Association between retinal neurodegeneration assessed by optical coherence tomography and PET-detected cerebral amyloid burden in Alzheimer's disease: A systematic review

Abstract

Background

Retinal imaging is proposed as a biomarker for Alzheimer's disease, but evidence linking retinal changes to cerebral amyloid remains inconsistent, particularly in preclinical populations.

Objective

This article evaluates whether retinal structural and microvascular changes measured by optical coherence tomography (OCT) and OCT-angiography (OCT-A) are associated with cerebral amyloid-β (Aβ) burden assessed by positron emission tomography (PET).

Methods

PubMed, Embase, Scopus, and Web of Science were searched on November 10, 2025. Two reviewers independently did screening, extraction, and quality assessment. Extracted outcomes included macular and peripapillary retinal nerve fiber layer and ganglion cell–inner plexiform layer thickness, vessel density, foveal avascular zone size, and PET tracers.

Results

Twenty-two studies including 1616 participants met inclusion criteria. Fifteen assessed OCT, five OCT-A, and two both. Among cognitively normal individuals, seven studies compared retinal nerve fiber layer thickness by Aβ status, with only one reporting a significant difference. Overall, OCT metrics showed no consistent differences by Aβ status. OCT-A results were similarly inconsistent, with one study linking larger foveal avascular zone to Aβ positivity. Substantial heterogeneity in imaging devices, segmentation methods, and PET quantification was observed, and few studies adjusted for these factors.

Conclusions

Current evidence does not demonstrate a consistent relationship between retinal OCT or OCT-A metrics and cerebral Aβ burden measured by PET. Methodological heterogeneity, small sample sizes, and limited statistical rigor contribute to inconclusive results. Standardized imaging protocols and longitudinal studies are required to determine whether retinal imaging can serve as a reliable non-invasive biomarker for early Alzheimer's disease.

Keywords

Alzheimer's disease amyloid-β neurodegeneration optical coherence tomography positron emission tomography retina

Introduction

Alzheimer's disease (AD) is the most prevalent type of dementia. AD is characterized by the accumulation of amyloid-β (Aβ)-containing senile plaques, primarily Aβ₄₂, and neurofibrillary tangles composed of hyperphosphorylated tau.¹ Diagnosis and monitoring of AD rely on both cognitive assessments and biomarkers, which provide critical insights into AD pathology..^2,3 Core biomarkers include Aβ and tau proteinopathies, and additional markers, such as neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP), help to capture broader neurodegenerative processes.⁴ Structural imaging, particularly MRI-detected hippocampal volume changes, further supports disease monitoring.⁵ Although positron emission tomography (PET) using Aβ-specific tracers can detect early Aβ deposition, its high cost and limited accessibility highlight the need for more available, non-invasive biomarkers.⁶

In this context, the retina emerges as a promising site for biomarker discovery. Because it is derived from the central nervous system (CNS), the retina shares fundamental structural and functional similarities with the brain, offering a distinctive vantage point for investigating cerebral pathology^7–9. Optical coherence tomography (OCT) and angiography (OCT-A) are increasingly recognized as valuable, non-invasive tools for detecting neurodegenerative changes associated with AD and mild cognitive impairment (MCI)^10–12. OCT enables high-resolution imaging of retinal layers, revealing thinning of the peripapillary retinal nerve fiber layer (pRNFL), ganglion cell–inner plexiform layer (GCIPL), and macular structures, changes that often correlate with the severity of cognitive decline.¹³ OCT-A extends this analysis by capturing microvascular alterations, including reduced vessel density in the superficial and deep capillary plexuses and enlargement of the foveal avascular zone (FAZ), which have been linked to cerebral small vessel disease and AD pathology.¹⁴ Recent meta-analyses and longitudinal studies suggest that retinal structural and vascular changes may precede measurable cognitive impairment, highlighting the potential of OCT and OCT-A as early biomarkers for AD and related dementias.¹⁵ These findings support the integration of retinal imaging in the broader framework of cognitive decline monitoring and AD risk stratification (Figure 1).

Figure 1.

Retinal structural changes detected by optical coherence tomography and amyloid deposition in the brain detected by positron emission tomography. OCT: optical coherence tomography; PET: positron emission tomography; BBB: blood-brain-barrier.

Despite growing interest in retinal imaging as a potential biomarker for AD, the existing literature remains heterogenous. Prior studies have reported inconsistent associations between retinal layer thickness and cerebral amyloid burden, often differing in study populations, cognitive staging, and OCT devices. Moreover, previous reviews have largely focused on established AD or mixed neurodegenerative cohorts, with limited emphasis on cognitively normal or preclinical individuals stratified based on the presence of early AD pathology, like amyloid status. Therefore, there remains a gap in contextualizing evidence specifically examining the relationship between retinal OCT changes and cerebral amyloid deposition across AD disease continuum. This systematic review aims to evaluate the association between retinal structural and microvascular alterations, as assessed by OCT and OCT-A, and cerebral Aβ deposition detected by PET in AD spectrum. By synthesizing current evidence, we seek to determine whether retinal structural and vascular changes provide a reliable, cost-effective, and non-invasive biomarker for early detection and longitudinal monitoring of AD, especially in preclinical and cognitively normal individuals with positive AD biomarkers, thereby complementing established cognitive assessments and other biomarker modalities.

Methods

This study was conducted in accordance with the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) (Supplemental Table 5)¹⁶ and was registered with PROSPERO (CRD 42023466410) prior to its commencement with an update on March 6, 2025. Covidence (https://www.covidence.org) was used in order to facilitate screening and data extraction.

Database and search strategy

A systematic search was carried out across PubMed, Embase, Scopus, and Web of Science by two independent reviewers (S.K. and M. F.) to identify studies assessing the association between retinal imaging changes (via OCT or OCT-A) and brain Aβ (via PET) in preclinical AD until November 10, 2025. The search strategy used combinations of terms related to dementia and AD, PET and its related terminology, and retinal imaging, with a particular focus of OCT and retina-related keywords. For the keywords used in search strategy refer to the (Supplemental Tables 1–4).

