Identification of Retinal Biomarkers in Alzheimer’s Disease Using Optical Coherence Tomography: Recent Insights, Challenges, and Opportunities

This review will highlight recent insights into measuring retinal structure in Alzheimer’s disease (AD). A growing body of evidence indicates that disturbances in retinal blood flow and structure are related to cognitive function, which can severely impair vision. Optical coherence tomography (OCT) is an optical imaging technology that may allow researchers and physicians to gain deeper insights into retinal morphology and clarify the impact of AD on retinal health and function. Direct and noninvasive measurement of retinal morphology using OCT has provided useful diagnostic and therapeutic indications in several central nervous system (CNS) diseases, including AD, multiple sclerosis, and Parkinson disease. Despite several limitations, morphology assessment in the retinal layers is a significant advancement in the understanding of ocular diseases. Nevertheless, additional studies are required to validate the use of OCT in AD and its complications in the eye.


Introduction
Optical coherence tomography (OCT) has transformed the diagnosis and treatment of ocular disease. Currently, OCT is one of the most widely used ophthalmic decision-making technologies [1]. OCT uses retroreflected light to provide micron-resolution, cross-sectional images of biological tissues. The first experimental OCT device developed to image the human retina in vivo was introduced in 1991 [2]. OCT has become a compelling medical imaging technology, particularly in ophthalmology, as it allows achieving the cross-sectional structure of the retina and anterior eye with higher resolutions than any other non-invasive imaging modality [2]. Even commercially available OCT systems exhibit extremely high resolution, typically in the micrometer-scale resolution; and the resulting images can be analyzed both qualitatively and objectively. The recent introduction of OCT angiography (OCTA) and wide-field imaging in clinical practice has led to significant advancements in retinal disease understanding and treatment [1]. Further developments of OCT technology may impact eye disease diagnosis and improve the management of the major clinical and public health problems associated with visual impairment. Also, there is growing evidence to incorporate the OCT technology into clinical settings managing cerebrovascular and neural diseases [3].  bottom row) in each 10 × 10 grid cell within a macular area of 6 × 6 mm. Images were obtained with a 3D OCT-2000 unit (software version 8.11, Topcon Corp., Tokyo, Japan). The horizontal OCT B scans (right column) reveal the corresponding boundaries (green lines) of the inner retinal layers. The scanned macular area (7 × 7 mm) is shown on the left column. Image was taken with permission from Cunha et al. [20].   bottom row) in each 10 × 10 grid cell within a macular area of 6 × 6 mm. Images were obtained with a 3D OCT-2000 unit (software version 8.11, Topcon Corp., Tokyo, Japan). The horizontal OCT B scans (right column) reveal the corresponding boundaries (green lines) of the inner retinal layers. The scanned macular area (7 × 7 mm) is shown on the left column. Image was taken with permission from Cunha et al. [20].

