The pathogenesis of AD

AD is an irreversible, highly complex, and progressive neurodegenerative disease. It was first described in 1907, with neurotic plaques, neurofibrillary tangles (NFTs), and amyloid angiopathy in the cerebral cortex [Reference Bondi, Edmonds and Salmon22]. These autopsy characterizations were then become the hallmarks of AD. Over the past several decades, numerous hypotheses have been made for AD pathogenesis. The amyloid cascade hypothesis postulates that neurodegeneration in AD is caused by abnormal accumulation of Aβ plaques in the cerebral cortex [Reference Barage and Sonawane23]. Aβ is cleavage from the amyloid precursor protein (APP), the defective intracellular cleavage of APP that is thought to be the predominant cause of plaque formation. The tau theory focuses on the aggregation of intracellular NFTs formed by accumulated tau protein in a hyperphosphorylated state [Reference Hampel, Mesulam and Cuello24]. The tau hypothesis suggests that the hyperphosphorylation of tau (P-tau) induces the formation of NFTs, which is another hallmark pathology of AD [Reference Mohandas, Rajmohan and Raghunath25,Reference Kametani and Hasegawa26]. Calcium dysregulation seems to generate large amounts of tau [Reference Arnsten, Datta, Del Tredici and Braak27]. The inflammatory hypothesis is based on the observation that the increased inflammatory markers in AD patients, such as pro-inflammatory cytokines interleukin-1β, interleukin-6, and tumor necrosis factor α [Reference Aberathne, Kulasiri and Samarasinghe28]. Increased oxidative stress in the AD brain is characterized by increased lipid peroxidation, increased protein, and DNA oxidation; these data support the oxidative stress hypothesis [Reference Markesbery29,Reference Pratico30]. Mitochondrial disturbances are also associated with both oxidative stress and AD plaque formation [Reference Aberathne, Kulasiri and Samarasinghe28].

The diagnosis of AD

In 2011, the NIA-AA conceptualized the pathophysiological process of AD in two stages, that is, preclinical stages and clinical stages based on whether the patients have overt symptoms of dementia and cognitive impairment [Reference Petersen, Wiste and Weigand7]. The diagnosis of AD in the clinical stage is clear in the guidelines, but the recommendations for preclinical AD which could potentially be decades before the symptoms become noticeable [Reference Monsell, Mock and Hassenstab31] were not established. Furthermore, accumulated evidence indicates that the neurodegeneration and cognitive decline in AD are a continuous pathophysiological process that should be regarded as a continuum rather than two distinct preclinical and clinical entities [Reference Jack, Bennett and Blennow6,Reference Dubois, Hampel and Feldman32]. This concept is recognized widely, indicating the importance of early AD diagnosis methods.

In 2011 NIA-AA guidelines, several methods have been established for identifying and assessing the progression of AD in clinical stages (Fig. 1). These methods can be clustered into two divisions: clinical assessments and neuroimaging techniques. The clinical assessments of AD rely on patient informant history, mental state examination, thorough physical examination, bedside cognitive testing including mini-mental state examination and Montreal cognitive assessment (MoCA), and some neuropsychologic and psychiatric tests such as memory, attention, language, problem-solving skills, and attention [Reference Eratne, Loi, Farrand, Kelso, Velakoulis and Looi33]. AD can be further diagnosed with firm clinical evidence provided by neuroimaging techniques, such as MRI and PET scans [Reference Islam and Zhang34,Reference Valotassiou, Malamitsi and Papatriantafyllou35], or direct biomarker tests including CSF analysis and blood tests [Reference Bouwman, Frisoni and Johnson36,Reference Schneider, Hampel and Buerger37]. All these traditional methods have a certain limitation, as these methods always work after the symptoms presented, such as memory loss and cognitive decline. Furthermore, these diagnostic techniques could be invasive procedures or require radiation exposure, posing a risk of side effects. Also, certain methods need specialized and costly equipment that is less available. Additionally, AD can only be noticed by families 4 years before a firm diagnosis is made, and this delay is often longer and includes misdiagnosis in the variants and early-onset groups [Reference Draper, Cations and White38].

Figure 1.

Overview of traditional diagnostic methods and early biomarkers in sensory systems. MMSE = mini-mental state examination; MoCA = Montreal cognitive assessment.