Inclusion and exclusion criteria

The inclusion criteria were as follows: Observational studies (including cohort and case-control studies) on human subjects published in in peer-reviewed journals. Studies were eligible if they included patients with confirmed Aβ status in the brain by PET imaging, with no restrictions on sex, ethnicity, or geographic location. Studies not published in English, published only as conference abstracts, or involving animal studies were excluded.

Two independent reviewers (S.K. and M.F.) screened the titles and abstracts of all retrieved records. The full texts of potentially eligible studies were then assessed according to the inclusion and exclusion criteria. Disagreements were resolved through discussion or, if necessary, consultation with a third reviewer (M.D.). Two reviewers (S.K. and M.F.) independently extracted data using a standardized data extraction form (Figure 2).

Figure 2.

PRISMA 2020 flow diagram depicting the study selection process for the systematic review.

Data extraction and outcomes

We extracted the following data from the studies: 1) Study characteristics: title, author, year, study design, sample size, total number of eyes, 2) Participant characteristics: age, sex, Aβ status, cognitive status, cognitive tests including Mini-Mental State Examination and Montreal Cognitive Assessment, and follow up time, 3) Device characteristics: type of OCT device, OCT and OCT-A protocol, tracer and dynamic versus static PET imaging. The primary outcomes of this study were differences in OCT imaging including retinal fiber layer (RNFL; prepapillary (p) or macular (m)), ganglion cell layer (GCL) and inner plexiform layer (IPL) (collectively GCIPL), total macular volume or thickness, and other retinal layers when relevant, such as retinal thickness, inner nuclear layer (INL), outer nuclear layer (ONL), macular thickness, FAZ, and vessel density (VD) between Aβ-positive and Aβ-negative individuals. The secondary outcomes of this study were the associations between OCT findings and cognitive function in Aβ-positive and Aβ-negative individuals.

Quality assessment and reporting

The risk of bias for each study was assessed independently by two reviewers using Newcastle-Ottawa Scale (NOS)¹⁷ for both cohort and cross-sectional studies, with studies scoring 0–3 considered low quality, 4–6 moderate quality, and 7–10 high quality. Data for non-significant findings are provided in the tables. For OCT analysis, RNFL measurements are annotated as macular (m) or peripapillary (p); for the rest of measurements, i.e., those not indicated as macular or peripapillary and any significant quadrant-specific values not detailed in the text, can be found in the corresponding table footnotes and legends.

Results

Study characteristics

After screening 323 studies, 22 studies were included in this systematic review, comprising 17 cross-sectional studies, 4 longitudinal cohort studies, and 1 case-control study. 15 studies only investigated the association between OCT metrics and Aβ status, 5 studies only investigated OCT-A, and 2 studies investigated both. The total sample investigating OCT metrics in Aβ+ and Aβ- cognitively normal individuals consisted of 92 Aβ+ with 116 eyes and 402 Aβ- cognitively normal subjects with 366 eyes (one of the studies did not report the number of the eyes).¹⁸ RNFL was reported in 16 studies, GCIPL in 13, INL and OPL in 5 each, ONL in 5 studies, macular thickness in 3 studies, and choroidal thickness in 2 studies^18–34. Seven studies^{23,32,35–39} investigated the association between OCT-A metrics and Aβ pathology. Four of them^35–38 only included cognitively normal individuals, two of them^23,39 only included cognitively impaired individuals and one study³² recruited both cognitively normal and impaired individuals. In total, 75 cognitively normal patients with Aβ positivity and 270 cognitively normal patients with Aβ negative status were included (one study did not report the number of patients in each subgroup). Given the limited number of available studies, along with heterogeneity in study designs and the stages of dementia among participants, the generalizability of this systematic review is limited.

Optical coherence tomography

Changes in retinal layer thicknesses in Aβ+ versus Aβ- cognitively normal individuals

Retinal nerve fiber layer: Six studies investigated RNFL (both mRNFL and pRNFL) thickness in cognitively normal individuals who were Aβ+ compared to those who were Aβ-. Marquie et al., Santos et al., Snyder et al., and Ravichandran et al. reported no significant differences in RNFL thickness between the two groups.^{18,19,24,25,27,31} In contrast, Byun et al. observed a significantly lower average RNFL thickness in Aβ+ individuals compared to healthy controls (HCs) (p = 0.006; Aβ+: 86.0 ± 31.5 µm, HCs: 104.8 ± 10.6 µm). RNFL thinning in Aβ+ individuals were significantly more pronounced in the superior (p = 0.04), temporal (p = 0.007), and inferior quadrants (p = 0.003), with the inferior quadrant showing the most marked difference. No significant difference was found in the nasal sector (p = 0.36).²⁷

Ganglion cell–inner plexiform layer: Byun et al., van de Kreeke et al., Santos et al., Snyder et al., and Marquie et al. found no significant differences in GCIPL thickness between Aβ+ and Aβ- individuals.^{18,24,25,27,31}

Inner nuclear layer: Two studies evaluated INL thickness in Aβ+ versus Aβ- individuals. Snyder et al. and Santos et al. did not report any significant differences in INL thickness between the groups.^24,25

Macular thickness: Byun et al. compared macular thickness in cognitively normal Aβ+ and Aβ- participants using swept-source OCT (SS-OCT, Topcon). They observed no significant differences in central fovea or outer ring (perifovea) thickness, (both p > 0.05), but the inner nasal portion of the inner ring was significantly lower in Aβ+ individuals (p = 0.012, Aβ+: 286.1 ± 22.5 µm; Aβ-: 308.9 ± 18.4 µm).²⁷

Table 1 shows an overview of the studies investigating the association between OCT findings and amyloid burden in cognitively normal individuals.