General Findings
The initial search for this review found 143 studies from the PubMed database. After content analysis, 96 articles dating from 2001 to 2018 were assessed as significant. The exclusion of non-English language publications might constitute a limitation of this study; however, it is commonly applied in review articles. The articles included in this review focused on a broad range of specific objectives with retinal biomarkers being one of several. Table 1 shows the study details for the selected studies. In particular, this review revealed that limited research had focused exclusively on screening the eyes of study subjects with and with no cognitive decline using optical coherence tomography, neuropsychological tests, and in vivo neuroimaging techniques. All studies screened the eyes with the OCT technology after pupil dilation. Also, axial length and refraction data were not collected in many of the studies. Therefore, based on the potential increase of retinal thinning with greater axial length, it is unclear how this effect impacted much of results reported [21]. Fourteen studies used the time-domain Stratus OCT device (Carl Zeiss Meditec, Dublin, CA, U.S.A.). However, most of the research was conducted on spectral-domain OCT platforms including Spectralis SD-OCT (Heidelberg Engineering GmbH, Heidelberg, Germany), Cirrus HD-OCT (Carl Zeiss Meditec, Dublin, CA, U.S.A.), Topcon 3D OCT (Maestro, 1000 and 2000 Series, Topcon Medical Systems Inc., Tokyo, Japan) and RTVue-100 (Optovue Inc., Fremont, CA, USA). In general, the OCT measurements of the ganglion cell layer (GCL)-inner plexiform layer (IPL) and retinal nerve fiber layer (RNFL) present good intra-visit repeatability and inter-visit reproducibility [22]. Less than 9% of the studies used MRI or PET for the diagnosis in the AD group. In most of the studies, women were most commonly predominant in all cohorts.
The studies varied in their sample sizes and the number of participants, which ranged from 8 to 2124. In summary, the OCT results of 7148 participants were evaluated. Most of the criteria for pathological participant selection included mild cognitive impairment (MCI), subjective memory complaint, and early AD. Certain health conditions such as hypertension, diabetes, chronic heart failure, cardiac insufficiency, stroke, heart arrhythmia, age-related macular degeneration, and glaucoma were considered in the exclusion criteria in most articles. In addition to these conditions, much of studies conducted the Mini-Mental State Examination (MMSE) and other neuropsychological tests used across the National Institute on Aging (NIA) to determine the mental condition of the patients [23,24]. Three studies used the Montreal Cognitive Assessment (MoCA), along with the NIA tests and MMSE. Overall, the cognitive methods used in each study varied, which may also elucidate some of the differences. Of note, even the recent studies commonly did not follow the Advised Protocol for OCT Study Terminology and Elements (APOSTEL) recommendations [25]. These guidelines, conceived by a group of researchers and members of the International Multiple Sclerosis Visual (IMSVISUAL) consortium, were developed to highlight the essential information that should be provided when reporting quantitative OCT.
In general, the variability in the overall study design, the use of different generations of OCTs from various manufacturers with potential differences in resultant measurements, as well as the lack of standardization when analyzing OCT data within studies, makes the straightforward comparability of measurements challenging. Also, the retinal abnormalities in AD are complicated by the fact that both neurodegenerative and chronic diseases and AD are strongly age-related, and that, consequently, the indicators overlap, and attribution of retinal changes to one pathology rather than the other may be challenging. Therefore, a common trend in the studies reviewed was the analysis of retinal features between individuals along with the existing confounding variables (e.g., diabetes, hypertension, aging) that may also result in retinal changes. Moreover, much of the studies assessed in this review comprise only cross-sectional data, which is a critical limitation for the investigation of imaging biomarkers, particularly for disease progression and treatment success. Longitudinal studies will better facilitate the analyses of progressive changes, but a larger sample size might be needed to delineate disease stage differences as well as assess the sensitivity and specificity to change. Longitudinal data on retinal changes with healthy aging is also fundamental to characterize retinal abnormalities associated with cognitive impairment within a single individual over time. Ultimately, investigating associations between retinal changes and aging-related brain processes in neurologically healthy older adults could reveal whether AD-related pathologic changes occurring in the retina and central visual pathways are specific to axonal and ganglionic cell body injury.