Structural overview of sensory pathways

Olfaction is a highly conservative primary sensory system [Reference Ache and Young47]. Olfactory can be clustered as the peripheral and central system [Reference Smith and Bhatnagar48]. Odorants are first received by olfactory sensory neurons (OSNs) located on the olfactory epithelium, where the chemical signals are converted to neuroelectrical signals [Reference Duchamp-Viret, Chaput and Duchamp49]. The electrical potentials are then transferred to the mitral cells and tuft cells on the surface of the olfactory bulb (OB) via OSNs projection [Reference Firestein50] and then handed over to olfactory cortical areas, including the piriform cortex (PCX), the medial olfactory cortex, and the amygdala [Reference Giessel and Datta51].

About half of the neuronal pathway in the brain subserves visual function, light signals are transduced to the neuronal signals via photoreceptors, the cones, and rods which are located on the outer nuclear layer of the retina. It is then passed to the inner nuclear layer, including [Reference Chaya, Matsumoto and Sugita52], followed by ganglion cells on the last layer. The axon of ganglion cells projects to the brain cortex via the lateral geniculate nucleus and superior colliculus and finally reaches the visual cortex.

The auditory system can be divided into two main components – peripheral and central systems – both are essential for normal hearing function. The peripheral auditory system includes external, middle, and inner ears; approximately 70% of hearing loss is caused by various pathologies in the inner ear and auditory nerve [Reference Strenzke, Pauli-Magnus, Meyer, Brandt, Maier and Moser53]. A specialized hearing organ in the inner ear, the cochlea, transduces the sound vibrations to neuronal electrical signals; this process is also known as mechanotransduction [Reference Stover and Diensthuber54]. The main sensory cells for mechanotransduction in the cochlea are hair cells. The ion homeostasis of the cochlea and hair cells (e.g., K⁺ and Ca²⁺) is required for normal hearing function and the development of the auditory system [Reference Schmiedt55–Reference Yang, Murtha and Climer58]. After mechanotransduction, the spiral ganglion neurons transmit the acoustic information from the cochlea to the central auditory system via the superior olivary complex, lateral lemniscus, inferior colliculus, medial geniculate nucleus, and the auditory cortex [Reference Jayakody, Friedland, Martins and Sohrabi59].

Preclinical research revealed early sensory biomarkers related to AD

AD animal models, primarily rodent models, have provided valuable insights into the disease and contributed to preclinical research. Since rodents do not develop AD, introducing human AD pathogenic genes is essential to model the AD pathology in mice. Thus, mouse models of AD can be generally classified as follows based on what gene is engineered, and each of these models is slightly different in AD pathophysiology. Here, we briefly summarized the pathogenesis of five main AD mouse models, including the APP transgenic model, APP/PS1 transgenic model, 5xFAD transgenic model, Tau transgenic model, and 3xTg mouse model.

Interestingly, numerous studies have also demonstrated that early sensory impairments are present in AD mouse models, highlighting the utility of these AD mouse models for exploring sensory deficits as potential early indicators of AD. The impairment of sensory systems can be assessed through behavioral tests, such as odor discrimination, identification, threshold test [Reference Hummel, Sekinger, Wolf, Pauli and Kobal60,Reference Zhang, Zhao and Sun61], Morris water maze (MWM) [Reference Lissner, Wartchow, Toniazzo, Goncalves and Rodrigues62,Reference Leinonen and Tanila63], acoustic startle response (ASR), and pre-pulse inhibition (PPI) tests [Reference Wendt, Kuhn, Hamm and Lonsdorf64]. Behavioral tests, however, may suffer from certain limitations such as variability in sensitivity and environmental factors and may not always directly correlate to the function being assessed. Electrophysiologic tests of sensory functions provide a noninvasive diagnosis [Reference Danjou, Viardot and Maurice65]. Here, we highlighted the earliest point at which the cerebral Aβ deposition, coupled with cognition decline, can be detected in each AD mouse model, which is, as previously described, the hallmark for the transition from the preclinical to clinical stages of AD (Table 1). We also summarized the earliest time points at which sensory impairments manifest, based on the primary research papers we have reviewed; this includes Aβ/P-tau aggregation in different sensory systems, functional-related behavioral deficits, and electrophysiologic changes within sensory systems including olfactory, auditory, and visual impairment (Table 2).