Table 1.

OCT-PET studies in cognitively normal individuals with amyloid status.

Author, year	Sample size (Aβ+/Aβ-)	Study design	Cognitive status	OCT device	Significant findings (Aβ+ versus Aβ-)	Key takeaways
Ravichandran et al., 2025¹⁹	18/14	Cross-sectional	Cognitively Normal	Spectralis (Heidelberg)	Putative retinal gliosis ↑	No significant differences in RNFL thickness. Possible early gliosis marker in Aβ+ individuals.
van de Kreeke et al., 2019¹⁸	18/147	Longitudinal	Cognitively Normal	Spectralis (Heidelberg)	No significant differences	Amyloid status did not affect rate of change in RNFL or GCIPL thickness over time.
Santos et al, 2018²⁴	15/41	Longitudinal	Cognitively Normal	Spectralis (Heidelberg)	OPL total volume ↓	Reduced outer plexiform layer volume in Aβ+. RNFL and GCIPL unchanged.
Snyder et al., 2016²⁵	10/53	Cross-sectional	Cognitively Normal	Spectralis (Heidelberg)	IPL volume ↑ (not significant after correction)	Initial increase in inner plexiform layer volume lost significance with multiple comparison correction. No differences in RNFL or GCIPL.
Byun et al, 2021²⁷	16/33	Cross-sectional	Cognitively Normal	Topcon	RNFL changes: Average RNFL ↓ Temporal RNFL ↓ Inferior RNFL ↓ Superior RNFL ↓ Macular changes: Inner inferior macula ↓ Inner nasal macula ↓	Significant thinning in multiple RNFL sectors and inner macular regions in Aβ+. GCIPL unchanged.
Marquie et al, 2020³¹	15/114	Longitudinal	Cognitively Normal	Topcon	Macular thickness ↑: Center Inner temporal Inner superior Inner nasal Inner inferior	Increased macular thickness across multiple inner regions in Aβ+. RNFL and GCIPL unchanged.

Aβ: amyloid-β; GCIPL: ganglion cell-inner plexiform layer; IPL: inner plexiform layer; OCT: optical coherence tomography; OPL: outer plexiform layer; PET: positron emission tomography; RNFL: retinal nerve fiber layer

↑ indicates increase; ↓ indicates decrease in Aβ+ compared to Aβ- groups. Sample sizes represent Aβ+/Aβ- participants.

Changes in retinal layer thicknesses in other subgroups (AD, MCI, preclinical AD)

Retinal nerve fiber layer: In symptomatic stages, the association between retinal thickness and amyloid burden remains inconsistent. Hadoux et al., den Haan et al., and Egle et al. reported no significant RNFL changes in Aβ+ patients with MCI or early onset AD (EOAD) compared to controls (p > 0.01)^21–23. However, Moon et al. compared pRNFL thickness across AD, MCI, and cognitively normal Aβ+ and Aβ− groups, observing significant quadrant-specific differences.³² López-de-Eguileta et al. reported that Aβ+ patient group consisting of MCI and AD subjects exhibited significantly lower overall RNFL thickness compared to HCs (p < 0.01, Aβ+ (AD + MCI): 97.7 ± 11.6 µm; Aβ- (HC): 99.9 ± 8.2 µm).³⁰

Ganglion cell-inner plexiform layer: Jorge et al., den Haan et al., and Egle et al. found no significant differences in GCL or IPL thickness between AD and HCs.^22,23,29 In contrast, López-de-Eguileta et al. observed lower GCL thickness in some quadrants in MCI and Aβ+ AD compared to HCs, where all GCL quadrants showed significantly lower thickness when the AD and MCI combined, were compared to HCs.³⁰ They also observed IPL thinning in some quadrants among Aβ+ individuals versus HCs as well as across all IPL quadrants in AD versus HCs.³⁰

Inner nuclear layer: Jorge et al. and den Haan et al. found no significant differences in INL thickness between AD and HCs.^22,29

Macular thickness: In the study by den Haan et al., macular volume did not demonstrate a significant association between macular thickness and Aβ burden.^23,40

Findings of longitudinal studies

Four studies investigated the correlation between longitudinal alterations in OCT/OCT-A metrics and Aβ burden in the brain, with three using OCT and one using OCT-A. Marquie et al. employed PET-MRI with an ¹⁸F-FFB tracer to assess the association between standardized uptake value ratio (SUVR) and retinal layer thicknesses. They found a significant positive correlation between global SUVR at baseline and the thickness of four macular subregions: the center of the macula, inner nasal, inner inferior, and outer inferior. However, only the inner nasal subregion demonstrated a significant association with FBB global SUVR when analyzing all macular layers together. They observed a correlation between increased global SUVR and increased nasal macular thickness both at baseline and after 24 months.³¹ van de Kreeke et al. found no significant correlation between the global non-displaceable binding potential (BPND) of [¹⁸F] flutemetamol and retinal layer alterations, except for the inner ring portion of IPL. They reported that higher BPND was associated with less thinning of IPL over 22 months of observation.²⁶ Santos et al. observed a significant decrease in RNFL thickness associated with neocortical Aβ accumulation after controlling for age, and a decrease in ONL and IPL volumes in preclinical AD compared to controls.²⁴ Curro et al. found no significant changes in vascular density or FAZ over time in Aβ+ and Aβ- groups.³⁸

Table 2 shows an overview of the studies investigating the association between OCT findings and amyloid burden in Alzheimer's disease spectrum.

Table 2.

OCT-PET studies in mixed cognitive status populations.