Although most reviewed studies identified retinal abnormalities in AD patients such as reduced RNFL thickness, and degeneration of retinal ganglion cells (RGC), these deficits are not AD-specific and are seen in patients with other degenerative diseases such as diabetic retinopathy, glaucoma, and multiple sclerosis. In particular, a common finding in all studies was the thinning of the peripapillary RNFL as well as macular thinning of the GCL, IPL, and the ganglion cell complex (RNFL + GCL + IPL). Interestingly, Lad et al. conducted a more complex statistical analysis and revealed areas of thinning adjacent to areas of thickening in the macula of MCI and AD patients (Figure 4), suggesting that these retinal layers might be undergoing dynamic changes during the development of AD progression [26]. Also, it was found that not only the retinal structure of patients with AD had been investigated, but that the choroid thickness measured with enhanced depth SD-OCT was also found to be reduced [18,19]. Moreover, some recent studies have reported significant choroidal thinning in patients with AD when compared with elderly subjects [27,28]. This finding may aid in the diagnosis of Alzheimer's choroidopathy not related to age. Overall, these general findings are of extreme importance to understanding how changes in the retinal neurovascular unit reflect neurovascular pathology in the brain. Although most reviewed studies identified retinal abnormalities in AD patients such as reduced RNFL thickness, and degeneration of retinal ganglion cells (RGC), these deficits are not AD-specific and are seen in patients with other degenerative diseases such as diabetic retinopathy, glaucoma, and multiple sclerosis. In particular, a common finding in all studies was the thinning of the peripapillary RNFL as well as macular thinning of the GCL, IPL, and the ganglion cell complex (RNFL + GCL + IPL). Interestingly, Lad et al. conducted a more complex statistical analysis and revealed areas of thinning adjacent to areas of thickening in the macula of MCI and AD patients (Figure 4), suggesting that these retinal layers might be undergoing dynamic changes during the development of AD progression [26]. Also, it was found that not only the retinal structure of patients with AD had been investigated, but that the choroid thickness measured with enhanced depth SD-OCT was also found to be reduced [18,19]. Moreover, some recent studies have reported significant choroidal thinning in patients with AD when compared with elderly subjects [27,28]. This finding may aid in the diagnosis of Alzheimer's choroidopathy not related to age. Overall, these general findings are of extreme importance to understanding how changes in the retinal neurovascular unit reflect neurovascular pathology in the brain. Multi-variate regression analysis results obtained after investigating the association between the NFL and GCIPL layer thicknesses (i.e., GCL + IPL complex) to the disease categories of the participants: control, mild cognitive impairment (MCI) or Alzheimer's disease (AD) by using a quasileast squares technique, adjusted for multiple comparisons. In this analysis, a total of 17 regional thickness measurements for both NFL and GCIPL were used. Also, note that areas in the macula were statistically significantly thinner (red) or thicker (green). (Image was taken with permission from Lad et al. [26]). Multi-variate regression analysis results obtained after investigating the association between the NFL and GCIPL layer thicknesses (i.e., GCL + IPL complex) to the disease categories of the participants: control, mild cognitive impairment (MCI) or Alzheimer's disease (AD) by using a quasi-least squares technique, adjusted for multiple comparisons. In this analysis, a total of 17 regional thickness measurements for both NFL and GCIPL were used. Also, note that areas in the macula were statistically significantly thinner (red) or thicker (green). (Image was taken with permission from Lad et al. [26]). (2) objective memory decline, as defined by a delayed recall score on the Seoul Verbal Learning test (SVLT) less than 1.5 standard deviations (SD) below the age-and education-adjusted normative means; (3) normal general cognitive function, as defined by CDR scale of 0.5, and MMSE scores more than 1.5 SD below the age-and education-adjusted normative means; (4) normal functional activities; and (5) lack of a dementia diagnosis.
The Clinical Dementia Rating Scale-Sum of Boxes (CDR-SB) score presented negative relationships with the average GCIPL thickness and the GCIPL thickness in the superotemporal, superonasal, and inferonasal sectors. The composite memory score exhibited significant positive associations with the average GCIPL thickness and the GCIPL thickness in the superotemporal, inferonasal, and inferotemporal sectors. The temporal RNFL thickness, the average and minimum GCIPL thicknesses, and the GCIPL thickness in the inferonasal, inferior, and inferotemporal sectors at baseline were significantly reduced in MCI patients who were converted to AD compared to stable MCI patients. The change of CDR-SB from baseline to 2 years exhibited significant negative associations with the average and minimum GCIPL thicknesses as well as GCIPL thickness in the superotemporal, superior, superonasal, and inferonasal sectors at baseline. Data suggest that macular GCIPL thickness represents a promising biomarker for monitoring the progression of MCI and AD. A total of 45 eyes from (n = 24) patients with AD and 48 eyes from (n = 24) healthy controls.