Table 1.

Earliest detection points of symptoms in AD mouse models

Table 1 Long description

A table comparing the earliest detection points of cerebral Aβ/P-tau and cognition decline in various Alzheimer’s disease (AD) mouse models. The table has four rows and three columns. The columns are labeled ’AD model mice’, ’Cerebral Aβ/P-tau’, and ’Cognition decline’. The rows are labeled with different AD mouse models: APP transgenic models (Tg2576), APP/PS1, 5xFAD, and Tau transgenic models (P301S). Each row lists the age in months at which cerebral Aβ/P-tau and cognition decline are first detected. For example, in APP transgenic models, cerebral Aβ/P-tau is detected at 6 months, and cognition decline is detected at 6 months. The table provides specific references for each detection point.

AD = Alzheimer’s disease; Aβ/P-tau = amyloid beta/hyperphosphorylated tau; APP/PS1 = amyloid precursor protein/presenilin 1; Mo = month of age.

Table 2.

Earliest detection points of sensory deficits in AD mouse models

Table 2 Long description

A table comparing the earliest detection points of sensory deficits in Alzheimer’s Disease (AD) mouse models across olfactory, visual, and auditory systems. The table has 4 rows and 12 columns. Column headers are: AD model mice, Olfactory APP/Aβ/P-tau aggregation, Olfactory Behaviors, Olfactory EP, Visual APP/Aβ/P-tau aggregation, Visual Behaviors, Visual EP, Auditory APP/Aβ/P-tau aggregation, Auditory Behaviors, Auditory EP. Row labels are: APP transgenic models (Tg2576), APP/PS1, 5xFAD, Tau transgenic models (P301S). Row 1: APP transgenic models (Tg2576), 2 Mo, 4 Mo, 3 Mo, 7.8 Mo, 9 Mo, NA, NA, NA, NA, NA. Row 2: APP/PS1, 3 Mo, 3 Mo, 3 Mo, 5 Mo, 8.5 Mo, 3 Mo, 2 Mo, 7 Mo, 2 Mo. Row 3: 5xFAD, 2 Mo, 3 Mo, 3 Mo, 6 Mo, 6 Mo, 3 Mo, 1 Mo, 2 Mo, 6 Mo. Row 4: Tau transgenic models (P301S), 1 Mo, 2 Mo, 3 Mo, 2 Mo, 2.5 Mo, NA, 4 Mo, 3 Mo, NA.

AD = Alzheimer’s disease; APP/Aβ/P-tau = amyloid precursor protein/amyloid beta/hyperphosphorylated tau; APP/PS1 = amyloid precursor protein/presenilin 1; EP = evoked potential; Mo = month of age; NA = not applicable.

Sensory deficits and aging in the symptomatic stage of AD

Transitioning from the exploration of sensory deficits in the research animal models to the symptomatic stage marks a critical juncture in understanding the disease’s progression. During the preclinical phase, subtle sensory changes may begin to emerge, often unnoticed or dismissed as benign age-related decline. However, as AD advances to its symptomatic stage, these sensory impairments become more pronounced, directly impacting patients’ quality of life and their ability to interact with the world around them (Table 3).

Table 3.

Sensory deficits in the symptomatic stage of Alzheimer’s disease

Table 3 Long description

A table with three columns and eight rows. The columns are labeled Sensory system, Sensory deficits, and Reference. The rows list different sensory systems and their associated deficits, along with references. Row 1: Vision, Difficulty reading, loss of contrast sensitivity, Lee et al. Row 2: Vision, Decreased visual field and/or contrast sensitivity and fixation problems, Ikram et al. Row 3: Vision, Changed topological organization of higher-level visual, Deng et al. Row 4: Vision, Loss of retinal ganglion cells, Kirby et al. Row 5: Olfactory, Increased olfactory threshold, Serby et al. Row 6: Olfactory, Smaller olfactory bulb volumes, Jobin et al. Row 7: Olfactory, Impaired olfactory identification, Vyhnalek et al. Row 8: Hearing, Cannot recognize or misinterpret environmental sounds, Dietz et al. Row 9: Hearing, Changes N200 and P300 latencies, Cintra et al.