Author, year	Study groups (Sample sizes)	Study design	OCT device	Significant findings (Direction of change)	Key takeaways
Golzan et al., 2017²⁰	Control (NAL<1.4): n = 27 Preclinical AD (NAL≥1.4): n = 23 AD: n = 28	Cross-sectional	Nidek	No significant differences in RNFL or GCIPL	No OCT changes detected across cognitive spectrum or by amyloid status.
Hadoux et al., 2019²¹	MCI: n = 15 Cognitively Unimpaired: n = 20	Cross-sectional	Nidek	No significant differences in RNFL	RNFL thickness unchanged between MCI and controls.
Egle et al., 2024⁴¹	Greater amyloid: n = 76 Less amyloid: n = 74 (CU, MCI, dementia)	Cross-sectional	RTVue-XR Avanti (AngioVue)	No significant differences in RNFL, GCIPL, or OCTA metrics	Retinal vascular and structural measures unchanged by amyloid burden across cognitive stages.
den Haan et al, 2017²²	AD: n = 15 Cognitively Unimpaired: n = 15	Cross-sectional	Spectralis (Heidelberg)	No significant differences in RNFL or GCIPL	No structural retinal differences between AD and controls.
López-De-Eguileta et al., 2020²⁸	AD: n = 12 MCI: n = 51 Control: n = 63	Cross-sectional	Spectralis (Heidelberg)	Choroidal thickness (CT) ↓: Temporal (1500 μm) Superior (500 μm) Subfoveal (vertical scan) Inferior (1000 μm) Inferior (1500 μm)	Reduced choroidal thickness at multiple locations in AD/MCI versus controls. May reflect vascular changes.
Jorge et al, 2020²⁹	AD: n = 20 Cognitively Unimpaired: n = 17	Cross-sectional	Cirrus HD-OCT (Zeiss)	No significant differences in RNFL or GCIPL	No retinal structural changes detected in AD.
López-de-Eguileta et al., 2019³⁰	AD: n = 12 MCI: n = 51 Control: n = 63	Cross-sectional	Spectralis (Heidelberg)	RNFL thickness ↓: Average ↓ Temporal, temporal-superior, temporal-inferior ↓ GCL thickness ↓: All quadrants (superior, inferior, temporal, nasal) ↓ IPL thickness ↓: Superior, nasal ↓ ONL thickness ↓: Superior ↓	Widespread thinning across multiple retinal layers (RNFL, GCL, IPL, ONL) in AD/MCI, particularly in temporal RNFL and all GCL sectors.
Moon et al, 2023³²	AD: n = 27 MCI: n = 35 Control: n = 9	Cross-sectional	Cirrus SD-OCT (Zeiss)	MCI: circumpapillary RNFL at 9 o'clock position ↓	Focal RNFL thinning at 9 o'clock in MCI. Otherwise, no significant differences in RNFL or GCIPL.
Jin et al, 2025³³	AD: n = 40 MCI: n = 22 Control: n = 20	Cross-sectional	Not specified	GCIPL, INL, OPL, and MEZ correlated with hippocampal volume	Multiple retinal layers (GCIPL, INL, OPL, MEZ) associated with hippocampal atrophy, suggesting retinal changes parallel brain degeneration.
Sastim et al., 2024³⁴	AD: n = 31 Cognitively Unimpaired: n = 31	Cross-sectional	Spectralis (Heidelberg)	Regional RNFL thickness ↓ correlated with decreased PET metabolism	Selective RNFL thinning in AD correlates with reduced brain glucose metabolism, supporting retinal-brain connection.

AD: Alzheimer's disease; CT: choroidal thickness; CU: cognitively unimpaired; GCL: ganglion cell layer; GCIPL: ganglion cell-inner plexiform layer; INL: inner nuclear layer; IPL: inner plexiform layer; MCI: mild cognitive impairment; MEZ: myoid and ellipsoid zone; NAL: neocortical amyloid load; OCT: optical coherence tomography; OCTA: optical coherence tomography angiography; ONL: outer nuclear layer; OPL: outer plexiform layer; PET: positron emission tomography; RNFL: retinal nerve fiber layer; SUVR: standardized uptake value ratio

↓ indicates decrease in AD/MCI compared to control groups. Sample sizes represent total participants in each diagnostic category.

Optical coherence tomography angiography

Seven studies explored the association between OCT-A findings and brain Aβ burden, with four focusing on cognitively normal individuals and those across the AD spectrum. Among cognitively normal participants, van de Kreeke et al. reported significant inverse correlation between retinal VD in the inner ring macula, outer ring macula, and around the optic nerve head, but found no association with FAZ (p = inner ring macula: <0.01; outer ring macula: 0.02; around optic nerve head: 0.02).³⁵ O’Bryhim et al. identified a significantly enlarged FAZ in Aβ-positive individuals (Aβ+: 0.4 ± 0.1 mm; Aβ−: 0.3 ± 0.1 mm).³⁷ In contrast, Moon et al. found no significant changes in FAZ associated with Aβ burden in cognitively normal controls.³² Elahi et al. reported no significant differences in VD with respect to Aβ burden.³⁶

In the broader AD spectrum, Egle et al. demonstrated a significant correlation between reduced intermediate capillary plexus VD and increased Aβ PET signal, although no significant difference was detected between Aβ+ versus Aβ- groups after adjusting for confounders. In addition, no significant associations were found with FAZ, deep capillary plexus, or superficial vascular complex.⁴¹ Moon et al. noted a significant decrease in para- and peri-foveal temporal microcirculation in Aβ-positive subjects.³² Bermudez et al. reported significantly lower superficial and deep foveal, as well as superficial parafoveal capillary densities in Aβ-positive subjects, along with a significantly enlarged FAZ, while other macular and parafoveal regions showed no significant differences.³⁹

Quality assessment

We assessed the quality of the included articles using the NOS criteria for cross-sectional and cohort studies, which evaluate three aspects: sample selection, comparability of subjects, and outcome.¹⁷ As shown in Supplemental Table 6, the included studies received total scores ranging from a minimum of 6 to a maximum of 10. All of the studies received at least one star in each domain.