YES
Each patient underwent a full neurological examination MMSE and MRI of the brain to rule out alternative diagnoses.
Most OCT peripapillary RNFL and macular full-thickness and segmented inner retinal layers parameters were reduced in AD compared to controls. Average, superior and inferior quadrant RNFL thickness parameters and all but one of the nine full-thickness macular measurements were significantly reduced in AD compared to controls. The segmented layers, GCL+ and GCL++ were significantly reduced in AD eyes. Significant correlation between most OCT parameters and MMSE scores, particularly in macular thickness. There was a significant reduction in peripapillary RNFL thickness and macular GCC thickness and a significant increase in the global loss volume (GLV) rate in both the NTG and AD patients when compared to the control subjects. The statistical evaluation showed no difference in any RNFL or GCC parameters between the AD and NTG groups. There was a negative correlation between disease duration and average RNFL and GCC thicknesses and a positive correlation between duration and GLV in the AD group. Thickness of RNFL in patient with AD was lower than that of controls. This suggests that SD-OCT has the potential to be used in the early diagnosis of AD as well as in the study of therapeutic agents. AD diagnosis according to DSM-IV criteria-(n = 54) subjects. Cognitively healthy age-matched volunteers were also examined as control subjects (n = 24)

YES
Ophthalmological evaluation included VA, IOP, slit lamp biomicroscopy and visual field examination. OCT measurements were performed by another ophthalmologist. All researchers were familiar with study protocol, but both ophthalmologists were blinded to cognitive status and diagnosis.
The total RNFL thicknesses were significantly different between the three groups: the RNFL was significantly thinner in the MCI Group compared to controls, as well as when the AD group was compared to MCI.

NO Zeiss Stratus
Gender and Race unknown (n = 17) subjects with AD, (n = 14) age-matched controls subjects.
The AD patients met the criteria for AD according to the NINCDS-ADRDA, DSM-IV and MMSE, MRI was performed.
AD patients have reduction of NFL thickness. This morphological abnormality is related to a retinal dysfunction as revealed by abnormal PERG responses.

OCT Findings and Normative Data
Concerning normative data, it was also found that less than a third of the studies used MRI or CT screening for healthy controls. Therefore, an important consideration is to carefully collect retinal imaging data from healthy subjects without subtle brain abnormalities. As the results of elderly patients are commonly analyzed, it is unknown whether the healthy subjects included in normative databases of retinal imaging devices may have discrete focal lesions (i.e., mild signs) in the brain without visible or detectable ocular manifestations ( Figure 5). Thus, caution should be exercised when considering normative data from retinal imaging, particularly if considering them for introducing new biological markers of non-ocular disease.

OCT Findings and Normative Data
Concerning normative data, it was also found that less than a third of the studies used MRI or CT screening for healthy controls. Therefore, an important consideration is to carefully collect retinal imaging data from healthy subjects without subtle brain abnormalities. As the results of elderly patients are commonly analyzed, it is unknown whether the healthy subjects included in normative databases of retinal imaging devices may have discrete focal lesions (i.e., mild signs) in the brain without visible or detectable ocular manifestations ( Figure 5). Thus, caution should be exercised when considering normative data from retinal imaging, particularly if considering them for introducing new biological markers of non-ocular disease.

OCT Findings Concerning Neuroimaging Features and Cognitive Tests
Current efforts are focused on detecting the presymptomatic phase (MCI) of the disease during clinical evaluations of healthy brain functioning. For example, Den Haan et al. showed that retinal thickness in early-onset AD correlated with parietal cortical atrophy [30]. Disease-modifying agents might be obtainable in the future while existing therapy for Alzheimer disease comprises enhancers of cholinergic function. Although medial temporal lobe atrophy was found to be a more valuable predictor of cognition than small-vessel disease in MCI [67], it is also essential to consider that "supposedly" healthy individuals with no systemic or chronic diseases may have beta-amyloid (Aβ) plaques in their brain and still have a normal healthy cognitive function. Also, not all MCI patients convert to AD [68]. It has also been reported that identifying the transition from the asymptomatic phase to symptomatic pre-dementia phase or from the symptomatic pre-dementia phase to dementia onset in the clinical setting is a non-trivial problem [69]. This matter creates a diagnostic uncertainty for the early stage of the disease. These caveats pointed out the need to identify new clinical biomarkers that could also be valuable in individuals without traditional risk factors (e.g., age, hypertension, diabetes, smoking habits, cardiovascular disease, hyperlipidemia (cholesterol, triglycerides). Most studies did not report quadrant-specific retinal OCT abnormalities. Those that found such a correlation revealed that peripapillary RNFL thickness in AD is lower in the superior and inferior quadrants than in nasal and temporal quadrants. It is uncertain why the superior and inferior quadrants could be favorably disturbed in some studies, but it might be because retinal