A variety of ocular abnormalities have been observed in AD patients in the MCI stage, indicating a strong association between AD and visual impairment [Reference Papageorgiou, Tsirelis, Lazari, Siokas, Dardiotis and Tsironi130]. In the clinical stage, a large number of visual disorders and their association with Aβ burden have been reported in AD patients [Reference Golzan, Goozee and Georgevsky131], such as lower visual acuity in recognition of low luminance and low spatial frequency pictures [Reference Lee and Pai132], reduced visual contrast sensitivity [Reference Gilmore, Groth and Thomas133], altered color vision [Reference Kim, Ryou, Choi, Kim, Kim and Han134], retinal hemodynamic parameters [Reference Berisha, Feke, Trempe, Mcmeel and Schepens135], visual field loss [Reference Bateman136], and poor ocular motor function [Reference Crawford, Devereaux, Higham and Kelly137]. A deteriorated olfactory function can be observed in AD patients [Reference Dibattista, Pifferi, Menini and Reisert138]. Furthermore, olfactory dysfunction is associated with Aβ burden. Several studies have reported early olfactory impairment in AD patients also known as hyposmia [Reference Vyhnalek, Magerova and Andel139], where these patients exhibit a reduced ability to detect, discriminate, and identify odors, coupled with abnormal odor coding [Reference Yaffe, Freimer and Chen140,Reference Li, Howard and Gottfried141]. As many as 60% of older adults with AD and ADRD have hearing loss [Reference Powell, Reed and Wolff142]. All these sensory impairments could be used to evaluate the AD pathophysiology and should be taken into consideration for the AD clinical diagnosis. Multiple epidemiological studies have revealed that hearing loss is correlated with pathological cognitive decline and dementia [Reference Jayakody, Friedland, Martins and Sohrabi59,Reference Tarawneh, Menegola, Peou, Tarawneh and Jayakody143]. Hearing loss is a risk factor for cognition impairment. It has been reported that mild to severe hearing loss significantly increases the dementia risk [Reference Lin, Metter, O’Brien, Resnick, Zonderman and Ferrucci144,Reference Livingston, Huntley and Sommerlad145]. Elderly individuals with hearing loss symptoms have a higher risk of AD, which is also associated with loss of brainstem and cerebellar volume [Reference Llano, Kwok and Devanarayan146]. Furthermore, AD could cause a faster onset of hearing loss. Individuals’ AD risks can be evaluated using a polygenic risk score (PRS) [Reference Jansen, Savage and Watanabe147]. High PRS individuals are more likely to experience hearing difficulty than those with lower PRS [Reference Mitchell, Thorp, Evans, Nyholt, Martin and Lupton148]. However, the causation between AD and hearing loss is still controversial. Several hypotheses might explain the relationship between AD and hearing loss. First, hearing loss could cause cognitive decline and eventually dementia. Indeed, hearing loss induced by noise exposure exacerbates cognitive decline [Reference Paciello, Rinaudo and Longo149], and expressing Aβ derivatives in the cochlear hair cells leads to an early onset of hearing loss, especially in high-frequency regions, similar to age-related hearing loss [Reference Omata, Tharasegaran and Lim111]. Overexpression of tau synergistically worsens the Aβ-induced hearing defects [Reference Omata, Tharasegaran and Lim111]. These data indicate that Aβ and tau are both toxic to the peripheral auditory system, and overexpression of Aβ might accelerate the process of age-related hearing loss. Furthermore, hearing loss leads to less engagement in social and leisure activities which results in dementia. Lastly, both hearing loss and cognitive decline could be caused by a “common factor” [Reference Amieva and Ouvrard150] and might share the same pathway with age-related hearing loss, or presbycusis [Reference Chern and Golub151]. Despite the unclear causation between AD and auditory impairment, it has been reported that inducing gamma oscillations in the auditory cortex improved spatial and recognition memory and reduced Aβ in AC [Reference Martorell, Paulson and Suk152], indicating that the auditory pathway could be a potential target to ameliorate AD-associated pathology.