Discussion

In this systematic review we included 22 studies with a total of 1616 participants and 92 Aβ+ subjects and 402 Aβ- cognitively normal subjects undergoing OCT. RNFL and GCIPL were the most investigated layers among the studies. Overall, retinal alterations associated with Aβ pathology are subtle and heterogenous in preclinical stages. RNFL and GCIPL were the most frequently investigated retinal layers; however, the association with cerebral amyloid burden differed substantially according to disease stage.

Research has demonstrated significant retinal degeneration, neuroretinal thinning, and both structural and functional alterations in the retina of patients with AD⁴² Histopathological studies have identified hallmark AD pathologies, including Aβ₄₀, Aβ₄₂ plaques and pTau inclusions within the retina tissue, mirroring the pathological processes observed in the cerebral cortex of AD patients.⁴³ A large postmortem study examining paired retinal and brain samples from 86 human donors demonstrated significantly elevated retinal Aβ₄₂ levels in individuals with MCI, particularly in those with increased amyloid burden in the entorhinal and temporal cortices, regions critically involved in memory and perception. Moreover, retinal Aβ₄₂ concentrations were higher in AD dementia compared with MCI, suggesting progressive amyloid accumulation in parallel with disease severity.⁴⁴ Supporting this concept, PET imaging studies have shown a direct quantitative relationship between retinal amyloid burden and cerebral amyloid load, indicating that amyloid deposition in the retina and brain likely arises from a shared pathological cascade rather than independent process.⁴⁵ From a vascular and neurodegenerative perspective, disruption of the retinal microvasculature and accumulation of Aβ₄₀ and Aβ₄₂ within retinal capillaries, analogous to cerebral amyloid angiography, further reinforce the notion of a parallel brain-retina disease trajectory in AD. These vascular changes are accompanied by reduced platelet-derived growth factor receptor-β expression and progressive pericyte loss, compromising the integrity of both the blood-retina and blood-brain barriers and contributing to neurovascular dysfunction.⁴⁶ These shared neurodegenerative and vascular mechanisms provide a compelling rationale for using OCT as a surrogate marker of cerebral pathology. Thinning of specific retinal layers measured by OCT may reflect underlying neuronal loss, synaptic dysfunction, and microvascular compromise associated with cerebral amyloid deposition. Consistent with this concept, prior studies have reported a positive correlation between macular thickness and grey matter volume in patients with AD,⁴⁷ underscoring the potential of retinal imaging as a non-invasive, accessible biomarker for cerebral neurodegeneration and amyloid-related disease burden.

Among cognitively normal individuals, most cross-sectional studies did not demonstrate significant differences in global RNFL or GCIPL thickness between Aβ+ and Aβ- participants.^{18,19,24,25,27,31} Where differences were observed, they were typically region-specific rather than diffuse,²⁷ suggesting that early amyloid-related changes may be spatially selective rather than generalized.

IPL is a retinal region rich in cholinergic activity and dense synaptic connectivity between bipolar and amacrine cells, making it of particular interest in neurodegenerative research.²⁵ Accordingly, the GCIPL, encompassing both the GCL and IPL, has been a primary focus of retinal investigations in AD and its preclinical stages. However, evidence regarding GCIPL alterations appears to differ between presymptomatic and symptomatic disease stages.

In cognitively normal individuals with cerebral amyloid positivity (Aβ+), most cross-sectional studies did not demonstrate significant differences in GCIPL thickness compared with Aβ− controls. Several investigations, including those by Snyder et al.,²⁵ Byun et al.,²⁷ and others, reported preserved GCIPL thickness in this pre-symptomatic population, suggesting relative structural stability of ganglion cell bodies early in the disease course. Snyder et al. did, however, observe a non-significant increase in IPL thickness alongside an increased number and surface area of retinal inclusion bodies localized to this layer, which correlated with neocortical amyloid burden measured by PET imaging.²⁵ Importantly, these inclusion bodies were distinguishable from drusen, and participants with age-related macular degeneration were excluded, strengthening the specificity of the findings.²⁵ However, a meta-analysis of three studies showed thinner GCIPL layer in AD patients, which emphasizes the need for further evaluations of this layer in the preclinical stages.⁴⁸ Taken together, according to the study by Snyder et al. and other relevant studies, neurologic pathologies such as AD, should be considered when alterations in this layer, either decrease or increase, are identified.^25,49

In contrast, findings in symptomatic stages of AD and MCI are more suggestive of neurodegenerative change, albeit still heterogeneous. While several studies reported no significant GCIPL or IPL thinning in clinically diagnosed AD compared with healthy controls,^22,23,29 others identified quadrant-specific or layer-specific thinning, particularly when MCI and AD groups were analyzed together.^30,32 These discrepancies likely reflect differences in disease severity, analytic approaches, and cohort composition. Nonetheless, the emergence of GCIPL and IPL thinning predominantly in symptomatic cohorts supports the concept that structural ganglion cell loss may be a later manifestation, following earlier functional or microstructural alterations.

A similar stage-dependent pattern emerges when examining the RNFL, which contains the axons of ganglion cells.²⁷ In pre-symptomatic Aβ+ individuals, most studies reported no significant RNFL thinning, although one study identified quadrant-specific reductions, particularly in superior, temporal, and inferior regions.²⁷ These findings may reflect early axonal vulnerability in select retinal regions rather than diffuse degeneration. In symptomatic populations, RNFL thinning has been reported more consistently, especially when MCI and AD subjects were combined, aligning with established evidence of axonal degeneration and white matter involvement in clinically manifest AD.^50,51 Notably, prior neuroimaging studies indicate that axonal pathology may precede overt cognitive symptoms,⁵⁰ which could explain why subtle RNFL changes are occasionally detectable even in cognitively normal Aβ+ individuals.

Regional analyses further underscore biological differences between AD-related retinal changes and other neurodegenerative conditions. For example, correlations between cortical thickness and superior or supero-nasal GCIPL subfields, as reported by Byun et al.,²⁷ contrast with the inferior and inferotemporal GCIPL thinning classically associated with glaucoma. This regional dissociation suggests distinct pathogenic mechanisms and highlights the importance of subfield-specific analyses when interpreting retinal biomarkers in pre-symptomatic versus symptomatic disease.