OCT Findings Concerning Neuroimaging Features and Cognitive Tests
Current efforts are focused on detecting the presymptomatic phase (MCI) of the disease during clinical evaluations of healthy brain functioning. For example, Den Haan et al. showed that retinal thickness in early-onset AD correlated with parietal cortical atrophy [30]. Disease-modifying agents might be obtainable in the future while existing therapy for Alzheimer disease comprises enhancers of cholinergic function. Although medial temporal lobe atrophy was found to be a more valuable predictor of cognition than small-vessel disease in MCI [67], it is also essential to consider that "supposedly" healthy individuals with no systemic or chronic diseases may have beta-amyloid (Aβ) plaques in their brain and still have a normal healthy cognitive function. Also, not all MCI patients convert to AD [68]. It has also been reported that identifying the transition from the asymptomatic phase to symptomatic pre-dementia phase or from the symptomatic pre-dementia phase to dementia onset in the clinical setting is a non-trivial problem [69]. This matter creates a diagnostic uncertainty for the early stage of the disease. These caveats pointed out the need to identify new clinical biomarkers that could also be valuable in individuals without traditional risk factors (e.g., age, hypertension, diabetes, smoking habits, cardiovascular disease, hyperlipidemia (cholesterol, triglycerides). Most studies did not report quadrant-specific retinal OCT abnormalities. Those that found such a correlation revealed that peripapillary RNFL thickness in AD is lower in the superior and inferior quadrants than in nasal and temporal quadrants. It is uncertain why the superior and inferior quadrants could be favorably disturbed in some studies, but it might be because retinal features such as axons and nerve fibers with bigger diameter degenerate more promptly and those features are more abundant in superior and inferior quadrants. A significant reason for the mixed results among these studies is the use of different OCT devices (e.g., time domain v. spectral domain) which appears to be the more significant variable causing heterogeneity in thickness measurements. Shi et al. reported that the inferior quadrant exhibited the strongest associations with results of cognitive tests, indicating that such results could show a higher risk to develop cognitive decline in older adults [55]. On the other hand, Uchida and associates revealed that cognitive testing scores correlated with the ellipsoid zone-retinal pigment epithelium volume [30].
Most of the studies employed the MMSE test. A drawback of the MMSE is its poor sensitivity for distinguishing MCI, which can be attributed to a lack of complexity as well as the absence of executive function items [70][71][72][73]. Its ceiling effect for healthy individuals is a common drawback that increases the likelihood that individuals in predementia phases score within the normal range (24 and above) [74,75]. MoCA is a more thought-provoking test that includes, higher-level language, executive function, and complex visuospatial processing to allow recognition of MCI with less ceiling effect [76,77]. A recent study reported that although MMSE and MoCA are more similar for dementia cases, the MoCA allocates MCI cases across a larger score range with less ceiling effect [78]. Therefore, MoCA might identify early and late MCI cases with more sensitivity than the MMSE. Also, it is crucial to consider that unusual presentation, or the existence of amnestic features in non-Alzheimer dementias could make to occasionally misdiagnosed AD as other dementias. Nonetheless, a thorough clinical history along with detailed neuroimaging assessment, and comprehensive mental evaluation will overcome this confusion.