Noticeably, all the sensory impairments mentioned above could also be the comorbidities of the natural aging process [Reference Liu, Tan, Zhang, Yi, Zhu and Peng153]. This confluence of sensory impairments with aging complicates the differential diagnosis of AD from normal age-related changes, necessitating a nuanced understanding of pathological versus physiological decline. Moreover, the overlap of sensory impairments with AD is not uniquely induced by the aging process but could also be induced by other neurodegenerative diseases such as Parkinson’s disease (PD) and dementia with Lewy bodies (DLB) [Reference Lad, Sedley and Griffiths154,Reference O’Callaghan, Firbank, Tomassini, Schumacher, O’Brien and Taylor155]. Critical to advancing this differentiation is the longitudinal study of sensory deficits, utilizing both animal models and human subjects, to delineate the specific markers of AD-related sensory degradation. The challenge remains to disentangle the intricate web of aging and disease-specific sensory changes, requiring innovative methodologies and interdisciplinary approaches. Integrating multisensory assessments with cognitive evaluations and biomarker analysis could enhance the specificity of early AD diagnoses. Ultimately, this comprehensive approach will not only improve our understanding of AD progression but also pave the way for the development of targeted interventions to mitigate sensory deficits and improve the quality of life for individuals with AD.

Sensory impairments in preclinical AD patients

The impairment of the sensory system in AD patients may occur in the preclinical stage, well before the emergence of the onset of clinical symptoms (Fig. 1). Olfactory dysfunction is associated with a higher Aβ burden in older people. Early olfactory impairment in AD patients also known as hyposmia [Reference Vyhnalek, Magerova and Andel139], where patients exhibit a reduced ability to detect, discriminate, and identify odors, coupled with abnormal odor coding [Reference Yaffe, Freimer and Chen140,Reference Li, Howard and Gottfried141], elucidating the potential contribution of olfactory testing to detect preclinical AD [Reference Growdon, Schultz and Dagley156]. In clinically normal elderly individuals, worse olfactory function was associated with decreased hippocampal volume, thinner EC, and a trend associated with greater Aβ burden [Reference O’Callaghan, Firbank, Tomassini, Schumacher, O’Brien and Taylor155]. Another study showed that reduced odor identification (OI) was associated with lower cognitive score and older age and increased CSF tau and Aβ, revealing that OI could be an affordable biomarker of AD pathology [Reference Lafaille-Magnan, Poirier and Etienne157]. The visual deficit is considered a high-risk factor for AD and is increasingly supported by evidence linking visual system alterations to the early stage of AD [Reference Hawkins, Goyal and Sergio158]. For the preclinical stage, substantial visual deficits have been reported in AD patients [Reference Javaid, Brenton, Guo and Cordeiro44,Reference Katz and Rimmer159]. Potential measurable functional, structural, metabolic, and vascular changes have been identified in the retina during the early stages of AD [Reference Gupta, Chitranshi and den Haan160], indicating that vision could be another biomarker for early detection of AD. As another high-risk factor of AD, hearing loss was also associated with higher Aβ and tau burden. A study indicated that worsened hearing was related to the increased Aβ and tau burden [Reference Zheng, Yan and Hao161].

Above all, the evidence supporting sensory deficits as early indicators of AD suggests a paradigm shift in the approach to diagnosing and managing this complex condition in the preclinical stages.

The imperative of incorporating sensory detection

In our comprehensive review of early changes in various AD mouse models, notable early alternations take place in sensory systems prior to the cognition decline and the cerebral Aβ/P-tau aggregation. Here we listed out the early time point of alternation in different sensory systems using different AD mouse models (Table 2). This trend is consistent in olfactory, visual, and auditory systems across different AD models. Further, we also found a trend of hyperactivity status across sensory systems before the onset of neurodegeneration and cognition decline. The soluble Aβ oligomers induced hyperactivity on memory performance, cell death, epileptiform activity, gamma oscillations, and slow wave activity can be observed in both AD mouse models and human patients [Reference Hector and Brouillette162], consistent with early detection of soluble Aβ in sensory organs. The hyperactivity status of sensory functions, also known as network hyperexcitability [Reference Tok, Ahnaou and Drinkenburg163], could be a biomarker for AD.