With respect to overall macular or retinal thickness, selective layer-specific changes, such as localized IPL thickening²⁵ or RNFL thinning,²⁷ may offset one another, resulting in preserved global thickness measures, particularly in pre-symptomatic individuals. This likely contributes to the inconsistent findings across studies and reinforces the limitation of relying on composite retinal metrics for early disease detection.

OCTA findings further emphasize stage-dependent heterogeneity. In cognitively normal Aβ+ individuals, studies reported mixed results, including reduced vessel density,³⁵ enlarged FAZ,^35,37 or no detectable differences.^32,36 This variability may reflect dynamic microvascular responses during early amyloid accumulation rather than fixed vascular loss. In symptomatic AD and across the broader disease spectrum, OCTA findings suggest selective, region-dependent microvascular alterations,²³ particularly within parafoveal and perifoveal regions,^32,39 rather than a uniform vascular signature. Importantly, several studies found that associations between vascular density and amyloid burden weakened after adjustment for confounders,^23,32,36 underscoring the complexity of interpreting OCTA changes in later disease stages.

Longitudinal studies provide critical insight into these stage-related discrepancies. While baseline retinal thickness measures often showed weak or absent associations with amyloid burden, rates of change over time, particularly within the IPL, RNFL, and specific macular subregions, demonstrated stronger correlations with cerebral Aβ accumulation.^24,26,31 These findings suggest that dynamic retinal measures may be more informative than static cross-sectional assessments, especially for monitoring disease progression rather than identifying pre-symptomatic amyloid pathology.

Taken together, current evidence indicates that retinal alterations associated with cerebral amyloid burden differ meaningfully between presymptomatic and symptomatic stages of AD. In cognitively normal Aβ+ individuals, retinal changes appear subtle, region-specific, and potentially non-degenerative, whereas symptomatic stages are more often characterized by focal neurodegenerative and microvascular alterations. This distinction is critical for interpreting existing data and for guiding future studies aimed at developing stage-specific retinal biomarkers for AD.

The results of this review should be interpreted considering its drawbacks. First, we focused on the studies using PET scans to evaluate Aβ status and did not review the cerebrospinal fluid-based studies, which is a more invasive method. Second, due to the limited number of the relevant studies and substantial heterogeneity regarding the OCT layers and the cognition status of the populations in each study, a meta-analysis was not feasible. Third, due to the relatively high level of heterogeneity in the included population (like cognitive status) among the studies, we should remain cautious when generalizing the findings. In addition, some studies did not report the number of total eyes, which is another source of heterogeneity. The results of this study should be interpreted with caution as most of the included studies had cross-sectional design which limits their reliability in long-term outcomes. Also, there are other cofounders interfering with retinal imaging results including other ocular and general comorbidities, vascular risk factors, and APOE ε4 status that was not adjusted or taken into account in all of the studies. Lastly, it is generally recommended to report GCL and IPL as a single entity, GCIPL; however, only few studies in this systematic review followed this recommendation.

Conclusions and future prospective

Cerebral Aβ deposition begins between 15 to 20 years prior to the onset of the clinical manifestations of AD. This biomarker can be detected via PET scan, which is a non-invasive neuroimaging technique; however, the availability of this method remains limited, particularly in underprivileged areas. Hence, more readily available modalities, such as ophthalmologic imaging, should be evaluated to at least partially substitute less feasible methods.

Upon this review, it was noticed that while some studies suggest retinal structural changes such as GCL-IPL reduction as a reflection of Aβ levels in the CNS, the exact association between retinal changes and brain Aβ status remains unclear. Hence, the application of OCT as a stand-alone biomarker for the early phases of AD remains inconclusive; however, its potential to be incorporated into the screening protocol for AD in a multimodal evaluation setting should be evaluated. Future research should prioritize large-scale, longitudinal studies combining retinal OCT with amyloid PET imaging and cerebrospinal fluid biomarkers to clarify temporal relationships and improve pathological specificity. Standardization of OCT acquisition protocols, segmentation algorithms, and retinal layer definitions across devices is also essential to enhance reproducibility and cross-study comparability. In addition, incorporating advanced retinal imaging techniques such as OCT angiography, layer-specific analyses, and longitudinal rate-of-change metrics may help disentangle neurodegenerative from vascular contributions to retinal alterations. Detailed phenotyping of participants based on cognitive status, genetic risk factors including APOE ε4, and vascular comorbidities will be critical to refining patient stratification and identifying subgroups in whom retinal biomarkers are most informative. Overall, these approaches may improve the utility of retinal imaging as a non-invasive window into early AD-related neurodegeneration, potentially preceding overt clinical manifestations.

Supplemental Material

sj-docx-1-alz-10.1177_13872877261433198 - Supplemental material for Association between retinal neurodegeneration assessed by optical coherence tomography and PET-detected cerebral amyloid burden in Alzheimer's disease: A systematic review

Supplemental material, sj-docx-1-alz-10.1177_13872877261433198 for Association between retinal neurodegeneration assessed by optical coherence tomography and PET-detected cerebral amyloid burden in Alzheimer's disease: A systematic review by Sara KamaliZonouzi, Mobina Amanollahi, Melina Farshbafnadi, Mahsa Dolatshahi and Cyrus A. Raji in Journal of Alzheimer's Disease

Footnotes

Acknowledgements

Figure 1 for this article was created in BioRender. .

Amanollahi 2025: This study was presented in abstract form at the Alzheimer's Association International Conference (AAIC) 2025.