OCT Findings and AD Biomarkers
The discovery of biomarkers is a difficult process that requires consideration of various factors and methods to obtain reliable biomarkers that could allow to predicting risk or response to treatment very early and with low false positive and false negative rates. To develop a biomarker-guided integrative AD modeling using precision medicine, the Alzheimer Precision Medicine Initiative group suggested applying a systematic neuropsychological and biological approach in exploratory translational neuroscience research on neurodegenerative diseases [79]. In ophthalmology, the continuous development of medical technology and methods to analyze and interpret ophthalmic data has facilitated the introduction of many retinal measures over time. For example, multiple optical-structural and functional parameters could be measured to characterize the retinal tissue, such as thickness, volume, optical scattering and polarization properties, texture measures, vasculature network metrics, blood flow dynamics, oximetry, electro-physiology metrics, fractal dimension, and lacunarity. Over the years, these retinal measures have been investigated and used as potential biomarkers of retinal abnormalities to elucidate potential applications in therapeutics [80]. Some aspects to be contemplated in the detection and assessment of biomarkers are the differences in phenotyping, the velocity of disease development in patients, the disease duration and impact of age, the subclinical ocular damage, the practicality of biomarkers to be confirmed in independent populations. Also, it should be noted that the prognosticators at the population level might not apply to all persons, and that large-scale studies are often needed. The goal would be to improve early detection by developing tools that aid in the identification of biomarkers that allow moving diagnosis backward in the temporal course.
In the search for AD biomarkers using OCT technology, more studies are necessary to assess whether choroidal, RNFL or GCL and IPL thickness represent an additional biomarker for the diagnosis and follow-up of AD pathology. The recent introduction of OCT angiography in Ophthalmology offers a novel approach to investigate retinal microvascular alterations in the foveal avascular zone and vessel density that may facilitate evaluating markers of non-perfusion and decreased capillary blood flow in the choriocapillaris as well as in the superficial and deep retinal vascular plexuses. These OCTA biomarkers could indicate microvascular network changes in the retina, mirroring changes in the cerebral microcirculation associated with cognitive function deterioration in the elderly. Feke et al. reported reduced blood flow in MCI patients in the presence of unchanged RNFL thickness [81]. Therefore, it might be presumed that blood flow abnormalities may precede the neurodegeneration in AD. Therefore, retinal hemodynamic measures could also be used as potential surrogate metrics for the cortex given the inaccessibility and the greater cost of imaging the vasculature of the cortex. Interestingly, retinal hemodynamics have not been practically explored concerning cognitive decline. Recently, contrary to previous understanding, the amyloid buildup is not the first sign of late-onset AD. Iturria-Medina et al. found that the first physiological sign of AD is a decrease in blood flow in the brain [82]. Also, another recent study reported that Polarization-Sensitive Optical Coherence Microscopy could assess amyloidosis based on intrinsic birefringent properties [83]. This finding could open a new venue for detecting Aβ plaques in the retina as another alternative for early diagnosis of AD. It is also essential to determine whether measurements investigated as potential biomarkers are repeatable and reproducible in patients with cognitive impairment and healthy controls to discard that OCT detected changes may be due to disease mechanisms or might be credited to measurement variability or typical progression due to aging and other factors not related to cognitive function decline [22].
The most recent research has identified Aβ deposits in eyes from both transgenic mouse models and human AD [84,85]. In another study, AD transgenic mice were administered curcumin, which binds to amyloid plaques. These studies confirmed the existence of amyloid deposits in the retina preceded amyloid plaque formation in the brain [84,85]. Both studies offer encouragement to the idea of imaging the retina of AD patients, as a potential biomarker that mirrors cerebral amyloid deposition. Cognoptix, Inc., and NeuroVision LLC. have been working on tests to detect amyloid beta plaques in the eye [86,87]. The Sapphire II test (Cognoptix, Inc., Concord, MA, U.S.) is based on detecting amyloid plaques in the crystalline lens using a fluorescent ligand marker applied topically to the inner eyelid of one eye (three ointment applications at home, 2 hours apart) the evening before the procedure. The Retinal Amyloid Index test (NeuroVision, LLC, Sacramento, CA, U.S.) detects amyloid plaques in the retina by administering a curcumin compound orally to the patient and imaging the patient's retina later with a device that is similar to a conventional retinal imaging scanner. NeuroVision's and Cognoptix's tests are promising, but they are not yet long-established as accurate, useful, and concrete using larger data.

