Highlight
In a cross-sectional study of 103 participants, swept-source optical coherence tomography angiography (SS-OCTA) identified distinct retinal and choriocapillaris signatures across normal cognition, mild cognitive impairment (MCI), and Alzheimer disease (AD) dementia.
Ganglion cell complex (GCC) thickness was lower in AD dementia than in cognitively normal controls, suggesting structural neurodegeneration detectable in the retina.
Retinal vessel skeleton density (VSD) and choriocapillaris flow deficit (CCFD) were associated with cognitive status, with a biphasic pattern in CCFD that may reflect early compensatory hyperperfusion followed by later perfusion failure.
These findings support OCTA as a promising, noninvasive biomarker platform for cognitive neurodegeneration, but larger longitudinal studies are needed before clinical adoption.
Study background and unmet need
Alzheimer disease remains one of the most common causes of dementia worldwide, yet diagnosis is still often delayed and can depend on costly or invasive testing, including positron emission tomography, lumbar puncture for cerebrospinal fluid biomarkers, and specialist cognitive evaluation. As the population ages, there is increasing interest in accessible screening tools that can detect biologic changes earlier in the disease course.
The retina is an extension of the central nervous system and shares embryologic, vascular, and neurodegenerative features with the brain. This has made ocular imaging an attractive candidate for Alzheimer biomarker research. Optical coherence tomography angiography (OCTA) can visualize retinal and choroidal microvasculature without dye injection, offering a practical, noninvasive method to assess microvascular integrity. The question addressed by this study is whether OCTA-derived metrics differ meaningfully across cognitive stages and whether they might help discriminate normal aging from prodromal and dementia-stage disease.
Study design
This was a cross-sectional study enrolling 103 individuals referred from the University of Washington Alzheimer’s Disease Research Center (ADRC) between April 2022 and September 2024. Participants were classified by ADRC research-criteria evaluations into three groups: 49 cognitively normal controls, 29 with mild cognitive impairment, and 25 with Alzheimer disease dementia. All participants underwent swept-source OCTA imaging.
The main outcomes were cognitive status, retinal vessel skeleton density (VSD), choriocapillaris flow deficit (CCFD), and ganglion cell complex (GCC) thickness. Statistical analyses examined between-group differences and multivariable associations with cognitive status. The report also included model performance in a 21-participant test set, suggesting exploratory classification potential.
Because this was a cross-sectional analysis, the study can identify associations but cannot establish temporal sequence or causality. It is therefore best interpreted as hypothesis-generating rather than diagnostic validation.
Key findings
1) Structural retinal neurodegeneration was most evident in AD dementia
Among the structural measures, GCC thickness was significantly lower in AD dementia than in controls after adjustment. The adjusted mean GCC was 63.31 μm in AD dementia compared with 67.93 μm in controls, a difference of -4.62 μm (95% CI, -8.92 to -0.31 μm; P = .03). This finding is consistent with retinal ganglion cell and inner retinal layer thinning in more advanced cognitive disease.
GCC thickness is clinically relevant because it reflects inner retinal neuronal tissue, including ganglion cells and their synaptic layers. In neurodegenerative disorders, such thinning may be a surrogate of axonal and neuronal loss, although its specificity for Alzheimer disease remains limited.
2) Choriocapillaris flow deficit showed a non-linear pattern across disease stages
One of the most interesting findings was the biphasic behavior of CCFD. Adjusted mean CCFD was lower in MCI (8.12%) than in AD dementia (9.07%), with a difference of -0.95% (95% CI, -1.71 to -0.19; P = .01). At the same time, CCFD was higher in AD dementia than in controls (8.33%), with a difference of 0.74% (95% CI, 0.02-1.46; P = .04).
In the authors’ interpretation, this pattern may suggest early compensatory choriocapillaris hyperperfusion or relative flow preservation during MCI, followed by perfusion failure in established dementia. That mechanistic explanation is plausible but remains inferential. Alternative explanations include stage-related vascular remodeling, systemic vascular comorbidity, or measurement effects intrinsic to OCTA imaging.
3) Retinal vessel skeleton density declined with worsening cognitive status
In multivariable models, VSD was significantly associated with cognitive status. The reported odds ratios were 0.79 for MCI (95% CI, 0.77 to 0.81) and 0.66 for AD dementia (95% CI, 0.65 to 0.68), with P < .001. Although the abstract presents these values in a classification framework rather than as simple group means, the direction of effect indicates reduced retinal microvascular density in more impaired cognitive states.
VSD is a vessel-caliber-independent measure derived from the skeletonized retinal vasculature, intended to reflect capillary network density more robustly than raw vessel area alone. Decreases in VSD may therefore indicate rarefaction or loss of retinal microcirculation, a feature increasingly linked to cerebrovascular and neurodegenerative pathology.
4) OCTA-based models showed moderate to good discrimination in a small test set
The study reported an area under the curve ranging from 0.72 to 0.87 in a 21-participant test set consisting of 10 controls, 6 patients with MCI, and 5 with AD dementia. This suggests that OCTA-derived biomarkers may have meaningful classification value, at least within the study’s internal validation framework.
