiliar accent [16]. A neuroanatomical substrate may be present in the posterior peri-Sylvian cortices [35].

It may be of value to measure BAEPs not only for auditory dysfunction in general but also for patients with MCI or suspected diagnosis of AD. Previous literature suggests that central auditory dysfunction may precede the onset of clinical dementia [36]. There is also growing evidence of vestibular loss as a contributing factor to the development of MCI and AD.

EVENT-RELATED POTENTIALS AND QUANTITATIVE EEG

ERPs comprise of voltage changes to sensory stimuli, for example auditory (Fig. 3), captured during an electroencephalogram (EEG) [37]. The sequence of 7 auditory evoked potentials (Fig. 3), as described by Starr and Achor [38], are directly related to neural activity within the inferior colliculi [39]. The inferior colliculi provide input to the medial geniculate nucleus of the thalamus which subsequently relay to the cortex for processing, directly influencing cortex activity more broadly. Abnormalities in interpeak latency and/or amplitude ratios of the BAEP may be present in patients with MCI. It is also plausible that EEG spectral power recorded at the cortex is associated with neural activity at the level of the inferior colliculi. Abnormalities in components of the auditory ERPs have been recently described in aging, depression, and cognitive impairment including AD [40]. There may also be a role for evaluating other pathways including visual ERPs and vestibular evoked myogenic potentials in MCI and AD [41].

OPEN IN VIEWER

Fig. 3

This figure shows voltage changes to auditory stimuli captured during an electroencephalography (EEG) recording. Auditory brain stem response (ABR) of less than 10 ms, followed by the auditory middle latency response (AMLR) of less than 100 ms and event related potentials (ERPs) of up to 1000 ms.

The P300 component of an auditory ERP is a positive deflection occurring 300 ms and the N400 component is a negative component peaking at 400 ms following a stimulus [42]. Research has demonstrated that EEG spectral power within the delta, theta, alpha, and beta frequency bands of pre- and post-auditory stimulus EEG is associated with both the amplitude and latency of various auditory ERP components [42]. The P300 amplitude has been shown to be reduced across the whole head and its distribution is altered in patients with AD with maximums found at frontal sites as opposed to parietal sites in healthy controls [43]. The P300 latency has also been shown to be increased in patients with AD [44] whereas the N400 component is diminished [45]. The P400 amplitude has also been shown to be significantly higher in the posterior head regions in patients with AD who are performing a memory workload task [46]. Mismatch negativity has been found to have a decreased amplitude in AD [47].

Research indicates that both broad and event specific brain activity changes may precede and possibly predict early cognitive decline [2]. Changes in beta and gamma activities are generally linked to cognitive decline [48]. Domain specific cognitive decline may also influence other EEG activities across all frequency bands. An increase in delta and theta frequency band activity on quantitative EEG is well recognized as a manifestation of cognitive impairment [49]. Changes in higher frequency EEG activity including decreased upper alpha band activity [50] and alpha reactivity [51] have also been correlated to changes in cognitive performance. Alterations in gamma oscillatory synchrony [52] and in beta [2] and gamma activity including increased theta/gamma ratios [53] have also been implicated.

FUTURE DIRECTIONS, SUGGESTIONS, AND CONCLUSIONS

Despite the plausible relationship between ¹H-MRS based investigations of neural degeneration with similarly based BAEP and ERP quantitative electroencephalography research, future confirmatory studies that directly compare and/or hybridize these measures are required. Such research could lead to the future development of a unique hybrid algorithm that could function as a novel non-invasive biomarker of early cognitive decline and which may be further utilized to identify individuals at risk of dementia, in particular AD.