OCT Findings in Relation to Advanced Imaging and Electrophysiological Techniques
Studies using the fundus images have revealed that major vessels in the retina are affected in patients with cognitive decline; however, advanced imaging technologies may be able to detect abnormalities in smaller vessels (e.g., capillaries). This potential advantage could make it possible to investigate subtle changes in static vascular parameters (e.g., capillary density) that may correlate with dynamic vascular parameters (e.g., pulse wave velocity and blood pressure) which have also been correlated with declining cognitive function [88][89][90][91]. There may also be the possibility to detect abnormal functional patterns of the retinal neurons by using electroretinography and other visual function tests. Notably, recent studies indicate that electrophysiological-based methodologies could become an alternate approach for the detection of abnormal neuronal activity and networking in the early stage of AD [92]. Interestingly, recent research reported that the scotopic response declined in C57/BL6 mice after subretinal injection of Aβ [93].
The integrity of the retinal tissue is not readily determined by fundus examination in the nonexistence of signs of vascular and structural abnormalities but can be assessed throughout both imaging analyses and functional testing. The confounding similarities between the pathogenesis of age-related macular degeneration (AMD), glaucoma and AD impose a significant challenge to early diagnose AD using changes in the retina. For example, AMD and AD share several pathogenic mechanisms such as oxidative stress and neuroinflammation, and also Aβ has been found to be located in drusen which are common findings in AMD [79]. While some studies have reported a non-significant correlation between retinal structures and AD severity [37,44,46,48,62], most have reported thinning of RNFL and GCL associated with AD severity [20,36,39,40,43,45,47,[49][50][51][52][53]59,64,65]. These findings are contradictory, pointing to the need to investigate not only static measures but dynamic metrics that may have a higher probability of being relevant in identifying those at risk. Considering that synaptic loss is one of the earliest functional signs of AD, as well as the best correlate of cognitive function decline [94,95], and that decreased a-and b-wave amplitudes were reported for mice carrying the ApoE-ε4 allele of apolipoprotein [96], ERG exploration could be a valid option to search for diagnostic markers of synaptic dysfunction within the retina that might be indicative of alterations in the brain [97].
Contemplating the above-mentioned findings, a multimodal integrative approach may offer a better venue to identify potential endpoints for neuroretinal impairment in patients with cognitive decline [97,98]. Also, it should be considered that different multimodal tests may be more useful for different stages of cognitive decline. Therefore, longitudinal studies would be important to determine how to best proceed with patient care and follow-up. Provided a clinical correlation between the brain and eye measures is established, screening of eyes in people considered at risk of AD could aid in the elaboration of a complementary low-cost approach for early diagnosis, as well as serving to monitor the effectiveness of developing therapies potentially.

Conclusions
The identification of retinal biomarkers in AD using OCT remains an area of active research. The main advantages of retinal imaging for diagnosing probable AD include its user-friendliness, low-cost, and the non-invasive nature of examinations. This review has identified a significant gap: the need for a clinically accessible framework to evaluate eye-brain measurements simultaneously across the lifespan. Crucially, this would enable baseline measurements before brain dysfunction happens and might allow the routine collection of retinal function metrics in the future, much like body temperature, pulse rate, and blood pressure measures today. However, successful longitudinal monitoring of eye-brain functional changes entails the establishment of accurate structural-functional baselines of eye-brain vitality before conditions of dysfunction. This framework should facilitate the translation of multimodal eye data into user-friendly metrics of brain function for broader clinical utilization, including the characterization of aging-related confounds. It is also worth noting that the use of big data and open-source models taking advantage of artificial intelligence are a critical step for more integrative, data-driven clinical studies to investigate into all the possible biological features involved, as well as the direct connections among these features. Also, it is crucial to improve the automation, ergonomics and user-friendliness of the designs of OCT technology for the elderly, who very frequently cannot follow instructions and complete an OCT examination successfully, making OCT assessment unreliable in this population.
In the coming decade, the aging research community, especially those investigating preclinical markers through the eye for a pathological cognitive decline, envisages to understanding better what eye-brain measures individually or collaboratively best predict future functional impairment. Also, longitudinal studies with larger sample sizes with a well-design stratified sampling and with a focus on the use of retinal measures by the geriatric population are needed, as this community group continues to grow and medical cost rise. A significant challenge will be to discover specific ocular biomarkers with sufficient sensitivity to reveal preclinical disorder and to monitor progression precisely. Ultimately, researchers will use the information from eye metrics to further examine the effectiveness of interventions in slowing or halting the development of cognitive function disorders.