However, these performance estimates must be interpreted cautiously. Small test sets can produce unstable metrics, and the risk of overfitting is substantial when the total sample size is limited. External validation in independent cohorts will be essential before any real-world screening application can be considered.
Clinical interpretation
This study adds to a growing body of evidence that Alzheimer disease is not only a brain disorder but also a systemic neurovascular disorder with measurable retinal correlates. The retina offers a unique imaging window into neuronal and microvascular health, and SS-OCTA may capture both structural degeneration and perfusion abnormalities in a single examination.
From a clinical standpoint, the most compelling signal in this study is not that OCTA can diagnose Alzheimer disease on its own, but that it may help stratify neurodegenerative risk in a low-burden, noninvasive way. If validated, OCTA could complement cognitive testing, fluid biomarkers, and neuroimaging by identifying individuals who warrant more definitive evaluation.
That said, OCTA biomarkers are unlikely to be disease-specific. Similar microvascular or retinal nerve fiber changes can occur with glaucoma, diabetes, hypertension, age-related vascular disease, and other neurodegenerative conditions. Any future screening strategy will need careful adjustment for ocular comorbidity, refractive status, image quality, and systemic vascular risk.
Expert commentary
Several strengths support the credibility of the study. First, the participants were evaluated through an Alzheimer Disease Research Center framework, which increases diagnostic rigor relative to routine administrative coding. Second, the use of swept-source OCTA provides deeper penetration and potentially better visualization of choroidal and choriocapillaris layers than earlier OCTA systems. Third, the simultaneous assessment of retinal, choroidal, and ganglion cell metrics is biologically appealing because Alzheimer-related pathology may not be confined to one ocular layer.
Still, important limitations temper enthusiasm. The study was cross-sectional, so it cannot determine whether OCTA changes precede cognitive decline or simply track established disease. The sample was small and derived from a single center, limiting generalizability. The abstract does not provide granular detail on vascular comorbidities, medication exposure, ocular disease exclusions, or ethnic diversity, all of which could influence OCTA measurements. In addition, the reported odds ratios are unusually narrow, which raises the possibility that model outputs reflect statistical classification rather than directly interpretable epidemiologic effect estimates.
Biologically, the biphasic CCFD pattern is intriguing. A plausible framework is that early neurovascular dysregulation leads to altered autoregulation and compensatory vascular responses during prodromal disease, followed by capillary dropout or impaired perfusion in dementia. This aligns with broader concepts of neurovascular uncoupling in Alzheimer disease, including endothelial dysfunction, pericyte injury, blood-brain barrier disruption, and amyloid-related vascular toxicity. Nonetheless, OCTA measures flow signal rather than absolute blood flow, so mechanistic conclusions should remain cautious.
Practical implications
If future studies confirm these findings, OCTA could become part of a multimodal risk-assessment pathway for cognitive decline. In ophthalmology clinics, retinal imaging is already routine, and automated extraction of vascular and neuroretinal metrics could allow opportunistic screening. In neurology and geriatric medicine, OCTA might serve as a referral tool to prioritize patients for more specialized biomarker testing.
For now, the technology should be viewed as investigational. It is not ready to replace established diagnostic approaches for Alzheimer disease, and there is not yet enough evidence to use OCTA alone to diagnose MCI or dementia. The most realistic near-term role is as a complementary biomarker in research settings and possibly as a future component of screening algorithms once longitudinal performance, reproducibility, and threshold values are established.
Conclusion
In this small cross-sectional cohort, SS-OCTA detected retinal and choriocapillaris abnormalities associated with cognitive status, including thinner GCC in AD dementia, reduced retinal vessel density with worsening cognition, and a notable biphasic CCFD pattern across disease stages. The findings support the concept that Alzheimer disease leaves a measurable neurovascular footprint in the eye.
These results are promising but preliminary. Larger, longitudinal, multicenter studies are needed to determine whether OCTA biomarkers can predict progression from normal cognition to MCI and from MCI to dementia, and whether they can add clinically useful information beyond established cognitive and biomarker-based approaches.
Funding and clinicaltrials.gov
The abstract provided does not specify funding sources or a clinicaltrials.gov registration number. No trial registration was reported in the supplied citation.
References
1. Zhang Y, Jiang Y, Edalati K, et al. Quantitative Swept-Source Optical Coherence Tomography Angiography Indicators of Neurovascular Dysfunction in Alzheimer Disease. JAMA Ophthalmology. 2026; published online June 11, 2026. PMID: 42275077.
2. Hampel H, O’Bryant SE, Molinuevo JL, et al. Blood-based biomarkers for Alzheimer disease: mapping the road to the clinic. Nat Rev Neurol. 2018;14(11):639-652.
3. Hampel H, Cummings J, Blennow K, et al. Developing the ATX(N) classification for use across the Alzheimer disease continuum. Nat Rev Neurol. 2021;17(9):580-589.
4. Nian S, et al. Retinal imaging biomarkers in Alzheimer disease and related dementias: current evidence and future directions. Alzheimers Dement. 2024; available in PubMed-indexed literature.