ORCID iDs

Sara KamaliZonouzi

Mobina Amanollahi

Mahsa Dolatshahi

Cyrus A. Raji

Ethical considerations

Not applicable

Consent to participate

Not applicable

Consent for publication

Not applicable

Author contribution(s)

Sara KamaliZonouzi: Conceptualization; Data curation; Formal analysis; Investigation; Project administration; Resources; Software; Visualization; Writing – original draft.

Mobina Amanollahi: Data curation; Writing – original draft.

Melina Farshbafnadi: Writing – original draft.

Mahsa Dolatshahi: Validation; Writing – review & editing.

Cyrus A. Raji: Validation; Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr. Raji has received grant support from the NIA R01AG072637 (P.I. Cyrus Raji, MD, PhD), NIA R01AG070883 (P.I. Amy Kind, MD, PhD), and NIA R01AG079241 (P.I. Jessica Alber, PhD).

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Cyrus Raji is an Editorial Board Member of this journal but was not involved in the peer-review process of this article nor had access to any information regarding its peer-review. The remaining authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

This systematic review is based entirely on previously published, peer-reviewed articles that are publicly available through academic databases.

Supplemental material

Supplemental material for this article is available online.

References

Lee

Jiang

McIlmoyle

, et al. Amyloid Beta immunoreactivity in the retinal ganglion cell layer of the Alzheimer’s eye. Front Neurosci 2020; 14: 758.

Watanabe

Kamijo

Nishino

, et al. The feasibility of using the go/no-go task as a dementia screening test assessed with a cross-sectional design. Sci Rep 2024; 14: 29834.

Kueper

Speechley

Montero-Odasso

. The Alzheimer’s disease assessment scale-cognitive subscale (ADAS-Cog): modifications and responsiveness in pre-dementia populations. A narrative review. J Alzheimers Dis 2018; 63: 423–444.

Park

Paneque

, et al. Tau, glial fibrillary acidic protein, and neurofilament light chain as brain protein biomarkers in cerebrospinal fluid and blood for diagnosis of neurobiological diseases. Int J Mol Sci 2024; 25: 6295.

Tosun

Schuff

Jagust

, et al. Discriminative power of arterial spin labeling magnetic resonance imaging and 18F-fluorodeoxyglucose positron emission tomography changes for amyloid-β-positive subjects in the Alzheimer’s disease continuum. Neurodegener Dis 2016; 16: 87–94.

Kaštelan

Gverović Antunica

Puzović

, et al. Non-invasive retinal biomarkers for early diagnosis of Alzheimer’s disease. Biomedicines 2025; 13: 283.

London

Benhar

Schwartz

. The retina as a window to the brain-from eye research to CNS disorders. Nat Rev Neurol 2013; 9: 44–53.

Kleerekooper

Houston

Dubis

, et al. Optical coherence tomography angiography (OCTA) in multiple sclerosis and neuromyelitis optica spectrum disorder. Front Neurol 2020; 11: 604049.

Alber

Goldfarb

Thompson

, et al. Developing retinal biomarkers for the earliest stages of Alzheimer’s disease: what we know, what we don’t, and how to move forward. Alzheimers Dement 2020; 16: 229–243.

10.

Huang

Swanson

Lin

, et al. Optical coherence tomography. Science 1991; 254: 1178–1181.

11.

Ibrahim

Xie

Macerollo

, et al. A systematic review on retinal biomarkers to diagnose dementia from OCT/OCTA images. J Alzheimers Dis Rep 2023; 7: 1201–1235.

12.

Biscetti

Lupidi

Luchetti

, et al. Novel noninvasive biomarkers of prodromal Alzheimer disease: the role of optical coherence tomography and optical coherence tomography-angiography. Eur J Neurol 2021; 28: 2185–2191.

13.

Coppola

Di Renzo

Ziccardi

, et al. Optical coherence tomography in Alzheimer’s disease: a meta-analysis. PLoS One 2015; 10: e0134750.

14.

Kutz

Obergassel

Mack

, et al. Retinal vessel density and cognitive function in healthy older adults. Exp Brain Res 2025; 243: 114.

15.

Wang

Chen

, et al. Association between retinal vascular geometric changes and cognitive impairment: a systematic review and meta-analysis. J Clin Neurol 2020; 16: 19–28.

16.

Page

Moher

Bossuyt

, et al. PRISMA 2020 Explanation and elaboration: updated guidance and exemplars for reporting systematic reviews. Br Med J 2021; 372: n160.

17.

Wells

Shea

O’Connell

, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp (2011).

18.

van de Kreeke

Nguyen

H-T

den Haan

, et al. Retinal layer thickness in preclinical Alzheimer’s disease. Acta Ophthalmol 2019; 97: 798–804.

19.

Ravichandran

Snyder

Alber

, et al. Association and multimodal model of retinal and blood-based biomarkers for detection of preclinical Alzheimer’s disease. Alzheimers Res Ther 2025; 17: 19.

20.

Golzan

Goozee

Georgevsky

, et al. Retinal vascular and structural changes are associated with amyloid burden in the elderly: ophthalmic biomarkers of preclinical Alzheimer’s disease. Alzheimers Res Ther 2017; 9: 13.

21.

Hadoux

Hui

Lim

JKH

, et al. Non-invasive in vivo hyperspectral imaging of the retina for potential biomarker use in Alzheimer’s disease. Nat Commun 2019; 10: 4227.

22.

den Haan

Janssen

van de Kreeke

, et al. Retinal thickness correlates with parietal cortical atrophy in early-onset Alzheimer’s disease and controls. Alzheimers Dement (Amst) 2017; 10: 49–55.

23.

Egle

Hamedani

Deal

, et al. Retinal microstructure and microvasculature in association with brain amyloid burden. Brain Commun 2025; 7: fcaf013.

24.

Santos

Johnson

Sinoff

, et al. Change in retinal structural anatomy during the preclinical stage of Alzheimer’s disease. Alzheimers Dement (Amst) 2018; 10: 196–209.

25.