However, we also found results may vary in different research due to the diversity of methodologies and sensory systems. Indeed, some tests have certain limitations. For example, the behavioral test can be affected easily by handling and environmental factors and can be stressful for animals [Reference Gouveia and Hurst164,Reference Fonio, Golani and Benjamini165]. Additionally, the accurate measurement and diagnosis of AD using sensory deficits may be complicated by other variables, including natural aging processes, PD, DLB, and various other neurodegenerative conditions, which can introduce significant variability into the assessment of sensory impairments. The variability and uncertainty indicate the necessity of integrating sensory function tests and the combination of sensory tests and other test methods to increase the accuracy of early AD diagnosis. To achieve this, multidisciplinary approaches that incorporate neuropsychological, neurobiological, and genetic analyses are essential. Implementing these strategies will improve the accuracy of early AD diagnosis through sensory deficit detection and contribute to a more nuanced understanding of the disease’s pathophysiology, for instance, the combination of vision and hearing deficits in older people experience higher rates of cognitive impairment [Reference Davidson and Guthrie166]. Thus, the simultaneous application of visual and auditory assessments could potentially enhance the precision of preclinical AD diagnosis, offering a more robust framework for identifying early indicators of cognitive impairment. Above all, sensory tests have demonstrated potential as supplementary tools for the diagnosis of AD and MCI. For instance, odor identification screening (Sniffin’ Sticks Odor Identification Test, SS-OIT) can improve the diagnostic accuracy of AD and MCI when combined with MoCA, providing additional information for clinical categorization of AD and MCI [Reference Quarmley, Moberg and Mechanic-Hamilton167]. Visuospatial measures show a significant diagnostic and prognostic potential in dementia, which means visuospatial tests could also improve the diagnosis of AD [Reference Salimi, Irish, Foxe, Hodges, Piguet and Burrell168]. Central auditory testing might be another suitable biomarker for identifying people at risk for dementia combined with electrophysiologic testing [Reference Gates, Anderson, Feeney, McCurry and Larson169].

Nevertheless, before these approaches can be adopted clinically, extensive validation steps are essential. This encompasses replicating findings across diverse studies to ensure reproducibility, alongside rigorous comparisons with established diagnostic benchmarks to clinical implementation. Additionally, defining precise diagnostic thresholds for sensory tests is crucial to delineate AD-related sensory impairments from those attributable to other causes. Moreover, evaluating the predictive value of these tests in foreseeing disease progression is essential. The pathway toward clinical adoption involves meticulous steps, including the elaboration of a clear context of use, submission of a comprehensive qualification package to regulatory bodies, and achieving a consensus on standards for test performance [Reference Cummings and Kinney170–Reference Goldsack, Coravos and Bakker173]. These processes are instrumental in ensuring the reliability, efficacy, and clinical relevance of sensory detection methods, thereby facilitating the integration into the diagnostic repertoire for early AD.

Integrating sensory change detection with artificial intelligence

Utilizing artificial intelligence (AI) in predicting AD is a rapidly evolving field. AI aids early detection and diagnosis based on a variety of machine learning-trained models. Based on the cognitive tests, imaging techniques, and CSF test, AI can improve the accuracy of traditional diagnostic methods, such as analyzing PET and MRI scan images [Reference Aberathne, Kulasiri and Samarasinghe28], and new biomarkers analysis beside Aβ42 and P-tau using the CSF database [Reference Chang, Lin and Lane174]. AI can also do end-to-end detection without using additional diagnostic methods. A pre-trained data2vec model can perform a self-supervised algorithm that works for speech, vision, and text [Reference Agbavor and Liang175]. An integrating pre-trained model that can access multimodal sensory test data would be a potential method for rapid, noninvasive AD diagnosis during the early detection window (Fig. 2).

Figure 2.

Schematic of alzheimer’s disease (AD) stages. Amyloid beta (Aβ) begins to accumulate many years before cognitive decline, marking the asymptomatic stage. This accumulation of Aβ starts to impair the sensory system’s function, including olfactory, visual, and auditory systems, before the onset of mild cognitive impairment (MCI). After the asymptomatic stage, some clinical diagnostic methods become applicable. However, by this point, patients start to experience cognitive decline, which, at this stage, is irreversible. Therefore, the window for early detection through sensory function tests is crucial. Integrating these tests with common clinical diagnostics and artificial intelligence could enhance the accuracy of early AD diagnosis before the onset of cognitive decline.