Snyder

Johnson

Lim

, et al. Nonvascular retinal imaging markers of preclinical Alzheimer’s disease. Alzheimers Dement (Amst) 2016; 4: 169–178.

26.

van de Kreeke

Nguyen

Konijnenberg

, et al. Longitudinal retinal layer changes in preclinical Alzheimer’s disease. Acta Ophthalmol 2021; 99: 538–544.

27.

Byun

Park

Lee

, et al. Association of retinal changes with Alzheimer disease neuroimaging biomarkers in cognitively normal individuals. JAMA Ophthalmol 2021; 139: 548–556.

28.

López-De-Eguileta

Lage

López-García

, et al. Evaluation of choroidal thickness in prodromal Alzheimer’s disease defined by amyloid PET. PLoS One 2020; 15: e0239484.

29.

Jorge

Canário

Martins

, et al. The retinal inner plexiform synaptic layer mirrors grey matter thickness of primary visual cortex with increased amyloid β load in early Alzheimer’s disease. Neural Plast 2020; 2020: 8826087.

30.

López-de-Eguileta

Lage

López-García

, et al. Ganglion cell layer thinning in prodromal Alzheimer’s disease defined by amyloid PET. . Alzheimers Dement (N Y) 2019; 5: 570–578.

31.

Marquie

Valero

Castilla-Marti

, et al. Association between retinal thickness and β-amyloid brain accumulation in individuals with subjective cognitive decline: fundacio ACE Healthy Brain Initiative. Alzheimers Res Ther 2020; 12: 37.

32.

Moon

Jeon

Seo

, et al. Comparison of retinal structural and neurovascular changes between patients with and without amyloid pathology. J Clin Med 2023; 12: 1310.

33.

Jin

Wang

Lang

, et al. Retinal optical coherence tomography intensity spatial correlation features as new biomarkers for confirmed Alzheimer’s disease. Alzheimers Res Ther 2025; 17: 33.

34.

Sastim

Elboga

, et al. Evaluation of the relationship between FDG-PET hypometabolism and retinal layer thickness in patients with Alzheimer’s disease. Acta Neurol Belg 2024; 124: 987–993.

35.

van de Kreeke

Nguyen

H-T

Konijnenberg

, et al. Optical coherence tomography angiography in preclinical Alzheimer’s disease. Br J Ophthalmol 2020; 104: 157–161.

36.

Elahi

Ashimatey

Bennett

, et al. Retinal imaging demonstrates reduced capillary density in clinically unimpaired APOE ε4 gene carriers. Alzheimers Dement (Amst) 2021; 13: e12181.

37.

O’Bryhim

Apte

Kung

, et al. Association of preclinical Alzheimer disease with optical coherence tomographic angiography findings. JAMA Ophthalmol 2018; 136: 1242–1248.

38.

Curro

van Nispen

RMA

den Braber

, et al. Longitudinal assessment of retinal microvasculature in preclinical Alzheimer’s disease. Invest Ophthalmol Vis Sci 2024; 65: 2.

39.

Bermudez

Lesnick

, et al. Optical coherence tomography angiography retinal imaging associations with burden of small vessel disease and amyloid positivity in the brain. J Neuroophthalmol 2025; 45: 63–70.

40.

Haan

van de Kreeke

Konijnenberg

, et al. Retinal thickness as a potential biomarker in patients with amyloid-proven early- and late-onset Alzheimer’s disease. Alzheimers Dement (Amst) 2019; 11: 463–471.

41.

Egle

Deal

Walker

, et al. Association between retinal microvascular abnormalities and late-life brain amyloid-β deposition: the ARIC-PET study. Alzheimers Res Ther 2024; 16: 100.

42.

Liao

Zhu

Peng

. Potential utility of retinal imaging for Alzheimer’s disease: a review. Front Aging Neurosci 2018; 10: 188.

43.

Hart de Ruyter

Morrema

THJ

den Haan

, et al. Phosphorylated tau in the retina correlates with tau pathology in the brain in Alzheimer’s disease and primary tauopathies. Acta Neuropathol 2023; 145: 197–218.

44.

Koronyo

Rentsendorj

Mirzaei

, et al. Retinal pathological features and proteome signatures of Alzheimer’s disease. Acta Neuropathol 2023; 145: 409–438.

45.

Dumitrascu

Lyden

Torbati

, et al. Sectoral segmentation of retinal amyloid imaging in subjects with cognitive decline. Alzheimers Dement (Amst) 2020; 12: e12109.

46.

Williams

McGowan

Cardwell

, et al. Retinal microvascular network attenuation in Alzheimer’s disease. Alzheimers Dement (Amst) 2015; 1: 229–235.

47.

Zhao

Yan

, et al. The correlation of retinal neurodegeneration and brain degeneration in patients with Alzheimer’s disease using optical coherence tomography angiography and MRI. Front Aging Neurosci 2023; 15: 1089188.

48.

Chan

VTT

Sun

Tang

, et al. Spectral-domain OCT measurements in Alzheimer’s disease: a systematic review and meta-analysis. Ophthalmology 2019; 126: 497–510.

49.

Liu

Y-L

Hsieh

Y-T

Chen

T-F

, et al. Retinal ganglion cell-inner plexiform layer thickness is nonlinearly associated with cognitive impairment in the community-dwelling elderly. Alzheimers Dement (Amst) 2019; 11: 19–27.

50.

Pillai

Bermel

Bonner-Jackson

, et al. Retinal nerve fiber layer thinning in Alzheimer’s disease: a case-control study in comparison to normal aging, Parkinson’s disease, and non-Alzheimer’s dementia. Am J Alzheimers Dis Other Demen 2016; 31: 430–436.

51.

Fortune

Cull

Reynaud

, et al. Relating retinal ganglion cell function and retinal nerve fiber layer (RNFL) retardance to progressive loss of RNFL thickness and optic nerve axons in experimental glaucoma. Invest Ophthalmol Vis Sci 2015; 56: 3936–3944.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB